- Email: [email protected]
Perceived influence of different information sources on the decision-making of internal medicine house staff and faculty
Sot. SCI. Med. Vol. 16. pp. 1361 to 1364. 1982 Printed in Great Brttatn
0277-9536 82.141361-04$03.00:0 Perwnon
Press Ltd
PERCEIVED INFLUENCE OF DIFFERENT INFORMATION SOURCES ON THE DECISION-MAKING OF INTERNAL MEDICINE HOUSE STAFF AND FACULTY STUART
Regenstrief
Institute
J.
COHEN,
for Health
Abstract-To
determine the medical decisions. degree of influence of house experience on their decisions
common
MORRIS WEINBERGER.STEVENA. MAZZUCA and CLEMENT J. MCDONALD
Care and Department of Medicine, Indiana cine, Indianapolis. IN 46202. U.S.A.
University
School
of Medi-
extent to which different sources of information are perceived to influence 10 interns. 22 senior residents and 9 faculty general internists rated the staff. general internists. subspecialists, conferences. journal reading and past concerning primary prevention (vaccination). secondary prevention (screen-
ing) and drug therapy, Analysis of variance of their questionnaire data supports the following conclusions: physicians at different stages of training rely on different sources of information; as physicians advance in training the influence of generalists wanes while that of subspecialists increases; subspecialists and past experience are perceived as primarily affecting therapeutic decisions; primary prevention .appears least subject to influence by prevailing information sources; and the preference for reading begins early and increases as physicians advance in training. These data suggest that designing effective medical education requires considering the level of the physician’s training and the nature of the medical decision
The
potential sources of updating medical expertise
information
available
for
are numerous. They vary from the more formal and group-oriented approaches, such as postgraduate courses and meetings of general and specialty medical societies. to the more informal and self-instructional sources. such as reading medical literature and consulting with colleagues. The extent to which these different sources of information actually influence practice patterns has not been elucidated [l]. What has been investigated is the preferences physicians have for the different sources of information. Invariably journal reading is the preferred source regardless of the medical specialty [2], the practice setting [3], or the medical topic [4.5]. Preferences for other sources of information such as professional meetings. colleagues and postgraduate courses varied. depending on what was to be learned and by whom. In order to determine the extent to which different sources of information may influence common medical decisions. the current study examines the perceived value of conventional sources of medical information in influencing physicians at different stages in their career. Perceptions involving both preventive and therapeutic medical decisions are examined. \IETHODS We developed a questionnaire to assess the influence of various sources of information in shaping physicians’ decisions about seven medical procedures. including two which involve primary prevention Please address all correspondence and requests for reprints to: Dr Stuart J. Cohen. Regenstrief Institute for Health Care. 1001 West Tenth Street. Indianapolis. IN 46202. U.S.A.
(pneumonia and influenza vaccination). three which involve.secondary prevention or screening (mammography to detect breast cancer, cervical Papanicolaou test to screen for cervical cancer and occult blood test to detect colon cancer) and two which are therapeutic (digitalis for congestive heart failure and beta adrenergic blockade for hypertension or arrhythmias), We selected these decisions based on a review of common ambulatory actions taken by general internal medicine faculty and medicine house officers who staff the General Medicine Clinic (GMC) of a city/council hospital affiliated with Indiana University School of Medicine. For each medical decision. the physician was asked to indicate the amount of influence they attributed to each of six information sources: interns and residents, general medicine staff. subspecialty staff, conferences (e.g. grand rounds. specialty clinics). reading and past experience. Each information source was rated by placing an ‘x’ on one of five horizontal line segments, the extremes of which were labeled ‘Major Influence’ and ‘No Influence’. Responses were coded on a five-point scale with major influences receiving a five. the next space given a four and so on, to no influence, which received a one. In April of 1980. one of the investigators gave the questionnaire to every full-time faculty physician in the general internal medicine section of the Department of Medicine at Indiana University School of Medicine. We asked the faculty to complete and return them in a week. That June. one of the investigators or a GMC faculty physician handed a packet containing the questionnaire to each of the house staff who was completing an internship or residency in general internal medicine. The packet contained a S10 bill and a letter indicating that the money was a token of appreciation for cooperation with all surveys during their stay at the medical center. This letter also 1361
STURT
1362
J.
