Cleveland Clinic Journal of Medicine

A 50-year-old African American woman with type 2 diabetes mellitus, hypertension, hyperlipidemia, and chronic kidney disease presents for a follow-up visit. The patient had been treated with hydrochlorothiazide 25 mg/day and enalapril (Vasotec) 20 mg twice daily until 6 weeks ago. At that time her blood pressure was 160/85 mm Hg, and amlodipine (Norvasc) 10 mg/day was added to her regimen. Her other medications include glipizide (Glucotrol), metformin (Glucophage), lovastatin (Mevacor), fish oils, aspirin, calcium, and vitamin D. Her current blood pressure is 145/80 mm Hg; her serum creatinine level is 1.5 mg/dL, and her urine albumin-to-creatinine ratio is 180 mg/g.

In hypertensive patients who have diabetes or chronic kidney disease, guidelines¹ call for intensification of antihypertensive therapy to reach a goal blood pressure of less than 130/80 mm Hg. What data exist to support these guidelines? And what should the clinician do?

IS MORE-INTENSE THERAPY IN THE PATIENT’S BEST INTEREST?

Often, clinicians are faced with hypertensive patients whose blood pressure, despite treatment, is higher than the accepted goal. Often, these patients are elderly and are already taking multiple medications that are costly and have significant potential adverse effects. The dilemma is whether to try to reach a target blood pressure listed in a guideline (by increasing the dosage of the current drugs or by adding a drug of a different class) or to “do no harm,” accept the patient’s blood pressure, and keep the regimen the same.^1,2

The current goal blood pressure is less than 140/90 mm Hg for all but the very elderly, with more intense control recommended for patients at high risk, ie, those with diabetes mellitus, chronic kidney disease, or atherosclerotic cardiovascular disease.¹

While it appears to be in the patient’s best interests to follow such guidelines, review of available data indicates that this it not necessarily so, and may even be harmful.

OBSERVATIONAL DATA AND EARLY RANDOMIZED TRIALS

Many observational studies have found that the higher one’s blood pressure, the greater one’s risk of cardiovascular events and death. Indeed, meta-analyses of these trials, which involved more than 1.5 million people, demonstrate a strong, positive, log-linear relationship between blood pressure and the incidence of cardiovascular disease and death.^3–5

Further, there is no evidence of a threshold pressure below which the risk is not lower (ie, a “J-point”), starting with 115/75 mm Hg. A J-point may exist for diastolic blood pressure in elderly patients with isolated systolic hypertension⁶ and in patients with coronary artery disease.⁷ Otherwise, the observation is clear: the lower the blood pressure the better. For every 20 mm Hg lower systolic blood pressure or 10 mm Hg lower diastolic blood pressure, the risk of a cardiovascular event is about 50% less.^4,5

Observational analyses also show a strong, graded relationship between blood pressure and future end-stage renal disease.^8,9 Post hoc analyses indicate that chronic kidney disease progresses more slowly with lower achieved blood pressures, especially in those with higher degrees of proteinuria.^10–12

However, observational data do not prove cause and effect, nor do they guarantee similar results with treatment. This requires randomized controlled trials.

RANDOMIZED TRIALS OF HYPERTENSION TREATMENT

Initial trials were aimed at determining whether hypertension should even be treated. A 1997 meta-analysis of 18 such trials comparing either low-dose diuretic therapy, high-dose diuretic therapy, or beta-blocker therapy with placebo involved 48,000 patients who were followed for an average of 5 years.¹³ The rates of stroke and congestive heart failure were consistently reduced, although only low-dose diuretic therapy reduced the risk of coronary heart disease and death from any cause.

More recent trials enrolled people not considered hypertensive who were randomized to receive either active drugs or placebo, or no treatment. Other trials attempted to assess non-pressure-related effects of specific agents, using other antihypertensive agents in the control group. Still other randomized controlled trials compared one agent or agents with other agents while attempting to attain equivalent blood pressure between groups. Frequently, however, there was some blood pressure difference.

Meta-analyses of most of these trials conclude that the major benefit of antihypertensive therapy—reducing rates of cardiovascular morbidity and mortality—comes from a lower attained blood pressure, irrespective of which agent is used.^14–18 Exceptions exist, however. For example, specific drug classes are indicated after myocardial infarction, and in congestive heart failure and proteinuric chronic kidney disease.^10,19–21

16 TRIALS OF DIFFERENT BLOOD PRESSURE TARGETS

The overriding theme of these observational data is that a lower blood pressure, whether attained naturally or with treatment, is better than a higher one from both the cardiovascular and the renal perspective.

What remains unclear is what blood pressure should be aimed for in a particular patient or group of patients. Is it a specific pressure (eg, 140/90 mm Hg), or does the change from baseline count more? Should other factors such as age or comorbidity alter this number?

Several randomized controlled trials have addressed these questions by targeting different levels of blood pressure. We are aware of at least 16 such trials in adults, including 13 with renal or cardiovascular primary end points and three with surrogate primary end points.

An unavoidable design flaw of all of these trials is their unblinded nature. Consequently, nearly all of them carry a Jadad score (a measure of quality, based on randomization and blinding)²² of 3 on a scale of 5.

NINE TRIALS WITH RENAL PRIMARY END POINTS

African American Study of Kidney Disease and Hypertension (AASK)²³

Patients: 1,094 African Americans with presumed hypertensive renal disease and a measured glomerular filtration rate between 20 and 65 mL/min/1.73 m².

Randomized blood pressure goals. Mean arterial pressure 92 mm Hg or less vs 102 to 107 mm Hg.

Results. At 4 years, the two groups had average blood pressures of 128/78 and 141/85 mm Hg, respectively. The groups did not differ in the rates of the primary end points—ie, the rate of change in the measured glomerular filtration rate over time or the composite of a 50% reduction in glomerular filtration rate, the onset of end-stage renal disease, or death.

Comments. Several issues have been raised about the internal validity of this trial.

So-called hypertensive kidney disease in African Americans (as opposed to European Americans) may be a genetic disorder related to polymorphisms of one or more genes on chromosome 22q. Initial data implicated the MYH9 gene, which encodes non-muscle myosin heavy chain II.^24,25 More recent data implicate the nearby APOL1 gene encoding apolipoprotein L126 as more relevant. These polymorphisms have a much greater prevalence in African Americans and appear responsible for the higher risk of idiopathic focal segmental glomerulosclerosis and HIV-associated nephropathy in this population.^24–26 Therefore, in African Americans, hypertension may in fact be the result of the kidney disease and not its primary cause, which may explain why in this and other African American populations stricter control of blood pressure did not produce a renal benefit.^27,28

Also, there is the possibility of misclassification bias. A secondary analysis of data obtained by ambulatory monitoring showed that of the 377 participants whose blood pressure appeared to be under control when measured in the clinic, 70% actually had masked hypertension, ie, uncontrolled hypertension outside the clinic.²⁹ The real difference in blood pressure between groups may well have been different than that determined in the clinic.

In addition, a prespecified secondary analysis showed no difference in the rates of cardiovascular events and death between the groups.³⁰ However, the study was not designed to have the statistical power to detect a difference in cardiovascular events. Moreover fewer cardiovascular events occurred than expected, further reducing the study’s power to detect a difference.

Toto et al³¹

Toto et al reported similar results in an earlier trial in 87 hypertensive patients (77 randomized), predominantly African American, and similar concerns apply.

Lewis et al³²

Patients: 129 patients with type 1 diabetes.

Randomized blood pressure goals. A mean arterial pressure of either no higher than 92 mm Hg or 100 to 107 mm Hg.

Results. At 2 years, despite a difference of 6 mm Hg in mean arterial pressure, the glomerular filtration rate (measured) had declined by the same amount in the two groups. The study was underpowered for this end point. Patients in the group with the lower goal pressure were excreting significantly less protein than those in the other group, but they were received higher doses of an angiotensin-converting enzyme (ACE) inhibitor—in this case, ramipril (Altace).

The Appropriate Blood Pressure Control in Diabetes (ABCD) trials^33–35

Patients: 950 patients with type 2 diabetes mellitus and either normal or high blood pressure.

Randomized blood pressure goals. Either intensive or moderate therapy (see Table 1).

Results. At 5 years, creatinine clearance (estimated) had declined by the same amount in the two groups. However, fewer of the hypertensive patients had died in the intensive-therapy group.³⁴ Similarly, normotensive patients had less progression of albuminuria if treated intensively.³³

In the ABCD Part 2 with Valsartan (ABCD-2V) trial in normotensive patients,³⁵ therapy with valsartan (Diovan) did not affect creatinine clearance but did reduce albuminuria. However, 75% of the patients in the moderate-treatment group were untreated.

Schrier et al³⁶

Patients. 75 hypertensive patients with autosomal-dominant polycystic kidney disease and left ventricular hypertrophy.

Randomized blood pressure targets. Less than 120/80 mm Hg vs 135/85 to 140/90 mm Hg.

Results. After 7 years, despite a difference in average mean arterial pressure of 11 mm Hg between the groups (90 vs 101 mm Hg), there was no difference in the rate of decline of creatinine clearance. The left ventricular mass index decreased by 21% in the lower-target group and by 35% in the higher-target group (P < .01).

Modification of Diet in Renal Disease (MDRD) trial^37,38

Patients: 840 patients whose measured glomerular filtration rate was between 13 and 55 mL/min/1.73 m².

Randomized blood pressure targets. A target mean arterial pressure of less than 92 mm Hg vs less than 107 mm Hg.^11,37

Results. After 2.2 years, the mean difference in mean arterial pressure was 4.7 mm Hg. There was, however, no difference in the rate of decline in the glomerular filtration rate.

In a 6-year follow-up, significantly fewer patients in the lower-blood-pressure group reached the end point of end-stage renal disease or the combined end point of end-stage renal disease or death.³⁸ The rate of death, however, was nearly twice as high in the lower-blood-pressure group (10% vs 6%). The blood pressure and treatment during follow-up were not reported.

Comments. Internal validity is an issue, since the blood pressure and therapy during follow-up were unknown, and more patients received ACE inhibitors in the lower-blood-pressure group during the trial. Further, the higher death rate in the lower-blood-pressure group is worrisome.

Patients: 338 nondiabetic patients who had proteinuria and reduced creatinine clearance.

Treatment and blood pressure goals. All were treated with ramipril and randomized to intensive (< 130/80 mm Hg) vs standard control (diastolic blood pressure < 90 mm Hg) with therapy based on felodipine (Plendil).

Results. The study was terminated early because of futility. Despite a mean difference of 4.1 mm Hg systolic and 2.8 mm Hg diastolic, the groups did not differ in the rate of progression to end-stage renal disease (23% with intensive therapy vs 20% with standard therapy) or in the rate of decline of the measured glomerular filtration rate (0.22 vs 0.24 mL/min/1.73 m²/month).

