Medicare Advantage Customer Service Is Used Most by Higher-Need Patients

ABSTRACT

Objectives: To examine characteristics of Medicare Advantage (MA) enrollees who use their plan’s customer service to help plans understand how to better meet members’ needs.

Study Design: National sample of 259,533 respondents to MA Consumer Assessment of Healthcare Providers and Systems survey enrolled in any of the 559 MA contracts in 2022.

Methods: We assessed the association between self-reported customer service use in the prior 6 months and enrollee demographic, coverage, health, and health care utilization characteristics. We used weighted linear regression models to test for bivariate and multivariate associations between customer service use and enrollee characteristics.

Results: Forty-two percent of MA enrollees reported using customer service in the prior 6 months. Use was 20 percentage points (PP) higher for those in poor vs excellent/very good general health, 13 PP higher for those in poor vs excellent/very good mental health, and 14 PP higher for those reporting 3 or more vs no chronic conditions. Those using customer service more often had lower educational attainment, had limited income and assets, preferred another language to English, and had greater health care utilization.

Conclusions: MA customer service supports a less healthy, higher-need population with greater-than-average barriers to health care, and so should be designed and staffed to effectively serve medically complex, high-need patients. Commercial plan evidence suggests that continuity in customer service support for a member or a given issue may be helpful. Customer service is an important mechanism for improving quality and addressing health equity.

Am J Manag Care. 2024;30(8):381-386. https://doi.org/10.37765/ajmc.2024.89589

Takeaway Points

Medicare Advantage (MA) enrollees who are in poorer general and mental health and who have greater medical complexity are more likely to use their plan’s customer service.

• Customer service use by MA enrollees was significantly higher for those with poorer general health, those with poorer mental health, and those reporting 3 or more chronic conditions.
• MA enrollees using customer service more often had lower educational attainment, had limited income and assets, preferred another language to English, and had high health care utilization.
• MA customer service should be designed and staffed to effectively serve complex, high-need patients.

Customer service (CS) systems play a crucial role in Medicare Advantage (MA) plans because they assist enrollees in accessing the information they need to navigate their health care coverage. Health plans have 2 major forms of CS: call center agents and automated systems. Call center agents inform enrollees about their health insurance coverage and assist in solving issues that arise in care delivery. In contrast, automated customer service through interactive voice recognition technology allows callers to quickly perform routine functions (eg, requesting provider directories or other plan documents, checking the status of claims or authorizations, verifying eligibility). Each of these types of CS interactions provides an opportunity for the health plan to influence the enrollee’s perception of the quality of care being provided, increase equity of access to care, and foster patient loyalty.^1,2 Within commercial health plans, people with Medicare typically represent 75% of all calls.^3,4

CS is particularly important for the MA population, who are lower in income, educational attainment, and health literacy, and higher in multimorbidity and age than the overall US population.^5-7 Their calls usually require the longest handling time and are typically the most complex because these callers are usually older than 65 years and often have special needs related to their age or other limitations. People with Medicare also are especially likely to prefer call center agents to interactive voice recognition.3 In 2023, 30.8 million people were enrolled in an MA plan, accounting for 51% of the eligible Medicare population.⁸ Furthermore, those with Medicare who have chronic conditions account for a disproportionate share of health care expenditures and may find it difficult to manage their care, increasing the likelihood of CS calls for assistance.⁹

CMS developed the MA Consumer Assessment of Healthcare Providers and Systems (CAHPS) surveys to collect data on health plan member experiences with MA (MA-only) and MA Prescription Drug (MA-PD) plans.¹⁰ MA CAHPS includes a 3-item CS measure for those who self-report CS use on a screener item. This CS measure is strongly predictive of the overall rating of the health plan,¹¹ the measure that most reliably differentiates plans¹² and is associated with voluntary disenrollment.¹³ Therefore, CS defines the subset of enrollees needing and wanting additional information and attention, which highlights that CS is a critical touch point in care delivery for quality and member retention.³

We examine which MA enrollees use their plan’s CS to understand how plans can better meet their members’ needs. Previous research has examined disparities in CS experience based on race, ethnicity, and sex.^14,15 By identifying and understanding different customer segments, plans can tailor their products, services, and communication efforts to better meet the specific needs of each segment.¹⁶ However, there have been no studies to date that examine the characteristics of Medicare enrollees who seek assistance from their plan’s CS.