COHEN er trl.
stated that individual responses would be kept cordiHouse staff returned the surveys by mail using a stamped envelope provided with the questionnaire. Any house officer not returning the questionnaire within a week was reminded in person at the GMC to mail back the survey. &ntial.
3. 5
RESULTS Questionnaires were returned from the 9 fadulty members. IO of the I2 house staff completing their internships and 21 of the 23 senior residents, representing a 91% return rate. In order to test the influence of each major factor (i.e. the level of physician training, the type of information source. and the kind of medical procedure) both singly and in combination, the results were analyzed using a 3 (level of training) by 6 (information source) by 7 (medical procedure) analysis of variance with repeated measures on the last two factors [6]. Table 1 displays the degrees of freedom, mean squares and F-ratios for the main or single effect of each factor by itself, the interaction between combinations of factors in pairs (first-order interaction), and the overall or higher-order interaction of all three factors. Only the main effect of the level of training was not statistically significant; therefore, no one group of physicians (i.e. interns, residents, faculty internists) perceived themselves to be influenced more than any other group. The significant main effect for the source of information (P < 0.01) indicates that some information sources were viewed as more influential than others. In general, reading was seen as the most influential information source and house staff, the least. The significant main effect for the medical procedures (P < 0.01) suggests that some medical decisions were viewed as subject to greater influence than others. Therapeutic decisions were reported to be influenced significantly more than decisions involving primary prevention. All first-order interactions (level by source. level by procedure, and source by procedure) were statistically significant (P < 0.05). To aid in the interpretation of these findings, we graphed the first-order interactions (Figs l-3). To identify the significant differences
Table I. Analysis of variance of the amount of influence as a function of the physicians’ level of training, the source of information and the type of medical procedure Source
Level of
training Error,
Information AxB
(A)
source
(B)
(C)
AxC Error, BxC AxBxC ErrorB,< 0.05; tP
2 37
20.17 12.43
I .62
5
71.75 18.45
27.10t 6.97t
185
Medical procedure
*p
MS
IO
Error,
< 0.01.
F
d.f.
6 I2 222 30 60 1110
2.65 Il.33 1.79 0.92 2.20 0.42 0.20
12.35t 1.95* 10.96t 2.07t
\
\ .\
1.0
\
1 5 hONL
1.0
Fig. 1. Perceived influence on clinical declslons as a function of the level of physician training and the source of information. Information sources: journal reading (0). past experience (0). house staff (A). general internists (C). conferences (W and subspecialty staff (A).
(P < 0.05) for each comparison (e.g. the influence of reading on interns vs faculty), we computed simple effect post hoc analyses using the procedures recommended by Winer [6]. The interaction between the level of training and the source of influence is displayed in Fig. 1. Interns. residents and faculty differed in the extent to which they reported being influenced by the different information sources. While all three groups rated journal reading as the most important source, only the faculty judged it to be significantly more influential than all the other sources. Residents viewed reading as significantly more important than all the other sources except conferences and subsequently staff. On the other hand, interns considered reading as significantly more influential than only conferences and subspecialty staff. In addition, interns viewed conferences and subspecialists as significantly less important than all the other information sources. The credibility of house staff and general internists as a source of information decreased significantly as one advanced in training. Only interns considered the generalist faculty as a valuable information source. Both senior residents and the general internists viewed the subspecialty staff as significantly more influential than their generalist colleagues. The interaction between the level of training and the medical procedure was significant (P < 0.01) and is displayed in Fig. 2. Faculty were subject to no more influence for one medical procedure than any other.
Perceived
influence
of different
Fig. 2. Perceived influence on clinical decisions as a function of the level of physician training and the medical procedure. Medical procedures: beta adrenergic blockade (0). digitalis (0). occult blood test (m). cervical Papanicolaou test (A). mammography (A). influenza vaccination (l ) and pneumococcal pneumonia vaccination (0).
