Mental Health Benefit Limits and Cost Sharing Under Managed Care: A National Survey of Employers

We sought information on health benefits from 1,433 employers with active long-term-disability insurance policies with UNUM, a major disability insurance provider. Selection criteria included being a UNUM customer over the preceding three years and having at least 300 covered lives on the long-term-disability policy at some time during that three-year period.

Employers received a mailed questionnaire in the latter half of 1996 and were asked to send in summary plan descriptions or a detailed plan abstract form for each health plan offered to their employees. Data were obtained on 577 health plans offered by 250 employers. Although the overall response rate was low, it is comparable to rates in other mail surveys of employers. Moreover, complete data on all mental health benefit questions were obtained for every respondent and every health plan. No evidence of response bias was found. Detailed comparisons showed essentially no respondent-nonrespondent differences in descriptive characteristics, such as size of the employer's organization, industry, or region.

Respondent firms were located in the following geographic regions: 38.8 percent were in the East, 27.2 percent in the East Central region, 17.2 percent in the West Central region, and 16.8 percent in the West. A total of 27.7 percent were services firms, and 24.9 percent were manufacturing firms. Other types of firms accounted for 47.4 percent of respondents.

Among the 241 respondents who reported their total number of employees, the mean was 1,749. Applying this figure to all respondent firms implies that more than 437,000 employees were represented in the survey data. Respondent firms ranged in size from 21 to 27,519 employees; more than 95 percent reported having more than 200 employees.

Each plan was initially classified as an HMO or a non-HMO based on the language in its summary plan description or plan abstract. We reviewed plan information to verify that all HMOs included restrictions on the providers from whom treatment could be sought under "in-network" benefit provisions. HMOs with coinsurance for routine ambulatory care were reclassified as non-HMOs. Preferred provider organizations (PPOs) with small copayments rather than coinsurance for routine ambulatory care were reclassified as HMOs. Plans were coded as mental health carve-outs if their summary plan description identified a different organization as the contact for authorization, review, coordination, and administration of mental health benefits.

HMOs constituted 37.4 percent of plans; non-HMOs were split fairly evenly between standard fee-for-service plans (28.4 percent of the plans) and PPOs (34.1 percent). In all, 111 plans included a mental health carve-out. The proportion of plans with carve-outs was twice as high for both HMO and PPO plans (22.2 percent and 23.4 percent, respectively) as for fee-for-service plans (10.4 percent).

We used chi square and t tests to analyze differences in use of limits and cost-sharing between HMOs and non-HMOs and between carve-outs and non-carve-outs. We excluded from our analysis four of the 577 plans, including one HMO and three non-HMOs, that did not provide any mental health coverage.

Table 1 shows the results of the comparison of HMOs and non-HMOs and the comparison of carve-outs and non-carve-outs. (Results of the comparison of PPOs and fee-for-service plans are available from the authors.) Carve-outs were significantly less likely to have an overall exclusion for substance abuse treatment. HMOs were significantly more likely to exclude coverage for substance abuse rehabilitation and to exclude coverage for outpatient substance abuse treatment. HMOs were more likely to limit mental health coverage to short-term treatment for acute episodes or crisis intervention. Carve-outs were less likely to exclude partial hospitalization services from coverage.

The overwhelming majority of plans included specific limitations. Carve-out plans were less likely to impose any specific limits, however, and HMOs were less likely to impose specific limits on inpatient services. (All data on HMOs' limits reported here are for in-network providers. See Sturm and McCulloch [4] for 1996 data on limits for carve-outs, including non-network providers.)

HMOs made less use of expenditure limits. Carve-out and non-carve-out plans looked alike in their use of expenditure limits, except that carve-outs were less likely to use annual dollar limits for outpatient services. In each plan category, less than 2 percent of plans had lifetime utilization limits. HMOs were more likely than non-HMOs to use annual visit and day limits, while the corresponding differential for carve-outs was significant only for visit limits.

HMOs had significantly more stringent levels of annual outpatient and inpatient utilization limits than non-HMOs. Analogous results for partial hospitalization were mixed. Both HMOs and carve-outs were significantly less likely to combine the partial hospitalization limit with their inpatient benefit limit, which suggests less stringent restrictions for these plans. Conversely, among plans with separate partial hospitalization limits, annual day limits were significantly higher for HMOs and significantly lower for carve-out plans. Annual dollar limits, however, were significantly higher for carve-out plans.

HMOs were more likely to use copayments and less likely to use coinsurance than were other types of plans. Carve-out plans were more likely to use outpatient copayments and mental health deductibles and less likely to use outpatient coinsurance and inpatient copayments.

HMOs using coinsurance had significantly higher coinsurance rates than non-HMOs using coinsurance. Outpatient coinsurance and inpatient copayment levels were significantly lower in carve-out plans.