Comment. The internal validity of this study can be questioned because of the low separation of achieved blood pressure and because of its early termination.

No benefit from a lower blood pressure goal in preserving kidney function

To summarize, these trials all showed no significant benefit from either targeting or achieving lower blood pressure in terms of slowing the decline of kidney function. Overall, they do not define a target and offer little support that a lower goal blood pressure is indicated with respect to the rate of loss of glomerular filtration rate in chronic kidney disease.

However, post hoc analysis of the MDRD trial indicates a statistical interaction between targeted blood pressure and degree of baseline proteinuria. At higher levels of proteinuria (≥ 1 g/day), the group with the lower blood pressure target had better outcomes.

In addition, long-term follow-up (mean of 12.2 years) of the AASK trial, including a 7-year cohort phase with nearly similar blood pressures in both groups, also indicated an interaction with targeted blood pressure and baseline proteinuria.⁴⁰ Although the overall analysis was negative, there was a significant reduction in the primary end point in the group originally assigned the low target when analysis was restricted to those in the highest tertile of proteinuria. These and other data¹⁰ suggest that patients with chronic kidney disease and proteinuria may represent a distinct subset of chronic kidney disease patients who benefit from more intensive blood-pressure-lowering. However, patients in the REIN-2 trial³⁴ and the macroalbuminuric patients in the ABCD hypertensive trial³⁵ did not benefit from a lower targeted blood pressure despite significant proteinuria.

FOUR TRIALS WITH CARDIOVASCULAR END POINTS

The Hypertension Optimal Treatment (HOT) trial⁴¹

Patients: 18,790 patients with diastolic blood pressure between 100 and 115 mm Hg.

Randomized blood pressure goals. Diastolic pressure of equal to or less than 80, 85, or 90 mm Hg.

Results. At an average of 3.8 years, the average blood pressures in the three groups were approximately 140/81, 141/83, and 144/85 mm Hg, respectively. There was no difference between the groups in the rate of the composite primary end point of all myocardial infarctions, all strokes, and cardiovascular death. Any conclusions from this trial were compromised by the small difference in achieved blood pressures between groups.

In the 1,501 patients with diabetes, the incidence of the primary end point was 50% lower with a goal of 80 mm Hg or less than with a goal of 90 mm Hg or less.

The UK Prospective Diabetes Study (UKPDS)^42,43

Patients: 1,148 hypertensive patients with type 2 diabetes mellitus.

Randomized blood pressure goals. Either “tight control” (aiming for < 150/85 mm Hg) or “less tight control” (aiming for < 180/105 mm Hg).

Results. At a median follow-up of 8.4 years, the attained blood pressures were 144/82 vs 154/87 mm Hg. The difference produced significant benefits, including a 24% lower rate of any diabetes-related end point, a 32% lower rate of death due to diabetes, and a nonsignificant 18% lower rate of total mortality—all co-primary end points.

The less-tight-control group had many patients with initial blood pressures below 180/105 mm Hg; hence, over 50% of patients received no antihypertensive therapy at the start of the trial. By the end of the trial 9 years later, 20% had still not been treated. This compares with only 5% of patients in the tight-control group who were not treated with antihypertensives throughout the trial. Therefore, this trial serves as better evidence for treating vs not treating, rather than defining a specific goal.

During a 10-year follow-up, blood pressure differences disappeared within 2 years.⁴³ There was no legacy effect, as the significant differences noted during the trial were no longer present 10 years later.

Action to Control Cardiovascular Risk in Diabetes (ACCORD)⁴⁴

Patients: 4,733 patients with type 2 diabetes.

Randomized blood pressure goals. Systolic blood pressure lower than either 120 or 140 mm Hg.

Results. At 4.7 years, despite a significant difference in mean systolic blood pressure of 14.2 mm Hg after the first year (119.3 vs 133.5 mm Hg), there was no difference in the primary end point of nonfatal myocardial infarction, nonfatal stroke, or cardiovascular death. There were fewer strokes in the lower-pressure group but no difference in myocardial infarctions, which were five times more common than strokes. Serious adverse events attributed to antihypertensive treatment occurred more frequently in the intensive-therapy group (3.3% vs 1.3%, P < .001).

Comment. There were fewer events than expected, possibly limiting the trial’s ability to detect a statistical difference. Compared with both the UKPDS and the diabetic population of HOT, ACCORD is much larger and more internally valid (unlike in UKPDS, nearly all patients in both groups were treated, and compared with HOT there was much greater separation of achieved pressure). It is more recent and better reflects current overall practice. It indicates that when specifically aiming for a target blood pressure, lower is not always better and comes at a price (more severe adverse events).

Patients: 4,418 patients, age 65 to 85 years, with a pretreatment systolic blood pressure above 160 mm Hg.

Randomized blood pressure goals. Systolic pressure either lower than 140 mm Hg or 140 to 160 mm Hg.

Results. At 2 years, despite a difference of 9.7/3.3 mm Hg, there was no difference in the primary end point (the combined incidence of cerebrovascular disease, cardiac and vascular disease, and renal failure). Fifty-four patients had died in the strict-treatment group and 42 in the mild-treatment group; the difference was not statistically significant.

Three other trials

Three other trials^46–48 had surrogate end points, but only one of them reported a composite cardiovascular secondary end point.⁴⁶ We will not discuss the other two.^47,48

Cardio-Sis. In the Studio Italiano Sugli Effetti Cardiovascolari del Controllo della Pressione Arteriosa Sistolica (Cardio-Sis) trial,⁴⁶ 1,111 people without diabetes with systolic pressure higher than 150 mm Hg were randomized to tight control (systolic pressure < 130 mm Hg) vs usual control (systolic pressure < 140 mm Hg) and followed for 2 years with electrocardiography to detect left ventricular hypertrophy.

At a median of 2 years, the systolic blood pressure had declined by an average of 3.8 mm Hg more in the tight-control group than in the usual-control group, and the diastolic pressure by an average of 1.5 mm Hg. There was significantly less left ventricular hypertrophy in the tight-control group. The incidence of the secondary end point of a composite of cardiovascular and renal events was also significantly lower. There was no difference individually in the rates of myocardial infarction, stroke, transient ischemic attack, admission for congestive heart failure, or death.

DISCUSSION: THE DILEMMA OF TREATING AN INDIVIDUAL PATIENT

These data illustrate the dilemma of treating an individual patient whose blood pressure is not at the currently accepted goal while on multiple antihypertensive medications. According to guidelines, therapy should be intensified in this situation. Observational data show a strong graded relationship between blood pressure and cardiovascular events and death, starting with a blood pressure of 115/75 mm Hg. The observational data relating blood pressure to kidney disease are similar. These data support the guideline recommendations that additional medications should be added to reach the promulgated target. Unfortunately, the targeting trials do not define a target, nor do they support the concept that lower is better.

Possible explanations for the negative results

Why does targeting a lower blood pressure not produce the benefit that the observational data lead us to expect?

One possibility is that blood pressure is merely a marker of cardiovascular risk, not a cause of it. This is unlikely, given the temporal relationship, reproducibility, and biologic plausibility that is supported by a very large body of experimental data. However, blood pressure is only one of multiple factors involved in the pathogenesis of vascular and renal disease, and perhaps better attention to other factors such as lipids and smoking may have made the targeting trials underpowered.

Another possibility is that these trials had such strict inclusion and exclusion criteria that they do not represent the general hypertensive population, reducing their external validity.⁴⁹ However, the trials generally enrolled populations at higher risk, in which end points were more likely to occur. This would have enhanced the chance to show a positive effect rather than mask it.

It is possible that antihypertensive medications themselves have unwanted side effects that offset their potential benefit. Medication-related side effects could directly contribute to vascular disease despite their beneficial effect of lowering pressure. There could also be reduced tissue perfusion due to lower blood pressure per se in the face of a diseased vasculature, with the lower pressure directly contributing to organ dysfunction.

Finally, these trials measured brachial pressures to monitor blood pressure. Brachial pressure does not always correlate with central aortic pressure, which is probably a better marker of the overall pressure burden.⁵⁰ It is possible that in these targeting trials, the peripheral blood pressure did not reflect the true central blood pressure and, therefore, significant separation of blood pressures may not have actually occurred.

Targeted vs achieved blood pressures: Analogies with other markers

This contradiction is not an exceptional circumstance in medicine.

For example, in chronic kidney disease, a graded observational relationship exists between decreasing levels of hemoglobin and various adverse outcomes.^51–53 However, targeting a more normal level of hemoglobin compared with a lower one has been shown to be detrimental.^54–57 This implies either that anemia is merely a marker of higher risk or, more likely, that the actual measures used to raise the hemoglobin to higher levels are the culprit. Notably, although targeting a higher hemoglobin concentration vs a lower one was detrimental, achieving a higher hemoglobin was beneficial within each targeted group.^54,58

Another example of harm caused by targeting goals based on observational data is tight glucose control, both acutely in the critically ill⁵⁹ and chronically in patients with type 2 diabetes.⁶⁰ In both cases higher mortality rates ensued.

The same concept may apply to lowering blood pressure. While achieving a lower blood pressure may be more beneficial, targeting a specific goal may be harmful. Given that perhaps 20% of those labeled as hypertensive have resistant hypertension,⁶¹ millions of patients are susceptible to potential harm from targeting a specific goal based solely on observational data. If lower is always better, the randomized trials outlined above should have had more positive outcomes.

It becomes problematic to assign a specific goal for all patients or even groups of patients. The targeting trials do not provide the answer. Based on the observational data it would be optimal to have a blood pressure less than 120/80 mm Hg. This is an observation, not a recommendation. Patients should be assessed on an individual basis, taking into consideration their starting blood pressure, age, medication burden (antihypertensive and otherwise), comorbidities, and ability to comply with a regimen. Given the available data, it is hard to be more specific. In the future it may be possible to identify specific blood pressure targets based on the patient’s genetic makeup, but today that is not possible. Even patients with lower initial blood pressure may benefit from therapy,^62,63 and some experts have advocated blood-pressure-lowering in all, irrespective of the baseline value.¹⁴

The first step in treating hypertension should be to avoid misclassification. Make sure the clinic blood pressure is measured correctly, using an appropriately sized cuff, positioning the patient properly, and following all the other recommendations.⁶⁴

However, the clinic blood pressure may not reflect true blood pressure load in up to one-third of all patients.⁶⁵ We recommend 24-hour ambulatory blood pressure monitoring⁶⁶ or home self-measurement, or both,⁶⁷ to better assess true blood pressure burden in several circumstances, including in patients with resistant hypertension (any patient who has not achieved acceptable clinic blood pressure on three or more antihypertensive medications including a diuretic or who requires four or more medications for adequate control), suspicion of white-coat hypertension (or effect), and any patient who has achieved acceptable clinic blood pressure but either has symptoms of hypotension or progressive end-organ damage.

