James W. Mold, MD

Failure to notify patients of abnormal laboratory test results or to ensure appropriate follow-up can result in inferior patient care and potential malpractice liability.^1-5 Failure to document physician review and communication of test results to patients can make defense of a malpractice claim more difficult. Our review of the English-language literature identified only a few studies related to the management and reporting of laboratory test results in primary care.^6-10

Boohaker and colleagues⁹ proposed that there are 4 basic steps involved in managing laboratory test results: tracking tests until results are received; notifying patients of the results; documenting that the notification occurred; and assuring that recommended follow-up occurs for abnormal test results. They surveyed 161 attending physicians and 101 internal medicine and family practice residents at the Henry Ford Hospital in Detroit and 21 suburban primary care practices in southeast Michigan for information on the perceived importance of each step, how each of the steps was handled, and the perceived reliability of the methods used. The percentages of respondents with fair or poor methods or no method for tracking test results were: 17% for blood studies, 27% for Papanicolaou tests, 29% for mammograms, and 32% for X-rays. Approximately 50% felt that it was moderately or extremely important to notify all patients of normal test results. However, only 28% said that they accomplished this. Thirty-six percent admitted that they do not always succeed in notifying patients of abnormal test results. Often this was because patients were expected to return to the clinic soon, or the results were considered trivial. However, patient unavailability, physician forgetfulness, and time constraints were contributing factors as well. Only 55% reported that they always document patient notification of laboratory results. Another 30% said they did so most of the time. Seventy-five percent had a fair or poor method or no method for tracking patients to see if they received recommended follow-up of abnormal test results. A patient survey was conducted in the Internal Medicine Teaching Clinic at Henry Ford Hospital. Seventy-nine percent of patients surveyed wanted to be notified of all test results, whether normal or abnormal.

No one has identified and published a unified method that accomplishes each of the 4 steps effectively and efficiently. We attempted to do so by tapping the collective wisdom of clinician members of the Oklahoma Physicians Resource/Research Network (OKPRN), a network of practicing family physicians throughout Oklahoma.

Methods

Our study was conducted between August 1998 and April 1999. We sent a questionnaire to all 24 physician members of OKPRN. The questions were designed to obtain information about methods being used to address the 4 steps in the process articulated by Boohaker and coworkers⁹ and their perceived effectiveness. Two clinicians reporting effective methods for a particular step were then chosen for further analysis, with an attempt to choose practices with different methods. Four different physicians’ practices were audited, with several physicians reporting effective methods for more than 1 step. A single research assistant was sent to each practice to perform the audits. During this visit, a 5-minute open-ended interview was also conducted with the physician or the physician’s staff to more precisely describe the method used for the management step being evaluated.

Step 1: Test Tracking

Twenty patients who had been seen by the physician 2 to 6 weeks earlier, who had laboratory tests ordered during that visit, and whose charts were available were chosen at random by the research assistant ([Table 1]). Each patient’s records were reviewed, as was the logbook or tracking device used, following the sequence of procedures described by the physician. All deviations from the procedure were documented.

Step 2: Patient Notification

Twenty patients who had been seen by the physician 2 weeks to 3 months earlier and who had laboratory tests ordered were chosen at random by the research assistant. A letter and a return postcard were sent to each of the patients asking whether they remembered having the test done, whether they were notified of the results, how soon they were notified, whether they were satisfied with the method used to notify them, and whether they had any suggestions for improving the notification process. No attempt was made to contact nonresponders.

Step 3: Documentation of Notification

The research assistant, using the same selection process as for Step 2, reviewed the patients’ charts for evidence of documentation that the results of tests ordered at the index visit had been communicated to the patients.

Step 4: Follow-Up Tracking

Twenty patients who had been seen by the physician 2 to 3 months earlier, who had abnormal test results, and for whom follow-up recommendation had been made, were identified from billing records and encounter forms by the research assistant. Patients asked to follow up at some point beyond the audit date were excluded. Charts were reviewed for evidence that the patients had either followed up as recommended or that the clinic had recognized their failure to follow up and had made an attempt to contact them.

Analysis

We tabulated the audit results. For each step, the 2 methods were compared with each other and with the following performance criteria: (1) >90% satisfactory accomplishment of the step; (2) for step 2, >90% patient notification rate within 2 weeks; and (3) for step 2, >90% patient satisfaction. It was possible to identify effective methods for the first 3 steps in the laboratory test management process. At that point, the research assistant went back to the practices using the more effective methods and observed the time and material costs of implementing each component of the strategy (time-motion study). Each step was directly observed and timed for 5 consecutive sets of laboratory tests for 5 patients seen on the same day. The results were averaged.

Results

Twenty-three of 24 physicians (96%) responded to the questionnaire. A summary of the results is shown in [Table 2]. Eleven of 23 physicians (48%) said that they had a method that worked well for at least 1 of the 4 management steps. Four respondents reported having an effective method for step 1 (tracking test results). All reported using either a laboratory test log, billing slips, or a computerized system. Ten respondents reported having an effective method for step 2 (patient notification of test results). Of these, 5 had their nurse call the patient with test results or discussed the results at a follow-up appointment, and 3 sent the laboratory result with a hand-written note from the physician with or without an additional explanatory form. One mailed a patient-friendly form without the actual test results, and 1 did not elaborate on his method. Ten respondents reported having an effective method for step 3 (documentation of patient notification). The method that all of them reported using for this step was documentation on the laboratory sheet. Only 3 of the respondents reported having an effective method for step 4 (follow-up tracking). Of these, 2 relied on attempts to call no-shows to their offices, and 1 did not elaborate on his method.

Practice Audits

Step 1: Test Tracking At the first practice chosen to evaluate this step, a manifest logbook was used to record all tests sent to an outside laboratory. The nurse who drew the blood was responsible for entering the information in the logbook. In-house laboratory tests were not recorded, as they were generally completed while the patient was in the office. The nurse checked the book daily and marked any outstanding test results with colored self-adhesive notes. When the results came in, the same nurse initialed and dated the entry in the book and removed the note. If the test results were not back within the expected time period (dependent on the test ordered), the nurse called the laboratory and notified the physician if the specimen needed to be obtained again. The nurse initialed laboratory test reports as they came in and took them to the physician for review.

The audit of this method revealed that 15% of the laboratory tests ordered were not recorded in the manifest logbook (the nurse said she must have forgotten to do so), and 5% of the tests sent to outside laboratories were marked received on the log but were not found in the chart. In addition, at least 1 in-house test result was missing from the charts of 15% of patients for whom these tests were ordered.

At the second practice, the physician ordered laboratory tests using a special laboratory billing slip. The nurse made and kept a copy of this slip, sending the original to the laboratory. Practice personnel recorded the laboratory tests in one of 2 logbooks, “path” and “other.” When results came back they were checked off in the logbooks. The books were checked daily (other) or monthly (path) to catch those that were missing. As results came in, the nurse checked them off the billing slip, and when all were back for a particular patient she took them to the physician along with a test results information sheet. The nurse also checked the billing slips every 3 days for most laboratory tests and every 5 to 6 days for “path” and called the laboratory when results were not back in a reasonable period of time.

The audit revealed that 100% of the laboratory tests were listed in the logbook, and 100% of the laboratory test results received were recorded. However, 15% of patients’ charts did not contain the results even though the reports were marked as having been received.

Steps 2 and 3: Notification and Documentation At the first practice chosen to evaluate these steps, the physician instructed the nurse to call all patients with abnormal results to give them instructions regarding follow-up. Patients with normal results or follow-up appointments were not contacted. Those without follow-up appointments were sent a copy of the laboratory report along with any necessary comments from the physician. Calls and mailings were generally handled on the day after the results were received by the clinic. For documentation, the nurse initialed and dated the laboratory form and placed it in the chart.

Of the 20 cards we mailed to patients, 13 were returned (65%). Of these, 92% indicated they had been notified of test results. Fifty-four percent of these had received the results within 1 week, 38% within 1 to 2 weeks, 0% within 2 to 3 weeks, and 8% within 4 weeks after the test was performed. Ninety-two percent of the respondents were satisfied with the method used to notify them of their test results, and no suggestions were offered for improvement. The audit revealed that 40% of the charts had inadequate documentation of patient notification. Of these, 50% of the test results were not initialed, and 50% were not dated.

At the second practice, the physician initialed and dated the laboratory test results form and wrote a note on it to the patient regarding the results. This was copied and mailed to the patient with a generic form describing commonly ordered tests and what they indicate. Patients could call for clarification. The original laboratory report was placed in the chart. Patients were only contacted for urgent matters or if a specimen had to be recollected. For documentation, laboratory sheets were initialed and dated by the nurse and marked “mailed” at the time they were copied and sent to the patient.

Of the 20 cards mailed to patients of the second practice, 9 were returned (45%). Of the patients who responded, 100% indicated they had been notified of test results. Seventy-five percent of these had received the results within 1 week and 25% within 1 to 2 weeks after the test was performed. All of the respondents were satisfied with the method used to notify them of their test results, and no specific suggestions were made. The results of the audit of the second practice for documentation of patient notification revealed 95% of the charts had adequate documentation of patient notification.

Step 4: Follow-Up Tracking At the first practice chosen to evaluate this step, the clinic receptionist had a list of patients scheduled each day to follow up abnormal results. If the patient did not show up, the nurse was notified and an attempt was made to contact the patient by phone or mail. A note was written in the chart that this had been done. If the situation was felt to be serious or if the patient failed to follow up after several attempts, the physician was notified. The results of the audit revealed that 40% of the charts had inadequate documentation of follow-up tracking. Several problems were identified. The longer the delay in recommended follow-up, the less well the tracking system worked. Situations in which the first attempt to contact the patient was unsuccessful were often lost to further attempts at follow-up. There also appeared to be no way to track outside consultations, and the method required a fair amount of judgment by the office staff.

At the second practice, an internal referral form was used to track patients referred within the practice for follow-up of abnormal results. When patients failed to show up for a scheduled follow-up appointment, attempts are made to contact them, and these attempts were recorded on the internal referral form. If several unsuccessful attempts were made to contact the patient, the form came back to the physician. If a patient did not show up for an outside consultation, the consultant usually notified the physician. The results of the audit revealed that 35% of the patients whose charts were reviewed had not followed up as recommended in 3 months, and 10% followed up much later than recommended. None of the charts of those patients who followed up late or not at all contained documentation of attempts to contact the patient. We did not believe that the other physicians’ practices were sufficiently different from these 2 approaches to warrant audit.

