Cleveland Clinic Journal of Medicine

This article aims to introduce the statistical methodology behind chi-square and Fisher’s exact tests, which are commonly used in medical research to assess associations between categorical variables. This discussion will use data from a study by Mrozek¹ in patients with acute respiratory distress syndrome (ARDS). This was a multicenter, prospective, observational study: multicenter because it included data from 10 intensive care units, prospective because the study collected the data moving forward in time, and observational because the study investigators did not have control over the group assignments but rather used the naturally occurring groups. The study objective was to characterize focal and nonfocal patterns of lung computed tomography (CT)-based imaging with plasma markers of lung injury.

The primary grouping variable was type of ARDS (focal vs nonfocal) as determined by CT scans and other lung imaging tools. In this study, there were 32 (27%) patients with focal ARDS and 87 (73%) patients with nonfocal ARDS. What will be important, however, is classifying the type of variables because this determines the type of analyses performed. Type of ARDS is a categorical variable with 2 levels.

The primary study endpoint was plasma levels of the soluble form of the receptor for advanced glycation end product. There were also a number of secondary study endpoints that can be grouped as either patient outcomes or biomarkers. Patient outcomes included the duration of mechanical ventilation and both 28- and 90-day mortality. Levels of other biomarkers included surfactant protein D, soluble intercellular adhesion molecule-1, and plasminogen activator inhibitor-1.

This article focused on the secondary outcome of 90-day mortality beginning at disease onset. Again, we are interested in classifying this variable, which is categorical with 2 levels (yes vs no). So the scenario is that we want to assess the relationship between the type of ARDS (focal vs nonfocal) and 90-day mortality (yes vs no). In its most basic form, this scenario is an investigation into the association among 2 categorical variables.

When there are 2 categorical variables, the data can be arranged in what is called a contingency table (Figure 1). Because both variables are binary (2 levels), it is called a 2 × 2 table. However, a contingency table can be generated for 2 categorical variables with any number of levels—in that case, it is called an r ×c table, where r is the number of levels for the row variable and c is the number of levels for the column variable. The actual raw counts or frequencies are recorded inside the table cells. The cell counts are often referred to as observed counts and thus the notation (O_ij) is used. The subscript i identifies the specific level of the row variable, and in this example it can equal 1 or 2 since the row variable is binary. Similarly, the subscript j identifies the specific level of the column variable and in this example it can equal 1 or 2 since the column variable is binary. Therefore, O₁₁ represents the number of patients who have the row variable = level 1 and the column variable = level 1.

In addition to the row and column variable cells, there are also the margin totals. These totals are either the row margin total (summing across the row) or the column margin total (summing down the column). For example, n₁₊ is the sum of the row where the row variable equal 1 (O₁₁ + O₁₂ = n₁₊). Finally, at the very bottom right corner is the grand total, which equals the sample size. 

The goal is to test whether or not these 2 categorical variables are associated with each other. The null hypothesis (H_o) is that there is no association between these 2 categorical variables and the alternative hypotheses (H_a) is that there is an association between these 2 categorical variables.

The next step is to translate the generic form of the hypotheses into hypotheses that are specific to the research question. In this case, the null hypothesis is that mortality is not associated with lung morphology and the alternative hypothesis is that mortality is associated with lung morphology.

The contingency table cells can be populated with the numbers found in the article. It has our outcome of focus—mortality at day 90—both the count and the percent. The results are broken down by type of ARDS (focal vs nonfocal) as follows:

• Focal ARDS = 6 patients (21.4%)
• Nonfocal ARDS = 35 patients (45.5%).

First, the row variable is lung morphology, and it has two levels (focal vs nonfocal). Next, the column variable is 90-day mortality and it has 2 levels (yes vs no). Finally, the table must be populated, but be careful not to assume that there are no missing data. Begin with the cell counts: there were 6 focal ARDS patients and 35 nonfocal ARDS patients who died within 90 days. These two numbers populate the first column and result in a column total of 41. Next, use the reported percentages to calculate the row totals. Six is 21.4% of 28, so the first row total is 28. Thirty-five is 45.5% of 77, so the second row total is 77. If there are 28 patients with focal ARDS and 77 with nonfocal ARDS, then the grand total is 28 + 77 = 105. The remaining values can be obtained by subtraction. If there are 105 total patients and 41 die within 90 days, then 105 − 41 = 64 patients who do not die within 90 days and this is the second column total. Similarly, if there are 28 focal ARDS patients and 6 die within 90 days, then 28 − 6 = 22 patients who do not die within 90 days. Lastly, if there are 77 nonfocal ARDS patients and 35 die within 90 days, then 77 − 35 = 42 patients who do not die within 90 days. Now the contingency table is complete.

Once the contingency table is built, the question becomes, “Is lung morphology associated with 90-day mortality?” To answer that question, we need to know how many patients one would expect in each table cell if the null hypothesis of no association is true. When conducting a hypothesis test, one always assumes that the null hypothesis is true and then gathers data to see how well the data aligns with that assumption.

So one must calculate how many patients to expect in each of these cells if lung morphology is not associated with 90-day mortality. One way to address this question is to ask these 2 questions:

(1) Overall, what proportion of patients die by day 90? Looking at the constructed contingency table, that answer would be 39%. This was calculated by taking the total number of patients who died by day 90 and dividing it by the total number of patients, 41/105 = 39%. This gives the overall proportion, based on the data, who would die by day 90.

(2) How many of the focal ARDS patients would be expected to die by day 90? Now it is not overall, but rather we are limiting the question to the focal ARDS group. To obtain the answer, multiply the overall proportion of patients who die by day 90 by how many focal ARDS patients are in the study. Essentially, take the answer from the previous question and multiply it by the total number of focal ARDS, which is 28. The result is (41/105) × 28 = 10.9. Thus, if there is no association among long morphology and 90-day mortality, one would expect 10.9 focal ARDS patients to die by day 90.

Now 10.9 is a very specific answer for a specific contingency table, but the answer could be written in general terms. Basically, 3 numbers were used in calculating the solution: the row margin, the column margin, and the grand total. The general formula is the following:

The notation E_ij is used to represent the expected count assuming the null hypothesis of no association among the row and column variables is true. To calculate the expected count, take the i^th row total times the j^th column total and divide by the grand total.

In the lung morphology and mortality example, what is the expected number of deaths within 90 days among the nonfocal ARDS patients? This is the second row and the first column (E₂₁). Applying the formula, one multiplies the total for the second row by the total for the first column and then divides by the grand total, (77 × 41)/105 = 30.1. This calculation is repeated for each of the 4 cells.

Because we now know the observed cell count and the expected cell count (under the null hypothesis), we can compare the observed and expected counts to see how well the data aligns with the null hypothesis. This is what the chi-square test does, and the test statistic is calculated as follows:

The sigma (Σ) means addition, so the calculation is performed on each individual cell in the contingency table and then the results are summed. A 2 × 2 table has 4 cells and thus 4 numbers will be summed. For each cell, the formula compares the observed to the expected. Basically, it computes how similar they are (that is the O minus E part). Because the differences will be positive for some cells and negative for others, the differences are squared to avoid cancellation when you add them. Finally, each squared difference is divided by the expected count to standardize the calculation.

Intuitively, if the observed counts (O_ij) are similar to the expected counts under the null hypothesis (E_ij), then these 2 numbers will be very close to each other. When taking the difference between them or subtracting them, the result is a small number. When squaring a small number, one obtains a really small number. And adding up a bunch of really small numbers results in a small number. So the test statistic is going to be small. That means that the resulting P value is going to be large. What is a P value? Think of it as an index of compatibility. How compatible is the data with the null hypothesis? Here, you get a large index of compatibility. That means that the data aligns nicely with the null hypothesis and one fails to reject the null.

Now, think about the alternative scenario. If the observed counts (O_ij) are wildly different from the expected counts under the null hypothesis (E_ij), then these 2 numbers will be quite different. When taking the difference between them or subtracting them, the result is a big number. When squaring a big number, one obtains a really big number, and adding up a bunch of really big numbers results in a large number. So the test statistic is going to be large. That means that the resulting P value is going to be small. And if you think of a P value as an index of compatibility, the data and the null hypothesis are not very compatible. That means that the data does not align nicely with the null hypothesis and one rejects the null. This is the general idea of the chi-square test. It assesses how compatible the data is with the null hypothesis that the 2 categorical variables are not associated.

To obtain the actual P value, the distribution of the test statistic (under the null hypothesis) is used to calculate the area under the curve for values equal to the test statistic or more extreme. The described test statistic has an approximate chi-square distribution with (r − 1)(c − 1) degree of freedom. Recall that r is the number of levels of the row variable and c is the number of levels of the column variable. Our example is a 2 × 2 table, so the test statistic has an approximate chi-square distribution with (2 − 1)(2 − 1) = 1 degree of freedom.

Now that the chi-square test has been fully described, the assumptions for the test must be discussed. It is important to know when you should or should not perform this test. The chi-square test assumes that observations are independent. This means that the outcome for one observation is not associated with the outcome of any other observation. This principle can be violated when multiple measurements are taken over time or when multiple measurements are taken from one patient.

Another assumption is that the chi-square large sample approximation just described is appropriate. In other words, no more than 20% of the expected counts (E_ij) are less than 5. For a 2 × 2 table, how many cells do you have? Four. So if even one of those 4 happens to have an expected count less than 5, this assumption is violated. For a 2 × 2 table, none of the expected counts can be less than 5.

Returning to the lung morphology and mortality example, were the assumptions met? The data consist of 105 unique patients. Thus, we can assume that they are independent. The minimum expected count was 10.9, which is not less than 5. Therefore, the assumptions for the chi-square test are met. Next, the test statistic is calculated using the observed and expected counts. For each cell, subtract the expected count from the observed count, square it, and divide by the expected count. Then, add the 4 resulting numbers to obtain the test statistic of 4.92.

Finally, compute the area under the chi-square distribution with 1 degree of freedom Χ²₍₁₎, at the test statistic and values more extreme. In this case, values more extreme are values greater than the test statistic. Here, the area under the curve to the right of 4.92 is .027 (Figure 3). This is the P value, which indicates that the data and the null hypothesis have very low compatibility. In this example, the area under the curve to the right of 4.92 is .027 (Figure 3). This is the P value, which indicates that the data and the null hypothesis have very low compatibility. Thus, the decision is to reject the null hypothesis. The conclusion is that lung morphology is associated with 90-day mortality (P = .027). To describe that association, one looks at the contingency table and finds a reduction in 90-day mortality with focal patterns compared to nonfocal patterns (21.4% vs 45.5%, respectively). The P value reported in the article is .026. Our hand calculation was .027, which is slightly off due to rounding. In summary, the scenario is an investigation into the association among 2 categorical variables, and, thus, a test to consider is the chi-square test, if assumptions are met.

In another example in the same study, the authors investigate whether any baseline characteristics are associated with lung morphology. For example, is neurology, specifically Parkinson disease (yes vs no), associated with lung morphology (focal vs nonfocal)? Again, the scenario is an investigation into the association between 2 categorical variables, so a chi-square test should be considered.

To start, build a contingency table arbitrarily placing lung morphology as the row variable and Parkinson disease as the column variable. Populate the contingency table based on the counts and percentages reported in the article (Figure 4). Next, check that the assumptions of the chi-square test are met. Are the observations independent? Again, because these are unique patients, we consider this assumption met. Since this is a 2 × 2 table, are all of the expected counts greater than 5? Calculations of the expected counts obtained the following: 1.1, 30.9, 2.9 and 84.1. Here, 2 of the 4 expected counts are less than 5. Therefore, methods that use large sample approximation, like the chi-squared test, may not be an appropriate choice.

Instead of using methodology that is an approximation, consider an exact test such as Fisher’s exact test. Again, refer to the contingency table where Fisher’s exact is going to calculate the exact probability (under the null hypothesis) of the observed data or results more extreme. This is the technical definition of a P value. It is, however, still quantifying how compatible the data are with the null hypothesis. The exact probability of a particular contingency table can be obtained using the hypergeometric distribution.

The symbols that resemble large parentheses are notations for a combinatorial. Because using combinatorials to calculate the probability is not user friendly, an equivalent version relies on factorials instead. Both techniques are presented above. Remember that the goal is to find the exact probability of the observed data or something more extreme.

The hypotheses are still testing whether these 2 categorical variables are associated with each other. In this particular example, we test if the proportion of patients with Parkinson disease is the same in the focal and nonfocal groups. Fisher’s exact test obtains its two-tailed P value by computing the probabilities associated with all possible tables that have the same row and column totals. Then, it identifies the alternative tables with a probability that is less than that of the observed table. Finally, it adds the probability of the observed table with the sum of the probabilities of each alternative table identified above, which results in the P value.

To explore each of those steps in detail, one must first enumerate how many tables can be built that all have the same row and column totals as the observed table. Figure 5 shows the 5 possible tables. Pick any one of the 5 2 × 2 tables; the margins are fixed. Each table has the same row totals, 32 focal and 87 nonfocal, and each table has the same column totals: 4 Parkinson and 115 non-Parkinson. Then, for each table, calculate the probability of that table. Figure 5 shows this calculation for the first 2 × 2 table, which happens to be the observed table. The probability of the table observed in the study is .2803. Such a calculation is performed on each of the other tables.

Next, one must identify the tables that have a probability smaller than the observed table. Here, we are looking for probabilities less than .2803. These are the tables deemed more extreme. Tables 3, 4, and 5 have probabilities less than .2803.

The final step is to sum the probability of the observed table and the more extreme tables (ie, those with probabilities < the observed table) (.2803 + .2337 + .0543 + .0045 = .5728). Thus, the resulting rounded P value is .57, which indicates a high level of compatibility between the data and the null hypothesis of no association. The decision is to fail to reject the null hypothesis and the conclusion is that the evidence does not support an association among lung morphology and Parkinson disease. In other words, there is insufficient evidence to claim that the proportion of Parkinson disease differs between the focal and nonfocal ARDS patients (0% vs 5%, P = .57). This matches the P value reported by Mrozek for this association.

The first objective of this article was to identify scenarios in which a chi-square or Fisher’s exact test should be considered. The general setting discussed was an investigation of the association between two categorical variables. Use of each test specifically depends on whether the assumptions have been met. Both of the examples used in our discussion happened to be binary, but that is not a restriction. Categorical variables can have more than 2 levels. All of the methods demonstrated for 2 × 2 tables can be generalized to r × c tables. 

The second objective of this article was to recognize when test assumptions have been violated. For simplicity, most researchers adhere to the following: if ≤ 20% of expected cell counts are less than 5, then use the chi-square test; if > 20% of expected cell counts are less than 5, then use Fisher’s exact test. Both methods assume that the observations are independent. Could one use the exact test when the chi-square assumptions are met? Yes, but it is more computationally expensive as it uses all possible fixed margin tables and their probabilities. If the chi-square assumptions are met, then the sample size is typically larger and these calculations become numerous. Also, it does not have to be that large of a sample for the chi-square to be a good approximation and do it very quickly.

The final objective of this article was to test claims made regarding the association of 2 independent categorical variables. We included examples from the medical literature showing step-by-step calculations of both the large sample approximation (chi-square) and exact (Fisher’s) methodologies providing insight into how these tests are conducted as well as when they are appropriate.

This article aims to introduce the statistical methodology behind chi-square and Fisher’s exact tests, which are commonly used in medical research to assess associations between categorical variables. This discussion will use data from a study by Mrozek¹ in patients with acute respiratory distress syndrome (ARDS). This was a multicenter, prospective, observational study: multicenter because it included data from 10 intensive care units, prospective because the study collected the data moving forward in time, and observational because the study investigators did not have control over the group assignments but rather used the naturally occurring groups. The study objective was to characterize focal and nonfocal patterns of lung computed tomography (CT)-based imaging with plasma markers of lung injury.

The primary grouping variable was type of ARDS (focal vs nonfocal) as determined by CT scans and other lung imaging tools. In this study, there were 32 (27%) patients with focal ARDS and 87 (73%) patients with nonfocal ARDS. What will be important, however, is classifying the type of variables because this determines the type of analyses performed. Type of ARDS is a categorical variable with 2 levels.

The primary study endpoint was plasma levels of the soluble form of the receptor for advanced glycation end product. There were also a number of secondary study endpoints that can be grouped as either patient outcomes or biomarkers. Patient outcomes included the duration of mechanical ventilation and both 28- and 90-day mortality. Levels of other biomarkers included surfactant protein D, soluble intercellular adhesion molecule-1, and plasminogen activator inhibitor-1.

This article focused on the secondary outcome of 90-day mortality beginning at disease onset. Again, we are interested in classifying this variable, which is categorical with 2 levels (yes vs no). So the scenario is that we want to assess the relationship between the type of ARDS (focal vs nonfocal) and 90-day mortality (yes vs no). In its most basic form, this scenario is an investigation into the association among 2 categorical variables.

When there are 2 categorical variables, the data can be arranged in what is called a contingency table (Figure 1). Because both variables are binary (2 levels), it is called a 2 × 2 table. However, a contingency table can be generated for 2 categorical variables with any number of levels—in that case, it is called an r ×c table, where r is the number of levels for the row variable and c is the number of levels for the column variable. The actual raw counts or frequencies are recorded inside the table cells. The cell counts are often referred to as observed counts and thus the notation (O_ij) is used. The subscript i identifies the specific level of the row variable, and in this example it can equal 1 or 2 since the row variable is binary. Similarly, the subscript j identifies the specific level of the column variable and in this example it can equal 1 or 2 since the column variable is binary. Therefore, O₁₁ represents the number of patients who have the row variable = level 1 and the column variable = level 1.

In addition to the row and column variable cells, there are also the margin totals. These totals are either the row margin total (summing across the row) or the column margin total (summing down the column). For example, n₁₊ is the sum of the row where the row variable equal 1 (O₁₁ + O₁₂ = n₁₊). Finally, at the very bottom right corner is the grand total, which equals the sample size. 

The goal is to test whether or not these 2 categorical variables are associated with each other. The null hypothesis (H_o) is that there is no association between these 2 categorical variables and the alternative hypotheses (H_a) is that there is an association between these 2 categorical variables.

The next step is to translate the generic form of the hypotheses into hypotheses that are specific to the research question. In this case, the null hypothesis is that mortality is not associated with lung morphology and the alternative hypothesis is that mortality is associated with lung morphology.

The contingency table cells can be populated with the numbers found in the article. It has our outcome of focus—mortality at day 90—both the count and the percent. The results are broken down by type of ARDS (focal vs nonfocal) as follows:

• Focal ARDS = 6 patients (21.4%)
• Nonfocal ARDS = 35 patients (45.5%).

First, the row variable is lung morphology, and it has two levels (focal vs nonfocal). Next, the column variable is 90-day mortality and it has 2 levels (yes vs no). Finally, the table must be populated, but be careful not to assume that there are no missing data. Begin with the cell counts: there were 6 focal ARDS patients and 35 nonfocal ARDS patients who died within 90 days. These two numbers populate the first column and result in a column total of 41. Next, use the reported percentages to calculate the row totals. Six is 21.4% of 28, so the first row total is 28. Thirty-five is 45.5% of 77, so the second row total is 77. If there are 28 patients with focal ARDS and 77 with nonfocal ARDS, then the grand total is 28 + 77 = 105. The remaining values can be obtained by subtraction. If there are 105 total patients and 41 die within 90 days, then 105 − 41 = 64 patients who do not die within 90 days and this is the second column total. Similarly, if there are 28 focal ARDS patients and 6 die within 90 days, then 28 − 6 = 22 patients who do not die within 90 days. Lastly, if there are 77 nonfocal ARDS patients and 35 die within 90 days, then 77 − 35 = 42 patients who do not die within 90 days. Now the contingency table is complete.

