Clinical Decision-Making Treatment of Insomnia
in the Evaluation and
DANIELE. EVERIT, M.D., JERRYAVORN,M.D., hton,
Massachusetts,
MERLW. BAKER,B.A.,
New York, New York
We interviewed a representative random sample of 501 office-based general physicians and 298 nurse practitioners to evaluate their approach to the symptoms of insomnia. Clinicians were presented with a standard case of a patient complaining of difficulty sleeping, with the age of the patient depicted as either 37 years or 77 years. Historical information was provided in response to practitioners’ questions. In evaluating the history, physicians asked an average of 2.5 questions and were most likely to ask about psychologic problems. Only 47% of the physicians who were presented with the elderly case vignette elicited a sleep history. By contrast, nurse practitioners asked an average of 3.2 questions, and 60% of them took a sleep history. Despite many possible non-pharmacologic therapies for the patients presented, 46% of physicians identified a prescription medication as the single most effective therapy for the older patient, compared with 17% of nurse practitioners. These findings suggest that physicians place inadequate emphasis on history-taking in the evaluation of insomnia and resort to the use of psychoactive drugs even when non-pharmacologic approaches might be more effective.
From the Program for the Analysis of Clinical Strategies and the Division on Aging, Harvard Medical School and Beth Israel Hospital (DEE, JA). Geriatric Research, Education and Clinical Center, Brockton/West Roxbury Veterans Administration Medical Center, and the Department of Medicine, Beth Israel Hospital (DEE), Boston, Massachusetts, and Louis Harris and Associates, Inc., (MWB). New York, New York. This research was made possible by a grant from the John A. Hartford Foundation. Requests for reprints should be addressed to Daniel E. Everitt. M.D., Program for the Analysis of Clinical Strategies, 333 Longwood Avenue, Boston, Massachusetts 02115. Manuscript submitted November 15, 1989. and accepted rn revised form April 24, 1990.
D
espite the emphasis on thorough history-taking in the training of medical students and house staff, there is evidence that physicians’ practices in these areas are frequently deficient [l-3]. This has led to calls for greater emphasis in initial and continuing medical training on developing the skills needed to gather data from the clinical history [4-61. However, not enough is known about the adequacy of physicians’ history-taking practices in typical clinical situations. The problem of insomnia is one of the most common complaints brought to physicians in office practice, and about one third of the population notes some problem with sleep [7-lo]. As a result, sedative/hypnotic agents are among the most frequently prescribed of all medications [ 111. It might be expected that primary care practitioners would be expert in evaluating and treating such a prevalent symptom complaint. A proportion of insomniac patients have underlying medical illnesses such as congestive heart failure or sleep apnea that cause their sleep disorders, and require careful physical examination and laboratory evaluation. Yet even when these conditions are absent, the evaluation of insomnia relies on a thorough history [12]. The knowledge base required for the diagnosis and initial treatment of most cases of insomnia is concise and has been defined by a National Institutes of Health Consensus Panel [18] as well as in several recent reviews [12-141. It is generally recommended that non-pharmacologic approaches to insomnia be attempted before resorting to prescription medications [8,12,15]. Hypnotics used on a long-term basis do not offer continuing efficacy in most patients and can be habituating; barbiturate and long halflife sedative drugs have been associated with an increased risk of falls and hip fractures in older patients [16,17]. We used the symptom of insomnia as a common outpatient problem for which a wide range of therapeutic strategies exists to study the nature of physicians’ clinical decision-making. Two nationwide surveys were conducted of 501 primary care physicians and 298 office-based nurse practitioners. We also sought to explore whether the approach to the problem was the same when clinicians were evaluat-
September
1990
The American
Journal
of Medicine
Volume
89
357
TABLE I Patient Information Patient’s sleep history: Medical problems: Medicatiions: Diet: Exercise: Psychologic problems: Prior insomnia therapy: Other medical history:
Typically goes to bed at 9:00 PM, wakes up at 4:OOAhf and is unable to get back to sleep Severe arthritis Tylenol for severe arthritis 2cups of coffee with dinner Getting in and out of a car several times a day Spouse diagnosed as having lung cancer 3 months ago None None of relevance
ing older versus younger patients. Because elderly paients are more likely to experience adverse drug effects for a variety of reasons [18], we reasoned that practitioners might take different therapeutic approaches with the two types of patients.
