Mental Healthcare Use After Leaving Medicare Advantage for traditional Medicare

Abstract

Objectives:

We evaluate changes in mental health visits and specialties among beneficiaries with at least one mental health visit before and after switching from Medicare Advantage (MA) to traditional Medicare.

Study Design:

This study examines Medicare beneficiaries with mental health diagnoses who switched from MA to traditional Medicare in 2018, analyzing their mental health utilization 12 months before and after the switch using MA encounter and traditional Medicare claims data.

Methods:

A longitudinal design was used, comparing mental health visits before and after the switch. We applied Wilcoxon Signed-Rank Tests to compare the total number of visits and McNemar’s tests for specific physician specialties. Statistical significance was defined as p<0.05.

Results:

Of the 32,710 beneficiaries who switched from MA to traditional Medicare in 2018, 1,184 beneficiaries (or 11,1015 claims) were included in our sample if they had at least one healthcare visit attributed to a mental health condition both before and after switching. We find a statistically significant difference in number of mental health visits before and after switching (p=0.014). For top 5 most prevalent specialties, we find no change in the use of psychiatrists (p=0.607) nor family medicine (p=0.696) specialties. However, we find increased use of nurse practitioners (p<0.001), alongside decreased use of internal medicine (p=0.003) and emergency medicine specialties (p=0.001) for mental healthcare after switching.

Conclusions:

Among beneficiaries with continued mental health utilization, switching from MA to traditional Medicare was associated with increased visits and a shift in provider compositions, which suggests potential care gaps or unmet needs.

PRECIS:

Medicare Advantage beneficiaries with mental health diagnoses see more nurse practitioner and fewer internal medicine and emergency physician specialties after switching to traditional Medicare.

INTRODUCTION

Enrollment in Medicare Advantage (MA) has grown rapidly,¹ suggesting that beneficiaries may prefer these privately run MA plans to traditional Medicare. One reason for this could be that these plans offer enhanced benefits and additional coverage, which often include dental and vision,^2,3 an out-of-pocket maximum spending cap, and non-medically related benefits, such as gym memberships.^4–6 While these benefits in MA can be valuable to beneficiaries, trade-offs exist, especially for beneficiaries with complex health needs. High-need and high-cost beneficiaries,^7,8 and those who have developed functional disability or Alzheimer disease and related dementias,^9,10 have increasingly switched from MA back to traditional Medicare.⁷ These findings suggest that MA’s advantages may diminish as beneficiaries age or their health conditions worsen. In particular, MA’s managed care strategies, including prior authorization requirements^11,12 and provider networks, can create barriers to timely access to care. These barriers may disproportionally affect beneficiaries with certain health conditions, including mental health conditions, prompting those beneficiaries to switch back to traditional Medicare.^10,13,14

Mental healthcare access is a critical issue for MA beneficiaries as the provider networks in MA contracts are often narrow, particularly for mental health services. Whereas beneficiaries in traditional Medicare can theoretically see any Medicare-accepting provider, MA enrollees are limited to providers in their plan’s network. Prior work has found that over two-third of MA enrollees’ plans included less than 25% of the local Medicare-accepting psychiatrists in their network.¹⁵ Additionally, MA plans include a low share of local psychiatrists even when compared to plans in the Medicaid and Affordable Care Act markets.¹⁶ Using a prediction algorithim, Feyman et al. find that network restrictiveness and access varibed by specialty, with psychiatry being in the top 10 most restricted specialties in MA. They predict that MA beneficiaries saw 49% as many psychiatrists as they would absent network restrictions.¹⁷ This discrepancy in provider access may be particularly burdensome for MA beneficiaries seeking mental healthcare, motivating them to switch to traditional Medicare, where they face fewer restrictions on provider choice.

This paper investigates the mental health utilization patterns among beneficiaries who had at least one mental health visit before and after switching from MA to traditional Medicare. By focusing on the intensive margin, we capture how care patterns shift among those who are already engaged in mental health treatment, rather than among enrollees who may develop mental health conditions after switching. Using MA encounter data and traditional Medicare claims, we examine the frequency of mental health visits and the specialty of providers rendering mental health services, before and after beneficiaries switch. We aim to better understand the impact of Medicare plan choice on access to mental healthcare.

