Community Mental Health Journal

, 45:209

Managed Care and Provider Satisfaction in Mental Health Settings


    • Mailman School of Public HealthColumbia University
  • Alan R. Ellis
    • Cecil G. Sheps Center for Health Services ResearchUniversity of North Carolina-Chapel Hill
  • Sharon Topping
    • University of Southern Mississippi
  • Joseph P. Morrissey
    • School of Public Health and Senior Research Fellow, Cecil G. Sheps Center for Health Services ResearchUniversity of North Carolina-Chapel Hill
Original Paper

DOI: 10.1007/s10597-008-9171-6

Cite this article as:
Isett, K.R., Ellis, A.R., Topping, S. et al. Community Ment Health J (2009) 45: 209. doi:10.1007/s10597-008-9171-6


We assess the satisfaction of mental health providers using four dimensions from the medical practice literature—degree of autonomy, relationship with patients, compensation, and administrative burden—and extend current work on professional satisfaction to include frontline service providers rather than only psychiatrists or other physicians. In contrast to results reported for primary care settings, we find that the impact of managed care on satisfaction is minimal for the mental health providers in our study of a Medicaid capitation demonstration in the southeastern US. Instead, variables relevant to everyday working conditions have an important effect on job satisfaction.


SatisfactionManaged care


Provider job satisfaction has been a concept of interest to health care researchers for the past two decades. From an organizational perspective, satisfied practitioners are more likely to remain in a given practice or position and to make the human resource training investments of the employing organization worthwhile. Moreover, a stable cadre of practitioners within an organization will positively affect the provider-patient relationship through greater continuity and quality of care (Murray et al. 2001; Warren et al. 1998). Continuity of care is particularly important for individuals with severe mental illness since these individuals tend to have difficulty maintaining a therapeutic relationship. This dynamic makes a positive provider-patient relationship an important precondition for engagement in on-going treatment.

Evidence from the primary care literature suggests that physicians working in managed care settings are becoming more and more dissatisfied with their professional life, resulting in physicians’ retiring early, moving out-of-state, joining large multi-specialty group practices, and leaving practice to become medical administrators (Brown et al. 2005; Burns 1996; Kletke et al. 2000; Landon et al. 2002). While the evidence of dissatisfaction under managed care in primary care has been well documented, research has not fully addressed whether the same negative effects are occurring for behavioral health providers. Generally, we know that psychiatrists, social workers, psychiatric nurses, and clinical psychologists are experiencing high levels of burnout, turnover, and general dissatisfaction with their jobs (Acker 1999; Farley 1994; Onyett and Pillinger 1997; Reid et al. 1999; Sturm 2001). In addition, mental health providers in hospitals are complaining about having little time to establish a therapeutic relationship with patients, while those in the community are frustrated with the lack of resources to treat persons with severe and long-term psychiatric problems (Bingham et al. 2002; Blankertz and Robinson 1997). What is unclear from these studies is whether managed behavioral health care intensifies the negative job satisfaction trends reported in the earlier behavioral health literature and in managed primary care settings. And as is the case in general healthcare, managed behavioral health care has many different models that affect incentives and ability to provide services, which complicates the assessment of satisfaction (Kongstvedt 1996).

The entry of managed care into behavioral health lagged behind its entry into primary care by about a decade (Anderson et al. 1996). Part of this lag was associated with the difficulty of developing a capitation rate for a population of individuals with hard-to-predict needs (Frank et al. 1995). However, due to high costs of treating persons with serious mental illnesses (SMI), states and private insurers alike have turned increasingly to managed care as a solution (Frank et al. 1995; Morris and Bloom 2002), and behavioral health became a major growth area for managed care through the 1990s (Koike et al. 2000).

The research question we seek to address here is whether managed care adversely affects provider job satisfaction in mental health settings. This study fills two gaps in the existing literature on provider satisfaction. First, it provides empirical results reporting the linkage between provider satisfaction and managed care in mental health settings. Second, in contrast to the primary care satisfaction studies that look only at physicians and other doctoral level staff, it expands the range of providers surveyed to include frontline care givers such as nurses, social workers, and case managers.


Using the primary care literature on physician satisfaction as our guide, we focus on four established dimensions of job satisfaction: degree of autonomy, relationship with patients, compensation, and administrative burden (Konrad et al. 1999; Lichtenstein 1984; Williams et al. 1999a). Each dimension is reviewed briefly below.


