key: cord-0937436-ut4eq6xf
authors: Baughman, Derek; Zain, Areeba; Waheed, Abdul
title: Patient Adherence to Hemoglobin A(1c) Testing Recommendations in Telemedicine and In-Office Cohorts During COVID-19
date: 2021-09-30
journal: JAMA Netw Open
DOI: 10.1001/jamanetworkopen.2021.27779
sha: 10ff19bac56f7c5a0718a815b262dc613abaf6f0
doc_id: 937436
cord_uid: ut4eq6xf

This cohort study assesses patient adherence to diabetes screening using hemoglobin A(1c) level at in-person vs telemedicine encounters during the COVID-19 pandemic in the US.

Controlling for sampling bias: the importance of avoiding timeframe selection during COVID-19 As desirable, larger national data set such as Medicare claims were difficult to acquire short term (thus precluding timely snapshots of quality), this introduced a sampling bias: a single integrated healthcare system comprising 20,000 employees, 8 hospitals and 2,600 clinicians in >200 outpatient care sites across South-Central Pennsylvania. This bias was addressed with timeframe stratification by encounter type pre-pandemic (1/1/17-1/1/20) to compare volume and adherence rates intra-pandemic (1/1/20-1/1/21) (figure 1). Due to insufficient encounter numbers to establish a telemedicine comparison group (only 17 total telemedicine encounters from 2019-2020 and none prior to that), we observed the trend in office-only volume (pre-pandemic into the pandemic timeframe) to facilitate context for comparing intra-pandemic telemedicine volume and adherence rates to the office encounter equivalents (reasonable considering similar data trends in other recent studies on volume of primary care utilization 1 ). Importantly, unlike recent studies 2 , time frame selection was avoided in our data set since the NQF states that this can impair validity and reliability of results 3, 4 . To control for additional COVID-19 confounding, office only encounter data was compared to a 3-year historical adherence rate (figure 1).

In addition to age, ethnicity, sex and SDOH, an "overall adult risk score" was used as an indicator of health risk based on: (1) number of diagnoses (asthma, coronary artery disease, congestive heart failure, chronic obstructive pulmonary disease, chronic kidney disease, diabetes and hypertension), (2) SDOH (age, depression risk, tobacco use, financial resource strain risk, food insecurity risk, transportation needs risk, physical inactivity risk, intimate partner violence risk, social isolation risk, and alcohol risk score), (3) healthcare utilization (recent emergency room visits or hospital admissions). Additional SDOH were stratified by race to specifically evaluate housing and financial strain. The "average overall adult risk score" although not a validated scoring system (as it was designed by the WellSpan health system) was plausibly superior to other less granular standardized scoring systems 5,6 because in addition to comorbid conditions, it accounted for SDOH and healthcare utilization ( Office volume

Office adherence % Telemed adherence %

The metric "Hemoglobin A1c (HbA1c) testing" was selected from the Comprehensive Diabetes Care 7 to evaluate quality of T2DM management and screening. The national committee for quality assurance (NCQA) describes this measure as the percentage of patients with diabetes who received an HbA1c test during the measurement year (excluding hospitalized patients) 8 . To evaluate percentage of adherence in screening versus management of HbA1c lab, the CQMC measure description was modified to include all patients in the denominator. For all group's proportions of adherence, the denominator was defined as "encounters meeting inclusion criteria" and the numerator was defined as "encounters with a linked subsequent completed HbA1c lab order within 6months".

Using Epic's slicer dicer tool to mine patient encounter data in separate data sessions facilitated the comparison of office and telemedicine encounters (which included both phone and video visits). Results were "sliced" for patients with and without a diagnosis of T2DM (ICD-10 E11* code grouper) to compare management and screening HbA1c lab adherence respectively. Additional data sessions sliced the same population to compare family medicine HbA1c adherence to all other outpatient specialties. We avoided case-mix adjustments or timeframe selections in our data set since the NQF states that this can impair validity and reliability of results 3,4 . Medcalc's "N-1" Chi-squared calculator 9 was used to detect statistical significance for proportions of HbA1c adherence between sub-groups. 

Use and Content of Primary Care Office-Based vs Telemedicine Care Visits During the COVID-19 Pandemic in the US

Patient Characteristics Associated With Telemedicine Access for Primary and Specialty Ambulatory Care During the COVID-19 Pandemic

NQF: Quality Positioning System TM

Practical considerations on the use of the Charlson comorbidity index with administrative data bases

MedCalc's Comparison of proportions calculator