Unlike faculty, the house staff judged themselves as being subject to greater influence in decisions involving therapeutic actions than they were in primary preventive measures. The interaction between the source of influence and the medical procedure is shown in Fig. 3. Because of the large number of posr hoc comparisons. the level of alpha for testing significance was set at 0.01. Across all procedures, reading was viewed uniformly as the strongest influence and house staff, the weakest. Subspecialists were perceived as influencing the use of beta blockers significantly more than all the other medical procedures except digitalis and occult blood testing. Conferences were reported to be significantly more valuable for the use of digitalis. occult blood and beta blockers than any of the other procedures. Past experience was seen as significantly more influential for therapeutic decisions and the use of occult blood testing than for the other screening and primary preventive decisions. The large number of degrees of freedom for the highest order interaction no doubt contributed to its statistical significance. The result is in large part explainable by the significant first-order interactions, and suggests that the amount of influence is a joint function of the source of information. the medical procedure in question and the level of training of the physiclan. DISCL’SSIOS
The results physicians in
of the general
questionnaire indicate that internal medicine furthest
information
sources
1363
advanced in their training view generalists as playing a less important role than subspecialists in influencing certain of their everyday medical decisions. Senior residents at the conclusion of their training share this perception. Apparently, a physician views another physician as influential if the latter has greater expertise in a special area. but not if the gap between their levels of knowledge is too great (e.g. between interns and subspecialists). In the United States. the Council on General Internal Medicine of the American Internal Medicine of the American Board of Internal Medicine has advocated training programs in which general internists serve as role models with major teaching responsibilities and that medical subspecialty rotations also be included in the comprehensive training of the general internist [7]. If our house staff are representative of the population of general medicine house staffs, then there are two possible errors which can occur in their training: (1) starting subspecialty rotations too early in training and (2) continuing major teaching responsibility by general internists beyond the time when more advanced residents are primarily heeding subspecialty advice. Our results also suggest that the perceived value of an information source depends upon the particular kind of medical decision being made. Interns and residents see therapeutic decisions as subject to much greater influence than are the primary preventive decisions. such as the use of influenza and pneumococcal pneumonia immunization. The house staff view their own past experience as being more important for therapeutic than for preventive decisions. The value of past experience may depend upon the speed and frequency with which physicians receive positive
Fig. 3. Perceived influence on clinical decisions as a function of the sources of information and the medical procedure. Information sources: journal reading (01. conferences (ml. subspecialty staff (A). general internists (3). past experience (0) and house staff (Al.
1364
STUART
J. COHES et trl.
feedback. The result of initiating a new therapy is usually clearer and more immediate than the feedback from immunization. For example. a patient may not manifest pneumococcal pneumonia because the vaccine was effective or there was no exposure. In contrast, the effectiveness of a beta blocker to lower blood pressure can be determined more immediately and unambiguously. The greater importance credited to past experience for occult blood testing than for mammography and cervical Papanicolaou testing may be due to the higher rate of positive feedback for occult blood testing (3”/;) than for the rate of the other two (0.5-l”,;). If past experience is to serve an edu: cational function. especially for nontherapeutic decisions, some practice audit or reporting system may be required to assist house staff in assessing the merit of their actions. For example, periodic circulation of the number of cases of pneumonia or influenza contracted among the high-risk vaccinated vs high-risk unvaccinated may lend objectivity to house staffs’ subjective impressions of past experience. The finding that physicians perceived their decisions to be influenced more by journal reading than by other sources of information complements other studies which report that reading is consistently preferred over other modes of continuing education [l-5]. These results do not prove that physician reading is the mode of choice because it most effectively improves care; reading may be selected simply because of its low cost and ready access. However, since attendance in continuing medical education courses is frequently mandated, the preference for reading is an important consideration. Moreover, reading is an inexpensive way to obtain information. Postgraduate courses cost physicians 1WlOOO times as much as obtaining the same information by reading medical journals [8]. Our cross sectional study indicates that differences exist at different levels of training regarding perceptions about the influence of common sources of medical information on clinical decisions. In the past, formal postgraduate courses were favored over reading by the mandators of continuing education because