According to Konrad and colleagues (1999), the autonomy dimension of provider satisfaction measures contentment with one’s independence of action, including elements such as level of input into important decisions in the workplace and freedom to use one’s own best clinical or professional judgment in treating patients. Both physicians and psychiatric social workers have ranked professional autonomy as the most critical element of job satisfaction in prior studies (Landon et al. 2003; Marriott et al. 1994; Schulz et al. 1997).

Current evidence about the effect of managed care on satisfaction with provider autonomy is mixed. On the one hand, provider satisfaction with autonomy has been shown to decrease with higher penetration of managed care in a given location (Baker and Cantor 1993; Collins et al. 1997; Hadley and Mitchell 2002; Murray et al. 2001; Warren et al. 1999), perhaps as a result of providers becoming more accountable to external controls (Mechanic 2003; Warren et al. 1999). Similarly, physicians in states with higher HMO penetration are less likely to say they have freedom in caring for patients (Burdi and Baker 1997; Hadley et al. 1999; Stoddard et al. 2001), and inpatient mental health professionals are more likely to report a lack of authority and responsibility than are those in community-based settings (Marriott et al. 1994; Reid et al. 1999).

On the other hand, there are several factors that might mitigate the negative aspects of managed care in behavioral health systems. These factors include practice models that offer more structured work environments, thereby providing a buffer from external regulations (Stamps 1995; Williams et al. 2001), shifts from fee-for-service to capitated models of Medicaid payment (Morris and Bloom 2002), scope of practice (e.g., generalists appear more satisfied with professional autonomy than specialists, especially psychiatrists) (Sturm and Ringel 2003), and physician age (Burdi and Baker 1997; Burns 1996; Keating et al. 2004; Landon et al. 2002; Warren et al. 1999).

Patient Relationships

The patient relationship dimension refers to satisfaction with the quality and duration of patient relationships, including being able to spend sufficient time with patients during an office visit (Baker and Cantor 1993; Bates et al. 1998; Konrad et al. 1999; Sturm and Ringel 2003; Warren et al. 1999). Research findings indicate that participation in managed care has a negative impact on physician satisfaction with time spent with patients (Baker and Cantor 1993; Collins et al. 1997; Donelan et al. 1997). Warren and colleagues (1999) and Keating and colleagues (2004) found that the greater the participation in managed care, the stronger the belief that patient load had increased, especially for younger physicians and those paid by capitation.1 However, other studies have found that regardless of practice model, all providers reported dissatisfaction with the time they had available to spend with patients during scheduled appointments (Collins et al. 1997; Keating et al. 2004; Murray et al. 2001; Farley 1994; Reid et al. 1999).

Other research studies suggest that social workers serving clients with severe mental illnesses may have higher levels of emotional exhaustion and dissatisfaction with their jobs than those mental health providers with less severe case mixes (Acker 1999; Blankertz and Robinson 1997; Reid et al. 1999). Findings also indicate that satisfaction with patient relationships is higher for providers in smaller hospital units where caseloads (patient/provider ratio) may be smaller, enabling the development of a deeper relationship with patients (Bingham et al. 2002; Farley 1994). However, patients in these units often had less severe disabilities and higher functioning than those in larger units, so social contact and interaction with clinical staff was enhanced. Other provider characteristics, such as education and position tenure, did not affect the patient relationship dimension of satisfaction (Bingham et al. 2002).

Satisfaction with Compensation

Income is a major predictor of work satisfaction (Cashman et al. 1990; Martin and Schinke 1998; Murray et al. 2001; Pathman et al. 2002; Schulz et al. 1997; Stoddard et al. 2001). This dimension is usually defined as “satisfaction with total compensation: direct pay, financial or non-financial fringe benefits, and future prospects for financial security” (Konrad et al. 1999, p. 1176). As with other dimensions of satisfaction, evidence of the effect of managed care on satisfaction with compensation is mixed.

A few studies have found that managed care exposure is associated with lower satisfaction with income (Baker and Cantor, 1993; Deckard 1995; Grembowski et al. 2003; Warren et al. 1999) and that dissatisfaction increased as HMO penetration increased (Hadley and Mitchell 1997, 2002). Other evidence suggests that providers in open model HMOs (i.e., models in which the physician remains independent and maintains his/her own office and accepts patients from multiple health plans and insurers) were less satisfied with their pay, whereas those in closed models (i.e., models in which physicians must be employed by the HMO [staff model] or by a group that contracts with the HMO and has an exclusive arrangement with an individual health plan) were becoming more satisfied with income and with their practices over time (Burdi and Baker 1997; Murray et al. 2001; Schulz et al. 1997, 1992).