METHODS

Data

We use the 2018 Master Beneficiary Summary File (MBSF) to identify beneficiary monthly enrollment in either traditional Medicare or MA and beneficiary characteristics. These data are supplemented by the 20% nationally representative sample of the 1) 2017 and 2018 MA encounter and 2) 2018 and 2019 traditional Medicare claims carrier (physician offices) files. We further use the 2019 OneKey provider data and merged with the Medicare claims and encounter data by physician National Provider Identifier (NPI), as the OneKey data included provider specialty information.¹⁹

Sample

Beneficiaries were included if they were 66 years or older in 2018, resided in 50 US states or Washington DC, and were not dually eligible for Medicare and Medicaid (Appendix Figure 1). “Switchers” (those who switched from MA to traditional Medicare) were identified using 2018 MBSF month-to-month enrollment in either MA or traditional Medicare.

We sought to build an analytical sample of Medicare beneficiaries with at least 1 mental health visit before and after switching. Subsetting to MA beneficiaries who already had a mental health diagnosis mitigates the new onset problem, that they are switching to traditional Medicare because of an incident mental health diagnosis. To identify changes in utilization before and after the switch, we required 12 months of enrollment in MA and traditional Medicare before and after the switch, respectively. In other words, each individual beneficiary was followed for a total of 24 months, centered around the month during which their switching occurred in 2018. Our sample included beneficiaries who switched in any month in 2018.

To find visits attributed to a mental health visit, we identified the visit’s primary diagnosis code using the International Classification of Diseases, Tenth Revision (ICD-10) codes. We identified beneficiaries who had a mental health visit by restricting to claims with ICD-10 codes starting with “F,” or “Mental, Behavioral and Neurodevelopmental Disorders.” The ICD-10 codes, which could be as specific as 5 digits, were grouped together into higher-level groups for analysis (Appendix Table 1).

Our sample of healthcare claims and encounter data were merged with the OneKey provider data by provider National Provider Identification (NPI), and mental health specialties were identified using the “provider specialty” information provided by OneKey. Leveraging the OneKey provider directory data was necessary as the physician NPI variable in the MA encounter data was sparsely populated.

Our final sample included beneficiaries who had at least 1 MA encounter for mental health before switching and at least 1 traditional Medicare claim for mental health after switching and included provider specialty information from the OneKey data. Additional analyses further restricted the sample to beneficiaries who did not visit a psychiatrist in MA prior to switching.

Outcome Variables

We measure the number of healthcare visits for mental health conditions as well as the physician’s primary specialty for those visits at any time during the 24-month period during which the beneficiary was first enrolled in MA (12 months) and then enrolled in traditional Medicare (12 months). For each of the physician primary specialties, we measure the variable as a dichotomous outcome identifying whether the visit was attributed to a specific provider specialty or not. For example, when examining the utilization of psychiatrists, the outcome was whether the mental health visit was attributed to provider specialty assigned as “psychiatrist.”

Independent Variables

Our primary independent variable is a binary variable, indicating before switching (while the beneficiary was enrolled in MA) or after switching (while the beneficiary was enrolled in traditional Medicare).

Sensitivity Analyses

To explore potential data quality differences between the traditional Medicare claims and MA encounter data that may influence our results,²⁰ we subset our data to beneficiaries enrolled in an MA contract identified as highly complete in 2018.^21,22 To do this, we identified the MA contract beneficiaries were enrolled in January 2018.

To further examine mental healthcare utilization with psychiatrists, we identify beneficiaries who did not see a psychiatrist while enrolled in MA before switching and then quantify the percentage of those beneficiaries who did see a psychiatrist in traditional Medicare after switching. To ensure that these results are not due purely to mean reversion, we apply the same inclusion/exclusion criteria to MA beneficiaries who remain in MA for 2 full years (2018 and 2019). Of those who did not see a psychiatrist in 2018, we quantify the percentage that did see a psychiatrist in 2019, while remaining in MA. While this does not directly mirror the analysis for Medicare switchers since MA stayers do not have a “switching month” to center their 24-month period of observed healthcare utilization, this approach generates an analogous sample to test for natural revision to the mean.

Finally, we perform a subgroup analysis by subsetting our data to the 2 most prevalent aggregated mental and behavioral health conditions.