attendance at courses could be verified: time spent reading could not. However. testing programs using newsletters to update medical knowledge should be greatly expanded to give the physician learning options equal in scope to those avallable from continuing education courses. This allows physicians to receive credit for learning etklently on their own. Such options would reflect the evidence that physicians prefer reading as a source of information. This preference is found early in their postgraduate medical career and remains strong while other information sources are seen as less influential for physicians more advanced in experience. Acfiilowledgr,,lerlts-Supporred
in part by grants from the National Institute for Arthritis. Metabolism and Dlgestlve Diseases (PHS P60 AM 205421 and the National Center for Health Services Research INo. 502185-HEW). REFERESCES D. B. and Werthelmer .A. I. Sources of I. Christensen lnformatlon and influence on new drug prescribmg among physxians in an HMO. Sw. %I. .Lfrd. 13A, 313, 1979. 2. Stinson E. R. and Mueller D. A. Survey of health professionals’ information habits and needs. JAM,-L 243. 140. 1980. F. E. Continumg educatmn 3. Guptill P. B. and Graham activities of physicians in solo and group practice. Mrd. Carr 14, 173. 1976. 4. Manning P. R. and Denson T. A. How cardiologists learn about echocardiography: a remmder for medical educators and legislators. Aw1. lfltern. Mrd. 92. 690. 1980. P. R. and Denson T. A. How lntermsts 5. Manning learned about cimetidine. Awl. itlrern. Mrd. 92, 690. 1980. B. J. Srtrtistictrl Prir~cipfrs 01 E.uprrimr~~tu/ 6. Winer Drsiyn. McGraw-Hill, New York. 1971. 7. Council on General Internal Medicine. American Board of Internal Medicine. Attributes of the general internist and recommendations for trainmg. Ann. intrrn. Med. 86, 472. 1977. 8. Huth E. Continuing medical education: Needs. costs and consequences. Ann. intern. Med. 93. 698. 1980.
0277-9536 82.141361-04$03.00:0 Perwnon
Press Ltd
PERCEIVED INFLUENCE OF DIFFERENT INFORMATION SOURCES ON THE DECISION-MAKING OF INTERNAL MEDICINE HOUSE STAFF AND FACULTY STUART
Regenstrief
Institute
J.
COHEN,
for Health
Abstract-To
determine the medical decisions. degree of influence of house experience on their decisions
common
MORRIS WEINBERGER.STEVENA. MAZZUCA and CLEMENT J. MCDONALD
Care and Department of Medicine, Indiana cine, Indianapolis. IN 46202. U.S.A.
University
School
of Medi-
extent to which different sources of information are perceived to influence 10 interns. 22 senior residents and 9 faculty general internists rated the staff. general internists. subspecialists, conferences. journal reading and past concerning primary prevention (vaccination). secondary prevention (screen-
ing) and drug therapy, Analysis of variance of their questionnaire data supports the following conclusions: physicians at different stages of training rely on different sources of information; as physicians advance in training the influence of generalists wanes while that of subspecialists increases; subspecialists and past experience are perceived as primarily affecting therapeutic decisions; primary prevention .appears least subject to influence by prevailing information sources; and the preference for reading begins early and increases as physicians advance in training. These data suggest that designing effective medical education requires considering the level of the physician’s training and the nature of the medical decision
The
potential sources of updating medical expertise
information
available
for
are numerous. They vary from the more formal and group-oriented approaches, such as postgraduate courses and meetings of general and specialty medical societies. to the more informal and self-instructional sources. such as reading medical literature and consulting with colleagues. The extent to which these different sources of information actually influence practice patterns has not been elucidated [l]. What has been investigated is the preferences physicians have for the different sources of information. Invariably journal reading is the preferred source regardless of the medical specialty [2], the practice setting [3], or the medical topic [4.5]. Preferences for other sources of information such as professional meetings. colleagues and postgraduate courses varied. depending on what was to be learned and by whom. In order to determine the extent to which different sources of information may influence common medical decisions. the current study examines the perceived value of conventional sources of medical information in influencing physicians at different stages in their career. Perceptions involving both preventive and therapeutic medical decisions are examined. \IETHODS We developed a questionnaire to assess the influence of various sources of information in shaping physicians’ decisions about seven medical procedures. including two which involve primary prevention Please address all correspondence and requests for reprints to: Dr Stuart J. Cohen. Regenstrief Institute for Health Care. 1001 West Tenth Street. Indianapolis. IN 46202. U.S.A.