Satisfaction with compensation may also be related to the ability to select compensation and incentive packages that fit one’s practice style (Zierler et al. 1998). Additionally, changes to compensation packages (e.g., changes to benefit structures, incentives, and retirement packages, but not necessarily direct pay) were associated with immediate reductions in satisfaction, while stable compensation practices were related to higher satisfaction (Freeborn and Hooker 1995; Zierler et al. 1998). Given these findings, it may be that dissatisfaction with income is associated with uncertainty and lack of control for providers. Dissatisfaction with compensation has also been identified as an issue for nurse practitioners and physician assistants. Evidence indicates that these groups are more dissatisfied with compensation when compared to primary care physicians in HMOs (Freeborn et al. 2002; Williams et al. 2001).

Administrative Burden

Administrative burden refers to the amount of documentation, especially paperwork, that an individual provider must complete. Providers are dissatisfied with the increased administrative burden that they attribute to managed care (Collins et al. 1997; Koike et al. 2000; McMurray et al. 1997; Murray et al. 2001; Stubbe and Thomas 2002; Warren et al. 1998, 1999). Increased burden was ascribed to the many different contracts, formularies, and guidelines from different managed care plans (Chehab et al. 2001), and the movement of patients in and out practices because of changes to insurance coverage, physician panels, and specialists (Donelan et al. 1997). Several studies have shown that healthcare workers feel that administrative requirements associated with managed care, such as increased documentation and reviews, are problematic and even infringe upon successful case management (Donelan et al. 1997; Farley 1994; Hromco et al. 1995).

In summarizing this literature, it appears that all dimensions of satisfaction are affected by managed care (Table 1). Provider characteristics have had less of an emphasis in the provider satisfaction literature than workplace context variables. Of the provider characteristics studied, time in practice and position title or occupation tended to have the most important influence on satisfaction (Keating et al. 2004; Landon et al. 2002, 2003; Warren et al. 1999, 1998; Williams et al. 2001). However, the influence of managed care on job satisfaction may change over time. For instance, managed care in its initial stages tended to decrease satisfaction, but as it evolved, buffers were developed (e.g., group practice models and large practices) that decreased the uncertainties associated with patient care, reimbursement, and compensation and mitigated dissatisfaction among providers. In contrast, severity of illness has continued to be a problem affecting all dimensions of provider satisfaction, especially with regard to inpatient “revolving door” situations (e.g., Farley 1994; Reid et al. 1999).
Table 1

Selected references: conditions influencing dimensions of satisfaction

Dimensions of satisfaction:


Patient relationship


Administrative burden

Workplace Characteristics:

Influence of Managed Care (MC)

Degree MC Affiliation & MC Penetration (Burdi and Baker 1997; Hadley et al. 1999; Hadley and Mitchell 2002; Murray et al. 2001; Stoddard et al. 2001; Warren et al. 1998, 1999)

Practice Model (Bates et al. 1998; Chehab et al. 2001; Deckard 1995; Grembowski et al. 2003; Grumbach et al. 1998; Keating et al. 2004; Stamps 1995; Williams et al. 2001)

Reimbursement Method (Bates et al. 1998; Keating et al. 2004; Landon et al. 2002, 2003; Morris and Bloom. 2002; Schulz et al. 1997; Williams et al. 1999b

Degree MC Affiliation & MC Penetration (Baker and Cantor 1993; Collins et al. 1997; Donelan et al. 1997; Farley 1994; Grumbach et al. 1998; Hadley and Mitchell 1997 Murray et al. 2001)

Practice Model (Deckard 1995; Grumbach et al. 1998)

Reimbursement Method (Sturm and Ringel 2003; Warren et al. 1999)

Degree MC Affiliation & MC Penetration (Baker and Cantor 1993; Deckard 1995; Hadley and Mitchell 1997, 2002; Hadley et al. 1999; Stoddard et al. 2001; Stubbe and Thomas 2002; Warren et al. 1999)

Practice Model (Burdi and Baker 1997; Chehab et al. 2001; Freeborn et al. 2002; Freeborn and Hooker 1995; Grembowski et al. 2003; Murray et al. 2001; Stamps 1995)

