The Cost of Primary Care Research

The need for a strong research base for primary care has been generally acknowledged in the United States and elsewhere for nearly 30 years.^1-5 Still, family physicians and other primary care physicians are not doing well in the competition for National Institutes of Health (NIH) funding6; this situation has changed little in more than a decade.⁷ Although it has been argued that the amount of NIH funding is proportional to the burden of disease,⁸ these analyses have not considered the potential impact of federal funding on disease burden if it were more available for research in different settings, such as those of community primary care. For example, the burden of liver disease may be reduced more by primary care research directed to helping physicians treat patients who drink too much9 than by research on liver transplantation.

A significant portion of research project costs is incurred before the receipt of grant funds. This poses a major problem for the development of a primary care research enterprise. Researchers, especially new ones, require financial support for investigator and staff time, training, pilot work, and grant proposal writing to compete successfully for grant funds. Although some institutional funding is available in academic medical centers to support the initiation of research, there is little or no comparable resource in community practices. These “laboratories” for most practice-based primary care research must find ways to integrate research activities with direct patient care needs, and a culture of support for inquiry needs to be developed and sustained along with a supporting infrastructure.

Start-up support is being reduced even within academic medical centers, and this support is especially critical for young faculty and for pilot work.^10,11 The negative impact of this, while significant for all research,¹² may be even more problematic for the nascent field of primary care research, which relies disproportionately on internal resources,¹³ and especially for practice-based research, since community physicians generally lack institutional resources. Furthermore, the critical importance of primary care clinical services to academic medical centers, with the resulting demands on faculty, can detract from allocation of time and money to research activities.¹⁴ To add to the problem of initiating research in primary care, there is evidence of a bias against patient-oriented research compared with that which is laboratory oriented, making obtaining grant funds for clinical researchers even more difficult and time consuming.¹⁵

Although the NIH has proposed solutions for increasing the number of physicians engaged in clinical research,¹⁶ the problem of developing researchers capable of maintaining research funding cycles in the primary care specialties has not been adequately addressed at the national level. Although concerned with clinical research, the NIH has not made primary care research a priority. Even programs that attempt to develop primary care researchers fail to take into account associated costs.¹⁷

In part, primary care’s disappointing research productivity has been due to a lack of the capital investment needed to develop robust research infrastructures (which include practice-based laboratories) and the limited availability of the financial and institutional support that investigators need to apply for project funding. This lack of an adequate infrastructure and sufficient funding to invest in the start-up process before the receipt of grant funding has proved to be a nearly insurmountable barrier to potential primary care investigators. It has been less a barrier to faculty in other disciplines in which long histories of grant-funded projects have enabled them to use resources from previously funded projects as the start-up capital for future ones. Although mechanisms are in place in some primary care departments for a capital investment of funds and for protected time and mentoring, these resources are generally limited and are threatened. A further problem exists in that even with start-up funding it may be difficult, if not impossible, to sustain a research program with a limited number of grant-funded projects running at any one time.¹⁸

Despite its critical role in the development of a research enterprise, the requirement for support for start-up efforts has not been quantified, and no scheme for such quantification has been published. Appropriately, grant budgets do not include the costs of getting to the point of grant submission. However, costs are incurred that must be covered by some source.

We provide a model that lists categories that will allow for the quantification of primary care research costs, including those incurred before the receipt of grant funds (Figure). In this model we assume that a significant portion of primary care research will be practice based, and thus we have included in our model costs related to the practices. One way of developing and using a “laboratory” for practice-based research is through the development and use of a practice-based research network, and thus these costs are also reflected in our model. We acknowledge that many aspects of the model would also apply to research in other settings, including basic science research and traditional clinical research in academic settings.

Methods

To develop our model, a group of University of Wisconsin and Wisconsin Research Network investigators experienced in primary care research explored the activities and associated costs necessary for conducting research projects. This group included an academic family physician (J.W.B.), a community family physician (D.H.), and 3 experienced research staff members (P.W., M.B.P., L.M.). We began with 4 core assumptions and then identified 10 phases of activity in the research process, the first 4 of which occur before the receipt of grant funding.

