Risk Aversion and Costs A Comparison of Family Physicians and General Internists

Many^1-5 studies (though not all^6-8) show that family physicians generate lower patient costs than internists. Family physicians have been reported to spend less time per patient,² order fewer laboratory tests,^1,2 charge less,⁹ refer patients less frequently to consultants,^1,3,10 and generate lower hospital¹ and overall costs.³ These findings have been replicated with residents^1,5 and practicing physicians,^2-5 using written patient scenarios,^5,11 in national samples,^1,2,10,12 after adjustment for differences in rates of morbidity (case mix),^1,3,9,10 and recently in a randomized trial.⁴

The reasons for these interspecialty differences in costs are not clear. However, research on physician attributes suggests potential explanations. The personality profile of medical students planning to go into family medicine shows a higher tolerance for ambiguity or uncertainty,¹³ which has been found to be inversely related to the likelihood of referral for genetic counseling.¹⁴ A related construct, risk aversion (a general avoidance of risk taking in life), has been linked to greater use of diagnostic testing and referrals^15,16 and greater likelihood of hospitalization for chest pain.¹⁷ Physicians participating in a Medicare health maintenance organization who reported less anxiety about uncertainty or who were less risk averse generated lower patient charges.¹⁸

For our study, we hypothesized that less anxiety generated by uncertainty and less risk aversion would explain differences in expenditures generated by family physicians and general internists.¹⁸ We used data from a panel of family physicians and internists in a local managed care organization (MCO) to explore these hypotheses.

Methods

Sample

We conducted our investigation in the Rochester, NY, metropolitan area using the claims database of the largest local MCO. Approximately 500,000 people (more than 50% of the local population) were enrolled in the MCO. It employed an independent practitioner association model in which the primary care physicians and the specialists were not capitated. Each patient was assigned to a primary care physician, and more than 95% of those physicians participated in the independent practitioner model. Our patient study sample included adults enrolled in the MCO who were aged 25 years or older and were assigned to a primary care physician (457 family physicians and internists) during 1995 or 1996. To facilitate comparisons between the 2 specialties, visits to obstetrician/gynecologists or for obstetric and gynecologic problems were excluded. Our final sample was made up of approximately 243,000 adult patients of whom 210,000 were active in the system during the year. We derived information on physician specialty, age, and sex from a database maintained by the independent practitioner association.

Physician Survey

Physicians were offered a $50 honorarium to encourage participation in a mailed survey. The 10-minute survey was sent to primary care physicians (internists and family physicians) in the independent practitioner association who had at least 100 patients in the MCO in both 1995 and 1996 (n = 274). There were survey responses from 182 physicians. Questionnaire survey data on all sampled physicians included demographics (age and sex), practice characteristics, specialty (family practice or internal medicine), years in practice (current site or any site), practice intensity (sessions and patients per week), and group size (number of partners). The questionnaire also included several psychometric scales, each with a 5- to 7-choice Likert-type response alternative. The scales were selected on the basis of the evidence of their reliability and their relationship to physician behavior. Physician satisfaction was measured using a scale developed by Linn and colleagues.¹⁹ The original 13-item scale was augmented with 3 additional questions about satisfaction with managed care (Cronbach’s a = 0.87 in this sample). Physicians’ anxiety generated by uncertainty was assessed with 3 items selected from the scales developed by Gerrity and coworkers^20,21 to measure physicians’ reactions to uncertainty. The scale includes items like: “The uncertainty of patient care often troubles me” and “I usually feel anxious when I am not sure of a diagnosis” (Cronbach’s a = 0.76). Attitude toward risk in general (risk aversion) was assessed using a 6-item scale developed by Pearson and colleagues¹⁷ that was found to predict the likelihood of physicians admitting patients with chest pain. The scale includes items such as: “I enjoy taking risks”; “I consider security an important element in every aspect of my life”; and “I try to avoid situations that have uncertain outcomes” (Cronbach’s a = 0.84). A 6-item scale to measure malpractice concerns was developed by one of the authors (GCW). Its items included, “Sometimes I ask for consultant opinions primarily to reduce my risk of being sued” and “Relying on clinical judgment rather than on technology to make a diagnosis is becoming riskier from a medicolegal perspective” (Cronbach’s a = 0.84). An 8-item version of the Physician Psychosocial Belief Scale²² was included. The questions included: “I do not focus on psychosocial problems until I have ruled out organic disease” and “Patients will reject the idea of my dealing with psychosocial issues” (Cronbach’s a = 0.84). Levinson and Roter²³ found that physicians’ scores on this scale correlated with their communication behaviors during audiotaped encounters. Two scales developed by one of the authors (GCW) assessed the extent to which the physician’s motivation for ordering tests or referring to specialists is controlled (7 items) and autonomous (4 items). Each item started with the stem “The reasons I order diagnostic tests or refer my patients to specialists are…” and included possible answers such as: “… because my reputation is at stake with each decision I make” and “…because my patients would be upset if I didn’t” to indicate controlled motivation and “…because it helps me fully understand what is causing my patients’ problem” and “because it’s in my patients’ best interests” to show autonomous motivation. (Cronbach’s a = 0.83 for the controlled scale and 0.82 for the autonomous scale).* The validity and reliability of these scales and constructs have been demonstrated in previous studies.^24-27

Complete survey information was available for 173 (63%) physicians. Those who did not complete the questionnaire were not statistically significantly different in age or sex from those who did. Physicians who completed the questionnaire were more likely to be family physicians (34% vs 21%, P = .02). Physicians who completed the questionnaire also had lower per-member observed expenditures ($1136 vs $1487, P = .02) and lower per-member expected expenditures ($1167 vs $1373, P = .02), but the ratio of observed to expected expenditures was not statistically significantly different (P = .80).