METHODS

Data came from the 2022 MA CAHPS survey (response rate, 35.2%). Poststratification analytic weights were used in all analyses, and matched the sample to the MA population on an extensive list of variables, including race and ethnicity, sex, dual eligibility (DE) for Medicare and Medicaid, and zip code–level US Census Bureau data on race and ethnicity, income, and education using iterative proportional fitting.¹⁷

Our outcome measure is self-reported use of CS in response to the CS screener item: “In the last 6 months, did you get information or help from your health plan’s customer service?” For this analysis, a “yes” response was recoded as 100 and a “no” response was recoded as 0. Of the 259,533 survey respondents, 17,881 (6.9%) did not respond to this item and were excluded, leaving 241,652.

We assessed the association between the use of CS and enrollee demographic, coverage, health, health care utilization, and geographic characteristics. Demographic characteristics included age, sex, race and ethnicity, survey language, and education. Coverage characteristics included DE, having a low-income subsidy (LIS), and MA coverage type (MA without Part D drug coverage [MA-only], MA with Part D drug coverage [MA-PD] but no Special Needs Plan [SNP]; chronic condition SNP; DE SNP); different forms of coverage could result in different kinds of customer service questions. Health characteristics included self-reported general and mental health and count of prior diagnoses of 6 chronic conditions (angina, cancer, chronic obstructive pulmonary disease, diabetes, high blood pressure, and heart attack). Health care utilization characteristics included having a personal doctor and use of specific health care services within the past 6 months (urgent care, routine care, care from more than 1 type of provider, specialist care, and hospitalization). Geographic characteristics included rurality and location (4 Census regions and Puerto Rico).

We used linear regression to test for bivariate and multivariate associations between CS use and enrollee characteristics. We used linear rather than logistic regression models for ease of interpretation because they are almost identical when sample sizes are large and outcomes are predominantly between 20% and 80% (as is the case here).¹⁸ At the sample sizes employed here, the central limit theorem ensures that the assumption that SEs of regression coefficients are normally distributed is satisfied.¹⁹ In bivariate analyses, 1 enrollee characteristic at a time was analyzed in a regression model. In multivariate analyses, all enrollee characteristics were included in a single regression model.

Supplementary analyses examined item nonresponse and the associations among CS use, experiences, and overall plan ratings.

RESULTS

Approximately 42% of all enrollees reported getting CS help in the last 6 months, with this rate varying by enrollee characteristics. Table 1 presents analytic results for the demographic and coverage characteristics. Bivariate models are summarized in column 5. Enrollees aged 18 to 64 years were more likely to contact CS than other age groups, with rates being 7 percentage points (PP) higher than among those aged 65 to 69 years. Enrollees 80 years or older were least likely to contact customer service, with rates 6 PP lower than in those aged 65 to 69 years. Male enrollees were 0.5PP more likely to contact customer service than female enrollees. There was marked variation by race and ethnicity, with American Indian/Alaska Native (5 PP), Asian American (6 PP), multiracial (12 PP), Black (15 PP), and Hispanic (19 PP) enrollees more likely than White enrollees to seek CS help. Enrollees who completed the survey in Spanish were 23 PP more likely to report CS use than enrollees who responded in English. Rates of CS use were 4 to 10 PP higher for enrollees without a high school degree than those with a high school degree. Enrollees who have DE or LIS coverage were 16 PP more likely to use CS than enrollees who did not have this coverage. Enrollees in SNPs were 12 to 18 PP more likely to use CS than other enrollees in MA-PD plans; enrollees with MA-only coverage were the least likely group to use CS (9 PP less likely than those with non-SNP MA-PD coverage).