Four Assumptions

On the basis of our collective experience, our group made 4 assumptions that are important for our analysis of the costs associated with primary care research. These assumptions are:

1. Expensive phases of the research process occur before the receipt of grant funds. It is therefore difficult to initiate major research solely through the grant-funding process. Many activities related to developing hypotheses, determining research methods, doing pilot work, and producing fundable proposals must occur before grant funding. Based on our collective experiences in applying for federal funding, we estimate that an investigator requires approximately a year of full-time equivalent work (eg, 25% time over 4 years) to get his or her first R01 grant funding.^19-21 An unsuccessful application by one skilled individual was estimated to cost 220 hours of effort, exclusive of the time of staff and colleagues.²²
2. We need community practice settings for primary care research, and their research activities should be rewarded at approximately the same rate as other practice activities. Not only are these settings critical when physician practice is the object of the research, but it also seems that patient recruitment may be better (possibly with less selection bias) in community practices,²³ since more patients are served by these than academic practices. Physicians and patients in community practice settings cannot participate without remuneration, especially when many studies and more time-consuming and sophisticated interventional studies are done. This is not only an issue of physician volunteerism. More physicians and staff are members of corporate medical entities that are beginning to insist on reimbursement for every corporate activity. Generally, pharmaceutical research is the only research that has budgeted adequate reimbursement for practices.
3. Patients must not incur costs for participating in research activities. Making patients financially support research will reduce participation and potentially increase selection bias.
4. An exact calculation for the budgeting of research activities across various primary care settings is not possible, since expenses are organization- and project-dependent. However, general estimates could be fit into the model.

The Research Process Before the Receipt of Grant Funding

The costs of research in primary care can be classified into 10 phases of operation by the series of tasks that have to be accomplished and by the organizations that will be required to accomplish them. Although our list may not exhaust the possible range of phases and organizations, it can serve as a framework for discussion of the resources needed to conceptualize a project, apply for funding, and carry out research. Not all phases are applicable to all projects, but in general most research will require activity in most phases.

• Phase 1: Develop and Maintain a Commitment to Research. Regardless of whether a formal research network is used, researchers must convince the practice community to support research efforts, especially since projects that originate within academic departments may be removed from the interests and concerns of physicians in community practice. If a research network is involved as a means to maintain the “laboratory,” an ongoing investment in network operations is required. We note that even ongoing successful research institutions generally cannot rely solely on grant funding.²⁴ These institutions often have some base support from other sources, such as state funding.
• Phase 2: Conceptualize and Design a Project. Time and energy must be invested in literature reviews, conferences, and consultations to design a research project that builds on existing experience and knowledge. The active involvement of clinicians at this stage helps to ensure buy-in and, more important, ensures that the results of the research are likely to be relevant to practice.
• Phase 3: Pilot-Test and Revise the Design. Generally, it is necessary to conduct some preliminary research to demonstrate feasibility and work out methodologic details before writing the grant proposal. Data collection instruments may have to be identified or developed, and some statistical analysis should be done.
• Phase 4: Write and Submit the Grant Proposal. Preparing a grant proposal requires time for writing, obtaining institutional review board approval, realistic budgeting, consultation, and revision. These efforts may not achieve the desired goal of funding, however; only 20% of NIH submissions receive funding on the first submission.

The Research Process After the Receipt of Grant Funding

• Phase 5: Recruit the Practice Sites, Clinicians, and Patients. The larger, more complex, and more intrusive the research project, the more difficult this stage of operation is. The expenses of this phase have recently been described^25-27 and quantified.²⁸
• Phase 6: Train the Researchers, Clinicians, and Staff. It may prove necessary to provide training for the personnel involved in the project. They may include research project staff, clinicians, and clinical staff at multiple or remote sites.
• Phase 7: Collect the Data. Data need to be collected at several steps along the way. If the trial is an interventional trial, pre- and postintervention data will be needed. Other types of research will require the tracking of patients over time. Data about physician-patient interactions may be difficult to collect, often requiring intensive qualitative methods.
• Phase 8: Intervention. Conducting the intervention will require clinician and patient time, staff time, materials, travel, and communication costs.
• Phase 9: Data Entry and Analysis. Database design and management and data entry are required. Data analysis is needed.
• Phase 10: Disseminate Results. The dissemination of study results must include feedback to practices, the preparation of grant reports, publication, and presentations. In addition, primary care research projects should include arrangements for the translation of the research results into practice.