Currently, we base therapy on out-of-office blood pressure (self-measured or by ambulatory monitoring) whenever there is a discrepancy with clinic blood pressure.

Whether therapy should be altered by other less traditional measures of blood pressure such as assessment of central aortic pressure by radial applanation tonometry,^68,69 or 24-hour ambulatory monitoring to assess nighttime blood pressures (specifically, “dipping”),⁷⁰ morning surge,⁷¹ or blood pressure variability^72,73 remains unclear and in need of randomized controlled trials.

In any patient requiring blood-pressure-lowering, we recommend lifestyle modifications.^1,2 These include exercise, weight loss, salt and alcohol restriction, evaluation for sleep apnea, and avoidance of medications known to elevate blood pressure such as nonsteroidal anti-inflammatory drugs and sympathomimetic decongestants.

Much needs to be learned

For the individual patient with unacceptably high blood pressure who is already taking multiple antihypertensive medications of different classes, it is unclear what to do. This type of patient with resistant hypertension would be an excellent candidate for a future targeting trial. Other cardiovascular risk factors should be appropriately addressed, including obesity, lipids, smoking, and poor glycemic control.⁷⁴ Each patient should be individually assessed with consideration of both global cardiovascular risk and quality-of-life issues.

Much still needs to be learned about the treatment of hypertension. The facts demonstrate that blood pressure is a strong modifiable risk factor of cardiovascular morbidity and mortality. Lowering it clearly produces benefits. It is unclear what treatment goals should be promulgated by official guidelines for large groups of patients. The resistant case remains a therapeutic dilemma with the potential for harm from overly aggressive treatment. The truly optimal level for an individual patient remains difficult to define. We anxiously await results of ongoing and future targeting trials.

CASE REVISITED

Regarding the initial case vignette, the patient is clearly not at her recommended goal blood pressure, especially given her high-risk status (diabetes mellitus and chronic kidney disease). Observational data support intensification of therapy, whereas targeting trials are essentially negative and indicate the potential for harm with overly aggressive treatment. Thus, we remain uncertain about what is correct or incorrect in terms of a targeted blood pressure, especially when applied to the individual patient.

Our approach would be to emphasize lifestyle modifications, to ensure accurate determination of her true blood pressure load (self-measurement at home or ambulatory blood pressure monitoring), to consider secondary causes of hypertension, and to educate the patient about the benefits and consequences of intensifying therapy with the aim of involving her in the decision.

A 50-year-old African American woman with type 2 diabetes mellitus, hypertension, hyperlipidemia, and chronic kidney disease presents for a follow-up visit. The patient had been treated with hydrochlorothiazide 25 mg/day and enalapril (Vasotec) 20 mg twice daily until 6 weeks ago. At that time her blood pressure was 160/85 mm Hg, and amlodipine (Norvasc) 10 mg/day was added to her regimen. Her other medications include glipizide (Glucotrol), metformin (Glucophage), lovastatin (Mevacor), fish oils, aspirin, calcium, and vitamin D. Her current blood pressure is 145/80 mm Hg; her serum creatinine level is 1.5 mg/dL, and her urine albumin-to-creatinine ratio is 180 mg/g.

In hypertensive patients who have diabetes or chronic kidney disease, guidelines¹ call for intensification of antihypertensive therapy to reach a goal blood pressure of less than 130/80 mm Hg. What data exist to support these guidelines? And what should the clinician do?

IS MORE-INTENSE THERAPY IN THE PATIENT’S BEST INTEREST?

Often, clinicians are faced with hypertensive patients whose blood pressure, despite treatment, is higher than the accepted goal. Often, these patients are elderly and are already taking multiple medications that are costly and have significant potential adverse effects. The dilemma is whether to try to reach a target blood pressure listed in a guideline (by increasing the dosage of the current drugs or by adding a drug of a different class) or to “do no harm,” accept the patient’s blood pressure, and keep the regimen the same.^1,2

The current goal blood pressure is less than 140/90 mm Hg for all but the very elderly, with more intense control recommended for patients at high risk, ie, those with diabetes mellitus, chronic kidney disease, or atherosclerotic cardiovascular disease.¹

While it appears to be in the patient’s best interests to follow such guidelines, review of available data indicates that this it not necessarily so, and may even be harmful.

OBSERVATIONAL DATA AND EARLY RANDOMIZED TRIALS

Many observational studies have found that the higher one’s blood pressure, the greater one’s risk of cardiovascular events and death. Indeed, meta-analyses of these trials, which involved more than 1.5 million people, demonstrate a strong, positive, log-linear relationship between blood pressure and the incidence of cardiovascular disease and death.^3–5

Further, there is no evidence of a threshold pressure below which the risk is not lower (ie, a “J-point”), starting with 115/75 mm Hg. A J-point may exist for diastolic blood pressure in elderly patients with isolated systolic hypertension⁶ and in patients with coronary artery disease.⁷ Otherwise, the observation is clear: the lower the blood pressure the better. For every 20 mm Hg lower systolic blood pressure or 10 mm Hg lower diastolic blood pressure, the risk of a cardiovascular event is about 50% less.^4,5

Observational analyses also show a strong, graded relationship between blood pressure and future end-stage renal disease.^8,9 Post hoc analyses indicate that chronic kidney disease progresses more slowly with lower achieved blood pressures, especially in those with higher degrees of proteinuria.^10–12

However, observational data do not prove cause and effect, nor do they guarantee similar results with treatment. This requires randomized controlled trials.

RANDOMIZED TRIALS OF HYPERTENSION TREATMENT

Initial trials were aimed at determining whether hypertension should even be treated. A 1997 meta-analysis of 18 such trials comparing either low-dose diuretic therapy, high-dose diuretic therapy, or beta-blocker therapy with placebo involved 48,000 patients who were followed for an average of 5 years.¹³ The rates of stroke and congestive heart failure were consistently reduced, although only low-dose diuretic therapy reduced the risk of coronary heart disease and death from any cause.

More recent trials enrolled people not considered hypertensive who were randomized to receive either active drugs or placebo, or no treatment. Other trials attempted to assess non-pressure-related effects of specific agents, using other antihypertensive agents in the control group. Still other randomized controlled trials compared one agent or agents with other agents while attempting to attain equivalent blood pressure between groups. Frequently, however, there was some blood pressure difference.

Meta-analyses of most of these trials conclude that the major benefit of antihypertensive therapy—reducing rates of cardiovascular morbidity and mortality—comes from a lower attained blood pressure, irrespective of which agent is used.^14–18 Exceptions exist, however. For example, specific drug classes are indicated after myocardial infarction, and in congestive heart failure and proteinuric chronic kidney disease.^10,19–21

16 TRIALS OF DIFFERENT BLOOD PRESSURE TARGETS

The overriding theme of these observational data is that a lower blood pressure, whether attained naturally or with treatment, is better than a higher one from both the cardiovascular and the renal perspective.

What remains unclear is what blood pressure should be aimed for in a particular patient or group of patients. Is it a specific pressure (eg, 140/90 mm Hg), or does the change from baseline count more? Should other factors such as age or comorbidity alter this number?

Several randomized controlled trials have addressed these questions by targeting different levels of blood pressure. We are aware of at least 16 such trials in adults, including 13 with renal or cardiovascular primary end points and three with surrogate primary end points.

An unavoidable design flaw of all of these trials is their unblinded nature. Consequently, nearly all of them carry a Jadad score (a measure of quality, based on randomization and blinding)²² of 3 on a scale of 5.

NINE TRIALS WITH RENAL PRIMARY END POINTS

African American Study of Kidney Disease and Hypertension (AASK)²³

Patients: 1,094 African Americans with presumed hypertensive renal disease and a measured glomerular filtration rate between 20 and 65 mL/min/1.73 m².

Randomized blood pressure goals. Mean arterial pressure 92 mm Hg or less vs 102 to 107 mm Hg.

Results. At 4 years, the two groups had average blood pressures of 128/78 and 141/85 mm Hg, respectively. The groups did not differ in the rates of the primary end points—ie, the rate of change in the measured glomerular filtration rate over time or the composite of a 50% reduction in glomerular filtration rate, the onset of end-stage renal disease, or death.

Comments. Several issues have been raised about the internal validity of this trial.

So-called hypertensive kidney disease in African Americans (as opposed to European Americans) may be a genetic disorder related to polymorphisms of one or more genes on chromosome 22q. Initial data implicated the MYH9 gene, which encodes non-muscle myosin heavy chain II.^24,25 More recent data implicate the nearby APOL1 gene encoding apolipoprotein L126 as more relevant. These polymorphisms have a much greater prevalence in African Americans and appear responsible for the higher risk of idiopathic focal segmental glomerulosclerosis and HIV-associated nephropathy in this population.^24–26 Therefore, in African Americans, hypertension may in fact be the result of the kidney disease and not its primary cause, which may explain why in this and other African American populations stricter control of blood pressure did not produce a renal benefit.^27,28

Also, there is the possibility of misclassification bias. A secondary analysis of data obtained by ambulatory monitoring showed that of the 377 participants whose blood pressure appeared to be under control when measured in the clinic, 70% actually had masked hypertension, ie, uncontrolled hypertension outside the clinic.²⁹ The real difference in blood pressure between groups may well have been different than that determined in the clinic.

In addition, a prespecified secondary analysis showed no difference in the rates of cardiovascular events and death between the groups.³⁰ However, the study was not designed to have the statistical power to detect a difference in cardiovascular events. Moreover fewer cardiovascular events occurred than expected, further reducing the study’s power to detect a difference.

Toto et al³¹

Toto et al reported similar results in an earlier trial in 87 hypertensive patients (77 randomized), predominantly African American, and similar concerns apply.

Lewis et al³²

Patients: 129 patients with type 1 diabetes.

Randomized blood pressure goals. A mean arterial pressure of either no higher than 92 mm Hg or 100 to 107 mm Hg.

Results. At 2 years, despite a difference of 6 mm Hg in mean arterial pressure, the glomerular filtration rate (measured) had declined by the same amount in the two groups. The study was underpowered for this end point. Patients in the group with the lower goal pressure were excreting significantly less protein than those in the other group, but they were received higher doses of an angiotensin-converting enzyme (ACE) inhibitor—in this case, ramipril (Altace).

The Appropriate Blood Pressure Control in Diabetes (ABCD) trials^33–35

Patients: 950 patients with type 2 diabetes mellitus and either normal or high blood pressure.

Randomized blood pressure goals. Either intensive or moderate therapy (see Table 1).

Results. At 5 years, creatinine clearance (estimated) had declined by the same amount in the two groups. However, fewer of the hypertensive patients had died in the intensive-therapy group.³⁴ Similarly, normotensive patients had less progression of albuminuria if treated intensively.³³

In the ABCD Part 2 with Valsartan (ABCD-2V) trial in normotensive patients,³⁵ therapy with valsartan (Diovan) did not affect creatinine clearance but did reduce albuminuria. However, 75% of the patients in the moderate-treatment group were untreated.