Combined Best Method

Considering effectiveness, efficiency, and ease of use, we concluded that a combination approach using the methods from the second practices audited for steps 1, 2, and 3 represented the best method that we could identify within our network for managing those steps. We were not happy with either method audited for step 4. The combined method for steps 1 to 3 is summarized in [Table 3]. The results of the time-motion studies are shown in [Table 4].

Discussion

This is the first attempt that we are aware of to describe an effective method for the comprehensive management of laboratory test results in the primary care setting. The results of the physician survey (Table 3) demonstrate the diversity of methods being used by practicing physicians. Laboratory test results are handled differently by individual practices and by physicians within the same practice. By self-report, a substantial percentage of these physicians fall short of optimal performance in this area.

The method that we propose on the basis of our findings involves a dual registry of tests ordered and returned (laboratory logbook and nurse billing slip copy) and a uniform system for patient notification and documentation. It is possible that in some practices a single-entry system may work, especially if there is a paid laboratory technician who is held accountable to manage this. However, our findings agree with 2 British studies^6,7 that also suggest that a single-entry system often does not work adequately. We believe that it is important to have a uniform system with very little decision making required on the part of staff. We also think it important to provide patients with their actual results, not just a report stating they were normal.

We were unsuccessful in identifying a method that is effective for the follow-up of abnormal laboratory test results. Our guess is that this will require some kind of tickler file system with postcards sent to patients who appear to have not followed up as suggested. It may also be that methods could be developed, such as the one described by Del Mar and Wright¹⁰ in which the outside laboratory participates in the follow-up process.

We do not claim that our best method is the only method that will work. The number of clinicians surveyed and audited was small, they all practice in the same geographic area, and many of them were trained in the same institution. The response rates on the patient audits for step 2 were only 65% and 45%, raising the possibility that those who did not respond were unhappy with the methods used. Anecdotally, however, we have subsequently implemented steps 2 and 3 in one of our University-based clinics, and they seem to work extremely well. Our nurse loves it, and we have had no patient complaints.

Telephonic methods of managing laboratory test results are now available. These allow physicians to dictate personalized messages to patients about their results. Patients access the information by calling a designated number and using their own personal identification code. Reports are generated for the physician indicating which patients called for results so that another notification method could be used for those who did not. These services typically do not address steps 1 and 4, and it was the feeling of the clinicians in our network who had tried them that the services failed for enough patients that a back-up method was necessary.

The largest portion (43%) of the cost of the method is physician time. In theory this portion is billable within the evaluation and management (E/M) codes. Most practices also charge a laboratory collection fee. Our experience has been that on average a practice can collect approximately $3.00 of an $11.00 collection fee billed to patients’ insurance. This varies by insurance carrier and does not apply to capitated systems. This must cover the costs of test result management and the venopuncture, packaging, and labeling of specimens as well. We estimate that primary care physicians who manage test results lose approximately $5.00 on each set of tests sent out on an individual patient unless they properly incorporate the costs into the E/M charge for the visit.

Since completion of this research, a number of OKPRN clinicians have adopted all or portions of our proposed method and have found that it works extremely well. A further modification has been suggested, which is supported by at least one published study.¹³ At the time when the test is obtained, patients are asked to write their own address on the envelope to be used to mail results. This assures that the address is correct and saves the nurse time later, reducing the cost by approximately $0.30 per set of tests.

Acknowledgments

We received financial support from the Presbyterian Health Foundation. We are grateful for the contributions of Ed Farrow, MD; Cary Fisher, MD; Marcia Matthews, MD; Paul Louis Preslar, DO; and the staff of the Oklahoma Center for Family Medicine Research.

Failure to notify patients of abnormal laboratory test results or to ensure appropriate follow-up can result in inferior patient care and potential malpractice liability.^1-5 Failure to document physician review and communication of test results to patients can make defense of a malpractice claim more difficult. Our review of the English-language literature identified only a few studies related to the management and reporting of laboratory test results in primary care.^6-10

Boohaker and colleagues⁹ proposed that there are 4 basic steps involved in managing laboratory test results: tracking tests until results are received; notifying patients of the results; documenting that the notification occurred; and assuring that recommended follow-up occurs for abnormal test results. They surveyed 161 attending physicians and 101 internal medicine and family practice residents at the Henry Ford Hospital in Detroit and 21 suburban primary care practices in southeast Michigan for information on the perceived importance of each step, how each of the steps was handled, and the perceived reliability of the methods used. The percentages of respondents with fair or poor methods or no method for tracking test results were: 17% for blood studies, 27% for Papanicolaou tests, 29% for mammograms, and 32% for X-rays. Approximately 50% felt that it was moderately or extremely important to notify all patients of normal test results. However, only 28% said that they accomplished this. Thirty-six percent admitted that they do not always succeed in notifying patients of abnormal test results. Often this was because patients were expected to return to the clinic soon, or the results were considered trivial. However, patient unavailability, physician forgetfulness, and time constraints were contributing factors as well. Only 55% reported that they always document patient notification of laboratory results. Another 30% said they did so most of the time. Seventy-five percent had a fair or poor method or no method for tracking patients to see if they received recommended follow-up of abnormal test results. A patient survey was conducted in the Internal Medicine Teaching Clinic at Henry Ford Hospital. Seventy-nine percent of patients surveyed wanted to be notified of all test results, whether normal or abnormal.

No one has identified and published a unified method that accomplishes each of the 4 steps effectively and efficiently. We attempted to do so by tapping the collective wisdom of clinician members of the Oklahoma Physicians Resource/Research Network (OKPRN), a network of practicing family physicians throughout Oklahoma.

Methods

Our study was conducted between August 1998 and April 1999. We sent a questionnaire to all 24 physician members of OKPRN. The questions were designed to obtain information about methods being used to address the 4 steps in the process articulated by Boohaker and coworkers⁹ and their perceived effectiveness. Two clinicians reporting effective methods for a particular step were then chosen for further analysis, with an attempt to choose practices with different methods. Four different physicians’ practices were audited, with several physicians reporting effective methods for more than 1 step. A single research assistant was sent to each practice to perform the audits. During this visit, a 5-minute open-ended interview was also conducted with the physician or the physician’s staff to more precisely describe the method used for the management step being evaluated.

Step 1: Test Tracking

Twenty patients who had been seen by the physician 2 to 6 weeks earlier, who had laboratory tests ordered during that visit, and whose charts were available were chosen at random by the research assistant ([Table 1]). Each patient’s records were reviewed, as was the logbook or tracking device used, following the sequence of procedures described by the physician. All deviations from the procedure were documented.

Step 2: Patient Notification

Twenty patients who had been seen by the physician 2 weeks to 3 months earlier and who had laboratory tests ordered were chosen at random by the research assistant. A letter and a return postcard were sent to each of the patients asking whether they remembered having the test done, whether they were notified of the results, how soon they were notified, whether they were satisfied with the method used to notify them, and whether they had any suggestions for improving the notification process. No attempt was made to contact nonresponders.

Step 3: Documentation of Notification

The research assistant, using the same selection process as for Step 2, reviewed the patients’ charts for evidence of documentation that the results of tests ordered at the index visit had been communicated to the patients.

Step 4: Follow-Up Tracking

Twenty patients who had been seen by the physician 2 to 3 months earlier, who had abnormal test results, and for whom follow-up recommendation had been made, were identified from billing records and encounter forms by the research assistant. Patients asked to follow up at some point beyond the audit date were excluded. Charts were reviewed for evidence that the patients had either followed up as recommended or that the clinic had recognized their failure to follow up and had made an attempt to contact them.

Analysis

We tabulated the audit results. For each step, the 2 methods were compared with each other and with the following performance criteria: (1) >90% satisfactory accomplishment of the step; (2) for step 2, >90% patient notification rate within 2 weeks; and (3) for step 2, >90% patient satisfaction. It was possible to identify effective methods for the first 3 steps in the laboratory test management process. At that point, the research assistant went back to the practices using the more effective methods and observed the time and material costs of implementing each component of the strategy (time-motion study). Each step was directly observed and timed for 5 consecutive sets of laboratory tests for 5 patients seen on the same day. The results were averaged.

Results

Twenty-three of 24 physicians (96%) responded to the questionnaire. A summary of the results is shown in [Table 2]. Eleven of 23 physicians (48%) said that they had a method that worked well for at least 1 of the 4 management steps. Four respondents reported having an effective method for step 1 (tracking test results). All reported using either a laboratory test log, billing slips, or a computerized system. Ten respondents reported having an effective method for step 2 (patient notification of test results). Of these, 5 had their nurse call the patient with test results or discussed the results at a follow-up appointment, and 3 sent the laboratory result with a hand-written note from the physician with or without an additional explanatory form. One mailed a patient-friendly form without the actual test results, and 1 did not elaborate on his method. Ten respondents reported having an effective method for step 3 (documentation of patient notification). The method that all of them reported using for this step was documentation on the laboratory sheet. Only 3 of the respondents reported having an effective method for step 4 (follow-up tracking). Of these, 2 relied on attempts to call no-shows to their offices, and 1 did not elaborate on his method.

Practice Audits

Step 1: Test Tracking At the first practice chosen to evaluate this step, a manifest logbook was used to record all tests sent to an outside laboratory. The nurse who drew the blood was responsible for entering the information in the logbook. In-house laboratory tests were not recorded, as they were generally completed while the patient was in the office. The nurse checked the book daily and marked any outstanding test results with colored self-adhesive notes. When the results came in, the same nurse initialed and dated the entry in the book and removed the note. If the test results were not back within the expected time period (dependent on the test ordered), the nurse called the laboratory and notified the physician if the specimen needed to be obtained again. The nurse initialed laboratory test reports as they came in and took them to the physician for review.

The audit of this method revealed that 15% of the laboratory tests ordered were not recorded in the manifest logbook (the nurse said she must have forgotten to do so), and 5% of the tests sent to outside laboratories were marked received on the log but were not found in the chart. In addition, at least 1 in-house test result was missing from the charts of 15% of patients for whom these tests were ordered.

At the second practice, the physician ordered laboratory tests using a special laboratory billing slip. The nurse made and kept a copy of this slip, sending the original to the laboratory. Practice personnel recorded the laboratory tests in one of 2 logbooks, “path” and “other.” When results came back they were checked off in the logbooks. The books were checked daily (other) or monthly (path) to catch those that were missing. As results came in, the nurse checked them off the billing slip, and when all were back for a particular patient she took them to the physician along with a test results information sheet. The nurse also checked the billing slips every 3 days for most laboratory tests and every 5 to 6 days for “path” and called the laboratory when results were not back in a reasonable period of time.

The audit revealed that 100% of the laboratory tests were listed in the logbook, and 100% of the laboratory test results received were recorded. However, 15% of patients’ charts did not contain the results even though the reports were marked as having been received.

