Managed Care, Physician Job Satisfaction, and the Quality of Primary Care
The authors have no conflicts of interest to report.
Address correspondence and requests for reprints to Dr. Grembowski: Department of Health Services, University of Washington, 1959 NE Pacific Street, Box 357660, Seattle, WA 98195-7660 (e-mail: firstname.lastname@example.org).
Objective: To determine the associations between managed care, physician job satisfaction, and the quality of primary care, and to determine whether physician job satisfaction is associated with health outcomes among primary care patients with pain and depressive symptoms.
Design: Prospective cohort study.
Setting: Offices of 261 primary physicians in private practice in Seattle.
Patients: We screened 17,187 patients in waiting rooms, yielding a sample of 1,514 patients with pain only, 575 patients with depressive symptoms only, and 761 patients with pain and depressive symptoms; 2,004 patients completed a 6-month follow-up survey.
Measurements And Results: For each patient, managed care was measured by the intensity of managed care controls in the patient's primary care office, physician financial incentives, and whether the physician read or used back pain and depression guidelines. Physician job satisfaction at baseline was measured through a 6-item scale. Quality of primary care at follow-up was measured by patient rating of care provided by the primary physician, patient trust and confidence in primary physician, quality-of-care index, and continuity of primary physician. Outcomes were pain interference and bothersomeness, Symptom Checklist for Depression, and restricted activity days. Pain and depression patients of physicians with greater job satisfaction had greater trust and confidence in their primary physicians. Pain patients of more satisfied physicians also were less likely to change physicians in the follow-up period. Depression patients of more satisfied physicians had higher ratings of the care provided by their physicians. These associations remained after controlling statistically for managed care. Physician job satisfaction was not associated with health outcomes.
Conclusions: For primary care patients with pain or depressive symptoms, primary physician job satisfaction is associated with some measures of patient-rated quality of care but not health outcomes.
Many primary physicians believe that managed care and market competition have eroded their satisfaction with medical practice.1–9 Little is known about whether physician job dissatisfaction—whether from managed care or other sources—undermines patient perceptions of quality care and health outcomes.10–12
A handful of studies suggest that when physicians are more satisfied with their jobs, quality of care benefits.13–16 Haas et al.13 report that patients of physicians who rated themselves to be very or extremely satisfied with their work were more satisfied with their health care and most recent physician visit. Managed care controls may partly explain this relationship.13,17,18 As the intensity of managed care controls increase, physicians may become more dissatisfied with their jobs,4,19–21 and patients may experience worse quality care,22–26 creating a spurious relationship between physician job satisfaction and quality of care.
If managed care and physician job dissatisfaction contribute to lower quality of primary care, health outcomes may be reduced.17,18,27–29 While patient dissatisfaction is associated with worse health outcomes,30–33 few studies have examined whether physician job dissatisfaction also is associated with worse health outcomes.
The aim of the study is to examine whether physician job satisfaction is associated with patient perceptions of the quality of primary care among patients with pain and depressive symptoms. A second aim is to determine whether physician job satisfaction is associated with health outcomes. We address patients with pain and depressive symptoms because they are common conditions in primary care, and physician job satisfaction may influence the quality of care differently for physical and mental health problems.
Design and Populations
Data for this analysis come from the Physician Referral Study.34,35 The physician population consisted of 832 primary care physicians (family practitioners, general internists, and general practitioners) in private practice at least 50% time in the Seattle metropolitan area in 1997. Of these, 261 physicians (31%) in 72 offices consented to participate. Office managers and participating physicians, as well as a random sample of 300 nonparticipating physicians, were asked to complete self-administered questionnaires at baseline.