Once the contingency table is built, the question becomes, “Is lung morphology associated with 90-day mortality?” To answer that question, we need to know how many patients one would expect in each table cell if the null hypothesis of no association is true. When conducting a hypothesis test, one always assumes that the null hypothesis is true and then gathers data to see how well the data aligns with that assumption.

So one must calculate how many patients to expect in each of these cells if lung morphology is not associated with 90-day mortality. One way to address this question is to ask these 2 questions:

(1) Overall, what proportion of patients die by day 90? Looking at the constructed contingency table, that answer would be 39%. This was calculated by taking the total number of patients who died by day 90 and dividing it by the total number of patients, 41/105 = 39%. This gives the overall proportion, based on the data, who would die by day 90.

(2) How many of the focal ARDS patients would be expected to die by day 90? Now it is not overall, but rather we are limiting the question to the focal ARDS group. To obtain the answer, multiply the overall proportion of patients who die by day 90 by how many focal ARDS patients are in the study. Essentially, take the answer from the previous question and multiply it by the total number of focal ARDS, which is 28. The result is (41/105) × 28 = 10.9. Thus, if there is no association among long morphology and 90-day mortality, one would expect 10.9 focal ARDS patients to die by day 90.

Now 10.9 is a very specific answer for a specific contingency table, but the answer could be written in general terms. Basically, 3 numbers were used in calculating the solution: the row margin, the column margin, and the grand total. The general formula is the following:

The notation E_ij is used to represent the expected count assuming the null hypothesis of no association among the row and column variables is true. To calculate the expected count, take the i^th row total times the j^th column total and divide by the grand total.

In the lung morphology and mortality example, what is the expected number of deaths within 90 days among the nonfocal ARDS patients? This is the second row and the first column (E₂₁). Applying the formula, one multiplies the total for the second row by the total for the first column and then divides by the grand total, (77 × 41)/105 = 30.1. This calculation is repeated for each of the 4 cells.

Because we now know the observed cell count and the expected cell count (under the null hypothesis), we can compare the observed and expected counts to see how well the data aligns with the null hypothesis. This is what the chi-square test does, and the test statistic is calculated as follows:

The sigma (Σ) means addition, so the calculation is performed on each individual cell in the contingency table and then the results are summed. A 2 × 2 table has 4 cells and thus 4 numbers will be summed. For each cell, the formula compares the observed to the expected. Basically, it computes how similar they are (that is the O minus E part). Because the differences will be positive for some cells and negative for others, the differences are squared to avoid cancellation when you add them. Finally, each squared difference is divided by the expected count to standardize the calculation.

Intuitively, if the observed counts (O_ij) are similar to the expected counts under the null hypothesis (E_ij), then these 2 numbers will be very close to each other. When taking the difference between them or subtracting them, the result is a small number. When squaring a small number, one obtains a really small number. And adding up a bunch of really small numbers results in a small number. So the test statistic is going to be small. That means that the resulting P value is going to be large. What is a P value? Think of it as an index of compatibility. How compatible is the data with the null hypothesis? Here, you get a large index of compatibility. That means that the data aligns nicely with the null hypothesis and one fails to reject the null.

Now, think about the alternative scenario. If the observed counts (O_ij) are wildly different from the expected counts under the null hypothesis (E_ij), then these 2 numbers will be quite different. When taking the difference between them or subtracting them, the result is a big number. When squaring a big number, one obtains a really big number, and adding up a bunch of really big numbers results in a large number. So the test statistic is going to be large. That means that the resulting P value is going to be small. And if you think of a P value as an index of compatibility, the data and the null hypothesis are not very compatible. That means that the data does not align nicely with the null hypothesis and one rejects the null. This is the general idea of the chi-square test. It assesses how compatible the data is with the null hypothesis that the 2 categorical variables are not associated.

To obtain the actual P value, the distribution of the test statistic (under the null hypothesis) is used to calculate the area under the curve for values equal to the test statistic or more extreme. The described test statistic has an approximate chi-square distribution with (r − 1)(c − 1) degree of freedom. Recall that r is the number of levels of the row variable and c is the number of levels of the column variable. Our example is a 2 × 2 table, so the test statistic has an approximate chi-square distribution with (2 − 1)(2 − 1) = 1 degree of freedom.

Now that the chi-square test has been fully described, the assumptions for the test must be discussed. It is important to know when you should or should not perform this test. The chi-square test assumes that observations are independent. This means that the outcome for one observation is not associated with the outcome of any other observation. This principle can be violated when multiple measurements are taken over time or when multiple measurements are taken from one patient.

Another assumption is that the chi-square large sample approximation just described is appropriate. In other words, no more than 20% of the expected counts (E_ij) are less than 5. For a 2 × 2 table, how many cells do you have? Four. So if even one of those 4 happens to have an expected count less than 5, this assumption is violated. For a 2 × 2 table, none of the expected counts can be less than 5.

Returning to the lung morphology and mortality example, were the assumptions met? The data consist of 105 unique patients. Thus, we can assume that they are independent. The minimum expected count was 10.9, which is not less than 5. Therefore, the assumptions for the chi-square test are met. Next, the test statistic is calculated using the observed and expected counts. For each cell, subtract the expected count from the observed count, square it, and divide by the expected count. Then, add the 4 resulting numbers to obtain the test statistic of 4.92.

Finally, compute the area under the chi-square distribution with 1 degree of freedom Χ²₍₁₎, at the test statistic and values more extreme. In this case, values more extreme are values greater than the test statistic. Here, the area under the curve to the right of 4.92 is .027 (Figure 3). This is the P value, which indicates that the data and the null hypothesis have very low compatibility. In this example, the area under the curve to the right of 4.92 is .027 (Figure 3). This is the P value, which indicates that the data and the null hypothesis have very low compatibility. Thus, the decision is to reject the null hypothesis. The conclusion is that lung morphology is associated with 90-day mortality (P = .027). To describe that association, one looks at the contingency table and finds a reduction in 90-day mortality with focal patterns compared to nonfocal patterns (21.4% vs 45.5%, respectively). The P value reported in the article is .026. Our hand calculation was .027, which is slightly off due to rounding. In summary, the scenario is an investigation into the association among 2 categorical variables, and, thus, a test to consider is the chi-square test, if assumptions are met.

In another example in the same study, the authors investigate whether any baseline characteristics are associated with lung morphology. For example, is neurology, specifically Parkinson disease (yes vs no), associated with lung morphology (focal vs nonfocal)? Again, the scenario is an investigation into the association between 2 categorical variables, so a chi-square test should be considered.

To start, build a contingency table arbitrarily placing lung morphology as the row variable and Parkinson disease as the column variable. Populate the contingency table based on the counts and percentages reported in the article (Figure 4). Next, check that the assumptions of the chi-square test are met. Are the observations independent? Again, because these are unique patients, we consider this assumption met. Since this is a 2 × 2 table, are all of the expected counts greater than 5? Calculations of the expected counts obtained the following: 1.1, 30.9, 2.9 and 84.1. Here, 2 of the 4 expected counts are less than 5. Therefore, methods that use large sample approximation, like the chi-squared test, may not be an appropriate choice.

Instead of using methodology that is an approximation, consider an exact test such as Fisher’s exact test. Again, refer to the contingency table where Fisher’s exact is going to calculate the exact probability (under the null hypothesis) of the observed data or results more extreme. This is the technical definition of a P value. It is, however, still quantifying how compatible the data are with the null hypothesis. The exact probability of a particular contingency table can be obtained using the hypergeometric distribution.

The symbols that resemble large parentheses are notations for a combinatorial. Because using combinatorials to calculate the probability is not user friendly, an equivalent version relies on factorials instead. Both techniques are presented above. Remember that the goal is to find the exact probability of the observed data or something more extreme.

The hypotheses are still testing whether these 2 categorical variables are associated with each other. In this particular example, we test if the proportion of patients with Parkinson disease is the same in the focal and nonfocal groups. Fisher’s exact test obtains its two-tailed P value by computing the probabilities associated with all possible tables that have the same row and column totals. Then, it identifies the alternative tables with a probability that is less than that of the observed table. Finally, it adds the probability of the observed table with the sum of the probabilities of each alternative table identified above, which results in the P value.

To explore each of those steps in detail, one must first enumerate how many tables can be built that all have the same row and column totals as the observed table. Figure 5 shows the 5 possible tables. Pick any one of the 5 2 × 2 tables; the margins are fixed. Each table has the same row totals, 32 focal and 87 nonfocal, and each table has the same column totals: 4 Parkinson and 115 non-Parkinson. Then, for each table, calculate the probability of that table. Figure 5 shows this calculation for the first 2 × 2 table, which happens to be the observed table. The probability of the table observed in the study is .2803. Such a calculation is performed on each of the other tables.

Next, one must identify the tables that have a probability smaller than the observed table. Here, we are looking for probabilities less than .2803. These are the tables deemed more extreme. Tables 3, 4, and 5 have probabilities less than .2803.

The final step is to sum the probability of the observed table and the more extreme tables (ie, those with probabilities < the observed table) (.2803 + .2337 + .0543 + .0045 = .5728). Thus, the resulting rounded P value is .57, which indicates a high level of compatibility between the data and the null hypothesis of no association. The decision is to fail to reject the null hypothesis and the conclusion is that the evidence does not support an association among lung morphology and Parkinson disease. In other words, there is insufficient evidence to claim that the proportion of Parkinson disease differs between the focal and nonfocal ARDS patients (0% vs 5%, P = .57). This matches the P value reported by Mrozek for this association.

The first objective of this article was to identify scenarios in which a chi-square or Fisher’s exact test should be considered. The general setting discussed was an investigation of the association between two categorical variables. Use of each test specifically depends on whether the assumptions have been met. Both of the examples used in our discussion happened to be binary, but that is not a restriction. Categorical variables can have more than 2 levels. All of the methods demonstrated for 2 × 2 tables can be generalized to r × c tables. 

The second objective of this article was to recognize when test assumptions have been violated. For simplicity, most researchers adhere to the following: if ≤ 20% of expected cell counts are less than 5, then use the chi-square test; if > 20% of expected cell counts are less than 5, then use Fisher’s exact test. Both methods assume that the observations are independent. Could one use the exact test when the chi-square assumptions are met? Yes, but it is more computationally expensive as it uses all possible fixed margin tables and their probabilities. If the chi-square assumptions are met, then the sample size is typically larger and these calculations become numerous. Also, it does not have to be that large of a sample for the chi-square to be a good approximation and do it very quickly.

The final objective of this article was to test claims made regarding the association of 2 independent categorical variables. We included examples from the medical literature showing step-by-step calculations of both the large sample approximation (chi-square) and exact (Fisher’s) methodologies providing insight into how these tests are conducted as well as when they are appropriate.

This article aims to introduce the statistical methodology behind chi-square and Fisher’s exact tests, which are commonly used in medical research to assess associations between categorical variables. This discussion will use data from a study by Mrozek¹ in patients with acute respiratory distress syndrome (ARDS). This was a multicenter, prospective, observational study: multicenter because it included data from 10 intensive care units, prospective because the study collected the data moving forward in time, and observational because the study investigators did not have control over the group assignments but rather used the naturally occurring groups. The study objective was to characterize focal and nonfocal patterns of lung computed tomography (CT)-based imaging with plasma markers of lung injury.

The primary grouping variable was type of ARDS (focal vs nonfocal) as determined by CT scans and other lung imaging tools. In this study, there were 32 (27%) patients with focal ARDS and 87 (73%) patients with nonfocal ARDS. What will be important, however, is classifying the type of variables because this determines the type of analyses performed. Type of ARDS is a categorical variable with 2 levels.

The primary study endpoint was plasma levels of the soluble form of the receptor for advanced glycation end product. There were also a number of secondary study endpoints that can be grouped as either patient outcomes or biomarkers. Patient outcomes included the duration of mechanical ventilation and both 28- and 90-day mortality. Levels of other biomarkers included surfactant protein D, soluble intercellular adhesion molecule-1, and plasminogen activator inhibitor-1.

This article focused on the secondary outcome of 90-day mortality beginning at disease onset. Again, we are interested in classifying this variable, which is categorical with 2 levels (yes vs no). So the scenario is that we want to assess the relationship between the type of ARDS (focal vs nonfocal) and 90-day mortality (yes vs no). In its most basic form, this scenario is an investigation into the association among 2 categorical variables.

When there are 2 categorical variables, the data can be arranged in what is called a contingency table (Figure 1). Because both variables are binary (2 levels), it is called a 2 × 2 table. However, a contingency table can be generated for 2 categorical variables with any number of levels—in that case, it is called an r ×c table, where r is the number of levels for the row variable and c is the number of levels for the column variable. The actual raw counts or frequencies are recorded inside the table cells. The cell counts are often referred to as observed counts and thus the notation (O_ij) is used. The subscript i identifies the specific level of the row variable, and in this example it can equal 1 or 2 since the row variable is binary. Similarly, the subscript j identifies the specific level of the column variable and in this example it can equal 1 or 2 since the column variable is binary. Therefore, O₁₁ represents the number of patients who have the row variable = level 1 and the column variable = level 1.

In addition to the row and column variable cells, there are also the margin totals. These totals are either the row margin total (summing across the row) or the column margin total (summing down the column). For example, n₁₊ is the sum of the row where the row variable equal 1 (O₁₁ + O₁₂ = n₁₊). Finally, at the very bottom right corner is the grand total, which equals the sample size. 

The goal is to test whether or not these 2 categorical variables are associated with each other. The null hypothesis (H_o) is that there is no association between these 2 categorical variables and the alternative hypotheses (H_a) is that there is an association between these 2 categorical variables.

The next step is to translate the generic form of the hypotheses into hypotheses that are specific to the research question. In this case, the null hypothesis is that mortality is not associated with lung morphology and the alternative hypothesis is that mortality is associated with lung morphology.

The contingency table cells can be populated with the numbers found in the article. It has our outcome of focus—mortality at day 90—both the count and the percent. The results are broken down by type of ARDS (focal vs nonfocal) as follows:

• Focal ARDS = 6 patients (21.4%)
• Nonfocal ARDS = 35 patients (45.5%).

First, the row variable is lung morphology, and it has two levels (focal vs nonfocal). Next, the column variable is 90-day mortality and it has 2 levels (yes vs no). Finally, the table must be populated, but be careful not to assume that there are no missing data. Begin with the cell counts: there were 6 focal ARDS patients and 35 nonfocal ARDS patients who died within 90 days. These two numbers populate the first column and result in a column total of 41. Next, use the reported percentages to calculate the row totals. Six is 21.4% of 28, so the first row total is 28. Thirty-five is 45.5% of 77, so the second row total is 77. If there are 28 patients with focal ARDS and 77 with nonfocal ARDS, then the grand total is 28 + 77 = 105. The remaining values can be obtained by subtraction. If there are 105 total patients and 41 die within 90 days, then 105 − 41 = 64 patients who do not die within 90 days and this is the second column total. Similarly, if there are 28 focal ARDS patients and 6 die within 90 days, then 28 − 6 = 22 patients who do not die within 90 days. Lastly, if there are 77 nonfocal ARDS patients and 35 die within 90 days, then 77 − 35 = 42 patients who do not die within 90 days. Now the contingency table is complete.

Once the contingency table is built, the question becomes, “Is lung morphology associated with 90-day mortality?” To answer that question, we need to know how many patients one would expect in each table cell if the null hypothesis of no association is true. When conducting a hypothesis test, one always assumes that the null hypothesis is true and then gathers data to see how well the data aligns with that assumption.

So one must calculate how many patients to expect in each of these cells if lung morphology is not associated with 90-day mortality. One way to address this question is to ask these 2 questions:

(1) Overall, what proportion of patients die by day 90? Looking at the constructed contingency table, that answer would be 39%. This was calculated by taking the total number of patients who died by day 90 and dividing it by the total number of patients, 41/105 = 39%. This gives the overall proportion, based on the data, who would die by day 90.

(2) How many of the focal ARDS patients would be expected to die by day 90? Now it is not overall, but rather we are limiting the question to the focal ARDS group. To obtain the answer, multiply the overall proportion of patients who die by day 90 by how many focal ARDS patients are in the study. Essentially, take the answer from the previous question and multiply it by the total number of focal ARDS, which is 28. The result is (41/105) × 28 = 10.9. Thus, if there is no association among long morphology and 90-day mortality, one would expect 10.9 focal ARDS patients to die by day 90.

Now 10.9 is a very specific answer for a specific contingency table, but the answer could be written in general terms. Basically, 3 numbers were used in calculating the solution: the row margin, the column margin, and the grand total. The general formula is the following:

The notation E_ij is used to represent the expected count assuming the null hypothesis of no association among the row and column variables is true. To calculate the expected count, take the i^th row total times the j^th column total and divide by the grand total.

In the lung morphology and mortality example, what is the expected number of deaths within 90 days among the nonfocal ARDS patients? This is the second row and the first column (E₂₁). Applying the formula, one multiplies the total for the second row by the total for the first column and then divides by the grand total, (77 × 41)/105 = 30.1. This calculation is repeated for each of the 4 cells.

Because we now know the observed cell count and the expected cell count (under the null hypothesis), we can compare the observed and expected counts to see how well the data aligns with the null hypothesis. This is what the chi-square test does, and the test statistic is calculated as follows:

The sigma (Σ) means addition, so the calculation is performed on each individual cell in the contingency table and then the results are summed. A 2 × 2 table has 4 cells and thus 4 numbers will be summed. For each cell, the formula compares the observed to the expected. Basically, it computes how similar they are (that is the O minus E part). Because the differences will be positive for some cells and negative for others, the differences are squared to avoid cancellation when you add them. Finally, each squared difference is divided by the expected count to standardize the calculation.

Intuitively, if the observed counts (O_ij) are similar to the expected counts under the null hypothesis (E_ij), then these 2 numbers will be very close to each other. When taking the difference between them or subtracting them, the result is a small number. When squaring a small number, one obtains a really small number. And adding up a bunch of really small numbers results in a small number. So the test statistic is going to be small. That means that the resulting P value is going to be large. What is a P value? Think of it as an index of compatibility. How compatible is the data with the null hypothesis? Here, you get a large index of compatibility. That means that the data aligns nicely with the null hypothesis and one fails to reject the null.

Now, think about the alternative scenario. If the observed counts (O_ij) are wildly different from the expected counts under the null hypothesis (E_ij), then these 2 numbers will be quite different. When taking the difference between them or subtracting them, the result is a big number. When squaring a big number, one obtains a really big number, and adding up a bunch of really big numbers results in a large number. So the test statistic is going to be large. That means that the resulting P value is going to be small. And if you think of a P value as an index of compatibility, the data and the null hypothesis are not very compatible. That means that the data does not align nicely with the null hypothesis and one rejects the null. This is the general idea of the chi-square test. It assesses how compatible the data is with the null hypothesis that the 2 categorical variables are not associated.

To obtain the actual P value, the distribution of the test statistic (under the null hypothesis) is used to calculate the area under the curve for values equal to the test statistic or more extreme. The described test statistic has an approximate chi-square distribution with (r − 1)(c − 1) degree of freedom. Recall that r is the number of levels of the row variable and c is the number of levels of the column variable. Our example is a 2 × 2 table, so the test statistic has an approximate chi-square distribution with (2 − 1)(2 − 1) = 1 degree of freedom.

Now that the chi-square test has been fully described, the assumptions for the test must be discussed. It is important to know when you should or should not perform this test. The chi-square test assumes that observations are independent. This means that the outcome for one observation is not associated with the outcome of any other observation. This principle can be violated when multiple measurements are taken over time or when multiple measurements are taken from one patient.

Another assumption is that the chi-square large sample approximation just described is appropriate. In other words, no more than 20% of the expected counts (E_ij) are less than 5. For a 2 × 2 table, how many cells do you have? Four. So if even one of those 4 happens to have an expected count less than 5, this assumption is violated. For a 2 × 2 table, none of the expected counts can be less than 5.

