Patient, facility, and community determinants of timely follow up after abnormal screening mammograms in Appalachia: A multilevel analysis.

PURPOSE: Examine the extent that community and organizational characteristics influence mammogram recall and biopsy rates in Appalachia.

METHODS

We collected facility survey and Medicare claims (2016-2018) data on 191 FDA-registered breast imaging organizations serving 2.65 million women in 5 Appalachian states (KY, OH, PA, VA, and WV). Primary outcomes were recall rates and time to recall and biopsy. Generalized linear mixed-effects models (GLMM) (patient level with facility random intercepts) and facility-level structural equation models were used to evaluate multilevel predictors of recall, biopsy, and timeliness. Imaging center service areas (SAs) were constructed using Medicare claims for patient, facility, and community characteristics utilization patterns. Mammogram service capacity was measured using a published tool (BIOPSI).

FINDINGS: Rural centers had SAs with higher area deprivation index (ADI) scores, lower capacity despite possessing more machines per capita, and were less likely to offer 3D tomosynthesis (3DT). Recall (8%) and biopsy (∼22%) rates were similar by metropolitan status; however, rural centers were more likely to provide same-day exams (P = .021). In GLMM, patient-level predictors included younger age and nonuse of 3DT for recall; and higher ADI and use of 3DT for biopsy. Days to recall was lengthened with dual Medicaid insurance, 3DT, and screening by a stand-alone facility; while days to biopsy was lengthened without the use of 3DT. Community-level pathways were found for days to recall, with rural and smaller facilities achieving significantly fewer days.

CONCLUSIONS

Characteristics of rural imaging centers influence efficiency through their resources and community contexts and should be considered in breast cancer screening outcomes research.