Effect of Outreach Messages on Medicaid Enrollment

Abstract

Objectives. To measure the impact of different outreach messages on health insurance enrollment among Medicaid-eligible adults.

Methods. Between March 2015 and April 2016, we conducted a series of experiments using mail-based outreach that encouraged individuals to enroll in Pennsylvania’s expanded Medicaid program. Recipients were randomized to receive 1 of 4 different messages describing the benefits of health insurance. The primary outcome was the response rate to each letter.

Results. We mailed outreach letters to 32 993 adults in Philadelphia. Messages that emphasized the dental benefits of insurance were significantly more likely to result in a response than messages emphasizing the health benefits (odds ratio = 1.33; 95% confidence interval = 1.10, 1.61).

Conclusions. Medicaid enrollment outreach messages that emphasized the dental benefits of insurance were more effective than those that emphasized the health-related benefits.

Public Health Implications. Although the structure and eligibility of the Medicaid program are likely to change, testing and identifying successful outreach and enrollment strategies remains important. Outreach messages that emphasize dental benefits may be more effective at motivating enrollment among individuals of low socioeconomic status.

As of July 2016, more than 72 million US persons were enrolled in Medicaid and the Children’s Health Insurance Program, an increase of 27% compared with before passage of the Patient Protection and Affordable Care Act (ACA).¹ Community-based enrollment specialists have played a key role in helping low-income adults enroll in Medicaid.² Despite outreach efforts, approximately 3.3 million eligible adults living in Medicaid expansion states are not enrolled.³ Many barriers to enrollment exist, including low health insurance literacy and not being aware that one is eligible.^4,5 Despite the uncertain future of the ACA, it is important to develop strategies to conduct outreach and to market health insurance benefits to this difficult-to-reach population.⁶ Enrollment in Medicaid is a key step toward improving uptake of preventive health services, reducing household financial stress, improving mental health, and addressing physical health care needs.⁷

One model to increase enrollment involves targeted outreach to individuals enrolled in other public benefit programs that may signal eligibility for Medicaid. In Philadelphia, Pennsylvania, Benefits Data Trust (BDT), a nonprofit organization dedicated to increasing access to public benefits, has conducted large-scale mail campaigns to increase enrollment in the Supplemental Nutrition Assistance Program (SNAP), in addition to Medicaid. Individuals are informed of their likely eligibility for a benefit and invited to contact BDT’s benefits hotline to receive eligibility screening and application assistance. As part of new health insurance outreach activities in 2015 to 2016, we tested the effectiveness of different messages to promote enrollment among Medicaid-eligible adults.

On the basis of preintervention qualitative interviews, we hypothesized that emphasizing the financial and dental benefits of insurance, both of which are frequently priorities for this population, would be more effective than messages focusing on the health benefits of health insurance.⁵ We also hypothesized that social norming (e.g., reporting how many people in the neighborhood are already insured) would increase responses by making enrollment normative.⁸ This technique has been successfully used on college campuses to decrease heavy alcohol intake, using a social marketing campaign to normalize lower consumption quantities.⁹

METHODS

We conducted a series of randomized controlled experiments of mail-based outreach messages that encouraged individuals to enroll in Pennsylvania’s expanded Medicaid program. In partnership with the Pennsylvania Department of Human Services, BDT identified adults who might be Medicaid-eligible on the basis of previous enrollment in the Low-Income Home Energy Assistance Program (the eligibility for which is ≤ 150% of the federal poverty level).¹⁰ BDT sent a 1-time outreach mailing asking recipients to call a benefits hotline for application assistance. In 4 sequential experiments (Table 1), we randomized individuals to 1 of 2 possible outreach messages. The primary outcome was the response rate to each letter. Using logistic regression, we determined the independent effects of each of these message components.

TABLE 1—

Summary of Experimental Design Testing Outreach Messages to Encourage Medicaid Enrollment: Philadelphia, PA, March 2015–April 2016

Experiment and Comparison	Description of Message
Experiment 1 (n = 8 998)
Health benefits of insurance	Narrative describing someone who by enrolling in health insurance avoided a serious health problem.
Financial benefits of insurance	Narrative describing someone who by enrolling in health insurance avoided a large financial cost from health care.
Experiment 2 (n = 11 997)
Health care benefits	Message emphasizing free or low-cost health coverage that can allow you to get regular health care and save you from having to pay for health care all on your own.
Dental care benefits	Message emphasizing free or low-cost dental care and dental programs that can help you get regular cleanings and allow you to address urgent tooth concerns and could allow you to see a dentist soon.
Experiment 3 (n = 2 399)
Health care benefits and social norming message	Health care message (experiment 2) that also included the number of people in the person’s neighborhood already signed up for health coverage.
Dental care benefits and social norming message	Dental care message (experiment 2) that also included the number of people in the person’s neighborhood already signed up for dental coverage.
Experiment 4 (n = 9 599)
Dental care benefits	Dental care message used in experiment 2.
Dental care benefits and social norming message	Dental care message used in experiment 3.