The costs related to each phase of operation must be funded (directly or indirectly) by one or more of 3 sources: (1) the academic or network infrastructure; (2) the practice site infrastructure, including the physician and patient; and (3) an extramural funding source.

The proportion of costs incurred by the various organizations or the extramural funding source will vary over time for any project. There is considerable overlap in how the costs of the various phases of research are distributed among the 3 types of sources. However, any extramural funding source will help support the research costs only in phase 5 and beyond.

Discussion

We have outlined the phases of research in a way that allows for some itemization of project expenses. The importance of the development and maintenance of a robust and well-funded support infrastructure is demonstrated by the fact that all activities up to and including those in phase 4 are performed before receiving grant funding.

Limitations

The limitations of this work include the lack of external validation of our model. Nonetheless, the basic outline provides one way to look at the costs of initiating and conducting research projects. We encourage other researchers to build on these ideas. In future work we intend to attach costs to the phases and activities outlined in this paper.

Conclusions

We hope that we have articulated the need for capital investment in the primary care research process, with special attention to practice-based research. This investment, which has paid great dividends in other areas of medicine, will help primary care investigators meet the challenge of developing a robust research enterprise. Such an investment will help us to ensure that the questions asked are important ones for primary care clinicians and that the answers will ultimately pay dividends in the health of our population in the same way research in other areas of medicine has.

The need for a strong research base for primary care has been generally acknowledged in the United States and elsewhere for nearly 30 years.^1-5 Still, family physicians and other primary care physicians are not doing well in the competition for National Institutes of Health (NIH) funding6; this situation has changed little in more than a decade.⁷ Although it has been argued that the amount of NIH funding is proportional to the burden of disease,⁸ these analyses have not considered the potential impact of federal funding on disease burden if it were more available for research in different settings, such as those of community primary care. For example, the burden of liver disease may be reduced more by primary care research directed to helping physicians treat patients who drink too much9 than by research on liver transplantation.

A significant portion of research project costs is incurred before the receipt of grant funds. This poses a major problem for the development of a primary care research enterprise. Researchers, especially new ones, require financial support for investigator and staff time, training, pilot work, and grant proposal writing to compete successfully for grant funds. Although some institutional funding is available in academic medical centers to support the initiation of research, there is little or no comparable resource in community practices. These “laboratories” for most practice-based primary care research must find ways to integrate research activities with direct patient care needs, and a culture of support for inquiry needs to be developed and sustained along with a supporting infrastructure.

Start-up support is being reduced even within academic medical centers, and this support is especially critical for young faculty and for pilot work.^10,11 The negative impact of this, while significant for all research,¹² may be even more problematic for the nascent field of primary care research, which relies disproportionately on internal resources,¹³ and especially for practice-based research, since community physicians generally lack institutional resources. Furthermore, the critical importance of primary care clinical services to academic medical centers, with the resulting demands on faculty, can detract from allocation of time and money to research activities.¹⁴ To add to the problem of initiating research in primary care, there is evidence of a bias against patient-oriented research compared with that which is laboratory oriented, making obtaining grant funds for clinical researchers even more difficult and time consuming.¹⁵

Although the NIH has proposed solutions for increasing the number of physicians engaged in clinical research,¹⁶ the problem of developing researchers capable of maintaining research funding cycles in the primary care specialties has not been adequately addressed at the national level. Although concerned with clinical research, the NIH has not made primary care research a priority. Even programs that attempt to develop primary care researchers fail to take into account associated costs.¹⁷

In part, primary care’s disappointing research productivity has been due to a lack of the capital investment needed to develop robust research infrastructures (which include practice-based laboratories) and the limited availability of the financial and institutional support that investigators need to apply for project funding. This lack of an adequate infrastructure and sufficient funding to invest in the start-up process before the receipt of grant funding has proved to be a nearly insurmountable barrier to potential primary care investigators. It has been less a barrier to faculty in other disciplines in which long histories of grant-funded projects have enabled them to use resources from previously funded projects as the start-up capital for future ones. Although mechanisms are in place in some primary care departments for a capital investment of funds and for protected time and mentoring, these resources are generally limited and are threatened. A further problem exists in that even with start-up funding it may be difficult, if not impossible, to sustain a research program with a limited number of grant-funded projects running at any one time.¹⁸

Despite its critical role in the development of a research enterprise, the requirement for support for start-up efforts has not been quantified, and no scheme for such quantification has been published. Appropriately, grant budgets do not include the costs of getting to the point of grant submission. However, costs are incurred that must be covered by some source.

