This Web site is a component of the SAMHSA Health Information Network

Enter Keywords Choose Website This Site Mental Health at HHS All Mental Health

Online Publications Order Publications National Library of Medicine National Academies Press Publications Homepage

printer friendly page e-mail this page bookmark this page shopping cart  current or new account

This Web site is a component of the SAMHSA Health Information Network.

Effects of the Vermont Mental Health and Substance Abuse Parity Law

Appendix D: Methods Used to Conduct the Survey of Vermont Employers

This appendix describes the methods used to conduct the Survey of Vermont Employers, including the sample design, data collection procedures, and analytic approach. This appendix also presents background information on the characteristics of Vermont employers and the attributes of employer-sponsored health insurance coverage in Vermont.

A. Sample Design

The sample for the Survey of Vermont Employers was drawn from the Unemployment Insurance (UI) File maintained by the Vermont Department of Employment and Training (DET). The file contains all employers who paid unemployment taxes in Vermont. Since businesses are mandated to report unemployment taxes annually, the UI file provided an up-to-date sample frame. The target population for the survey was businesses currently in operation in Vermont, excluding (1) those that had, on average, fewer than five employees across establishments in calendar year 1999; (2) those not in business before January 1, 1998 (when the parity law was enacted); and (3) Federal and State government entities.

Employer surveys can be conducted at the enterprise or establishment level, and the sampling unit may depend on the objectives of the survey (Zarkin et al., 1995).³¹ Because most insurance decisions typically are made at the level of the enterprise, the sampling unit for the Vermont employer survey was defined as the "Vermont portion of the business enterprise." The DET sampling frame was used to identify those Vermont establishments associated with each enterprise operating in Vermont as of December 31, 1999.

The sample was selected using a stratified, simple random sample without replacement of businesses in Vermont enterprises. The records were divided into three strata: small (5 to 25 employees), medium (26 to 50 employees), and large (more than 50 employees). Each stratum was then divided into seven substrata: six substrata were formed based on three Standard Industrial Classification (SIC) codes (retail trade, services, and other) and two locations (Chittenden County and other counties), and a seventh substratum that included all local government entities. Thus, 21 substrata were defined for the study.

A total of 1,311 records originally were designated and separated into three waves. The goal was to complete 200 interviews with insured businesses in each of the three size strata (uninsured businesses would naturally distribute themselves across the strata as a result of screening for insurance status). The eligibility rate was projected at 85 percent and the response rate at 80 percent. However, after fielding two waves, the eligibility rate was much higher than expected (97.5 percent) and, as a result, the size of the third wave was reduced, such that a total of 1,040 records actually were released for interview: 421 in stratum 1 (small), 326 in stratum 2 (medium), and 293 in stratum 3 (large).

Altogether, 806 employers completed the survey (674 insured and 132 uninsured), and the overall response rate was 80.7 percent. Across the three size strata, the number of completed cases and response rates are shown in Table D.1.

B. Data Collection Procedures

Interviews were conducted using computer-assisted telephone interviewing (CATI). Prior to conducting the interview, pre-field locating was performed to confirm that the sampled businesses were still in operation in Vermont, to verify that sampled businesses were at the enterprise level, and to identify the appropriate respondent for the survey (defined as the head of Vermont operations). Contact information was verified, and an advance letter and information packet were mailed to each employer prior to the CATI interview. In addition, the survey was publicized to Vermont employers through stories in local newspapers and trade magazines and through an informational Web site.

The questionnaire included the following topics: (1) employer eligibility for the survey, including health insurance status (insured, uninsured); (2) eligibility for and participation in employer health plans; (3) characteristics of health insurance coverage; (4) costs of health insurance coverage; (5) awareness of the parity law; (6) effects of the parity law; (7) satisfaction with parity; (8) concerns and recommendations about parity; and (9) firm characteristics. Once an employer had been determined eligible for the survey, the interviewer identified the respondent who was most familiar with the Vermont parity law to conduct the remainder of the interview.

Uninsured businesses completed two sections of the instrument - the eligibility screener and firm characteristics. In addition, a brief set of questions was administered to newly uninsured businesses (uninsured since January 1, 1998) to determine the role of the MH/SA parity law in their decision to discontinue coverage.

Quality control was performed throughout the data collection process, including the use of supervisors and interviewers with experience on surveys of professionals; a one-day training session, including general instruction on data collection procedures and survey-specific training on the instrument and the project; consistency checks within the CATI system; random monitoring by the project director, survey director, and survey supervisor; and automated editing for skip patterns following completion of the survey.

C. Analytic Approach

Weights were developed for analysis, to adjust for the disproportionate probability of selection by size of employer. Medium and large businesses were oversampled, while small businesses were undersampled. Table D.2 shows the unweighted and weighted distributions of responding businesses across the three strata.