Returning to the lung morphology and mortality example, were the assumptions met? The data consist of 105 unique patients. Thus, we can assume that they are independent. The minimum expected count was 10.9, which is not less than 5. Therefore, the assumptions for the chi-square test are met. Next, the test statistic is calculated using the observed and expected counts. For each cell, subtract the expected count from the observed count, square it, and divide by the expected count. Then, add the 4 resulting numbers to obtain the test statistic of 4.92.

Finally, compute the area under the chi-square distribution with 1 degree of freedom Χ²₍₁₎, at the test statistic and values more extreme. In this case, values more extreme are values greater than the test statistic. Here, the area under the curve to the right of 4.92 is .027 (Figure 3). This is the P value, which indicates that the data and the null hypothesis have very low compatibility. In this example, the area under the curve to the right of 4.92 is .027 (Figure 3). This is the P value, which indicates that the data and the null hypothesis have very low compatibility. Thus, the decision is to reject the null hypothesis. The conclusion is that lung morphology is associated with 90-day mortality (P = .027). To describe that association, one looks at the contingency table and finds a reduction in 90-day mortality with focal patterns compared to nonfocal patterns (21.4% vs 45.5%, respectively). The P value reported in the article is .026. Our hand calculation was .027, which is slightly off due to rounding. In summary, the scenario is an investigation into the association among 2 categorical variables, and, thus, a test to consider is the chi-square test, if assumptions are met.

In another example in the same study, the authors investigate whether any baseline characteristics are associated with lung morphology. For example, is neurology, specifically Parkinson disease (yes vs no), associated with lung morphology (focal vs nonfocal)? Again, the scenario is an investigation into the association between 2 categorical variables, so a chi-square test should be considered.

To start, build a contingency table arbitrarily placing lung morphology as the row variable and Parkinson disease as the column variable. Populate the contingency table based on the counts and percentages reported in the article (Figure 4). Next, check that the assumptions of the chi-square test are met. Are the observations independent? Again, because these are unique patients, we consider this assumption met. Since this is a 2 × 2 table, are all of the expected counts greater than 5? Calculations of the expected counts obtained the following: 1.1, 30.9, 2.9 and 84.1. Here, 2 of the 4 expected counts are less than 5. Therefore, methods that use large sample approximation, like the chi-squared test, may not be an appropriate choice.

Instead of using methodology that is an approximation, consider an exact test such as Fisher’s exact test. Again, refer to the contingency table where Fisher’s exact is going to calculate the exact probability (under the null hypothesis) of the observed data or results more extreme. This is the technical definition of a P value. It is, however, still quantifying how compatible the data are with the null hypothesis. The exact probability of a particular contingency table can be obtained using the hypergeometric distribution.

The symbols that resemble large parentheses are notations for a combinatorial. Because using combinatorials to calculate the probability is not user friendly, an equivalent version relies on factorials instead. Both techniques are presented above. Remember that the goal is to find the exact probability of the observed data or something more extreme.

The hypotheses are still testing whether these 2 categorical variables are associated with each other. In this particular example, we test if the proportion of patients with Parkinson disease is the same in the focal and nonfocal groups. Fisher’s exact test obtains its two-tailed P value by computing the probabilities associated with all possible tables that have the same row and column totals. Then, it identifies the alternative tables with a probability that is less than that of the observed table. Finally, it adds the probability of the observed table with the sum of the probabilities of each alternative table identified above, which results in the P value.

To explore each of those steps in detail, one must first enumerate how many tables can be built that all have the same row and column totals as the observed table. Figure 5 shows the 5 possible tables. Pick any one of the 5 2 × 2 tables; the margins are fixed. Each table has the same row totals, 32 focal and 87 nonfocal, and each table has the same column totals: 4 Parkinson and 115 non-Parkinson. Then, for each table, calculate the probability of that table. Figure 5 shows this calculation for the first 2 × 2 table, which happens to be the observed table. The probability of the table observed in the study is .2803. Such a calculation is performed on each of the other tables.

Next, one must identify the tables that have a probability smaller than the observed table. Here, we are looking for probabilities less than .2803. These are the tables deemed more extreme. Tables 3, 4, and 5 have probabilities less than .2803.

The final step is to sum the probability of the observed table and the more extreme tables (ie, those with probabilities < the observed table) (.2803 + .2337 + .0543 + .0045 = .5728). Thus, the resulting rounded P value is .57, which indicates a high level of compatibility between the data and the null hypothesis of no association. The decision is to fail to reject the null hypothesis and the conclusion is that the evidence does not support an association among lung morphology and Parkinson disease. In other words, there is insufficient evidence to claim that the proportion of Parkinson disease differs between the focal and nonfocal ARDS patients (0% vs 5%, P = .57). This matches the P value reported by Mrozek for this association.

The first objective of this article was to identify scenarios in which a chi-square or Fisher’s exact test should be considered. The general setting discussed was an investigation of the association between two categorical variables. Use of each test specifically depends on whether the assumptions have been met. Both of the examples used in our discussion happened to be binary, but that is not a restriction. Categorical variables can have more than 2 levels. All of the methods demonstrated for 2 × 2 tables can be generalized to r × c tables. 

The second objective of this article was to recognize when test assumptions have been violated. For simplicity, most researchers adhere to the following: if ≤ 20% of expected cell counts are less than 5, then use the chi-square test; if > 20% of expected cell counts are less than 5, then use Fisher’s exact test. Both methods assume that the observations are independent. Could one use the exact test when the chi-square assumptions are met? Yes, but it is more computationally expensive as it uses all possible fixed margin tables and their probabilities. If the chi-square assumptions are met, then the sample size is typically larger and these calculations become numerous. Also, it does not have to be that large of a sample for the chi-square to be a good approximation and do it very quickly.

The final objective of this article was to test claims made regarding the association of 2 independent categorical variables. We included examples from the medical literature showing step-by-step calculations of both the large sample approximation (chi-square) and exact (Fisher’s) methodologies providing insight into how these tests are conducted as well as when they are appropriate.

Concussion, also known as mild traumatic brain injury, affects more than 600 adults per 100,000 each year and is commonly treated by nonneurologists.¹ Public attention to concussion has been increasing, particularly to concussion sustained during sports. Coincident with this increased attention, the diagnosis of concussion continues to increase in the outpatient setting. Thus, a review of the topic is timely.

ACCELERATION OF THE BRAIN DUE TO TRAUMA

The definition of concussion has changed considerably over the years. It is currently defined as a pathophysiologic process that results from an acceleration or deceleration of the brain induced by trauma.² It is largely a temporary, functional problem, as opposed to a gross structural injury.^2–5

The acceleration of the brain that results in a concussion is usually initiated by a direct blow to the head, although direct impact is not required.⁶ As the brain rotates, different areas accelerate at different rates, resulting in a shear strain imparted to the parenchyma.

This shear strain causes deformation of axonal membranes and opening of membrane-associated sodium-potassium channels. This in turn leads to release of excitatory neurotransmitters, ultimately culminating in a wave of neuronal depolarization and a spreading depression-like phenomenon that may mediate the loss of consciousness, posttraumatic amnesia, confusion, and many of the other immediate signs and symptoms associated with concussion.

The sudden metabolic demand created by the massive excitatory phenomena triggers an increased utilization of glucose to restore cellular homeostasis. At the same time, cerebral blood flow decreases after concussion, which, in the setting of increased glucose demand, leads to an “energy crisis”: an increased need for adenosine triphosphate with a concomitant decreased delivery of glucose.⁷ This mismatch between energy demand and supply is thought to underlie the most common signs and symptoms of concussion.

ASSESSMENT

History

The history of present illness is essential to a diagnosis of concussion. In the classic scenario, an otherwise asymptomatic person sustains some trauma to the head that is followed immediately by the signs and symptoms of concussion.

Many of these signs and symptoms are nonspecific and may occur without concussion or other trauma.^8,9 Thus, the diagnosis of concussion cannot be made on the basis of symptoms alone, but only in the overall context of history, physical examination, and, at times, additional clinical assessments.

The symptoms of concussion should gradually improve. While they may be exacerbated by certain activities or stimuli, the overall trend should be one of symptom improvement. If symptoms are worsening over time, alternative explanations for the patient’s symptoms should be considered.

Physical examination

A thorough neurologic examination should be conducted in all patients with suspected concussion and include the following.

A mental status examination should include assessment of attention, memory, and recall. Orientation is normal except in the most acute examinations.

Cranial nerve examination must include careful assessment of eye-movement control, including smooth pursuit and saccades. However, even in patients with prominent subjective dizziness, considerable experience may be needed to actually demonstrate abnormalities.

Balance testing. Balance demands careful assessment and, especially for young athletes, this testing should be more difficult than the tandem gait and eyes-closed, feet-together tests.

Standard strength, sensory, reflex, and coordination testing is usually normal.

Any focal neurologic findings should prompt consideration of other causes or of a more serious injury and should lead to further evaluation, including brain imaging.

Diagnostic tests

Current clinical brain imaging cannot diagnose a concussion. The purpose of neuroimaging is to assess for other etiologies or injuries, such as hemorrhage or contusion, that may cause similar symptoms but require different management.

Several guidelines are available to assess the need for imaging in the setting of recent trauma, of which 2 are typically used^10–12:

The Canadian CT Head Rule¹⁰ states that computed tomography (CT) is indicated in any of the following situations:

• The patient fails to reach a Glasgow Coma Scale score of 15—on a scale of 3 (worst) to 15 (best)—within 2 hours
• There is a suspected open skull fracture
• There is any sign of basal skull fracture
• The patient has 2 or more episodes of vomiting
• The patient is 65 or older
• The patient has retrograde amnesia (ie, cannot remember events that occurred before the injury) for 30 minutes or more
• The mechanism of injury was dangerous (eg, a pedestrian was struck by a motor vehicle, or the patient fell from > 3 feet or > 5 stairs).

The New Orleans Criteria¹¹ state that a patient warrants CT of the head if any of the following is present:

• Severe headache
• Vomiting
• Age over 60
• Drug or alcohol intoxication
• Deficit in short-term memory
• Physical evidence of trauma above the clavicles
• Seizure.

Caveats: these imaging guidelines apply to adults; those for pediatric patients differ.¹² Also, because they were designed for use in an emergency department, their utility in clinical practice outside the emergency department is unclear.

Electroencephalography is not necessary in the evaluation of concussion unless a seizure disorder is believed to be the cause of the injury.

Concussion in athletes

Athletes who participate in contact and collision sports are at higher risk of concussion than the nonathletic population. Therefore, specific assessments of symptoms, balance, oculomotor function, cognitive function, and reaction time have been developed for athletes.

Ideally, these measures are taken at preseason baseline, so that they are available for comparison with postinjury assessments after a known or suspected concussion. These assessments can be used to help make the diagnosis of concussion in cases that are unclear and to help monitor recovery. Objective measures of injury are especially useful for athletes who may be reluctant to report symptoms in order to return to play.

Like most medical tests, these assessments need to be properly interpreted in the overall context of the medical history and physical examination by those who know how to administer them. It is important to remember that the natural history of concussion recovery differs between sport-related concussion and concussion that occurs outside of sports.⁸

The symptoms and signs after concussion are so variable and multidimensional that they make a generally applicable treatment hard to define.

Rest: Physical and cognitive

Treatment depends on the specifics of the injury, but there are common recommendations for the acute days after injury. Lacking hard data, the consensus among experts is that patients should undergo a period of physical and cognitive rest.^13,14 Exactly what “rest” means and how long it should last are unknown, leading to a wide variation in its application.

Rest aids recovery but also may have adverse effects: fatigue, diurnal sleep disruption, reactive depression, anxiety, and physiologic deconditioning.^15,16 Many guidelines recommend physical and cognitive rest until symptoms resolve,¹⁴ but this is likely too cautious. Even without a concussion, inactivity is associated with many of the nonspecific symptoms also associated with concussion. As recovery progresses, the somatic symptoms of concussion improve, while emotional symptoms worsen, likely in part due to prolonged rest.¹⁷

We recommend a period of rest lasting 3 to 5 days after injury, followed by a gradual resumption of both physical and cognitive activities as tolerated, remaining below the level at which symptoms are exacerbated.

Not surprisingly, many guidelines for returning to physical activity are focused on athletes. Yet the same principles apply to management of concussion in the general population who exercise: light physical activity (typically walking or stationary bicycling), followed by more vigorous aerobic activity, followed by some resistance activities. Mild aerobic exercise (to below the threshold of symptoms) may speed recovery from refractive postconcussion syndrome, even in those who did not exercise before the injury.¹⁸

Athletes require specific and strict instructions to avoid increased trauma to the head during the gradual increase of physical activities. The National Collegiate Athletic Association has published an algorithm for a gradual return to sport-specific training that is echoed in recent consensus statements on concussion.¹⁹ Once aerobic reconditioning produces no symptoms, then noncontact, sport-specific activities are begun, followed by contact activities. We have patients return to the clinic once they are symptom-free for repeat evaluation before clearing them for high-risk activities (eg, skiing, bicycling) or contact sports (eg, basketball, soccer, football, ice hockey).

Cognitive rest

While physical rest is fairly straightforward, cognitive rest is more challenging. The concept of cognitive rest is hard to define and even harder to enforce. Patients are often told to minimize any activities that require attention or concentration. This often includes, but is not limited to, avoiding reading, texting, playing video games, and using computers.¹³

In the modern world, full avoidance of these activities is difficult and can be profoundly socially isolating. Further, complete cognitive rest may be associated with symptoms of its own.^15,16,20 Still, some reasonable limitation of cognitive activities, at least initially, is likely beneficial.²¹ For patients engaged in school or academic work, often the daily schedule needs to be adjusted and accommodations made to help them return to a full academic schedule and level of activity. It is reasonable to have patients return gradually to work or school rather than attempt to immediately return to their preinjury level.

With these interventions, most patients have full resolution of their symptoms and return to preinjury levels of performance.

TREATING SOMATIC SYMPTOMS

Posttraumatic headache

Posttraumatic headache is the most common sequela of concussion.²² Surprisingly, it is more common after concussion than after moderate or severe traumatic brain injury.²³ A prior history of headache, particularly migraine, is a known risk factor for development of posttraumatic headache.²⁴

Posttraumatic headache is usually further defined by headache type using the International Classification of Headache Disorders criteria (www.ichd-3.org). Migraine or probable migraine is the most common type of posttraumatic headache; tension headache is less common.²⁵

Analgesics such as nonsteroidal anti-inflammatory drugs (NSAIDs) are often used initially by patients to treat posttraumatic headache. One study found that 70% of patients used acetaminophen or an NSAID.²⁶

Treating early with effective therapy is the most important tenet of posttraumatic headache treatment, since 80% of those who self-treat have incomplete relief, and almost all of them are using over-the-counter products.²⁷ Overuse of over-the-counter abortive medications can lead to medication overuse headache, also known as rebound headache, thus complicating the treatment of posttraumatic headache.²⁶

Earlier treatment with a preventive medication can often limit the need for and overuse of over-the-counter analgesics and can minimize the occurrence of subsequent medication overuse headache. However, in pediatric populations, nonpharmacologic interventions such as rest and sleep hygiene are typically used first, then medications after 4 to 6 weeks if this is ineffective.

A number of medications have been studied for prophylactic treatment of posttraumatic headache, including topiramate, amitriptyline, and divalproex sodium,^28–30 but there is little compelling evidence for use of one over the other. If posttraumatic headache is migrainous, beta-blockers, calcium-channel blockers, selective serotonin reuptake inhibitors, serotonin-norepinephrine reuptake inhibtors, and gabapentin are other prophylactic medication options under the appropriate circumstances.^27,31,32 In adults, we have clinically had success with nortriptyline 20 mg or gabapentin 300 mg at night as an initial prophylactic headache medication, increasing as tolerated or until pain is controlled, though there are no high-quality data to guide this decision.

The ideal prophylactic medication depends on headache type, patient tolerance, comorbidities, allergies, and medication sensitivities. Gabapentin, amitriptyline, and nortriptyline can produce sedation, which can help those suffering from sleep disturbance.

If a provider is not comfortable prescribing these medications or doesn’t prescribe them regularly, the patient should be referred to a concussion or headache specialist more familiar with their use.

In some patients, even some athletes, headache may be related to a cervical strain injury—whiplash—that should be treated with an NSAID (or acetaminophen), perhaps with a short course of a muscle relaxant in adults, and with physical therapy.³²

Some patients have chronic headache despite oral medications.²⁶ Therefore, alternatives to oral medications and complementary therapies should be considered. Especially for protracted cases requiring more complicated headache management or injectable treatments, patients should be referred to a pain clinic, headache specialist, or concussion specialist.

Dizziness is also common after concussion. But what the patient means by dizziness requires a little probing. Some have paroxysms of vertigo. This typically represents a peripheral vestibular injury, usually benign paroxysmal positional vertigo. The latter can be elicited with a Hallpike maneuver and treated in the office with the Epley maneuver.³³

Usually, dizziness is a subjective sense of poor coordination, gait instability, or dysequilibrium. Patients may also complain of associated nausea and motion sensitivity. This may all be secondary to a mechanism in the middle or inner ear or the brain. Patients should be encouraged to begin movement—gradually and safely—to help the vestibular system accommodate, which it will do with gradual stimulation. It usually resolves spontaneously.

Specific treatment is unfortunately limited. There is no established benefit from vestibular suppressants such as meclizine. Vestibular rehabilitation may accelerate improvement and decrease symptoms.³³ Referral for a comprehensive balance assessment or to vestibular therapy (a subset of physical therapy) should be considered and is something we typically undertake in our clinic if there is no recovery from dizziness 4 to 6 weeks after the concussion.

Visual symptoms can contribute to dizziness. Convergence spasm or convergence insufficiency (both related to muscle spasm of the eye) can occur after concussion, with some studies estimating that up to 69% of patients have these symptoms.³⁴ This can interfere with visual tracking and contribute to a feeling of dysequilibrium.³⁴ Referral to a concussion specialist or vestibular rehabilitation physical therapist can be helpful in treating this issue if it does not resolve spontaneously.

Orthostasis and lightheadedness also contribute to dizziness and are associated with cerebrovascular autoregulation. Available data suggest that dysregulation of neurovascular coupling, cerebral vasoreactivity, and cerebral autoregulation contribute to some of the chronic symptoms of concussion, including dizziness. A gradual return to exercise may help regulate cerebral blood flow and improve this type of dizziness.³⁵

Sleep disturbance

Sleep disturbance is common after concussion, but the form is variable: insomnia, excessive daytime somnolence, and alteration of the sleep-wake cycle are all seen and may themselves affect recovery.³⁶

Sleep hygiene education should be the first intervention for postconcussive sleep issues. For example, the patient should be encouraged to do the following:

• Minimize “screen time” an hour before going to bed: cell phone, tablet, and computer screens emit a wavelength of light that suppresses endogenous melatonin release^37,38
• Go to bed and wake up at the same time each day
• Minimize or avoid caffeine, nicotine, and alcohol
• Avoid naps.³⁹

Melatonin is a safe and effective treatment that could be added.⁴⁰ In addition, some studies suggest that melatonin may improve recovery from traumatic brain injury.^41,42

Mild exercise (to below the threshold of causing or exacerbating symptoms) may also improve sleep quality.

Amitriptyline or nortriptyline may reduce headache frequency and intensity and also help treat insomnia.

Trazodone is recommended by some as a first-line agent,³⁹ but we usually reserve it for protracted insomnia refractory to the above treatments.

Benzodiazepines should be avoided, as they reduce arousal, impair cognition, and exacerbate motor impairments.⁴³

Emotional symptoms

Acute-onset anxiety or depression often occurs after concussion.^44,45 There is abundant evidence that emotional effects of injury may be the most significant factor in recovery.⁴⁶ A preinjury history of anxiety may be a prognostic factor.⁹ Patients with a history of anxiety or depression are more likely to develop emotional symptoms after a concussion, but emotional problems may develop in any patient after a concussion.^47,48

The circumstances under which an injury is sustained may be traumatic (eg, car accident, assault), leading to an acute stress reaction or disorder and, if untreated, may result in a more chronic condition—posttraumatic stress disorder. Moreover, the injury and subsequent symptoms may have repercussions in many aspects of the patient’s life, leading to further psychologic stress (eg, loss of wages or the inability to handle normal work, school, and family responsibilities).