Observed Expenditures

Total expenditures per physician for each member who was enrolled for all of 1995 were calculated from the “allowed amount” variable in the claims files. The allowed amount is the sum of the amount paid, the copayment, the deductible, and the amount withheld for the risk pool. Since the allowed amounts varied across providers, we standardized prices using the claims data. For physician claims, our standardized prices were the average amounts allowed for each Current Procedural Terminology, 4th edition (CPT-4) code and provider specialty. For inpatient hospital claims, our standardized price was the average or allowed amounts by diagnosis-related group. For all other claims, our standardized price was the average of the amounts allowed by CPT-4 code, with separate facility and nonfacility categories.

We defined the total expenditures for each patient as the sum of the standardized prices for all services listed on the patient’s claims for the calendar year. And we calculated the observed expenditures per panel member for each physician using the mean of the total expenditures for each patient in the physician’s panel, with nonusers assigned 0 expenditures.

Case-Mix Adjustment

We developed an expected expenditures per panel member measure to adjust for possible case-mix differences between family physicians and internists. Following the standard method of Duan and colleagues,²⁸ we used a 2-part model to derive anticipated expenditures per panel member: The first part predicted the probability that services were used, and the second predicted expenditures contingent on use.²⁸ The product of those 2 predictions was considered to be the anticipated expenditures per panel member. Since little was known about panel members who did not use any services, the proportions of patients who used some services for 10 age/sex categories were used to define the predicted probability of using some services. Predicted expenditures among users were made on the basis of patient age, sex, and case-mix methodology using the ambulatory care groups system.²⁹ The logarithm of expenditures contingent on some use of services was predicted with an ordinary least squares regression using ambulatory diagnostic groups, age group, and sex. We used ambulatory diagnostic groups instead of ambulatory care groups because they explained more variance. The logarithmic transformation of expenditures was used in the regression analyses because of the extremely skewed distribution of expenditures.

The result was retransformed into anticipated expenditures for those who used services with the “smearing” estimate recommended by Duan and coworkers.³⁰ Smearing is recommended because simply retransforming the logarithms results in an underestimate of expenditures above the median.* The anticipated expenditures per panel member for each physician was calculated as the mean of the anticipated expenditures for each patient in the physician’s panel, including those who did and did not use services.

Analyses

We conducted the analyses in 3 parts. First, we sought to identify psychological factors that would distinguish family physicians from internists. Second, we examined whether there was a relationship between generated physician costs and specialty or psychological factors. Third, we explored the extent to which the psychological factors found to distinguish family physicians from internists also accounted for the differences in generated costs.

We used logistic regression to compare the demographic, practice, and psychometric profiles of family physicians and internists. In this analysis, physician specialty (family physician or internist) was the dependent variable and the demographic, practice, and psychometric scale scores were the independent variables. We used stepwise regression to determine which factors made a statistically significant contribution to distinguishing family physicians from internists. Second, we used physician-level ordinary least squares regression analyses to examine the relationships between observed logarithmic expenditures per panel member and specialty or the psychological factors discriminating between the 2 specialties. Covariate adjustment included logarithmic anticipated expenditures per panel member and significant physician demographic and practice variables. A final analysis with logarithmic observed expenditures as the dependent variable included both specialty and the psychological factors found to differ between the 2 specialties as independent variables. Covariate adjustment again included logarithmic anticipated expenditures and significant physician demographic and practice variables. We used logarithms of expenditures per panel member because both the total and anticipated expenditures exhibited skewed distributions that were normalized by logarithmic transformation. The expenditure analyses were weighted by the physician panel size.

Results

Table 1 shows the comparisons between family physicians and internists for the 173 physicians providing complete information. Family physicians were younger and had fewer office sessions per week. Among the psychological factors, family physicians reported less anxiety generated by uncertainty and were less risk averse; none of the other psychological factors showed specialty differences. Family physicians also had lower observed and anticipated expenditures per panel member.

After using the stepwise logistic regression Table 2, anxiety generated by uncertainty was no longer significantly associated with specialty. The receiver operator curve analysis for this logistic regression revealed an area below the curve of 0.74, indicating a good fit for the regression model for independent predictors of specialty.

In the physician-level ordinary linear regression Table 3, family physicians had observed per-panel-member expenditures that were 5.3% lower (95% confidence interval [CI], 1.8% - 9.0%) after adjustment for anticipated expenditures. None of the physician demographic or practice variables made statistically significant contributions to explaining expenditures. After adjustment for case mix, risk-averse physicians generated higher expenditures Table 3; a 1 standard deviation increase in risk aversion was associated with a 2.4% increase in expenditures (95% CI, 0.5% - 4.4%). After adjustment for case mix and risk aversion Table 3, family physicians’ costs were no longer significantly lower (3.3%; 95% CI, -1.2 to 8.1%). The effect of risk aversion remained significant; a 1 standard deviation increase in risk aversion was associated with a 2.1% increase in expenditures (95% CI, 0.1% to 4.1%). None of the other physician demographic, practice, or psychological factors were significantly associated with case-mix adjusted expenditures.

Discussion

Family physicians had significantly less anxiety generated by uncertainty, less risk aversion, and worked fewer sessions per week than internists; after multivariate adjustment, the effect of anxiety generated by uncertainty was no longer statistically significant. Consistent with previous studies,^1-5 we found that family physicians generated lower overall costs than general internists. After adjusting for patient case mix, the costs per enrolled member for family physicians were 5% lower. When the risk aversion measure was included in the regression model, the expenditures generated by family physicians were no longer significantly lower than those of internists. Risk aversion, however, was associated with significantly lower patient costs both before and after adjustment for specialty. These findings suggest that interspecialty differences in costs may be related to the greater risk aversion of internists.

