Optimizing Enrollment in Employer Health Programs: A Comparison of Enrollment Strategies in the Diabetes Health Plan

September 4, 2014
Lindsay B. Kimbro, MPP
Lindsay B. Kimbro, MPP

,
Jinnan Li, MPH
Jinnan Li, MPH

,
Norman Turk, MS
Norman Turk, MS

,
Susan L. Ettner, PhD
Susan L. Ettner, PhD

,
Tannaz Moin, MD, MBA, MSHS
Tannaz Moin, MD, MBA, MSHS

,
Carol M. Mangione, MD
Carol M. Mangione, MD

,
O. Kenrik Duru, MD, MSHS
O. Kenrik Duru, MD, MSHS

Volume 20, Issue 8

An automatic enrollment strategy for health insurance programs may not only increase the total number of enrollees but may also decrease some enrollment disparities.

Background

Many health programs struggle with low enrollment rates.

Objectives

To compare the characteristics of populations enrolled in a new

health plan when employer groups implement voluntary versus

automatic enrollment approaches.

Study Design

We analyzed enrollment rates resulting from 2 different strategies:

voluntary and automatic enrollment. We used regression

modeling to estimate the associations of patient characteristics

with the probability of enrolling within each strategy. The subjects

were 5014 eligible employees from 11 self-insured employers who

had purchased the Diabetes Health Plan (DHP), which offers free

or discounted copayments for diabetes related medications, testing

supplies, and physician visits. Six employers used voluntary

enrollment while 5 used automatic enrollment.

The main outcome of interest was enrollment into the DHP.

Predictors were gender, age, race/ethnicity, dependent status,

household income, education level, number of comorbidities, and

employer group.

Results

Overall, the proportion of eligible members who were enrolled

within the automatic enrollment strategy was 91%, compared

with 35% for voluntary enrollment. Income was a significant predictor

for voluntary enrollment but not for automatic enrollment.

Within automatic enrollment, covered dependents, Hispanics,

and persons with 1 nondiabetes comorbidity were more likely to

enroll than other subgroups.

Employer group was also a significant correlate of enrollment.

Notably, all demographic groups had higher DHP enrollment

rates under automatic enrollment than under voluntary

enrollment.

Conclusions

For employer-based programs that struggle with low enrollment

rates, especially among certain employee subgroups, an automatic

enrollment strategy may not only increase the total number

of enrollees but may also decrease some enrollment disparities.

Am J Manag Care. 2014;20(8):e311-e319

The Diabetes Health Plan included variation in enrollment strategy across employer groups, with some using voluntary enrollment and others using automatic enrollment.

• Utilizing voluntary enrollment, 35% of eligible employees were enrolled, and annual income ≥$125,000 and black race were associated with much higher rates of enrollment.

• Utilizing automatic enrollment, 91% of eligible employees were enrolled, and dependents and employees of Hispanic ethnicity were somewhat more likely to be enrolled than other employees.

• Automatic enrollment strategies can increase overall enrollment and provide participants with the opportunity to “opt out,” and may also decrease some enrollment disparities.

Despite extensive recruitment efforts by health

plans, state and local governments, and other

stakeholders, many eligible individuals do not

voluntarily enroll in health promotion or insurance benefit

programs designed to improve health outcomes.1 Employers

are increasingly sponsoring wellness programs as a way to

possibly decrease costs and increase productivity across a

large component of the workforce.2 However, despite the

use of various approaches, enrollment in wellness programs

often remains low.3-5 Although many of these programs and

benefits may improve access and outcomes among the subset

of persons who are enrolled, with limited reach they

are unlikely to improve the health of the overall targeted

population.6

Many employer health programs use a voluntary enrollment

approach, which employees must actively join in order

to be enrolled. However, voluntary program enrollees may

have different demographic characteristics than the underlying

population, in terms of gender, age, race/ethnicity,

income, risk for chronic conditions or disability, and other

factors.7 Voluntary program enrollees may also have different

clinical characteristics than the underlying population,

potentially representing either the “worried well” who may

have less need for services or a sicker subgroup motivated

to enroll because of the severity of their underlying condition.