Schrier et al³⁶

Patients. 75 hypertensive patients with autosomal-dominant polycystic kidney disease and left ventricular hypertrophy.

Randomized blood pressure targets. Less than 120/80 mm Hg vs 135/85 to 140/90 mm Hg.

Results. After 7 years, despite a difference in average mean arterial pressure of 11 mm Hg between the groups (90 vs 101 mm Hg), there was no difference in the rate of decline of creatinine clearance. The left ventricular mass index decreased by 21% in the lower-target group and by 35% in the higher-target group (P < .01).

Modification of Diet in Renal Disease (MDRD) trial^37,38

Patients: 840 patients whose measured glomerular filtration rate was between 13 and 55 mL/min/1.73 m².

Randomized blood pressure targets. A target mean arterial pressure of less than 92 mm Hg vs less than 107 mm Hg.^11,37

Results. After 2.2 years, the mean difference in mean arterial pressure was 4.7 mm Hg. There was, however, no difference in the rate of decline in the glomerular filtration rate.

In a 6-year follow-up, significantly fewer patients in the lower-blood-pressure group reached the end point of end-stage renal disease or the combined end point of end-stage renal disease or death.³⁸ The rate of death, however, was nearly twice as high in the lower-blood-pressure group (10% vs 6%). The blood pressure and treatment during follow-up were not reported.

Comments. Internal validity is an issue, since the blood pressure and therapy during follow-up were unknown, and more patients received ACE inhibitors in the lower-blood-pressure group during the trial. Further, the higher death rate in the lower-blood-pressure group is worrisome.

Patients: 338 nondiabetic patients who had proteinuria and reduced creatinine clearance.

Treatment and blood pressure goals. All were treated with ramipril and randomized to intensive (< 130/80 mm Hg) vs standard control (diastolic blood pressure < 90 mm Hg) with therapy based on felodipine (Plendil).

Results. The study was terminated early because of futility. Despite a mean difference of 4.1 mm Hg systolic and 2.8 mm Hg diastolic, the groups did not differ in the rate of progression to end-stage renal disease (23% with intensive therapy vs 20% with standard therapy) or in the rate of decline of the measured glomerular filtration rate (0.22 vs 0.24 mL/min/1.73 m²/month).

Comment. The internal validity of this study can be questioned because of the low separation of achieved blood pressure and because of its early termination.

No benefit from a lower blood pressure goal in preserving kidney function

To summarize, these trials all showed no significant benefit from either targeting or achieving lower blood pressure in terms of slowing the decline of kidney function. Overall, they do not define a target and offer little support that a lower goal blood pressure is indicated with respect to the rate of loss of glomerular filtration rate in chronic kidney disease.

However, post hoc analysis of the MDRD trial indicates a statistical interaction between targeted blood pressure and degree of baseline proteinuria. At higher levels of proteinuria (≥ 1 g/day), the group with the lower blood pressure target had better outcomes.

In addition, long-term follow-up (mean of 12.2 years) of the AASK trial, including a 7-year cohort phase with nearly similar blood pressures in both groups, also indicated an interaction with targeted blood pressure and baseline proteinuria.⁴⁰ Although the overall analysis was negative, there was a significant reduction in the primary end point in the group originally assigned the low target when analysis was restricted to those in the highest tertile of proteinuria. These and other data¹⁰ suggest that patients with chronic kidney disease and proteinuria may represent a distinct subset of chronic kidney disease patients who benefit from more intensive blood-pressure-lowering. However, patients in the REIN-2 trial³⁴ and the macroalbuminuric patients in the ABCD hypertensive trial³⁵ did not benefit from a lower targeted blood pressure despite significant proteinuria.

FOUR TRIALS WITH CARDIOVASCULAR END POINTS

The Hypertension Optimal Treatment (HOT) trial⁴¹

Patients: 18,790 patients with diastolic blood pressure between 100 and 115 mm Hg.

Randomized blood pressure goals. Diastolic pressure of equal to or less than 80, 85, or 90 mm Hg.

Results. At an average of 3.8 years, the average blood pressures in the three groups were approximately 140/81, 141/83, and 144/85 mm Hg, respectively. There was no difference between the groups in the rate of the composite primary end point of all myocardial infarctions, all strokes, and cardiovascular death. Any conclusions from this trial were compromised by the small difference in achieved blood pressures between groups.

In the 1,501 patients with diabetes, the incidence of the primary end point was 50% lower with a goal of 80 mm Hg or less than with a goal of 90 mm Hg or less.

The UK Prospective Diabetes Study (UKPDS)^42,43

Patients: 1,148 hypertensive patients with type 2 diabetes mellitus.

Randomized blood pressure goals. Either “tight control” (aiming for < 150/85 mm Hg) or “less tight control” (aiming for < 180/105 mm Hg).

Results. At a median follow-up of 8.4 years, the attained blood pressures were 144/82 vs 154/87 mm Hg. The difference produced significant benefits, including a 24% lower rate of any diabetes-related end point, a 32% lower rate of death due to diabetes, and a nonsignificant 18% lower rate of total mortality—all co-primary end points.

The less-tight-control group had many patients with initial blood pressures below 180/105 mm Hg; hence, over 50% of patients received no antihypertensive therapy at the start of the trial. By the end of the trial 9 years later, 20% had still not been treated. This compares with only 5% of patients in the tight-control group who were not treated with antihypertensives throughout the trial. Therefore, this trial serves as better evidence for treating vs not treating, rather than defining a specific goal.

During a 10-year follow-up, blood pressure differences disappeared within 2 years.⁴³ There was no legacy effect, as the significant differences noted during the trial were no longer present 10 years later.

Action to Control Cardiovascular Risk in Diabetes (ACCORD)⁴⁴

Patients: 4,733 patients with type 2 diabetes.

Randomized blood pressure goals. Systolic blood pressure lower than either 120 or 140 mm Hg.

Results. At 4.7 years, despite a significant difference in mean systolic blood pressure of 14.2 mm Hg after the first year (119.3 vs 133.5 mm Hg), there was no difference in the primary end point of nonfatal myocardial infarction, nonfatal stroke, or cardiovascular death. There were fewer strokes in the lower-pressure group but no difference in myocardial infarctions, which were five times more common than strokes. Serious adverse events attributed to antihypertensive treatment occurred more frequently in the intensive-therapy group (3.3% vs 1.3%, P < .001).

Comment. There were fewer events than expected, possibly limiting the trial’s ability to detect a statistical difference. Compared with both the UKPDS and the diabetic population of HOT, ACCORD is much larger and more internally valid (unlike in UKPDS, nearly all patients in both groups were treated, and compared with HOT there was much greater separation of achieved pressure). It is more recent and better reflects current overall practice. It indicates that when specifically aiming for a target blood pressure, lower is not always better and comes at a price (more severe adverse events).

Patients: 4,418 patients, age 65 to 85 years, with a pretreatment systolic blood pressure above 160 mm Hg.

Randomized blood pressure goals. Systolic pressure either lower than 140 mm Hg or 140 to 160 mm Hg.

Results. At 2 years, despite a difference of 9.7/3.3 mm Hg, there was no difference in the primary end point (the combined incidence of cerebrovascular disease, cardiac and vascular disease, and renal failure). Fifty-four patients had died in the strict-treatment group and 42 in the mild-treatment group; the difference was not statistically significant.

Three other trials

Three other trials^46–48 had surrogate end points, but only one of them reported a composite cardiovascular secondary end point.⁴⁶ We will not discuss the other two.^47,48

Cardio-Sis. In the Studio Italiano Sugli Effetti Cardiovascolari del Controllo della Pressione Arteriosa Sistolica (Cardio-Sis) trial,⁴⁶ 1,111 people without diabetes with systolic pressure higher than 150 mm Hg were randomized to tight control (systolic pressure < 130 mm Hg) vs usual control (systolic pressure < 140 mm Hg) and followed for 2 years with electrocardiography to detect left ventricular hypertrophy.

At a median of 2 years, the systolic blood pressure had declined by an average of 3.8 mm Hg more in the tight-control group than in the usual-control group, and the diastolic pressure by an average of 1.5 mm Hg. There was significantly less left ventricular hypertrophy in the tight-control group. The incidence of the secondary end point of a composite of cardiovascular and renal events was also significantly lower. There was no difference individually in the rates of myocardial infarction, stroke, transient ischemic attack, admission for congestive heart failure, or death.

DISCUSSION: THE DILEMMA OF TREATING AN INDIVIDUAL PATIENT

These data illustrate the dilemma of treating an individual patient whose blood pressure is not at the currently accepted goal while on multiple antihypertensive medications. According to guidelines, therapy should be intensified in this situation. Observational data show a strong graded relationship between blood pressure and cardiovascular events and death, starting with a blood pressure of 115/75 mm Hg. The observational data relating blood pressure to kidney disease are similar. These data support the guideline recommendations that additional medications should be added to reach the promulgated target. Unfortunately, the targeting trials do not define a target, nor do they support the concept that lower is better.

Possible explanations for the negative results

Why does targeting a lower blood pressure not produce the benefit that the observational data lead us to expect?

One possibility is that blood pressure is merely a marker of cardiovascular risk, not a cause of it. This is unlikely, given the temporal relationship, reproducibility, and biologic plausibility that is supported by a very large body of experimental data. However, blood pressure is only one of multiple factors involved in the pathogenesis of vascular and renal disease, and perhaps better attention to other factors such as lipids and smoking may have made the targeting trials underpowered.

Another possibility is that these trials had such strict inclusion and exclusion criteria that they do not represent the general hypertensive population, reducing their external validity.⁴⁹ However, the trials generally enrolled populations at higher risk, in which end points were more likely to occur. This would have enhanced the chance to show a positive effect rather than mask it.

It is possible that antihypertensive medications themselves have unwanted side effects that offset their potential benefit. Medication-related side effects could directly contribute to vascular disease despite their beneficial effect of lowering pressure. There could also be reduced tissue perfusion due to lower blood pressure per se in the face of a diseased vasculature, with the lower pressure directly contributing to organ dysfunction.

Finally, these trials measured brachial pressures to monitor blood pressure. Brachial pressure does not always correlate with central aortic pressure, which is probably a better marker of the overall pressure burden.⁵⁰ It is possible that in these targeting trials, the peripheral blood pressure did not reflect the true central blood pressure and, therefore, significant separation of blood pressures may not have actually occurred.

Targeted vs achieved blood pressures: Analogies with other markers

This contradiction is not an exceptional circumstance in medicine.