Steps 2 and 3: Notification and Documentation At the first practice chosen to evaluate these steps, the physician instructed the nurse to call all patients with abnormal results to give them instructions regarding follow-up. Patients with normal results or follow-up appointments were not contacted. Those without follow-up appointments were sent a copy of the laboratory report along with any necessary comments from the physician. Calls and mailings were generally handled on the day after the results were received by the clinic. For documentation, the nurse initialed and dated the laboratory form and placed it in the chart.

Of the 20 cards we mailed to patients, 13 were returned (65%). Of these, 92% indicated they had been notified of test results. Fifty-four percent of these had received the results within 1 week, 38% within 1 to 2 weeks, 0% within 2 to 3 weeks, and 8% within 4 weeks after the test was performed. Ninety-two percent of the respondents were satisfied with the method used to notify them of their test results, and no suggestions were offered for improvement. The audit revealed that 40% of the charts had inadequate documentation of patient notification. Of these, 50% of the test results were not initialed, and 50% were not dated.

At the second practice, the physician initialed and dated the laboratory test results form and wrote a note on it to the patient regarding the results. This was copied and mailed to the patient with a generic form describing commonly ordered tests and what they indicate. Patients could call for clarification. The original laboratory report was placed in the chart. Patients were only contacted for urgent matters or if a specimen had to be recollected. For documentation, laboratory sheets were initialed and dated by the nurse and marked “mailed” at the time they were copied and sent to the patient.

Of the 20 cards mailed to patients of the second practice, 9 were returned (45%). Of the patients who responded, 100% indicated they had been notified of test results. Seventy-five percent of these had received the results within 1 week and 25% within 1 to 2 weeks after the test was performed. All of the respondents were satisfied with the method used to notify them of their test results, and no specific suggestions were made. The results of the audit of the second practice for documentation of patient notification revealed 95% of the charts had adequate documentation of patient notification.

Step 4: Follow-Up Tracking At the first practice chosen to evaluate this step, the clinic receptionist had a list of patients scheduled each day to follow up abnormal results. If the patient did not show up, the nurse was notified and an attempt was made to contact the patient by phone or mail. A note was written in the chart that this had been done. If the situation was felt to be serious or if the patient failed to follow up after several attempts, the physician was notified. The results of the audit revealed that 40% of the charts had inadequate documentation of follow-up tracking. Several problems were identified. The longer the delay in recommended follow-up, the less well the tracking system worked. Situations in which the first attempt to contact the patient was unsuccessful were often lost to further attempts at follow-up. There also appeared to be no way to track outside consultations, and the method required a fair amount of judgment by the office staff.

At the second practice, an internal referral form was used to track patients referred within the practice for follow-up of abnormal results. When patients failed to show up for a scheduled follow-up appointment, attempts are made to contact them, and these attempts were recorded on the internal referral form. If several unsuccessful attempts were made to contact the patient, the form came back to the physician. If a patient did not show up for an outside consultation, the consultant usually notified the physician. The results of the audit revealed that 35% of the patients whose charts were reviewed had not followed up as recommended in 3 months, and 10% followed up much later than recommended. None of the charts of those patients who followed up late or not at all contained documentation of attempts to contact the patient. We did not believe that the other physicians’ practices were sufficiently different from these 2 approaches to warrant audit.

Combined Best Method

Considering effectiveness, efficiency, and ease of use, we concluded that a combination approach using the methods from the second practices audited for steps 1, 2, and 3 represented the best method that we could identify within our network for managing those steps. We were not happy with either method audited for step 4. The combined method for steps 1 to 3 is summarized in [Table 3]. The results of the time-motion studies are shown in [Table 4].

Discussion

This is the first attempt that we are aware of to describe an effective method for the comprehensive management of laboratory test results in the primary care setting. The results of the physician survey (Table 3) demonstrate the diversity of methods being used by practicing physicians. Laboratory test results are handled differently by individual practices and by physicians within the same practice. By self-report, a substantial percentage of these physicians fall short of optimal performance in this area.

The method that we propose on the basis of our findings involves a dual registry of tests ordered and returned (laboratory logbook and nurse billing slip copy) and a uniform system for patient notification and documentation. It is possible that in some practices a single-entry system may work, especially if there is a paid laboratory technician who is held accountable to manage this. However, our findings agree with 2 British studies^6,7 that also suggest that a single-entry system often does not work adequately. We believe that it is important to have a uniform system with very little decision making required on the part of staff. We also think it important to provide patients with their actual results, not just a report stating they were normal.

We were unsuccessful in identifying a method that is effective for the follow-up of abnormal laboratory test results. Our guess is that this will require some kind of tickler file system with postcards sent to patients who appear to have not followed up as suggested. It may also be that methods could be developed, such as the one described by Del Mar and Wright¹⁰ in which the outside laboratory participates in the follow-up process.

We do not claim that our best method is the only method that will work. The number of clinicians surveyed and audited was small, they all practice in the same geographic area, and many of them were trained in the same institution. The response rates on the patient audits for step 2 were only 65% and 45%, raising the possibility that those who did not respond were unhappy with the methods used. Anecdotally, however, we have subsequently implemented steps 2 and 3 in one of our University-based clinics, and they seem to work extremely well. Our nurse loves it, and we have had no patient complaints.

Telephonic methods of managing laboratory test results are now available. These allow physicians to dictate personalized messages to patients about their results. Patients access the information by calling a designated number and using their own personal identification code. Reports are generated for the physician indicating which patients called for results so that another notification method could be used for those who did not. These services typically do not address steps 1 and 4, and it was the feeling of the clinicians in our network who had tried them that the services failed for enough patients that a back-up method was necessary.

The largest portion (43%) of the cost of the method is physician time. In theory this portion is billable within the evaluation and management (E/M) codes. Most practices also charge a laboratory collection fee. Our experience has been that on average a practice can collect approximately $3.00 of an $11.00 collection fee billed to patients’ insurance. This varies by insurance carrier and does not apply to capitated systems. This must cover the costs of test result management and the venopuncture, packaging, and labeling of specimens as well. We estimate that primary care physicians who manage test results lose approximately $5.00 on each set of tests sent out on an individual patient unless they properly incorporate the costs into the E/M charge for the visit.

Since completion of this research, a number of OKPRN clinicians have adopted all or portions of our proposed method and have found that it works extremely well. A further modification has been suggested, which is supported by at least one published study.¹³ At the time when the test is obtained, patients are asked to write their own address on the envelope to be used to mail results. This assures that the address is correct and saves the nurse time later, reducing the cost by approximately $0.30 per set of tests.

Acknowledgments

We received financial support from the Presbyterian Health Foundation. We are grateful for the contributions of Ed Farrow, MD; Cary Fisher, MD; Marcia Matthews, MD; Paul Louis Preslar, DO; and the staff of the Oklahoma Center for Family Medicine Research.

Failure to notify patients of abnormal laboratory test results or to ensure appropriate follow-up can result in inferior patient care and potential malpractice liability.^1-5 Failure to document physician review and communication of test results to patients can make defense of a malpractice claim more difficult. Our review of the English-language literature identified only a few studies related to the management and reporting of laboratory test results in primary care.^6-10

Boohaker and colleagues⁹ proposed that there are 4 basic steps involved in managing laboratory test results: tracking tests until results are received; notifying patients of the results; documenting that the notification occurred; and assuring that recommended follow-up occurs for abnormal test results. They surveyed 161 attending physicians and 101 internal medicine and family practice residents at the Henry Ford Hospital in Detroit and 21 suburban primary care practices in southeast Michigan for information on the perceived importance of each step, how each of the steps was handled, and the perceived reliability of the methods used. The percentages of respondents with fair or poor methods or no method for tracking test results were: 17% for blood studies, 27% for Papanicolaou tests, 29% for mammograms, and 32% for X-rays. Approximately 50% felt that it was moderately or extremely important to notify all patients of normal test results. However, only 28% said that they accomplished this. Thirty-six percent admitted that they do not always succeed in notifying patients of abnormal test results. Often this was because patients were expected to return to the clinic soon, or the results were considered trivial. However, patient unavailability, physician forgetfulness, and time constraints were contributing factors as well. Only 55% reported that they always document patient notification of laboratory results. Another 30% said they did so most of the time. Seventy-five percent had a fair or poor method or no method for tracking patients to see if they received recommended follow-up of abnormal test results. A patient survey was conducted in the Internal Medicine Teaching Clinic at Henry Ford Hospital. Seventy-nine percent of patients surveyed wanted to be notified of all test results, whether normal or abnormal.

No one has identified and published a unified method that accomplishes each of the 4 steps effectively and efficiently. We attempted to do so by tapping the collective wisdom of clinician members of the Oklahoma Physicians Resource/Research Network (OKPRN), a network of practicing family physicians throughout Oklahoma.

Methods

Our study was conducted between August 1998 and April 1999. We sent a questionnaire to all 24 physician members of OKPRN. The questions were designed to obtain information about methods being used to address the 4 steps in the process articulated by Boohaker and coworkers⁹ and their perceived effectiveness. Two clinicians reporting effective methods for a particular step were then chosen for further analysis, with an attempt to choose practices with different methods. Four different physicians’ practices were audited, with several physicians reporting effective methods for more than 1 step. A single research assistant was sent to each practice to perform the audits. During this visit, a 5-minute open-ended interview was also conducted with the physician or the physician’s staff to more precisely describe the method used for the management step being evaluated.

Step 1: Test Tracking

Twenty patients who had been seen by the physician 2 to 6 weeks earlier, who had laboratory tests ordered during that visit, and whose charts were available were chosen at random by the research assistant ([Table 1]). Each patient’s records were reviewed, as was the logbook or tracking device used, following the sequence of procedures described by the physician. All deviations from the procedure were documented.

Step 2: Patient Notification

Twenty patients who had been seen by the physician 2 weeks to 3 months earlier and who had laboratory tests ordered were chosen at random by the research assistant. A letter and a return postcard were sent to each of the patients asking whether they remembered having the test done, whether they were notified of the results, how soon they were notified, whether they were satisfied with the method used to notify them, and whether they had any suggestions for improving the notification process. No attempt was made to contact nonresponders.

Step 3: Documentation of Notification

The research assistant, using the same selection process as for Step 2, reviewed the patients’ charts for evidence of documentation that the results of tests ordered at the index visit had been communicated to the patients.

Step 4: Follow-Up Tracking

Twenty patients who had been seen by the physician 2 to 3 months earlier, who had abnormal test results, and for whom follow-up recommendation had been made, were identified from billing records and encounter forms by the research assistant. Patients asked to follow up at some point beyond the audit date were excluded. Charts were reviewed for evidence that the patients had either followed up as recommended or that the clinic had recognized their failure to follow up and had made an attempt to contact them.