RESULTS

Outreach letters were mailed to 32 993 adults in Philadelphia between March 2015 and April 2016. In experiment 1 (health vs financial benefits of insurance), we found no significant differences in response rate between those who received the health benefits– versus the financial benefits–focused message (1.69% vs 1.93%; P = .39). In experiment 2, recipients of the dental benefits–focused message were more likely to respond than recipients of the health benefits–focused message (3.62% vs 2.82%; P = .02). No significant differences were found in response rates in experiment 3 (health benefits with social norming vs dental benefits with social norming; 2.17% vs 3.33%; P = .09) or experiment 4 (dental benefits vs dental benefits with social norming; 3.02% vs 2.58%; P = .21). In an adjusted analysis in which the independent effects of each message component were determined using a logistic regression model that included fixed effects for each experiment to account for differences over time, messages that emphasized the dental benefits of insurance were significantly more likely to result in a response than a message emphasizing the health benefits (odds ratio = 1.33; 95% confidence interval = 1.10, 1.61). Including social norming or financial-focused messages did not lead to higher response rates.

DISCUSSION

Medicaid enrollment outreach messages that emphasized dental benefits were modestly more effective than those that emphasized health-related benefits. Low-income populations have a high burden of oral health problems, which may explain why these messages were more salient to this population.¹¹ More importantly, unlike in many other states in which Medicaid provides only emergency dental care, comprehensive dental coverage is included in adult Medicaid plans in Pennsylvania. Meeting the oral health needs of new beneficiaries will depend on accessible dental services as well as ensuring that beneficiaries are aware of such benefits.¹² In states with comprehensive dental coverage, outreach messages that emphasize these benefits may be more effective.

Although response rates were generally low for all messages, these results should be viewed in the context of a low-intensity, pragmatic intervention that involved a single outreach letter. A central challenge of this campaign was that outreach data were not thoroughly cleaned of individuals who were already receiving Medicaid. As such, an unknown percentage of individuals contacted were already insured and would not have been expected to respond. Because information provided by Pennsylvania state agencies to BDT was limited to a name and address for outreach purposes, we could not confirm that randomization resulted in a balanced distribution of demographic characteristics. The response rates we observed across these outreach message tests were somewhat lower than the response rates BDT receives to outreach campaigns focused on other public benefits. For example, during the same period as this study, a mail outreach campaign to individuals receiving unemployment benefits that informed them of potential SNAP eligibility resulted in a response rate of 5% in Philadelphia and 16% statewide. SNAP outreach letters may elicit a greater response than health insurance letters because of the more tangible monthly financial benefit that meets an immediate need, whereas health insurance needs vary over time. Despite the low response rate, mail outreach may therefore still be a feasible complement to more effective but relatively expensive enrollment strategies, such as insurance navigators and application counselors.^2,13

PUBLIC HEALTH IMPLICATIONS

Our study has several implications. First, outreach messages that emphasize dental benefits may be more effective at motivating enrollment. Dental care may resonate more than medical care given the acute oral health needs of low-income adults.¹⁴ Second, low-intensity, low-cost direct mail outreach can be successful in enrolling socially vulnerable populations into Medicaid, but to date response rates have been low. However, partnerships with Medicaid programs that allow for more tailored messaging and precise targeting of the uninsured could improve response. Third, although previous studies have suggested that social norming messages can motivate behavior, we did not observe that in our study.

With the Medicaid program’s structure and eligibility likely to change, it is important to test and identify successful outreach and enrollment strategies. Enrollment assistance programs that market other public benefits and then screen for health insurance eligibility are likely to be more effective. Our experience bears this out: during the study period, BDT conducted a separate outreach program for SNAP benefits, and more than half of the people who responded to that letter ultimately completed an application for Medicaid. Alternatively, states may adopt fast-track enrollment programs that enable state Medicaid agencies to systematically identify individuals who are Medicaid-eligible using other state data (e.g., SNAP recipients) and then directly facilitate enrollment through a streamlined application process.^2,15 BDT is currently working in partnership with the Pennsylvania Department of Human Services to operate fast-track enrollment in Pennsylvania with success.