Although causality cannot be established through an observational study, the findings were consistent with our hypothesis that attitude toward risk influences physician behavior. Previous studies have shown that physicians who are more averse to risk order more tests,¹⁵ refer more often,¹⁶ hospitalize more frequently,¹⁴ and generate higher overall costs.¹⁸ The aspects of risk aversion that affect utilization are not clear. There was no effect of fear of malpractice or of anxiety generated by uncertainty on costs. Although these 2 scales address some aspects of risk aversion, they were focused on medical issues. The risk aversion scale we used is not specific to clinical risk taking but taps the broader domain of attitude toward risk taking in general.

There are 2 plausible explanations for the lower risk aversion of family physicians than that of general internists: self-selection and residency training. Medical students who are less risk averse may select family medicine because of its biopsychosocial approach, attention to the health of families, and a broad focus that cuts across traditional specialties defined by age or organ system. Less risk-averse students may also be drawn to family medicine’s countercultural roots.³¹ Differences in residency training between family medicine and general internal medicine may also contribute to specialty differences in risk aversion. Historically, internal medicine residency training has been more hospital-based than family medicine and has emphasized differential diagnosis and thorough diagnostic evaluations. Much of internal medicine residency teaching is done by subspecialists who emphasize diagnosis and management of diseases that are seen less frequently in primary care. Our study cannot address the relative importance of self-selection or training in accounting for differences in risk aversion between the 2 specialties; the nature of our scale, however, suggests that self-selection plays a larger role. The scale measures attitude toward risk taking in general, and it seems less likely that residency training would influence such a broad construct. It is possible that as family medicine becomes more mainstream and general internal medicine becomes more primary care based³² cost differences between the specialties will decline.

The validity of the these findings is strengthened by our use of a community sample of board-certified physicians, a comprehensive MCO claims database, and careful case-mix adjustment. Although these findings are not necessarily generalizable to other communities, the finding of lower expenditures for family physicians than internists is consistent with results from national samples.^1,2,10,12 The finding that these differences are related to risk taking is new and warrants replication.

Limitations

Our study has several limitations. First, we did not address the question of whether specialty differences in costs or risk aversion were associated with differences in patient outcomes. Second, our study was conducted in a community where there had been an emphasis on the biopsychosocial model among both internists and family physicians.³³ It is not clear how this tradition might affect the generalizability of these findings to other communities. Third, the specialty difference in expenditures both before and after adjustment for risk aversion was modest. This small difference reflects, in part, the contradictory findings previously reported in the literature.^1-8 Fourth, it is possible that differences in case mix between specialties not fully captured through ambulatory diagnostic groups explain the observed differences between specialties. Other unmeasured patient factors could also account for our findings. For example, patients who are more risk averse and desire more health care may seek physicians with a similar psychological orientation. Finally, it is possible that response bias accounted for these findings. Slightly less than two thirds of the physicians submitted fully completed surveys. If family physicians with lower costs responded disproportionately to internists with lower costs, then response bias would produce a spurious relationship between specialty and costs. However, an analysis that included both completion of the survey as a variable and the interaction of that variable with specialty revealed that neither completion of the survey nor the interaction effects were statistically significant. A similar potential bias exists for risk aversion; this, however, could not be examined.

Conclusions

We found that the family physicians compared had significantly lower total adjusted medical expenditures than general internists for managed care patients. These differences were no longer statistically significant after adjustment for specialty differences in risk aversion. Further study is needed to determine the origin of these specialty differences and whether those differences affect patient outcomes.

Acknowledgments

This study was supported by grant R01 HS09397-01 from the Agency for Health Care Policy and Research.

Many^1-5 studies (though not all^6-8) show that family physicians generate lower patient costs than internists. Family physicians have been reported to spend less time per patient,² order fewer laboratory tests,^1,2 charge less,⁹ refer patients less frequently to consultants,^1,3,10 and generate lower hospital¹ and overall costs.³ These findings have been replicated with residents^1,5 and practicing physicians,^2-5 using written patient scenarios,^5,11 in national samples,^1,2,10,12 after adjustment for differences in rates of morbidity (case mix),^1,3,9,10 and recently in a randomized trial.⁴

The reasons for these interspecialty differences in costs are not clear. However, research on physician attributes suggests potential explanations. The personality profile of medical students planning to go into family medicine shows a higher tolerance for ambiguity or uncertainty,¹³ which has been found to be inversely related to the likelihood of referral for genetic counseling.¹⁴ A related construct, risk aversion (a general avoidance of risk taking in life), has been linked to greater use of diagnostic testing and referrals^15,16 and greater likelihood of hospitalization for chest pain.¹⁷ Physicians participating in a Medicare health maintenance organization who reported less anxiety about uncertainty or who were less risk averse generated lower patient charges.¹⁸

For our study, we hypothesized that less anxiety generated by uncertainty and less risk aversion would explain differences in expenditures generated by family physicians and general internists.¹⁸ We used data from a panel of family physicians and internists in a local managed care organization (MCO) to explore these hypotheses.

