We examined Medicare FFS claims data and Part D drug event data (Appendix) to identify LD diagnoses. We restricted the analyzed population to Medicare FFS beneficiaries ≥65 years of age who participated in Parts A, B, and D for at least 12 months of a calendar year or until their month of death during 2016–2019.

Aligning with previously described methods (2), we defined LD diagnosis as an International Classification of Diseases, 10th Revision, Clinical Modification code for LD combined with a drug claim for an antibiotic indicated for LD (Appendix) within 30 days of the LD code and prescribed for >7 days. We defined an inpatient LD diagnosis as hospitalization with a primary code for LD or a primary code for a known manifestation or plausible co-infection of LD plus a secondary code (A69.2x) for LD (Appendix) (2,3).

We compared the restricted Medicare FFS study population with 2019 US Census estimation data for persons >65 years of age to ensure the 2 groups were similar with respect to sex, age, race, ethnicity, and region (4). We grouped states into incidence regions based on previous definitions (3).

We compared LD diagnoses identified in the Medicare FFS data to confirmed and probable cases among persons >65 years of age reported through national LD surveillance (5). State and local health departments investigate and tabulate LD surveillance cases and classify them according to the Council of State and Territorial Epidemiologists (https://www.cste.org) case definition in effect during the reporting year (5). We used SAS 9.4 and SAS Enterprise Guide 7.1 (SAS Institute Inc, https://www.sas.com/en_us) for analyses. The Centers for Disease Control and Prevention deemed this activity not research.

Seasonality

Figure 1

Figure 1. Analysis of Lyme disease among older adults, United States, 2016–2019. A) Percentage of Lyme disease diagnoses by month according to Medicare fee-for-service data. B) Percentage of Lyme disease cases by...

Most (57.8%) LD diagnoses occurred during May–August, but most (72.6%) LD surveillance cases had onset during the summer months (Table; Appendix Figure 4). Compared with Medicare data, the peak in surveillance cases was more prominent for all regions. In addition, a larger proportion of LD diagnoses occurred in winter months among residents of low-incidence areas (Figure 1).

Age and Sex by Region

Figure 2

Figure 2. Lyme disease incidence for older adults, United States, 2016–2019. Results according to age group, sex, and geographic category of Lyme disease endemicity based on Medicare fee-for-service beneficiary data (A,C,F) and...

In high-incidence states, men had the highest incidence of LD for all age groups in both Medicare and surveillance data (Figure 2). In neighboring states, according to Medicare data, women had a slightly higher incidence than men in only 1 age group (65–69 years), whereas according to surveillance data men had a higher incidence of LD across all age groups. In low-incidence states, according to Medicare data, women had a slightly higher incidence than men in 1 age group (65–69 years); in surveillance data, women had a higher incidence than men in a single age group (75–79 years).

Among persons >65 years of age, epidemiologic trends for sex, age, and region are similar for Medicare diagnoses and cases identified through public health surveillance. Nevertheless, overall diagnoses per person-year are ≈7-fold higher than for LD incidence in surveillance data. Those findings align with findings reported in our previous claims analysis (3). Seasonality of LD differed somewhat by region when comparing Medicare FFS and surveillance data: high-incidence states and neighboring states exhibited similar patterns for diagnoses and surveillance cases, but low-incidence states demonstrated a more muted peak for LD diagnoses in summer. Those findings also aligned with past claims analyses (2,3).

Some differences exist between this study and previous claims analyses (2,3). Within the Medicare population, men had a higher incidence of LD compared with women in all age groups in the high-incidence and neighboring states. In past claims analyses, male incidence was higher in children (both regions) and older adults (high-incidence states). Overdiagnosis of LD has been previously reported (6–9) and may contribute to some of those differences.

We used methods similar to past claims analyses to identify LD diagnoses (2,3); those stated limitations also apply here. However, the source from which our present data are derived is different from past analyses, and its representativeness is a strength. Nearly all adults >65 years of age enroll in either the Medicare FFS program or the Medicare Advantage program, making Medicare FFS a reliable and consistent, though still incomplete, source of data for most US citizens in this age group. We noted some differences between the Medicare FFS population and the Census population regarding race, ethnicity, and sex, but differences were small and did not fluctuate over the study period.

In conclusion, we found that LD diagnoses identified from the Medicare FFS databases exhibit similar patterns to those of surveillance data, and that most diagnoses occur among residents of high-incidence states, in summer months, and among male beneficiaries. These findings, combined with data gathered in similar analyses, add insight into LD patterns that are unique to this older population in the United States.