Statistical Analysis

We analyzed mental healthcare utilization using a paired design, as each beneficiary had healthcare visits recorded both before and after switching from MA to traditional Medicare. To compare the total number of mental health visits between these two periods, we conducted a Wilcoxon Signed-Rank Test, appropriate for paired non-parametric data. To assess changes in the utilization of specific physician specialties, we performed McNemar’s tests for each specialty separately. Statistical significance was defined as p<0.05.

Analyses were conducted in Stata version 16 (StataCorp LLC) and R version 4.2.2. This study was approved by the Johns Hopkins Bloomberg School of Public Health Institutional Review Board (approval #11318).

Limitations

This study has limitations. First, healthcare utilization patterns for mental health can change for various reasons, and our analysis does not isolate these underlying drivers. For instance, beneficiaries may switch from MA to traditional Medicare in response to a new mental health diagnosis, which could independently influence their utilization of services and specialists. We mitigate this limitation by subsetting our analytic sample to beneficiaries who have at least one mental health visit in MA prior to switching. While future research is needed to disentangle these pathways, this study provides foundational empirical evidence on mental healthcare utilization before and after switching Medicare coverage—a critical first step in understanding these patterns. Second, while the use of traditional Medicare claims and MA encounter data allow us to capture a large sample, we can only observe healthcare utilization, which is a proxy for demand. There may be beneficiaries who seek mental health resources but are unable to do so, and these cases would not be captured in our data. Third and finally, merging the data with OneKey to identify provider specialty drops many of the claims and encounters, so results may not be generalizable outside of the sample included in our analysis.

RESULTS

Of the 32,710 beneficiaries who switched from MA to traditional Medicare in 2018, 7,226 beneficiaries had a healthcare encounter with a mental health ICD-10 code, and 5,730 beneficiaries were included after merging claim-level physician information with OneKey provider directory using physician NPI. Of those 5,730 beneficiaries, 1,184 beneficiaries (or 11,1015 claims) were included in our sample if they had both a MA encounter for mental health before switching and a traditional Medicare claim for mental health after switching (Figure 1). Since our sample decreases with each exclusion criteria, we compare key beneficiary characteristics across the different sample groups. We find no large differences between the sample groups for the sex, age group, or race distribution, or mean number of comorbidities (Table 1).

Figure 1.

Flow chart illustrating sample before and after exclusion criteria

Table 1.

Beneficiary characteristics across samples

	Merge with OneKey Provider data	Beneficiary had claim before and after switching	Did not visit psychiatrist in MA
Beneficiaries, N	5,730	1,184	821
Total Claims	23,031	11,015	5,514
Claims (Avg, SD)	4.02 (7.10)	9.30 (12.71)	6.72 (6.07)
Female, %	64.1	65.5	68.0
Age Group, %
65–74	50.3	50.0	46.0
75–84	32.5	30.3	33.0
85+	17.2	19.7	21.0
Race, %
non-Hispanic White	77.5	82.4	83.2
Black	11.4	10.1	9.7
Hispanic	7.7	4.7	4.9
Other	3.3	2.7	2.2
Comorbidities	21.08 (21.09)	23.82 (20.66)	22.85 (20.64)

We compare the distribution of mental health visits by ICD-10 grouping for the 1,184 beneficiaries that includes both MA encounter and traditional Medicare claims before and after switching, respectively. For that sample, the majority of encounters was for Mood disorders, which include Major depressive disorder and Bipolar disorder. Physiological conditions, anxiety and stress, and schizophrenia were the second, third, and fourth most prevalent mental health visits, respectively (Figure 2). Next, we compare the distribution of specialists for the healthcare visits for mental health conditions. The most prevalent provider specialty, regardless of in MA or traditional Medicare, is Psychiatrists, followed by Family Medicine, Internal Medicine, Nurse Practitioner, and Emergency Medicine physicians (Figure 3).

Figure 2.

Distribution of visits by ICD-10 specialty for beneficiaries with mental health visits before and after switching

Figure 3.