Methods

Sample

We conducted our investigation in the Rochester, NY, metropolitan area using the claims database of the largest local MCO. Approximately 500,000 people (more than 50% of the local population) were enrolled in the MCO. It employed an independent practitioner association model in which the primary care physicians and the specialists were not capitated. Each patient was assigned to a primary care physician, and more than 95% of those physicians participated in the independent practitioner model. Our patient study sample included adults enrolled in the MCO who were aged 25 years or older and were assigned to a primary care physician (457 family physicians and internists) during 1995 or 1996. To facilitate comparisons between the 2 specialties, visits to obstetrician/gynecologists or for obstetric and gynecologic problems were excluded. Our final sample was made up of approximately 243,000 adult patients of whom 210,000 were active in the system during the year. We derived information on physician specialty, age, and sex from a database maintained by the independent practitioner association.

Physician Survey

Physicians were offered a $50 honorarium to encourage participation in a mailed survey. The 10-minute survey was sent to primary care physicians (internists and family physicians) in the independent practitioner association who had at least 100 patients in the MCO in both 1995 and 1996 (n = 274). There were survey responses from 182 physicians. Questionnaire survey data on all sampled physicians included demographics (age and sex), practice characteristics, specialty (family practice or internal medicine), years in practice (current site or any site), practice intensity (sessions and patients per week), and group size (number of partners). The questionnaire also included several psychometric scales, each with a 5- to 7-choice Likert-type response alternative. The scales were selected on the basis of the evidence of their reliability and their relationship to physician behavior. Physician satisfaction was measured using a scale developed by Linn and colleagues.¹⁹ The original 13-item scale was augmented with 3 additional questions about satisfaction with managed care (Cronbach’s a = 0.87 in this sample). Physicians’ anxiety generated by uncertainty was assessed with 3 items selected from the scales developed by Gerrity and coworkers^20,21 to measure physicians’ reactions to uncertainty. The scale includes items like: “The uncertainty of patient care often troubles me” and “I usually feel anxious when I am not sure of a diagnosis” (Cronbach’s a = 0.76). Attitude toward risk in general (risk aversion) was assessed using a 6-item scale developed by Pearson and colleagues¹⁷ that was found to predict the likelihood of physicians admitting patients with chest pain. The scale includes items such as: “I enjoy taking risks”; “I consider security an important element in every aspect of my life”; and “I try to avoid situations that have uncertain outcomes” (Cronbach’s a = 0.84). A 6-item scale to measure malpractice concerns was developed by one of the authors (GCW). Its items included, “Sometimes I ask for consultant opinions primarily to reduce my risk of being sued” and “Relying on clinical judgment rather than on technology to make a diagnosis is becoming riskier from a medicolegal perspective” (Cronbach’s a = 0.84). An 8-item version of the Physician Psychosocial Belief Scale²² was included. The questions included: “I do not focus on psychosocial problems until I have ruled out organic disease” and “Patients will reject the idea of my dealing with psychosocial issues” (Cronbach’s a = 0.84). Levinson and Roter²³ found that physicians’ scores on this scale correlated with their communication behaviors during audiotaped encounters. Two scales developed by one of the authors (GCW) assessed the extent to which the physician’s motivation for ordering tests or referring to specialists is controlled (7 items) and autonomous (4 items). Each item started with the stem “The reasons I order diagnostic tests or refer my patients to specialists are…” and included possible answers such as: “… because my reputation is at stake with each decision I make” and “…because my patients would be upset if I didn’t” to indicate controlled motivation and “…because it helps me fully understand what is causing my patients’ problem” and “because it’s in my patients’ best interests” to show autonomous motivation. (Cronbach’s a = 0.83 for the controlled scale and 0.82 for the autonomous scale).* The validity and reliability of these scales and constructs have been demonstrated in previous studies.^24-27

Complete survey information was available for 173 (63%) physicians. Those who did not complete the questionnaire were not statistically significantly different in age or sex from those who did. Physicians who completed the questionnaire were more likely to be family physicians (34% vs 21%, P = .02). Physicians who completed the questionnaire also had lower per-member observed expenditures ($1136 vs $1487, P = .02) and lower per-member expected expenditures ($1167 vs $1373, P = .02), but the ratio of observed to expected expenditures was not statistically significantly different (P = .80).

Observed Expenditures

Total expenditures per physician for each member who was enrolled for all of 1995 were calculated from the “allowed amount” variable in the claims files. The allowed amount is the sum of the amount paid, the copayment, the deductible, and the amount withheld for the risk pool. Since the allowed amounts varied across providers, we standardized prices using the claims data. For physician claims, our standardized prices were the average amounts allowed for each Current Procedural Terminology, 4th edition (CPT-4) code and provider specialty. For inpatient hospital claims, our standardized price was the average or allowed amounts by diagnosis-related group. For all other claims, our standardized price was the average of the amounts allowed by CPT-4 code, with separate facility and nonfacility categories.

We defined the total expenditures for each patient as the sum of the standardized prices for all services listed on the patient’s claims for the calendar year. And we calculated the observed expenditures per panel member for each physician using the mean of the total expenditures for each patient in the physician’s panel, with nonusers assigned 0 expenditures.

Case-Mix Adjustment

We developed an expected expenditures per panel member measure to adjust for possible case-mix differences between family physicians and internists. Following the standard method of Duan and colleagues,²⁸ we used a 2-part model to derive anticipated expenditures per panel member: The first part predicted the probability that services were used, and the second predicted expenditures contingent on use.²⁸ The product of those 2 predictions was considered to be the anticipated expenditures per panel member. Since little was known about panel members who did not use any services, the proportions of patients who used some services for 10 age/sex categories were used to define the predicted probability of using some services. Predicted expenditures among users were made on the basis of patient age, sex, and case-mix methodology using the ambulatory care groups system.²⁹ The logarithm of expenditures contingent on some use of services was predicted with an ordinary least squares regression using ambulatory diagnostic groups, age group, and sex. We used ambulatory diagnostic groups instead of ambulatory care groups because they explained more variance. The logarithmic transformation of expenditures was used in the regression analyses because of the extremely skewed distribution of expenditures.