We provide a model that lists categories that will allow for the quantification of primary care research costs, including those incurred before the receipt of grant funds (Figure). In this model we assume that a significant portion of primary care research will be practice based, and thus we have included in our model costs related to the practices. One way of developing and using a “laboratory” for practice-based research is through the development and use of a practice-based research network, and thus these costs are also reflected in our model. We acknowledge that many aspects of the model would also apply to research in other settings, including basic science research and traditional clinical research in academic settings.

Methods

To develop our model, a group of University of Wisconsin and Wisconsin Research Network investigators experienced in primary care research explored the activities and associated costs necessary for conducting research projects. This group included an academic family physician (J.W.B.), a community family physician (D.H.), and 3 experienced research staff members (P.W., M.B.P., L.M.). We began with 4 core assumptions and then identified 10 phases of activity in the research process, the first 4 of which occur before the receipt of grant funding.

Four Assumptions

On the basis of our collective experience, our group made 4 assumptions that are important for our analysis of the costs associated with primary care research. These assumptions are:

1. Expensive phases of the research process occur before the receipt of grant funds. It is therefore difficult to initiate major research solely through the grant-funding process. Many activities related to developing hypotheses, determining research methods, doing pilot work, and producing fundable proposals must occur before grant funding. Based on our collective experiences in applying for federal funding, we estimate that an investigator requires approximately a year of full-time equivalent work (eg, 25% time over 4 years) to get his or her first R01 grant funding.^19-21 An unsuccessful application by one skilled individual was estimated to cost 220 hours of effort, exclusive of the time of staff and colleagues.²²
2. We need community practice settings for primary care research, and their research activities should be rewarded at approximately the same rate as other practice activities. Not only are these settings critical when physician practice is the object of the research, but it also seems that patient recruitment may be better (possibly with less selection bias) in community practices,²³ since more patients are served by these than academic practices. Physicians and patients in community practice settings cannot participate without remuneration, especially when many studies and more time-consuming and sophisticated interventional studies are done. This is not only an issue of physician volunteerism. More physicians and staff are members of corporate medical entities that are beginning to insist on reimbursement for every corporate activity. Generally, pharmaceutical research is the only research that has budgeted adequate reimbursement for practices.
3. Patients must not incur costs for participating in research activities. Making patients financially support research will reduce participation and potentially increase selection bias.
4. An exact calculation for the budgeting of research activities across various primary care settings is not possible, since expenses are organization- and project-dependent. However, general estimates could be fit into the model.

The Research Process Before the Receipt of Grant Funding

The costs of research in primary care can be classified into 10 phases of operation by the series of tasks that have to be accomplished and by the organizations that will be required to accomplish them. Although our list may not exhaust the possible range of phases and organizations, it can serve as a framework for discussion of the resources needed to conceptualize a project, apply for funding, and carry out research. Not all phases are applicable to all projects, but in general most research will require activity in most phases.

• Phase 1: Develop and Maintain a Commitment to Research. Regardless of whether a formal research network is used, researchers must convince the practice community to support research efforts, especially since projects that originate within academic departments may be removed from the interests and concerns of physicians in community practice. If a research network is involved as a means to maintain the “laboratory,” an ongoing investment in network operations is required. We note that even ongoing successful research institutions generally cannot rely solely on grant funding.²⁴ These institutions often have some base support from other sources, such as state funding.
• Phase 2: Conceptualize and Design a Project. Time and energy must be invested in literature reviews, conferences, and consultations to design a research project that builds on existing experience and knowledge. The active involvement of clinicians at this stage helps to ensure buy-in and, more important, ensures that the results of the research are likely to be relevant to practice.
• Phase 3: Pilot-Test and Revise the Design. Generally, it is necessary to conduct some preliminary research to demonstrate feasibility and work out methodologic details before writing the grant proposal. Data collection instruments may have to be identified or developed, and some statistical analysis should be done.
• Phase 4: Write and Submit the Grant Proposal. Preparing a grant proposal requires time for writing, obtaining institutional review board approval, realistic budgeting, consultation, and revision. These efforts may not achieve the desired goal of funding, however; only 20% of NIH submissions receive funding on the first submission.