SUBJECTSAND METHODS A random sample of all office-based physicians in the continental United States with primary specialties in family practice, general practice, and internal medicine was identified from the American Medical Association census files. The sample was stratified additionally so that each type of physician was interviewed in proportion to its frequency in the population of office-based adult primary care physicians. Sample physicians were sent a letter by Louis Harris and Associates that requested their cooperation in a foundation-sponsored national study of clinical decision-making. Physicians were excluded if they saw no patients over the age of 65 years or if they spent fewer than 20 hours per week in office practice. An honorarium of $25 was provided to compensate physicians for their time in participating in a telephone interview. A random sample of nurse practitioners was identified from a list of all certified nurse practitioners provided by the American Academy of Nurse Practitioners. Nurses were also required to have an active office practice of at least 20 hours per week that included some patients over age 65. In the interview, a brief clinical vignette was presented to the respondents: “Now, I’d like to ask you to consider a hypothetical case. Let’s say a (age specified) man you’ve never seen before comes to your office seeking help for difficulty sleeping, a problem he has noted for several weeks.” The identical scenario and survey questions were presented to the sample of physicians and nurse practitioners. The patient was identified as being 77 years old to the nurse practitioners and to a randomly chosen two thirds of physicians. However, the patient was identified as being 37 years old to the remaining third of the physicians in the sample to evaluate the effect of age on the clinical decision-making of physicians. At the end of the inter358
September
1990
The American
Journal
of Medicine
Volume
view, nurses were asked about their authority to prescribe medications under the laws and regulations of the state in which they practiced. In presenting the case scenario, care was taken to identify the survey as a study of how practitioners approach common clinical problems, rather than as a study of medication use per se. Practitioners were given the opportunity to ask for historical information at several points, and the authors prepared answers for the most likely questions a clinician might ask. After the initial scenario was presented, the respondent was asked, “Before deciding on a treatment plan, what would you be most likely to ask the patient ?” Interviewers offered certain specific information about the patient, if asked (Table I). When a clinician requested no further information, he or she was asked, “Armed with this information, what do you think you could do for this patient that would be effective in helping his sleeping problem?” The clinician was asked after each response if there was anything else that would be effective, and which one intervention was likely to be the single most effective. Experienced surveyors employed by Louis Harris and Associates conducted the interviews. The survey instrument was pilot-tested extensively, and coding was refined so that little discretion or interpretation was needed by the telephone surveyors. All interviewers were frequently monitored by supervisors at random intervals to assure consistency in technique. This scenario was presented within the first several minutes of the interview. At the end of the interview, physicians and nurses were asked questions about the nature of their practice, details of their training, and the length of time they had spent in practice. Ninety-five percent confidence limits on the proportions of sample responses were calculated assuming a normal approximation to the binomial distribution [ 191.