Table 2 presents parallel results for health, health care utilization, and geographic characteristics. In bivariate analysis, we found that a higher percentage of enrollees with worse self-reported health report getting CS help (20 PP higher for enrollees indicating poor vs excellent or very good general health; 13 PP for enrollees indicating poor vs excellent or very good mental health). Having 1 of 6 specific chronic conditions was associated with a higher likelihood of using CS, with rates especially high for those with 3 or more conditions (14 PP higher compared with enrollees with no chronic condition). Enrollees who had a personal doctor were 14 PP more likely to seek CS help than enrollees who did not. Health care utilization measures were also associated with higher rates of getting CS help. Enrollees who required urgent care, made an appointment for routine care, received care from more than 1 type of provider, saw a specialist, or spent 1 or more nights in the hospital in the past 6 months were 11 to 16 PP more likely to report using CS than enrollees who did not have those health care needs. Rates of contacting CS varied by Census region and were highest in Puerto Rico (25 PP higher than in New England).

The last columns of Tables 1 and 2 summarize multivariate results from a single multivariate model. Asian American (7 PP), multiracial (8 PP), Hispanic (9 PP), and Black (10 PP) enrollees were more likely than White enrollees to seek CS help. Spanish survey response was associated with a 7 PP higher rate of CS compared with English-surveyresponse. DE/LIS enrollees were 7 PP more likely to contact CS after adjustment for all enrollee characteristics. SNP enrollees had a 6 PP higher rate of getting CS help than other enrollees with MA-PD coverage. Enrollees with MA-only coverage were 6 PP less likely to get CS help than enrollees with non-SNP MA-PD coverage.

Turning to Table 2, enrollees with poor general health had a 4 PP higher rate of getting CS help than those with excellent or very good general health. Enrollees with 3 or more chronic conditions had a 5 PP higher rate of getting CS help than those with no chronic conditions. Enrollees with a personal doctor had a 10 PP higher rate of getting CS help than enrollees who did not. Enrollees who made an appointment for routine or specialist care had 9 to 10 PP higher rates than other enrollees.

eAppendix A (eAppendices available at ajmc.com) describes parallel results for 2019. Notable is that reported use of CS declined by approximately 1.1 PP in 2019-2022. The unadjusted and adjusted patterns are broadly similar by patient characteristics. eAppendix B shows that plans with higher CS use had slightly worse CS experiences but that, taken together, contracts with better CS experiences had markedly higher ratings and those with more CS use had slightly higher ratings. eAppendix C indicates that nonresponse to the CS item was similar to that for other items.^20,21

DISCUSSION

MA enrollees with poorer general and mental health, greater medical complexity, limited income and assets, and higher health care utilization, as well as those who belong to a racial or ethnic minority group or prefer another language to English, were more likely than others to report using MA CS. This evidence adds to the current knowledge about the higher importance placed on CS among some patient groups. CS users may have higher-than-average needs and barriers to care or require clarification about specific aspects of coverage or help with access or care.

We found that plans with high CS use had lower CS performance and that CS performance is a strong driver of plan ratings, underscoring that better serving the population of CS users can improve care. As such, our findings suggest that health plan CS needs to be staffed and trained to meet the needs of a racially, ethnically, and linguistically diverse population with lower-than-average income and assets and worse-than-average health, with multiple chronic conditions. Furthermore, representatives’ conversations with members should include asking how to best support the caller, including whether the caller has access to others who can help or prefers information provided in writing or in another language. Any written material provided for follow-up should be simple and clear, at most at a 6th-grade reading level,^22,23 culturally appropriate, and in multiple languages (as needed). The best method to exchange information may differ across patient groups.

Further, findings suggest that health plans can differentiate themselves in service quality and experience by using call center metrics to analyze communications and member needs to assess health risks and status of various segments of their member population and to better target process improvements that support high users’ needs. This may require health plans to improve their database management and integrate information management and call center strategies.