A recent review of enrollment into a variety of public

benefit programs identified multiple barriers to voluntary enrollment

and suggested automatic enrollment of all eligible

participants as a preferential strategy.8 There is little current,

“real-world” data on patient-level differences comparing

“voluntary” and “automatic” enrollment approaches. Such

information may be useful in the design of future health promotion

or insurance benefit programs.

Data from the rollout of the SHP at 11 self-insured employers

of the Diabetes Health Plan (DHP)—the first disease-specific

health insurance plan for employees and their covered

dependents with diabetes or prediabetes—provides a unique

opportunity to assess the effectiveness of these 2 enrollment

strategies. The DHP is offered by different employer

groups, some using a voluntary enrollment approach requiring

employees to sign up, others an automatic enrollment

approach that directly enrolls all eligible employees.

We hypothesized that the automatic enrollment strategy

would enroll a larger and more representative sample of

the underlying population, compared with the voluntary

enrollment strategy.

METHODS

Study Design, Setting, and Participants

The Diabetes Health Plan (DHP), initiated in 2009, represents an innovative approach to care for individuals

with diabetes or prediabetes.9 Purchased by several medium

and large self-insured employers across the United

States, the DHP eliminates or reduces co-payments for

medications and physician visits in order to incentivize

evidence-based care. Eligible employees and their eligible

covered dependents have the option of maintaining their

standard plan or switching to a DHP plan. The latter adds

DHP benefits to the standard plan while maintaining the

same premium cost to the employee. The DHP also includes

enhanced access to wellness programs at no additional

Table 1

cost to the employee. shows the variations

in features between the DHP and the standard plan.

In addition to these program benefits, the DHP was

originally designed by the health plan to include several

requirements to be met each year in order to maintain

enrollment for the following year. These “compliance

criteria” were ultimately determined by each employer,

but potentially included a combination of the following:

laboratory evaluations such as biannual A1C testing, annual

cholesterol blood testing and/or annual microalbuminuria

screening, biannual primary care visits, annual

retinal exams, biannual mammography, and/or colon

cancer screening for persons aged 50 years or more. Required

tests were offered free to the enrolled DHP member.

Although the DHP enrolled both

employees with diabetes and prediabetes,

the current analysis is limited

to the sample with diabetes. In order

to be considered eligible for the DHP,

an employee (or dependent) with diabetes

had to meet at least 1 of the following

criteria during the prior 1 year:

(1) 1 or more medical claims with an

International Classification of Diseases,

Ninth Revision, Clinical Modification

ICD-9-CM

() diagnosis code of 250.xx from a doctor’s office,

clinic visit, or inpatient hospital stay; (2) any glycated

hemoglobin (A1C) value of ≥6.5%, fasting plasma glucose

>125 mg/dL, or 2-hour oral glucose tolerance test ≥200

mg/dL; (3) any prescription filled for insulin or an oral

antiglycemic agent other than metformin; (4) direct referral

from a medical provider or as a result of onsite biometric

screenings.

When the DHP was first introduced in early 2009, all

participating employer groups offered the plan under a

voluntary enrollment strategy. Some employers limited

DHP eligibility to persons who had existing diagnoses of

diabetes and these persons could voluntarily enroll. Other

employers offered on-site biometric screenings to detect

new cases of diabetes, and allowed those with either new or existing diagnoses to voluntarily enroll.

Employer groups initially offering the DHP in late 2009

and 2010 had the option to enroll employees using an

automatic enrollment approach. Each employer identified

the eligible employees based on the criteria described

above and notified them of their eligibility. Eligible individuals

were automatically enrolled in the DHP at the beginning

of the next enrollment period unless they made

an active decision to opt out in favor of having a standard

health plan. The opt-out process was relatively simple

for individuals who preferred to remain in the standard

health plan, usually involving a short form that could be

returned to their designated DHP representative.

Using a cross-sectional design, we compared the 6 employer

groups that offered voluntary enrollment and the

5 groups that used automatic enrollment to evaluate differences

in an employee’s likelihood of DHP enrollment.