The Research Process After the Receipt of Grant Funding

• Phase 5: Recruit the Practice Sites, Clinicians, and Patients. The larger, more complex, and more intrusive the research project, the more difficult this stage of operation is. The expenses of this phase have recently been described^25-27 and quantified.²⁸
• Phase 6: Train the Researchers, Clinicians, and Staff. It may prove necessary to provide training for the personnel involved in the project. They may include research project staff, clinicians, and clinical staff at multiple or remote sites.
• Phase 7: Collect the Data. Data need to be collected at several steps along the way. If the trial is an interventional trial, pre- and postintervention data will be needed. Other types of research will require the tracking of patients over time. Data about physician-patient interactions may be difficult to collect, often requiring intensive qualitative methods.
• Phase 8: Intervention. Conducting the intervention will require clinician and patient time, staff time, materials, travel, and communication costs.
• Phase 9: Data Entry and Analysis. Database design and management and data entry are required. Data analysis is needed.
• Phase 10: Disseminate Results. The dissemination of study results must include feedback to practices, the preparation of grant reports, publication, and presentations. In addition, primary care research projects should include arrangements for the translation of the research results into practice.

The costs related to each phase of operation must be funded (directly or indirectly) by one or more of 3 sources: (1) the academic or network infrastructure; (2) the practice site infrastructure, including the physician and patient; and (3) an extramural funding source.

The proportion of costs incurred by the various organizations or the extramural funding source will vary over time for any project. There is considerable overlap in how the costs of the various phases of research are distributed among the 3 types of sources. However, any extramural funding source will help support the research costs only in phase 5 and beyond.

Discussion

We have outlined the phases of research in a way that allows for some itemization of project expenses. The importance of the development and maintenance of a robust and well-funded support infrastructure is demonstrated by the fact that all activities up to and including those in phase 4 are performed before receiving grant funding.

Limitations

The limitations of this work include the lack of external validation of our model. Nonetheless, the basic outline provides one way to look at the costs of initiating and conducting research projects. We encourage other researchers to build on these ideas. In future work we intend to attach costs to the phases and activities outlined in this paper.

Conclusions

We hope that we have articulated the need for capital investment in the primary care research process, with special attention to practice-based research. This investment, which has paid great dividends in other areas of medicine, will help primary care investigators meet the challenge of developing a robust research enterprise. Such an investment will help us to ensure that the questions asked are important ones for primary care clinicians and that the answers will ultimately pay dividends in the health of our population in the same way research in other areas of medicine has.

The need for a strong research base for primary care has been generally acknowledged in the United States and elsewhere for nearly 30 years.^1-5 Still, family physicians and other primary care physicians are not doing well in the competition for National Institutes of Health (NIH) funding6; this situation has changed little in more than a decade.⁷ Although it has been argued that the amount of NIH funding is proportional to the burden of disease,⁸ these analyses have not considered the potential impact of federal funding on disease burden if it were more available for research in different settings, such as those of community primary care. For example, the burden of liver disease may be reduced more by primary care research directed to helping physicians treat patients who drink too much9 than by research on liver transplantation.

A significant portion of research project costs is incurred before the receipt of grant funds. This poses a major problem for the development of a primary care research enterprise. Researchers, especially new ones, require financial support for investigator and staff time, training, pilot work, and grant proposal writing to compete successfully for grant funds. Although some institutional funding is available in academic medical centers to support the initiation of research, there is little or no comparable resource in community practices. These “laboratories” for most practice-based primary care research must find ways to integrate research activities with direct patient care needs, and a culture of support for inquiry needs to be developed and sustained along with a supporting infrastructure.