Reimbursement Method (Landon et al. 2002; Morris and Bloom 2002; Zierler et al. 1998)

Degree MC Affiliation & MC Penetration (Collins et al. 1997; Koike et al. 2000; McMurray et al. 1997; Murray et al. 2001; Warren et al. 1999)

Practice Model (Chehab et al. 2001; Deckard 1995; Stamps 1995)

Caseload: Size & Type of Patient

% MC Patients (Collins et al. 1997; Warren et al. 1999)

Patient Severity (Acker 1999; Blankertz and Robinson 1997; Onyett and Pillinger 1997; Reid et al. 1999) Patient Load (Bingham et al. 2002; Farley 1994 Warren et al. 1999, 1998)

Patient Severity (Acker 1999)

Patient Severity (Blankertz and Robinson 1997)

Unit/Organizational Type

Inpatient/Outpatient (Marriott et al. 1994; Reid et al. 1999) Structured/Unstructured (Stamps 1995; Williams et al. 2001)

Inpatient/Outpatient (Bingham et al. 2002; Donelan et al. 1997; Farley 1994; Reid et al. 1999) Structured/Unstructured (Williams et al. 2001)

Structured/Unstructured (Stamps 1995; Williams et al. 2001)

Inpatient/Outpatient (Farley 1994; Hromco et al. 1995; Reid et al. 1999)

Provider Characteristics:



Graduate Degrees (Hromco et al. 1995)

Professional tenure

Time in Practice (Kletke et al. 2000; Landon et al. 2002; Warren et al. 1999)

Time in Practice (Farley 1994; Keating et al. 2004; Warren et al. 1998)

Time in Practice (Acker 1999)


Position tenure

Tenure with Employers (Burns 1996)


Position title/Occupation

Generalist/Specialists (Burdi and Baker 1997; Landon et al. 2002, 2003; Schulz et al. 1992; Sturm 2001)

Generalist/Specialists (Sturm 2001) PA/NPs & PCPs (Freeborn et al. 2002)

Social Workers (Farley 1994)

Generalist/Specialists (Burdi and Baker 1997; Landon et al. 2003; Schulz et al. 1992; Sturm 2001) PA/NPs & PCPs (Freeborn et al. 2002)

Case Managers & Social Workers (Hromco et al. 1995; Onyett and Pillinger 1997)

Data and Methods

Study Sites

The Tidewater Managed Care Study (TMCS) was part of a SAMHSA managed care initiative that examined various models and arrangements of managed mental health care in the public sector (Leff et al. 2005). TMCS compared two different organizational and financing arrangements for services to adults with SMI who were enrolled in Medicaid in two regions of eastern Virginia (Fried et al. 2000; Isett et al. 2006; Morrissey et al. 2002; Stroup et al. 2001). In 1996 under a Sect. 1915b waiver, Virginia implemented a mandatory HMO enrollment program known as Medallion II in the Tidewater region of the state. Four HMOs participated in the Medallion II program at its initiation. Each of the HMOs received a capitation payment based upon a contracted per-member-per-month fee. However, the HMOs were free to make their provider contracts under different arrangements, and sub-contract arrangements varied among providers from sub-capitation to discounted fee contracts. Table 2 provides an overview of these arrangements and compares the managed care site to the fee-for-service site. Because we have only two study sites, we were unable to parse out the effects of the two different managed care subcontract arrangements, representing a limitation of our dataset and analyses.
Table 2

Comparison of arrangements in study sitesa



Managed care


-Utilization Management (UM) contract with private vendor of medical-psychiatric benefit

-CSBs case rate for social-rehab benefit (including case management)

-Primary care gatekeeper, but none for mental health

-Four HMOs with public and private providers

-1915b waiver authority

-Mandatory HMO enrollment

-CSBs case rate for social-rehab benefit (including case management)


-Fee-for-service based on Medicaid fee schedule

-UM paid as an administrative services organization with no performance incentives

-HMOs paid on a capitated basis by Medicaid agency

-1 HMO (70% of enrollment) subcapitates a subsidiary managed behavioral health care organization (MBHCO)

-3 HMOs pay a discounted fee-for-service


-No risk is shared, retained by Medicaid agency

-Each HMO is at full risk

-1 HMO transfers full risk to MBHCO and providers are subcapitated by MBHCO and are at risk