The result was retransformed into anticipated expenditures for those who used services with the “smearing” estimate recommended by Duan and coworkers.³⁰ Smearing is recommended because simply retransforming the logarithms results in an underestimate of expenditures above the median.* The anticipated expenditures per panel member for each physician was calculated as the mean of the anticipated expenditures for each patient in the physician’s panel, including those who did and did not use services.

Analyses

We conducted the analyses in 3 parts. First, we sought to identify psychological factors that would distinguish family physicians from internists. Second, we examined whether there was a relationship between generated physician costs and specialty or psychological factors. Third, we explored the extent to which the psychological factors found to distinguish family physicians from internists also accounted for the differences in generated costs.

We used logistic regression to compare the demographic, practice, and psychometric profiles of family physicians and internists. In this analysis, physician specialty (family physician or internist) was the dependent variable and the demographic, practice, and psychometric scale scores were the independent variables. We used stepwise regression to determine which factors made a statistically significant contribution to distinguishing family physicians from internists. Second, we used physician-level ordinary least squares regression analyses to examine the relationships between observed logarithmic expenditures per panel member and specialty or the psychological factors discriminating between the 2 specialties. Covariate adjustment included logarithmic anticipated expenditures per panel member and significant physician demographic and practice variables. A final analysis with logarithmic observed expenditures as the dependent variable included both specialty and the psychological factors found to differ between the 2 specialties as independent variables. Covariate adjustment again included logarithmic anticipated expenditures and significant physician demographic and practice variables. We used logarithms of expenditures per panel member because both the total and anticipated expenditures exhibited skewed distributions that were normalized by logarithmic transformation. The expenditure analyses were weighted by the physician panel size.

Results

Table 1 shows the comparisons between family physicians and internists for the 173 physicians providing complete information. Family physicians were younger and had fewer office sessions per week. Among the psychological factors, family physicians reported less anxiety generated by uncertainty and were less risk averse; none of the other psychological factors showed specialty differences. Family physicians also had lower observed and anticipated expenditures per panel member.

After using the stepwise logistic regression Table 2, anxiety generated by uncertainty was no longer significantly associated with specialty. The receiver operator curve analysis for this logistic regression revealed an area below the curve of 0.74, indicating a good fit for the regression model for independent predictors of specialty.

In the physician-level ordinary linear regression Table 3, family physicians had observed per-panel-member expenditures that were 5.3% lower (95% confidence interval [CI], 1.8% - 9.0%) after adjustment for anticipated expenditures. None of the physician demographic or practice variables made statistically significant contributions to explaining expenditures. After adjustment for case mix, risk-averse physicians generated higher expenditures Table 3; a 1 standard deviation increase in risk aversion was associated with a 2.4% increase in expenditures (95% CI, 0.5% - 4.4%). After adjustment for case mix and risk aversion Table 3, family physicians’ costs were no longer significantly lower (3.3%; 95% CI, -1.2 to 8.1%). The effect of risk aversion remained significant; a 1 standard deviation increase in risk aversion was associated with a 2.1% increase in expenditures (95% CI, 0.1% to 4.1%). None of the other physician demographic, practice, or psychological factors were significantly associated with case-mix adjusted expenditures.

Discussion

Family physicians had significantly less anxiety generated by uncertainty, less risk aversion, and worked fewer sessions per week than internists; after multivariate adjustment, the effect of anxiety generated by uncertainty was no longer statistically significant. Consistent with previous studies,^1-5 we found that family physicians generated lower overall costs than general internists. After adjusting for patient case mix, the costs per enrolled member for family physicians were 5% lower. When the risk aversion measure was included in the regression model, the expenditures generated by family physicians were no longer significantly lower than those of internists. Risk aversion, however, was associated with significantly lower patient costs both before and after adjustment for specialty. These findings suggest that interspecialty differences in costs may be related to the greater risk aversion of internists.

Although causality cannot be established through an observational study, the findings were consistent with our hypothesis that attitude toward risk influences physician behavior. Previous studies have shown that physicians who are more averse to risk order more tests,¹⁵ refer more often,¹⁶ hospitalize more frequently,¹⁴ and generate higher overall costs.¹⁸ The aspects of risk aversion that affect utilization are not clear. There was no effect of fear of malpractice or of anxiety generated by uncertainty on costs. Although these 2 scales address some aspects of risk aversion, they were focused on medical issues. The risk aversion scale we used is not specific to clinical risk taking but taps the broader domain of attitude toward risk taking in general.

There are 2 plausible explanations for the lower risk aversion of family physicians than that of general internists: self-selection and residency training. Medical students who are less risk averse may select family medicine because of its biopsychosocial approach, attention to the health of families, and a broad focus that cuts across traditional specialties defined by age or organ system. Less risk-averse students may also be drawn to family medicine’s countercultural roots.³¹ Differences in residency training between family medicine and general internal medicine may also contribute to specialty differences in risk aversion. Historically, internal medicine residency training has been more hospital-based than family medicine and has emphasized differential diagnosis and thorough diagnostic evaluations. Much of internal medicine residency teaching is done by subspecialists who emphasize diagnosis and management of diseases that are seen less frequently in primary care. Our study cannot address the relative importance of self-selection or training in accounting for differences in risk aversion between the 2 specialties; the nature of our scale, however, suggests that self-selection plays a larger role. The scale measures attitude toward risk taking in general, and it seems less likely that residency training would influence such a broad construct. It is possible that as family medicine becomes more mainstream and general internal medicine becomes more primary care based³² cost differences between the specialties will decline.

