key: cord-0074547-e2ywr167
authors: Ko, Hansoo; Glied, Sherry A.
title: Associations Between a New York City Paid Sick Leave Mandate and Health Care Utilization Among Medicaid Beneficiaries in New York City and New York State
date: 2021-05-06
journal: JAMA Health Forum
DOI: 10.1001/jamahealthforum.2021.0342
sha: cc74d4a9e8b2bec72f8ec0063a941b7e16748e2a
doc_id: 74547
cord_uid: e2ywr167

IMPORTANCE: More evidence on associations between mandated paid sick leave and health service utilization among low-income adults is needed to guide health policy and legislation nationwide. OBJECTIVE: To evaluate the association between New York City’s 2014 paid sick leave mandate and health care utilization among Medicaid-enrolled adults. DESIGN, SETTING, AND PARTICIPANTS: This retrospective cohort study used New York State Medicaid administrative data for adults 18 to 64 years old continuously enrolled in Medicaid from August 1, 2011, through July 31, 2017. A difference-in-differences approach with entropy balancing weights was used to compare New York City with the rest of New York State to assess the association of the paid sick leave mandate with health care utilization, and for those 40 to 64 years old, with preventive care utilization. The data analysis was performed from June through August 2020. EXPOSURES: Temporal and spatial variation in exposure to the mandate. MAIN OUTCOMES AND MEASURES: Annual health care utilization (emergency care, specialist visits, and primary care clinician visits) per Medicaid-enrolled adult. Secondary outcomes include categories of emergency utilization and utilization of 5 preventive services. RESULTS: Of 552 857 individuals (mean [SD] age, 43 [12] years; 351 130 [64%] women) who met inclusion criteria, 99 181 (18%) were White, 162 492 (29%) Black, and 138 061 (25%) Hispanic. Paid sick leave was significantly associated with a reduction in the probability of emergency care (−0.6 percentage points [pp]; 95% CI, −0.7 to −0.5 pp; P < .001), including a 0.3 pp reduction (95% CI, −0.4 to −0.2; P < .001) in care for conditions treatable in a primary care setting and an increase in annual outpatient visits (0.124 pp; 95% CI, 0.040 to 0.208 pp; P < .001). Among those 40 to 64 years old, the mandate was significantly associated with increased probabilities of glycated hemoglobin A(1c) level testing (2.9 pp; 95% CI, 2.5-3.3 pp; P < .001), blood cholesterol testing (2.7 pp; 95% CI, 2.5-2.9 pp; P < .001), and colon cancer screening (0.4 pp; 95% CI, 0.2-0.6 pp; P < .001). CONCLUSIONS AND RELEVANCE: This retrospective cohort study of nonelderly adults enrolled in Medicaid New York State showed that mandated paid sick leave in New York City was significantly associated with differences in several dimensions of health care services use.

The key assumption of our analyses is that the effects should be concentrated in the second half of 2014 because accumulation of sick leave hours began in April 2014 but paid sick leave became available to use beginning in August 2014. In other words, our utilization measures would have followed the same trends without the policy and we would find smooth trend lines at the cutoff date. To explore whether there were discrete changes in utilization when paid sick leave was implemented in New York City in August 2014, we specify a regression discontinuity model as follows: (1) Unit of analysis is person-month. Our outcome measure is monthly utilization per person including primary care clinician visit and emergency room visit. In particular, classifying emergent utilization into the four types of visit, 1 we expect to find significant reductions in emergency room utilization for conditions classified as primary care treatable and insignificant changes in not-preventable emergent utilization. A binary indicator equals 1 in and after

August 2014 ( 0, ranges from -12 to 11) and a parameter shows discontinuous changes in outcome measures at month zero attributed to the implementation of paid sick leave -for our best knowledge, no other changes in Medicaid utilization policy were concurrently made in August 2014. is slopes of linear trend and a parameter represents a gradual change in trends (the difference in the slopes between the pre-policy period and the post-policy period).

We are interested in the statistical significance and direction of coefficient estimates on and For graphical presentations shown in eFigure2, we first estimate seasonality-adjusted monthly utilization rate using the following regression:

Taking regression coefficients on calendar month dummies ( ), we calculate seasonality- Regressions also included month dummies (February through December) and individual fixed effects. Standard errors were clustered at the individual level. Detailed information on the regression framework was provided in eAppendix2.

Emergency room use: The

Medicaid increases emergency-department use: Evidence from Oregon's Health Insurance Experiment

Lower incidence of myocardial infarction after smoke-free legislation enforcement in Chile

Abbreviations: ED, emergency department; DID, difference-in-differences. Regressions included year dummies, individual fixed effects, and Charlson comorbidity index