RESULTS PhysicianSample A total of 1,743 physicians were initially contacted by telephone. Of these, 277 did not meet the entry criteria of being in general office practice more than 20 hours per week and seeing at least some patients over age 65. Of the remaining 1,466 physicians, 501 agreed to participate in the complete interview, including 140 in general practice, 151 in family practice, and 210 in internal medicine. Respondents practiced in a range of geographic areas: 36% were in small town or rural areas, 24% were in suburbs, and 40% practiced in urban areas. Forty-one percent of the physicians stated that 89
TREATMENT
more than half of their practice was made up of patients over age 65. Most practiced in a fee-forservice setting (83%). The physicians interviewed were demographically very similar to the sample of all physicians from which they were drawn (Table II), although they slightly over-represented the south and under-represented the west. The length of time that physicians interviewed had been in practice was nearly identical to that of all the physicians in the relevant specialties. Nurse Practitioner Sample Three hundred fifty-one nurse practitioners, contacted by telephone, met the study inclusion criteria; 53 declined to participate in the study and 298 completed the interview. Of these, 40% were authorized to prescribe certain medications under the laws and regulations of their state. More than a third (37%) held Masters of Science degrees and 60% held a nurse practitioner’s certificate without an advanced degree. The nurses were employed in a wide range of practice settings that were diverse geographically; 61% had been in practice for 6 years or longer. Decision to Initiate Therapy Fewer than half of all physicians asked for any information about the patient’s sleep pattern (which indicated that he slept 7 hours each night). Fewer than one quarter learned about the patient’s evening caffeine consumption. The average number of questions asked was similar for physicians who were presented with the older patient as for those who were presented with the younger patient (2.6 versus 2.5). Overall, 58% of respondents asked zero to two questions, and 42% asked three or more questions. Table III indicates the frequency with which particular historical questions were raised, stratified by age of the “patient.” Physicians who requested additional information before making a therapy decision were most likely to ask about psychologic problems. Although physicians presented with the older patient were significantly more likely to ask about other medical problems (52% versus 32%), the frequency of other questions did not differ significantly by “patient” age. Doctors in practice 21 years or more asked substantially fewer questions than did those in practice fewer years: 34% of those in practice over 20 years asked three or more questions, compared with 54% of those practicing medicine for 5 years or less. Choice of Therapy The patients presented in the scenario had many contributing causes for sleep problems such as un-
OF INSOMNIA
/ EVERITT
ET AL
TABLEii DemographicCharacteristicsof PhysiciansSampledversusAil Physiciansin GeneralMedicalPractice* I Physicians Sampled (%I (n = 501)
All Physicians (%)
Ge;f;taphicarea South Midwest West Time in practice
in the Evaluation and
DANIELE. EVERIT, M.D., JERRYAVORN,M.D., hton,
Massachusetts,
MERLW. BAKER,B.A.,
New York, New York
We interviewed a representative random sample of 501 office-based general physicians and 298 nurse practitioners to evaluate their approach to the symptoms of insomnia. Clinicians were presented with a standard case of a patient complaining of difficulty sleeping, with the age of the patient depicted as either 37 years or 77 years. Historical information was provided in response to practitioners’ questions. In evaluating the history, physicians asked an average of 2.5 questions and were most likely to ask about psychologic problems. Only 47% of the physicians who were presented with the elderly case vignette elicited a sleep history. By contrast, nurse practitioners asked an average of 3.2 questions, and 60% of them took a sleep history. Despite many possible non-pharmacologic therapies for the patients presented, 46% of physicians identified a prescription medication as the single most effective therapy for the older patient, compared with 17% of nurse practitioners. These findings suggest that physicians place inadequate emphasis on history-taking in the evaluation of insomnia and resort to the use of psychoactive drugs even when non-pharmacologic approaches might be more effective.
From the Program for the Analysis of Clinical Strategies and the Division on Aging, Harvard Medical School and Beth Israel Hospital (DEE, JA). Geriatric Research, Education and Clinical Center, Brockton/West Roxbury Veterans Administration Medical Center, and the Department of Medicine, Beth Israel Hospital (DEE), Boston, Massachusetts, and Louis Harris and Associates, Inc., (MWB). New York, New York. This research was made possible by a grant from the John A. Hartford Foundation. Requests for reprints should be addressed to Daniel E. Everitt. M.D., Program for the Analysis of Clinical Strategies, 333 Longwood Avenue, Boston, Massachusetts 02115. Manuscript submitted November 15, 1989. and accepted rn revised form April 24, 1990.
