Employers' Use of Value-Based Purchasing Strategies

Context Value-based purchasing by employers has often been portrayed as the lynchpin to quality improvement in a market-based health care system. Although a small group of the largest national employers has been actively engaged in promoting quality measurement, reporting, and pay for performance, it is unknown whether these ideas have significantly permeated employer-sponsored health benefit purchasing.

Objective To provide systematic descriptions and analyses of value-based purchasing and related efforts to improve quality of care by health care purchasers.

Design, Setting, and Participants We conducted telephone interviews with executives at 609 of the largest employers across 41 US markets between July 2005 and March 2006. The 41 randomly selected markets have at least 100 000 persons enrolled in health maintenance organizations, include approximately 91% of individuals enrolled in health maintenance organizations nationally, and represent roughly 78% of the US metropolitan population. Using the Dun & Bradstreet database of US employers, we identified the 26 largest firms in each market. Firms ranged in size from 60 to 250 000 employees.

Main Outcome Measure The degree to which value-based purchasing and related strategies are reported being used by employers. Percentages were weighted by number of employees.

Results Of 1041 companies contacted, 609 employer representatives completed the survey (response rate, 64%). A large percentage of surveyed executives reported that they examine health plan quality data (269 respondents; 65% [95% confidence interval {CI}, 57%-74%]; P<.001), but few reported using it for performance rewards (49 respondents; 17% [95% CI, 7%-27%]; P=.008) or to influence employees (71 respondents; 23% [95% CI, 13%-33%]). Physician quality information is even less commonly examined (71 respondents; 16% [95% CI, 9%-23%]) or used by employers to reward performance (8 respondents; 2% [95% CI, 0%-3%]) or influence employee choice of providers (34 respondents; 8% [95% CI, 3%-12%]).

Conclusion Surveyed employers as a whole do not appear to be individually implementing incentives and programs in line with value-based purchasing ideals.

Value-based purchasing has often been portrayed as the lynchpin to quality improvement in a market-based health care system.^1-6 Under this paradigm as it was originally conceived, employers and other large purchasers of health care are expected to contract with health plans according to quality and cost. Other key elements of value-based purchasing include the promotion of quality improvement in negotiations with health plans and facilitating informed choice of health plan through dissemination of comparative cost and quality information to employees.

With the release of several landmark reports on the poor quality of health care in the United States, there have been a number of recent, high-profile, employer-based efforts to play an active role in improving the quality of care, with a more direct focus on providers of health care rather than health plans alone.^7,8 These efforts include the Leapfrog Group's profiling of hospital patient safety measures, Bridges to Excellence's physician pay-for-performance program, and numerous business coalition endeavors.^9-11 Although a small group of the largest employers has been vocal and actively engaged in improving health care quality through the promotion of quality measurement,¹² reporting, and pay for performance, it is unknown whether these ideas have significantly permeated employer-sponsored health benefit purchasing.

In this study, we report on a national survey of large employers to ascertain their conduct of value-based purchasing of health care and related efforts to improve the quality of health care and employee health. Like many observers, we view quality improvement as an inherently local problem. Therefore, we considered the largest employers within a sample of metropolitan areas as the possible locus for value-based purchasing within their community.

Metropolitan Statistical Areas were identified and population data obtained according to 2000 Census definitions established by the Office of Management and Budget.^13,14 Data on health maintenance organization (HMO) enrollment were obtained from InterStudy.¹⁵ Firms were primarily identified by using the Dun & Bradstreet database of US employers, which catalogs both private and public employment information. Information about value-based purchasing initiatives and related efforts was obtained through telephone surveys of employers, as described below.

We randomly sampled 40 Metropolitan Statistical Areas with at least 100 000 HMO enrollees. These markets (Metropolitan Statistical Areas) include approximately 91% of individuals enrolled in HMOs and represent roughly 78% of the US metropolitan population. To produce a nationally representative sample, the number of markets selected in each of the 4 census regions was determined according to the region's proportion of the total US population. Specific markets were then randomly selected in each region, with a probability proportional to the size of the market within the region. To ensure that we did not sample contiguous Metropolitan Statistical Areas that were effectively a single market, only 1 primary Metropolitan Statistical Area per consolidated Metropolitan Statistical Area was selected. Primary Metropolitan Statistical Areas are the individual metropolitan areas that make up consolidated Metropolitan Statistical Areas. New Orleans was in our original sample, but after Hurricane Katrina, data collection was suspended in this market (after 7 employers had been contacted) and San Antonio was substituted for the remainder of the study; as a result, a total of 41 Metropolitan Statistical Areas are within our sampling frame.