Analysis

We tabulated the audit results. For each step, the 2 methods were compared with each other and with the following performance criteria: (1) >90% satisfactory accomplishment of the step; (2) for step 2, >90% patient notification rate within 2 weeks; and (3) for step 2, >90% patient satisfaction. It was possible to identify effective methods for the first 3 steps in the laboratory test management process. At that point, the research assistant went back to the practices using the more effective methods and observed the time and material costs of implementing each component of the strategy (time-motion study). Each step was directly observed and timed for 5 consecutive sets of laboratory tests for 5 patients seen on the same day. The results were averaged.

Results

Twenty-three of 24 physicians (96%) responded to the questionnaire. A summary of the results is shown in [Table 2]. Eleven of 23 physicians (48%) said that they had a method that worked well for at least 1 of the 4 management steps. Four respondents reported having an effective method for step 1 (tracking test results). All reported using either a laboratory test log, billing slips, or a computerized system. Ten respondents reported having an effective method for step 2 (patient notification of test results). Of these, 5 had their nurse call the patient with test results or discussed the results at a follow-up appointment, and 3 sent the laboratory result with a hand-written note from the physician with or without an additional explanatory form. One mailed a patient-friendly form without the actual test results, and 1 did not elaborate on his method. Ten respondents reported having an effective method for step 3 (documentation of patient notification). The method that all of them reported using for this step was documentation on the laboratory sheet. Only 3 of the respondents reported having an effective method for step 4 (follow-up tracking). Of these, 2 relied on attempts to call no-shows to their offices, and 1 did not elaborate on his method.

Practice Audits

Step 1: Test Tracking At the first practice chosen to evaluate this step, a manifest logbook was used to record all tests sent to an outside laboratory. The nurse who drew the blood was responsible for entering the information in the logbook. In-house laboratory tests were not recorded, as they were generally completed while the patient was in the office. The nurse checked the book daily and marked any outstanding test results with colored self-adhesive notes. When the results came in, the same nurse initialed and dated the entry in the book and removed the note. If the test results were not back within the expected time period (dependent on the test ordered), the nurse called the laboratory and notified the physician if the specimen needed to be obtained again. The nurse initialed laboratory test reports as they came in and took them to the physician for review.

The audit of this method revealed that 15% of the laboratory tests ordered were not recorded in the manifest logbook (the nurse said she must have forgotten to do so), and 5% of the tests sent to outside laboratories were marked received on the log but were not found in the chart. In addition, at least 1 in-house test result was missing from the charts of 15% of patients for whom these tests were ordered.

At the second practice, the physician ordered laboratory tests using a special laboratory billing slip. The nurse made and kept a copy of this slip, sending the original to the laboratory. Practice personnel recorded the laboratory tests in one of 2 logbooks, “path” and “other.” When results came back they were checked off in the logbooks. The books were checked daily (other) or monthly (path) to catch those that were missing. As results came in, the nurse checked them off the billing slip, and when all were back for a particular patient she took them to the physician along with a test results information sheet. The nurse also checked the billing slips every 3 days for most laboratory tests and every 5 to 6 days for “path” and called the laboratory when results were not back in a reasonable period of time.

The audit revealed that 100% of the laboratory tests were listed in the logbook, and 100% of the laboratory test results received were recorded. However, 15% of patients’ charts did not contain the results even though the reports were marked as having been received.

Steps 2 and 3: Notification and Documentation At the first practice chosen to evaluate these steps, the physician instructed the nurse to call all patients with abnormal results to give them instructions regarding follow-up. Patients with normal results or follow-up appointments were not contacted. Those without follow-up appointments were sent a copy of the laboratory report along with any necessary comments from the physician. Calls and mailings were generally handled on the day after the results were received by the clinic. For documentation, the nurse initialed and dated the laboratory form and placed it in the chart.

Of the 20 cards we mailed to patients, 13 were returned (65%). Of these, 92% indicated they had been notified of test results. Fifty-four percent of these had received the results within 1 week, 38% within 1 to 2 weeks, 0% within 2 to 3 weeks, and 8% within 4 weeks after the test was performed. Ninety-two percent of the respondents were satisfied with the method used to notify them of their test results, and no suggestions were offered for improvement. The audit revealed that 40% of the charts had inadequate documentation of patient notification. Of these, 50% of the test results were not initialed, and 50% were not dated.

At the second practice, the physician initialed and dated the laboratory test results form and wrote a note on it to the patient regarding the results. This was copied and mailed to the patient with a generic form describing commonly ordered tests and what they indicate. Patients could call for clarification. The original laboratory report was placed in the chart. Patients were only contacted for urgent matters or if a specimen had to be recollected. For documentation, laboratory sheets were initialed and dated by the nurse and marked “mailed” at the time they were copied and sent to the patient.

Of the 20 cards mailed to patients of the second practice, 9 were returned (45%). Of the patients who responded, 100% indicated they had been notified of test results. Seventy-five percent of these had received the results within 1 week and 25% within 1 to 2 weeks after the test was performed. All of the respondents were satisfied with the method used to notify them of their test results, and no specific suggestions were made. The results of the audit of the second practice for documentation of patient notification revealed 95% of the charts had adequate documentation of patient notification.

Step 4: Follow-Up Tracking At the first practice chosen to evaluate this step, the clinic receptionist had a list of patients scheduled each day to follow up abnormal results. If the patient did not show up, the nurse was notified and an attempt was made to contact the patient by phone or mail. A note was written in the chart that this had been done. If the situation was felt to be serious or if the patient failed to follow up after several attempts, the physician was notified. The results of the audit revealed that 40% of the charts had inadequate documentation of follow-up tracking. Several problems were identified. The longer the delay in recommended follow-up, the less well the tracking system worked. Situations in which the first attempt to contact the patient was unsuccessful were often lost to further attempts at follow-up. There also appeared to be no way to track outside consultations, and the method required a fair amount of judgment by the office staff.

At the second practice, an internal referral form was used to track patients referred within the practice for follow-up of abnormal results. When patients failed to show up for a scheduled follow-up appointment, attempts are made to contact them, and these attempts were recorded on the internal referral form. If several unsuccessful attempts were made to contact the patient, the form came back to the physician. If a patient did not show up for an outside consultation, the consultant usually notified the physician. The results of the audit revealed that 35% of the patients whose charts were reviewed had not followed up as recommended in 3 months, and 10% followed up much later than recommended. None of the charts of those patients who followed up late or not at all contained documentation of attempts to contact the patient. We did not believe that the other physicians’ practices were sufficiently different from these 2 approaches to warrant audit.

Combined Best Method

Considering effectiveness, efficiency, and ease of use, we concluded that a combination approach using the methods from the second practices audited for steps 1, 2, and 3 represented the best method that we could identify within our network for managing those steps. We were not happy with either method audited for step 4. The combined method for steps 1 to 3 is summarized in [Table 3]. The results of the time-motion studies are shown in [Table 4].

Discussion

This is the first attempt that we are aware of to describe an effective method for the comprehensive management of laboratory test results in the primary care setting. The results of the physician survey (Table 3) demonstrate the diversity of methods being used by practicing physicians. Laboratory test results are handled differently by individual practices and by physicians within the same practice. By self-report, a substantial percentage of these physicians fall short of optimal performance in this area.

The method that we propose on the basis of our findings involves a dual registry of tests ordered and returned (laboratory logbook and nurse billing slip copy) and a uniform system for patient notification and documentation. It is possible that in some practices a single-entry system may work, especially if there is a paid laboratory technician who is held accountable to manage this. However, our findings agree with 2 British studies^6,7 that also suggest that a single-entry system often does not work adequately. We believe that it is important to have a uniform system with very little decision making required on the part of staff. We also think it important to provide patients with their actual results, not just a report stating they were normal.

We were unsuccessful in identifying a method that is effective for the follow-up of abnormal laboratory test results. Our guess is that this will require some kind of tickler file system with postcards sent to patients who appear to have not followed up as suggested. It may also be that methods could be developed, such as the one described by Del Mar and Wright¹⁰ in which the outside laboratory participates in the follow-up process.

We do not claim that our best method is the only method that will work. The number of clinicians surveyed and audited was small, they all practice in the same geographic area, and many of them were trained in the same institution. The response rates on the patient audits for step 2 were only 65% and 45%, raising the possibility that those who did not respond were unhappy with the methods used. Anecdotally, however, we have subsequently implemented steps 2 and 3 in one of our University-based clinics, and they seem to work extremely well. Our nurse loves it, and we have had no patient complaints.

Telephonic methods of managing laboratory test results are now available. These allow physicians to dictate personalized messages to patients about their results. Patients access the information by calling a designated number and using their own personal identification code. Reports are generated for the physician indicating which patients called for results so that another notification method could be used for those who did not. These services typically do not address steps 1 and 4, and it was the feeling of the clinicians in our network who had tried them that the services failed for enough patients that a back-up method was necessary.

The largest portion (43%) of the cost of the method is physician time. In theory this portion is billable within the evaluation and management (E/M) codes. Most practices also charge a laboratory collection fee. Our experience has been that on average a practice can collect approximately $3.00 of an $11.00 collection fee billed to patients’ insurance. This varies by insurance carrier and does not apply to capitated systems. This must cover the costs of test result management and the venopuncture, packaging, and labeling of specimens as well. We estimate that primary care physicians who manage test results lose approximately $5.00 on each set of tests sent out on an individual patient unless they properly incorporate the costs into the E/M charge for the visit.

Since completion of this research, a number of OKPRN clinicians have adopted all or portions of our proposed method and have found that it works extremely well. A further modification has been suggested, which is supported by at least one published study.¹³ At the time when the test is obtained, patients are asked to write their own address on the envelope to be used to mail results. This assures that the address is correct and saves the nurse time later, reducing the cost by approximately $0.30 per set of tests.

Acknowledgments

We received financial support from the Presbyterian Health Foundation. We are grateful for the contributions of Ed Farrow, MD; Cary Fisher, MD; Marcia Matthews, MD; Paul Louis Preslar, DO; and the staff of the Oklahoma Center for Family Medicine Research.

