Managed Care Organizations’ Use of Treatment Management Strategies for Outpatient Mental Health Care
- First Online:
- Cite this article as:
- Merrick, E., Horgan, C., Garnick, D. et al. Adm Policy Ment Health (2006) 33: 104. doi:10.1007/s10488-005-0024-0
- 84 Views
A nationally representative sample of managed care organizations was surveyed (response rate=92%) regarding use of treatment management techniques for outpatient mental health care in their commercial products in 1999. Bivariate tests and logistic regression models were used to examine the relationship between product type, behavioral health contracting arrangement and treatment management techniques (prior authorization, standards for time to initial appointment, concurrent review, standards for follow-up after discharge, case management, practice guidelines). Prevalence varied from 43% to 87% depending on the technique. HMO products and products with specialty behavioral health contracts were more likely to use the techniques. Product type and contracting arrangement had independent effects.
KEY WORDS:mental healthmanaged careutilization managementcontracting
Managed care entities—whether general managed care organizations (MCOs) or specialty managed behavioral health care organizations (MBHOs)—have a variety of mechanisms available to manage mental health care, that is, to control or influence treatment services. These organizations can use techniques aimed at providers by applying utilization management techniques, carefully selecting their networks, or using financial incentives; they can also focus on patient-oriented constraints and incentives such as benefit features (Grembowski et al., 2000). Landon, Wilson, and Cleary (1998) identify four ways that MCOs influence physicians: financial incentives, structural characteristics, information or normative influences such as interactions with colleagues, and management (administrative) strategies. The last realm, within which our study falls, includes case management practices, critical pathways or guidelines, and other approaches designed to directly influence physician behavior. We use the term “treatment management” to describe the wide range of administrative techniques that can be used to influence care.
These mechanisms can apply not only to general medical care but also to specialty mental health services (recognizing that specialty mental health professionals are frequently psychologists, clinical social workers, or other disciplines rather than physicians). In the mental health field, additional organizational complexity arises from the recent tendency to “carve out” behavioral health care, i.e. contract for it separately with specialty managed behavioral health care organizations (MBHOs), at either employer or health plan levels (Hodgkin, Horgan, & Garnick, 1997). There are also clinical and societal complexities in the mental health area due to lingering stigma and discriminatory coverage compared to medical conditions (USDHHS, 1999), the fact that a large proportion of people with mental health disorders are untreated or treated inadequately (Kessler et al. 2003; Regier et al., 1993; Wang, Berglund, & Kessler, 2000) and difficulties in quantifying appropriate care for some conditions and treatments. Organizational variation and the challenges of mental health treatment suggest the need to understand how mental health treatment is managed within major product types and contracting arrangements.
Utilization management may be defined broadly to include prior authorization, concurrent review, case management, medical necessity criteria, and practice guidelines (Mihalik & Scherer, 1998). Tischler (1990) describes utilization management as the cost-containment strategy that has the most immediate impact on care, since a third party directly intervenes in treatment decisions. Certainly, much controversy has been generated by techniques such as prior authorization (in which authorization is required prior to initiating a treatment), concurrent review (in which periodic reviews of ongoing treatment are conducted to determine whether to authorize additional treatment), and case management (which focuses particularly on potential or actual high-cost users). There are concerns that these strategies may prevent people from receiving needed care, incur major time costs (“hassle factor”) for providers without significant benefit to patients, intrude into the confidential therapeutic process, fail to be cost-effective for outpatient care, and/or represent an excessive focus on cost at the expense of quality (Borenstein, 1990; Hennessy & Green-Hennessy, 1997; Miller, 1996). On the other hand, utilization management can serve to triage patients into appropriate care, facilitate access to services, and eliminate unnecessary or inappropriate care, enabling limited resources to be used efficiently while containing costs. Furthermore, there are related treatment management mechanisms such as standards for timely access to care that clearly aim to facilitate service delivery.
There is little nationally representative data regarding how often commercial managed care plans use such treatment management approaches for mental health services and how they apply them. One exception is a study by Barry and colleagues (2003) that found prior authorization for outpatient mental health was more common among HMOs and carve-out plans. Numerous investigators have, however, reported on the characteristics and/or effects of utilization management mechanisms such as prior authorization and concurrent review in behavioral health care, for particular populations or plans. Most studies have found that some utilization management strategies are associated with lower costs and/or quantities of treatment (e.g., Frank & Brookmeyer, 1995; Gotowka & Smith, 1991; Howard, 1998; Liu, Sturm, & Cuffel, 2000; Wickizer & Lessler, 1998; reviews by Hodgkin, 1992; Mechanic, Schlesinger, & McAlpine, 1995), though a smaller number have found only weak or no effects (e.g., Dickey & Azeni, 1992) or that effects are often short-term (Frank & Brookmeyer, 1995). Reductions can occur directly through denials or approving less treatment than requested (e.g., Wickizer & Lessler, 1998) and/or through a “sentinel effect” in which the existence of the utilization management system itself may deter care despite typically low denial rates (e.g., Liu et al., 2000; and Howard, 1998 in regard to prior authorization for outpatient care). Some findings have raised concerns about negative impacts on quality of care (e.g., increased readmission rates found by Wickizer & Lessler, 1998; Frank & Brookmeyer, 1995).
Case management is often used in two distinct ways (Mechanic et al., 1995). In the public sector, case management typically focuses on persons with serious and persistent mental illness and is often intensive, including approaches such as assertive community treatment. There is considerable literature on public-sector mental health case management (e.g., Jinnett, Alexander, & Ullman, 2001; Kuno, Rothbard, & Sands, 1999; Olfson, 1990), but the degree to which this relates to private-sector case management programs is unknown. In the private sector, case management tends to focus on high-cost or high-utilizing enrollees. There is little scientific literature on private-sector case management (e.g., Frank & Brookmeyer, 1995), though the recent adoption of “disease management” programs (which generally go beyond case management) for depression suggests that evolution is occurring.
A substantial body of research on practice guidelines, some of it specific to mental health, focuses on adherence rates and outcomes rather than prevalence of guidelines in health plans. Findings include frequently low rates of guideline-concordant treatment, some (inconsistent) evidence for improved outcomes in some diagnostic groups when guidelines are followed, and the need for comprehensive, ongoing adherence interventions (e.g., review by Bauer, 2002; Fortney, Rost, Zhang, & Pyne, 2001; Wang et al., 2000). Regarding timely access to care and follow-up after discharge, studies have reported rates in various populations (e.g., Merrick, 1999; Sturm, 1999) but have not focused on the prevalence of standards.
This study uses data from a nationally representative survey to describe the prevalence of treatment management practices for outpatient mental health care in MCOs, and to analyze how it may vary by product type and contracting arrangement. We include both traditional utilization management techniques focused on individual cases (prior authorization, concurrent review, case management) and additional tools that can theoretically support higher-quality care (practice guidelines, standards for time to initial appointment, standards for time to follow-up after discharge).
The results of this study will inform the debate on methods of managing mental health treatment by providing a national picture of how MCOs choose to manage mental health care. While the perception is that, overall, managed care is becoming looser (Robinson, 2001), it is not clear what has occurred in regard to mental health. Further, when purchasers, policy makers, and others consider alternative product types and contracting arrangements, it will be useful to know the treatment management parameters associated with them and to understand which characteristics appear to have independent effects. In addition, documenting the prevalence and variation in treatment management techniques will help to guide further research by indicating where additional focus is needed. The 1999 data will provide an essential baseline from which to measure changes in this critical area.
Data and Sample
The primary data source is the 1999 Brandeis Survey on Alcohol, Drug Abuse and Mental Health Services, a nationally representative telephone survey of MCOs. We surveyed 434 MCOs in 60 market areas (response rate=92%). Each MCO was asked about its top 3 commercial managed care products in terms of enrollment in 1999. The survey included an administrative module addressing contracting arrangements, benefits, and provider payment, and a clinical module that addressed utilization management, treatment entry mechanisms, prescription drug formularies, quality improvement, and other topics. Typically there were two respondents per MCO (executive director and medical director, or their designees). The survey interviews were conducted by Mathematica Policy Research (Princeton, NJ) on behalf of Brandeis University.
The study is linked to the Community Tracking Study (CTS), a longitudinal study of health system change, funded by the Robert Wood Johnson Foundation and described in Kemper et al. (1996). The primary sampling units for our survey were the 60 market areas selected for the CTS to be nationally representative. Our second sampling stage consisted of selecting MCOs within market areas. MCOs serving multiple markets were defined as separate MCOs for the study and data were collected with reference to the specific market area. From our CTS-derived sample frame of 944 MCOs, 720 market-specific MCOs were selected with equal probability and without replacement across the 60 sites. The MCO sample was allocated to each market area based on the weighted estimate of MCOs represented by each site, a minimum sample allocation per site, and the weighted estimate of MCO enrollees in each site. The allocation to each site was proportionally distributed between PPO-only and HMO/multi-type MCOs. Of the sampled MCOs, 247 were categorized as ineligible after screening because they had closed, merged, or were otherwise unreachable; had less than 300 subscribers in the market area; did not offer comprehensive health care products; served only Medicaid/Medicare; or for other reasons. This left 473 eligible MCOs, of which 434 (92%) responded and reported on 787 eligible products. This analysis is based on the 752 products for which 417 MCOs completed the clinical module. Overall, on a weighted basis, 40.6% of these products were health maintenance organization (HMO) products, 35.7% were preferred provider organization (PPO) products, and 23.7% were point-of-service (POS) products. The products varied by behavioral health contracting arrangement: 59.8% had specialty MBHO contracts, 15.5% had comprehensive contracts which included general medical as well as behavioral health services, and 24.7% had internal-only arrangements.
We used data from other sources to describe market-level characteristics and to supplement our enrollment data. For HMO penetration rates, we used 1998 county-level estimates developed by others based on Interstudy data (Baker, 2001) and aggregated these to the market-area level. For market area population, we used 1998 Census Bureau estimates from the 1999 Area Resource File maintained by the U.S. Health Resources and Services Administration (2001) aggregated from county level to match our market areas. For enrollment, we used survey information for MCOs that provided valid responses; when missing, we used sources including MCO website searches and industry directories.
Dependent variables include those based on survey items asking whether prior authorization was required to initiate specialty mental health outpatient treatment, and if so, what was the typical number of outpatient visits authorized (or alternatively, whether it varied depending on the patient’s diagnosis). We asked similarly about concurrent review. Regarding standards for maximum time to initial appointment and for follow-up after discharge from inpatient/residential care, we asked whether the product had each standard and if so, what the time period was. For practice guidelines, we chose three mental health conditions as markers (major depressive disorder, panic disorder, and schizophrenia) and asked whether the organization had practice guidelines for each within the product. Finally, we asked whether the product had a behavioral health case management program.
Key independent variables include product type and contracting arrangement. We categorized managed care products as HMO, PPO or POS. 1 We did not distinguish between different models within a product type. We categorized contracting arrangements as specialty (contracts with a managed behavioral health care vendor or provider organization for behavioral health services), comprehensive (contracts with a comprehensive network including both general medical and specialty behavioral health providers), or internal (provides behavioral health services via salaried staff or network managed directly by the MCO).
Results are weighted to be representative of MCOs’ commercial managed care products in the continental United States. Statistical analyses were implemented using SUDAAN software (Shah, Barnwell, & Bieler, 1997) to allow correction of standard errors for our complex survey design and consequent clustering. We performed pairwise chi-square tests for categorical variables and t-tests for continuous variables to determine the significance of differences by product type and by contracting arrangement. We corrected for multiple comparisons (three pairwise tests for each bivariate comparison by product type or contracting arrangement) by using the Bonferroni correction, and only corrected p values are reported based on corrected critical chi-square values of 5.731 (for p<.05) and 8.615 (for p<.01).
We constructed logistic regression models to explain the use of each treatment management tool: required prior authorization for regular outpatient mental health care; concurrent review for regular outpatient mental health care; standards for time to initial routine mental health appointment; standards for time to follow-up after discharge; case management for behavioral health; and practice guidelines for any of the three marker mental health conditions. Analyses are conducted at the product level. Products with missing data are excluded, so N varies slightly across models. Each dependent variable is binary, indicating whether the product employed the management technique. Key explanatory variables in all models are product type (dummy variables for PPO and POS, with HMO products as the omitted reference category) and contracting arrangement (dummy variables for specialty- and comprehensive-contract products, with products with internal delivery as the reference group).
Distribution of MCO Products by MCO and Market Characteristics (Percent of Products, Weighted)
Subsidiary/division of multi-state organization
For-profit, privately held
For-profit, publicly held
MCO enrollment in market
Market area population
Outpatient Mental Health Treatment Management Tools Used in MCO Products: Prevalence by Product Type and Contracting Arrangement (percent of products, weighted)
By product type
By contracting arrangement
Preauthorization required for regular outpatient care
Concurrent review required for regular outpatient care
Standards for time to initial appointment(routine)
Standards for follow-up after discharge
Practice guidelines(any of 3 MH)
Number of Treatment Management Techniques Employed in MCO Products (Weighted Percent of Products, by Product Type and Contracting Arrangement)
By product type
By contracting arrangement
# Strategies used
Content of Treatment Authorization and Standards When Required by MCO Products (Percent of Products, Weighted)
By product type
By contracting arrangement
Preauthorization for regular outpatient mental health care
Varies by diagnosis
By typical # visits
Mean (SE) visits
Median # visits
Modal # visits
Concurrent review (regular outpatient)
By typical #visits
Mean (SE) visits
Median # visits
Modal # visits
Standards for time to initial appointment
Mean (SE) days
Median # days
Modal # days
Standards for follow- up after MH discharge
Mean (SE) days
Median # days
Modal # days
Use of Treatment Management Tools in MCO Products: Logistic Regression Results
Initial appointment time standards
Follow-up after discharge standards
Weighted N (products)
Product type (HMO omitted)
Contracting arrangement (Internal omitted)
Tax status (not-for-profit omitted)
For-profit, privately held
For-profit, publicly held
Subsidiary/division of multi- state organization
Region (West omitted)
MCO enrollment in market (>50,000 omitted)
HMO penetration rate in market Area
Market area population (>4 million omitted)
Contracting arrangement variables are significant in every model. The specialty-contract variable is positive and significant in prior authorization, concurrent review, and case management analyses, indicating that this characteristic is associated with higher likelihood of using the strategies relative to products with internal behavioral health arrangements. The comprehensive-contract variable is negative and significant for standards to initial appointment time, practice guidelines, and standards for follow-up after discharge. Some of the covariates are also significant in some of the logistic regression models. For example, for several management tools, products in smaller market areas are more likely to use them relative to those in the largest market areas (population over 2 million).
The results of this study provide evidence of a fairly high degree of treatment management for outpatient mental health services by MCOs overall, and also demonstrates that their prevalence varies significantly by type of managed care product and by contracting arrangement. In particular, HMO products—and sometimes POS products—are more likely to employ the management strategies we analyze here, in comparison to PPO products. Specialty-contract products are more likely to use these management tools relative to products with other contracting arrangements. The bivariate results were largely confirmed in multivariate analyses, which showed independent effects of both product type and contracting arrangement. Furthermore, among products that employ the techniques, HMOs and specialty-contract products appear to usually apply them more stringently, for example by typically approving a lower number of outpatient mental health visits at prior authorization.
There are several reasons that non-PPO products and/or products with specialty contracts might be more likely to use these treatment management approaches. In regard to product type, higher use in HMOs and POS products could partially result from external influences. These include participation in National Committee on Quality Assurance (NCQA) accreditation or Health Plan Employer Data and Information Set (HEDIS) performance measurement, which directly or indirectly involve several of our measures, and in which PPOs have only recently begun to participate. At the same time, it is possible that HMO and POS products, as well as specialized vendor systems, are more geared to implement these approaches because of their information systems and provider network structures.
PPOs and non-specialized products may rely more heavily on benefit restrictions such as cost sharing, rather than utilization management techniques, to control utilization. This would preserve consumers’ perception of choice and freedom while providing financial incentives to use less care. It has, in fact, been noted that a hallmark of managed care is that it controls utilization by influencing providers or directly intervening through utilization review, rather than relying on consumer cost sharing (Frank & McGuire, 1998), but this trait may vary depending where the managed care product falls on the management spectrum. In other analyses of these survey data, it was found that higher cost sharing for mental health compared to general medical care was equally prevalent among product types (Hodgkin, Horgan, Garnick, & Merrick, 2003). However, HMOs were less likely to use coinsurance, a form of cost sharing in which burdensome levels were frequently reached. Future work will investigate the direct relationship between behavioral health cost sharing and aspects of treatment management.
It is noteworthy that the same product characteristics are associated with higher use of two different types of management strategies: those unequivocally geared towards facilitating access (such as specifying maximum wait time to initial appointment) and those which some view as oriented towards controlling utilization (such as prior authorization and concurrent review). This suggests that HMOs and specialty-contracted products take a more assertive treatment management stance in general. In fact, previous study findings that quality management activities such as conducting clinical outcomes assessment were more likely in the same product groups raises the possibility that the degree of management in a very broad range of domains may be correlated (Merrick, Garnick, Horgan, & Hodgkin, 2002). Thus, there may be a trade-off here between perceived threats and benefits of tighter management, the evaluation of which could depend partly on individual consumer preferences. Behavioral health service users whose major concerns revolve around choice and privacy may well prefer low management all around. Those who seek guidance regarding levels of care or providers, have complex treatment needs, or think it a safeguard to have providers potentially held accountable by a larger system, may be more willing to live with tighter controls. However, most enrollees probably do not know about treatment management mechanisms in any detail prior to accessing services. Furthermore, consumer costs have been found to factor into the larger and probably overriding trade-off of management versus price. For example, Kemper, Tu, Reschovsky, and Schaefer (2002), in a study based partly on the Community Tracking Study survey which formed our sample frame, found a trade-off between paying more out of pocket and experiencing more administrative barriers to health care in general. This may similarly prove to be the case in relation to mental health.
Exactly how, and on what basis, the treatment management strategies are implemented could also make a big difference in their effects (and desirability from various stakeholders’ perspectives). For instance, if concurrent review decisions are based on solid clinical expertise and valid medical necessity criteria, require only essential information, and are processed in a way that places little burden on providers while guarding patients against unnecessary or inappropriate care, the results may be quite different than in another system. Similarly, a well-run and clinically sound case management program could help people access additional services they might need and coordinate complex care. Alternatively, there could be an excessive focus on cost containment.
In addition to the particular characteristics of utilization review practices used by MCOs, practitioners’ responses—including various types and frequency of advocacy behavior—have implications for the ultimate effects on treatment (Schlesinger, Gary, & Perreira, 1997; Wolff & Schlesinger, 2002). Provider behavior is also critical in determining the effects of standards such as maximum wait time to first appointment: if many practitioners do not know about and/or adhere to the policy, the practical effects of such a standard would be negligible. Practice-level variables may also have an important effect on how a management approach works (Landon, Wilson, & Cleary, 1998).
Limitations of our study include the lack of more detailed information regarding implementation of these management tools, for example how frequently treatment requests are denied in the utilization management process or how often prior authorization serves a useful triage function. However, as noted earlier, there is some evidence suggesting a sentinel effect in which it is the existence of a utilization management system with certain milestones rather than the actual denial rate that appears to affect care (Howard, 1998; Liu et al., 2000). To the extent that a sentinel effect occurs, the basic findings of requiring particular types of review at various intervals are meaningful. We do not know the actual effects of these treatment management strategies as our survey data did not include individual-level data or aggregated quality measures; future research on this linkage would be useful. Also, in this analysis we examined only outpatient mental health services; patterns may vary for other mental health or substance abuse services. Additional fruitful areas for continued research include how different medical necessity criteria relate to the variation we have documented, and multiple stakeholder analyses of provider and patient experiences of the management techniques we studied here.
Our study does, however, provide national estimates of prevalence and characteristics of treatment management mechanisms that, according to the literature, may have important ramifications for treatment. It further identifies the independent effects of product type and contracting arrangement on behavioral health treatment management. This knowledge should help to guide health care decision makers at every level in focusing on aspects of treatment management that vary systematically and perhaps warrant closer attention. This also represents an essential step in the ongoing research effort to understand how the increasingly complex structures of managed behavioral health care matter in terms of the services that people receive, and provides a baseline from which to measure ongoing change.
We used the Community Tracking Study Follow-back Survey definitions for product type as follows:
HMO: A product in which enrolled individuals are provided health care services by a network of affiliated providers. Services provided to enrollees outside the network are generally not covered, other than for some specialized services or in emergencies.
POS: A product in which enrollees may select in-network or out-of-network physicians at the “point-of-service” usually with significant differences in coinsurance or deductibles. Some POS products are also referred to as “open-ended” HMOs or “triple option” plans.
PPO: A product in which enrollees are given a financial incentive to use a “preferred” network of providers, usually through differences in coinsurance or deductibles.
This research was supported by grants from the National Institute on Drug Abuse (Grant #R01 DA10915), the National Institute on Alcohol Abuse and Alcoholism (Grant #R01 AA10869), and the Center for Mental Health Services in the Substance Abuse and Mental Health Services Administration. The authors thank all of the respondents for their participation in this survey. The authors acknowledge the contributions of Robert Cenczyk, Della Faulkner, David Goldin, Will Lusenhop, Frank Potter and all of the project team at Mathematica Policy Research, Inc., Grant Ritter, Kathleen Carley Skwara, and Galina Zolotusky, and thank Michele King for manuscript preparation assistance. The authors also thank Laurence Baker for sharing his HMO enrollment data.