D
espite the emphasis on thorough history-taking in the training of medical students and house staff, there is evidence that physicians’ practices in these areas are frequently deficient [l-3]. This has led to calls for greater emphasis in initial and continuing medical training on developing the skills needed to gather data from the clinical history [4-61. However, not enough is known about the adequacy of physicians’ history-taking practices in typical clinical situations. The problem of insomnia is one of the most common complaints brought to physicians in office practice, and about one third of the population notes some problem with sleep [7-lo]. As a result, sedative/hypnotic agents are among the most frequently prescribed of all medications [ 111. It might be expected that primary care practitioners would be expert in evaluating and treating such a prevalent symptom complaint. A proportion of insomniac patients have underlying medical illnesses such as congestive heart failure or sleep apnea that cause their sleep disorders, and require careful physical examination and laboratory evaluation. Yet even when these conditions are absent, the evaluation of insomnia relies on a thorough history [12]. The knowledge base required for the diagnosis and initial treatment of most cases of insomnia is concise and has been defined by a National Institutes of Health Consensus Panel [18] as well as in several recent reviews [12-141. It is generally recommended that non-pharmacologic approaches to insomnia be attempted before resorting to prescription medications [8,12,15]. Hypnotics used on a long-term basis do not offer continuing efficacy in most patients and can be habituating; barbiturate and long halflife sedative drugs have been associated with an increased risk of falls and hip fractures in older patients [16,17]. We used the symptom of insomnia as a common outpatient problem for which a wide range of therapeutic strategies exists to study the nature of physicians’ clinical decision-making. Two nationwide surveys were conducted of 501 primary care physicians and 298 office-based nurse practitioners. We also sought to explore whether the approach to the problem was the same when clinicians were evaluat-
September
1990
The American
Journal
of Medicine
Volume
89
357
TABLE I Patient Information Patient’s sleep history: Medical problems: Medicatiions: Diet: Exercise: Psychologic problems: Prior insomnia therapy: Other medical history:
Typically goes to bed at 9:00 PM, wakes up at 4:OOAhf and is unable to get back to sleep Severe arthritis Tylenol for severe arthritis 2cups of coffee with dinner Getting in and out of a car several times a day Spouse diagnosed as having lung cancer 3 months ago None None of relevance
ing older versus younger patients. Because elderly paients are more likely to experience adverse drug effects for a variety of reasons [18], we reasoned that practitioners might take different therapeutic approaches with the two types of patients.
SUBJECTSAND METHODS A random sample of all office-based physicians in the continental United States with primary specialties in family practice, general practice, and internal medicine was identified from the American Medical Association census files. The sample was stratified additionally so that each type of physician was interviewed in proportion to its frequency in the population of office-based adult primary care physicians. Sample physicians were sent a letter by Louis Harris and Associates that requested their cooperation in a foundation-sponsored national study of clinical decision-making. Physicians were excluded if they saw no patients over the age of 65 years or if they spent fewer than 20 hours per week in office practice. An honorarium of $25 was provided to compensate physicians for their time in participating in a telephone interview. A random sample of nurse practitioners was identified from a list of all certified nurse practitioners provided by the American Academy of Nurse Practitioners. Nurses were also required to have an active office practice of at least 20 hours per week that included some patients over age 65. In the interview, a brief clinical vignette was presented to the respondents: “Now, I’d like to ask you to consider a hypothetical case. Let’s say a (age specified) man you’ve never seen before comes to your office seeking help for difficulty sleeping, a problem he has noted for several weeks.” The identical scenario and survey questions were presented to the sample of physicians and nurse practitioners. The patient was identified as being 77 years old to the nurse practitioners and to a randomly chosen two thirds of physicians. However, the patient was identified as being 37 years old to the remaining third of the physicians in the sample to evaluate the effect of age on the clinical decision-making of physicians. At the end of the inter358
September
1990
The American
Journal
of Medicine
Volume