We asked Dun & Bradstreet to provide us with a list of the top 50 employers by local employment in each selected market, from which we attempted to identify the largest 26 that met our selection criteria. We excluded federal agencies (because they offer benefits through the Federal Employees Health Benefit Program) and companies that appeared more than twice in our sample. We also excluded redundant state or local agencies that purchased through a common public commission; that is, if there were several municipalities, school districts, and state agencies whose health insurance was purchased by a single public or quasi-public agency, we counted the agency only once and attributed to it all of the employees it covered. In some markets, these exclusions left us with fewer than 26 employers, so we supplemented Dun & Bradstreet data with information from local chambers of commerce and public agencies (45 companies were added this way). Our sample included the largest 26 private and public employers in each of the selected markets, with the exception of New Orleans and San Antonio, in which we sampled 7 and 20 employers, respectively. Any employer that reported fewer than 50 employees in the market sampled was considered ineligible for the survey. The study protocol was reviewed by the Harvard School of Public Health Institutional Review Board and determined to be exempt from human subjects oversight. Specific survey content is described below.

Respondents from firms that had employees in more than 1 Metropolitan Statistical Area were asked to answer the survey for each market in which they had been sampled. To reduce respondent burden, no corporation was included in the sample more than twice. In cases in which a corporation appeared more than twice in the original sample, we selected the 2 markets in which they had the largest local employment. Survey respondents were directors of benefits, vice presidents of human resources, or other high-level personnel in charge of contracting for and administering health benefits. The survey, which averaged 30 minutes to complete, was administered by telephone by trained survey researchers between July 2005 and December 2005. The general strategy used was to first fax 4 pages of information (cover letter, fact sheet, 2 pages of sample questions) to each respondent for review. Subsequently, an interview was pursued with the appropriate individual, with a relatively low level of telephone contact (1-3 calls per week) for a prolonged period (8 weeks or more). Toward the end of the field period, researchers increased the frequency of telephone activity for all remaining nonrespondents. In November 2005, internal project staff began calling refusal cases with potential of producing an interview and continued to interview respondents until March 2006.

We developed our survey instrument to ascertain the use of key components of value-based purchasing directed at health plans, physicians and medical groups, and hospitals. We also elicited information on participation in coalitions such as the Leapfrog Group and efforts to engage employees directly in health improvement efforts such as health risk assessment and promotion programs. Specific survey content is described below.

To establish the characteristics of our sample with regard to employment and health benefit purchasing, we first asked for the total number of employees within the selected market (at all locations within the market) and the number of current employees for which medical benefits were purchased, the percentage of the firm's local workforce in unions, and the percentage of employees in self-insured arrangements. We then asked a series of questions to ascertain the perceived importance of specific health plan characteristics when plans were selected to offer to employees. These included premium rates and expected cost of care, National Commission for Quality Assurance accreditation, quality of customer service provided, performance on clinical quality or patient satisfaction measures, efforts to reduce racial/ethnic disparities in care, geographic distribution of the network, and flexibility in plan design. The level of importance was captured with a 4-point scale: a lot, a moderate amount, a little, or none at all.

We then asked about value-based purchasing activities focused on health plans. We requested information about data collection on clinical quality or patient satisfaction indicators (Health Employer Data and Information Set and Consumer Assessment of Health Plans Study), the use of bonuses or penalties in health plan contracts, dissemination of collected information through employee report cards, whether employee plan contributions were adjusted to encourage selection of higher-quality plans, and efforts to work with health plans in selecting projects to foster quality improvement. Next, information was elicited on the number of plans offered, number of health plan carriers contracted with, and participation in a purchasing coalition or any national or regional formal collective quality-improvement efforts (eg, Leapfrog Group, Bridges to Excellence).

With regard to efforts focused on providers, we asked about strategies to improve the quality of physician and hospital services. For physicians, we asked whether employers examined data on quality of care and, if so, how these data were used. In particular, we documented whether quality performance data were reported to employees, used to tier co-payments, or used for performance rewards provided directly to physicians. We also asked whether the employer was subsidizing information technology infrastructure for physicians in the local market. In parallel, we asked about value-based purchasing initiatives targeted at hospitals, including Leapfrog membership and use of Leapfrog standards (ie, intensivist staffing of intensive care units, computerized physician order entry, or evidence-based referral for volume-sensitive procedures), nurse staffing ratios, and other measures to evaluate and reward hospitals or define hospital tiers. Employers were also asked about programs to engage and support employees in improving their own health. These programs include disease management, health promotion and education programs, health risk appraisals, and on-site health care services. Finally, we asked respondents for their views on the usefulness of the specific value-based purchasing approaches described above for improving the quality of health care for workers. Responses were captured with a 4-point scale, ranging from “very useful” to “not useful.”