-3 HMOs do not pass on risk to providers

Utilization management

-UM contractor responsible for full range of UM functions -UM contractor has no responsibility for alternative care arrangements

-HMOs/MBHCO responsible for full range of UM functions

-HMOs/MBHCO responsible for alternative care arrangements

aTable adapted from Fried et al. (2000)

The Tidewater region includes the southeastern Virginia cities of Norfolk, Chesapeake, Newport News, Hampton, and Virginia Beach. Other parts of the state, including the greater Richmond area, continued to operate under the Medallion I program, a fee-for-service primary care system implemented in 1992. In this study, the Richmond region serves as the comparison site to Tidewater. The sites were fairly evenly matched in terms of demographics with the exception of population size. The Tidewater region had a larger population (1.5 million) than the Richmond area (943,000), but the two populations were similar with respect to percentage that were black (29.7% and 29.9%, respectively) and percentage below poverty level (14.1% and 12.1%). (For a detailed description of the TMCS and the specific sites, see Fried et al. 2000).

Survey Instrument

A survey instrument was developed to obtain the views of frontline caregivers and physicians on satisfaction with work and client relationships. The instrument was pre-tested at another site in Virginia with both community service providers and hospital staff. Analysis from the pre-testing showed that the instrument was reliable and valid for the target population of this study. Job satisfaction was measured by a 15-question scale, with four subscales corresponding to the four dimensions of satisfaction discussed earlier. The response format was a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree) with a neutral midpoint. The items addressed specific constructs related to each dimension of satisfaction explored in the study, such as “my total compensation package is fair”. An important consideration is that the scales for this study were constructed originally to measure dissatisfaction. Thus, many of our individual items were worded negatively. For ease of interpretation, the items were reverse coded where appropriate so that a higher number reflected higher satisfaction.

Sampling and Data Collection

The survey was distributed to 274 individuals who met eligibility requirements for the study. Eligible individuals included direct care staff such as psychiatrists, psychologists, social workers, therapists, triage specialists, case managers, and program administrators who worked with individuals with severe mental illnesses in one of the two locations (Tidewater or Richmond). The survey was administered at four inpatient facilities (two in each site) known to have a large number of inpatient Medicaid psychiatric admissions annually, and at all nine publicly funded outpatient facilities (five in Tidewater and four in Richmond).

An internal organizational representative distributed surveys to the eligible individuals in each organization. The survey was self-administered and then mailed back to the researchers. While the distribution of our survey instrument through an organizational representative was efficient, it may provide two sources of bias for respondents. First, the internal representative could have selectively distributed the surveys, introducing selection bias. We have no check for this type of bias in our data, but we asked that the survey be distributed to all eligible individuals and have no reason to believe the survey was distributed otherwise. Second, the internal handling of the survey could lead to individuals providing desirable, rather than honest, answers to our questions. However, we believe that this potential bias was mitigated by the fact that the surveys were mailed directly back to the researchers.

Data collection lasted for 2 months and the final response rate achieved was 72% (198 of 274). There was moderate variation in response rate by site: fee-for service outpatient 75% (74 out of 99), fee-for-service inpatient 86% (25 out of 29), managed care outpatient 70% (85 out of 122), and managed care inpatient 58% (14 out of 24). The variation also was evident by job title: social workers, therapists, and nurses 83% (83 out of 100) and case managers, psychiatrists, and administrators 65% (88 out of 135) (see Fried et al. 2000 for a detailed breakdown of response rate variation by job title and site).

Statistical Methods

A principal components analysis was run on the survey data to determine the cohesiveness of the concepts of provider satisfaction explored in the survey as well as to produce scales reflecting the four dimensions of interest in this analysis. Because the items pertaining to autonomy and patient care were more relevant for direct care staff, the sample for the principal components analysis was limited to direct care staff (n = 106). Five questions from the survey were eliminated from the analysis because of their lack of consistency with the factors that were found. Ultimately, four factors were extracted from the data corresponding to the concepts and subscales identified by Konrad and colleagues (Konrad et al. 1999; Lichtenstein 1984; Williams et al. 1999a). Each of the factors had acceptable internal reliability with Cronbach’s alpha coefficients of .60 or above (see Table 3).
Table 3

Summary of principal components analysis for direct care staff (n = 106)

Promax Oblique Rotation—Rotated Factor Pattern (Standardized Regression Coefficients)