Distribution of visits by physician specialty for beneficiaries with mental health visits before and after switching

In our sample, beneficiaries had a median of 8 visits (25%–75% percentile: 4–18) in MA before switching and a median of 9 visits (25%–75% percentile: 4–21) in traditional Medicare after switching. The average number of visits per beneficiary with psychiatrists is 1.4 (SD: 4.2) and 1.7 (SD: 5.3) in MA before switching and in traditional Medicare after switching, respectively (Table 2). To contextualize these utilization patterns, we compare mental healthcare utilization for beneficiaries who stay in traditional Medicare or MA only, who are outside of our sample. We find that beneficiaries have a median of 11 (25%–75% percentile: 4–24) and 7 (25%–75%: 3–16) mental health visits for traditional Medicare and MA beneficiaries, respectively (Appendix Table 2).

Table 2:

Beneficiary-level total number of visits and number of visits by physician specialty before and after switching

	Medicare Advantage	Traditional Medicare	Test of Significancea
Beneficiaries, #	1,184	1,184
Number of visits, median (25%-75% percentile)	8 (4–18)	9 (4–21)	p=0.014**
Specialty, mean (SD)
Psychiatry	1.41 (4.22)	1.67 (5.28)	p=0.607
Family Medicine	.63 (1.50)	.63 (1.67)	p=0.696
Internal Medicine	.62 (1.53)	.47 (1.72)	p=0.003**
Nurse Practitioner	.54 (1.82)	1.01 (2.42)	p<0.001***
Emergency Medicine	.23 (.91)	.17 (1.15)	p=0.001**

^*p<0.100

^**p<0.050

^***p<0.001 Statistical significance was defined as p<0.05

^aTo test whether the total number of visits were the same of different, we used a Wilcoxon Signed-Rank Test appropriate for paired non-parametric data. To assess changes on the intensive margins and compare the number of visits for specific physician specialties individually, we used McNemar’s test.

Using the sample of 1,184 beneficiaries (or 11,015 claims) who had mental healthcare claims both in MA before switching and in traditional Medicare after switching, we find a statistically significant difference in number of mental health visits before and after switching (Wilcoxon signed-rank test: p=0.014). Leveraging McNemar’s test for top 5 most prevalent physician specialties, we find no statistically significant difference for psychiatrists (p=0.6071) or family medicine (p=0.6956) specialties. However, we do find a statistically significant increase in visits with nurse practitioners (p<0.001), alongside decrease in visits with internal medicine (p=0.003) and emergency medicine specialties (p=0.0010) (Table 2). The average number of visits with a Nurse Practitioner is 0.54 before switching and 1.01 after switching, or 85% higher after switching to traditional Medicare (Table 2). Relatedly, we find that the share of mental healthcare services delivered by Nurse Practitioners increases by 8.1 percentage points after switching from MA to traditional Medicare (from 12.3% before switching to 20.5% after) (Figure 3). Visits with internal medicine and emergency medicine specialists decreased by 24.2% and 26.1%, respectively.

Sensitivity Analyses

When we restrict to beneficiaries enrolled in a MA contract identified as highly complete, the number of unique beneficiaries is reduced from 1,184 to 859 (72.6% unique beneficiaries remained in our sample). We find no large differences between average number of mental healthcare claims, percentage female, distribution of age group or race categories, or measure of comorbidity (Appendix Table 3). Additionally, our results remain largely consistent. We find that the number of healthcare visits increases after switching (p=0.017) and visits with nurse practitioners (p<0.001) increase while emergency medicine specialties (p<0.001) decrease (Appendix Table 4).

For the 821 beneficiaries who did not see a psychiatrist in MA prior to switching, we observe that 10% of them see a psychiatrist after switching. Compared to the 258,997 MA stayers who did not see a psychiatrist in 2018, we observe that 2.1% of them see a psychiatrist in the subsequent year.

Subsetting to beneficiaries diagnosed with “Physiological Conditions,” we find slightly different results. Overall number of visits remained consistent (p=0.339), while visits with nurse practitioners (p=0.001) increase while emergency medicine specialties (p<0.001) decrease after switching. Appendix Table 5 and Appendix Table 6 show subgroup analysis results for our sample restricted to “physiological conditions” and “mood disorders”, respectively.

DISCUSSION

In this retrospective, claims-based analysis, we identify MA beneficiaries who received a healthcare service coded to a mental health ICD-10 code both before and after switching to traditional Medicare. We observe an increase in total number of mental health visits. While we observe no statistically significant increase in the utilization of psychiatry or family medicine specialties, we do find greater utilization of nurse practitioner alongside less utilization of internal medicine and emergency medicine specialties for mental health services after enrollees switch to traditional Medicare.

