Association of Utilization Management and Treatment Plan Modifications Among Practicing U.S. Psychiatrists

This study used data from the American Psychiatric Institute for Research and Education Practice Research Network’s study of psychiatric patients and treatments, which in 1999 administered a mailed survey to a rotating cohort of 784 APA member psychiatrists. Demographic and practice data were collected from 615 psychiatrist members of APA who spent at least 15 hours per week providing direct face-to-face patient care (response rate=78.4%). The sample consisted of 291 responding psychiatrists selected at random from the APA membership list and 324 self-identified volunteers who responded to announcements placed in a national publication mailed to APA members. Sample weights were developed to generate nationally representative estimates and to account for the sampling design, nonresponse, and random/volunteer status.

Participating psychiatrists completed a patient log reporting general information on the first 12 consecutive unique patients seen after an assigned start time. Detailed information was collected on every fourth patient in the log (up to three patients total). The survey collected data on demographics, diagnostic information, clinical information, treatment settings, medications, treatment characteristics, financial arrangements, and utilization management and quality assurance for 1,843 patients. The survey methods and weights are described in detail by West et al. (12) and Pincus et al. (13).

The patient sample (N=1,843) was almost equally split by gender, predominantly white, and either currently or formerly married. Overall, most patients were being treated in an outpatient setting. Patients for whom utilization management techniques were in effect were less educated, had relatively more inpatient care, and received lower Global Assessment of Functioning scale scores (axis V) than patients whose care was not subject to utilization management (Table 1).

Frequency distributions were calculated to determine the proportion of patients whose care was subject to any utilization management technique. Nine hundred forty-seven (52.6%) of the patients had utilization management applied to their care (Table 1), of whom 309 (31.5%) had associated changes in treatment provisions reported by their psychiatrists because of financial considerations. Among those with no utilization management, only 15.5% had associated changes in treatment.

The overall odds ratio of a change in treatment when any utilization management technique was present was 2.6 (Table 3). The odds of reducing the frequency and/or number of visits and the odds of providing different treatment when any utilization management technique was present were 2.3 and 3.7, respectively.

Table 3 also presents the odds ratios for changes in treatment in the presence of different utilization management techniques. The presence of utilization review or treatment algorithms was not associated with psychiatrists’ reported changes in treatment. In contrast, the three remaining utilization management techniques had significant associations with treatment modifications. For example, the presence of formulary restrictions was associated with a threefold increase in the odds of a physician providing different treatment. Although only 1% of the patient group had treatment recommendations restricted by financial incentives or penalties to limit referrals, when present, these techniques were associated with an increased odds of reducing the frequency or number of visits the psychiatrist would have preferred by a factor of almost four. When treatment recommendations were restricted by limiting referrals to selected hospitals or specialist panels, there was a significant increase in the odds of the psychiatrist providing different treatment than he or she otherwise would have provided.

Two psychiatrist characteristics were associated with the likelihood of modifying treatment (Table 2). Psychiatrists who were compensated by a nonsalaried income source were 1.61 times more likely to modify their treatment plans when caring for patients under any form of utilization management, relative to psychiatrists with a salaried income source (with control for other significant patient and clinician factors). Similarly, clinicians in individual practice settings were 1.9 times more likely to change their treatment plans than clinicians in other settings.

More than half of the patients in a nationally representative sample of psychiatric patients were receiving care that was subject to one or more utilization management techniques. Patients treated under utilization management were significantly more likely to receive a different treatment or less intensive treatment than the patients who were not treated under utilization management. Approximately one-third of these patients received a different form of treatment than their psychiatrist would optimally have liked to provide, compared to 15.5% of the patients who were not treated under utilization management. It is notable that the utilization management methods found to have the greatest association with changes in treatment were those that have financial implications: formulary restrictions, incentives to limit referrals, and limits in referrals to hospitals and specialty panels. In contrast, there was no statistical association between utilization review or treatment algorithms and psychiatrists’ reported changes in treatment provisions. Although the application of treatment algorithms is still uncommon, utilization review is commonly used in managed care. There may be two reasons for this absence of statistical significance. It is possible that psychiatrists are acting as effective advocates for their patients and have become increasingly sophisticated in requesting authorization for continued care, thereby reducing the actual impact of utilization review. Alternatively, the confidence interval of the 40% increase in the odds of treatment change in the presence of utilization review suggests that this relationship might be significant with a larger sample size. An even greater effect size was found in the association between utilization review and reducing the frequency and/or number of visits.

Two psychiatrist characteristics were associated with treatment changes. Psychiatrists in individual practice were significantly more likely to change their treatment plans relative to the psychiatrists practicing in other settings. Physicians who practice in settings other than individual practice may be more removed from the utilization management process and decisions than physicians in solo practice (e.g., a nurse or other member of the staff may be more likely to coordinate utilization management review processes). Practicing with others may provide greater bargaining power when dealing with managed care organizations that implement utilization management. Alternatively, psychiatrists in group or other nonindividual settings may be practicing in a manner that is already more consonant with managed care guidelines and may thus have less need to alter their treatment plans in response to utilization management. Similarly, psychiatrists receiving a source of income other than salary were more likely to reflect an association between treatment modifications and utilization management techniques. This is a logical association because salary is generally considered fixed and independent of either productivity or profit, whereas other types of income are often tied to the generation of income or to practice behaviors. The latter are more characteristic of a practice environment influenced by managed care techniques.