In total, 17,187 English-speaking patients aged 18 and over were screened in the waiting rooms of the offices for 2 weeks. Of these, 691 patients were ineligible due to age below 18 or language, physical, or mental limitations, and 4,107 eligible patients refused to participate. Of the remaining 12,389 patients, 2,850 consenting patients had depressive symptoms (6 items from the Symptom Checklist for Depression) and/or at least 1 of 8 common, often persistent pain symptoms (back and neck pain, chest pain, abdominal pain, sinus or facial pain, headache or migraine, pain from indigestion/constipation, pain or arthritis in arms/legs/joints, and pelvic pain from female problems).36,37 Three patient cohorts were recruited: 1) patients with pain only (n=1,514; 53%); 2) patients with pain and depressive symptoms (n=761; 27%); and 3) patients with depressive symptoms only (n=575; 20%). Patients received mail or telephone surveys at 6 months to collect personal characteristics, measures of patient-rated quality of care, and health status.
Physician Job Satisfaction. Primary physician job satisfaction was measured with a 6-item scale adapted from Greenfield et al.38 and validated.39 Physicians rated their satisfaction on a 1 (very dissatisfied) to 5 (very satisfied) scale for: the care you provide to your patients (mean, 4.4); degree of personal autonomy you have (mean, 3.7); the way you are paid for your services (mean, 3.3); current volume of patients that you see (mean, 3.5); the way that your practice is managed (mean, 3.3); and your current work setting overall (mean, 3.7). A physician's job satisfaction was measured by averaging the 6 items in the scale (mean, 3.7; standard deviation, [SD], 0.73).
Quality of Care. Four quality-of-care measures were constructed. Patients rated the health care provided by their primary physicians at the 6-month follow-up on a 6-point scale of poor (1), fair, good, very good, excellent, and outstanding (6).40
The quality-of-care (QC) index was constructed from patients' reports to 4 questions, derived from Picker Institute survey instruments, at the 6-month follow-up (see Table 1).41 Factor analysis revealed a single factor with factor loadings between 0.66 and 0.87 and Cronbach's α of .73. The QC index was constructed by summing the 4 (0, 1) items, and the index ranged between 0 and 4. The QC index was correlated with all patients' rating of care (0.63).
Table 1. Descriptive Statistics for the Items in the Quality of Care Index at 6-Month Follow-up
|Were you involved in decisions about your health care as much as you wanted?|
|Involved as much as wanted||87||86||82|
|Involved too little or too much||13||13||18|
|Did your doctor explain what to do if problems or symptoms continued, got worse, or came back?|
|Doctor explained not at all, little, or somewhat||47||30||36|
|Doctor explained completely||53||70||64|
|Did you get as much information about your condition and treatment as you wanted from this doctor?|
| No, a little, or some information||51||36||45|
| Complete information||49||64||55|
|Did you spend as much time with your doctor as you wanted?|
From Picker Institute instruments, patient trust and confidence in the primary physician at the 6-month follow-up was measured on a 5-point scale of (1) none, a little, some, quite a lot, and total confidence (5). Patient trust was correlated with the patient rating of care (0.73) and the QC index (0.59) among all patients.
The fourth measure, continuity of primary physician, indicated whether the patient reported having the same primary care physician at baseline and 6-month follow-up.42 Patients who changed physicians had lower ratings of their primary care physicians, less trust, and lower QC index scores (P=.000).
Health Status. For patients with depressive symptoms, the severity of symptoms was measured at the waiting room screen and 6-month follow-up by the 20-item Symptom Checklist for Depression (SCL-20), where scores ≥1.70 indicate severe depressive symptoms.37,43,44 Disability was measured by the number of restricted activity days due to emotional problems in the past 4 weeks.45
For patients with pain, the severity of pain symptoms was measured at the waiting room screen and the 6-month follow-up by a 10-point scale indicating the bothersomeness of the pain in the past 4 weeks, from “not bothersome” (0) to “extremely bothersome” (10).46 Functional health status was measured by the 3-item pain interference scale, from “no interference” (0) to “unable to carry on activities” (10).47 Disability was measured by the number of days the patient was limited in usual activities due to physical health problems in the past 4 weeks.45
For each measure, health outcome was calculated as the change in health status between the waiting room screen and the 6-month follow-up, where bigger values indicated more improvement.