Since the early 1970s, concerns have been raised about the adverse consequences associated with diagnostic labeling. Some of the earliest research examined parental responses to childhood cardiac diagnoses and misdiagnoses.^1,2 Subsequent studies extended those findings to the results of newborn screening tests,^3,4 preschool developmental screening,⁵ and sickle cell disease screening.⁶ The greatest body of published research in this regard involves the diagnosis of hypertension. Haynes and colleagues⁷ first observed that Canadian steel mill workers found to have hypertension through workplace screening had increased absenteeism from work that persisted for at least 4 years, and Johnston and coworkers⁸ found that this diagnosis was associated with lower mean annual incomes at 5 years postscreen. Subsequently, researchers have consistently found some patients, after being told that their blood pressures are too high, perceive their overall health to be worse and report more depressive symptoms and lower quality of life.⁹ In 1 study, even close relatives of patients with hypertension seemed to be adversely affected.¹⁰

To our knowledge, no one has looked at the effect of the diagnosis of hypertension on patients’ perceptions of their ability to recover from completely unrelated illnesses, such as infections and injuries. Such information may be important, particularly if labeling may affect health care utilization.

Methods

We recruited 22 patients waiting to see family practice residents and faculty in a family medicine clinic at an academic medical center during the summer of 1995. To participate, patients had to be aged 21 years or older and not mentally retarded, severely depressed, schizophrenic, demented, or acutely ill. We constructed the sample to include 11 patients with diagnosed hypertension within the past 5 years and 11 with no diagnosis of hypertension.

After signing informed consent, each participant was asked questions designed to elicit demographic information, recent experience with upper respiratory tract infection (URI), urinary tract infection (UTI), and ankle sprain; self-rated overall health using a single COOP chart; and estimates of how long it would take for them to recover from a URI, a UTI, and an ankle sprain now and 5 years ago. These conditions were chosen because they have no known pathophysiologic relationship to hypertension and represent 3 different kinds of acute illness: a self-limited viral infection, a bacterial infection that is usually treated with antibiotics, and an acute musculoskeletal injury. Medical records were reviewed for possible confounders, such as cigarette smoking, chronic lung disease, asthma, allergic rhinitis, diabetes, congestive heart failure, and payer source.

We entered the data into a standard statistical software program (Statistix, Analytical Software, Tallahassee, Fla) and used the chi-square test to make comparisons between the hypertensive and nonhypertensive groups with respect to age, sex, race, marital status, educational attainment, and self-rated health. Mean estimated times required to recover from each of the 3 acute conditions, both current and past, were compared using Student t tests.

We considered 6 separate linear regression models with estimated time to resolution for each of the 3 conditions at present and in the past as the dependent variables. Age, sex, race, education, and diagnosis group were initially entered into each model and removed 1 at a time until the best model was obtained in each case. We defined the best model as the one associated with an overall P value <.05 that maximized R2 while minimizing the number of variables with individual P values >.05.

Results

The characteristics of patients in each of the 2 groups are shown in Table 1. Patients with hypertension were likely to be older, male, African American, and less educated. However, none of these differences was statistically significant. The control group of nonhypertensives had more comorbidities, were more likely to smoke cigarettes, and had worse perceived health (also nonsignificant). There was no difference in payer source. In the control group, 5 of 11 reported having a URI within the past 6 months, 1 had experienced a UTI, and 2 had ankle sprains. In the hypertensive group, 3 of 11 reported a URI, none had a UTI, and 1 had a sprained ankle.

Table 2 shows the estimates of time to resolution of a URI, a UTI, and an ankle sprain both now and 5 years ago. Patients with hypertension estimated that it took them almost twice as long, on average, to recover from a URI both in the present and in the past. This was a significant difference. They also seemed to believe that it would take them longer to heal an ankle sprain. This, however, did not reach statistical significance, primarily because of the substantial variability of the estimates. There appeared to be no difference at all and very little variability in the perceived times for recovery from a UTI.

Table 3 shows the linear regression models for URI, current and past. Both URI models included diagnosis group as a significant predictor of estimated recovery time. No satisfactory regression models could be constructed for present or past UTI, or for past ankle sprain. The present ankle sprain model did not include hypertensive status as a predictor.

We constructed a model for estimated time to recover from a URI in the present as a function of the demographic variables (age, sex, race, and education), estimated recovery time for URI in the past, UTI in the past and present, and ankle sprain in the past and present . A model that included past recovery from URI (P <.001), present (P=.004) and past (P <.001) recovery from ankle sprain, and hypertension diagnosis (P=.008) explained 93% of the variability.

Discussion

Despite the small sample size, the results of our study are striking. Not only did patients with hypertension estimate that it took them twice as long to recover from a URI now, but they believed that it had taken them twice as long to recover even before receiving the diagnosis. These findings seem consistent with previous research on the adverse effects of labeling. Alternative explanations for our results include an unknown biological association between hypertension and the ability to fight viral infections, and other unknown confounders. For example, patients with hypertension may be more attuned to the medical system and their own health status and therefore more accurate in their estimates of recovery times.

It is more difficult to interpret the data about ankle sprains. The standard deviations of the estimates were large, and age, race, and education were significant predictors for ankle sprain in the present. Study participants were less likely to have experienced an ankle sprain than a URI, and ankle sprain severity is more likely to range from mild to severe, making estimation of recovery time more difficult.

There was no effect of hypertension status on perceived time to recover from a UTI. Estimates of time to recovery from UTI were not correlated with estimated recovery times for either of the other 2 conditions. This may be because UTIs are believed to be predictably cured by antibiotics regardless of a person’s general medical condition.

Although there were baseline differences between the 2 groups, none was statistically significant. Controlling for these differences did not eliminate the significant effect of a diagnosis of hypertension on a patient’s perceived time to recovery from a URI. We conclude that being given the diagnosis of hypertension may change patients’ perceptions of physical resiliency.

Although this apparent adverse effect of labeling is troubling, we have no direct evidence that it has an adverse impact on health or health care utilization. If, however, patients with hypertension believe that it will take them twice as long to recover from a URI, they may be more likely to seek medical attention or use medications for URI episodes. Consistent with the findings of Haynes and colleagues,⁷ they may stay out of work longer.

Is there a way to diagnose hypertension and lower blood pressure without stigmatizing the patient? Do the potential benefits of giving a diagnosis of hypertension outweigh the hazards? We suggest further research to confirm our findings and to explore these questions.

Since the early 1970s, concerns have been raised about the adverse consequences associated with diagnostic labeling. Some of the earliest research examined parental responses to childhood cardiac diagnoses and misdiagnoses.^1,2 Subsequent studies extended those findings to the results of newborn screening tests,^3,4 preschool developmental screening,⁵ and sickle cell disease screening.⁶ The greatest body of published research in this regard involves the diagnosis of hypertension. Haynes and colleagues⁷ first observed that Canadian steel mill workers found to have hypertension through workplace screening had increased absenteeism from work that persisted for at least 4 years, and Johnston and coworkers⁸ found that this diagnosis was associated with lower mean annual incomes at 5 years postscreen. Subsequently, researchers have consistently found some patients, after being told that their blood pressures are too high, perceive their overall health to be worse and report more depressive symptoms and lower quality of life.⁹ In 1 study, even close relatives of patients with hypertension seemed to be adversely affected.¹⁰

To our knowledge, no one has looked at the effect of the diagnosis of hypertension on patients’ perceptions of their ability to recover from completely unrelated illnesses, such as infections and injuries. Such information may be important, particularly if labeling may affect health care utilization.

Methods

We recruited 22 patients waiting to see family practice residents and faculty in a family medicine clinic at an academic medical center during the summer of 1995. To participate, patients had to be aged 21 years or older and not mentally retarded, severely depressed, schizophrenic, demented, or acutely ill. We constructed the sample to include 11 patients with diagnosed hypertension within the past 5 years and 11 with no diagnosis of hypertension.

After signing informed consent, each participant was asked questions designed to elicit demographic information, recent experience with upper respiratory tract infection (URI), urinary tract infection (UTI), and ankle sprain; self-rated overall health using a single COOP chart; and estimates of how long it would take for them to recover from a URI, a UTI, and an ankle sprain now and 5 years ago. These conditions were chosen because they have no known pathophysiologic relationship to hypertension and represent 3 different kinds of acute illness: a self-limited viral infection, a bacterial infection that is usually treated with antibiotics, and an acute musculoskeletal injury. Medical records were reviewed for possible confounders, such as cigarette smoking, chronic lung disease, asthma, allergic rhinitis, diabetes, congestive heart failure, and payer source.

We entered the data into a standard statistical software program (Statistix, Analytical Software, Tallahassee, Fla) and used the chi-square test to make comparisons between the hypertensive and nonhypertensive groups with respect to age, sex, race, marital status, educational attainment, and self-rated health. Mean estimated times required to recover from each of the 3 acute conditions, both current and past, were compared using Student t tests.

We considered 6 separate linear regression models with estimated time to resolution for each of the 3 conditions at present and in the past as the dependent variables. Age, sex, race, education, and diagnosis group were initially entered into each model and removed 1 at a time until the best model was obtained in each case. We defined the best model as the one associated with an overall P value <.05 that maximized R2 while minimizing the number of variables with individual P values >.05.

Results

The characteristics of patients in each of the 2 groups are shown in Table 1. Patients with hypertension were likely to be older, male, African American, and less educated. However, none of these differences was statistically significant. The control group of nonhypertensives had more comorbidities, were more likely to smoke cigarettes, and had worse perceived health (also nonsignificant). There was no difference in payer source. In the control group, 5 of 11 reported having a URI within the past 6 months, 1 had experienced a UTI, and 2 had ankle sprains. In the hypertensive group, 3 of 11 reported a URI, none had a UTI, and 1 had a sprained ankle.

Table 2 shows the estimates of time to resolution of a URI, a UTI, and an ankle sprain both now and 5 years ago. Patients with hypertension estimated that it took them almost twice as long, on average, to recover from a URI both in the present and in the past. This was a significant difference. They also seemed to believe that it would take them longer to heal an ankle sprain. This, however, did not reach statistical significance, primarily because of the substantial variability of the estimates. There appeared to be no difference at all and very little variability in the perceived times for recovery from a UTI.

Table 3 shows the linear regression models for URI, current and past. Both URI models included diagnosis group as a significant predictor of estimated recovery time. No satisfactory regression models could be constructed for present or past UTI, or for past ankle sprain. The present ankle sprain model did not include hypertensive status as a predictor.

We constructed a model for estimated time to recover from a URI in the present as a function of the demographic variables (age, sex, race, and education), estimated recovery time for URI in the past, UTI in the past and present, and ankle sprain in the past and present . A model that included past recovery from URI (P <.001), present (P=.004) and past (P <.001) recovery from ankle sprain, and hypertension diagnosis (P=.008) explained 93% of the variability.