For example, in chronic kidney disease, a graded observational relationship exists between decreasing levels of hemoglobin and various adverse outcomes.^51–53 However, targeting a more normal level of hemoglobin compared with a lower one has been shown to be detrimental.^54–57 This implies either that anemia is merely a marker of higher risk or, more likely, that the actual measures used to raise the hemoglobin to higher levels are the culprit. Notably, although targeting a higher hemoglobin concentration vs a lower one was detrimental, achieving a higher hemoglobin was beneficial within each targeted group.^54,58

Another example of harm caused by targeting goals based on observational data is tight glucose control, both acutely in the critically ill⁵⁹ and chronically in patients with type 2 diabetes.⁶⁰ In both cases higher mortality rates ensued.

The same concept may apply to lowering blood pressure. While achieving a lower blood pressure may be more beneficial, targeting a specific goal may be harmful. Given that perhaps 20% of those labeled as hypertensive have resistant hypertension,⁶¹ millions of patients are susceptible to potential harm from targeting a specific goal based solely on observational data. If lower is always better, the randomized trials outlined above should have had more positive outcomes.

It becomes problematic to assign a specific goal for all patients or even groups of patients. The targeting trials do not provide the answer. Based on the observational data it would be optimal to have a blood pressure less than 120/80 mm Hg. This is an observation, not a recommendation. Patients should be assessed on an individual basis, taking into consideration their starting blood pressure, age, medication burden (antihypertensive and otherwise), comorbidities, and ability to comply with a regimen. Given the available data, it is hard to be more specific. In the future it may be possible to identify specific blood pressure targets based on the patient’s genetic makeup, but today that is not possible. Even patients with lower initial blood pressure may benefit from therapy,^62,63 and some experts have advocated blood-pressure-lowering in all, irrespective of the baseline value.¹⁴

The first step in treating hypertension should be to avoid misclassification. Make sure the clinic blood pressure is measured correctly, using an appropriately sized cuff, positioning the patient properly, and following all the other recommendations.⁶⁴

However, the clinic blood pressure may not reflect true blood pressure load in up to one-third of all patients.⁶⁵ We recommend 24-hour ambulatory blood pressure monitoring⁶⁶ or home self-measurement, or both,⁶⁷ to better assess true blood pressure burden in several circumstances, including in patients with resistant hypertension (any patient who has not achieved acceptable clinic blood pressure on three or more antihypertensive medications including a diuretic or who requires four or more medications for adequate control), suspicion of white-coat hypertension (or effect), and any patient who has achieved acceptable clinic blood pressure but either has symptoms of hypotension or progressive end-organ damage.

Currently, we base therapy on out-of-office blood pressure (self-measured or by ambulatory monitoring) whenever there is a discrepancy with clinic blood pressure.

Whether therapy should be altered by other less traditional measures of blood pressure such as assessment of central aortic pressure by radial applanation tonometry,^68,69 or 24-hour ambulatory monitoring to assess nighttime blood pressures (specifically, “dipping”),⁷⁰ morning surge,⁷¹ or blood pressure variability^72,73 remains unclear and in need of randomized controlled trials.

In any patient requiring blood-pressure-lowering, we recommend lifestyle modifications.^1,2 These include exercise, weight loss, salt and alcohol restriction, evaluation for sleep apnea, and avoidance of medications known to elevate blood pressure such as nonsteroidal anti-inflammatory drugs and sympathomimetic decongestants.

Much needs to be learned

For the individual patient with unacceptably high blood pressure who is already taking multiple antihypertensive medications of different classes, it is unclear what to do. This type of patient with resistant hypertension would be an excellent candidate for a future targeting trial. Other cardiovascular risk factors should be appropriately addressed, including obesity, lipids, smoking, and poor glycemic control.⁷⁴ Each patient should be individually assessed with consideration of both global cardiovascular risk and quality-of-life issues.

Much still needs to be learned about the treatment of hypertension. The facts demonstrate that blood pressure is a strong modifiable risk factor of cardiovascular morbidity and mortality. Lowering it clearly produces benefits. It is unclear what treatment goals should be promulgated by official guidelines for large groups of patients. The resistant case remains a therapeutic dilemma with the potential for harm from overly aggressive treatment. The truly optimal level for an individual patient remains difficult to define. We anxiously await results of ongoing and future targeting trials.

CASE REVISITED

Regarding the initial case vignette, the patient is clearly not at her recommended goal blood pressure, especially given her high-risk status (diabetes mellitus and chronic kidney disease). Observational data support intensification of therapy, whereas targeting trials are essentially negative and indicate the potential for harm with overly aggressive treatment. Thus, we remain uncertain about what is correct or incorrect in terms of a targeted blood pressure, especially when applied to the individual patient.

Our approach would be to emphasize lifestyle modifications, to ensure accurate determination of her true blood pressure load (self-measurement at home or ambulatory blood pressure monitoring), to consider secondary causes of hypertension, and to educate the patient about the benefits and consequences of intensifying therapy with the aim of involving her in the decision.

A 50-year-old African American woman with type 2 diabetes mellitus, hypertension, hyperlipidemia, and chronic kidney disease presents for a follow-up visit. The patient had been treated with hydrochlorothiazide 25 mg/day and enalapril (Vasotec) 20 mg twice daily until 6 weeks ago. At that time her blood pressure was 160/85 mm Hg, and amlodipine (Norvasc) 10 mg/day was added to her regimen. Her other medications include glipizide (Glucotrol), metformin (Glucophage), lovastatin (Mevacor), fish oils, aspirin, calcium, and vitamin D. Her current blood pressure is 145/80 mm Hg; her serum creatinine level is 1.5 mg/dL, and her urine albumin-to-creatinine ratio is 180 mg/g.

In hypertensive patients who have diabetes or chronic kidney disease, guidelines¹ call for intensification of antihypertensive therapy to reach a goal blood pressure of less than 130/80 mm Hg. What data exist to support these guidelines? And what should the clinician do?

IS MORE-INTENSE THERAPY IN THE PATIENT’S BEST INTEREST?

Often, clinicians are faced with hypertensive patients whose blood pressure, despite treatment, is higher than the accepted goal. Often, these patients are elderly and are already taking multiple medications that are costly and have significant potential adverse effects. The dilemma is whether to try to reach a target blood pressure listed in a guideline (by increasing the dosage of the current drugs or by adding a drug of a different class) or to “do no harm,” accept the patient’s blood pressure, and keep the regimen the same.^1,2

The current goal blood pressure is less than 140/90 mm Hg for all but the very elderly, with more intense control recommended for patients at high risk, ie, those with diabetes mellitus, chronic kidney disease, or atherosclerotic cardiovascular disease.¹

While it appears to be in the patient’s best interests to follow such guidelines, review of available data indicates that this it not necessarily so, and may even be harmful.

OBSERVATIONAL DATA AND EARLY RANDOMIZED TRIALS

Many observational studies have found that the higher one’s blood pressure, the greater one’s risk of cardiovascular events and death. Indeed, meta-analyses of these trials, which involved more than 1.5 million people, demonstrate a strong, positive, log-linear relationship between blood pressure and the incidence of cardiovascular disease and death.^3–5

Further, there is no evidence of a threshold pressure below which the risk is not lower (ie, a “J-point”), starting with 115/75 mm Hg. A J-point may exist for diastolic blood pressure in elderly patients with isolated systolic hypertension⁶ and in patients with coronary artery disease.⁷ Otherwise, the observation is clear: the lower the blood pressure the better. For every 20 mm Hg lower systolic blood pressure or 10 mm Hg lower diastolic blood pressure, the risk of a cardiovascular event is about 50% less.^4,5

Observational analyses also show a strong, graded relationship between blood pressure and future end-stage renal disease.^8,9 Post hoc analyses indicate that chronic kidney disease progresses more slowly with lower achieved blood pressures, especially in those with higher degrees of proteinuria.^10–12

However, observational data do not prove cause and effect, nor do they guarantee similar results with treatment. This requires randomized controlled trials.

RANDOMIZED TRIALS OF HYPERTENSION TREATMENT

Initial trials were aimed at determining whether hypertension should even be treated. A 1997 meta-analysis of 18 such trials comparing either low-dose diuretic therapy, high-dose diuretic therapy, or beta-blocker therapy with placebo involved 48,000 patients who were followed for an average of 5 years.¹³ The rates of stroke and congestive heart failure were consistently reduced, although only low-dose diuretic therapy reduced the risk of coronary heart disease and death from any cause.

More recent trials enrolled people not considered hypertensive who were randomized to receive either active drugs or placebo, or no treatment. Other trials attempted to assess non-pressure-related effects of specific agents, using other antihypertensive agents in the control group. Still other randomized controlled trials compared one agent or agents with other agents while attempting to attain equivalent blood pressure between groups. Frequently, however, there was some blood pressure difference.

Meta-analyses of most of these trials conclude that the major benefit of antihypertensive therapy—reducing rates of cardiovascular morbidity and mortality—comes from a lower attained blood pressure, irrespective of which agent is used.^14–18 Exceptions exist, however. For example, specific drug classes are indicated after myocardial infarction, and in congestive heart failure and proteinuric chronic kidney disease.^10,19–21

16 TRIALS OF DIFFERENT BLOOD PRESSURE TARGETS

The overriding theme of these observational data is that a lower blood pressure, whether attained naturally or with treatment, is better than a higher one from both the cardiovascular and the renal perspective.

What remains unclear is what blood pressure should be aimed for in a particular patient or group of patients. Is it a specific pressure (eg, 140/90 mm Hg), or does the change from baseline count more? Should other factors such as age or comorbidity alter this number?

Several randomized controlled trials have addressed these questions by targeting different levels of blood pressure. We are aware of at least 16 such trials in adults, including 13 with renal or cardiovascular primary end points and three with surrogate primary end points.

An unavoidable design flaw of all of these trials is their unblinded nature. Consequently, nearly all of them carry a Jadad score (a measure of quality, based on randomization and blinding)²² of 3 on a scale of 5.

NINE TRIALS WITH RENAL PRIMARY END POINTS

African American Study of Kidney Disease and Hypertension (AASK)²³

Patients: 1,094 African Americans with presumed hypertensive renal disease and a measured glomerular filtration rate between 20 and 65 mL/min/1.73 m².

Randomized blood pressure goals. Mean arterial pressure 92 mm Hg or less vs 102 to 107 mm Hg.

Results. At 4 years, the two groups had average blood pressures of 128/78 and 141/85 mm Hg, respectively. The groups did not differ in the rates of the primary end points—ie, the rate of change in the measured glomerular filtration rate over time or the composite of a 50% reduction in glomerular filtration rate, the onset of end-stage renal disease, or death.

Comments. Several issues have been raised about the internal validity of this trial.