view, nurses were asked about their authority to prescribe medications under the laws and regulations of the state in which they practiced. In presenting the case scenario, care was taken to identify the survey as a study of how practitioners approach common clinical problems, rather than as a study of medication use per se. Practitioners were given the opportunity to ask for historical information at several points, and the authors prepared answers for the most likely questions a clinician might ask. After the initial scenario was presented, the respondent was asked, “Before deciding on a treatment plan, what would you be most likely to ask the patient ?” Interviewers offered certain specific information about the patient, if asked (Table I). When a clinician requested no further information, he or she was asked, “Armed with this information, what do you think you could do for this patient that would be effective in helping his sleeping problem?” The clinician was asked after each response if there was anything else that would be effective, and which one intervention was likely to be the single most effective. Experienced surveyors employed by Louis Harris and Associates conducted the interviews. The survey instrument was pilot-tested extensively, and coding was refined so that little discretion or interpretation was needed by the telephone surveyors. All interviewers were frequently monitored by supervisors at random intervals to assure consistency in technique. This scenario was presented within the first several minutes of the interview. At the end of the interview, physicians and nurses were asked questions about the nature of their practice, details of their training, and the length of time they had spent in practice. Ninety-five percent confidence limits on the proportions of sample responses were calculated assuming a normal approximation to the binomial distribution [ 191.
RESULTS PhysicianSample A total of 1,743 physicians were initially contacted by telephone. Of these, 277 did not meet the entry criteria of being in general office practice more than 20 hours per week and seeing at least some patients over age 65. Of the remaining 1,466 physicians, 501 agreed to participate in the complete interview, including 140 in general practice, 151 in family practice, and 210 in internal medicine. Respondents practiced in a range of geographic areas: 36% were in small town or rural areas, 24% were in suburbs, and 40% practiced in urban areas. Forty-one percent of the physicians stated that 89
TREATMENT
more than half of their practice was made up of patients over age 65. Most practiced in a fee-forservice setting (83%). The physicians interviewed were demographically very similar to the sample of all physicians from which they were drawn (Table II), although they slightly over-represented the south and under-represented the west. The length of time that physicians interviewed had been in practice was nearly identical to that of all the physicians in the relevant specialties. Nurse Practitioner Sample Three hundred fifty-one nurse practitioners, contacted by telephone, met the study inclusion criteria; 53 declined to participate in the study and 298 completed the interview. Of these, 40% were authorized to prescribe certain medications under the laws and regulations of their state. More than a third (37%) held Masters of Science degrees and 60% held a nurse practitioner’s certificate without an advanced degree. The nurses were employed in a wide range of practice settings that were diverse geographically; 61% had been in practice for 6 years or longer. Decision to Initiate Therapy Fewer than half of all physicians asked for any information about the patient’s sleep pattern (which indicated that he slept 7 hours each night). Fewer than one quarter learned about the patient’s evening caffeine consumption. The average number of questions asked was similar for physicians who were presented with the older patient as for those who were presented with the younger patient (2.6 versus 2.5). Overall, 58% of respondents asked zero to two questions, and 42% asked three or more questions. Table III indicates the frequency with which particular historical questions were raised, stratified by age of the “patient.” Physicians who requested additional information before making a therapy decision were most likely to ask about psychologic problems. Although physicians presented with the older patient were significantly more likely to ask about other medical problems (52% versus 32%), the frequency of other questions did not differ significantly by “patient” age. Doctors in practice 21 years or more asked substantially fewer questions than did those in practice fewer years: 34% of those in practice over 20 years asked three or more questions, compared with 54% of those practicing medicine for 5 years or less. Choice of Therapy The patients presented in the scenario had many contributing causes for sleep problems such as un-
OF INSOMNIA
/ EVERITT
ET AL
TABLEii DemographicCharacteristicsof PhysiciansSampledversusAil Physiciansin GeneralMedicalPractice* I Physicians Sampled (%I (n = 501)
All Physicians (%)
Ge;f;taphicarea South Midwest West Time in practice