To provide more personal assistance, health plans have instituted several CS approaches; these also could assist high-need patients. For example, some health plans use a personal service representative, pairing each member of certain patient groups with a single dedicated service representative to personalize the member’s service for all their calls. This approach led to lower costs through greater efficiencies in plan member referrals to needed services based on the representative’s familiarity with the member.^3,24 Health plans could provide a dedicated representative for high-need patients and frequent callers to emphasize continuity and active follow-up; this could include an option to have a service representative who remains on a given case, including any needed follow-up, assistance in navigating the health system, and follow-up calls to gain resolution. Another innovative CS approach is to set up community-based locations where Medicare members can attend free events and gain answers to their Medicare questions.²⁵

Limitations

Our study has limitations. The national MA survey data provide information about the CS experience and user characteristics but do not include the content of the CS inquiries, which are data that only health plans can examine. However, our findings point out the importance of examining call center data and member characteristics to improve and target services. High CS use could indicate difficulties with the plan, but it is also the case that good CS may encourage more CS use.

CONCLUSIONS

MA CS supports a less healthy, higher-need population with greater-than-average barriers to health care, as well as a culturally and linguistically diverse population, so it should be designed and staffed to effectively serve complex, high-need patients. Effective CS that is aligned with the needs of users results in better patient experiences with their health plan and their care. Measuring and analyzing CS data are vital for improving quality and addressing health equity.

Acknowledgments

The authors would like to thank Biayna Darabidian, MPP, and Katherine Osby, BA, for the preparation of the manuscript.

Author Affiliations: RAND Corporation, Santa Monica, CA (DDQ, MNE, NO), and Pittsburgh, PA (AH, AMH); Carnegie Mellon University (AMH), Pittsburgh, PA; CMS (SG), Baltimore, MD; University of Alabama, Birmingham (RW-M), Birmingham, AL.

Source of Funding: This research was funded by CMS under contract No. GS-10F-0275P/task order 75FCMC20F0101.

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 (DDQ, MNE, AMH, NO, RW-M); acquisition of data (MNE, SG); analysis and interpretation of data (DDQ, AH, MNE, AMH, NO, SG, RW-M); drafting of the manuscript (DDQ, AH, NO, RW-M); critical revision of the manuscript for important intellectual content (AH, MNE, AMH, NO, SG, RW-M); statistical analysis (AH, AMH); obtaining funding (MNE); and supervision (MNE).

Address Correspondence to: Marc N. Elliott, PhD, RAND Corporation, 1776 Main St, Santa Monica, CA 90407. Email: elliott@rand.org.

REFERENCES

1. Strategy 6P: service recovery programs. Agency for Healthcare Research and Quality. Accessed September 6, 2023. https://www.ahrq.gov/cahps/quality-improvement/improvement-guide/6-strategies-for-improving/customer-service/strategy6p-service-recovery.html

2. Kennedy DM. Creating an excellent patient experience through service education: content and methods for engaging and motivating frontline staff. J Patient Exp. 2017;4(4):156-161. doi:10.1177/2374373517718351

3. Strother JB. Call centers in health care: effect on patient satisfaction. In: 2006 IEEE International Professional Communication Conference. IEEE; 2006:291-298. doi:10.1109/IPCC.2006.320362

4. Medicare Beneficiary Telephone Customer Service. HHS Office of Inspector General;2005. Accessed August 11, 2023. https://oig.hhs.gov/oei/reports/oei-07-04-00030.pdf

5. Haviland AM, Ma S, Klein DJ, Orr N, Elliott MN. Association of Medicare Advantage premiums with measures of quality and patient experience. JAMA Health Forum. 2022;3(8):e222826. doi:10.1001/jamahealthforum.2022.2826

6. Thomas KS, Durfey SNM, Gadbois EA, et al. Perspectives of Medicare Advantage plan representatives on addressing social determinants of health in response to the CHRONIC Care Act. JAMA Netw Open. 2019;2(7):e196923. doi:10.1001/jamanetworkopen.2019.6923