Managed Care Controls. Based on our conceptual model of managed care,48 we identified managed care controls in primary care offices and controls targeting primary physicians. Managed care controls in patients' health plans were excluded because they were not associated with our quality-of-care measures.22 Further, while primary physicians generally experienced a single set of office and practice controls, physicians typically saw patients with many health plans, and we lacked data for all plans of each physician.
Office managed care was measured through the following controls: utilization management (the office's referral preauthorization requirements), financial incentives (percentage of office revenue from capitation), and whether the office uses referral guidelines or clinical guidelines for specific conditions. Because the office variables were correlated strongly, we created an office managed care index using principal component analysis. A single factor explained 60% of the total variation of the 5 variables; factor loadings were positive and ranged between 0.62 and 0.87. Factor scores were transformed to create a 0 to 100 office managed care index, where higher scores indicated more managed offices.
To validate the index, we hypothesized that, on average, solo physician offices would be the least managed, primary group offices would be more managed, and multispecialty group offices would be the most managed. As expected, mean index scores increased from solo to multispecialty group offices (mean scores: solo, 9; primary group, 30; multispecialty group, 55; P<.000).
Physician managed care was measured by financial incentives (how the primary physician was paid, whether the physician received a productivity bonus, or whether the physician had a financial withhold for referrals) and the number of Agency for Healthcare Policy and Research (AHCPR) clinical guidelines read or used by the physician.49,50
Patient Characteristics. Patient measures included age, gender, race, living alone, employment status, education, and annual household income. The number of comorbidities at baseline was assessed using a checklist of 21 comorbid conditions based on the Medical Outcomes Study.51 We also measured the context of care: whether the primary physician at baseline was the patient's usual source of care, whether the baseline visit was the patient's first visit with the primary physician for the pain problem, and whether the patient had sought care for the pain problem in the 6 months before the baseline visit.
Primary Physician Characteristics. Physician characteristics included gender, years in practice, and whether the physician's race was white or not. Specialty and board certification were measured using the American Medical Association Physician Masterfile. Physicians rated their tolerance for uncertainty in patient care, indicating agreement or disagreement on a 1-to-4 scale with 2 statements from an instrument by Gerrity et al.52: 1) the uncertainty of patient care often troubles me; and 2) uncertainty in patient care makes me uneasy. Scores ranged from 2 to 8, where 8 indicates strong disagreement, or greater tolerance for uncertainty.
Medical Office and Physician Practice Characteristics. Office characteristics included office type (solo, primary group, or multispecialty group practice), the number of physicians in the office, and whether the office was owned privately. Physicians also rated how difficult or easy it was to refer a patient to a specialist on a 1-to-5 scale, where 1 indicates very difficult and 5 indicates very easy.
Physician workload was measured by patient visits per hour, administrative hours per week, and percentage of patients referred in a typical month. Patient mix was measured by the percentages of patients who were female, nonwhite race, aged 18 and under, aged 65 and above, and from middle- or upper-class households.
With patients as the unit of analysis, separate ordinary least squares and logistic regression models were estimated to determine the association between physician job satisfaction and each quality-of-care variable. Covariates for both conditions included the patient's age, gender, race, marital status, education, annual household income, employment status, and the number of comorbid conditions. Covariates for patients with pain also included the following baseline health characteristics: pain interference, pain bothersomeness, restricted activity days due to physical health, presence or absence of depressive symptoms, whether the primary care physician was the patient's usual source of care, whether the patient was seeing the physician for the first time about the pain problem, and whether the patient reported seeing a health professional for the pain problem in the 6 months prior to the waiting room screen. Additional covariates for patients with depressive symptoms included the following baseline health characteristics: SCL depression score, restricted activity days due to emotional health, presence or absence of pain, and whether the primary care physician was the patient's usual source of care. If no association was detected between job satisfaction and quality of care, physician characteristics were entered as control variables, and the regressions were reestimated.
An association between physician job satisfaction and patient-rated quality of care may be due to managed care. In this case, we reestimated the regressions, controlling for the managed care variables.