So-called hypertensive kidney disease in African Americans (as opposed to European Americans) may be a genetic disorder related to polymorphisms of one or more genes on chromosome 22q. Initial data implicated the MYH9 gene, which encodes non-muscle myosin heavy chain II.^24,25 More recent data implicate the nearby APOL1 gene encoding apolipoprotein L126 as more relevant. These polymorphisms have a much greater prevalence in African Americans and appear responsible for the higher risk of idiopathic focal segmental glomerulosclerosis and HIV-associated nephropathy in this population.^24–26 Therefore, in African Americans, hypertension may in fact be the result of the kidney disease and not its primary cause, which may explain why in this and other African American populations stricter control of blood pressure did not produce a renal benefit.^27,28

Also, there is the possibility of misclassification bias. A secondary analysis of data obtained by ambulatory monitoring showed that of the 377 participants whose blood pressure appeared to be under control when measured in the clinic, 70% actually had masked hypertension, ie, uncontrolled hypertension outside the clinic.²⁹ The real difference in blood pressure between groups may well have been different than that determined in the clinic.

In addition, a prespecified secondary analysis showed no difference in the rates of cardiovascular events and death between the groups.³⁰ However, the study was not designed to have the statistical power to detect a difference in cardiovascular events. Moreover fewer cardiovascular events occurred than expected, further reducing the study’s power to detect a difference.

Toto et al³¹

Toto et al reported similar results in an earlier trial in 87 hypertensive patients (77 randomized), predominantly African American, and similar concerns apply.

Lewis et al³²

Patients: 129 patients with type 1 diabetes.

Randomized blood pressure goals. A mean arterial pressure of either no higher than 92 mm Hg or 100 to 107 mm Hg.

Results. At 2 years, despite a difference of 6 mm Hg in mean arterial pressure, the glomerular filtration rate (measured) had declined by the same amount in the two groups. The study was underpowered for this end point. Patients in the group with the lower goal pressure were excreting significantly less protein than those in the other group, but they were received higher doses of an angiotensin-converting enzyme (ACE) inhibitor—in this case, ramipril (Altace).

The Appropriate Blood Pressure Control in Diabetes (ABCD) trials^33–35

Patients: 950 patients with type 2 diabetes mellitus and either normal or high blood pressure.

Randomized blood pressure goals. Either intensive or moderate therapy (see Table 1).

Results. At 5 years, creatinine clearance (estimated) had declined by the same amount in the two groups. However, fewer of the hypertensive patients had died in the intensive-therapy group.³⁴ Similarly, normotensive patients had less progression of albuminuria if treated intensively.³³

In the ABCD Part 2 with Valsartan (ABCD-2V) trial in normotensive patients,³⁵ therapy with valsartan (Diovan) did not affect creatinine clearance but did reduce albuminuria. However, 75% of the patients in the moderate-treatment group were untreated.

Schrier et al³⁶

Patients. 75 hypertensive patients with autosomal-dominant polycystic kidney disease and left ventricular hypertrophy.

Randomized blood pressure targets. Less than 120/80 mm Hg vs 135/85 to 140/90 mm Hg.

Results. After 7 years, despite a difference in average mean arterial pressure of 11 mm Hg between the groups (90 vs 101 mm Hg), there was no difference in the rate of decline of creatinine clearance. The left ventricular mass index decreased by 21% in the lower-target group and by 35% in the higher-target group (P < .01).

Modification of Diet in Renal Disease (MDRD) trial^37,38

Patients: 840 patients whose measured glomerular filtration rate was between 13 and 55 mL/min/1.73 m².

Randomized blood pressure targets. A target mean arterial pressure of less than 92 mm Hg vs less than 107 mm Hg.^11,37

Results. After 2.2 years, the mean difference in mean arterial pressure was 4.7 mm Hg. There was, however, no difference in the rate of decline in the glomerular filtration rate.

In a 6-year follow-up, significantly fewer patients in the lower-blood-pressure group reached the end point of end-stage renal disease or the combined end point of end-stage renal disease or death.³⁸ The rate of death, however, was nearly twice as high in the lower-blood-pressure group (10% vs 6%). The blood pressure and treatment during follow-up were not reported.

Comments. Internal validity is an issue, since the blood pressure and therapy during follow-up were unknown, and more patients received ACE inhibitors in the lower-blood-pressure group during the trial. Further, the higher death rate in the lower-blood-pressure group is worrisome.

Patients: 338 nondiabetic patients who had proteinuria and reduced creatinine clearance.

Treatment and blood pressure goals. All were treated with ramipril and randomized to intensive (< 130/80 mm Hg) vs standard control (diastolic blood pressure < 90 mm Hg) with therapy based on felodipine (Plendil).

Results. The study was terminated early because of futility. Despite a mean difference of 4.1 mm Hg systolic and 2.8 mm Hg diastolic, the groups did not differ in the rate of progression to end-stage renal disease (23% with intensive therapy vs 20% with standard therapy) or in the rate of decline of the measured glomerular filtration rate (0.22 vs 0.24 mL/min/1.73 m²/month).

Comment. The internal validity of this study can be questioned because of the low separation of achieved blood pressure and because of its early termination.

No benefit from a lower blood pressure goal in preserving kidney function

To summarize, these trials all showed no significant benefit from either targeting or achieving lower blood pressure in terms of slowing the decline of kidney function. Overall, they do not define a target and offer little support that a lower goal blood pressure is indicated with respect to the rate of loss of glomerular filtration rate in chronic kidney disease.

However, post hoc analysis of the MDRD trial indicates a statistical interaction between targeted blood pressure and degree of baseline proteinuria. At higher levels of proteinuria (≥ 1 g/day), the group with the lower blood pressure target had better outcomes.

In addition, long-term follow-up (mean of 12.2 years) of the AASK trial, including a 7-year cohort phase with nearly similar blood pressures in both groups, also indicated an interaction with targeted blood pressure and baseline proteinuria.⁴⁰ Although the overall analysis was negative, there was a significant reduction in the primary end point in the group originally assigned the low target when analysis was restricted to those in the highest tertile of proteinuria. These and other data¹⁰ suggest that patients with chronic kidney disease and proteinuria may represent a distinct subset of chronic kidney disease patients who benefit from more intensive blood-pressure-lowering. However, patients in the REIN-2 trial³⁴ and the macroalbuminuric patients in the ABCD hypertensive trial³⁵ did not benefit from a lower targeted blood pressure despite significant proteinuria.

FOUR TRIALS WITH CARDIOVASCULAR END POINTS

The Hypertension Optimal Treatment (HOT) trial⁴¹

Patients: 18,790 patients with diastolic blood pressure between 100 and 115 mm Hg.

Randomized blood pressure goals. Diastolic pressure of equal to or less than 80, 85, or 90 mm Hg.

Results. At an average of 3.8 years, the average blood pressures in the three groups were approximately 140/81, 141/83, and 144/85 mm Hg, respectively. There was no difference between the groups in the rate of the composite primary end point of all myocardial infarctions, all strokes, and cardiovascular death. Any conclusions from this trial were compromised by the small difference in achieved blood pressures between groups.

In the 1,501 patients with diabetes, the incidence of the primary end point was 50% lower with a goal of 80 mm Hg or less than with a goal of 90 mm Hg or less.

The UK Prospective Diabetes Study (UKPDS)^42,43

Patients: 1,148 hypertensive patients with type 2 diabetes mellitus.

Randomized blood pressure goals. Either “tight control” (aiming for < 150/85 mm Hg) or “less tight control” (aiming for < 180/105 mm Hg).

Results. At a median follow-up of 8.4 years, the attained blood pressures were 144/82 vs 154/87 mm Hg. The difference produced significant benefits, including a 24% lower rate of any diabetes-related end point, a 32% lower rate of death due to diabetes, and a nonsignificant 18% lower rate of total mortality—all co-primary end points.

The less-tight-control group had many patients with initial blood pressures below 180/105 mm Hg; hence, over 50% of patients received no antihypertensive therapy at the start of the trial. By the end of the trial 9 years later, 20% had still not been treated. This compares with only 5% of patients in the tight-control group who were not treated with antihypertensives throughout the trial. Therefore, this trial serves as better evidence for treating vs not treating, rather than defining a specific goal.

During a 10-year follow-up, blood pressure differences disappeared within 2 years.⁴³ There was no legacy effect, as the significant differences noted during the trial were no longer present 10 years later.

Action to Control Cardiovascular Risk in Diabetes (ACCORD)⁴⁴

Patients: 4,733 patients with type 2 diabetes.

Randomized blood pressure goals. Systolic blood pressure lower than either 120 or 140 mm Hg.

Results. At 4.7 years, despite a significant difference in mean systolic blood pressure of 14.2 mm Hg after the first year (119.3 vs 133.5 mm Hg), there was no difference in the primary end point of nonfatal myocardial infarction, nonfatal stroke, or cardiovascular death. There were fewer strokes in the lower-pressure group but no difference in myocardial infarctions, which were five times more common than strokes. Serious adverse events attributed to antihypertensive treatment occurred more frequently in the intensive-therapy group (3.3% vs 1.3%, P < .001).

Comment. There were fewer events than expected, possibly limiting the trial’s ability to detect a statistical difference. Compared with both the UKPDS and the diabetic population of HOT, ACCORD is much larger and more internally valid (unlike in UKPDS, nearly all patients in both groups were treated, and compared with HOT there was much greater separation of achieved pressure). It is more recent and better reflects current overall practice. It indicates that when specifically aiming for a target blood pressure, lower is not always better and comes at a price (more severe adverse events).

Patients: 4,418 patients, age 65 to 85 years, with a pretreatment systolic blood pressure above 160 mm Hg.

Randomized blood pressure goals. Systolic pressure either lower than 140 mm Hg or 140 to 160 mm Hg.

Results. At 2 years, despite a difference of 9.7/3.3 mm Hg, there was no difference in the primary end point (the combined incidence of cerebrovascular disease, cardiac and vascular disease, and renal failure). Fifty-four patients had died in the strict-treatment group and 42 in the mild-treatment group; the difference was not statistically significant.

Three other trials

Three other trials^46–48 had surrogate end points, but only one of them reported a composite cardiovascular secondary end point.⁴⁶ We will not discuss the other two.^47,48

Cardio-Sis. In the Studio Italiano Sugli Effetti Cardiovascolari del Controllo della Pressione Arteriosa Sistolica (Cardio-Sis) trial,⁴⁶ 1,111 people without diabetes with systolic pressure higher than 150 mm Hg were randomized to tight control (systolic pressure < 130 mm Hg) vs usual control (systolic pressure < 140 mm Hg) and followed for 2 years with electrocardiography to detect left ventricular hypertrophy.

At a median of 2 years, the systolic blood pressure had declined by an average of 3.8 mm Hg more in the tight-control group than in the usual-control group, and the diastolic pressure by an average of 1.5 mm Hg. There was significantly less left ventricular hypertrophy in the tight-control group. The incidence of the secondary end point of a composite of cardiovascular and renal events was also significantly lower. There was no difference individually in the rates of myocardial infarction, stroke, transient ischemic attack, admission for congestive heart failure, or death.