Employer characteristics and reported use of value-based purchasing strategies were summarized and examined with employee-weighted frequencies. For ease of exposition, we report in the text the percentage of employers, but in the tables, percentages are weighted by employment. Subjective views on value-based purchasing were summarized without weighting by firm size because a single respondent's views may not be ascribed to the employer as a whole. For viewpoints on the importance of health plan characteristics, we computed and report the percentage of respondents indicating either “a lot” or “a moderate amount” because the patterns of responses for these 2 categories were consistently related across the items to be rated. For viewpoints on the usefulness of individual value-based purchasing strategies, we report the percentage of employers that responded “very useful” because the other 3 categories accounted for relatively consistent shares of the residual.

According to previous research^16-18 and the theory that because of fixed costs associated with many value-based purchasing activities there would be economies of scale, we focused on employer size as the key predictor of activity. Employer size was defined by the number of employees at all sites within the relevant market: small, 51 through 1000; midsized, 1001 to 5000; and large, greater than 5000. The statistical significance of size as a predictor of value-based purchasing was determined with multivariate analysis, in which indicators for region, percentage unionized, and percentage self-insured were included, in addition to indicator variables for size. Adjusted frequencies by firm size were computed as the predicted values from the multivariate analyses; 95% confidence intervals (CIs) were constructed with bootstrapping techniques. Analyses were conducted with SUDAAN (version 8.01; Research Triangle Institute, Research Triangle Park, North Carolina) to enable weighting observations by employees and account for clustering at the market level. P values for the employer size indicator variables collectively were calculated for each value-based purchasing strategy, and P < .05 was considered statistically significant.

The sample consisted of 1041 firms across 41 markets. Of this sample, 79 were found to be ineligible because they had fewer than 50 employees for which health benefits were purchased (47), were owned and controlled by companies already represented twice in the sample (22), or no longer had operations in the market area specified (10). Eligibility was undetermined for 165 respondents who did not pick up or return telephone calls and 7 firms for which we were unable to obtain valid contact information for a potential respondent. Of the 790 firms determined to be eligible for the survey, 181 refused to participate, yielding an interview completion rate of 77% and an overall response rate of 64%, according to Council of American Survey Research Organizations standards. Nonrespondents did not differ from respondents in terms of size (P = .18) or region of the country (P = .81). Among the completed surveys, there were 20 firms represented in 2 markets.

According to reported local employment, 103 survey respondents, representing 1% (95% CI, 1%-8%) of employees, were from firms with 1000 or fewer employees; 281 respondents, representing 14% (95% CI, 11%-18%) of employees, were from firms with 1001 to 5000 employees; and 225 respondents, representing 84% (95% CI, 81%-88%) of employees, were from firms with more than 5000 employees (Table 1). The majority of employers (510; 88% [95% CI, 83%-93%]) reported offering self-insured plans to employees, with 248 (30%; 95% CI, 22%-37%) reporting only self-insured arrangements. Approximately one-third of the sample reported 50% or more of its workforce in unions, whereas roughly another third reported no unionized employees. Most employers offered a choice of plan, although 244 (24%; 95% CI, 16%-32%) contracted with a single carrier. Preferred provider organization plans were offered by the greatest number of employers (510; 90% [95% CI, 86%-94%]), followed by HMO plans (444; 87% [95% CI, 83%-91%]), indemnity (137; 24% [95% CI, 14%-34%]), and high-deductible plans (124; 18% [95% CI, 12%-24%]). Sixty-eight percent of employers offered preferred provider organization and HMO health plan options (data not shown).

Approximately 85% (95% CI, 76%-93%) of respondents, regardless of firm size, perceived geographic distribution and premium rates to be important characteristics when choosing health plans to offer employees (Table 2). Flexibility in plan design and customer service were also rated as highly important, although this pattern varied by employer size. Characteristics more directly reflective of the quality of clinical care (Health Employer Data and Information Set/Consumer Assessment of Health Plans Study scores and National Commission for Quality Assurance accreditation) were reportedly given far less weight during health plan selection.