Separate ordinary least squares regression models were estimated to determine the association between physician job satisfaction and health outcomes. Covariates included the baseline score of the dependent variable, age, gender, race, marital status, education, annual household income, employment status, number of comorbidities, whether the patient had pain and depressive symptoms, whether the primary care physician was the patient's usual source of care, and whether the patient reported seeing a health professional for pain or depression in the 6 months prior to the waiting room screen.
Models were estimated with Stata statistical software (Stata, College Station, TX)53 using general estimated equations (GEE) to account for correlations among patients in the same medical offices.
About 95% of the participating physicians and 96% of office managers completed their respective questionnaires. About 33% of the physicians were general internists, 64% were family medicine practitioners, and 3% were general practitioners. About 82% of the nonparticipating physicians completed their questionnaires. Participating and nonparticipating physicians had similar job satisfaction scores, referral rates, board certification, specialty and racial mix, but participants had a higher percentage of group practice and female physicians who had fewer years in practice, fewer office hours per week, and fewer patients aged 65 and over than nonparticipating physicians (P<.05).
Follow-up surveys were collected for 2,004 insured patients (70% response rate; 1,062 with pain only, 518 patients with pain and depressive symptoms, and 424 patients with depressive symptoms only). Patients with follow-up data were older and had less pain interference with activities or fewer depressive symptoms than excluded patients without follow-ups. Depressed patients with follow-up data were less likely to have seen a psychiatrist in the past than patients without follow-ups.
Table 2 presents baseline patient characteristics. The average age of patients was 49 years. A majority of patients were female, white, married, had education beyond high school, had moderate family incomes, and were seeing their usual primary care physician at the waiting room screen. Patients averaged 2.6 comorbidities. For patients with pain, musculoskeletal pains were the most common. About 2% (n=36) of the pain cohort reported a cancer diagnosis in the past 3 years. About half the patients were seeing their primary care physician the first time for their pain symptom, and over half had seen a health professional for their pain problem in the past 6 months. For patients with depressive symptoms, about 30% had seen a mental health specialist in the past 6 months before the waiting room screen.
Table 2. Patient Characteristics at Baseline
| Mean age, y (SD)||51 (16.5)||45 (15.3)||48 (14.9)|
| Female, %||62||72||76|
| Nonwhite, %||10||13||15|
| Living alone, %||29||46||42|
| Employed, %||63||66||60|
| Mean years of education (SD)||15 (2.5)||14 (2.4)||14 (2.5)|
| Mean annual household income (SD)||$50,169 (28,008)||$39,092 (26,525)||$39,039 (27,334)|
| Mean number of comorbidities (SD)||2.3 (1.9)||2.5 (2.2)||3.1 (2.2)|
| Percent of patients whose physician at waiting room screen is patient's usual source of health care||79||83||84|
|Baseline pain symptoms (percent of patients with symptoms)|
| Joint, arm, or leg pain||36||—||29|
| Back pain||21||—||26|
| Sinus, ear, or facial pain||13||—||14|
| Abdominal pain||10||—||8|
| Chest pain||8||—||7|
| Headache and migraine||7||—||11|
| Pain from indigestion and||4||—||2|
| Pelvic pain||3||—||4|
| Mean pain interference (SD)*||4.3 (2.9)||—||6.0 (2.8)|
| Mean pain bothersomeness (SD)†||6.4 (2.8)||—||7.3 (2.5)|
| Mean restricted activity days due to physical health (SD)||3.7 (6.5)||—||9.3 (10.2)|
| Percent of patients seeing primary physician first time for pain problem at baseline visit||52||—||40|
| Percent of patients with visits to any health professional for pain problem in 6 months before baseline visit||52||—||73|
| Mean SCL depression score (SD)‡||—||1.8 (.7)||1.7 (.6)|
| Mean restricted activity days due to emotional health (SD)||—||6.0 (7.8)||6.1 (8.2)|
| Percent of patients with visits to a mental health specialist in past 6 months before baseline visit||—||32||30|
On average, patients rated their primary care physicians “very good” (mean, 4.1; SD, 1.31) and trusted their physicians (mean, 4.0; SD, 0.89) at the 6-month follow-up. The QC index averaged 2.83 (SD, 1.26). Compared to the pain cohorts, patients who only had depressive symptoms rated their physicians higher, trusted their physicians more, and had greater QC index scores (P=.000). About 80% of the patients had the same primary care physician throughout the 6-month follow-up period, and about 95% of the patients received primary care at the same medical office throughout the follow-up period.