Item #


Patient care

Administrative burden








Clinical guidelines restrict my freedom to practice






Gatekeeping requirements often conflict with my clinical judgment






Formularies or prescription limits restrict the quality of care I provide






I am overwhelmed by the needs of my clients






Time pressures keep me from developing good client relationships






I often feel like what I do for my clients is just a drop in the bucket






Paperwork required by payers is a burden on me






I have too much administrative work to do






I am not well compensated given my training and experience






My total compensation package is faira







Inter-Factor Correlations



Patient care

Administrative burden









Patient care





Administrative burden






aitem not reverse coded; all other items were reverse coded

Bolded entries indicate the factors that were used in the final analysis

The four scales developed in the principal components analysis were then used as dependent variables in a series of regression models. The independent variables in the models (Table 4) were related to workplace characteristics and provider characteristics found in the literature. Workplace characteristics included caseload size (small and large, coded as a dummy variable with “large” as “1”), treatment setting (inpatient versus outpatient, coded as a dummy variable with “inpatient” as “1”), and study site (managed care versus fee-for-service, coded as a dummy variable with “managed care” as “1”). Provider characteristics included position title (psychiatrist, nurse, or case manager, coded as a series of dummy variables, with “other clinicians” as the suppressed category, representing staff such as social workers, therapists, and triage specialists), professional tenure/experience (years of work experience, coded as a dummy variable with “≥10 years” as “1”), and position tenure/longevity (years in current position, coded as a dummy variable, with “≥5 years” as “1”). Caseload, professional tenure, and position tenure were coded as dummy variables rather than as continuous variables because linear relationships with provider satisfaction were not expected. Table 5 presents basic demographic information by respondent type for all staff with complete data on the satisfaction items and predictor variables (n = 150). Level of education is included in the table but was not included in the regression models because it is strongly related to job title and caseload.
Table 4

Variables used in multivariate regression models



Job title



Case manager

Other cliniciana

Size of caseload

Large (≥25)

Small (1–24)a

Temporal dimension

Experience in profession ≥10 years (1 = yes, 0 = no)

Longevity at current organization ≥5 years (1 = yes, 0 = no)

Treatment setting



Study site

Managed care


aReference category. Other categories coded using dummy variables with 1 = yes, 0 = no

Table 5

Respondent demographic information by job type


Non-direct care staffa



Case manager

Other clinicians

Caseload size







Small (<25)






Large (25+)






Level of education













Bachelors or less






Temporal dimension

Experience (years)






Longevity (years)






Treatment setting













Study site

Managed care


















aTwo non-direct-care staff in the analysis sample did not report level of education

Some important issues emerge from the demographic data reported in Table 5. Fifty-seven percent (20/35) of case managers and about half of nurses and psychiatrists (6/12 and 5/9 respectively) have large caseloads, while “other clinicians” tend to have small caseloads. Case managers, other clinicians, and nurses are mostly in outpatient settings (100, 96, and 92%, respectively), while psychiatrists and nondirect care staff are more evenly distributed between inpatient and outpatient settings. Only 3 of the 9 psychiatrists (33%) are in the managed care site, however.

After model specification, full models were run on each of the four dependent variables and a process of backward elimination was completed on each model. Because of the strong association of job title with both treatment setting and study site, it was especially important to examine collinearity diagnostics. Based on collinearity indicators (e.g., maximum generalized variance inflation factor of 2.15), we concluded that there was limited reason to be concerned about multicollinearity in this analysis.

Our literature review suggests that managed care may have differential effects in different settings and perhaps for different classes of employees. To fully test for these dynamics, our model would have to include interaction effects for some key variables, such as job title and treatment setting. However, our data do not allow meaningful tests of these interactions due to the sample size (especially the small numbers of psychiatrists and nurses) and to the uneven distribution of providers across study sites and treatment settings.