The validity of the these findings is strengthened by our use of a community sample of board-certified physicians, a comprehensive MCO claims database, and careful case-mix adjustment. Although these findings are not necessarily generalizable to other communities, the finding of lower expenditures for family physicians than internists is consistent with results from national samples.^1,2,10,12 The finding that these differences are related to risk taking is new and warrants replication.

Limitations

Our study has several limitations. First, we did not address the question of whether specialty differences in costs or risk aversion were associated with differences in patient outcomes. Second, our study was conducted in a community where there had been an emphasis on the biopsychosocial model among both internists and family physicians.³³ It is not clear how this tradition might affect the generalizability of these findings to other communities. Third, the specialty difference in expenditures both before and after adjustment for risk aversion was modest. This small difference reflects, in part, the contradictory findings previously reported in the literature.^1-8 Fourth, it is possible that differences in case mix between specialties not fully captured through ambulatory diagnostic groups explain the observed differences between specialties. Other unmeasured patient factors could also account for our findings. For example, patients who are more risk averse and desire more health care may seek physicians with a similar psychological orientation. Finally, it is possible that response bias accounted for these findings. Slightly less than two thirds of the physicians submitted fully completed surveys. If family physicians with lower costs responded disproportionately to internists with lower costs, then response bias would produce a spurious relationship between specialty and costs. However, an analysis that included both completion of the survey as a variable and the interaction of that variable with specialty revealed that neither completion of the survey nor the interaction effects were statistically significant. A similar potential bias exists for risk aversion; this, however, could not be examined.

Conclusions

We found that the family physicians compared had significantly lower total adjusted medical expenditures than general internists for managed care patients. These differences were no longer statistically significant after adjustment for specialty differences in risk aversion. Further study is needed to determine the origin of these specialty differences and whether those differences affect patient outcomes.

Acknowledgments

This study was supported by grant R01 HS09397-01 from the Agency for Health Care Policy and Research.

Many^1-5 studies (though not all^6-8) show that family physicians generate lower patient costs than internists. Family physicians have been reported to spend less time per patient,² order fewer laboratory tests,^1,2 charge less,⁹ refer patients less frequently to consultants,^1,3,10 and generate lower hospital¹ and overall costs.³ These findings have been replicated with residents^1,5 and practicing physicians,^2-5 using written patient scenarios,^5,11 in national samples,^1,2,10,12 after adjustment for differences in rates of morbidity (case mix),^1,3,9,10 and recently in a randomized trial.⁴

The reasons for these interspecialty differences in costs are not clear. However, research on physician attributes suggests potential explanations. The personality profile of medical students planning to go into family medicine shows a higher tolerance for ambiguity or uncertainty,¹³ which has been found to be inversely related to the likelihood of referral for genetic counseling.¹⁴ A related construct, risk aversion (a general avoidance of risk taking in life), has been linked to greater use of diagnostic testing and referrals^15,16 and greater likelihood of hospitalization for chest pain.¹⁷ Physicians participating in a Medicare health maintenance organization who reported less anxiety about uncertainty or who were less risk averse generated lower patient charges.¹⁸

For our study, we hypothesized that less anxiety generated by uncertainty and less risk aversion would explain differences in expenditures generated by family physicians and general internists.¹⁸ We used data from a panel of family physicians and internists in a local managed care organization (MCO) to explore these hypotheses.

Methods

Sample

We conducted our investigation in the Rochester, NY, metropolitan area using the claims database of the largest local MCO. Approximately 500,000 people (more than 50% of the local population) were enrolled in the MCO. It employed an independent practitioner association model in which the primary care physicians and the specialists were not capitated. Each patient was assigned to a primary care physician, and more than 95% of those physicians participated in the independent practitioner model. Our patient study sample included adults enrolled in the MCO who were aged 25 years or older and were assigned to a primary care physician (457 family physicians and internists) during 1995 or 1996. To facilitate comparisons between the 2 specialties, visits to obstetrician/gynecologists or for obstetric and gynecologic problems were excluded. Our final sample was made up of approximately 243,000 adult patients of whom 210,000 were active in the system during the year. We derived information on physician specialty, age, and sex from a database maintained by the independent practitioner association.

Physician Survey

Physicians were offered a $50 honorarium to encourage participation in a mailed survey. The 10-minute survey was sent to primary care physicians (internists and family physicians) in the independent practitioner association who had at least 100 patients in the MCO in both 1995 and 1996 (n = 274). There were survey responses from 182 physicians. Questionnaire survey data on all sampled physicians included demographics (age and sex), practice characteristics, specialty (family practice or internal medicine), years in practice (current site or any site), practice intensity (sessions and patients per week), and group size (number of partners). The questionnaire also included several psychometric scales, each with a 5- to 7-choice Likert-type response alternative. The scales were selected on the basis of the evidence of their reliability and their relationship to physician behavior. Physician satisfaction was measured using a scale developed by Linn and colleagues.¹⁹ The original 13-item scale was augmented with 3 additional questions about satisfaction with managed care (Cronbach’s a = 0.87 in this sample). Physicians’ anxiety generated by uncertainty was assessed with 3 items selected from the scales developed by Gerrity and coworkers^20,21 to measure physicians’ reactions to uncertainty. The scale includes items like: “The uncertainty of patient care often troubles me” and “I usually feel anxious when I am not sure of a diagnosis” (Cronbach’s a = 0.76). Attitude toward risk in general (risk aversion) was assessed using a 6-item scale developed by Pearson and colleagues¹⁷ that was found to predict the likelihood of physicians admitting patients with chest pain. The scale includes items such as: “I enjoy taking risks”; “I consider security an important element in every aspect of my life”; and “I try to avoid situations that have uncertain outcomes” (Cronbach’s a = 0.84). A 6-item scale to measure malpractice concerns was developed by one of the authors (GCW). Its items included, “Sometimes I ask for consultant opinions primarily to reduce my risk of being sued” and “Relying on clinical judgment rather than on technology to make a diagnosis is becoming riskier from a medicolegal perspective” (Cronbach’s a = 0.84). An 8-item version of the Physician Psychosocial Belief Scale²² was included. The questions included: “I do not focus on psychosocial problems until I have ruled out organic disease” and “Patients will reject the idea of my dealing with psychosocial issues” (Cronbach’s a = 0.84). Levinson and Roter²³ found that physicians’ scores on this scale correlated with their communication behaviors during audiotaped encounters. Two scales developed by one of the authors (GCW) assessed the extent to which the physician’s motivation for ordering tests or referring to specialists is controlled (7 items) and autonomous (4 items). Each item started with the stem “The reasons I order diagnostic tests or refer my patients to specialists are…” and included possible answers such as: “… because my reputation is at stake with each decision I make” and “…because my patients would be upset if I didn’t” to indicate controlled motivation and “…because it helps me fully understand what is causing my patients’ problem” and “because it’s in my patients’ best interests” to show autonomous motivation. (Cronbach’s a = 0.83 for the controlled scale and 0.82 for the autonomous scale).* The validity and reliability of these scales and constructs have been demonstrated in previous studies.^24-27

Complete survey information was available for 173 (63%) physicians. Those who did not complete the questionnaire were not statistically significantly different in age or sex from those who did. Physicians who completed the questionnaire were more likely to be family physicians (34% vs 21%, P = .02). Physicians who completed the questionnaire also had lower per-member observed expenditures ($1136 vs $1487, P = .02) and lower per-member expected expenditures ($1167 vs $1373, P = .02), but the ratio of observed to expected expenditures was not statistically significantly different (P = .80).

Observed Expenditures

Total expenditures per physician for each member who was enrolled for all of 1995 were calculated from the “allowed amount” variable in the claims files. The allowed amount is the sum of the amount paid, the copayment, the deductible, and the amount withheld for the risk pool. Since the allowed amounts varied across providers, we standardized prices using the claims data. For physician claims, our standardized prices were the average amounts allowed for each Current Procedural Terminology, 4th edition (CPT-4) code and provider specialty. For inpatient hospital claims, our standardized price was the average or allowed amounts by diagnosis-related group. For all other claims, our standardized price was the average of the amounts allowed by CPT-4 code, with separate facility and nonfacility categories.

We defined the total expenditures for each patient as the sum of the standardized prices for all services listed on the patient’s claims for the calendar year. And we calculated the observed expenditures per panel member for each physician using the mean of the total expenditures for each patient in the physician’s panel, with nonusers assigned 0 expenditures.

Case-Mix Adjustment

We developed an expected expenditures per panel member measure to adjust for possible case-mix differences between family physicians and internists. Following the standard method of Duan and colleagues,²⁸ we used a 2-part model to derive anticipated expenditures per panel member: The first part predicted the probability that services were used, and the second predicted expenditures contingent on use.²⁸ The product of those 2 predictions was considered to be the anticipated expenditures per panel member. Since little was known about panel members who did not use any services, the proportions of patients who used some services for 10 age/sex categories were used to define the predicted probability of using some services. Predicted expenditures among users were made on the basis of patient age, sex, and case-mix methodology using the ambulatory care groups system.²⁹ The logarithm of expenditures contingent on some use of services was predicted with an ordinary least squares regression using ambulatory diagnostic groups, age group, and sex. We used ambulatory diagnostic groups instead of ambulatory care groups because they explained more variance. The logarithmic transformation of expenditures was used in the regression analyses because of the extremely skewed distribution of expenditures.

The result was retransformed into anticipated expenditures for those who used services with the “smearing” estimate recommended by Duan and coworkers.³⁰ Smearing is recommended because simply retransforming the logarithms results in an underestimate of expenditures above the median.* The anticipated expenditures per panel member for each physician was calculated as the mean of the anticipated expenditures for each patient in the physician’s panel, including those who did and did not use services.

Analyses

We conducted the analyses in 3 parts. First, we sought to identify psychological factors that would distinguish family physicians from internists. Second, we examined whether there was a relationship between generated physician costs and specialty or psychological factors. Third, we explored the extent to which the psychological factors found to distinguish family physicians from internists also accounted for the differences in generated costs.

We used logistic regression to compare the demographic, practice, and psychometric profiles of family physicians and internists. In this analysis, physician specialty (family physician or internist) was the dependent variable and the demographic, practice, and psychometric scale scores were the independent variables. We used stepwise regression to determine which factors made a statistically significant contribution to distinguishing family physicians from internists. Second, we used physician-level ordinary least squares regression analyses to examine the relationships between observed logarithmic expenditures per panel member and specialty or the psychological factors discriminating between the 2 specialties. Covariate adjustment included logarithmic anticipated expenditures per panel member and significant physician demographic and practice variables. A final analysis with logarithmic observed expenditures as the dependent variable included both specialty and the psychological factors found to differ between the 2 specialties as independent variables. Covariate adjustment again included logarithmic anticipated expenditures and significant physician demographic and practice variables. We used logarithms of expenditures per panel member because both the total and anticipated expenditures exhibited skewed distributions that were normalized by logarithmic transformation. The expenditure analyses were weighted by the physician panel size.