Table 3 presents descriptive statistics of the office and physician managed care variables for the 3 patient cohorts.
Table 3. Descriptive Statistics of Managed Care Measures for Primary Care Offices and Physicians at Baseline
|Primary care offices||(n=997)||(n=380)||(n=480)|
| Mean office managed care index*||40||37||37|
| Percent of patients seeing primary|
|physicians with these characteristics|
| Payment by salary||66||62||64|
| Productivity bonus||52||57||59|
| Financial withhold for referral||29||29||38|
| Depression patients|
| Patient's primary physician has read|
| or uses AHCPR depression|
| guidelines (percent of patients)†||—||21||29|
| Pain patients|
| Patient's primary physician has read|
| or uses AHCPR back pain or|
| depression guidelines (mean number|
| of guidelines)†||.68||—||.65|
Physician Job Satisfaction and Quality of Care
Table 4 summarizes the relationships between physician job satisfaction at baseline and patient-rated quality of care at follow-up. For pain patients, greater physician job satisfaction was associated with greater patient trust (coefficient, 0.06; P=.034) and greater continuity of primary physician (odds ratio, 1.64; P=.000). Controlling for the managed care variables did not change these relationships. Physician job satisfaction was not associated with patient ratings of care from their primary physician and the QC index.
Table 4. Summary of Associations Between Physician Job Satisfaction and Patient-rated Quality of Primary Care
|Patient trust and confidence in primary physician||*||*|
|Quality of care index||NS||NS|
|Patient rating of care provided by his or her primary physician||NS||*|
|Continuity of primary physician at follow-up||*||NS|
For depression patients, greater physician job satisfaction was associated with higher patient ratings of care provided by their primary physician (coefficient, 0.14; P=.041). Controlling for the managed care variables did not change this relationship. Greater physician job satisfaction also was associated with greater patient trust (coefficient, 0.10; P=.024) only after the office managed care index and physician characteristics were added into the regression.
For depression patients, physician satisfaction also had a weak association with greater continuity of primary physician (odds ratio, 1.32; P=.054), but this association disappeared when controlling for the office managed care index or physician characteristics. Physician job satisfaction was not associated with the QC index.
Physician Job Satisfaction and Health Outcomes
Table 5 describes the health status of patients at the waiting room screen and the 6-month follow-up. On average, most patients improved. For pain and depression patients, physician job satisfaction was not associated with any of the change in health status measures.
Table 5. Health Status at Waiting Room Screen and 6-Month Follow-up: Unadjusted Descriptive Statistics
|Patients with pain|
| Pain interference†||4.84 (2.97)||2.11* (2.74)||2.73 (3.22)|
| Pain bothersomeness‡||6.64 (2.75)||2.98* (2.99)||3.67 (3.55)|
| Restricted activity days due to physical health||5.44 (8.20)||3.60* (7.57)||1.79 (8.11)|
|Patients with depressive symptoms|
| SCL Depression Score§||1.71 (.65)||.92* (.74)||.79 (.75)|
| Restricted activity days due to emotional health||6.04 (8.03)||2.73* (6.22)||3.24 (8.60)|
Our study has three major findings. First, we found that physician job satisfaction at baseline is related to some but not all of our measures of patient-rated quality of primary care at the 6-month follow-up. For patients with pain or depressive symptoms, greater physician job satisfaction is associated with greater patient trust and confidence in their primary physicians. If the association is causal, the finding suggests that patient trust can be increased by reducing physicians' job dissatisfaction. Our regression results imply that if physician job satisfaction increased from very dissatisfied to very satisfied, patient trust would increase, on average, by 0.24 for pain patients and 0.40 for depression patients on the 1-to-5 trust/confidence scale.