Discussion

Despite the small sample size, the results of our study are striking. Not only did patients with hypertension estimate that it took them twice as long to recover from a URI now, but they believed that it had taken them twice as long to recover even before receiving the diagnosis. These findings seem consistent with previous research on the adverse effects of labeling. Alternative explanations for our results include an unknown biological association between hypertension and the ability to fight viral infections, and other unknown confounders. For example, patients with hypertension may be more attuned to the medical system and their own health status and therefore more accurate in their estimates of recovery times.

It is more difficult to interpret the data about ankle sprains. The standard deviations of the estimates were large, and age, race, and education were significant predictors for ankle sprain in the present. Study participants were less likely to have experienced an ankle sprain than a URI, and ankle sprain severity is more likely to range from mild to severe, making estimation of recovery time more difficult.

There was no effect of hypertension status on perceived time to recover from a UTI. Estimates of time to recovery from UTI were not correlated with estimated recovery times for either of the other 2 conditions. This may be because UTIs are believed to be predictably cured by antibiotics regardless of a person’s general medical condition.

Although there were baseline differences between the 2 groups, none was statistically significant. Controlling for these differences did not eliminate the significant effect of a diagnosis of hypertension on a patient’s perceived time to recovery from a URI. We conclude that being given the diagnosis of hypertension may change patients’ perceptions of physical resiliency.

Although this apparent adverse effect of labeling is troubling, we have no direct evidence that it has an adverse impact on health or health care utilization. If, however, patients with hypertension believe that it will take them twice as long to recover from a URI, they may be more likely to seek medical attention or use medications for URI episodes. Consistent with the findings of Haynes and colleagues,⁷ they may stay out of work longer.

Is there a way to diagnose hypertension and lower blood pressure without stigmatizing the patient? Do the potential benefits of giving a diagnosis of hypertension outweigh the hazards? We suggest further research to confirm our findings and to explore these questions.

Since the early 1970s, concerns have been raised about the adverse consequences associated with diagnostic labeling. Some of the earliest research examined parental responses to childhood cardiac diagnoses and misdiagnoses.^1,2 Subsequent studies extended those findings to the results of newborn screening tests,^3,4 preschool developmental screening,⁵ and sickle cell disease screening.⁶ The greatest body of published research in this regard involves the diagnosis of hypertension. Haynes and colleagues⁷ first observed that Canadian steel mill workers found to have hypertension through workplace screening had increased absenteeism from work that persisted for at least 4 years, and Johnston and coworkers⁸ found that this diagnosis was associated with lower mean annual incomes at 5 years postscreen. Subsequently, researchers have consistently found some patients, after being told that their blood pressures are too high, perceive their overall health to be worse and report more depressive symptoms and lower quality of life.⁹ In 1 study, even close relatives of patients with hypertension seemed to be adversely affected.¹⁰

To our knowledge, no one has looked at the effect of the diagnosis of hypertension on patients’ perceptions of their ability to recover from completely unrelated illnesses, such as infections and injuries. Such information may be important, particularly if labeling may affect health care utilization.

Methods

We recruited 22 patients waiting to see family practice residents and faculty in a family medicine clinic at an academic medical center during the summer of 1995. To participate, patients had to be aged 21 years or older and not mentally retarded, severely depressed, schizophrenic, demented, or acutely ill. We constructed the sample to include 11 patients with diagnosed hypertension within the past 5 years and 11 with no diagnosis of hypertension.

After signing informed consent, each participant was asked questions designed to elicit demographic information, recent experience with upper respiratory tract infection (URI), urinary tract infection (UTI), and ankle sprain; self-rated overall health using a single COOP chart; and estimates of how long it would take for them to recover from a URI, a UTI, and an ankle sprain now and 5 years ago. These conditions were chosen because they have no known pathophysiologic relationship to hypertension and represent 3 different kinds of acute illness: a self-limited viral infection, a bacterial infection that is usually treated with antibiotics, and an acute musculoskeletal injury. Medical records were reviewed for possible confounders, such as cigarette smoking, chronic lung disease, asthma, allergic rhinitis, diabetes, congestive heart failure, and payer source.

We entered the data into a standard statistical software program (Statistix, Analytical Software, Tallahassee, Fla) and used the chi-square test to make comparisons between the hypertensive and nonhypertensive groups with respect to age, sex, race, marital status, educational attainment, and self-rated health. Mean estimated times required to recover from each of the 3 acute conditions, both current and past, were compared using Student t tests.

We considered 6 separate linear regression models with estimated time to resolution for each of the 3 conditions at present and in the past as the dependent variables. Age, sex, race, education, and diagnosis group were initially entered into each model and removed 1 at a time until the best model was obtained in each case. We defined the best model as the one associated with an overall P value <.05 that maximized R2 while minimizing the number of variables with individual P values >.05.

Results

The characteristics of patients in each of the 2 groups are shown in Table 1. Patients with hypertension were likely to be older, male, African American, and less educated. However, none of these differences was statistically significant. The control group of nonhypertensives had more comorbidities, were more likely to smoke cigarettes, and had worse perceived health (also nonsignificant). There was no difference in payer source. In the control group, 5 of 11 reported having a URI within the past 6 months, 1 had experienced a UTI, and 2 had ankle sprains. In the hypertensive group, 3 of 11 reported a URI, none had a UTI, and 1 had a sprained ankle.

Table 2 shows the estimates of time to resolution of a URI, a UTI, and an ankle sprain both now and 5 years ago. Patients with hypertension estimated that it took them almost twice as long, on average, to recover from a URI both in the present and in the past. This was a significant difference. They also seemed to believe that it would take them longer to heal an ankle sprain. This, however, did not reach statistical significance, primarily because of the substantial variability of the estimates. There appeared to be no difference at all and very little variability in the perceived times for recovery from a UTI.

Table 3 shows the linear regression models for URI, current and past. Both URI models included diagnosis group as a significant predictor of estimated recovery time. No satisfactory regression models could be constructed for present or past UTI, or for past ankle sprain. The present ankle sprain model did not include hypertensive status as a predictor.

We constructed a model for estimated time to recover from a URI in the present as a function of the demographic variables (age, sex, race, and education), estimated recovery time for URI in the past, UTI in the past and present, and ankle sprain in the past and present . A model that included past recovery from URI (P <.001), present (P=.004) and past (P <.001) recovery from ankle sprain, and hypertension diagnosis (P=.008) explained 93% of the variability.

Discussion

Despite the small sample size, the results of our study are striking. Not only did patients with hypertension estimate that it took them twice as long to recover from a URI now, but they believed that it had taken them twice as long to recover even before receiving the diagnosis. These findings seem consistent with previous research on the adverse effects of labeling. Alternative explanations for our results include an unknown biological association between hypertension and the ability to fight viral infections, and other unknown confounders. For example, patients with hypertension may be more attuned to the medical system and their own health status and therefore more accurate in their estimates of recovery times.

It is more difficult to interpret the data about ankle sprains. The standard deviations of the estimates were large, and age, race, and education were significant predictors for ankle sprain in the present. Study participants were less likely to have experienced an ankle sprain than a URI, and ankle sprain severity is more likely to range from mild to severe, making estimation of recovery time more difficult.

There was no effect of hypertension status on perceived time to recover from a UTI. Estimates of time to recovery from UTI were not correlated with estimated recovery times for either of the other 2 conditions. This may be because UTIs are believed to be predictably cured by antibiotics regardless of a person’s general medical condition.

Although there were baseline differences between the 2 groups, none was statistically significant. Controlling for these differences did not eliminate the significant effect of a diagnosis of hypertension on a patient’s perceived time to recovery from a URI. We conclude that being given the diagnosis of hypertension may change patients’ perceptions of physical resiliency.

Although this apparent adverse effect of labeling is troubling, we have no direct evidence that it has an adverse impact on health or health care utilization. If, however, patients with hypertension believe that it will take them twice as long to recover from a URI, they may be more likely to seek medical attention or use medications for URI episodes. Consistent with the findings of Haynes and colleagues,⁷ they may stay out of work longer.

Is there a way to diagnose hypertension and lower blood pressure without stigmatizing the patient? Do the potential benefits of giving a diagnosis of hypertension outweigh the hazards? We suggest further research to confirm our findings and to explore these questions.

Too often the questions of basic biomedical research have been mistaken to represent the critical scope of all medical research, and traditional laboratory methods have been seen as necessary and sufficient methods for understanding human health and illness. As a result, approximately 90% of National Institutes of Health (NIH) funding is spent on research within the traditional biomedical sciences (anatomy, biochemistry, genetics, microbiology, molecular biology, physiology, and so forth). The smaller amount of federal funding available for clinical research has been spent primarily on specific disease entities, such as cancer and heart disease. These funding decisions have resulted in the neglect of a large proportion of the problems and issues that confront primary care physicians and their patients.

Clinical research

Perhaps not surprisingly in an era of academic medical centers’ increasing dependence on clinical revenues, the amount of clinical research and the number of clinical investigators have decreased. It has been difficult to track the magnitude of this trend, however, because a concise definition of clinical research is lacking. To address this concern, a national consensus-building conference, the Clinical Research Summit, was held in November 1998, and a working definition was adopted.¹ Clinical research was defined as “a component of medical and health research intended to produce knowledge valuable for understanding human disease, preventing and treating illness, and promoting health.” Attendees further agreed that “clinical research embraces a continuum of studies involving interactions with patients, diagnostic clinical materials or data, or populations in any of the following categories: (1) disease mechanisms; (2) bidirectional integrative research; (3) clinical knowledge, detection, diagnosis, and natural history of disease; (4) therapeutic interventions including clinical trials of drugs, biologics devices, and instruments; (5) prevention (primary and secondary) and health promotion; (6) behavioral research; (7) health services research, including outcomes and cost effectiveness; (8) epidemiology; and (9) community-based and managed care trials.”

Primary care research

In the broader context of clinical research, primary care research remains a fledgling enterprise with many institutions and funding agencies still having difficulty understanding what it includes and what its potential value might be despite a considerable body of supportive literature. The need for primary care research,^2-4 the research traditions of relevance to primary care and the scope of its methods,⁵ and efforts to elevate primary care research to a national priority^6-8 have been documented. However, only 0.4% of NIH funding (up from 0.3% in 1984) and only 4% of the funding from the Agency for Health Care Policy and Research (down from 4.4% in 1984) goes to departments of family medicine.⁹ Both private (eg, the American Academy of Family Physicians) and public (eg, the Agency for Health Care Policy and Research) organizations have formulated working descriptions of primary care research. Nonetheless, primary care researchers still lack a coherent definition with which most can agree. We believe it is particularly important to address this problem now, during a time of concern about and advocacy for clinical research. The definition and examples provided here should be useful for those who are trying to explain the importance of this work to academic administrators and funding agencies.

