• 2019-07
  • 2019-08
  • 2019-09
  • 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • 2021-03
  • br Cancer screening br Our


    2.2. Cancer screening
    Our dependent variables were binary outcomes denoting status of receiving cervical cancer screening [Papanicolaou (Pap) test], and FOBT/FIT colorectal cancer screening. Because data on resulted sig-moidoscopy or colonoscopy in this population is often missing from the EHR (these services are conducted outside of the clinics), we focused on FOBT/FIT tests only and did not assess sigmoidoscopy or colonoscopy screenings. We based eligibility for each screening on current guidelines from the US Preventive Services Task Force recommendations relevant
    during the study time FSL-1 (https://www. The eligible population for cer-vical cancer screening were females aged 21–64 throughout period, without history of hysterectomy. Human papillomavirus co-testing with a 5-year coverage period was taken into account for women ages 30–64. The eligible population for colorectal cancer screening were patients aged 50–64 throughout period, without history of total colectomy. Our analysis assessed whether each screening was received among patients who were eligible pre- and/or post-ACA Medicaid expansion, including an appropriate look-back period for each type of screening (e.g., three years for Pap). In other words, patients could be eligible for the pre-period only, post-period only, or both pre- and post-periods. Specifically, for cervical cancer screening, a woman eligible during the pre-period who received the test at any point during the 24 months would be in the numerator (received service) for the pre-period and the post-period. If a woman was eligible in the pre-period and received the screening in the post-period, this woman would be in the denominator in the pre-period (not received) and the numerator in the post-period (received). We used a similar process for FOBT/FIT received over the 24 months period even though FOBT/FIT is an annual screening. This study does not assess whether patients due for a screening in the pre-
    period received it in the post-period as the likelihood of receiving a screening would increase with time, independent of the ACA Medicaid expansion. Breast cancer screening was excluded because results from mammograms are often missing from EHRs and thus difficult to accu-rately identify overdue or received screenings. We excluded clinical breast exam alone, as the USPSTF does not recommend for or against its use as an effective breast cancer screening test (U.S. Preventive Services Task Force, 2014; U.S. Preventive Services Task Force, 2018).
    2.3. Health insurance
    EHR data contain information on coverage types and billable codes for services performed at each visit; as these data are used for billing purposes, they provide reliable information on insurance status and services received. Our analyses were stratified to assess the differential impact of Medicaid expansion on cancer screening among patients with Medicaid coverage, privately-purchased insurance, and those who re-mained uninsured. Other insurance types (e.g., Medicare for disability-eligible patients or grant programs that cover specific services HIV/ AIDS care) were excluded from the stratified analyses as eligibility is unrelated to the ACA Medicaid expansion. We used the payer at the last visit within the pre-ACA period and the post-ACA period. Of note, most patients seen in CHCs with private insurance likely directly purchased an individual plan rather than having employer-sponsored coverage (National Association of Community Health Centers, 2018).
    For race/ethnicity stratified results, we used the following cate-gories: non-Hispanic white, non-Hispanic black, and Hispanic. Other races were excluded as they represent < 10% of the patient population. CHCs are federally required to collect and report many individual-level demographic data variables to the US Health Resources and Services Administration to receive funding or designation under the Health Center Program. Therefore, EHR data from CHCs contain self-reported data on race/ethnicity, language, and FPL on most patients.
    We computed descriptive statistics for all patients eligible for cer-vical and/or colorectal cancer screening comparing patient character-istics between those living in expansion and non-expansion states. We calculated the adjusted prevalence of eligible adults who had appro-priate cervical and/or colorectal cancer screening in the pre- and post-periods, by expansion status. We then fitted logistic generalized esti-mating equation (GEE) models with robust sandwich variance estima-tors to obtain adjusted odds ratios (aORs) comparing post- versus pre-changes in screening within expansion groups, and difference-in-dif-ference (DID) estimates with 95% confidence intervals (CI) to test pre-post change between expansion groups. GEE models included an in-dicator for ACA period (pre vs post), Medicaid expansion status (ex-panded vs. did not expand), and the interaction between these vari-ables. Models were adjusted for age, race/ethnicity, FPL, health insurance type, urban/rural status, number of ambulatory visits, and sex (colorectal model only). GEE models implemented a robust sand-wich standard error estimator with a working independent correlation structure of health systems nested within states. We performed an overall model for each outcome that included all eligible patients and we further considered models stratified by insurance status and race/ ethnicity. We conducted a sensitivity analysis to assess whether the prevalence of screening changed in expansion and non-expansion states among patients with ≤138% FPL (Medicaid expansion eligibility). Since most patients in CHCs have an FPL ≤138%, the patterns of results were not altered (data not shown). Analyses were performed using SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA); all statistical tests were two-sided and significance was defined as a p-value < 0.05.