Figure

As shown in , we restricted the sample to persons

who: (1) had no gestational diabetes; (2) were continuously

enrolled with the health insurer 1 year before and 1

year after the DHP was offered; (3) were between the ages

of 18 and 64 years at baseline; (4) were not missing key

demographic variables or information on employee/dependent

status; and (5) had enrollment status in the DHP

Variables

The outcome variable for this analysis was enrollment in the DHP. Enrollment data was provided by the health insurer, and individual-level information about reasons why employees did or did not enroll was not available. Predictor variables for this analysis included gender, age group (aged 18-34, 35-44, 45-54, and 55-64 years), and employee vs covered dependent (>18 years) status, all of which were member-reported and were acquired from the eligibility file provided by the health insurer. Other variables included education level (high school graduate or less, some college, bachelor’s degree, and above), race (white, Hispanic, black, Asian, other), and annual household income (<$30K, $30K-$49K, $50K-$74K, $75K- $124K, ≥$125K), all of which were obtained by the health insurer from a third-party consumer marketing services firm that derived them from a combination of census data, an algorithm analyzing first and last names, and an income database. A count of comorbidities was derived from administrative claims data provided by the health insurer; it included each of 15 conditions based on ICD-9-CM codes: hypertension, hyperlipidemia, coronary artery disease, congestive heart failure, atrial fibrillation, end-stage renal disease, osteoarthritis, rheumatoid arthritis, cancer, chronic obstructive pulmonary disease, stroke, peripheral vascular disease, dementia, and schizophrenia and other mental health diagnoses (eg, depression, anxiety).

Statistical Analysis

We compared the unadjusted differences in enrollment within the voluntary and automatic enrollment groups. Using a multivariate logistic regression model to control for demographic and health variables, we determined the marginal predicted probabilities of being enrolled in the DHP. In addition to controlling for various demographic characteristics, we also included “enrollment method” to control for those who were offered the plan under voluntary versus automatic enrollment.

Finally, we conducted the same analysis using employer fixed effects with stratified models to compare the associations estimated among employer groups who offered voluntary enrollment with those who offered automatic enrollment. We chose this specification because of the inherent flexibility, as fixed effects control for any confounding of patient-level effects with employer characteristics, and stratification allows enrollment strategy to fully interact with the other predictors in the model.

RESULTS

None of the employer-level characteristics were significantly different between the automatic and voluntary enrollment groups (Table 2). Of persons meeting our study criteria (aged between 18 and 64 years and continuously enrolled in a UnitedHealth plan for 2 years), 8.7% had diabetes. Of the 1549 eligible persons in the voluntary enrollment group, only 35% enrolled in the DHP, by opting into the program. Of the 3465 persons in the automatic enrollment group, 91% enrolled in the DHP, by not opting out. Chi-square tests revealed significant unadjusted differences by race/ethnicity in the voluntary enrollment groups, with a higher percentage of white and Hispanic individuals and a lower percentage of black individuals represented among those enrolled (Table 3). There were also differences by income and education in the voluntary enrollment groups, with a higher percentage of individuals with annual household income ≥$75K, a lower percentage with a high school diploma or less, and a higher percentage with a bachelor’s degree enrolled in the DHP compared with the sample that did not enroll. Among the automatic enrollment groups, there was a higher percentage of men among DHP (Table 3). There was also a higher percentage of Hispanic individuals among the enrolled as compared with the nonenrolled.

Within the pooled regression controlling for demographics and enrollment strategy, we found that individuals within an automatic enrollment setting were 58 percentage points more likely (P <.01) to enroll than those in a voluntary enrollment group. In the stratified adjusted analyses with all predictor variables simultaneously included (Table 4), within voluntary enrollment groups we found that black individuals were more likely to be enrolled in the DHP (+8 percentage points, P = .01) compared with white individuals. We also found that covered dependents were less likely to be enrolled in the DHP than employees (—10, P <.001), and individuals with annual household incomes of ≥$125K were more likely to be enrolled in the DHP than individuals with incomes of under $30K (+17, P = .04). Individuals in the aged 45-54 years group were also more likely to be enrolled in the DHP (+10, P = .03) compared with individuals between 18 and 35 years of age. Examining the automatic enrollment groups, we found no significant differences by patient income or education, but Hispanics were more likely to remain enrolled in the DHP than white individuals (+5, P <.001), and covered dependents were more likely to remain enrolled in the DHP than employees (+2, P =.02). Individuals aged between 55 and 64 years were less likely to remain enrolled in the DHP (–6, P = .02) compared with individuals aged between 18 and 35 years.