On a scale from 1 (low) to 5 (high), the raw means (SD) for the four dimensions of satisfaction—autonomy, patient relationships, compensation, and administrative burden—were 3.0 (0.7), 3.0 (0.8), 2.6 (1.0), and 2.4 (0.9) respectively, all indicating neutrality or dissatisfaction. Table 6 describes the bivariate relationships among our continuous predictors and dependent variables. The dependent variables have small correlations with each other. The correlation between work experience and longevity is noteworthy but does not in itself give cause for concern about multicollinearity.
Table 6

Bivariate correlations of nondichotomous measures among direct care staff (n = 106)



Pt. Care


Admin burden












Pt. care













Admin burden












The final model for the autonomy variable was reduced to two independent variables: caseload size (large) and treatment setting (inpatient). The results for this model (Table 7) suggest that having a large caseload decreases satisfaction on this dimension (P < .05). There also was a tendency for those who worked in inpatient settings to be less satisfied with their job autonomy than those who worked in outpatient settings (P < .01).
Table 7

Final regression models for satisfaction measuresa direct care staff (n = 106)


Parameter estimate (standard error)


Patient care


Administrative burden

Job title







Case manager






Large caseload (25+)










Employment characteristics

10+ years in practice


5+ years in current position






Site characteristics

Inpatient setting





Managed care site











aOnly variables significant at the P < .1 level or lower are shown: * P < .1, ** P < .05, *** P < .01. In the model for satisfaction with compensation, the P-value for large caseload is .1019

Satisfaction with patient relationships had one significant predictor in common with the autonomy scale: staff with large caseloads tended to have lower satisfaction (P < .1). Also, those with at least 5 years of position tenure reported higher satisfaction (P < .1). There were no significant effects of job title, site, or setting.

Satisfaction with compensation was the only type of satisfaction associated with job title. Nurses (P < .05) and case managers (P < .01) reported lower satisfaction than did “other clinicians”, while psychiatrists did not differ from other clinicians. Also, consistent with the first two models, staff with large caseloads tended to report lower satisfaction (P < .1).

The final dimension of job satisfaction—administrative burden—yielded the only significant difference for the managed care variable. Those who worked in managed care settings reported less satisfaction with their administrative burden than did those who worked in fee-for-service settings (P < .01). Here, there was no effect of caseload, inpatient setting, or personal characteristics.

We were interested in determining whether non-direct-care staff differed from direct care staff with regard to satisfaction with autonomy and administrative burden. A principal components analysis that included all staff yielded the same four factors as did the analysis for direct care staff, with results very similar to those displayed in Table 3. However, in regression models that tested for a difference between direct care and non-direct-care staff while controlling for the other predictors (caseload, longevity, inpatient setting, and managed care site), the effect of interest was not significant.


The main question addressed in this study was whether the introduction of managed care into a publicly funded mental health system had a negative impact on provider satisfaction. Interestingly, and contrary to empirical findings in the general healthcare literature, managed behavioral health care was not uniformly associated with lowered job satisfaction compared to fee-for-service settings. Administrative burden was the only dimension with which managed care had a significant association; managed care did not influence ratings of satisfaction with worker autonomy, the quality of patient care relationships, or compensation for behavioral health care employees.

One possible explanation for these results is the recognition that many managed care practices such as prior authorization, strict medical necessity criteria, length of stay reviews, and staffing patterns had gradually diffused throughout the public mental health system prior to the adoption of formal managed care contracts by many state Medicaid programs (Essock and Goldman 1995). The net result of this trend was to lessen the administrative differences between managed care and fee-for-service reimbursement systems. To the extent that these practices had already been adopted in public service delivery settings, it is not surprising that workers in both fee-for-service and managed care systems would rate many aspects of their job satisfaction in similar ways. Even so, the findings reported here on administrative burden suggest that there may well be a net increase in the perception of the level of paperwork and other documentation requirements (such as prior authorizations and utilization reviews) under managed care over and above what occurs under fee-for-service reimbursement. This burden increase is consistent with findings reported in numerous other studies both in mental health (cf. Stubbe and Thomas 2002) and general healthcare (cf. Chehab et al. 2001; Murray et al. 2001).

The factors that did make a difference in job satisfaction in this study relate to more traditional occupational and work setting distinctions. The most consistent findings across the four satisfaction dimensions occurred for the caseload variable, with individuals with larger caseloads experiencing lower satisfaction on three of the four dimensions (autonomy, patient care, and compensation) than those with smaller caseloads. With that said, it is unclear whether or not the dissatisfaction related to caseload is actually directly related to the individuals and job titles they hold, or if it is actually related to some other aspect of their job responsibilities not captured in our model—particularly in light of nurse and case manager dissatisfaction with their compensation. This relationship deserves greater scrutiny and further investigation in future studies of provider satisfaction.