Start-up support is being reduced even within academic medical centers, and this support is especially critical for young faculty and for pilot work.^10,11 The negative impact of this, while significant for all research,¹² may be even more problematic for the nascent field of primary care research, which relies disproportionately on internal resources,¹³ and especially for practice-based research, since community physicians generally lack institutional resources. Furthermore, the critical importance of primary care clinical services to academic medical centers, with the resulting demands on faculty, can detract from allocation of time and money to research activities.¹⁴ To add to the problem of initiating research in primary care, there is evidence of a bias against patient-oriented research compared with that which is laboratory oriented, making obtaining grant funds for clinical researchers even more difficult and time consuming.¹⁵

Although the NIH has proposed solutions for increasing the number of physicians engaged in clinical research,¹⁶ the problem of developing researchers capable of maintaining research funding cycles in the primary care specialties has not been adequately addressed at the national level. Although concerned with clinical research, the NIH has not made primary care research a priority. Even programs that attempt to develop primary care researchers fail to take into account associated costs.¹⁷

In part, primary care’s disappointing research productivity has been due to a lack of the capital investment needed to develop robust research infrastructures (which include practice-based laboratories) and the limited availability of the financial and institutional support that investigators need to apply for project funding. This lack of an adequate infrastructure and sufficient funding to invest in the start-up process before the receipt of grant funding has proved to be a nearly insurmountable barrier to potential primary care investigators. It has been less a barrier to faculty in other disciplines in which long histories of grant-funded projects have enabled them to use resources from previously funded projects as the start-up capital for future ones. Although mechanisms are in place in some primary care departments for a capital investment of funds and for protected time and mentoring, these resources are generally limited and are threatened. A further problem exists in that even with start-up funding it may be difficult, if not impossible, to sustain a research program with a limited number of grant-funded projects running at any one time.¹⁸

Despite its critical role in the development of a research enterprise, the requirement for support for start-up efforts has not been quantified, and no scheme for such quantification has been published. Appropriately, grant budgets do not include the costs of getting to the point of grant submission. However, costs are incurred that must be covered by some source.

We provide a model that lists categories that will allow for the quantification of primary care research costs, including those incurred before the receipt of grant funds (Figure). In this model we assume that a significant portion of primary care research will be practice based, and thus we have included in our model costs related to the practices. One way of developing and using a “laboratory” for practice-based research is through the development and use of a practice-based research network, and thus these costs are also reflected in our model. We acknowledge that many aspects of the model would also apply to research in other settings, including basic science research and traditional clinical research in academic settings.

Methods

To develop our model, a group of University of Wisconsin and Wisconsin Research Network investigators experienced in primary care research explored the activities and associated costs necessary for conducting research projects. This group included an academic family physician (J.W.B.), a community family physician (D.H.), and 3 experienced research staff members (P.W., M.B.P., L.M.). We began with 4 core assumptions and then identified 10 phases of activity in the research process, the first 4 of which occur before the receipt of grant funding.

Four Assumptions

On the basis of our collective experience, our group made 4 assumptions that are important for our analysis of the costs associated with primary care research. These assumptions are:

1. Expensive phases of the research process occur before the receipt of grant funds. It is therefore difficult to initiate major research solely through the grant-funding process. Many activities related to developing hypotheses, determining research methods, doing pilot work, and producing fundable proposals must occur before grant funding. Based on our collective experiences in applying for federal funding, we estimate that an investigator requires approximately a year of full-time equivalent work (eg, 25% time over 4 years) to get his or her first R01 grant funding.^19-21 An unsuccessful application by one skilled individual was estimated to cost 220 hours of effort, exclusive of the time of staff and colleagues.²²
2. We need community practice settings for primary care research, and their research activities should be rewarded at approximately the same rate as other practice activities. Not only are these settings critical when physician practice is the object of the research, but it also seems that patient recruitment may be better (possibly with less selection bias) in community practices,²³ since more patients are served by these than academic practices. Physicians and patients in community practice settings cannot participate without remuneration, especially when many studies and more time-consuming and sophisticated interventional studies are done. This is not only an issue of physician volunteerism. More physicians and staff are members of corporate medical entities that are beginning to insist on reimbursement for every corporate activity. Generally, pharmaceutical research is the only research that has budgeted adequate reimbursement for practices.
3. Patients must not incur costs for participating in research activities. Making patients financially support research will reduce participation and potentially increase selection bias.
4. An exact calculation for the budgeting of research activities across various primary care settings is not possible, since expenses are organization- and project-dependent. However, general estimates could be fit into the model.