(pneumonia and influenza vaccination). three which involve.secondary prevention or screening (mammography to detect breast cancer, cervical Papanicolaou test to screen for cervical cancer and occult blood test to detect colon cancer) and two which are therapeutic (digitalis for congestive heart failure and beta adrenergic blockade for hypertension or arrhythmias), We selected these decisions based on a review of common ambulatory actions taken by general internal medicine faculty and medicine house officers who staff the General Medicine Clinic (GMC) of a city/council hospital affiliated with Indiana University School of Medicine. For each medical decision. the physician was asked to indicate the amount of influence they attributed to each of six information sources: interns and residents, general medicine staff. subspecialty staff, conferences (e.g. grand rounds. specialty clinics). reading and past experience. Each information source was rated by placing an ‘x’ on one of five horizontal line segments, the extremes of which were labeled ‘Major Influence’ and ‘No Influence’. Responses were coded on a five-point scale with major influences receiving a five. the next space given a four and so on, to no influence, which received a one. In April of 1980. one of the investigators gave the questionnaire to every full-time faculty physician in the general internal medicine section of the Department of Medicine at Indiana University School of Medicine. We asked the faculty to complete and return them in a week. That June. one of the investigators or a GMC faculty physician handed a packet containing the questionnaire to each of the house staff who was completing an internship or residency in general internal medicine. The packet contained a S10 bill and a letter indicating that the money was a token of appreciation for cooperation with all surveys during their stay at the medical center. This letter also 1361
STURT
1362
J.
COHEN er trl.
stated that individual responses would be kept cordiHouse staff returned the surveys by mail using a stamped envelope provided with the questionnaire. Any house officer not returning the questionnaire within a week was reminded in person at the GMC to mail back the survey. &ntial.
3. 5
RESULTS Questionnaires were returned from the 9 fadulty members. IO of the I2 house staff completing their internships and 21 of the 23 senior residents, representing a 91% return rate. In order to test the influence of each major factor (i.e. the level of physician training, the type of information source. and the kind of medical procedure) both singly and in combination, the results were analyzed using a 3 (level of training) by 6 (information source) by 7 (medical procedure) analysis of variance with repeated measures on the last two factors [6]. Table 1 displays the degrees of freedom, mean squares and F-ratios for the main or single effect of each factor by itself, the interaction between combinations of factors in pairs (first-order interaction), and the overall or higher-order interaction of all three factors. Only the main effect of the level of training was not statistically significant; therefore, no one group of physicians (i.e. interns, residents, faculty internists) perceived themselves to be influenced more than any other group. The significant main effect for the source of information (P < 0.01) indicates that some information sources were viewed as more influential than others. In general, reading was seen as the most influential information source and house staff, the least. The significant main effect for the medical procedures (P < 0.01) suggests that some medical decisions were viewed as subject to greater influence than others. Therapeutic decisions were reported to be influenced significantly more than decisions involving primary prevention. All first-order interactions (level by source. level by procedure, and source by procedure) were statistically significant (P < 0.05). To aid in the interpretation of these findings, we graphed the first-order interactions (Figs l-3). To identify the significant differences
Table I. Analysis of variance of the amount of influence as a function of the physicians’ level of training, the source of information and the type of medical procedure Source
Level of
training Error,
Information AxB
(A)
source
(B)
(C)
AxC Error, BxC AxBxC ErrorB,< 0.05; tP
2 37
20.17 12.43
I .62
5
71.75 18.45
27.10t 6.97t
185
Medical procedure
*p
MS
IO
Error,
< 0.01.