DISCUSSION: THE DILEMMA OF TREATING AN INDIVIDUAL PATIENT

These data illustrate the dilemma of treating an individual patient whose blood pressure is not at the currently accepted goal while on multiple antihypertensive medications. According to guidelines, therapy should be intensified in this situation. Observational data show a strong graded relationship between blood pressure and cardiovascular events and death, starting with a blood pressure of 115/75 mm Hg. The observational data relating blood pressure to kidney disease are similar. These data support the guideline recommendations that additional medications should be added to reach the promulgated target. Unfortunately, the targeting trials do not define a target, nor do they support the concept that lower is better.

Possible explanations for the negative results

Why does targeting a lower blood pressure not produce the benefit that the observational data lead us to expect?

One possibility is that blood pressure is merely a marker of cardiovascular risk, not a cause of it. This is unlikely, given the temporal relationship, reproducibility, and biologic plausibility that is supported by a very large body of experimental data. However, blood pressure is only one of multiple factors involved in the pathogenesis of vascular and renal disease, and perhaps better attention to other factors such as lipids and smoking may have made the targeting trials underpowered.

Another possibility is that these trials had such strict inclusion and exclusion criteria that they do not represent the general hypertensive population, reducing their external validity.⁴⁹ However, the trials generally enrolled populations at higher risk, in which end points were more likely to occur. This would have enhanced the chance to show a positive effect rather than mask it.

It is possible that antihypertensive medications themselves have unwanted side effects that offset their potential benefit. Medication-related side effects could directly contribute to vascular disease despite their beneficial effect of lowering pressure. There could also be reduced tissue perfusion due to lower blood pressure per se in the face of a diseased vasculature, with the lower pressure directly contributing to organ dysfunction.

Finally, these trials measured brachial pressures to monitor blood pressure. Brachial pressure does not always correlate with central aortic pressure, which is probably a better marker of the overall pressure burden.⁵⁰ It is possible that in these targeting trials, the peripheral blood pressure did not reflect the true central blood pressure and, therefore, significant separation of blood pressures may not have actually occurred.

Targeted vs achieved blood pressures: Analogies with other markers

This contradiction is not an exceptional circumstance in medicine.

For example, in chronic kidney disease, a graded observational relationship exists between decreasing levels of hemoglobin and various adverse outcomes.^51–53 However, targeting a more normal level of hemoglobin compared with a lower one has been shown to be detrimental.^54–57 This implies either that anemia is merely a marker of higher risk or, more likely, that the actual measures used to raise the hemoglobin to higher levels are the culprit. Notably, although targeting a higher hemoglobin concentration vs a lower one was detrimental, achieving a higher hemoglobin was beneficial within each targeted group.^54,58

Another example of harm caused by targeting goals based on observational data is tight glucose control, both acutely in the critically ill⁵⁹ and chronically in patients with type 2 diabetes.⁶⁰ In both cases higher mortality rates ensued.

The same concept may apply to lowering blood pressure. While achieving a lower blood pressure may be more beneficial, targeting a specific goal may be harmful. Given that perhaps 20% of those labeled as hypertensive have resistant hypertension,⁶¹ millions of patients are susceptible to potential harm from targeting a specific goal based solely on observational data. If lower is always better, the randomized trials outlined above should have had more positive outcomes.

It becomes problematic to assign a specific goal for all patients or even groups of patients. The targeting trials do not provide the answer. Based on the observational data it would be optimal to have a blood pressure less than 120/80 mm Hg. This is an observation, not a recommendation. Patients should be assessed on an individual basis, taking into consideration their starting blood pressure, age, medication burden (antihypertensive and otherwise), comorbidities, and ability to comply with a regimen. Given the available data, it is hard to be more specific. In the future it may be possible to identify specific blood pressure targets based on the patient’s genetic makeup, but today that is not possible. Even patients with lower initial blood pressure may benefit from therapy,^62,63 and some experts have advocated blood-pressure-lowering in all, irrespective of the baseline value.¹⁴

The first step in treating hypertension should be to avoid misclassification. Make sure the clinic blood pressure is measured correctly, using an appropriately sized cuff, positioning the patient properly, and following all the other recommendations.⁶⁴

However, the clinic blood pressure may not reflect true blood pressure load in up to one-third of all patients.⁶⁵ We recommend 24-hour ambulatory blood pressure monitoring⁶⁶ or home self-measurement, or both,⁶⁷ to better assess true blood pressure burden in several circumstances, including in patients with resistant hypertension (any patient who has not achieved acceptable clinic blood pressure on three or more antihypertensive medications including a diuretic or who requires four or more medications for adequate control), suspicion of white-coat hypertension (or effect), and any patient who has achieved acceptable clinic blood pressure but either has symptoms of hypotension or progressive end-organ damage.

Currently, we base therapy on out-of-office blood pressure (self-measured or by ambulatory monitoring) whenever there is a discrepancy with clinic blood pressure.

Whether therapy should be altered by other less traditional measures of blood pressure such as assessment of central aortic pressure by radial applanation tonometry,^68,69 or 24-hour ambulatory monitoring to assess nighttime blood pressures (specifically, “dipping”),⁷⁰ morning surge,⁷¹ or blood pressure variability^72,73 remains unclear and in need of randomized controlled trials.

In any patient requiring blood-pressure-lowering, we recommend lifestyle modifications.^1,2 These include exercise, weight loss, salt and alcohol restriction, evaluation for sleep apnea, and avoidance of medications known to elevate blood pressure such as nonsteroidal anti-inflammatory drugs and sympathomimetic decongestants.

Much needs to be learned

For the individual patient with unacceptably high blood pressure who is already taking multiple antihypertensive medications of different classes, it is unclear what to do. This type of patient with resistant hypertension would be an excellent candidate for a future targeting trial. Other cardiovascular risk factors should be appropriately addressed, including obesity, lipids, smoking, and poor glycemic control.⁷⁴ Each patient should be individually assessed with consideration of both global cardiovascular risk and quality-of-life issues.

Much still needs to be learned about the treatment of hypertension. The facts demonstrate that blood pressure is a strong modifiable risk factor of cardiovascular morbidity and mortality. Lowering it clearly produces benefits. It is unclear what treatment goals should be promulgated by official guidelines for large groups of patients. The resistant case remains a therapeutic dilemma with the potential for harm from overly aggressive treatment. The truly optimal level for an individual patient remains difficult to define. We anxiously await results of ongoing and future targeting trials.

CASE REVISITED

Regarding the initial case vignette, the patient is clearly not at her recommended goal blood pressure, especially given her high-risk status (diabetes mellitus and chronic kidney disease). Observational data support intensification of therapy, whereas targeting trials are essentially negative and indicate the potential for harm with overly aggressive treatment. Thus, we remain uncertain about what is correct or incorrect in terms of a targeted blood pressure, especially when applied to the individual patient.

Our approach would be to emphasize lifestyle modifications, to ensure accurate determination of her true blood pressure load (self-measurement at home or ambulatory blood pressure monitoring), to consider secondary causes of hypertension, and to educate the patient about the benefits and consequences of intensifying therapy with the aim of involving her in the decision.

With early treatment of human immunodeficiency virus (HIV) infection, we can now expect patients to live a much longer life and, in some situations, have a near-normal lifespan.¹ Unfortunately, in screening for HIV infection, the United States lags behind many regions of the world, and infection is often not diagnosed until patients present with advanced disease, ie, the acquired immunodeficiency syndrome (AIDS). In this country there is a critical need to make HIV screening a routine part of medical care in all health settings in order to give patients their best chance for a healthy life, to prevent mother-to-child transmission, and to reduce the spread of HIV in the community.

HIV infection meets the criteria that justify routine screening, as laid out by the World Health Organization²:

• It is a serious health disorder that can be detected before symptoms develop
• Treatment is more beneficial if begun before symptoms develop
• Reliable, inexpensive, and acceptable screening tests exist
• The costs of screening are reasonable in relation to the anticipated benefits.

This article will review the epidemiology of the HIV epidemic, present the benefits of early treatment, and make the case for widely expanding screening for HIV infection in the US health care system.

HIV INFECTION CONTINUES TO BE A LARGE BURDEN

In 2008, an estimated 33.4 million people worldwide were HIV-positive. The vast majority of infected people—more than 22 million—live in sub-Saharan Africa.³

The United States has approximately 1.2 million cases.⁴ Although this is a small proportion of cases worldwide, it still represents a significant health care burden. In this country, the number of AIDS cases peaked in 1993, and the rate of deaths from AIDS began to decrease over the ensuing years as adequate therapy for HIV was developed. Standard therapy then and now consists of at least three drugs from two different classes.

Unfortunately, we have made little progress on the incidence of this disease. The estimated number of new HIV infections in the United States in 2008 was 56,000 and had remained about the same over the previous 15 years.^5,6 Because of improved rates of survival, the prevalence has risen steadily since the mid-1990s to the current estimate of 1.2 million persons living with HIV/AIDS in the US.

About 25% of people infected with HIV are unaware of it. This group accounts for more than half of all new infections annually, which highlights the importance of enhanced screening. Once people know they are infected, they tend to change their behavior and are less likely to spread the disease.⁷

HIV disproportionately affects minority populations and gay men

Cases of HIV infection are reported among all age groups, although most patients tend to have been infected as young adults. Currently, the largest age group living with HIV is middle-aged. As this cohort grows older, an increasing burden of comorbidities due to aging can be expected. In 5 years, about half of the people with HIV in this country are expected to be 50 years of age or older. Although survival rates have steadily increased due to better treatment, survival tends to be shorter for older people newly diagnosed with HIV.

Worldwide, about an equal number of men and women are infected with HIV, but in the United States infected men outnumber women. In this country, about half the cases of HIV transmission among adults are by male-to-male sexual contact, about 30% are by high-risk heterosexual contact (ie, with a partner known to be HIV-infected or at high risk for being infected), and about 10% are by injection drug use.

In the United States, AIDS is predominantly and disproportionately a disease of minorities and those who live in poverty. African Americans account for the largest number of cases, followed by whites and then by Hispanics. Combined, African Americans and Hispanics account for two-thirds to three-fourths of all new cases, although they make up less than one-fourth of the US population. The incidence rate is nearly 137 per 100,000 for African Americans, 56 per 100,000 for Hispanics, and 19 per 100,000 for whites. The incidence is highest in New York and in the southeast, the geographic areas where the greatest number of minorities and people living in poverty reside. These groups also often lack access to health care.