7. Tarazi W, Welch WP, Nguyen N, et al. Medicare beneficiary enrollment trends and demographic characteristics. HHS Office of the Assistant Secretary for Planning and Evaluation. March 2, 2022. Accessed August 13, 2023. https://aspe.hhs.gov/sites/default/files/documents/b9ac26a13b4fdf30c16c24e79df0c99c/medicare-beneficiary-enrollment-ib.pdf

8. Ochieng N, Fuglesten Biniek J, Freed M, Damico A, Neuman T. Medicare Advantage in 2023: enrollment update and key trends. KFF. August 9, 2023. Accessed September 6, 2023. https://www.kff.org/medicare/issue-brief/medicare-advantage-in-2023-enrollment-update-and-key-trends/

9. Cotton P, Newhouse JP, Volpp KG, et al. Medicare Advantage: issues, insights, and implications for the future. Popul Health Manag. 2016;19(suppl 3):S1-S8. doi:10.1089/pop.2016.29013.pc

10. Medicare Advantage and Prescription Drug Plan CAHPS Survey. Accessed September 6, 2023. https://www.ma-pdpcahps.org/

11. Zaslavsky AM, Beaulieu ND, Landon BE, Cleary PD. Dimensions of consumer-assessed quality of Medicare managed-care health plans. Med Care. 2000;38(2):162-174. doi:10.1097/00005650-200002000-00006

12. Zaslavsky AM, Landon BE, Beaulieu ND, Cleary PD. How consumer assessments of managed care vary within and among markets. Inquiry. 2000;37(2):146-161.

13. Lied TR, Sheingold SH, Landon BE, Shaul JA, Cleary PD. Beneficiary reported experience and voluntary disenrollment in Medicare managed care. Health Care Financ Rev. 2003;25(1):55-66.

14. Martino SC, Elliott MN, Hambarsoomian K, et al. Disparities in care experienced by American Indian and Alaska Native Medicare beneficiaries. Med Care. 2020;58(11):981-987. doi:10.1097/mlr.0000000000001392

15. Burkhart Q, Elliott MN, Haviland AM, et al. Gender differences in patient experience across Medicare Advantage plans. Womens Health Issues. 2020;30(6):477-483. doi:10.1016/j.whi.2020.08.009

16. Cooil B, Aksoy L, Keiningham TL. Approaches to customer segmentation. J Relatsh Mark. 2008;6(3-4):9-39. doi:10.1300/J366v06n03_02

17. Purcell NJ, Kish L. Postcensal estimates for local areas (or domains). Int Stat Rev. 1980;48(1):3-18. doi:10.2307/1402400

18. Hosmer DW, Lemeshow S, Cook E. Applied Logistic Regression. 2nd ed. John Wiley & Sons Inc; 2000.

19. Chen L. Testing the mean of skewed distributions. J Am Stat Assoc. 1995;90(430):767-772. doi:10.2307/2291090

20. Dembosky JW, Haviland AM, Haas A, et al. Indirect estimation of race/ethnicity for survey respondents who do not report race/ethnicity. Med Care. 2019;57(5):e28-e33. doi:10.1097/MLR.0000000000001011

21. Klein DJ, Elliott MN, Haviland AM, et al. Understanding nonresponse to the 2007 Medicare CAHPS survey. Gerontologist. 2011;51(6):843-855. doi:10.1093/geront/gnr046

22. Safeer RS, Keenan J. Health literacy: the gap between physicians and patients. Am Fam Physician. 2005;72(3):463-468.

23. Improving print communication to promote health literacy. Center for Health Care Strategies, Inc. Updated March 2024. Accessed August 28, 2023. https://www.chcs.org/media/Improving_Print_Communication.pdf

24. Hagland M. Making the connection. Healthcare Informatics. October 1999. Accessed July 3, 2024. https://web.archive.org/web/20060426232034/http://www.healthcare-informatics.com/issues/1999/10_99/hagland.htm

25. Viva Health Cafés. Viva Health. Accessed September 6, 2023. https://www.vivahealth.com/cafes/