Referral to a therapist trained in skills-based psychotherapy (eg, cognitive-behavioral therapy, exposure-based treatment) is often helpful.

Pharmacologic treatment can be a useful adjunct. Several studies have shown that selective serotonin reuptake inhibitors, serotonin-norepinephrine reuptake inhibitors, and tricyclic antidepressants may improve depression after concussion.⁴⁹ The prescription of antidepressants, however, is best left to providers with experience in treating anxiety and depression.

As with sleep disorders after concussion, benzodiazepines should be avoided, as they can impair cognition.⁴³

Cognitive problems

Cognitive problems are also common after concussion. Patients complain about everyday experiences of forgetfulness, distractibility, loss of concentration, and mental fatigue. Although patients often subjectively perceive these symptoms as quite limiting, the impairments can be difficult to demonstrate in office testing.

A program of gradual increase in mental activity, parallel to recovery of physical capacity, should be undertaken. Most patients make a gradual recovery within a few weeks.⁵⁰

When cognitive symptoms cause significant school or vocational problems or become persistent, patients should be referred to a specialty clinic. As with most of the consequences of concussion, there are few established treatments. When cognitive difficulties persist, it is important to consider the complications of concussion mentioned above: headache, pain, sleep disturbance, and anxiety, all of which may cause subjective cognitive problems and are treatable.

If cognitive symptoms are prolonged despite improvement of other issues like headache and sleep disturbance, a low-dose stimulant medication such as amphetamine salts or methylphenidate may be useful for symptoms of poor attention.⁴⁹ They should be only a temporary measure after concussion to carry the patient through a cognitively challenging period, unless there was a history of attention-deficit disorder before the injury. A variety of other agents, including amantadine,⁵¹ have been proposed based on limited studies; all are off-label uses. Before considering these types of interventions, referral to a specialist or a specialty program would be appropriate.

With the interventions suggested above, most patients with concussion have a resolution of symptoms and can return to preinjury levels of performance. But some have prolonged symptoms and sequelae. Approximately 10% of athletes have persistent signs and symptoms of concussion beyond 2 weeks. If concussion is not sport-related, most patients recover completely within the first 3 months, but up to 33% may have symptoms beyond that.⁵²

Four types of patients have persistent symptoms:

Patients who sustained a high-force mechanism of injury. These patients simply need more time and accommodation.

Patients who sustained multiple concussions. These patients may also need more time and accommodation.

Patients with an underlying neurologic condition, recognized prior to injury or not, may have delayed or incomplete recovery. Even aging may be an “underlying condition” in concussion.

Patients whose symptoms from an apparently single mild concussion do not resolve despite appropriate treatments may have identifiable factors, but intractable pain (usually headache) or significant emotional disturbance or both are common. Once established and persistent, this is difficult to treat. Referral to a specialty practice is appropriate, but even in that setting effective treatment may be elusive.

PATIENT EDUCATION

Most important for patient education is reassurance. Ultimately, concussion is a self-limited phenomenon, and reinforcing this is helpful for patients. If concussion is not sport-related, most patients recover completely within 3 months.

The next important tenet in patient education is that they should rest for 3 to 5 days, then resume gradual physical and cognitive activities. If resuming activities too soon results in symptoms, then they should rest for a day and gradually resume activity. If their recovery is prolonged (ie, longer than 6 weeks), they likely need to be referred to a concussion specialist.

Concussion, also known as mild traumatic brain injury, affects more than 600 adults per 100,000 each year and is commonly treated by nonneurologists.¹ Public attention to concussion has been increasing, particularly to concussion sustained during sports. Coincident with this increased attention, the diagnosis of concussion continues to increase in the outpatient setting. Thus, a review of the topic is timely.

ACCELERATION OF THE BRAIN DUE TO TRAUMA

The definition of concussion has changed considerably over the years. It is currently defined as a pathophysiologic process that results from an acceleration or deceleration of the brain induced by trauma.² It is largely a temporary, functional problem, as opposed to a gross structural injury.^2–5

The acceleration of the brain that results in a concussion is usually initiated by a direct blow to the head, although direct impact is not required.⁶ As the brain rotates, different areas accelerate at different rates, resulting in a shear strain imparted to the parenchyma.

This shear strain causes deformation of axonal membranes and opening of membrane-associated sodium-potassium channels. This in turn leads to release of excitatory neurotransmitters, ultimately culminating in a wave of neuronal depolarization and a spreading depression-like phenomenon that may mediate the loss of consciousness, posttraumatic amnesia, confusion, and many of the other immediate signs and symptoms associated with concussion.

The sudden metabolic demand created by the massive excitatory phenomena triggers an increased utilization of glucose to restore cellular homeostasis. At the same time, cerebral blood flow decreases after concussion, which, in the setting of increased glucose demand, leads to an “energy crisis”: an increased need for adenosine triphosphate with a concomitant decreased delivery of glucose.⁷ This mismatch between energy demand and supply is thought to underlie the most common signs and symptoms of concussion.

ASSESSMENT

History

The history of present illness is essential to a diagnosis of concussion. In the classic scenario, an otherwise asymptomatic person sustains some trauma to the head that is followed immediately by the signs and symptoms of concussion.

Many of these signs and symptoms are nonspecific and may occur without concussion or other trauma.^8,9 Thus, the diagnosis of concussion cannot be made on the basis of symptoms alone, but only in the overall context of history, physical examination, and, at times, additional clinical assessments.

The symptoms of concussion should gradually improve. While they may be exacerbated by certain activities or stimuli, the overall trend should be one of symptom improvement. If symptoms are worsening over time, alternative explanations for the patient’s symptoms should be considered.

Physical examination

A thorough neurologic examination should be conducted in all patients with suspected concussion and include the following.

A mental status examination should include assessment of attention, memory, and recall. Orientation is normal except in the most acute examinations.

Cranial nerve examination must include careful assessment of eye-movement control, including smooth pursuit and saccades. However, even in patients with prominent subjective dizziness, considerable experience may be needed to actually demonstrate abnormalities.

Balance testing. Balance demands careful assessment and, especially for young athletes, this testing should be more difficult than the tandem gait and eyes-closed, feet-together tests.

Standard strength, sensory, reflex, and coordination testing is usually normal.

Any focal neurologic findings should prompt consideration of other causes or of a more serious injury and should lead to further evaluation, including brain imaging.

Diagnostic tests

Current clinical brain imaging cannot diagnose a concussion. The purpose of neuroimaging is to assess for other etiologies or injuries, such as hemorrhage or contusion, that may cause similar symptoms but require different management.

Several guidelines are available to assess the need for imaging in the setting of recent trauma, of which 2 are typically used^10–12:

The Canadian CT Head Rule¹⁰ states that computed tomography (CT) is indicated in any of the following situations:

• The patient fails to reach a Glasgow Coma Scale score of 15—on a scale of 3 (worst) to 15 (best)—within 2 hours
• There is a suspected open skull fracture
• There is any sign of basal skull fracture
• The patient has 2 or more episodes of vomiting
• The patient is 65 or older
• The patient has retrograde amnesia (ie, cannot remember events that occurred before the injury) for 30 minutes or more
• The mechanism of injury was dangerous (eg, a pedestrian was struck by a motor vehicle, or the patient fell from > 3 feet or > 5 stairs).

The New Orleans Criteria¹¹ state that a patient warrants CT of the head if any of the following is present:

• Severe headache
• Vomiting
• Age over 60
• Drug or alcohol intoxication
• Deficit in short-term memory
• Physical evidence of trauma above the clavicles
• Seizure.

Caveats: these imaging guidelines apply to adults; those for pediatric patients differ.¹² Also, because they were designed for use in an emergency department, their utility in clinical practice outside the emergency department is unclear.

Electroencephalography is not necessary in the evaluation of concussion unless a seizure disorder is believed to be the cause of the injury.

Concussion in athletes

Athletes who participate in contact and collision sports are at higher risk of concussion than the nonathletic population. Therefore, specific assessments of symptoms, balance, oculomotor function, cognitive function, and reaction time have been developed for athletes.

Ideally, these measures are taken at preseason baseline, so that they are available for comparison with postinjury assessments after a known or suspected concussion. These assessments can be used to help make the diagnosis of concussion in cases that are unclear and to help monitor recovery. Objective measures of injury are especially useful for athletes who may be reluctant to report symptoms in order to return to play.

Like most medical tests, these assessments need to be properly interpreted in the overall context of the medical history and physical examination by those who know how to administer them. It is important to remember that the natural history of concussion recovery differs between sport-related concussion and concussion that occurs outside of sports.⁸

The symptoms and signs after concussion are so variable and multidimensional that they make a generally applicable treatment hard to define.

Rest: Physical and cognitive

Treatment depends on the specifics of the injury, but there are common recommendations for the acute days after injury. Lacking hard data, the consensus among experts is that patients should undergo a period of physical and cognitive rest.^13,14 Exactly what “rest” means and how long it should last are unknown, leading to a wide variation in its application.

Rest aids recovery but also may have adverse effects: fatigue, diurnal sleep disruption, reactive depression, anxiety, and physiologic deconditioning.^15,16 Many guidelines recommend physical and cognitive rest until symptoms resolve,¹⁴ but this is likely too cautious. Even without a concussion, inactivity is associated with many of the nonspecific symptoms also associated with concussion. As recovery progresses, the somatic symptoms of concussion improve, while emotional symptoms worsen, likely in part due to prolonged rest.¹⁷

We recommend a period of rest lasting 3 to 5 days after injury, followed by a gradual resumption of both physical and cognitive activities as tolerated, remaining below the level at which symptoms are exacerbated.

Not surprisingly, many guidelines for returning to physical activity are focused on athletes. Yet the same principles apply to management of concussion in the general population who exercise: light physical activity (typically walking or stationary bicycling), followed by more vigorous aerobic activity, followed by some resistance activities. Mild aerobic exercise (to below the threshold of symptoms) may speed recovery from refractive postconcussion syndrome, even in those who did not exercise before the injury.¹⁸

Athletes require specific and strict instructions to avoid increased trauma to the head during the gradual increase of physical activities. The National Collegiate Athletic Association has published an algorithm for a gradual return to sport-specific training that is echoed in recent consensus statements on concussion.¹⁹ Once aerobic reconditioning produces no symptoms, then noncontact, sport-specific activities are begun, followed by contact activities. We have patients return to the clinic once they are symptom-free for repeat evaluation before clearing them for high-risk activities (eg, skiing, bicycling) or contact sports (eg, basketball, soccer, football, ice hockey).

Cognitive rest

While physical rest is fairly straightforward, cognitive rest is more challenging. The concept of cognitive rest is hard to define and even harder to enforce. Patients are often told to minimize any activities that require attention or concentration. This often includes, but is not limited to, avoiding reading, texting, playing video games, and using computers.¹³

In the modern world, full avoidance of these activities is difficult and can be profoundly socially isolating. Further, complete cognitive rest may be associated with symptoms of its own.^15,16,20 Still, some reasonable limitation of cognitive activities, at least initially, is likely beneficial.²¹ For patients engaged in school or academic work, often the daily schedule needs to be adjusted and accommodations made to help them return to a full academic schedule and level of activity. It is reasonable to have patients return gradually to work or school rather than attempt to immediately return to their preinjury level.

With these interventions, most patients have full resolution of their symptoms and return to preinjury levels of performance.

TREATING SOMATIC SYMPTOMS

Posttraumatic headache

Posttraumatic headache is the most common sequela of concussion.²² Surprisingly, it is more common after concussion than after moderate or severe traumatic brain injury.²³ A prior history of headache, particularly migraine, is a known risk factor for development of posttraumatic headache.²⁴

Posttraumatic headache is usually further defined by headache type using the International Classification of Headache Disorders criteria (www.ichd-3.org). Migraine or probable migraine is the most common type of posttraumatic headache; tension headache is less common.²⁵

Analgesics such as nonsteroidal anti-inflammatory drugs (NSAIDs) are often used initially by patients to treat posttraumatic headache. One study found that 70% of patients used acetaminophen or an NSAID.²⁶

Treating early with effective therapy is the most important tenet of posttraumatic headache treatment, since 80% of those who self-treat have incomplete relief, and almost all of them are using over-the-counter products.²⁷ Overuse of over-the-counter abortive medications can lead to medication overuse headache, also known as rebound headache, thus complicating the treatment of posttraumatic headache.²⁶

Earlier treatment with a preventive medication can often limit the need for and overuse of over-the-counter analgesics and can minimize the occurrence of subsequent medication overuse headache. However, in pediatric populations, nonpharmacologic interventions such as rest and sleep hygiene are typically used first, then medications after 4 to 6 weeks if this is ineffective.

A number of medications have been studied for prophylactic treatment of posttraumatic headache, including topiramate, amitriptyline, and divalproex sodium,^28–30 but there is little compelling evidence for use of one over the other. If posttraumatic headache is migrainous, beta-blockers, calcium-channel blockers, selective serotonin reuptake inhibitors, serotonin-norepinephrine reuptake inhibtors, and gabapentin are other prophylactic medication options under the appropriate circumstances.^27,31,32 In adults, we have clinically had success with nortriptyline 20 mg or gabapentin 300 mg at night as an initial prophylactic headache medication, increasing as tolerated or until pain is controlled, though there are no high-quality data to guide this decision.

The ideal prophylactic medication depends on headache type, patient tolerance, comorbidities, allergies, and medication sensitivities. Gabapentin, amitriptyline, and nortriptyline can produce sedation, which can help those suffering from sleep disturbance.

If a provider is not comfortable prescribing these medications or doesn’t prescribe them regularly, the patient should be referred to a concussion or headache specialist more familiar with their use.

In some patients, even some athletes, headache may be related to a cervical strain injury—whiplash—that should be treated with an NSAID (or acetaminophen), perhaps with a short course of a muscle relaxant in adults, and with physical therapy.³²

Some patients have chronic headache despite oral medications.²⁶ Therefore, alternatives to oral medications and complementary therapies should be considered. Especially for protracted cases requiring more complicated headache management or injectable treatments, patients should be referred to a pain clinic, headache specialist, or concussion specialist.

Dizziness is also common after concussion. But what the patient means by dizziness requires a little probing. Some have paroxysms of vertigo. This typically represents a peripheral vestibular injury, usually benign paroxysmal positional vertigo. The latter can be elicited with a Hallpike maneuver and treated in the office with the Epley maneuver.³³

Usually, dizziness is a subjective sense of poor coordination, gait instability, or dysequilibrium. Patients may also complain of associated nausea and motion sensitivity. This may all be secondary to a mechanism in the middle or inner ear or the brain. Patients should be encouraged to begin movement—gradually and safely—to help the vestibular system accommodate, which it will do with gradual stimulation. It usually resolves spontaneously.

Specific treatment is unfortunately limited. There is no established benefit from vestibular suppressants such as meclizine. Vestibular rehabilitation may accelerate improvement and decrease symptoms.³³ Referral for a comprehensive balance assessment or to vestibular therapy (a subset of physical therapy) should be considered and is something we typically undertake in our clinic if there is no recovery from dizziness 4 to 6 weeks after the concussion.

Visual symptoms can contribute to dizziness. Convergence spasm or convergence insufficiency (both related to muscle spasm of the eye) can occur after concussion, with some studies estimating that up to 69% of patients have these symptoms.³⁴ This can interfere with visual tracking and contribute to a feeling of dysequilibrium.³⁴ Referral to a concussion specialist or vestibular rehabilitation physical therapist can be helpful in treating this issue if it does not resolve spontaneously.

Orthostasis and lightheadedness also contribute to dizziness and are associated with cerebrovascular autoregulation. Available data suggest that dysregulation of neurovascular coupling, cerebral vasoreactivity, and cerebral autoregulation contribute to some of the chronic symptoms of concussion, including dizziness. A gradual return to exercise may help regulate cerebral blood flow and improve this type of dizziness.³⁵

Sleep disturbance

Sleep disturbance is common after concussion, but the form is variable: insomnia, excessive daytime somnolence, and alteration of the sleep-wake cycle are all seen and may themselves affect recovery.³⁶

Sleep hygiene education should be the first intervention for postconcussive sleep issues. For example, the patient should be encouraged to do the following:

• Minimize “screen time” an hour before going to bed: cell phone, tablet, and computer screens emit a wavelength of light that suppresses endogenous melatonin release^37,38
• Go to bed and wake up at the same time each day
• Minimize or avoid caffeine, nicotine, and alcohol
• Avoid naps.³⁹

Melatonin is a safe and effective treatment that could be added.⁴⁰ In addition, some studies suggest that melatonin may improve recovery from traumatic brain injury.^41,42

Mild exercise (to below the threshold of causing or exacerbating symptoms) may also improve sleep quality.

Amitriptyline or nortriptyline may reduce headache frequency and intensity and also help treat insomnia.

Trazodone is recommended by some as a first-line agent,³⁹ but we usually reserve it for protracted insomnia refractory to the above treatments.

Benzodiazepines should be avoided, as they reduce arousal, impair cognition, and exacerbate motor impairments.⁴³

Emotional symptoms

Acute-onset anxiety or depression often occurs after concussion.^44,45 There is abundant evidence that emotional effects of injury may be the most significant factor in recovery.⁴⁶ A preinjury history of anxiety may be a prognostic factor.⁹ Patients with a history of anxiety or depression are more likely to develop emotional symptoms after a concussion, but emotional problems may develop in any patient after a concussion.^47,48

The circumstances under which an injury is sustained may be traumatic (eg, car accident, assault), leading to an acute stress reaction or disorder and, if untreated, may result in a more chronic condition—posttraumatic stress disorder. Moreover, the injury and subsequent symptoms may have repercussions in many aspects of the patient’s life, leading to further psychologic stress (eg, loss of wages or the inability to handle normal work, school, and family responsibilities).

Referral to a therapist trained in skills-based psychotherapy (eg, cognitive-behavioral therapy, exposure-based treatment) is often helpful.

Pharmacologic treatment can be a useful adjunct. Several studies have shown that selective serotonin reuptake inhibitors, serotonin-norepinephrine reuptake inhibitors, and tricyclic antidepressants may improve depression after concussion.⁴⁹ The prescription of antidepressants, however, is best left to providers with experience in treating anxiety and depression.

As with sleep disorders after concussion, benzodiazepines should be avoided, as they can impair cognition.⁴³

Cognitive problems

Cognitive problems are also common after concussion. Patients complain about everyday experiences of forgetfulness, distractibility, loss of concentration, and mental fatigue. Although patients often subjectively perceive these symptoms as quite limiting, the impairments can be difficult to demonstrate in office testing.

A program of gradual increase in mental activity, parallel to recovery of physical capacity, should be undertaken. Most patients make a gradual recovery within a few weeks.⁵⁰

When cognitive symptoms cause significant school or vocational problems or become persistent, patients should be referred to a specialty clinic. As with most of the consequences of concussion, there are few established treatments. When cognitive difficulties persist, it is important to consider the complications of concussion mentioned above: headache, pain, sleep disturbance, and anxiety, all of which may cause subjective cognitive problems and are treatable.

If cognitive symptoms are prolonged despite improvement of other issues like headache and sleep disturbance, a low-dose stimulant medication such as amphetamine salts or methylphenidate may be useful for symptoms of poor attention.⁴⁹ They should be only a temporary measure after concussion to carry the patient through a cognitively challenging period, unless there was a history of attention-deficit disorder before the injury. A variety of other agents, including amantadine,⁵¹ have been proposed based on limited studies; all are off-label uses. Before considering these types of interventions, referral to a specialist or a specialty program would be appropriate.