We speculate that in both managed and fee-for-service mental health service settings, professionals with low job satisfaction might have limited interest in supporting initiatives such as the adoption of evidence-based practices (Drake et al. 2001) or the transformation of mental health services (Davidson et al. 2006; Day 2006). Clinicians who occupy key clinical and leadership roles in mental health treatment organizations are critical to generating enthusiasm for change efforts (Kelman 2005; Rogers 2003). But if these individuals lack commitment and satisfaction in their jobs, their motivation to improve the workplace might be absent (Wright and Davis 2003; Kim 2002). So low job satisfaction of the scope reported here might signal a fundamental demoralization and apathy toward any constructive change. On the other hand, the low satisfaction of mental health professionals with their jobs and the current service delivery system could foster readiness to embrace new practices or initiate improvements (Kelman 2005). It is likely that both of these dynamics exist, but it is unclear which has more salience or intensity in behavioral health care settings.

A particularly troubling finding in the paper is the strong effect of inpatient setting on satisfaction with autonomy. Controlling for other variables in the model, being located in an inpatient setting drops satisfaction with autonomy by two-thirds of a scale point on average, compared to outpatient settings. This might be explained by greater external controls on utilization in inpatient settings than in outpatient settings. However, the literature review suggests that we might see a similar effect of inpatient setting on the patient care dimension of satisfaction, given the “revolving door” phenomenon. We do not, and therefore we struggle to understand the meaning of this relationship in terms of the existing literature. Our results suggest that more work needs to be done on understanding these two related, but distinct, dimensions of satisfaction.

One of the goals of this paper was to extend the provider satisfaction literature to non-psychiatrist staff to say something about front-line caregivers’ satisfaction. We did not find much of a relationship between job title and the four dimensions of satisfaction measured here, except that nurses and caseworkers were less satisfied with compensation than were “other clinicians”. This result is similar to findings from studies of physician assistants and nurse practitioners in primary care, who reported experiencing stress on a daily basis and were dissatisfied with amount of time with patients and compensation (e.g., Freeborn et al. 2002). The lack of a difference in satisfaction between “other clinicians” (mostly masters-level therapists) and psychiatric staff is puzzling. On the one hand, the substitution of auxiliary and para-professional staff for psychiatric staff under many managed behavioral health care arrangements has elevated the professional status of many of these employees (Cohen 2003). This suggests that this group might be expected to experience improved job satisfaction. On the other hand, increased status has come with increased treatment responsibility for these clinicians—often with constrained resources. This suggests that “other clinician” attitudes may now mimic the negative attitudes of psychiatrists because they now face the same treatment constraints. Either way, additional research is needed on the determinants of job satisfaction among this group and the consequences for mental health organizations and their clients.

Since a happy workforce is a stable workforce, and workers who are both satisfied and stable contribute to client outcomes through quality of care, these are non-trivial findings (Morris and Bloom 2002; Trotter 1996). The results of the study must be taken with caution, however. The models run in this research had low coefficients of determination (i.e., R2 values ranged from .14 to .23). This indicates that there are unmeasured variables that contribute to these ratings.


While managed care might lead to greater dissatisfaction with administrative burden among mental health professionals, it was not associated with reduced job satisfaction on other key dimensions such as autonomy, patient relationships, and compensation in this study. Current efforts to improve the quality and performance of the public mental health system cannot ignore workforce dissatisfaction, but may move forward with the knowledge that managed care is not necessarily a roadblock to change efforts as it relates to increased workforce dissatisfaction. Both of the interpretations offered above for the impact of low job satisfaction on change initiatives suggest that the antecedents and consequences of low morale and low job satisfaction among mental health clinicians in mental health agencies must be addressed in current systems change efforts throughout the public mental health system much more centrally than they have been to date. Over and above the impact of job satisfaction attitudes on change efforts, low satisfaction with large caseloads and compensation for some key clinical groups—important proximal variables in the daily lives of the mental health care workforce (Isett et al. 2006; Strauss 1981)—could have an impact on retention and growth of the current workforce.


While most of these studies focus on the provider’s perception of time with patients, one study examined the change in actual length of office visits from 1989 through 1998 (Mechanic et al. 2001). Contrary to popular opinion, the growth of managed care was not associated with a reduction in the length of office visits. The duration of visits actually increased for both prepaid and non-prepaid visits during this time period. Non-prepaid visits were consistently longer than prepaid visits, but even this gap declined from 1 min in 1989 to 0.6 min in 1998.


Copyright information

© Springer Science+Business Media, LLC 2008