Primary care has most recently been defined as “the provision of integrated, accessible, health-care services by clinicians that are accountable for addressing a large majority of personal health-care needs, developing a sustained partnership with patients, and practicing within the context of family and community.”⁶ This definition was debated andwas based on a comprehensive literature review and multidisciplinary input. It has been widely adopted since its publication and offers a sensible, if not perfect, definition to guide research in primary care. One of its most important aspects is the concept that primary care is a function that does not necessarily belong to a particular discipline but is dependent on knowledge from many sources. On the basis of this definition, primary care research can be as defined as research directed toward the better understanding and practice of the primary care function.

Categories of research

Primary care research has traditionally included studies that fall into the following overlapping categories.

Theoretical and Methodologic Research. This category includes the development and testing of theoretical models, operational definitions, and measurement tools relevant to the primary care function. Included in this category are such things as: classification systems designed to capture the phenomena of primary care; ways to measure concepts, such as integrated, accessible, and accountable care; methods for distinguishing the separate and combined effects of primary care on individuals, family units, or the community; and ways to observe and measure important relationships (eg, the physician-patient relationship) and their impact on outcomes. Other investigations involve generating and testing alternative conceptualizations of the tasks and methods of primary care (eg, goal-directed care, family-centered care) and efforts to expand the methods available to primary care researchers (eg, mixed methods research, complexity theories).

Health Care Research. This is the large and important area of primary care research focused on direct investigation of the primary care function itself. Questions directed toward improving the quality and effectiveness of primary care practices fit in this category. Although health care research has historically emphasized the analysis of large data sets, primary care research often employs both qualitative and quantitative methods to examine relatively small numbers of individual physicians and practices to determine which methods seem to work best.

Clinical Research. Studies in this category are focused directly on the effects of the primary care function on patients. It includes research on the factors that determine why patients become ill and seek medical attention, the meanings of presenting symptoms and signs in a primary care setting, the most effective and efficient diagnostic and treatment strategies, and the natural histories of health problems with and without intervention. The outcome measures used are often those that are directly meaningful to patients, such as quality of life, mortality, health of the family unit, and cost and convenience of care (outcomes research).

Health Systems Research. This category extends the first 3 to the larger systems level. It encompasses educational research, research on dissemination and adoption of new discoveries, implementation of quality improvement systems into primary care settings, and health policy.

Three examples

Night Sweats. Many diseases and health states are thought to beassociated with night sweats (eg, tuberculosis, menopause, nocturnal hypoglycemia in people with diabetes, autoimmune diseases, malignancies, and medications). However, for the primary care physician who is faced with an undiagnosed patient with night sweats, the following questions arise that currently lack answers: (1) What are the incidence and prevalence of night sweats in a primary care patient population? How often does this symptom go unreported? (clinical); (2) What are the most likely causes of night sweats in a patient presenting in a primary care setting? (clinical); (3) What is the most effective and efficient path to the correct diagnosis? (clinical); (4) What is the natural history of idiopathic night sweats in otherwise healthy patients? (clinical); (5) How much of an impact do night sweats have on the quality of life of patients and their bed partners? (clinical); (6) If common and significant but underrecognized, how can clinicians reorganize their assessment methods to systematically screen patients for night sweats? (health care); and (7) If night sweats are an important symptom of potentially serious disease, how can the population be educated to pay attention to and report them? (health systems).

Cognitive Impairment. Cognitive impairment is a prevalent and serious problem for older people. Early diagnosis and treatment are becoming more important as effective treatment strategies have become available. To provide optimal care for these patients, primary care physicians need answers to the following questions:¹¹ (1) Is systematic formal screening for cognitive impairment necessary in primary care settings in which patients and their families are well known by their physicians and staff and are seen frequently? (clinical); (2) Does earlier detection result in better outcomes? (clinical); (3) What are the outcomes of greatest importance, and how can they best be measured? (theoretical/methodologic); (4) What would it cost to screen all patients [over] older than a specified age? (health systems); (5) Should we train primary care physicians to complete the evaluation and direct treatment, or should patients who screen positive be seen by a neurologist, geriatrician, or psychiatrist or psychologist? (health care); and (6) How can caregiver education programs, which have been shown to delay institutionalization and save money, be organized and funded? (health systems).

Laboratory Test Results

Another set of questions involves laboratory test results. What is the best way to manage (tracking, notification, documentation, follow-up) laboratory test results in a primary care setting? Before this health care question can be answered, several others must be addressed: (1) What is meant by “best”? How should patient preference, physician preference, cost efficiency, and legal requirements be balanced? (theoretical/methodologic); (2) How would we answer such a question even if “best” could be adequately defined? (theoretical/methodologic); and (3) Assuming that an answer can be found, what is the most efficient way to help clinicians incorporate the method into their practices? (health systems).

Education for the future

Primary care research is directed toward the better understanding and practice of the primary care function. It is further distinguished from other types of research by an emphasis on effectiveness rather than efficacy. It is often immediately applicable to primary care practice and widely generalizable.

The science base of medicine, including primary care, has improved significantly during the 20th century. Overall, however, primary care has been neglected because it was not seen as requiring an intellectual engine. The mismatch between the focus of research efforts to date and the need for research in primary care can be understood to derive from an unbalanced effort focused on understanding specific diseases and molecular mechanisms. To improve primary care, a more robust enterprise embracing the full scope of research in this setting is required. We hope that our attempt to define and describe the scope of primary care research will help its advocates educate those who are able to make this happen.

Too often the questions of basic biomedical research have been mistaken to represent the critical scope of all medical research, and traditional laboratory methods have been seen as necessary and sufficient methods for understanding human health and illness. As a result, approximately 90% of National Institutes of Health (NIH) funding is spent on research within the traditional biomedical sciences (anatomy, biochemistry, genetics, microbiology, molecular biology, physiology, and so forth). The smaller amount of federal funding available for clinical research has been spent primarily on specific disease entities, such as cancer and heart disease. These funding decisions have resulted in the neglect of a large proportion of the problems and issues that confront primary care physicians and their patients.

Clinical research

Perhaps not surprisingly in an era of academic medical centers’ increasing dependence on clinical revenues, the amount of clinical research and the number of clinical investigators have decreased. It has been difficult to track the magnitude of this trend, however, because a concise definition of clinical research is lacking. To address this concern, a national consensus-building conference, the Clinical Research Summit, was held in November 1998, and a working definition was adopted.¹ Clinical research was defined as “a component of medical and health research intended to produce knowledge valuable for understanding human disease, preventing and treating illness, and promoting health.” Attendees further agreed that “clinical research embraces a continuum of studies involving interactions with patients, diagnostic clinical materials or data, or populations in any of the following categories: (1) disease mechanisms; (2) bidirectional integrative research; (3) clinical knowledge, detection, diagnosis, and natural history of disease; (4) therapeutic interventions including clinical trials of drugs, biologics devices, and instruments; (5) prevention (primary and secondary) and health promotion; (6) behavioral research; (7) health services research, including outcomes and cost effectiveness; (8) epidemiology; and (9) community-based and managed care trials.”

Primary care research

In the broader context of clinical research, primary care research remains a fledgling enterprise with many institutions and funding agencies still having difficulty understanding what it includes and what its potential value might be despite a considerable body of supportive literature. The need for primary care research,^2-4 the research traditions of relevance to primary care and the scope of its methods,⁵ and efforts to elevate primary care research to a national priority^6-8 have been documented. However, only 0.4% of NIH funding (up from 0.3% in 1984) and only 4% of the funding from the Agency for Health Care Policy and Research (down from 4.4% in 1984) goes to departments of family medicine.⁹ Both private (eg, the American Academy of Family Physicians) and public (eg, the Agency for Health Care Policy and Research) organizations have formulated working descriptions of primary care research. Nonetheless, primary care researchers still lack a coherent definition with which most can agree. We believe it is particularly important to address this problem now, during a time of concern about and advocacy for clinical research. The definition and examples provided here should be useful for those who are trying to explain the importance of this work to academic administrators and funding agencies.

Primary care has most recently been defined as “the provision of integrated, accessible, health-care services by clinicians that are accountable for addressing a large majority of personal health-care needs, developing a sustained partnership with patients, and practicing within the context of family and community.”⁶ This definition was debated andwas based on a comprehensive literature review and multidisciplinary input. It has been widely adopted since its publication and offers a sensible, if not perfect, definition to guide research in primary care. One of its most important aspects is the concept that primary care is a function that does not necessarily belong to a particular discipline but is dependent on knowledge from many sources. On the basis of this definition, primary care research can be as defined as research directed toward the better understanding and practice of the primary care function.

Categories of research

Primary care research has traditionally included studies that fall into the following overlapping categories.

Theoretical and Methodologic Research. This category includes the development and testing of theoretical models, operational definitions, and measurement tools relevant to the primary care function. Included in this category are such things as: classification systems designed to capture the phenomena of primary care; ways to measure concepts, such as integrated, accessible, and accountable care; methods for distinguishing the separate and combined effects of primary care on individuals, family units, or the community; and ways to observe and measure important relationships (eg, the physician-patient relationship) and their impact on outcomes. Other investigations involve generating and testing alternative conceptualizations of the tasks and methods of primary care (eg, goal-directed care, family-centered care) and efforts to expand the methods available to primary care researchers (eg, mixed methods research, complexity theories).

Health Care Research. This is the large and important area of primary care research focused on direct investigation of the primary care function itself. Questions directed toward improving the quality and effectiveness of primary care practices fit in this category. Although health care research has historically emphasized the analysis of large data sets, primary care research often employs both qualitative and quantitative methods to examine relatively small numbers of individual physicians and practices to determine which methods seem to work best.

Clinical Research. Studies in this category are focused directly on the effects of the primary care function on patients. It includes research on the factors that determine why patients become ill and seek medical attention, the meanings of presenting symptoms and signs in a primary care setting, the most effective and efficient diagnostic and treatment strategies, and the natural histories of health problems with and without intervention. The outcome measures used are often those that are directly meaningful to patients, such as quality of life, mortality, health of the family unit, and cost and convenience of care (outcomes research).

Health Systems Research. This category extends the first 3 to the larger systems level. It encompasses educational research, research on dissemination and adoption of new discoveries, implementation of quality improvement systems into primary care settings, and health policy.