With the interventions suggested above, most patients with concussion have a resolution of symptoms and can return to preinjury levels of performance. But some have prolonged symptoms and sequelae. Approximately 10% of athletes have persistent signs and symptoms of concussion beyond 2 weeks. If concussion is not sport-related, most patients recover completely within the first 3 months, but up to 33% may have symptoms beyond that.⁵²

Four types of patients have persistent symptoms:

Patients who sustained a high-force mechanism of injury. These patients simply need more time and accommodation.

Patients who sustained multiple concussions. These patients may also need more time and accommodation.

Patients with an underlying neurologic condition, recognized prior to injury or not, may have delayed or incomplete recovery. Even aging may be an “underlying condition” in concussion.

Patients whose symptoms from an apparently single mild concussion do not resolve despite appropriate treatments may have identifiable factors, but intractable pain (usually headache) or significant emotional disturbance or both are common. Once established and persistent, this is difficult to treat. Referral to a specialty practice is appropriate, but even in that setting effective treatment may be elusive.

PATIENT EDUCATION

Most important for patient education is reassurance. Ultimately, concussion is a self-limited phenomenon, and reinforcing this is helpful for patients. If concussion is not sport-related, most patients recover completely within 3 months.

The next important tenet in patient education is that they should rest for 3 to 5 days, then resume gradual physical and cognitive activities. If resuming activities too soon results in symptoms, then they should rest for a day and gradually resume activity. If their recovery is prolonged (ie, longer than 6 weeks), they likely need to be referred to a concussion specialist.

Concussion, also known as mild traumatic brain injury, affects more than 600 adults per 100,000 each year and is commonly treated by nonneurologists.¹ Public attention to concussion has been increasing, particularly to concussion sustained during sports. Coincident with this increased attention, the diagnosis of concussion continues to increase in the outpatient setting. Thus, a review of the topic is timely.

ACCELERATION OF THE BRAIN DUE TO TRAUMA

The definition of concussion has changed considerably over the years. It is currently defined as a pathophysiologic process that results from an acceleration or deceleration of the brain induced by trauma.² It is largely a temporary, functional problem, as opposed to a gross structural injury.^2–5

The acceleration of the brain that results in a concussion is usually initiated by a direct blow to the head, although direct impact is not required.⁶ As the brain rotates, different areas accelerate at different rates, resulting in a shear strain imparted to the parenchyma.

This shear strain causes deformation of axonal membranes and opening of membrane-associated sodium-potassium channels. This in turn leads to release of excitatory neurotransmitters, ultimately culminating in a wave of neuronal depolarization and a spreading depression-like phenomenon that may mediate the loss of consciousness, posttraumatic amnesia, confusion, and many of the other immediate signs and symptoms associated with concussion.

The sudden metabolic demand created by the massive excitatory phenomena triggers an increased utilization of glucose to restore cellular homeostasis. At the same time, cerebral blood flow decreases after concussion, which, in the setting of increased glucose demand, leads to an “energy crisis”: an increased need for adenosine triphosphate with a concomitant decreased delivery of glucose.⁷ This mismatch between energy demand and supply is thought to underlie the most common signs and symptoms of concussion.

ASSESSMENT

History

The history of present illness is essential to a diagnosis of concussion. In the classic scenario, an otherwise asymptomatic person sustains some trauma to the head that is followed immediately by the signs and symptoms of concussion.

Many of these signs and symptoms are nonspecific and may occur without concussion or other trauma.^8,9 Thus, the diagnosis of concussion cannot be made on the basis of symptoms alone, but only in the overall context of history, physical examination, and, at times, additional clinical assessments.

The symptoms of concussion should gradually improve. While they may be exacerbated by certain activities or stimuli, the overall trend should be one of symptom improvement. If symptoms are worsening over time, alternative explanations for the patient’s symptoms should be considered.

Physical examination

A thorough neurologic examination should be conducted in all patients with suspected concussion and include the following.

A mental status examination should include assessment of attention, memory, and recall. Orientation is normal except in the most acute examinations.

Cranial nerve examination must include careful assessment of eye-movement control, including smooth pursuit and saccades. However, even in patients with prominent subjective dizziness, considerable experience may be needed to actually demonstrate abnormalities.

Balance testing. Balance demands careful assessment and, especially for young athletes, this testing should be more difficult than the tandem gait and eyes-closed, feet-together tests.

Standard strength, sensory, reflex, and coordination testing is usually normal.

Any focal neurologic findings should prompt consideration of other causes or of a more serious injury and should lead to further evaluation, including brain imaging.

Diagnostic tests

Current clinical brain imaging cannot diagnose a concussion. The purpose of neuroimaging is to assess for other etiologies or injuries, such as hemorrhage or contusion, that may cause similar symptoms but require different management.

Several guidelines are available to assess the need for imaging in the setting of recent trauma, of which 2 are typically used^10–12:

The Canadian CT Head Rule¹⁰ states that computed tomography (CT) is indicated in any of the following situations:

• The patient fails to reach a Glasgow Coma Scale score of 15—on a scale of 3 (worst) to 15 (best)—within 2 hours
• There is a suspected open skull fracture
• There is any sign of basal skull fracture
• The patient has 2 or more episodes of vomiting
• The patient is 65 or older
• The patient has retrograde amnesia (ie, cannot remember events that occurred before the injury) for 30 minutes or more
• The mechanism of injury was dangerous (eg, a pedestrian was struck by a motor vehicle, or the patient fell from > 3 feet or > 5 stairs).

The New Orleans Criteria¹¹ state that a patient warrants CT of the head if any of the following is present:

• Severe headache
• Vomiting
• Age over 60
• Drug or alcohol intoxication
• Deficit in short-term memory
• Physical evidence of trauma above the clavicles
• Seizure.

Caveats: these imaging guidelines apply to adults; those for pediatric patients differ.¹² Also, because they were designed for use in an emergency department, their utility in clinical practice outside the emergency department is unclear.

Electroencephalography is not necessary in the evaluation of concussion unless a seizure disorder is believed to be the cause of the injury.

Concussion in athletes

Athletes who participate in contact and collision sports are at higher risk of concussion than the nonathletic population. Therefore, specific assessments of symptoms, balance, oculomotor function, cognitive function, and reaction time have been developed for athletes.

Ideally, these measures are taken at preseason baseline, so that they are available for comparison with postinjury assessments after a known or suspected concussion. These assessments can be used to help make the diagnosis of concussion in cases that are unclear and to help monitor recovery. Objective measures of injury are especially useful for athletes who may be reluctant to report symptoms in order to return to play.

Like most medical tests, these assessments need to be properly interpreted in the overall context of the medical history and physical examination by those who know how to administer them. It is important to remember that the natural history of concussion recovery differs between sport-related concussion and concussion that occurs outside of sports.⁸

The symptoms and signs after concussion are so variable and multidimensional that they make a generally applicable treatment hard to define.

Rest: Physical and cognitive

Treatment depends on the specifics of the injury, but there are common recommendations for the acute days after injury. Lacking hard data, the consensus among experts is that patients should undergo a period of physical and cognitive rest.^13,14 Exactly what “rest” means and how long it should last are unknown, leading to a wide variation in its application.

Rest aids recovery but also may have adverse effects: fatigue, diurnal sleep disruption, reactive depression, anxiety, and physiologic deconditioning.^15,16 Many guidelines recommend physical and cognitive rest until symptoms resolve,¹⁴ but this is likely too cautious. Even without a concussion, inactivity is associated with many of the nonspecific symptoms also associated with concussion. As recovery progresses, the somatic symptoms of concussion improve, while emotional symptoms worsen, likely in part due to prolonged rest.¹⁷

We recommend a period of rest lasting 3 to 5 days after injury, followed by a gradual resumption of both physical and cognitive activities as tolerated, remaining below the level at which symptoms are exacerbated.

Not surprisingly, many guidelines for returning to physical activity are focused on athletes. Yet the same principles apply to management of concussion in the general population who exercise: light physical activity (typically walking or stationary bicycling), followed by more vigorous aerobic activity, followed by some resistance activities. Mild aerobic exercise (to below the threshold of symptoms) may speed recovery from refractive postconcussion syndrome, even in those who did not exercise before the injury.¹⁸

Athletes require specific and strict instructions to avoid increased trauma to the head during the gradual increase of physical activities. The National Collegiate Athletic Association has published an algorithm for a gradual return to sport-specific training that is echoed in recent consensus statements on concussion.¹⁹ Once aerobic reconditioning produces no symptoms, then noncontact, sport-specific activities are begun, followed by contact activities. We have patients return to the clinic once they are symptom-free for repeat evaluation before clearing them for high-risk activities (eg, skiing, bicycling) or contact sports (eg, basketball, soccer, football, ice hockey).

Cognitive rest

While physical rest is fairly straightforward, cognitive rest is more challenging. The concept of cognitive rest is hard to define and even harder to enforce. Patients are often told to minimize any activities that require attention or concentration. This often includes, but is not limited to, avoiding reading, texting, playing video games, and using computers.¹³

In the modern world, full avoidance of these activities is difficult and can be profoundly socially isolating. Further, complete cognitive rest may be associated with symptoms of its own.^15,16,20 Still, some reasonable limitation of cognitive activities, at least initially, is likely beneficial.²¹ For patients engaged in school or academic work, often the daily schedule needs to be adjusted and accommodations made to help them return to a full academic schedule and level of activity. It is reasonable to have patients return gradually to work or school rather than attempt to immediately return to their preinjury level.

With these interventions, most patients have full resolution of their symptoms and return to preinjury levels of performance.

TREATING SOMATIC SYMPTOMS

Posttraumatic headache

Posttraumatic headache is the most common sequela of concussion.²² Surprisingly, it is more common after concussion than after moderate or severe traumatic brain injury.²³ A prior history of headache, particularly migraine, is a known risk factor for development of posttraumatic headache.²⁴

Posttraumatic headache is usually further defined by headache type using the International Classification of Headache Disorders criteria (www.ichd-3.org). Migraine or probable migraine is the most common type of posttraumatic headache; tension headache is less common.²⁵

Analgesics such as nonsteroidal anti-inflammatory drugs (NSAIDs) are often used initially by patients to treat posttraumatic headache. One study found that 70% of patients used acetaminophen or an NSAID.²⁶

Treating early with effective therapy is the most important tenet of posttraumatic headache treatment, since 80% of those who self-treat have incomplete relief, and almost all of them are using over-the-counter products.²⁷ Overuse of over-the-counter abortive medications can lead to medication overuse headache, also known as rebound headache, thus complicating the treatment of posttraumatic headache.²⁶

Earlier treatment with a preventive medication can often limit the need for and overuse of over-the-counter analgesics and can minimize the occurrence of subsequent medication overuse headache. However, in pediatric populations, nonpharmacologic interventions such as rest and sleep hygiene are typically used first, then medications after 4 to 6 weeks if this is ineffective.

A number of medications have been studied for prophylactic treatment of posttraumatic headache, including topiramate, amitriptyline, and divalproex sodium,^28–30 but there is little compelling evidence for use of one over the other. If posttraumatic headache is migrainous, beta-blockers, calcium-channel blockers, selective serotonin reuptake inhibitors, serotonin-norepinephrine reuptake inhibtors, and gabapentin are other prophylactic medication options under the appropriate circumstances.^27,31,32 In adults, we have clinically had success with nortriptyline 20 mg or gabapentin 300 mg at night as an initial prophylactic headache medication, increasing as tolerated or until pain is controlled, though there are no high-quality data to guide this decision.

The ideal prophylactic medication depends on headache type, patient tolerance, comorbidities, allergies, and medication sensitivities. Gabapentin, amitriptyline, and nortriptyline can produce sedation, which can help those suffering from sleep disturbance.

If a provider is not comfortable prescribing these medications or doesn’t prescribe them regularly, the patient should be referred to a concussion or headache specialist more familiar with their use.

In some patients, even some athletes, headache may be related to a cervical strain injury—whiplash—that should be treated with an NSAID (or acetaminophen), perhaps with a short course of a muscle relaxant in adults, and with physical therapy.³²

Some patients have chronic headache despite oral medications.²⁶ Therefore, alternatives to oral medications and complementary therapies should be considered. Especially for protracted cases requiring more complicated headache management or injectable treatments, patients should be referred to a pain clinic, headache specialist, or concussion specialist.

Dizziness is also common after concussion. But what the patient means by dizziness requires a little probing. Some have paroxysms of vertigo. This typically represents a peripheral vestibular injury, usually benign paroxysmal positional vertigo. The latter can be elicited with a Hallpike maneuver and treated in the office with the Epley maneuver.³³

Usually, dizziness is a subjective sense of poor coordination, gait instability, or dysequilibrium. Patients may also complain of associated nausea and motion sensitivity. This may all be secondary to a mechanism in the middle or inner ear or the brain. Patients should be encouraged to begin movement—gradually and safely—to help the vestibular system accommodate, which it will do with gradual stimulation. It usually resolves spontaneously.

Specific treatment is unfortunately limited. There is no established benefit from vestibular suppressants such as meclizine. Vestibular rehabilitation may accelerate improvement and decrease symptoms.³³ Referral for a comprehensive balance assessment or to vestibular therapy (a subset of physical therapy) should be considered and is something we typically undertake in our clinic if there is no recovery from dizziness 4 to 6 weeks after the concussion.

Visual symptoms can contribute to dizziness. Convergence spasm or convergence insufficiency (both related to muscle spasm of the eye) can occur after concussion, with some studies estimating that up to 69% of patients have these symptoms.³⁴ This can interfere with visual tracking and contribute to a feeling of dysequilibrium.³⁴ Referral to a concussion specialist or vestibular rehabilitation physical therapist can be helpful in treating this issue if it does not resolve spontaneously.

Orthostasis and lightheadedness also contribute to dizziness and are associated with cerebrovascular autoregulation. Available data suggest that dysregulation of neurovascular coupling, cerebral vasoreactivity, and cerebral autoregulation contribute to some of the chronic symptoms of concussion, including dizziness. A gradual return to exercise may help regulate cerebral blood flow and improve this type of dizziness.³⁵

Sleep disturbance

Sleep disturbance is common after concussion, but the form is variable: insomnia, excessive daytime somnolence, and alteration of the sleep-wake cycle are all seen and may themselves affect recovery.³⁶

Sleep hygiene education should be the first intervention for postconcussive sleep issues. For example, the patient should be encouraged to do the following:

• Minimize “screen time” an hour before going to bed: cell phone, tablet, and computer screens emit a wavelength of light that suppresses endogenous melatonin release^37,38
• Go to bed and wake up at the same time each day
• Minimize or avoid caffeine, nicotine, and alcohol
• Avoid naps.³⁹

Melatonin is a safe and effective treatment that could be added.⁴⁰ In addition, some studies suggest that melatonin may improve recovery from traumatic brain injury.^41,42

Mild exercise (to below the threshold of causing or exacerbating symptoms) may also improve sleep quality.

Amitriptyline or nortriptyline may reduce headache frequency and intensity and also help treat insomnia.

Trazodone is recommended by some as a first-line agent,³⁹ but we usually reserve it for protracted insomnia refractory to the above treatments.

Benzodiazepines should be avoided, as they reduce arousal, impair cognition, and exacerbate motor impairments.⁴³

Emotional symptoms

Acute-onset anxiety or depression often occurs after concussion.^44,45 There is abundant evidence that emotional effects of injury may be the most significant factor in recovery.⁴⁶ A preinjury history of anxiety may be a prognostic factor.⁹ Patients with a history of anxiety or depression are more likely to develop emotional symptoms after a concussion, but emotional problems may develop in any patient after a concussion.^47,48

The circumstances under which an injury is sustained may be traumatic (eg, car accident, assault), leading to an acute stress reaction or disorder and, if untreated, may result in a more chronic condition—posttraumatic stress disorder. Moreover, the injury and subsequent symptoms may have repercussions in many aspects of the patient’s life, leading to further psychologic stress (eg, loss of wages or the inability to handle normal work, school, and family responsibilities).

Referral to a therapist trained in skills-based psychotherapy (eg, cognitive-behavioral therapy, exposure-based treatment) is often helpful.

Pharmacologic treatment can be a useful adjunct. Several studies have shown that selective serotonin reuptake inhibitors, serotonin-norepinephrine reuptake inhibitors, and tricyclic antidepressants may improve depression after concussion.⁴⁹ The prescription of antidepressants, however, is best left to providers with experience in treating anxiety and depression.

As with sleep disorders after concussion, benzodiazepines should be avoided, as they can impair cognition.⁴³

Cognitive problems

Cognitive problems are also common after concussion. Patients complain about everyday experiences of forgetfulness, distractibility, loss of concentration, and mental fatigue. Although patients often subjectively perceive these symptoms as quite limiting, the impairments can be difficult to demonstrate in office testing.

A program of gradual increase in mental activity, parallel to recovery of physical capacity, should be undertaken. Most patients make a gradual recovery within a few weeks.⁵⁰

When cognitive symptoms cause significant school or vocational problems or become persistent, patients should be referred to a specialty clinic. As with most of the consequences of concussion, there are few established treatments. When cognitive difficulties persist, it is important to consider the complications of concussion mentioned above: headache, pain, sleep disturbance, and anxiety, all of which may cause subjective cognitive problems and are treatable.

If cognitive symptoms are prolonged despite improvement of other issues like headache and sleep disturbance, a low-dose stimulant medication such as amphetamine salts or methylphenidate may be useful for symptoms of poor attention.⁴⁹ They should be only a temporary measure after concussion to carry the patient through a cognitively challenging period, unless there was a history of attention-deficit disorder before the injury. A variety of other agents, including amantadine,⁵¹ have been proposed based on limited studies; all are off-label uses. Before considering these types of interventions, referral to a specialist or a specialty program would be appropriate.

With the interventions suggested above, most patients with concussion have a resolution of symptoms and can return to preinjury levels of performance. But some have prolonged symptoms and sequelae. Approximately 10% of athletes have persistent signs and symptoms of concussion beyond 2 weeks. If concussion is not sport-related, most patients recover completely within the first 3 months, but up to 33% may have symptoms beyond that.⁵²

Four types of patients have persistent symptoms:

Patients who sustained a high-force mechanism of injury. These patients simply need more time and accommodation.

Patients who sustained multiple concussions. These patients may also need more time and accommodation.

Patients with an underlying neurologic condition, recognized prior to injury or not, may have delayed or incomplete recovery. Even aging may be an “underlying condition” in concussion.

Patients whose symptoms from an apparently single mild concussion do not resolve despite appropriate treatments may have identifiable factors, but intractable pain (usually headache) or significant emotional disturbance or both are common. Once established and persistent, this is difficult to treat. Referral to a specialty practice is appropriate, but even in that setting effective treatment may be elusive.

PATIENT EDUCATION

Most important for patient education is reassurance. Ultimately, concussion is a self-limited phenomenon, and reinforcing this is helpful for patients. If concussion is not sport-related, most patients recover completely within 3 months.

The next important tenet in patient education is that they should rest for 3 to 5 days, then resume gradual physical and cognitive activities. If resuming activities too soon results in symptoms, then they should rest for a day and gradually resume activity. If their recovery is prolonged (ie, longer than 6 weeks), they likely need to be referred to a concussion specialist.

We well recall, back in the day, getting our “bell rung” from some form of sports-related head contact. If we could count the coach’s fingers clearly, run fast and straight, and know the plays, we could happily go back into the game. There was little additional thought given to short-term or lasting effects. I recall hearing tales from my grandfather, a boxing enthusiast, of retired punch-drunk fighters working as bouncers and greeters at sports-focused restaurants and clubs. I certainly didn’t draw any link to a few episodes of personally feeling spacey or dizzy after playing football.