For all patients, we also found that physician job satisfaction was not associated with the quality-of-care index. This finding suggests that physicians' views about their work are not related to their interactions with patients. Information sharing, patient participation in decision making, and the amount of time with patients are similar for satisfied and dissatisfied physicians in our sample.
The other relationships between physician job satisfaction and quality of care were different for patients with pain from those with depressive symptoms. For patients with pain, physician job dissatisfaction was associated with discontinuity of primary physician in the follow-up period. This association emerged likely because pain patients who changed physicians had lower ratings of their primary care physicians, less trust, and lower QC index scores, which may have precipitated the changes. The findings are similar to those in Federman et al.,54 who found associations between patient dissatisfaction and discontinuity of primary physician. However, no relationship was found for patients with depressive symptoms.
Different patterns also were found for patient ratings of the care delivered by their primary physicians. For patients with depressive symptoms, greater physician job satisfaction was associated with better patient ratings of the care provided by their primary physicians. Our regression results imply that if physician job satisfaction increased from very dissatisfied to very satisfied, patient ratings of care would increase, on average, by 0.56 on the 1-to-6 patient-rating scale. However, no association existed for patients with pain.
The reasons for these opposite findings for pain versus depressive patients are unclear. We speculate that depression patients may rate their care based more on affective aspects of the patient-physician relationship,42 such as patient trust and physician job satisfaction, which may result in higher ratings of care provided by their physicians. In contrast, pain patients may rate the care from their primary physicians based on the amount of pain relief,22 which may not build similar affective ties between physician job satisfaction and ratings of care. The relationship between physician job satisfaction and continuity of primary physician was weaker for depression patients than for pain patients, perhaps due to the smaller sample size for depression patients.
The second major finding is that managed care controls do not account for observed relationships between physician job satisfaction and patient-rated quality of primary care. This finding is supported by primary physician perceptions that managed care has little impact on their ability to provide quality care.10
The study's prospective design and the ruling out of managed care as an alternative explanation increase our confidence that physician job satisfaction may be causing better patient ratings of their care and greater continuity of primary physician. However, other explanations for this relationship may exist. Physicians with greater job satisfaction may have greater competence in technical and interpersonal aspects of primary care, and patients may be able to detect better competence, resulting in higher ratings of care provided by their primary physicians.13,55,56
The third major finding is that primary physician satisfaction at baseline is not associated with health outcomes. Physician job satisfaction may not have a direct, causal connection with health outcomes. At best, physician satisfaction might indirectly improve health outcomes through its effects on quality of primary care, which has closer links to health outcomes.32,33,57
Limitations and Conclusions
Findings are limited to our samples of physicians and patients with pain and depressive symptoms in the Seattle area and may not be generalizable to other cities with different mixes of managed care and delivery systems. Primary physicians in small practices were less likely to participate, and our findings may not apply to those settings.
Another limitation is that we measured the quality of primary care based solely on patient perceptions. The relationship between physician satisfaction and quality of care may be different for quality measures based on physician perceptions, chart reviews, medical claims, or other sources.10,58 However, Meredith et al.42 report that patient ratings of the patient-provider relationship are correlated with technical measures of the quality of care for patients with depression.
In conclusion, for both patients with pain or depressive symptoms, greater physician job satisfaction at baseline was related to greater patient trust and confidence in their primary physicians at the 6-month follow-up. Otherwise, the patient cohorts had different associations: pain patients of more satisfied physicians also were less likely to change physicians between baseline and the 6-month follow-up, while depression patients of more satisfied physicians had higher ratings of the care provided by their physicians. Physician satisfaction at baseline was not associated with health outcomes.
Funding support was from Agency for Healthcare Research and Quality grant HS11712.