Results

Table 1 shows the comparisons between family physicians and internists for the 173 physicians providing complete information. Family physicians were younger and had fewer office sessions per week. Among the psychological factors, family physicians reported less anxiety generated by uncertainty and were less risk averse; none of the other psychological factors showed specialty differences. Family physicians also had lower observed and anticipated expenditures per panel member.

After using the stepwise logistic regression Table 2, anxiety generated by uncertainty was no longer significantly associated with specialty. The receiver operator curve analysis for this logistic regression revealed an area below the curve of 0.74, indicating a good fit for the regression model for independent predictors of specialty.

In the physician-level ordinary linear regression Table 3, family physicians had observed per-panel-member expenditures that were 5.3% lower (95% confidence interval [CI], 1.8% - 9.0%) after adjustment for anticipated expenditures. None of the physician demographic or practice variables made statistically significant contributions to explaining expenditures. After adjustment for case mix, risk-averse physicians generated higher expenditures Table 3; a 1 standard deviation increase in risk aversion was associated with a 2.4% increase in expenditures (95% CI, 0.5% - 4.4%). After adjustment for case mix and risk aversion Table 3, family physicians’ costs were no longer significantly lower (3.3%; 95% CI, -1.2 to 8.1%). The effect of risk aversion remained significant; a 1 standard deviation increase in risk aversion was associated with a 2.1% increase in expenditures (95% CI, 0.1% to 4.1%). None of the other physician demographic, practice, or psychological factors were significantly associated with case-mix adjusted expenditures.

Discussion

Family physicians had significantly less anxiety generated by uncertainty, less risk aversion, and worked fewer sessions per week than internists; after multivariate adjustment, the effect of anxiety generated by uncertainty was no longer statistically significant. Consistent with previous studies,^1-5 we found that family physicians generated lower overall costs than general internists. After adjusting for patient case mix, the costs per enrolled member for family physicians were 5% lower. When the risk aversion measure was included in the regression model, the expenditures generated by family physicians were no longer significantly lower than those of internists. Risk aversion, however, was associated with significantly lower patient costs both before and after adjustment for specialty. These findings suggest that interspecialty differences in costs may be related to the greater risk aversion of internists.

Although causality cannot be established through an observational study, the findings were consistent with our hypothesis that attitude toward risk influences physician behavior. Previous studies have shown that physicians who are more averse to risk order more tests,¹⁵ refer more often,¹⁶ hospitalize more frequently,¹⁴ and generate higher overall costs.¹⁸ The aspects of risk aversion that affect utilization are not clear. There was no effect of fear of malpractice or of anxiety generated by uncertainty on costs. Although these 2 scales address some aspects of risk aversion, they were focused on medical issues. The risk aversion scale we used is not specific to clinical risk taking but taps the broader domain of attitude toward risk taking in general.

There are 2 plausible explanations for the lower risk aversion of family physicians than that of general internists: self-selection and residency training. Medical students who are less risk averse may select family medicine because of its biopsychosocial approach, attention to the health of families, and a broad focus that cuts across traditional specialties defined by age or organ system. Less risk-averse students may also be drawn to family medicine’s countercultural roots.³¹ Differences in residency training between family medicine and general internal medicine may also contribute to specialty differences in risk aversion. Historically, internal medicine residency training has been more hospital-based than family medicine and has emphasized differential diagnosis and thorough diagnostic evaluations. Much of internal medicine residency teaching is done by subspecialists who emphasize diagnosis and management of diseases that are seen less frequently in primary care. Our study cannot address the relative importance of self-selection or training in accounting for differences in risk aversion between the 2 specialties; the nature of our scale, however, suggests that self-selection plays a larger role. The scale measures attitude toward risk taking in general, and it seems less likely that residency training would influence such a broad construct. It is possible that as family medicine becomes more mainstream and general internal medicine becomes more primary care based³² cost differences between the specialties will decline.

The validity of the these findings is strengthened by our use of a community sample of board-certified physicians, a comprehensive MCO claims database, and careful case-mix adjustment. Although these findings are not necessarily generalizable to other communities, the finding of lower expenditures for family physicians than internists is consistent with results from national samples.^1,2,10,12 The finding that these differences are related to risk taking is new and warrants replication.

Limitations

Our study has several limitations. First, we did not address the question of whether specialty differences in costs or risk aversion were associated with differences in patient outcomes. Second, our study was conducted in a community where there had been an emphasis on the biopsychosocial model among both internists and family physicians.³³ It is not clear how this tradition might affect the generalizability of these findings to other communities. Third, the specialty difference in expenditures both before and after adjustment for risk aversion was modest. This small difference reflects, in part, the contradictory findings previously reported in the literature.^1-8 Fourth, it is possible that differences in case mix between specialties not fully captured through ambulatory diagnostic groups explain the observed differences between specialties. Other unmeasured patient factors could also account for our findings. For example, patients who are more risk averse and desire more health care may seek physicians with a similar psychological orientation. Finally, it is possible that response bias accounted for these findings. Slightly less than two thirds of the physicians submitted fully completed surveys. If family physicians with lower costs responded disproportionately to internists with lower costs, then response bias would produce a spurious relationship between specialty and costs. However, an analysis that included both completion of the survey as a variable and the interaction of that variable with specialty revealed that neither completion of the survey nor the interaction effects were statistically significant. A similar potential bias exists for risk aversion; this, however, could not be examined.

Conclusions

We found that the family physicians compared had significantly lower total adjusted medical expenditures than general internists for managed care patients. These differences were no longer statistically significant after adjustment for specialty differences in risk aversion. Further study is needed to determine the origin of these specialty differences and whether those differences affect patient outcomes.

Acknowledgments

This study was supported by grant R01 HS09397-01 from the Agency for Health Care Policy and Research.