But now, as parents, we are all highly tuned in to the issue of wrongly minimized “minor” head contact and concussion in our children playing sports. There is a growing research-based understanding of the mechanisms of concussion, which remains a clinical syndrome diagnosed on the basis of symptoms and sometimes subtle objective findings that occur in the appropriate environmental context. Intracranial brain impact sets the stage for locally spreading firing of neurons outside their usual pattern. This can result in a diffuse jamming of some normal electrochemical pathways of cognitive function, as well as create additional mismatch between neuronal metabolic needs and the local blood flow providing oxygen and nutrients. This disruption in autoregulation of blood flow sets the stage for enhanced brain sensitivity to any second injurious event, even a minimal one. Hence the aggressive implementation of enforced rest and recovery time for athletes and others with concussion.

It is critical to realize that the patient may not have had a loss of consciousness. Equally important is to consider the need for imaging and protection of patients who are not recovering as expected in 7 to 10 days, as well as for initial imaging of those with severe head impact or baseline neurologic disease, the aged, and those on anticoagulation.

We well recall, back in the day, getting our “bell rung” from some form of sports-related head contact. If we could count the coach’s fingers clearly, run fast and straight, and know the plays, we could happily go back into the game. There was little additional thought given to short-term or lasting effects. I recall hearing tales from my grandfather, a boxing enthusiast, of retired punch-drunk fighters working as bouncers and greeters at sports-focused restaurants and clubs. I certainly didn’t draw any link to a few episodes of personally feeling spacey or dizzy after playing football.

But now, as parents, we are all highly tuned in to the issue of wrongly minimized “minor” head contact and concussion in our children playing sports. There is a growing research-based understanding of the mechanisms of concussion, which remains a clinical syndrome diagnosed on the basis of symptoms and sometimes subtle objective findings that occur in the appropriate environmental context. Intracranial brain impact sets the stage for locally spreading firing of neurons outside their usual pattern. This can result in a diffuse jamming of some normal electrochemical pathways of cognitive function, as well as create additional mismatch between neuronal metabolic needs and the local blood flow providing oxygen and nutrients. This disruption in autoregulation of blood flow sets the stage for enhanced brain sensitivity to any second injurious event, even a minimal one. Hence the aggressive implementation of enforced rest and recovery time for athletes and others with concussion.

It is critical to realize that the patient may not have had a loss of consciousness. Equally important is to consider the need for imaging and protection of patients who are not recovering as expected in 7 to 10 days, as well as for initial imaging of those with severe head impact or baseline neurologic disease, the aged, and those on anticoagulation.

We well recall, back in the day, getting our “bell rung” from some form of sports-related head contact. If we could count the coach’s fingers clearly, run fast and straight, and know the plays, we could happily go back into the game. There was little additional thought given to short-term or lasting effects. I recall hearing tales from my grandfather, a boxing enthusiast, of retired punch-drunk fighters working as bouncers and greeters at sports-focused restaurants and clubs. I certainly didn’t draw any link to a few episodes of personally feeling spacey or dizzy after playing football.

But now, as parents, we are all highly tuned in to the issue of wrongly minimized “minor” head contact and concussion in our children playing sports. There is a growing research-based understanding of the mechanisms of concussion, which remains a clinical syndrome diagnosed on the basis of symptoms and sometimes subtle objective findings that occur in the appropriate environmental context. Intracranial brain impact sets the stage for locally spreading firing of neurons outside their usual pattern. This can result in a diffuse jamming of some normal electrochemical pathways of cognitive function, as well as create additional mismatch between neuronal metabolic needs and the local blood flow providing oxygen and nutrients. This disruption in autoregulation of blood flow sets the stage for enhanced brain sensitivity to any second injurious event, even a minimal one. Hence the aggressive implementation of enforced rest and recovery time for athletes and others with concussion.

It is critical to realize that the patient may not have had a loss of consciousness. Equally important is to consider the need for imaging and protection of patients who are not recovering as expected in 7 to 10 days, as well as for initial imaging of those with severe head impact or baseline neurologic disease, the aged, and those on anticoagulation.

A 22-year old man with no cardiac history presented to our emergency department after 5 days of dyspnea, cough, vomiting, and sharp intermittent epigastric pain. He used marijuana chronically and had inhaled it in unusually high amounts for several days before the onset of his symptoms.

The physical examination was unremarkable. Diagnostic tests including a complete blood cell count, complete metabolic panel, lipase level, urinalysis, and chest radiography showed no notable abnormalities. A urine drug screen revealed marijuana use.

Given this clinical picture, the question was whether cardiac catheterization was needed. Our young, previously healthy patient lacked risk factors for coronary artery disease and did not present with chest pain. Though dyspnea and epigastric pain are angina equivalents, he did not have the profile of patients commonly presenting with angina. Further, acute marijuana intoxication has been reported to be associated with reversible changes affecting the P and T waves and ST segments.^1,2 The likelihood of critical occlusion of the left anterior descending artery in this patient was deemed low.

PSEUDO-WELLENS SYNDROME

Wellens syndrome is characterized by biphasic or deeply inverted T waves in leads V₂ and V₃, normal precordial R-wave progression, and the absence of pathologic Q waves, in addition to a history of angina and minimal or no elevation of cardiac enzymes in a patient with or without ongoing chest pain.^3,4 This ominous syndrome is associated with critical occlusion of the proximal left anterior descending artery whose natural history is anterior myocardial infarction in the next few days. Stress testing is contraindicated, and urgent catheterization is warranted to prevent progression to myocardial infarction, even in patients without known heart disease or multiple cardiac risk factors.⁵

This case shows that acute marijuana intoxication may present with symptoms typical of Wellens syndrome. Because Wellens syndrome is considered highly specific for impending anterior myocardial infarction, urgent cardiac catheterization typically would be recommended. In this age of increasing use and legalization of marijuana, knowledge of the electrocardiographic findings associated with heavy marijuana use may prevent unnecessary cardiac catheterization procedures, especially in patients at low risk.

A 22-year old man with no cardiac history presented to our emergency department after 5 days of dyspnea, cough, vomiting, and sharp intermittent epigastric pain. He used marijuana chronically and had inhaled it in unusually high amounts for several days before the onset of his symptoms.

The physical examination was unremarkable. Diagnostic tests including a complete blood cell count, complete metabolic panel, lipase level, urinalysis, and chest radiography showed no notable abnormalities. A urine drug screen revealed marijuana use.

Given this clinical picture, the question was whether cardiac catheterization was needed. Our young, previously healthy patient lacked risk factors for coronary artery disease and did not present with chest pain. Though dyspnea and epigastric pain are angina equivalents, he did not have the profile of patients commonly presenting with angina. Further, acute marijuana intoxication has been reported to be associated with reversible changes affecting the P and T waves and ST segments.^1,2 The likelihood of critical occlusion of the left anterior descending artery in this patient was deemed low.

PSEUDO-WELLENS SYNDROME

Wellens syndrome is characterized by biphasic or deeply inverted T waves in leads V₂ and V₃, normal precordial R-wave progression, and the absence of pathologic Q waves, in addition to a history of angina and minimal or no elevation of cardiac enzymes in a patient with or without ongoing chest pain.^3,4 This ominous syndrome is associated with critical occlusion of the proximal left anterior descending artery whose natural history is anterior myocardial infarction in the next few days. Stress testing is contraindicated, and urgent catheterization is warranted to prevent progression to myocardial infarction, even in patients without known heart disease or multiple cardiac risk factors.⁵

This case shows that acute marijuana intoxication may present with symptoms typical of Wellens syndrome. Because Wellens syndrome is considered highly specific for impending anterior myocardial infarction, urgent cardiac catheterization typically would be recommended. In this age of increasing use and legalization of marijuana, knowledge of the electrocardiographic findings associated with heavy marijuana use may prevent unnecessary cardiac catheterization procedures, especially in patients at low risk.

A 22-year old man with no cardiac history presented to our emergency department after 5 days of dyspnea, cough, vomiting, and sharp intermittent epigastric pain. He used marijuana chronically and had inhaled it in unusually high amounts for several days before the onset of his symptoms.

The physical examination was unremarkable. Diagnostic tests including a complete blood cell count, complete metabolic panel, lipase level, urinalysis, and chest radiography showed no notable abnormalities. A urine drug screen revealed marijuana use.

Given this clinical picture, the question was whether cardiac catheterization was needed. Our young, previously healthy patient lacked risk factors for coronary artery disease and did not present with chest pain. Though dyspnea and epigastric pain are angina equivalents, he did not have the profile of patients commonly presenting with angina. Further, acute marijuana intoxication has been reported to be associated with reversible changes affecting the P and T waves and ST segments.^1,2 The likelihood of critical occlusion of the left anterior descending artery in this patient was deemed low.

PSEUDO-WELLENS SYNDROME

Wellens syndrome is characterized by biphasic or deeply inverted T waves in leads V₂ and V₃, normal precordial R-wave progression, and the absence of pathologic Q waves, in addition to a history of angina and minimal or no elevation of cardiac enzymes in a patient with or without ongoing chest pain.^3,4 This ominous syndrome is associated with critical occlusion of the proximal left anterior descending artery whose natural history is anterior myocardial infarction in the next few days. Stress testing is contraindicated, and urgent catheterization is warranted to prevent progression to myocardial infarction, even in patients without known heart disease or multiple cardiac risk factors.⁵

This case shows that acute marijuana intoxication may present with symptoms typical of Wellens syndrome. Because Wellens syndrome is considered highly specific for impending anterior myocardial infarction, urgent cardiac catheterization typically would be recommended. In this age of increasing use and legalization of marijuana, knowledge of the electrocardiographic findings associated with heavy marijuana use may prevent unnecessary cardiac catheterization procedures, especially in patients at low risk.

A black 37-year-old man has gradually lost 100 lb (45 kg) over the past 2 years, and reports progressive fatigue and malaise as well. He has not noted swollen lymph nodes, fever, or night sweats. He denies dyspnea, cough, or chest pain. He has no skin rashes, and no dry or red eyes or visual changes. He reports no flank pain, dysuria, frank hematuria, foamy urine, decline in urine output, or difficulty voiding.

He has no history of significant medical conditions. He does not drink, smoke, or use recreational drugs. He is not taking any prescription medications and has not been using nonsteroidal anti-inflammatory drugs (NSAIDs) or combination analgesics. He does not have a family history of kidney disease.

Physical examination. He appears relaxed and comfortable. He does not have nasal polyps or signs of pharyngeal inflammation. He has no apparent lymphadenopathy. His breath sounds are normal without rales or wheezes. Cardiac examination reveals a regular rhythm, with no rub or murmurs. The abdomen is soft and nontender with no flank pain or groin tenderness. The skin is intact with no rash or nodules.

• Temperature 98.4ºF (36.9ºC)
• Blood pressure 125/70 mm Hg
• Heart rate 102 beats per minute
• Respiratory rate 19 per minute
• Oxygen saturation 99% while breathing room air
• Weight 194 lb (88 kg)
• Body mass index 28 kg/m².

Laboratory testing (Table 1) reveals severe renal insufficiency with anemia:

• Serum creatinine 9 mg/dL (reference range 0.5–1.2)
• Estimated glomerular filtration rate (eGFR) 8 mL/min/1.73m² (using the Modification of Diet in Renal Disease Study equation).

His serum calcium level is normal, but his serum phosphorus is 5.3 mg/dL (reference range 2.5–4.6), and his parathyroid hormone level is 317 pg/mL (12–88), consistent with hyperparathyroidism secondary to chronic kidney disease. His 25-hydroxyvitamin D level is less than 13 ng/mL (30–80), and angiotensin-converting enzyme (ACE) is 129 U/L (9–67 U/L). His urinary calcium level is less than 3.0 mg/dL.

Urinalysis:

• Urine protein 100 mg/dL (0–20)
• No urine crystals
• 3 to 5 coarse granular urine casts per high-power field
• No hematuria or pyuria.

Renal ultrasonography shows normal kidneys with no hydronephrosis.

Renal biopsy study demonstrates noncaseating granulomatous interstitial nephritis (Figure 1).

GRANULOMATOUS INTERSTITIAL NEPHRITIS

1. Based on the information above, what is the most likely cause of this patient’s kidney disease?

• Medication
• Granulomatosis with polyangiitis
• Sarcoidosis
• Infection

Granulomatous interstitial nephritis is a histologic diagnosis that is present in up to 1% of renal biopsies. It has been associated with medications, infections, sarcoidosis, crystal deposits, paraproteinemia, and granulomatosis with polyangiitis and also is seen in an idiopathic form.

Medicines implicated include anticonvulsants, antibiotics, NSAIDs, allopurinol, and diuretics.

Mycobacteria and fungi are the main infective causes and seem to be the main causative factor in cases of renal transplant.¹ Granulomas are usually not found on kidney biopsy in granulomatosis with polyangiitis, and that diagnosis is usually made by the presence of antiproteinase 3 antibodies.²

Sarcoidosis is the most likely diagnosis in this patient after excluding implicated medications, infection, and vasculitis and confirming the presence of granulomatous interstitial nephritis on renal biopsy.

SARCOIDOSIS: A MULTISYSTEM DISEASE

Sarcoidosis is a multisystem inflammatory disease of unknown cause, characterized by noncaseating epithelioid granulomas. It can involve any organ but most commonly the thoracic and peripheral lymph nodes.^3,4 Involvement of the eyes and skin is also relatively common.

Extrapulmonary involvement occurs in more than 30% of cases of sarcoidosis, almost always with concomitant thoracic involvement.^5,6 Isolated extrathoracic sarcoidosis is unusual, found in only 2% of patients in a sarcoidosis case-control study.⁵

Current theory suggests that sarcoidosis develops from a cell-mediated immune response triggered by one or more unidentified antigens in people with a genetic predisposition.⁷

Sarcoidosis affects men and women of all ages, most often adults ages 20 to 40; but more recently, it has increased in US adults over age 55.⁸ The condition is more prevalent in Northern Europe and Japan, and in blacks in the United States.⁷

2. What is the likelihood of finding clinically apparent renal involvement in a patient with sarcoidosis?

• Greater than 70%
• Greater than 50%
• Up to 50%
• Less than 10%

The prevalence of renal involvement in sarcoidosis is hard to determine due to differences in study design and patient populations included in the available reports, and because renal involvement may be silent for many years. Recent studies have reported impaired renal function in 0.7% to 9.7% of cases: eg, a case-control study of 736 patients reported clinically apparent renal involvement in 0.7% of patients,⁵ and in a series of 818 patients, the incidence was 1%.⁹ In earlier studies, depending on the diagnostic criteria, the incidence ranged from 1.1% to 9.7%.¹⁰

The prevalence of renal involvement may also be underestimated because it can be asymp­tomatic, and the number of granulomas may be so few that they are absent in a small biopsy specimen. A higher prevalence of renal involvement in sarcoidosis is reported from autopsy studies, although many cases remained clinically silent. These studies have reported renal noncaseating granulomas in 7% to 23% of sarcoidosis patients.^11–13

PRESENTATION OF RENAL SARCOIDOSIS

3. What is the most common presentation in isolated renal sarcoidosis?

• Sterile pyuria
• Elevated urine eosinophils
• Renal insufficiency
• Painless hematuria

Renal manifestations of sarcoidosis include hypercalcemia, hypercalciuria, nephrocalcinosis, nephrolithiasis, and impaired renal function.¹⁴ Renal involvement can occur during the course of existing sarcoidosis, at the time of first presentation, or even as the sole presentation of the disease.^1,11,15 In patients with isolated renal sarcoidosis, the most common presentation is renal insufficiency.^15,16

Two main pathways for nephron insult that have been validated are granulomatous infiltration of the renal interstitium and disordered calcium homeostasis.^11,17 Though extremely rare, various types of glomerular disease, renal tubular defects, and renal vascular involvement such as renal artery granulomatous angiitis have been documented.¹⁸

Hypercalcemia in sarcoidosis

Sarcoidosis is known to cause hypercalcemia by increasing calcium absorption secondary to 1,25-dihydroxyvitamin D production from granulomas. Our patient’s case is unusual, as renal failure was the sole extrapulmonary manifestation of sarcoidosis without hypercalcemia.

In sarcoidosis, extrarenal production of 1-alpha-hydroxylase by activated macrophages inappropriately increases levels of 1,25-dihydroxyvitamin D (calcitriol). Subsequently, serum calcium levels are increased. Unlike its renal equivalent, granulomatous 1-alpha-hydroxylase evades the normal negative feedback of hypercalcemia, so that increased calcitriol levels are sustained, leading to hypercalcemia, often accompanied by hypercalciuria.¹⁹

Disruption of calcium homeostasis affects renal function through several mechanisms. Hypercalcemia promotes vasoconstriction of the afferent arteriole, leading to a reduction in the GFR. Intracellular calcium overload can contribute to acute tubular necrosis and intratubular precipitation of calcium, leading to tubular obstruction. Hypercalciuria predisposes to nephrolithiasis and obstructive uropathy. Chronic hypercalcemia and hypercalciuria, if untreated, cause progressive interstitial inflammation and deposition of calcium in the kidney parenchyma and tubules, resulting in nephrocalcinosis. In some cases, nephrocalcinosis leads to chronic kidney injury and renal dysfunction.

HISTOLOGIC FEATURES

4. What is the characteristic histologic feature of renal sarcoidosis?

• Membranous glomerulonephritis
• Mesangioproliferative glomerulonephritis
• Minimal change disease
• Granulomatous interstitial nephritis
• Immunoglobulin (Ig) A nephropathy

Granulomatous interstitial nephritis is the most typical histologic feature of renal sarcoidosis.^4,20–22 However, interstitial nephritis without granulomas is found in up to one-third of patients with sarcoid interstitial nephritis.^15,23

Patients with sarcoid granulomatous interstitial nephritis usually present with elevated serum creatinine with or without mild proteinuria (< 1 g/24 hours).^1,15,16 Advanced renal failure (stage 4 or 5 chronic kidney disease) is relatively common at the time of presentation.^1,15,16 In the 2 largest case series of renal sarcoidosis to date, the mean presenting serum creatinine levels were 3.0 and 4.8 mg/dL.^11,15 The most common clinical syndrome associated with sarcoidosis and granulomatous interstitial nephritis is chronic kidney disease with a decline in renal function, which if untreated can occur over weeks to months.²¹ Acute renal failure as an initial presentation is also well documented.^15,24

Even though glomerular involvement in sarcoidosis is rare, different kinds of glomerulonephritis have been reported, including membranous glomerulonephritis, mesangio­proliferative glomerulonephritis, IgA nephropathy, minimal change disease, focal segmental sclerosis, and crescentic glomerulonephritis.²⁵

DIAGNOSIS OF RENAL SARCOIDOSIS

5. How is renal sarcoidosis diagnosed?

• By exclusion
• Complete urine analysis and renal function assessment
• Renal biopsy
• Computed tomography
• Renal ultrasonography

The diagnosis of renal sarcoidosis is one of exclusion. Sarcoidosis must be considered in the differential diagnosis of renal failure of unknown origin, especially if disordered calcium homeostasis is also present. If clinically suspected, diagnosis usually requires pathohistologic demonstration of typical granulomatous lesions in the kidneys or in one or more organ systems.²⁶

In cases of sarcoidosis with granulomatous interstitial nephritis with isolated renal failure as a presenting feature, other causes of granulomatous interstitial nephritis must be ruled out. A number of drug reactions are associated with interstitial nephritis, most commonly with antibiotics, NSAIDs, and diuretics. Although granulomatous interstitial nephritis may develop as a reaction to some drugs, most cases of drug-induced interstitial nephritis do not involve granulomatous interstitial nephritis.

Other causes of granulomatous interstitial infiltrates include granulomatous infection by mycobacteria, fungi, or Brucella; foreign-body reaction such as cholesterol atheroemboli; heroin; lymphoma; or autoimmune disease such as tubulointerstitial nephritis with uveitis syndrome, granulomatosis with polyangiitis, or Crohn disease.^27,28 The absence of characteristic kidney biopsy findings does not exclude the diagnosis because renal sarcoidosis can be focal and easily missed on biopsy.²⁹

Urinary manifestations of renal sarcoidosis are usually not specific. In renal sarcoidosis with interstitial nephritis with or without granulomas, proteinuria is mild or absent, usually less than 1.0 g/day.^11,15,16 Urine studies may show a “bland” sediment (ie, without red or white blood cells) or may show sterile pyuria or microscopic hematuria. In glomerular disease, more overt proteinuria or the presence of red blood cell casts is more typical.