The Research Process Before the Receipt of Grant Funding

The costs of research in primary care can be classified into 10 phases of operation by the series of tasks that have to be accomplished and by the organizations that will be required to accomplish them. Although our list may not exhaust the possible range of phases and organizations, it can serve as a framework for discussion of the resources needed to conceptualize a project, apply for funding, and carry out research. Not all phases are applicable to all projects, but in general most research will require activity in most phases.

• Phase 1: Develop and Maintain a Commitment to Research. Regardless of whether a formal research network is used, researchers must convince the practice community to support research efforts, especially since projects that originate within academic departments may be removed from the interests and concerns of physicians in community practice. If a research network is involved as a means to maintain the “laboratory,” an ongoing investment in network operations is required. We note that even ongoing successful research institutions generally cannot rely solely on grant funding.²⁴ These institutions often have some base support from other sources, such as state funding.
• Phase 2: Conceptualize and Design a Project. Time and energy must be invested in literature reviews, conferences, and consultations to design a research project that builds on existing experience and knowledge. The active involvement of clinicians at this stage helps to ensure buy-in and, more important, ensures that the results of the research are likely to be relevant to practice.
• Phase 3: Pilot-Test and Revise the Design. Generally, it is necessary to conduct some preliminary research to demonstrate feasibility and work out methodologic details before writing the grant proposal. Data collection instruments may have to be identified or developed, and some statistical analysis should be done.
• Phase 4: Write and Submit the Grant Proposal. Preparing a grant proposal requires time for writing, obtaining institutional review board approval, realistic budgeting, consultation, and revision. These efforts may not achieve the desired goal of funding, however; only 20% of NIH submissions receive funding on the first submission.

The Research Process After the Receipt of Grant Funding

• Phase 5: Recruit the Practice Sites, Clinicians, and Patients. The larger, more complex, and more intrusive the research project, the more difficult this stage of operation is. The expenses of this phase have recently been described^25-27 and quantified.²⁸
• Phase 6: Train the Researchers, Clinicians, and Staff. It may prove necessary to provide training for the personnel involved in the project. They may include research project staff, clinicians, and clinical staff at multiple or remote sites.
• Phase 7: Collect the Data. Data need to be collected at several steps along the way. If the trial is an interventional trial, pre- and postintervention data will be needed. Other types of research will require the tracking of patients over time. Data about physician-patient interactions may be difficult to collect, often requiring intensive qualitative methods.
• Phase 8: Intervention. Conducting the intervention will require clinician and patient time, staff time, materials, travel, and communication costs.
• Phase 9: Data Entry and Analysis. Database design and management and data entry are required. Data analysis is needed.
• Phase 10: Disseminate Results. The dissemination of study results must include feedback to practices, the preparation of grant reports, publication, and presentations. In addition, primary care research projects should include arrangements for the translation of the research results into practice.

The costs related to each phase of operation must be funded (directly or indirectly) by one or more of 3 sources: (1) the academic or network infrastructure; (2) the practice site infrastructure, including the physician and patient; and (3) an extramural funding source.

The proportion of costs incurred by the various organizations or the extramural funding source will vary over time for any project. There is considerable overlap in how the costs of the various phases of research are distributed among the 3 types of sources. However, any extramural funding source will help support the research costs only in phase 5 and beyond.

Discussion

We have outlined the phases of research in a way that allows for some itemization of project expenses. The importance of the development and maintenance of a robust and well-funded support infrastructure is demonstrated by the fact that all activities up to and including those in phase 4 are performed before receiving grant funding.

Limitations

The limitations of this work include the lack of external validation of our model. Nonetheless, the basic outline provides one way to look at the costs of initiating and conducting research projects. We encourage other researchers to build on these ideas. In future work we intend to attach costs to the phases and activities outlined in this paper.

Conclusions

We hope that we have articulated the need for capital investment in the primary care research process, with special attention to practice-based research. This investment, which has paid great dividends in other areas of medicine, will help primary care investigators meet the challenge of developing a robust research enterprise. Such an investment will help us to ensure that the questions asked are important ones for primary care clinicians and that the answers will ultimately pay dividends in the health of our population in the same way research in other areas of medicine has.