Sixty-five percent (269; 95% CI, 57%-74%) of surveyed employers reported examining data on clinical or patient satisfaction indicators for health plans (Table 3). Among those who reported considering such data, however, few reported disseminating this information to employees in the form of report cards (71; 23% [95% CI, 13%-33%]), using it to determine bonuses or penalties in plan contracts (49; 17% [95% CI, 7%-27%]), or using it to differentiate employee premium contributions to encourage enrollees to choose higher-quality plans (32; 9% [95% CI, 4%-14%]). Only a minority, 95 (28%; 95% CI, 18%-38%), reported working with health plans to select quality improvement projects. Finally, roughly one-fifth of employers reported participation in a local or regional health care purchasing coalition. Controlling for other factors, employer size was a significant predictor of reported adoption for each tactic except participation in a purchasing coalition (Table 3), with larger employers being more engaged in value-based purchasing efforts.

Only 71 respondents (16%; 95% CI, 9%-23%) reported examining performance data at the physician or medical group level, 34 (8%; 95% CI, 3%-12%) reported offering incentives for improving information technology infrastructure, and 23 (8%; 95% CI, 3%-13%) reported providing employees with provider report cards (Table 4). Approximately 36% (95% CI, 19%-54%) of employers reported that they were members of the Leapfrog Group or used the Leapfrog quality standards for contracting or informing employees. On average, 5% (95% CI, 2%-8%) of employers reported using the Leapfrog measures for adjusting hospital co-payments or rewarding hospitals (data not shown), 42 employers (17%; 95% CI, 7%-27%) reported evaluating nurse staffing ratios at local hospitals, and 63 employers (18%; 95% CI, 10%-25%) reported examining other quality measures used to evaluate hospitals. Employer size was a significant predictor of examining and reporting physician and medical group quality data (P = .02 and P = .006, respectively) and Leapfrog membership (P = .002) and evaluating nurse staffing ratios of hospitals (P = .01). Finally, 147 respondents (40%; 95% CI, 30%-50%) reported participating in formal collective quality improvement efforts.

The majority of respondents (446) reported adopting strategies to influence employee health through disease management programs (83%; 95% CI, 76%-89%) and health promotion programs (437; 70% [95% CI, 59%-82%]) (Table 5). Seventy percent (423; 95% CI, 59%-80%) of responding employers reported providing a clinical help line to employees. Of the 293 (56%; 95% CI, 46%-67%) responding employers reporting provision of health risk appraisals to employees, 121 (21%; 95% CI, 14%-29%) reported offering an incentive for completing the health risk appraisal. Nearly half of surveyed employers reported providing on-site screening or treatment to their employees. Employer size was positively and significantly associated with reporting a health risk appraisal (P = .03), health promotion and education programs (P = .003), and on-site screening and treatment programs (P = .008).

When respondents were asked to assess the usefulness of selected quality improvement strategies asked about in the survey, with the exception of using Health Employer Data and Information Set/Consumer Assessment of Health Plans Study scores in plan selection (17%; 95% CI, 14%-20%), all were considered very useful by at least a quarter of respondents (Table 6). Sharing physician or hospital quality scores with consumers (42%; 95% CI, 40%-44%), use of financial incentives for consumers to choose higher-quality plans or providers (36%; 95% CI, 31%-42%), and sharing health plan scores with consumers (36%; 95% CI, 31%-41%) were considered very or modestly useful by most employers surveyed.

Despite the increasing visibility of national efforts such as those of the Leapfrog Group to catalyze large employers and other purchasers of care, our findings suggest that many large employers are not using their purchasing power with health plans and providers to improve the quality of health care received by their employees. Although a significant number of surveyed employers reported examining health plan quality data, many attach a low level of importance to it, and they infrequently report using it for performance rewards or to influence employees. Moreover, our data suggest provider quality information is even less commonly examined or used by employers to reward provider performance or influence employee choice of providers.

Although overall patterns of value-based purchasing suggest limited deployment, larger employers in our sample reported more activity in all areas than smaller employers reported. This result supports the theory that the fixed costs associated with, for example, developing a health plan report card are a barrier to wider adoption of these methods. That is, the perceived benefits of measuring quality and using that information for contracting, employee education, and benefit design might be simply too small for employers of modest size to justify the cost. If this hypothesis is true, then coalition efforts may have a critical role to play. Approximately one-fifth of employers reported belonging to a purchasing coalition, which may undertake many of the strategies that we asked about on their behalf. Moreover, 40% of employers reported that they participate in national or regional coalition efforts to improve quality. These collective efforts, which may ultimately prove more effective than a piecemeal alternative, could become the major means by which employers seek to influence health plans and providers.