Hypercalciuria, nephrocalcinosis, and nephrolithiasis are nonspecific abnormalities that may be present in patients with sarcoidosis. In this regard, an elevated urine calcium level may support the diagnosis of renal sarcoidosis.

Computed tomography and renal ultrasonography may aid in diagnosis by detecting nephrocalcinosis or nephrolithiasis.

The serum ACE level is elevated in 55% to 60% of patients with sarcoidosis, but it may also be elevated in other granulomatous diseases or in chronic kidney disease from various causes.⁵ Therefore, considering its nonspecificity, the serum ACE level has a limited role in the diagnosis of sarcoidosis.³⁰ Using the ACE level as a marker for disease activity and response to treatment remains controversial because levels do not correlate with disease activity.^5,11

6. Which is a first-line therapy for renal sarcoidosis?

• Corticosteroids
• Azathioprine
• Mycophenolate mofetil
• Infliximab
• Adalimumab

Treatment of impaired calcium homeostasis in sarcoidosis includes hydration; reducing intake of calcium, vitamin D, and oxalate; and limiting sun exposure.^11,31 For more significant hypercalcemia (eg, serum calcium levels > 11 mg/dL) or nephrolithiasis, corticosteroid therapy is the first choice and should be implemented at the first sign of renal involvement. Corticosteroids inhibit the activity of 1-alpha-hydroxylase in macrophages, thereby reducing the production of 1,25-dihydroxyvitamin D.

Chloroquine and hydroxychloroquine have been mentioned in the literature as alternatives to corticosteroids.³² But the effect of these agents is less predictable and is slower than treatment with corticosteroids. Ketoconazole has no effect on granuloma formation but corrects hypercalcemia by inhibiting calcitriol production, and can be used as an adjunct for treating hypercalcemia and hypercalciuria.

Corticosteroids are the mainstay of treatment for renal sarcoidosis, including granulomatous interstitial nephritis and interstitial nephritis without granulomas. Most patients experience significant improvement in renal function. However, full recovery is rare, likely as a result of long-standing disease with some degree of already established irreversible renal injury.¹⁶

Corticosteroid dosage

There is no standard dosing protocol, but patients with impaired renal function due to biopsy-proven renal sarcoidosis should receive prednisone 0.5 to 1 mg/kg/day, depending on the severity of the disease, in a single dose every morning.

The optimal dosing and duration of maintenance therapy are unknown. Based on studies to date, the initial dosing should be maintained for 4 weeks, after which it can be tapered by 5 mg each week down to a maintenance dosage of 5 to 10 mg/day.⁴

Patients with a poor response after 4 weeks tend to have a worse renal outcome and are more susceptible to relapse.¹⁵ Fortunately, relapse often responds to increased corticosteroid doses.^11,15 In the case of relapse, the dose should be increased to the lowest effective dose and continued for 4 weeks, then tapered more gradually.

A total of 24 months of treatment seems necessary to be effective and to prevent relapse.¹⁵ Some authors have proposed a lifelong maintenance dose for patients with frequent relapses, and some propose it for all patients.⁴

Other agents

Tumor necrosis factor (TNF)-blocking agents. Considering the critical role TNF plays in granuloma formation, anti-TNF-alpha agents are useful in steroid-resistant sarcoidosis.³³ A thorough workup is necessary before starting these agents because of the increased risk of serious infection, including reactivation of latent tuberculosis. Of the current TNF-blocking agents, infliximab is most often used in renal sarcoidosis.³⁴ Experience with adalimumab is more limited, though promising results indicate it could be an alternative for patients who do not tolerate infliximab.³⁵

Azathioprine, mycophenolate mofetil, or methotrexate may also be used as a second-line agent if treatment with corticosteroids is not tolerated or does not control the disease. The evidence in support of these agents is limited. In small series, they have allowed sustainable control of renal function while reducing the steroid dose. Currently, these agents are used for patients resistant to corticosteroid therapy, who would otherwise need prolonged high-dose corticosteroid treatment, or who have corticosteroid intolerance; they allow a more effective steroid taper and maintenance of stable renal function.^15,36

The data supporting a standardized treatment of renal sarcoidosis are limited. For steroid intolerance or resistance, cytotoxic drugs and selected anti-TNF-alpha agents, as mentioned above, have shown promise in improving or stabilizing serum creatinine levels. Further exploration is required as to which agent or combination is better at limiting the disease process with fewer adverse effects.

Our patient was initially treated with corticosteroids and was ultimately weaned to a maintenance dose of 5 mg/day. He was followed as an outpatient and was started on mycophenolate mofetil in place of higher steroid doses. His renal function stabilized, but he was lost to follow-up after 2 years.

KEY POINTS

• Sarcoidosis is a multisystem granulomatous disease that most commonly involves the lungs, skin, and reticuloendothelial system.
• Renal involvement in sarcoidosis is likely underestimated due to its often clinically silent nature and the possibility of missing typical granulomatous lesions in a small or less-than-optimal biopsy sample.
• Manifestations of renal sarcoidosis include disrupted calcium homeostasis, nephrocalcinosis, nephrolithiasis, and renal failure.
• Because the clinical and histopathologic manifestations of renal sarcoidosis are nonspecific, the diagnosis is one of exclusion. In patients with renal failure or with hypercalcemia or hypercalciuria of unknown cause, renal sarcoidosis should be included in the differential diagnosis. Patients with chronic sarcoidosis should also be screened for renal impairment.
• Granulomatous interstitial nephritis is the classic histologic finding of renal sarcoidosis. Nonetheless, up to one-third of patients have interstitial nephritis without granulomas.
• Corticosteroids are the mainstay of treatment for renal sarcoidosis. An initial dose of oral prednisone 0.5 to 1 mg/kg/day should be maintained for 4 weeks and then gradually tapered to 5 to 10 mg/day for a total of 24 months. Some patients require lifelong therapy.
• Several immunosuppressive and cytotoxic agents may be used in cases of corticosteroid intolerance or to aid in effective taper of corticosteroids.

A black 37-year-old man has gradually lost 100 lb (45 kg) over the past 2 years, and reports progressive fatigue and malaise as well. He has not noted swollen lymph nodes, fever, or night sweats. He denies dyspnea, cough, or chest pain. He has no skin rashes, and no dry or red eyes or visual changes. He reports no flank pain, dysuria, frank hematuria, foamy urine, decline in urine output, or difficulty voiding.

He has no history of significant medical conditions. He does not drink, smoke, or use recreational drugs. He is not taking any prescription medications and has not been using nonsteroidal anti-inflammatory drugs (NSAIDs) or combination analgesics. He does not have a family history of kidney disease.

Physical examination. He appears relaxed and comfortable. He does not have nasal polyps or signs of pharyngeal inflammation. He has no apparent lymphadenopathy. His breath sounds are normal without rales or wheezes. Cardiac examination reveals a regular rhythm, with no rub or murmurs. The abdomen is soft and nontender with no flank pain or groin tenderness. The skin is intact with no rash or nodules.

• Temperature 98.4ºF (36.9ºC)
• Blood pressure 125/70 mm Hg
• Heart rate 102 beats per minute
• Respiratory rate 19 per minute
• Oxygen saturation 99% while breathing room air
• Weight 194 lb (88 kg)
• Body mass index 28 kg/m².

Laboratory testing (Table 1) reveals severe renal insufficiency with anemia:

• Serum creatinine 9 mg/dL (reference range 0.5–1.2)
• Estimated glomerular filtration rate (eGFR) 8 mL/min/1.73m² (using the Modification of Diet in Renal Disease Study equation).

His serum calcium level is normal, but his serum phosphorus is 5.3 mg/dL (reference range 2.5–4.6), and his parathyroid hormone level is 317 pg/mL (12–88), consistent with hyperparathyroidism secondary to chronic kidney disease. His 25-hydroxyvitamin D level is less than 13 ng/mL (30–80), and angiotensin-converting enzyme (ACE) is 129 U/L (9–67 U/L). His urinary calcium level is less than 3.0 mg/dL.

Urinalysis:

• Urine protein 100 mg/dL (0–20)
• No urine crystals
• 3 to 5 coarse granular urine casts per high-power field
• No hematuria or pyuria.

Renal ultrasonography shows normal kidneys with no hydronephrosis.

Renal biopsy study demonstrates noncaseating granulomatous interstitial nephritis (Figure 1).

GRANULOMATOUS INTERSTITIAL NEPHRITIS

1. Based on the information above, what is the most likely cause of this patient’s kidney disease?

• Medication
• Granulomatosis with polyangiitis
• Sarcoidosis
• Infection

Granulomatous interstitial nephritis is a histologic diagnosis that is present in up to 1% of renal biopsies. It has been associated with medications, infections, sarcoidosis, crystal deposits, paraproteinemia, and granulomatosis with polyangiitis and also is seen in an idiopathic form.

Medicines implicated include anticonvulsants, antibiotics, NSAIDs, allopurinol, and diuretics.

Mycobacteria and fungi are the main infective causes and seem to be the main causative factor in cases of renal transplant.¹ Granulomas are usually not found on kidney biopsy in granulomatosis with polyangiitis, and that diagnosis is usually made by the presence of antiproteinase 3 antibodies.²

Sarcoidosis is the most likely diagnosis in this patient after excluding implicated medications, infection, and vasculitis and confirming the presence of granulomatous interstitial nephritis on renal biopsy.

SARCOIDOSIS: A MULTISYSTEM DISEASE

Sarcoidosis is a multisystem inflammatory disease of unknown cause, characterized by noncaseating epithelioid granulomas. It can involve any organ but most commonly the thoracic and peripheral lymph nodes.^3,4 Involvement of the eyes and skin is also relatively common.

Extrapulmonary involvement occurs in more than 30% of cases of sarcoidosis, almost always with concomitant thoracic involvement.^5,6 Isolated extrathoracic sarcoidosis is unusual, found in only 2% of patients in a sarcoidosis case-control study.⁵

Current theory suggests that sarcoidosis develops from a cell-mediated immune response triggered by one or more unidentified antigens in people with a genetic predisposition.⁷

Sarcoidosis affects men and women of all ages, most often adults ages 20 to 40; but more recently, it has increased in US adults over age 55.⁸ The condition is more prevalent in Northern Europe and Japan, and in blacks in the United States.⁷

2. What is the likelihood of finding clinically apparent renal involvement in a patient with sarcoidosis?

• Greater than 70%
• Greater than 50%
• Up to 50%
• Less than 10%

The prevalence of renal involvement in sarcoidosis is hard to determine due to differences in study design and patient populations included in the available reports, and because renal involvement may be silent for many years. Recent studies have reported impaired renal function in 0.7% to 9.7% of cases: eg, a case-control study of 736 patients reported clinically apparent renal involvement in 0.7% of patients,⁵ and in a series of 818 patients, the incidence was 1%.⁹ In earlier studies, depending on the diagnostic criteria, the incidence ranged from 1.1% to 9.7%.¹⁰

The prevalence of renal involvement may also be underestimated because it can be asymp­tomatic, and the number of granulomas may be so few that they are absent in a small biopsy specimen. A higher prevalence of renal involvement in sarcoidosis is reported from autopsy studies, although many cases remained clinically silent. These studies have reported renal noncaseating granulomas in 7% to 23% of sarcoidosis patients.^11–13

PRESENTATION OF RENAL SARCOIDOSIS

3. What is the most common presentation in isolated renal sarcoidosis?

• Sterile pyuria
• Elevated urine eosinophils
• Renal insufficiency
• Painless hematuria

Renal manifestations of sarcoidosis include hypercalcemia, hypercalciuria, nephrocalcinosis, nephrolithiasis, and impaired renal function.¹⁴ Renal involvement can occur during the course of existing sarcoidosis, at the time of first presentation, or even as the sole presentation of the disease.^1,11,15 In patients with isolated renal sarcoidosis, the most common presentation is renal insufficiency.^15,16

Two main pathways for nephron insult that have been validated are granulomatous infiltration of the renal interstitium and disordered calcium homeostasis.^11,17 Though extremely rare, various types of glomerular disease, renal tubular defects, and renal vascular involvement such as renal artery granulomatous angiitis have been documented.¹⁸

Hypercalcemia in sarcoidosis

Sarcoidosis is known to cause hypercalcemia by increasing calcium absorption secondary to 1,25-dihydroxyvitamin D production from granulomas. Our patient’s case is unusual, as renal failure was the sole extrapulmonary manifestation of sarcoidosis without hypercalcemia.

In sarcoidosis, extrarenal production of 1-alpha-hydroxylase by activated macrophages inappropriately increases levels of 1,25-dihydroxyvitamin D (calcitriol). Subsequently, serum calcium levels are increased. Unlike its renal equivalent, granulomatous 1-alpha-hydroxylase evades the normal negative feedback of hypercalcemia, so that increased calcitriol levels are sustained, leading to hypercalcemia, often accompanied by hypercalciuria.¹⁹

Disruption of calcium homeostasis affects renal function through several mechanisms. Hypercalcemia promotes vasoconstriction of the afferent arteriole, leading to a reduction in the GFR. Intracellular calcium overload can contribute to acute tubular necrosis and intratubular precipitation of calcium, leading to tubular obstruction. Hypercalciuria predisposes to nephrolithiasis and obstructive uropathy. Chronic hypercalcemia and hypercalciuria, if untreated, cause progressive interstitial inflammation and deposition of calcium in the kidney parenchyma and tubules, resulting in nephrocalcinosis. In some cases, nephrocalcinosis leads to chronic kidney injury and renal dysfunction.

HISTOLOGIC FEATURES

4. What is the characteristic histologic feature of renal sarcoidosis?

• Membranous glomerulonephritis
• Mesangioproliferative glomerulonephritis
• Minimal change disease
• Granulomatous interstitial nephritis
• Immunoglobulin (Ig) A nephropathy

Granulomatous interstitial nephritis is the most typical histologic feature of renal sarcoidosis.^4,20–22 However, interstitial nephritis without granulomas is found in up to one-third of patients with sarcoid interstitial nephritis.^15,23

Patients with sarcoid granulomatous interstitial nephritis usually present with elevated serum creatinine with or without mild proteinuria (< 1 g/24 hours).^1,15,16 Advanced renal failure (stage 4 or 5 chronic kidney disease) is relatively common at the time of presentation.^1,15,16 In the 2 largest case series of renal sarcoidosis to date, the mean presenting serum creatinine levels were 3.0 and 4.8 mg/dL.^11,15 The most common clinical syndrome associated with sarcoidosis and granulomatous interstitial nephritis is chronic kidney disease with a decline in renal function, which if untreated can occur over weeks to months.²¹ Acute renal failure as an initial presentation is also well documented.^15,24

Even though glomerular involvement in sarcoidosis is rare, different kinds of glomerulonephritis have been reported, including membranous glomerulonephritis, mesangio­proliferative glomerulonephritis, IgA nephropathy, minimal change disease, focal segmental sclerosis, and crescentic glomerulonephritis.²⁵

DIAGNOSIS OF RENAL SARCOIDOSIS

5. How is renal sarcoidosis diagnosed?

• By exclusion
• Complete urine analysis and renal function assessment
• Renal biopsy
• Computed tomography
• Renal ultrasonography

The diagnosis of renal sarcoidosis is one of exclusion. Sarcoidosis must be considered in the differential diagnosis of renal failure of unknown origin, especially if disordered calcium homeostasis is also present. If clinically suspected, diagnosis usually requires pathohistologic demonstration of typical granulomatous lesions in the kidneys or in one or more organ systems.²⁶

In cases of sarcoidosis with granulomatous interstitial nephritis with isolated renal failure as a presenting feature, other causes of granulomatous interstitial nephritis must be ruled out. A number of drug reactions are associated with interstitial nephritis, most commonly with antibiotics, NSAIDs, and diuretics. Although granulomatous interstitial nephritis may develop as a reaction to some drugs, most cases of drug-induced interstitial nephritis do not involve granulomatous interstitial nephritis.

Other causes of granulomatous interstitial infiltrates include granulomatous infection by mycobacteria, fungi, or Brucella; foreign-body reaction such as cholesterol atheroemboli; heroin; lymphoma; or autoimmune disease such as tubulointerstitial nephritis with uveitis syndrome, granulomatosis with polyangiitis, or Crohn disease.^27,28 The absence of characteristic kidney biopsy findings does not exclude the diagnosis because renal sarcoidosis can be focal and easily missed on biopsy.²⁹

Urinary manifestations of renal sarcoidosis are usually not specific. In renal sarcoidosis with interstitial nephritis with or without granulomas, proteinuria is mild or absent, usually less than 1.0 g/day.^11,15,16 Urine studies may show a “bland” sediment (ie, without red or white blood cells) or may show sterile pyuria or microscopic hematuria. In glomerular disease, more overt proteinuria or the presence of red blood cell casts is more typical.

Hypercalciuria, nephrocalcinosis, and nephrolithiasis are nonspecific abnormalities that may be present in patients with sarcoidosis. In this regard, an elevated urine calcium level may support the diagnosis of renal sarcoidosis.

Computed tomography and renal ultrasonography may aid in diagnosis by detecting nephrocalcinosis or nephrolithiasis.

The serum ACE level is elevated in 55% to 60% of patients with sarcoidosis, but it may also be elevated in other granulomatous diseases or in chronic kidney disease from various causes.⁵ Therefore, considering its nonspecificity, the serum ACE level has a limited role in the diagnosis of sarcoidosis.³⁰ Using the ACE level as a marker for disease activity and response to treatment remains controversial because levels do not correlate with disease activity.^5,11

6. Which is a first-line therapy for renal sarcoidosis?

• Corticosteroids
• Azathioprine
• Mycophenolate mofetil
• Infliximab
• Adalimumab

Treatment of impaired calcium homeostasis in sarcoidosis includes hydration; reducing intake of calcium, vitamin D, and oxalate; and limiting sun exposure.^11,31 For more significant hypercalcemia (eg, serum calcium levels > 11 mg/dL) or nephrolithiasis, corticosteroid therapy is the first choice and should be implemented at the first sign of renal involvement. Corticosteroids inhibit the activity of 1-alpha-hydroxylase in macrophages, thereby reducing the production of 1,25-dihydroxyvitamin D.

Chloroquine and hydroxychloroquine have been mentioned in the literature as alternatives to corticosteroids.³² But the effect of these agents is less predictable and is slower than treatment with corticosteroids. Ketoconazole has no effect on granuloma formation but corrects hypercalcemia by inhibiting calcitriol production, and can be used as an adjunct for treating hypercalcemia and hypercalciuria.

Corticosteroids are the mainstay of treatment for renal sarcoidosis, including granulomatous interstitial nephritis and interstitial nephritis without granulomas. Most patients experience significant improvement in renal function. However, full recovery is rare, likely as a result of long-standing disease with some degree of already established irreversible renal injury.¹⁶

Corticosteroid dosage

There is no standard dosing protocol, but patients with impaired renal function due to biopsy-proven renal sarcoidosis should receive prednisone 0.5 to 1 mg/kg/day, depending on the severity of the disease, in a single dose every morning.

The optimal dosing and duration of maintenance therapy are unknown. Based on studies to date, the initial dosing should be maintained for 4 weeks, after which it can be tapered by 5 mg each week down to a maintenance dosage of 5 to 10 mg/day.⁴

Patients with a poor response after 4 weeks tend to have a worse renal outcome and are more susceptible to relapse.¹⁵ Fortunately, relapse often responds to increased corticosteroid doses.^11,15 In the case of relapse, the dose should be increased to the lowest effective dose and continued for 4 weeks, then tapered more gradually.