A significant association was found between the presence of formulary restrictions and the provision of different treatment. In evaluating this association, it is important to consider whether this change in psychiatrist behavior is good or bad for patient care. There is a growing body of literature addressing the influence of formulary restrictions on clinical outcomes. Randomized clinical trials have repeatedly shown that there are no significant differences in clinical effectiveness among individual selective serotonin reuptake inhibitors (SSRIs) or between SSRIs and other classes of antidepressants in the treatment of depression in outpatient settings (14–16). The fact that SSRI drugs are equally effective on average, however, does not mean that they are equally effective for individual patients (17). Previous research has shown that among patients who do not respond to one SSRI antidepressant, at least half will experience benefit from another drug in the same class (18, 19). Based on the conclusions of these studies, it would not be unreasonable to restrict formularies for the first-line treatment of depression in an outpatient setting. The restriction of other antidepressants for use as second or third-line treatments cannot be similarly supported by the literature. Furthermore, it is useful to specifically study the use of formulary restrictions in actual managed care settings. In one study examining the effect of “single-drug” formulary restrictions on the likelihood of the completion of drug therapy for new patients, it was found that patients from a health maintenance organization (HMO) with a single preferred SSRI (paroxetine) were 80% less likely to complete therapy than were patients from an HMO with two preferred SSRIs (fluoxetine and paroxetine). The formulary effect was found to be independent of the initial drug used to treat the patient (20). These findings suggest that the use of formulary restrictions may have a substantial impact on the likelihood of patients completing therapy. Undertreated or untreated depression has well-documented adverse effects on the individual as well as on society.

Another significant association was found between the incentive of utilization management techniques to limit referrals and the likelihood of reducing the number or frequency of visits. The question remains whether reduced visits to the psychiatrist actually affect quality of care and in which direction. Although there are no specific evidence-based guidelines that reflect the influence of too few visits on care, it is worth examining some nationally accepted standards. The Health Employer Data and Information Set is used to rate health plans that cover the majority of Americans enrolled in HMOs (21). One recent study (22) examined the rates and correlates of mental health care performance measures in the Health Employer Data and Information Set (22). Among the mental health care quality measures recommended by the Health Employer Data and Information Set and used for HMO “report cards” are 1) ongoing medication treatment in the 6 months after a new depressive episode and 2) optimal practitioner contact as defined by at least three follow-up mental health care visits in the 3 months after a new depressive episode. The mean rate of all mental health care performance measures was found to be significantly lower than for non-mental-health care domains. Worse quality of these domains was also predictive of poor mental health care performance. The findings suggest that commitment of resources to clinical care predicts better mental health care performance for U.S. HMOs. In this context, it is reasonable to consider that fewer visits to a psychiatrist, at least in the case of the treatment of depression, may result in a compromise of optimal care. In the aggregate, it appears that the presence of utilization management techniques is associated with changes in treatment behaviors and that these changes may result in less than optimal care. Clearly, managed care as a whole may have some benefits because it has reduced harmful practices that were common in the era before managed care, such as unnecessary hospitalization, poor follow-up after discharge, and care that was incongruent with practice guidelines. However, the specific utilization management mechanisms described in this study appear, taken together, to be associated with changes in provider treatment behavior that may have led to less than optimal care.

The strengths (23) and generalizability (13) of a practice-based research network for observational health services research have been documented in detail elsewhere. Particularly germane to this study is the advantage of capturing data on representative patients receiving representative treatments from psychiatrists with varying clinical backgrounds across the full range of managed and nonmanaged treatment settings. Several limitations should be noted, however. First, in the absence of a randomized design, the observed effects of utilization management may be confounded by other important health plan features, including mental health benefit designs (e.g., mental health coverage levels, copayments, deductibles) (13). Second, both the utilization management mechanisms and psychiatrist treatment plans are self-reported by the psychiatrist. This leads to the possibility of bias in either of two directions: dissatisfied psychiatrists could overreport changes in treatment plans or, alternatively, social desirability (i.e., not wanting to believe that as physicians they would ever be affected by financial incentives) could keep them from reporting actual changes in treatment. This use of self-report also leads to the possibility that psychiatrists may have misinterpreted survey questions. For instance, they may have understood the phrase “financial considerations” in the dependent measure to either narrowly reflect only their considerations or the broader financial considerations embodied in managed care limitations, the patient’s resources, or limitations of a public system. Third, it would be erroneous to assume that the utilization management techniques included as the independent variable had a causal relationship with identified treatment changes in this sample. We have been careful to describe these relationships as associations. However, despite the limitations inherent in trying to infer what an individual psychiatrist may have been thinking when he or she answered the survey question, the associations do highlight the importance of co-occurrence of psychiatrists’ perceptions of utilization management techniques and changes in treatment.

This study’s findings demonstrate that utilization management has a substantial and significant association with psychiatrists’ treatment patterns. With the goal of ensuring care that is guideline concordant and evidence based while promoting quality of care rather than cost containment (24), further research will have to examine the link between utilization management techniques, associated psychiatrist treatment changes, and subsequent patient and population outcome.

Received Feb. 21, 2002; revisions received July 11, 2002, and May 6 and Aug. 26, 2003; accepted Oct. 27, 2003. From the University of Miami School of Medicine; the University of Pennsylvania School of Social Work, Philadelphia; Emory University, Atlanta; and the American Psychiatric Institute for Research and Education, Arlington, Va. Address reprint requests to Dr. Brickman, Department of Psychiatry and Behavioral Sciences, University of Miami School of Medicine (M861), P.O. Box 016960, Miami, FL 33101; [email protected] (e-mail). The Practice Research Network is funded by the John D. and Catherine T. MacArthur Foundation, the American Psychiatric Foundation, the Center for Mental Health Services, and the Center for Substance Abuse Treatment. The authors thank the psychiatrists from the Practice Research Network who participated in the American Psychiatric Association Practice Research Network’s 1999 study of psychiatric patients and treatments (Harold Alan Pincus, M.D., and Ivan D. Montoya, M.D., M.P.H., coprincipal investigators).