Three examples

Night Sweats. Many diseases and health states are thought to beassociated with night sweats (eg, tuberculosis, menopause, nocturnal hypoglycemia in people with diabetes, autoimmune diseases, malignancies, and medications). However, for the primary care physician who is faced with an undiagnosed patient with night sweats, the following questions arise that currently lack answers: (1) What are the incidence and prevalence of night sweats in a primary care patient population? How often does this symptom go unreported? (clinical); (2) What are the most likely causes of night sweats in a patient presenting in a primary care setting? (clinical); (3) What is the most effective and efficient path to the correct diagnosis? (clinical); (4) What is the natural history of idiopathic night sweats in otherwise healthy patients? (clinical); (5) How much of an impact do night sweats have on the quality of life of patients and their bed partners? (clinical); (6) If common and significant but underrecognized, how can clinicians reorganize their assessment methods to systematically screen patients for night sweats? (health care); and (7) If night sweats are an important symptom of potentially serious disease, how can the population be educated to pay attention to and report them? (health systems).

Cognitive Impairment. Cognitive impairment is a prevalent and serious problem for older people. Early diagnosis and treatment are becoming more important as effective treatment strategies have become available. To provide optimal care for these patients, primary care physicians need answers to the following questions:¹¹ (1) Is systematic formal screening for cognitive impairment necessary in primary care settings in which patients and their families are well known by their physicians and staff and are seen frequently? (clinical); (2) Does earlier detection result in better outcomes? (clinical); (3) What are the outcomes of greatest importance, and how can they best be measured? (theoretical/methodologic); (4) What would it cost to screen all patients [over] older than a specified age? (health systems); (5) Should we train primary care physicians to complete the evaluation and direct treatment, or should patients who screen positive be seen by a neurologist, geriatrician, or psychiatrist or psychologist? (health care); and (6) How can caregiver education programs, which have been shown to delay institutionalization and save money, be organized and funded? (health systems).

Laboratory Test Results

Another set of questions involves laboratory test results. What is the best way to manage (tracking, notification, documentation, follow-up) laboratory test results in a primary care setting? Before this health care question can be answered, several others must be addressed: (1) What is meant by “best”? How should patient preference, physician preference, cost efficiency, and legal requirements be balanced? (theoretical/methodologic); (2) How would we answer such a question even if “best” could be adequately defined? (theoretical/methodologic); and (3) Assuming that an answer can be found, what is the most efficient way to help clinicians incorporate the method into their practices? (health systems).

Education for the future

Primary care research is directed toward the better understanding and practice of the primary care function. It is further distinguished from other types of research by an emphasis on effectiveness rather than efficacy. It is often immediately applicable to primary care practice and widely generalizable.

The science base of medicine, including primary care, has improved significantly during the 20th century. Overall, however, primary care has been neglected because it was not seen as requiring an intellectual engine. The mismatch between the focus of research efforts to date and the need for research in primary care can be understood to derive from an unbalanced effort focused on understanding specific diseases and molecular mechanisms. To improve primary care, a more robust enterprise embracing the full scope of research in this setting is required. We hope that our attempt to define and describe the scope of primary care research will help its advocates educate those who are able to make this happen.

Too often the questions of basic biomedical research have been mistaken to represent the critical scope of all medical research, and traditional laboratory methods have been seen as necessary and sufficient methods for understanding human health and illness. As a result, approximately 90% of National Institutes of Health (NIH) funding is spent on research within the traditional biomedical sciences (anatomy, biochemistry, genetics, microbiology, molecular biology, physiology, and so forth). The smaller amount of federal funding available for clinical research has been spent primarily on specific disease entities, such as cancer and heart disease. These funding decisions have resulted in the neglect of a large proportion of the problems and issues that confront primary care physicians and their patients.

Clinical research

Perhaps not surprisingly in an era of academic medical centers’ increasing dependence on clinical revenues, the amount of clinical research and the number of clinical investigators have decreased. It has been difficult to track the magnitude of this trend, however, because a concise definition of clinical research is lacking. To address this concern, a national consensus-building conference, the Clinical Research Summit, was held in November 1998, and a working definition was adopted.¹ Clinical research was defined as “a component of medical and health research intended to produce knowledge valuable for understanding human disease, preventing and treating illness, and promoting health.” Attendees further agreed that “clinical research embraces a continuum of studies involving interactions with patients, diagnostic clinical materials or data, or populations in any of the following categories: (1) disease mechanisms; (2) bidirectional integrative research; (3) clinical knowledge, detection, diagnosis, and natural history of disease; (4) therapeutic interventions including clinical trials of drugs, biologics devices, and instruments; (5) prevention (primary and secondary) and health promotion; (6) behavioral research; (7) health services research, including outcomes and cost effectiveness; (8) epidemiology; and (9) community-based and managed care trials.”

Primary care research

In the broader context of clinical research, primary care research remains a fledgling enterprise with many institutions and funding agencies still having difficulty understanding what it includes and what its potential value might be despite a considerable body of supportive literature. The need for primary care research,^2-4 the research traditions of relevance to primary care and the scope of its methods,⁵ and efforts to elevate primary care research to a national priority^6-8 have been documented. However, only 0.4% of NIH funding (up from 0.3% in 1984) and only 4% of the funding from the Agency for Health Care Policy and Research (down from 4.4% in 1984) goes to departments of family medicine.⁹ Both private (eg, the American Academy of Family Physicians) and public (eg, the Agency for Health Care Policy and Research) organizations have formulated working descriptions of primary care research. Nonetheless, primary care researchers still lack a coherent definition with which most can agree. We believe it is particularly important to address this problem now, during a time of concern about and advocacy for clinical research. The definition and examples provided here should be useful for those who are trying to explain the importance of this work to academic administrators and funding agencies.

Primary care has most recently been defined as “the provision of integrated, accessible, health-care services by clinicians that are accountable for addressing a large majority of personal health-care needs, developing a sustained partnership with patients, and practicing within the context of family and community.”⁶ This definition was debated andwas based on a comprehensive literature review and multidisciplinary input. It has been widely adopted since its publication and offers a sensible, if not perfect, definition to guide research in primary care. One of its most important aspects is the concept that primary care is a function that does not necessarily belong to a particular discipline but is dependent on knowledge from many sources. On the basis of this definition, primary care research can be as defined as research directed toward the better understanding and practice of the primary care function.

Categories of research

Primary care research has traditionally included studies that fall into the following overlapping categories.

Theoretical and Methodologic Research. This category includes the development and testing of theoretical models, operational definitions, and measurement tools relevant to the primary care function. Included in this category are such things as: classification systems designed to capture the phenomena of primary care; ways to measure concepts, such as integrated, accessible, and accountable care; methods for distinguishing the separate and combined effects of primary care on individuals, family units, or the community; and ways to observe and measure important relationships (eg, the physician-patient relationship) and their impact on outcomes. Other investigations involve generating and testing alternative conceptualizations of the tasks and methods of primary care (eg, goal-directed care, family-centered care) and efforts to expand the methods available to primary care researchers (eg, mixed methods research, complexity theories).

Health Care Research. This is the large and important area of primary care research focused on direct investigation of the primary care function itself. Questions directed toward improving the quality and effectiveness of primary care practices fit in this category. Although health care research has historically emphasized the analysis of large data sets, primary care research often employs both qualitative and quantitative methods to examine relatively small numbers of individual physicians and practices to determine which methods seem to work best.

Clinical Research. Studies in this category are focused directly on the effects of the primary care function on patients. It includes research on the factors that determine why patients become ill and seek medical attention, the meanings of presenting symptoms and signs in a primary care setting, the most effective and efficient diagnostic and treatment strategies, and the natural histories of health problems with and without intervention. The outcome measures used are often those that are directly meaningful to patients, such as quality of life, mortality, health of the family unit, and cost and convenience of care (outcomes research).

Health Systems Research. This category extends the first 3 to the larger systems level. It encompasses educational research, research on dissemination and adoption of new discoveries, implementation of quality improvement systems into primary care settings, and health policy.

Three examples

Night Sweats. Many diseases and health states are thought to beassociated with night sweats (eg, tuberculosis, menopause, nocturnal hypoglycemia in people with diabetes, autoimmune diseases, malignancies, and medications). However, for the primary care physician who is faced with an undiagnosed patient with night sweats, the following questions arise that currently lack answers: (1) What are the incidence and prevalence of night sweats in a primary care patient population? How often does this symptom go unreported? (clinical); (2) What are the most likely causes of night sweats in a patient presenting in a primary care setting? (clinical); (3) What is the most effective and efficient path to the correct diagnosis? (clinical); (4) What is the natural history of idiopathic night sweats in otherwise healthy patients? (clinical); (5) How much of an impact do night sweats have on the quality of life of patients and their bed partners? (clinical); (6) If common and significant but underrecognized, how can clinicians reorganize their assessment methods to systematically screen patients for night sweats? (health care); and (7) If night sweats are an important symptom of potentially serious disease, how can the population be educated to pay attention to and report them? (health systems).

Cognitive Impairment. Cognitive impairment is a prevalent and serious problem for older people. Early diagnosis and treatment are becoming more important as effective treatment strategies have become available. To provide optimal care for these patients, primary care physicians need answers to the following questions:¹¹ (1) Is systematic formal screening for cognitive impairment necessary in primary care settings in which patients and their families are well known by their physicians and staff and are seen frequently? (clinical); (2) Does earlier detection result in better outcomes? (clinical); (3) What are the outcomes of greatest importance, and how can they best be measured? (theoretical/methodologic); (4) What would it cost to screen all patients [over] older than a specified age? (health systems); (5) Should we train primary care physicians to complete the evaluation and direct treatment, or should patients who screen positive be seen by a neurologist, geriatrician, or psychiatrist or psychologist? (health care); and (6) How can caregiver education programs, which have been shown to delay institutionalization and save money, be organized and funded? (health systems).

Laboratory Test Results

Another set of questions involves laboratory test results. What is the best way to manage (tracking, notification, documentation, follow-up) laboratory test results in a primary care setting? Before this health care question can be answered, several others must be addressed: (1) What is meant by “best”? How should patient preference, physician preference, cost efficiency, and legal requirements be balanced? (theoretical/methodologic); (2) How would we answer such a question even if “best” could be adequately defined? (theoretical/methodologic); and (3) Assuming that an answer can be found, what is the most efficient way to help clinicians incorporate the method into their practices? (health systems).

Education for the future

Primary care research is directed toward the better understanding and practice of the primary care function. It is further distinguished from other types of research by an emphasis on effectiveness rather than efficacy. It is often immediately applicable to primary care practice and widely generalizable.

The science base of medicine, including primary care, has improved significantly during the 20th century. Overall, however, primary care has been neglected because it was not seen as requiring an intellectual engine. The mismatch between the focus of research efforts to date and the need for research in primary care can be understood to derive from an unbalanced effort focused on understanding specific diseases and molecular mechanisms. To improve primary care, a more robust enterprise embracing the full scope of research in this setting is required. We hope that our attempt to define and describe the scope of primary care research will help its advocates educate those who are able to make this happen.