These results suggest several important dynamics related to mental healthcare utilization and access for beneficiaries who switch from MA to traditional Medicare. First, we observe slightly greater number of mental health visits for our sample after beneficiaries switch away from MA to traditional Medicare. Beneficiaries in traditional Medicare are not limited by provider networks and are generally subject to fewer prior authorization policies to see specialists.

Second, the lack of significant differences the use of psychiatry or family medicine specialists suggests that beneficiaries who switch from MA to traditional Medicare do not drastically alter overall care-seeking behavior within these specialties. Beneficiaries who switch from MA to traditional Medicare may have established a preference or relationship with a mental health provider that persisted across systems. While this may be true, we do find that for the 821 beneficiaries who did not see a psychiatrist while in MA, 10% of these beneficiaries did see a psychiatrist after switching to traditional Medicare. For the group who stayed in MA, only 2% of beneficiaries who did not see a psychiatrist in the year prior then saw a psychiatrist. Taken together, these results suggest that beneficiaries in MA may switch to traditional Medicare for broader access to psychiatrists.

Third, our results suggests that there is an increased role of non-physician providers in traditional Medicare. The increased utilization of nurse practitioners suggests that traditional Medicare may provide greater access to non-physician mental health providers, which could be critical to addressing gaps in care or improving timeliness of services. In traditional Medicare, Medicare reimburses nurse practitioners at 85% of the physician fee schedule rate for the same services, meaning that nurse practitioners receive a lower payment for comparable procedures compared to doctors, although the exact payment differential varies at the state-level. A scoping review found that care provided by psychiatric mental health nurse practitioners was generally associated with positive patient outcomes,²³ although more research is needed to understand the Medicare setting.

Another important aspect is the beneficiary’s access to supplemental insurance in traditional Medicare, i.e. Medigap. Most beneficiaries who disenroll from MA are subject to individual rating of Medigap premiums and Medigap insurers can deny coverage outright or effectively deny coverage through high premiums.²⁴ Thus, those beneficiaries who still choose to disenroll, even when faced with potentially high Medigap premiums, may be different than the beneficiaries who remain “trapped” in MA.²⁵ Future work could examine how beneficiaries mental health utilization patterns compare between switchers and non-switchers by state-level Medigap policies.

This work is, to the best of our knowledge, the first to examine the intensive margin of mental health utilization focusing on physician specialties utilized for Medicare beneficiaries who switch from MA to traditional Medicare. Given the movement of mental and behavioral health providers to cash-pay models, future research should explore the role of cash-pay mental health providers on Medicare beneficiaries’ care access. Finally, the MA encounter data are incomplete and many of the NPI variables were missing or did not successfully merge with provider directory data. Comprehensive and accurate MA encounter data will significantly improve the research landscape around the topic of MA beneficiary’s access to specialty care. While this descriptive work is an important first step in understanding specialty care access for Medicare beneficiaries, future work could include qualitative interviews with Medicare beneficiaries and quantitative, causal analyses.

CONCLUSION

Using claims-level data, we analyzed longitudinal beneficiary-level mental healthcare visits and specialties used for beneficiaries who had at least one mental healthcare visit before and after switching from MA to traditional Medicare. We find greater number of mental health visits in traditional Medicare after switching. Examining physician specialties, we find an increase in visits with nurse practitioners, alongside a decrease in visits with internal medicine and emergency medicine specialists. We further find that some beneficiaries who do not see a psychiatrist before switching do see a psychiatrist after switching, underscoring the potential barriers MA beneficiaries might face in accessing specialty mental healthcare. Future research should explore these regulatory dynamics around mental healthcare access, particularly at the insurer and plan level, to ensure access to mental healthcare for all Medicare beneficiaries.

TAKEAWAY POINTS:

• Medicare Advantage (MA) insurers use provider networks which may restrict access to mental healthcare. This study examines changes in mental health visits and specialties used before and after beneficiaries switch from MA to traditional Medicare.

• We find an increase in the number of mental health visits after switching. We find increased visits with nurse practitioners, decreased visits with emergency medicine and internal medicine specialists, and no change with psychiatrists and family medicine specialists.

• These findings provide descriptive evidence of how transitioning from MA to traditional Medicare is associated with type of provider specialties used.