HIV TREATMENT IS MORE EFFECTIVE IF STARTED EARLY

Treatment guidelines from the US Department of Health and Human Services (DHHS) have changed over the years. When effective medications were first introduced in the 1990s, the trend was to treat everyone as soon as they were diagnosed. As the burden of therapy began to unfold (side effects, cost, adherence, and drug resistance), the consensus was to wait until the CD4 T-cell count dropped to a lower level. As the medications have improved and have become better tolerated, the pendulum has swung back to treating earlier in the course of the disease. Currently, the DHHS recommends that therapy be started at CD4 counts of 350 cells/mL or lower (level of evidence: A1).⁸ It also recommends therapy for CD4 counts between 350 and 500 cells/mL, but the level of evidence is lower.⁸

The CD4 T cell is the prime target of the HIV virus and also an important marker of the health of the immune system. The lower the CD4 count at the start of therapy, the more challenging it is to normalize.⁹ If HIV infection is diagnosed early and therapy is started early, the likelihood is higher of normalizing the CD4 count and preserving immune function.

Progress is being made toward diagnosing HIV earlier. The CD4 count at presentation is increasing, but patients in the United States still present for care later than in other countries. In 1997, the median CD4 count at presentation was 234 cells/mL; in 2007, it was 327 (normal is about 550–1,000). Although this is a significant improvement, more than 50% of patients still have fewer than 350 cells/mL at presentation, which is the current threshold for beginning therapy, according to the most recent guidelines.¹⁰

Before triple therapy was available, almost all HIV-infected patients died of AIDS-related diseases. Now, about half of treated HIV-infected patients in Europe and North America die of other causes.¹¹ However, many diseases not previously attributed to AIDS are now also known to be exacerbated by HIV infection.

The cumulative incidence of AIDS-defining cancers (Kaposi sarcoma, non-Hodgkin lymphoma, cervical carcinoma) has decreased steadily from 8.7% in the 1980s to 6.4% during the years 1990 to 1995, and to 2.1% between 1996 and 2006. This is attributable to improved immune function as a result of treatment success with antiviral therapy.¹²

But the incidence of non-AIDS-defining cancers (Hodgkin disease, anal cancer, oral and respiratory cancers) has increased.¹¹ As therapy has regenerated the immune system, patients are surviving longer and are developing the more common cancers but with higher rates than in the general population.

Higher cancer risk is attributed to reduced immune surveillance. Many of these cancers are associated with viruses, such as human papillomavirus (anal and oral or pharyngeal cancers) and Epstein-Barr virus (Hodgkin disease), which can usually be controlled by a fully functioning immune system. The lower the CD4 count, the higher the risk of cancer, which highlights the need to diagnose HIV and start treatment early.¹³

Cardiovascular disease increases with lower CD4 counts

Associations have recently been identified between coronary disease and HIV as well as with HIV medications. Protease inhibitors tend to raise the levels of triglycerides, low-density lipoprotein cholesterol, and total cholesterol and increase the risk of heart attack.¹⁴

Regardless of therapy, HIV appears to be an independent risk factor for coronary disease. Arterial stiffness, as measured by carotid femoral pulse-wave velocity, was found to be increased among a sample of 80 HIV-infected men. This was associated with the usual risk factors of increasing age, blood pressure, and diabetes, as well as with lower nadir CD4 count.¹⁵

Fractures and neurocognitive disorders increase with HIV

Osteoporotic fractures are also more common in patients with HIV than in the general population. Risk factors include the traditional risks of older age, hepatitis C infection, diabetes, and substance abuse, but also nadir CD4 count less than 200.¹⁶

The risk of neurocognitive disorders is also associated with lower nadir CD4 counts. The lower the CD4 count, the higher the risk of developing neurocognitive deficits.¹⁷ The potential benefits of earlier diagnosis and treatment are obvious based upon the multiple recent findings outlined above.

CLINICAL PRESENTATION OF PRIMARY HIV INFECTION

During primary HIV infection, when patients are first infected, 50% to 90% are symptomatic. Symptoms usually appear in the first 6 weeks. The viral load tends to be highest at this time. Higher viral loads appear directly correlated with the degree of infectivity, highlighting the urgency of finding and treating new infections promptly to help avoid transmission to others.¹⁸

The clinical picture during primary infection is similar to that of acute mononucleosis. Signs and symptoms include fever, fatigue, rash, headache, lymphadenopathy, sore throat, and muscle aches. Although this presentation is common to many viral infections, questioning the patient about high-risk behavior (unprotected sex, multiple partners, intravenous drug use) will lead the astute physician to the correct testing and diagnosis.

Other early manifestations include mucocutaneous signs, such as seborrheic dermatitis, psoriasis, folliculitis, and thrush. Laboratory test results demonstrating leukopenia, thrombocytopenia, elevated total protein levels, proteinuria, and transaminitis are also suggestive of HIV infection.

THE CASE FOR INCREASED TESTING AND TREATMENT

The estimated prevalence of HIV in the United States is approximately 0.3%. However, its prevalence in Washington, DC, is 3%, which rivals rates in some areas of the developing world. From 2004 to 2008, health officials made a concerted effort in Washington, DC, to screen more people, particularly those at high risk. The number of publicly funded HIV tests performed increased by a factor of 3.7, and the number of newly reported cases increased by 17%. There was also a significant increase in the median CD4 count at the time of HIV diagnosis and a significant delay in time to progression to AIDS after HIV diagnosis.¹⁹

A study in British Columbia expanded access to highly active antiretroviral therapy during 2004 through 2009. High-risk individuals were targeted for increased screening. All those diagnosed with HIV were provided free medication. This resulted in a 50% reduction in new diagnoses of HIV infection throughout the community, especially among injectable drug users, a usually marginalized population. The proportion of patients with HIV-1 RNA levels above 1,500 copies/mL fell from about 50% to about 20%, indicating that the viral load—a measure of infectivity throughout the community—was reduced. Interestingly, this trend occurred during a time of increased rates of gonorrhea, syphilis, and other sexually transmitted diseases known to be associated with enhanced HIV transmission.²⁰

In Africa, antiretroviral therapy was offered to discordant couples (one partner was infected with HIV and the other was not). Among those who chose therapy, the rate of HIV transmission was 92% lower than in those not receiving antiretroviral drugs,²¹ once again demonstrating that control of HIV by treatment can lead to decreased transmission.

US HIV testing is inadequate

The current state of HIV testing in the United States needs to be improved. Testing is not performed routinely, leading to delayed diagnosis when patients present with symptomatic, advanced disease. Patients who are tested late (within 12 months before being diagnosed with AIDS) tend to be younger and less educated and are more likely to be heterosexual and either African American or Hispanic than patients who are tested earlier.²² When retrospectively evaluated, these patients often have been in the health care system but not tested. Routine universal screening and targeted testing could lead to a much earlier diagnosis and potential better long-term outcomes.

A 1996 survey of 95 academic emergency departments found that for patients with suspected sexually transmitted infections, 93% of physicians said they screen for gonorrhea, 88% for Chlamydia infection, 58% for syphilis, but only 3% for HIV.²³ Sexually transmitted infections and HIV are often transmitted together.

A similar 2002 survey of 154 emergency department providers who saw an average of 13 patients with sexually transmitted infections per week found that only 10% always recommend HIV testing to these patients. Reasons given for not testing were concern about follow-up (51%), not having a “certified” counselor (45%), HIV testing being too time-consuming (19%), and HIV testing being unavailable (27%).²⁴

Although most HIV tests are given by private doctors and health maintenance organizations, the likelihood of finding patients with HIV is greatest in hospitals, emergency departments, outpatient clinics, and public community clinics.

The Advancing HIV Prevention initiative of the US Centers for Disease Control and Prevention (CDC) has four priorities:

• To make voluntary HIV testing a routine part of medical care
• To implement new models for diagnosing HIV infection outside medical settings
• To prevent HIV infection by working with patients with HIV and their partners
• To further decrease the rate of perinatal HIV transmission.

There is a public health need to have rapid HIV testing available in all health care settings. With standard HIV tests, which can take 48 to 72 hours to run, about one-third of patients do not return for results. Subsequently locating them can be a huge challenge and is sometimes impossible. The ability to have rapid test results can improve this situation. It is especially important in prenatal care settings, where the mother can be immediately treated to reduce the risk of transmission to the child. Rapid testing increases the feasibility of testing in multiple venues, particularly acute-care settings with almost immediate results and linkage to care.

Several rapid tests are available and can be performed on whole blood, serum, plasma, and oral fluid. The tests provide reliable results in minutes, with 99% sensitivity and specificity. Positive results must be confirmed by subsequent two-stage laboratory testing, enzyme-linked immunosorbent assay, and Western blot. Patients who have negative or have indeterminate results on Western blot testing should be tested again after 4 weeks.

The cost-effectiveness of routine screening for HIV, even in populations with a low prevalence, is similar to that of commonly accepted interventions.²⁵ In populations with a 1% prevalence of HIV, the cost is $15,078 per quality-adjusted life-year.²⁶ Even if the prevalence is less than 0.05%, the cost is less than $50,000 per quality-adjusted life-year, which is normally the cutoff for acceptability for screening tests.^25,26

‘OPT-OUT’ TESTING

In the past, patients were asked if they would like to have HIV testing (“opt-in” testing). It is now recommended that physicians request testing to be performed (“opt-out” testing). This still allows the patient to decline but also conveys a “matter of fact” nonjudgmental message, indicative of a routine procedure no different than other screening tests. When testing was done on an opt-in basis, only 35% of pregnant women agreed to be tested. Some women felt that accepting an HIV test indicated that they engage in high-risk behavior. When testing was instead offered as routine but with an opportunity to decline, 88% accepted testing, and they were significantly less anxious about testing.²⁷

CDC RECOMMENDATIONS

The CDC now recommends that routine, voluntary HIV screening be done for all persons ages 13 to 64 in health care settings, regardless of risk.²⁸ Screening should be repeated at least annually in persons with known risk. Screening should be done on an opt-out basis, with the opportunity to ask questions and the option to decline. Consent for HIV testing should be included with general consent for care. A separate signed informed consent is not recommended, and verbal consent can merely be documented in the medical record. Prevention counseling in conjunction with HIV screening in health care settings is not required.

Testing should be done in all health care settings, including primary care settings, inpatient services, emergency departments, urgent care clinics, and sexually transmitted disease clinics. Test results should be communicated in the same manner as other diagnostic and screening care. Clinical HIV care should be available onsite or reliable referral to qualified providers should be established.

For pregnant women, the CDC recommends universal opt-out HIV screening, with HIV testing as part of the routine panel of prenatal screening tests. The consent for prenatal care includes HIV testing, with notification and the option to decline. Women should be tested again in the third trimester if they are known to be at risk for HIV, and in areas and health care facilities in which the prevalence of HIV is high.

In women whose HIV status is undocumented in labor and delivery, opt-out rapid testing should be performed, and antiretroviral prophylaxis should be given on the basis of the rapid test result. Rapid testing of the newborn is recommended if the mother’s status is unknown at delivery, and antiretroviral prophylaxis should be started within 12 hours of birth on the basis of the rapid test result.

Widespread routine screening and earlier treatment could significantly reduce the incidence and improve the outcomes of HIV in this country. Health care providers are encouraged to adopt these practices.