A total of 24 months of treatment seems necessary to be effective and to prevent relapse.¹⁵ Some authors have proposed a lifelong maintenance dose for patients with frequent relapses, and some propose it for all patients.⁴

Other agents

Tumor necrosis factor (TNF)-blocking agents. Considering the critical role TNF plays in granuloma formation, anti-TNF-alpha agents are useful in steroid-resistant sarcoidosis.³³ A thorough workup is necessary before starting these agents because of the increased risk of serious infection, including reactivation of latent tuberculosis. Of the current TNF-blocking agents, infliximab is most often used in renal sarcoidosis.³⁴ Experience with adalimumab is more limited, though promising results indicate it could be an alternative for patients who do not tolerate infliximab.³⁵

Azathioprine, mycophenolate mofetil, or methotrexate may also be used as a second-line agent if treatment with corticosteroids is not tolerated or does not control the disease. The evidence in support of these agents is limited. In small series, they have allowed sustainable control of renal function while reducing the steroid dose. Currently, these agents are used for patients resistant to corticosteroid therapy, who would otherwise need prolonged high-dose corticosteroid treatment, or who have corticosteroid intolerance; they allow a more effective steroid taper and maintenance of stable renal function.^15,36

The data supporting a standardized treatment of renal sarcoidosis are limited. For steroid intolerance or resistance, cytotoxic drugs and selected anti-TNF-alpha agents, as mentioned above, have shown promise in improving or stabilizing serum creatinine levels. Further exploration is required as to which agent or combination is better at limiting the disease process with fewer adverse effects.

Our patient was initially treated with corticosteroids and was ultimately weaned to a maintenance dose of 5 mg/day. He was followed as an outpatient and was started on mycophenolate mofetil in place of higher steroid doses. His renal function stabilized, but he was lost to follow-up after 2 years.

KEY POINTS

• Sarcoidosis is a multisystem granulomatous disease that most commonly involves the lungs, skin, and reticuloendothelial system.
• Renal involvement in sarcoidosis is likely underestimated due to its often clinically silent nature and the possibility of missing typical granulomatous lesions in a small or less-than-optimal biopsy sample.
• Manifestations of renal sarcoidosis include disrupted calcium homeostasis, nephrocalcinosis, nephrolithiasis, and renal failure.
• Because the clinical and histopathologic manifestations of renal sarcoidosis are nonspecific, the diagnosis is one of exclusion. In patients with renal failure or with hypercalcemia or hypercalciuria of unknown cause, renal sarcoidosis should be included in the differential diagnosis. Patients with chronic sarcoidosis should also be screened for renal impairment.
• Granulomatous interstitial nephritis is the classic histologic finding of renal sarcoidosis. Nonetheless, up to one-third of patients have interstitial nephritis without granulomas.
• Corticosteroids are the mainstay of treatment for renal sarcoidosis. An initial dose of oral prednisone 0.5 to 1 mg/kg/day should be maintained for 4 weeks and then gradually tapered to 5 to 10 mg/day for a total of 24 months. Some patients require lifelong therapy.
• Several immunosuppressive and cytotoxic agents may be used in cases of corticosteroid intolerance or to aid in effective taper of corticosteroids.

A black 37-year-old man has gradually lost 100 lb (45 kg) over the past 2 years, and reports progressive fatigue and malaise as well. He has not noted swollen lymph nodes, fever, or night sweats. He denies dyspnea, cough, or chest pain. He has no skin rashes, and no dry or red eyes or visual changes. He reports no flank pain, dysuria, frank hematuria, foamy urine, decline in urine output, or difficulty voiding.

He has no history of significant medical conditions. He does not drink, smoke, or use recreational drugs. He is not taking any prescription medications and has not been using nonsteroidal anti-inflammatory drugs (NSAIDs) or combination analgesics. He does not have a family history of kidney disease.

Physical examination. He appears relaxed and comfortable. He does not have nasal polyps or signs of pharyngeal inflammation. He has no apparent lymphadenopathy. His breath sounds are normal without rales or wheezes. Cardiac examination reveals a regular rhythm, with no rub or murmurs. The abdomen is soft and nontender with no flank pain or groin tenderness. The skin is intact with no rash or nodules.

• Temperature 98.4ºF (36.9ºC)
• Blood pressure 125/70 mm Hg
• Heart rate 102 beats per minute
• Respiratory rate 19 per minute
• Oxygen saturation 99% while breathing room air
• Weight 194 lb (88 kg)
• Body mass index 28 kg/m².

Laboratory testing (Table 1) reveals severe renal insufficiency with anemia:

• Serum creatinine 9 mg/dL (reference range 0.5–1.2)
• Estimated glomerular filtration rate (eGFR) 8 mL/min/1.73m² (using the Modification of Diet in Renal Disease Study equation).

His serum calcium level is normal, but his serum phosphorus is 5.3 mg/dL (reference range 2.5–4.6), and his parathyroid hormone level is 317 pg/mL (12–88), consistent with hyperparathyroidism secondary to chronic kidney disease. His 25-hydroxyvitamin D level is less than 13 ng/mL (30–80), and angiotensin-converting enzyme (ACE) is 129 U/L (9–67 U/L). His urinary calcium level is less than 3.0 mg/dL.

Urinalysis:

• Urine protein 100 mg/dL (0–20)
• No urine crystals
• 3 to 5 coarse granular urine casts per high-power field
• No hematuria or pyuria.

Renal ultrasonography shows normal kidneys with no hydronephrosis.

Renal biopsy study demonstrates noncaseating granulomatous interstitial nephritis (Figure 1).

GRANULOMATOUS INTERSTITIAL NEPHRITIS

1. Based on the information above, what is the most likely cause of this patient’s kidney disease?

• Medication
• Granulomatosis with polyangiitis
• Sarcoidosis
• Infection

Granulomatous interstitial nephritis is a histologic diagnosis that is present in up to 1% of renal biopsies. It has been associated with medications, infections, sarcoidosis, crystal deposits, paraproteinemia, and granulomatosis with polyangiitis and also is seen in an idiopathic form.

Medicines implicated include anticonvulsants, antibiotics, NSAIDs, allopurinol, and diuretics.

Mycobacteria and fungi are the main infective causes and seem to be the main causative factor in cases of renal transplant.¹ Granulomas are usually not found on kidney biopsy in granulomatosis with polyangiitis, and that diagnosis is usually made by the presence of antiproteinase 3 antibodies.²

Sarcoidosis is the most likely diagnosis in this patient after excluding implicated medications, infection, and vasculitis and confirming the presence of granulomatous interstitial nephritis on renal biopsy.

SARCOIDOSIS: A MULTISYSTEM DISEASE

Sarcoidosis is a multisystem inflammatory disease of unknown cause, characterized by noncaseating epithelioid granulomas. It can involve any organ but most commonly the thoracic and peripheral lymph nodes.^3,4 Involvement of the eyes and skin is also relatively common.

Extrapulmonary involvement occurs in more than 30% of cases of sarcoidosis, almost always with concomitant thoracic involvement.^5,6 Isolated extrathoracic sarcoidosis is unusual, found in only 2% of patients in a sarcoidosis case-control study.⁵

Current theory suggests that sarcoidosis develops from a cell-mediated immune response triggered by one or more unidentified antigens in people with a genetic predisposition.⁷

Sarcoidosis affects men and women of all ages, most often adults ages 20 to 40; but more recently, it has increased in US adults over age 55.⁸ The condition is more prevalent in Northern Europe and Japan, and in blacks in the United States.⁷

2. What is the likelihood of finding clinically apparent renal involvement in a patient with sarcoidosis?

• Greater than 70%
• Greater than 50%
• Up to 50%
• Less than 10%

The prevalence of renal involvement in sarcoidosis is hard to determine due to differences in study design and patient populations included in the available reports, and because renal involvement may be silent for many years. Recent studies have reported impaired renal function in 0.7% to 9.7% of cases: eg, a case-control study of 736 patients reported clinically apparent renal involvement in 0.7% of patients,⁵ and in a series of 818 patients, the incidence was 1%.⁹ In earlier studies, depending on the diagnostic criteria, the incidence ranged from 1.1% to 9.7%.¹⁰

The prevalence of renal involvement may also be underestimated because it can be asymp­tomatic, and the number of granulomas may be so few that they are absent in a small biopsy specimen. A higher prevalence of renal involvement in sarcoidosis is reported from autopsy studies, although many cases remained clinically silent. These studies have reported renal noncaseating granulomas in 7% to 23% of sarcoidosis patients.^11–13

PRESENTATION OF RENAL SARCOIDOSIS

3. What is the most common presentation in isolated renal sarcoidosis?

• Sterile pyuria
• Elevated urine eosinophils
• Renal insufficiency
• Painless hematuria

Renal manifestations of sarcoidosis include hypercalcemia, hypercalciuria, nephrocalcinosis, nephrolithiasis, and impaired renal function.¹⁴ Renal involvement can occur during the course of existing sarcoidosis, at the time of first presentation, or even as the sole presentation of the disease.^1,11,15 In patients with isolated renal sarcoidosis, the most common presentation is renal insufficiency.^15,16

Two main pathways for nephron insult that have been validated are granulomatous infiltration of the renal interstitium and disordered calcium homeostasis.^11,17 Though extremely rare, various types of glomerular disease, renal tubular defects, and renal vascular involvement such as renal artery granulomatous angiitis have been documented.¹⁸

Hypercalcemia in sarcoidosis

Sarcoidosis is known to cause hypercalcemia by increasing calcium absorption secondary to 1,25-dihydroxyvitamin D production from granulomas. Our patient’s case is unusual, as renal failure was the sole extrapulmonary manifestation of sarcoidosis without hypercalcemia.

In sarcoidosis, extrarenal production of 1-alpha-hydroxylase by activated macrophages inappropriately increases levels of 1,25-dihydroxyvitamin D (calcitriol). Subsequently, serum calcium levels are increased. Unlike its renal equivalent, granulomatous 1-alpha-hydroxylase evades the normal negative feedback of hypercalcemia, so that increased calcitriol levels are sustained, leading to hypercalcemia, often accompanied by hypercalciuria.¹⁹

Disruption of calcium homeostasis affects renal function through several mechanisms. Hypercalcemia promotes vasoconstriction of the afferent arteriole, leading to a reduction in the GFR. Intracellular calcium overload can contribute to acute tubular necrosis and intratubular precipitation of calcium, leading to tubular obstruction. Hypercalciuria predisposes to nephrolithiasis and obstructive uropathy. Chronic hypercalcemia and hypercalciuria, if untreated, cause progressive interstitial inflammation and deposition of calcium in the kidney parenchyma and tubules, resulting in nephrocalcinosis. In some cases, nephrocalcinosis leads to chronic kidney injury and renal dysfunction.

HISTOLOGIC FEATURES

4. What is the characteristic histologic feature of renal sarcoidosis?

• Membranous glomerulonephritis
• Mesangioproliferative glomerulonephritis
• Minimal change disease
• Granulomatous interstitial nephritis
• Immunoglobulin (Ig) A nephropathy

Granulomatous interstitial nephritis is the most typical histologic feature of renal sarcoidosis.^4,20–22 However, interstitial nephritis without granulomas is found in up to one-third of patients with sarcoid interstitial nephritis.^15,23

Patients with sarcoid granulomatous interstitial nephritis usually present with elevated serum creatinine with or without mild proteinuria (< 1 g/24 hours).^1,15,16 Advanced renal failure (stage 4 or 5 chronic kidney disease) is relatively common at the time of presentation.^1,15,16 In the 2 largest case series of renal sarcoidosis to date, the mean presenting serum creatinine levels were 3.0 and 4.8 mg/dL.^11,15 The most common clinical syndrome associated with sarcoidosis and granulomatous interstitial nephritis is chronic kidney disease with a decline in renal function, which if untreated can occur over weeks to months.²¹ Acute renal failure as an initial presentation is also well documented.^15,24

Even though glomerular involvement in sarcoidosis is rare, different kinds of glomerulonephritis have been reported, including membranous glomerulonephritis, mesangio­proliferative glomerulonephritis, IgA nephropathy, minimal change disease, focal segmental sclerosis, and crescentic glomerulonephritis.²⁵

DIAGNOSIS OF RENAL SARCOIDOSIS

5. How is renal sarcoidosis diagnosed?

• By exclusion
• Complete urine analysis and renal function assessment
• Renal biopsy
• Computed tomography
• Renal ultrasonography

The diagnosis of renal sarcoidosis is one of exclusion. Sarcoidosis must be considered in the differential diagnosis of renal failure of unknown origin, especially if disordered calcium homeostasis is also present. If clinically suspected, diagnosis usually requires pathohistologic demonstration of typical granulomatous lesions in the kidneys or in one or more organ systems.²⁶

In cases of sarcoidosis with granulomatous interstitial nephritis with isolated renal failure as a presenting feature, other causes of granulomatous interstitial nephritis must be ruled out. A number of drug reactions are associated with interstitial nephritis, most commonly with antibiotics, NSAIDs, and diuretics. Although granulomatous interstitial nephritis may develop as a reaction to some drugs, most cases of drug-induced interstitial nephritis do not involve granulomatous interstitial nephritis.

Other causes of granulomatous interstitial infiltrates include granulomatous infection by mycobacteria, fungi, or Brucella; foreign-body reaction such as cholesterol atheroemboli; heroin; lymphoma; or autoimmune disease such as tubulointerstitial nephritis with uveitis syndrome, granulomatosis with polyangiitis, or Crohn disease.^27,28 The absence of characteristic kidney biopsy findings does not exclude the diagnosis because renal sarcoidosis can be focal and easily missed on biopsy.²⁹

Urinary manifestations of renal sarcoidosis are usually not specific. In renal sarcoidosis with interstitial nephritis with or without granulomas, proteinuria is mild or absent, usually less than 1.0 g/day.^11,15,16 Urine studies may show a “bland” sediment (ie, without red or white blood cells) or may show sterile pyuria or microscopic hematuria. In glomerular disease, more overt proteinuria or the presence of red blood cell casts is more typical.

Hypercalciuria, nephrocalcinosis, and nephrolithiasis are nonspecific abnormalities that may be present in patients with sarcoidosis. In this regard, an elevated urine calcium level may support the diagnosis of renal sarcoidosis.

Computed tomography and renal ultrasonography may aid in diagnosis by detecting nephrocalcinosis or nephrolithiasis.

The serum ACE level is elevated in 55% to 60% of patients with sarcoidosis, but it may also be elevated in other granulomatous diseases or in chronic kidney disease from various causes.⁵ Therefore, considering its nonspecificity, the serum ACE level has a limited role in the diagnosis of sarcoidosis.³⁰ Using the ACE level as a marker for disease activity and response to treatment remains controversial because levels do not correlate with disease activity.^5,11

6. Which is a first-line therapy for renal sarcoidosis?

• Corticosteroids
• Azathioprine
• Mycophenolate mofetil
• Infliximab
• Adalimumab

Treatment of impaired calcium homeostasis in sarcoidosis includes hydration; reducing intake of calcium, vitamin D, and oxalate; and limiting sun exposure.^11,31 For more significant hypercalcemia (eg, serum calcium levels > 11 mg/dL) or nephrolithiasis, corticosteroid therapy is the first choice and should be implemented at the first sign of renal involvement. Corticosteroids inhibit the activity of 1-alpha-hydroxylase in macrophages, thereby reducing the production of 1,25-dihydroxyvitamin D.

Chloroquine and hydroxychloroquine have been mentioned in the literature as alternatives to corticosteroids.³² But the effect of these agents is less predictable and is slower than treatment with corticosteroids. Ketoconazole has no effect on granuloma formation but corrects hypercalcemia by inhibiting calcitriol production, and can be used as an adjunct for treating hypercalcemia and hypercalciuria.

Corticosteroids are the mainstay of treatment for renal sarcoidosis, including granulomatous interstitial nephritis and interstitial nephritis without granulomas. Most patients experience significant improvement in renal function. However, full recovery is rare, likely as a result of long-standing disease with some degree of already established irreversible renal injury.¹⁶

Corticosteroid dosage

There is no standard dosing protocol, but patients with impaired renal function due to biopsy-proven renal sarcoidosis should receive prednisone 0.5 to 1 mg/kg/day, depending on the severity of the disease, in a single dose every morning.

The optimal dosing and duration of maintenance therapy are unknown. Based on studies to date, the initial dosing should be maintained for 4 weeks, after which it can be tapered by 5 mg each week down to a maintenance dosage of 5 to 10 mg/day.⁴

Patients with a poor response after 4 weeks tend to have a worse renal outcome and are more susceptible to relapse.¹⁵ Fortunately, relapse often responds to increased corticosteroid doses.^11,15 In the case of relapse, the dose should be increased to the lowest effective dose and continued for 4 weeks, then tapered more gradually.

A total of 24 months of treatment seems necessary to be effective and to prevent relapse.¹⁵ Some authors have proposed a lifelong maintenance dose for patients with frequent relapses, and some propose it for all patients.⁴

Other agents

Tumor necrosis factor (TNF)-blocking agents. Considering the critical role TNF plays in granuloma formation, anti-TNF-alpha agents are useful in steroid-resistant sarcoidosis.³³ A thorough workup is necessary before starting these agents because of the increased risk of serious infection, including reactivation of latent tuberculosis. Of the current TNF-blocking agents, infliximab is most often used in renal sarcoidosis.³⁴ Experience with adalimumab is more limited, though promising results indicate it could be an alternative for patients who do not tolerate infliximab.³⁵

Azathioprine, mycophenolate mofetil, or methotrexate may also be used as a second-line agent if treatment with corticosteroids is not tolerated or does not control the disease. The evidence in support of these agents is limited. In small series, they have allowed sustainable control of renal function while reducing the steroid dose. Currently, these agents are used for patients resistant to corticosteroid therapy, who would otherwise need prolonged high-dose corticosteroid treatment, or who have corticosteroid intolerance; they allow a more effective steroid taper and maintenance of stable renal function.^15,36

The data supporting a standardized treatment of renal sarcoidosis are limited. For steroid intolerance or resistance, cytotoxic drugs and selected anti-TNF-alpha agents, as mentioned above, have shown promise in improving or stabilizing serum creatinine levels. Further exploration is required as to which agent or combination is better at limiting the disease process with fewer adverse effects.

Our patient was initially treated with corticosteroids and was ultimately weaned to a maintenance dose of 5 mg/day. He was followed as an outpatient and was started on mycophenolate mofetil in place of higher steroid doses. His renal function stabilized, but he was lost to follow-up after 2 years.

KEY POINTS

• Sarcoidosis is a multisystem granulomatous disease that most commonly involves the lungs, skin, and reticuloendothelial system.
• Renal involvement in sarcoidosis is likely underestimated due to its often clinically silent nature and the possibility of missing typical granulomatous lesions in a small or less-than-optimal biopsy sample.
• Manifestations of renal sarcoidosis include disrupted calcium homeostasis, nephrocalcinosis, nephrolithiasis, and renal failure.
• Because the clinical and histopathologic manifestations of renal sarcoidosis are nonspecific, the diagnosis is one of exclusion. In patients with renal failure or with hypercalcemia or hypercalciuria of unknown cause, renal sarcoidosis should be included in the differential diagnosis. Patients with chronic sarcoidosis should also be screened for renal impairment.
• Granulomatous interstitial nephritis is the classic histologic finding of renal sarcoidosis. Nonetheless, up to one-third of patients have interstitial nephritis without granulomas.
• Corticosteroids are the mainstay of treatment for renal sarcoidosis. An initial dose of oral prednisone 0.5 to 1 mg/kg/day should be maintained for 4 weeks and then gradually tapered to 5 to 10 mg/day for a total of 24 months. Some patients require lifelong therapy.
• Several immunosuppressive and cytotoxic agents may be used in cases of corticosteroid intolerance or to aid in effective taper of corticosteroids.