To account for the complex sample design, SUDAAN software was used to compute the standard errors for significance testing. Two types of significance tests were performed: a t-test for continuous variables and chi-square test for categorical variables. Unless otherwise specified, all reported differences are significant at the .10 level or higher.

Most analyses compared employer perceptions of, and responses to, parity by the size of the firm. A measure of firm size was created based on the number of permanent full-time and part-time Vermont employees, as of December 31, 1999, as reported in the survey. Four size categories were analyzed: fewer than 10 (very small), 10 to 25 (small), 26 to 50 (medium), and more than 50 (large). Thus, for the purpose of this analysis, we divided the stratum containing 25 or fewer employees into two analytic categories because of the differences in characteristics and responses of the small and very small businesses.

Most analyses are performed at the employer level to ascertain differences in employer attitudes and responses to parity. In addition, some analyses are presented at the employee level; that is, they are weighted by the number of employees in the firm. These analyses estimate the proportion of Vermont employees affected by various changes in employer-sponsored health insurance coverage (such as the percent affected by the discontinuation of coverage or by the shift to self-insured coverage).

D. Background Information on the Characteristics of Vermont Employers

Because the analysis in Chapter IV focuses on variations in employer perspectives on parity according to firm size, this appendix provides background information on the characteristics of Vermont businesses by firm size. The definition of "large business" used in this survey - more than 50 employees - is different from that used in many other surveys.³² This analysis focused on businesses with more than 50 employees, as distinct from those with 50 or fewer, for two reasons. First, the Federal mental health parity law (along with many State laws) exempts businesses with 50 or fewer employees from compliance. Second, Vermont's small business market is subject to different rating requirements than companies with more than 50 employees. Moreover, there is substantial evidence that virtually all businesses with more than 50 employees offer coverage, but there is considerable heterogeneity among smaller firms (KFF/HRET, 2001).

As shown in Table D.3, significant differences existed in the characteristics of businesses in Vermont along all dimensions other than urban/rural location. The vast majority (83 percent) of Vermont employers represented in the survey were for-profit enterprises; another 10 percent were not-for-profit, and the remaining 7 percent were local government entities (such as school districts). Businesses with 25 employees or fewer were more likely to be for-profit enterprises, whereas firms with more than 25 employees included a disproportionate representation of not-for-profit and publicly owned businesses.

Among the nongovernmental firms, more than one-third (37 percent) were associated with service industries, while about one-fourth (26 percent) were involved in retail trade. In general, businesses with 50 or fewer employees were more likely to specialize in retail trade, agriculture/forestry/fishing/mining, construction, and wholesale trade, while large businesses (more than 50 employees) were more likely to concentrate on manufacturing and services.

About 93 percent of businesses were headquartered in Vermont, though the likelihood of having a headquarters outside Vermont increased with size - 18 percent of firms with more than 50 employees had their center of operations in another State or even outside the United States. Large businesses also were more likely to have a union presence - 23 percent of those with more than 50 employees, versus 2 percent of those with fewer than 10 employees, employed staff with collective bargaining agreements.

The vast majority of Vermont firms had been in operation for more than 5 years, including 46 percent for 5 to 20 years and 45 percent for more than 20 years. A higher share of the large firms (79 percent) than small firms had been in business for more than 20 years. In general, the self-reported financial status was stronger in medium and large firms than in small or very small firms - 21 to 26 percent of firms with 25 or fewer employees reported that they were in fair or poor financial status, compared to 9 to 15 percent of firms with more than 25 employees. Finally, two-thirds of Vermont businesses were located in rural areas (outside Chittenden County), and there were no significant differences in the geographic distribution by firm size.

E. Characteristics of Employer-Sponsored Health Insurance Coverage in Vermont

This section provides background information to set the context for the discussion of the effects of parity on employers, including the rates and characteristics of employer offers of insurance coverage and employee participation; the number and types of health plan choices offered by employers; and the extent of employer monitoring of health care costs.

1. Employer Offers of Insurance Coverage

As shown in Table D. 4, three out of four Vermont employers offered employer-sponsored insurance (ESI) coverage to their employees at the time the survey was conducted. The likelihood of offering coverage increased significantly with employer size, ranging from 62 percent of employers with fewer than 10 employees in Vermont to 97 percent among those with more than 50 Vermont employees. Virtually all firms - 91 percent - that offered coverage to employees also offered coverage to their dependents. However, firms with 25 employees or fewer were less likely to offer dependent coverage than firms with more than 25 employees.

Firms offering ESI may restrict coverage based on the number of hours worked. Overall, about one-third of Vermont businesses offered coverage to part-time employees. Large firms (53 percent) were more likely to offer coverage to part-time workers than smaller firms (25 to 40 percent).