Although respondents from larger employers report being more involved in value-based purchasing, their reported adoption of these strategies is still limited; why is value-based purchasing not more widespread among very large firms? Conceivably, these large employers believe that the benefits from value-based purchasing do not justify the cost or that other entities are better positioned to undertake these activities. Our study suggests that skepticism about the benefits of value-based purchasing may be important because only about one-third of employers viewed each value-based purchasing strategy we asked about as “very useful.” This perception may be due to the lack of a “business case” for the intended outcomes of value-based purchasing in terms of the effects on workforce productivity, benefit cost savings, or the ability to attract and retain employees. Alternatively, some employers may (correctly) perceive that the evidence to support the effectiveness of strategies such as pay for performance and report cards is mixed at best.

If high fixed costs and questions about benefits deter employers from pursuing certain components of value-based purchasing, they are still well situated to promote and support desirable employee health behavior. Indeed, respondents reported substantial use of employee-centered quality improvement strategies, including health risk appraisals, disease management, and the provision of on-site health care services. The widespread emphasis on influencing employee health behavior found in our survey is consistent with that found in other recent reports.¹⁹ It remains to be seen, however, which of these strategies are effective and whether employers will realize gains from reduced premiums, improvements in labor productivity, or other advantages such as employee loyalty.²⁰

Although our study focusing on value-based purchasing is the first that we are aware of during this decade, it nevertheless builds on previous work. For example, previous studies have documented^17,21,22 that employers have not put much weight on measures of quality in selecting health plans. Our study shows that this has not changed despite the expanding efforts nationally in measuring health plan quality. We are unaware of any previous efforts to document nationally employers' efforts in value-based purchasing focused on physicians and medical groups and hospitals.

We underscore several study strengths and limitations. Our data reflect 2005 activity and to our knowledge represent the only systematic national survey of value-based purchasing by employers this decade. We focused on metropolitan areas with a minimum of 100 000 total HMO enrollees because of our previous assumptions that value-based purchasing activity would be focused in markets with at least some minimum of HMO penetration. As such, our results may not generalize to nonmetropolitan areas or markets with lower levels of HMO activity. Similarly, our sample selection approach identified the largest employers in the areas we studied, and our findings do not likely generalize to smaller employers. Given the size gradient that we found in reported value-based purchasing activity, however, we surmise that there is little value-based purchasing among smaller employers. We documented reported participation in purchasing coalitions but not what these coalitions do; thus, our data pertain only to direct efforts in value-based purchasing by employers. Finally, the validity of self-reported data, particularly for the subjective aspects of the study, may reasonably be questioned.

During the last decade, a number of high-profile employers have become involved in delivery system reform and quality improvement in national and local spheres. Nonetheless, our findings suggest that employers as a whole do not appear to be directly implementing contracting strategies and programs to improve the quality and value of health benefits, except as they relate to supporting improved employee health behavior. Efforts to alter the dynamics of health plan and provider competition will likely have to come from other sources, including private employer coalitions, multistakeholder collaborative organizations, and the public sector. At the national level, the Centers for Medicare & Medicaid Services is collecting and disseminating quality information and is proceeding toward implementing pay for performance. Because Medicare is the largest single purchaser of health care services, many believe that it can and should effectively lead value-based purchasing.²³ Likewise, some states such as Pennsylvania have exerted their regulatory authority to require public reporting of quality and patient safety information for hospitals. In the private sector, organizations such as Care Focused Purchasing plan to take advantage of economies of scale in data collection and quality measurement and reporting, as well as the statistical advantages of all-payer data analysis. It remains to be seen whether these large-scale quality measurement and reporting efforts can yield their full potential benefit if employers—the sponsors of insurance for 54% of the US population²⁴—do not begin using such information in health plan contracting, benefit design, or other ways that might stimulate quality improvement.

Corresponding Author: Meredith B. Rosenthal, PhD, Department of Health Policy and Management, Harvard School of Public Health, 677 Huntington Ave, Boston, MA 02115 (mrosenth@hsph.harvard.edu).

Author Contributions: Dr Rosenthal had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Rosenthal, Landon, Normand, Frank, Epstein.

Acquisition of data: Rosenthal, Landon, Ahmad, Epstein.

Analysis and interpretation of data: Rosenthal, Landon, Normand, Frank, Ahmad, Epstein.

Drafting of the manuscript: Rosenthal, Normand, Ahmad.

Critical revision of the manuscript for important intellectual content: Rosenthal, Landon, Normand, Frank, Ahmad, Epstein.

Statistical analysis: Rosenthal, Normand, Frank, Ahmad.

Obtained funding: Rosenthal, Normand, Frank, Epstein.

Administrative, technical, or material support: Ahmad, Epstein.

Study supervision: Rosenthal, Ahmad.

Funding/Support: Financial support for this project was provided by the Agency for Healthcare Research and Quality (R01 HS 13335).

Role of the Sponsor: The Agency for Healthcare Research and Quality had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript.