(84) Population, Outcome, and Treatment Changes in Secondary Data Early in the COVID-19 Pandemic

Publication Number: 415

Background: The COVID-19 pandemic has had an enormous impact on public health directly via infections (and associated morbidity and mortality) and indirectly via lockdowns and social distancing. Understanding alterations in populations, outcome measurement, and treatment initiation is necessary to use secondary data from 2020, 2021, and 2022 for pharmacoepidemiologic research. Because 2020 North Carolina (NC) Medicaid data is already available, it may be useful for identifying trends common in secondary data.

Objectives: Understand changes in the secondary data collected in NC Medicaid in 2020 with respect to 1) population size and demographics, 2) rates of a variety of outcomes, and 3) new use of common treatments.

Methods: We identified all adults (age >18) with any coverage by North Carolina Medicaid from March through September 2018, 2019, and 2020. We then calculated the person-time overall and stratified by age, sex, and race. This person-time was used to calculate rates per 100 person-years (PY) of nine commonly used outcomes in pharmacoepidemiology. We also compared rates of new use per 100 PYs for five common drugs and drug classes. Finally, we compared monthly rates to identify when 2020 data differed the most from the other calendar years.

Results: The total number of individuals and amount of person-time covered by NC Medicaid from March-September was greatest in 2020 (n=1,275,189; PY=698,076) compared to 2018 (n=1,216,473; PY=649,667) or 2019 (n=1,243,958; PY=661,992), with the larger relative increase in person-years because laws were passed to prevent Medicaid coverage from lapsing during the pandemic. Demographics were similar in each calendar period. Despite this similarity, rates of several outcomes differed substantially in 2020 compared to 2018 and 2019, including higher rates of death (2.35/2.51/2.74 per 100 PY in 2018/2019/2020), lower rates of breast cancer screening in women (3.30/3.27/2.77), and fewer myocardial infarctions (2.11/2.22/1.74), with the largest differences observed in March, April, and May. Other outcomes showed similar trends. While new use of drugs also appeared lower during April (1.99/2.02/1.73 for metformin, 3.45/3.35/3.13 for statins) and May (2.00/1.87/1.66 for metformin, 3.58/3.42/2.92 for statins), a rebound in prescriptions in later months resulted in similar overall rates of new use; this pattern was common to most drugs studied.

Conclusions: The pandemic and associated public health measures resulted in substantial changes in the healthcare system that are reflected in the 2020 NC Medicaid data. Establishing how these changes impact outcome misclassification and propensity score estimation will help studies using similar real-world data reach valid conclusions.