Finally, statistically significant differences in enrollment by employer group were observed within both the voluntary and automatic enrollment groups. In particular, rates of DHP enrollment among employers using the voluntary enrollment approach varied from 14% to 88%.

DISCUSSION

In summary, enrollment rates varied within the groups of employers using voluntary and automatic enrollment approaches, as well as between employers using voluntary enrollment and those using automatic enrollment approaches. In the voluntary enrollment groups, black individuals and high-income individuals were more likely, and covered dependents less likely, to “opt in.” In the automatic enrollment groups, Hispanic individuals and covered dependents were less likely to “opt out.” We also observed significantly higher rates of enrollment acrossall subgroups in the automatic enrollment approach as compared with voluntary enrollment.

The 2 enrollment strategies that we compared require very different levels of patient engagement and initiative. In workplaces offering voluntary enrollment, eligible individuals must take personal initiative to enroll. They must first become aware of the program and then proceed through the proper channels or complete tasks required for enrollment (ie, contact the appropriate representative to request an application form for enrollment, then complete and return the form). However, employees who are automatically enrolled are only required to take any action if they choose not to participate.

In addition, the reasons for not being enrolled under each strategy are likely very different. There are numerous potential barriers to entry in a voluntary enrollment system, which may include poor communication about the program, lack of understanding of the program, and/ or the opportunity cost of the time associated with the enrollment process. Within an automatic enrollment system, employees who choose to opt out may do so because they have an existing competing insurance plan, are insured under another family member’s plan, or for another financial or health reason.

We found that covered dependents were significantly less likely than employees to be enrolled within voluntary enrollment, but were significantly more likely to be enrolled under the automatic enrollment strategy. Covered dependents were required to meet the same eligibility requirements as eligible employees. It is possible that covered dependents were less likely to be aware of the DHP and voluntarily enroll, since they may not have received promotional communications distributed at the workplace. On the other hand, covered dependents may be less likely to have typical “opt out” reasons such as a better benefit through a spouse or a choice of a different health insurance plan. Therefore, they may be less likely to opt out under automatic enrollment.

We also found that affluent individuals (annual household incomes of $125K or more) were much more likely to voluntarily enroll in the DHP than individuals with annual household incomes of less than $30K. Co-payment reductions are more likely to influence adherence among individuals for whom the out-of-pocket cost of medications is a greater burden.10 To the extent that programrelated resources are disproportionately devoted to higher income groups, the DHP goal of reduction in cost-related nonadherence may be less pronounced with use of a voluntary enrollment approach.

Among racial/ethnic groups, we found that Hispanics were least likely to make an “active” enrollment choice about their health insurance by opting out. This is consistent with prior studies showing lower levels of initiating use of outpatient health services by Hispanics compared with other racial/ethnic groups.10-11 Language barriers or beliefs about healthcare have been listed as possible causes of these differences.12-13 An automatic enrollment approach may be one way to overcome racial disparities in employer health program participation. Research has shown that the type of health insurance an individual is offered has the strongest effect on healthcare utilization among Hispanics, as rates of preventive care used among Hispanics were much greater among those enrolled in HMO as opposed to fee-for-service plans, suggesting that co-payment or coinsurance costs may be a major deterrent for seeking preventive care.15 Since the DHP, is designed to minimize or eliminate co-payments, this type of value-based benefit feature combined with an automatic enrollment approach could potentially increase use of both diabetes-specific services (eg, routine A1C checks and retinal exams) as well as general preventive services (eg, mammograms).

The strongest determinant of enrollment rates within both voluntary and automatic enrollment groups was the employer. We believe that these differences may be driven by variation in marketing the plan to employees, use of financial incentives for participation at the employee level, and implementation of compliance criteria. Although the study team did not collect this information in a standardized manner, communications with the health plan design team indicate that employers with the highest voluntary enrollment rates tend to be those that offer multiple wellness programs or incentives, and have designated wellness champions or wellness committees who take an active role in decisions about health programs.

Our analysis has 2 notable limitations. First, no small or medium-sized employers (<1000 employees) purchased the DHP so the analysis is limited to large employers. We are therefore unable to generalize these results to smaller companies. However, because of this, the results are unlikely to be affected by changes caused by the Affordable Care Act, as these individuals eligible for the DHP were with larger companies already offering insurance. Secondly, potential ceiling effects may limit the ability of our analyses to detect differences in enrollment rates with the automatic enrollment strategy.

The first and arguably most important barrier to access for any health plan or health program is enrollment of eligible individuals. Our findings of increased enrollment of blacks and higher income individuals with voluntary enrollment, as well as a higher probability of remaining enrolled for Hispanics and covered dependents with automatic enrollment, may help inform future policies that involve employer health programs. It is important to note that employees who are automatically enrolled may be less likely to fully engage with the benefits and features available, which may diminish the overall effectiveness of the program. The administrative burden of including these less engaged individuals in a health program may present an excessive burden in certain cases. Conversely, programs that use a voluntary enrollment approach are likely to have more engagement among participants but will likely need to make a large upfront investment in time and financial resources to drive enrollment.

Forthcoming analyses will evaluate the effectiveness of the DHP in terms of key outcomes such as control of cardiovascular risk factors, utilization of care, and total costs. However, based on previous research, we expect that lower cost-sharing applied across entire populations will enhance medication adherence and may also drive these more distal outcomes.17-18 In a quickly evolving health policy environment, innovative ideas and a strong push in the direction of wellness and population management will likely result in millions of dollars being spent on new health promotion programs. If these programs enroll only small and unrepresentative proportions of targeted individuals, it will be very difficult not only to evaluate the likely impact on the larger population, but also to disseminate effective programs to a broad spectrum of eligible individuals. An automatic enrollment approach may prove critical in overcoming entrance barriers that hinder participation in health promotion programs that may ultimately decrease costs and lead to better health outcomes.4-6,19

Author Affiliations: Department of Medicine, David Geffen School of Medicine at University of California, Los Angeles (LBK, JL, NT, SLE, CMM, OKD); Department of Health Policy and Management, Jonathan and Karin Fielding School of Public Health at UCLA, Los Angeles (CMM, SLE); and Department of Medicine and HSR&D Center of Excellence for the Study of Healthcare Provider Behavior, Veterans Affairs Greater Los Angeles Healthcare System, Los Angeles (TM).

Funding Source: Funding received from the Centers for Disease Control and Prevention and the National Institute of Diabetes and Digestive and Kidney Diseases as part of the Natural Experiments for the Translation of Diabetes (NEXT-D) Study (Grant number DP002722). Dr Moin is supported by VA Office of Academic Affiliations, Health Services Research and Development, through the Health Services Fellowship Training Program (TPM65-010), VA Greater Los Angeles Healthcare System. Dr Mangione and Dr Duru are supported in part by the University of

California, Los Angeles, Resource Centers for Minority Aging Research Center for Health Improvement of Minority Elderly (RCMAR/CHIME) under NIH/NIA Grant P30-AG021684. Dr Duru is supported in part by the Career Development award K08 AG033630.

Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (LK, OKD, SLE, CMM, JL, NT); acquisition of data (LK, OKD, CMM, JL, TM, NT); analysis and interpretation of data (LK, SLE, SMM, JL, TM, NT); drafting of the manuscript (LK,); critical revision of the manuscript for important intellectual content (LK, OKD, SLE, CMM, TM); statistical analysis (LK, SLE, JL); provision of study materials or patients (LK); obtaining funding (LK, OKD); administrative, technical, or logistic support (LK, CMM); and supervision (LK, OKD, CMM).

Address correspondence to: Lindsay B. Kimbro, MPP, 10940 Wilshire Blvd, Ste 700, Los Angeles, CA 90095. E-mail: lkimbro@mednet.ucla.edu.1. Cooper PF, Schone BS. More offers, fewer takers for employment-based health insurance: 1987 and 1996. Health Aff (Millwood). 1997;16(6):142—149.

2. Thomas B. Wellness in the Workplace 2012: An Optum Research Update. Eden Prairie, MN: Optum; 2012. www.optum.com.

3. Rula E, Sacks R. Incentives for health and wellness programs: strategies, evidence and best practice. Outcomes & Insights in Health Management. 2009;1(3):1-7.

4. Mattke S, Schnyer C, Van Busum K. A Review of the U.S. Workplace Wellness Market. Online only: Rand Corporation, 2012. www.rand.org/pubs/occasional_papers/OP373.html. Accessed July 30, 2014.

5. Diehr P, Madden CW, Cheadle A, Martin DP, Patrick DL, Skillman S. Will uninsured people volunteer for voluntary health insurance? experience from Washington state. Amer J Public Health. 1996;86(4):529-532.

6. Glasgow RE, Vogt TM, Boles SM. Evaluating the public health impact of health promotion interventions: the RE-AIM framework. Amer J Public Health. 1999; 89(9):1322-1327.

7. Terry PE, Fowles JB, Harvey L. Employee engagement factors that affect enrollment compared with retention in two coaching programs—the ACTIVATE study. Popul Health Manag. 2010;13(3):115-122.

8. Remler DK, Glied SA. What other programs can teach us: increasing participation in health insurance programs. Amer J Public Health. 2003;93(1):67-74.

9. Duru OK, Mangione CM, Chan C, et al. Evaluation of the diabetes health plan to improve diabetes care and prevention. Prev Chronic Dis. 2013;10:E16.

10. Andersen R, Lewis SZ, Giachello AL, Aday LA, Chiu G. Access to medical care among the Hispanic population of the southwestern United States. J Health Soc Behav. 1981;22(1):78-89.

11. Solis JM, Marks G, Garcia M, Shelton D. Acculturation, access to care, and use of preventive services by Hispanics: findings from HHANES 1982-84. Amer J Public Health, 1990;80 (Suppl);11-19.

12. Chesney AP, Chavira JA, Hall RP, Gary HE Jr. Barriers to medical care of Mexican-Americans: the role of social class, acculturation, and social isolation. Med Care. 1982; 20(9):883-891.13. Nall FC II, Spielberg J. Social and cultural factors in the responses of Mexican-Americans to medical treatment. J Health Soc Behav. 1967;8(4):299-308.

14. Guendelman S, Wagner TH. Health services utilization among Latinos and white non-Latinos: results from a national survey. J Health Care Poor Underserved. 2000;11(2):179-194.

15. Tseng CW, Tierney EF, Gerzoff RB, et al. Race/ethnicity and economic differences in cost-related medication underuse among insured adults with diabetes: the Translating Research Into Action for Diabetes study. Diabetes Care. 2007;31(2):261-266.

16. Fung V, Mangione CM, Huang J, et.al. Falling into the coverage gap: Part D drug costs and adherence for Medicare Advantage prescription drug plan beneficiaries with diabetes. Health Serv Res. 2010;45(2);355-375.

17. Kazerooni R, Bounthavong M, Watanabe JH. Association of copayment and statin adherence stratified by socioeconomic status. Ann Pharmacother. 2013;47(11):1463-1470.

18. Chernew M, Gibson TB, Yu-Isenberg K, Sokol MC, Rosen AB, Fendrick AM. Effects of increased patient cost sharing on socioeconomic disparities in health care. J Gen Intern Med. 2008;23(8):1131-1136.

19. Consensus Statement of the Health Enhancement Research Organization; American College of Occupational and Environmental Medicine; American Cancer Society and American Cancer Society Cancer Action Network; American Diabetes Association; American Heart Association. Guidance for a reasonably designed, employer-sponsored wellness program using outcomes-based incentives. J Occup Environ Med. 2012;54(7):889-896.