Firms may require a minimum length of employment prior to offering coverage to employees. About two-thirds of Vermont employers had a waiting period for eligibility, although the rate was slightly higher among employers with more than 25 employees. Only one-fourth of Vermont employers had a waiting period for preexisting conditions (PEC); employers with more than 25 employees were at least twice as likely as smaller businesses to have a PEC clause.

2. Employee Participation

In addition to finding considerable variation among firms in whether they offered coverage - and to whom they offered coverage - there was significant variation in the participation rate among eligible workers (Table D.4). Across all firms that offered coverage, the participation rate among eligible workers was about 72 percent. In other words, nearly three-fourths of workers who were eligible to participate in ESI actually obtained coverage. The participation rate was higher among eligible employees in businesses with more than 50 employees (78 percent) than among employees in small or very small businesses (70 to 72 percent). In part, this may reflect the tendency of employees in small businesses to obtain coverage through a spouse or partner who is employed by a larger firm (Cromwell et al., 1994).

The Vermont Family Health Insurance Survey provides insights into why some workers may decline coverage when it is offered (BISHCA, 2000). By far the most common reason - reported by 47 percent of employees who declined coverage - was that they had obtained coverage through a spouse's or partner's employer. Ineligibility due to part-time status was reported by 19 percent, while 11 percent reported that they were ineligible due to a waiting period. Sixteen percent cited cost as a barrier.

3. Health Plan Choices Offered by Vermont Employers

Among employers that offered health insurance coverage, there was considerable variation in the number and types of health plan choices (Table D.5). Nearly one-third of large firms (more than 50 employees) offered more than one choice, compared to about 5 percent of firms with 25 employees or fewer.

Vermont employers were most likely to report that they offered their employees a preferred provider organization (PPO) plan (42 percent), and less likely to report offering a health maintenance organization (HMO), either with or without a point-of-service (POS) option (38 percent combined). Only 21 percent of employers offered a traditional fee-for-service, or indemnity, option. There was no significant variation by employer size in the percentage that offered a managed care plan (HMO or POS), but large employers were more likely to report that they offered a PPO plan. These data mirror national trends in two respects - the entry of managed care into the small group market (Jensen et al., 1997) and the strong emergence of PPOs, as "heavier" forms of managed care retreat (Gabel et al., 2001).

To the extent that most employers offered only one health plan, many employees could not choose an alternative health plan follow ing implementation of parity. Moreover, when that one plan was a managed care plan, network composition or care management strategies may have affected the choice of providers and accessibility of care. Among businesses that offered only one plan, there were significant differences by firm size in the type of health plan offered. Businesses with 50 employees or fewer were more likely to offer a managed care plan (HMO or POS) as the only option, while large businesses (more than 50 employees) were more likely to offer a PPO plan as the only option (data not shown).

Approximately 15 percent of Vermont employers reported that they provided health insurance through a self-insured plan at the time of the survey (Table D.5). Large businesses were three times more likely to offer at least one self-insured plan. Thus, 30 percent of employees were employed by firms offering only a self-insured plan, while 12 percent were in firms with a choice between fully insured and self-insured plans (Table D.6). This means that, because self-insured plans are exempt from Vermont's parity law, nearly one in three Vermont employees worked for insured businesses that were beyond the reach of the Vermont MH/SA parity law.

4. Employer Monitoring of Health Care Costs

The majority of Vermont employers reported that they monitor their health care costs at least once or twice a year or upon contract renewal (Table D.7). The frequency of monitoring varied by firm size, however. Large firms were more likely to monitor their health care costs at least quarterly, while other firms were more likely to report that they never monitored their costs.

About half of Vermont employers relied on outside sources for monitoring, although the likelihood of using an outside source was about twice as high among the large firms (77 percent) as among the very small firms (39 percent). Among those using outside sources to assist in monitoring, the most common sources were insurance brokers (57 percent), benefits consultants (24 percent), and health plans (23 percent). Firms of all sizes relied most often on insurance brokers, although other differences were observed by firm size. Large firms were more likely than very small firms to hire benefits consultants (37 versus 16 percent) and more likely to rely on health plans or third-party administrators (30 versus 12 percent). Very small businesses were more than three times as likely to call on trade or professional associations, many of which sponsored association plans for small businesses in Vermont (24 versus 7 percent).

Despite these efforts to monitor health care costs periodically - often with the assistance of outside sources - few employers were able to estimate what percentage of their health care costs were attributable to MH/SA services. Between 78 and 93 percent of firms reported that the share of costs attributable to MH/SA services was either zero or unknown. A sizable proportion - about 40 percent - were unable to report the direction of the change in costs attributable to MH/SA services over the past 3 years. Another 41 percent reported that costs stayed the same, while 18 percent reported that they increased.

Previous | Table of Contents