F
d.f.
6 I2 222 30 60 1110
2.65 Il.33 1.79 0.92 2.20 0.42 0.20
12.35t 1.95* 10.96t 2.07t
\
\ .\
1.0
\
1 5 hONL
1.0
Fig. 1. Perceived influence on clinical declslons as a function of the level of physician training and the source of information. Information sources: journal reading (0). past experience (0). house staff (A). general internists (C). conferences (W and subspecialty staff (A).
(P < 0.05) for each comparison (e.g. the influence of reading on interns vs faculty), we computed simple effect post hoc analyses using the procedures recommended by Winer [6]. The interaction between the level of training and the source of influence is displayed in Fig. 1. Interns. residents and faculty differed in the extent to which they reported being influenced by the different information sources. While all three groups rated journal reading as the most important source, only the faculty judged it to be significantly more influential than all the other sources. Residents viewed reading as significantly more important than all the other sources except conferences and subsequently staff. On the other hand, interns considered reading as significantly more influential than only conferences and subspecialty staff. In addition, interns viewed conferences and subspecialists as significantly less important than all the other information sources. The credibility of house staff and general internists as a source of information decreased significantly as one advanced in training. Only interns considered the generalist faculty as a valuable information source. Both senior residents and the general internists viewed the subspecialty staff as significantly more influential than their generalist colleagues. The interaction between the level of training and the medical procedure was significant (P < 0.01) and is displayed in Fig. 2. Faculty were subject to no more influence for one medical procedure than any other.
Perceived
influence
of different
Fig. 2. Perceived influence on clinical decisions as a function of the level of physician training and the medical procedure. Medical procedures: beta adrenergic blockade (0). digitalis (0). occult blood test (m). cervical Papanicolaou test (A). mammography (A). influenza vaccination (l ) and pneumococcal pneumonia vaccination (0).
Unlike faculty, the house staff judged themselves as being subject to greater influence in decisions involving therapeutic actions than they were in primary preventive measures. The interaction between the source of influence and the medical procedure is shown in Fig. 3. Because of the large number of posr hoc comparisons. the level of alpha for testing significance was set at 0.01. Across all procedures, reading was viewed uniformly as the strongest influence and house staff, the weakest. Subspecialists were perceived as influencing the use of beta blockers significantly more than all the other medical procedures except digitalis and occult blood testing. Conferences were reported to be significantly more valuable for the use of digitalis. occult blood and beta blockers than any of the other procedures. Past experience was seen as significantly more influential for therapeutic decisions and the use of occult blood testing than for the other screening and primary preventive decisions. The large number of degrees of freedom for the highest order interaction no doubt contributed to its statistical significance. The result is in large part explainable by the significant first-order interactions, and suggests that the amount of influence is a joint function of the source of information. the medical procedure in question and the level of training of the physiclan. DISCL’SSIOS
The results physicians in
of the general
questionnaire indicate that internal medicine furthest
information
sources
1363
advanced in their training view generalists as playing a less important role than subspecialists in influencing certain of their everyday medical decisions. Senior residents at the conclusion of their training share this perception. Apparently, a physician views another physician as influential if the latter has greater expertise in a special area. but not if the gap between their levels of knowledge is too great (e.g. between interns and subspecialists). In the United States. the Council on General Internal Medicine of the American Internal Medicine of the American Board of Internal Medicine has advocated training programs in which general internists serve as role models with major teaching responsibilities and that medical subspecialty rotations also be included in the comprehensive training of the general internist [7]. If our house staff are representative of the population of general medicine house staffs, then there are two possible errors which can occur in their training: (1) starting subspecialty rotations too early in training and (2) continuing major teaching responsibility by general internists beyond the time when more advanced residents are primarily heeding subspecialty advice. Our results also suggest that the perceived value of an information source depends upon the particular kind of medical decision being made. Interns and residents see therapeutic decisions as subject to much greater influence than are the primary preventive decisions. such as the use of influenza and pneumococcal pneumonia immunization. The house staff view their own past experience as being more important for therapeutic than for preventive decisions. The value of past experience may depend upon the speed and frequency with which physicians receive positive
Fig. 3. Perceived influence on clinical decisions as a function of the sources of information and the medical procedure. Information sources: journal reading (01. conferences (ml. subspecialty staff (A). general internists (3). past experience (0) and house staff (Al.
1364
STUART
J. COHES et trl.
feedback. The result of initiating a new therapy is usually clearer and more immediate than the feedback from immunization. For example. a patient may not manifest pneumococcal pneumonia because the vaccine was effective or there was no exposure. In contrast, the effectiveness of a beta blocker to lower blood pressure can be determined more immediately and unambiguously. The greater importance credited to past experience for occult blood testing than for mammography and cervical Papanicolaou testing may be due to the higher rate of positive feedback for occult blood testing (3”/;) than for the rate of the other two (0.5-l”,;). If past experience is to serve an edu: cational function. especially for nontherapeutic decisions, some practice audit or reporting system may be required to assist house staff in assessing the merit of their actions. For example, periodic circulation of the number of cases of pneumonia or influenza contracted among the high-risk vaccinated vs high-risk unvaccinated may lend objectivity to house staffs’ subjective impressions of past experience. The finding that physicians perceived their decisions to be influenced more by journal reading than by other sources of information complements other studies which report that reading is consistently preferred over other modes of continuing education [l-5]. These results do not prove that physician reading is the mode of choice because it most effectively improves care; reading may be selected simply because of its low cost and ready access. However, since attendance in continuing medical education courses is frequently mandated, the preference for reading is an important consideration. Moreover, reading is an inexpensive way to obtain information. Postgraduate courses cost physicians 1WlOOO times as much as obtaining the same information by reading medical journals [8]. Our cross sectional study indicates that differences exist at different levels of training regarding perceptions about the influence of common sources of medical information on clinical decisions. In the past, formal postgraduate courses were favored over reading by the mandators of continuing education because
attendance at courses could be verified: time spent reading could not. However. testing programs using newsletters to update medical knowledge should be greatly expanded to give the physician learning options equal in scope to those avallable from continuing education courses. This allows physicians to receive credit for learning etklently on their own. Such options would reflect the evidence that physicians prefer reading as a source of information. This preference is found early in their postgraduate medical career and remains strong while other information sources are seen as less influential for physicians more advanced in experience. Acfiilowledgr,,lerlts-Supporred
in part by grants from the National Institute for Arthritis. Metabolism and Dlgestlve Diseases (PHS P60 AM 205421 and the National Center for Health Services Research INo. 502185-HEW). REFERESCES D. B. and Werthelmer .A. I. Sources of I. Christensen lnformatlon and influence on new drug prescribmg among physxians in an HMO. Sw. %I. .Lfrd. 13A, 313, 1979. 2. Stinson E. R. and Mueller D. A. Survey of health professionals’ information habits and needs. JAM,-L 243. 140. 1980. F. E. Continumg educatmn 3. Guptill P. B. and Graham activities of physicians in solo and group practice. Mrd. Carr 14, 173. 1976. 4. Manning P. R. and Denson T. A. How cardiologists learn about echocardiography: a remmder for medical educators and legislators. Aw1. lfltern. Mrd. 92. 690. 1980. P. R. and Denson T. A. How lntermsts 5. Manning learned about cimetidine. Awl. itlrern. Mrd. 92, 690. 1980. B. J. Srtrtistictrl Prir~cipfrs 01 E.uprrimr~~tu/ 6. Winer Drsiyn. McGraw-Hill, New York. 1971. 7. Council on General Internal Medicine. American Board of Internal Medicine. Attributes of the general internist and recommendations for trainmg. Ann. intrrn. Med. 86, 472. 1977. 8. Huth E. Continuing medical education: Needs. costs and consequences. Ann. intern. Med. 93. 698. 1980.