The Ongoing Impact of COVID-19 on US Dermatology Practices

Abstract

Background: COVID-19 is significantly impacting healthcare delivery worldwide. Chen et al anecdotally reported the impact on dermatology outpatient care at the outbreak epicenter in Wuhan, China, but nothing has yet been assessed for the US. The purpose of this study was to determine the magnitude of the ongoing impact of COVID-19 on US dermatology outpatient care. Methods: After pre-validation, 2 surveys comparing outpatient volumes and scheduling issues for the weeks of February 17th versus the week of March 16th, 2020 (Survey 1) and April 13th, 2020 (Survey 2) and for estimation of trends in the next several weeks was emailed to 9,891 US Dermatologists on 3/21(Survey 1) and 4/18(Survey 2)(Tables S1a & S1b). Because of the importance of this information and the need for rapid dissemination, only data from the first 1,000 respondents (collected in the initial 36 hours) were included in each survey. In Survey 1, 30 responses were removed due to ineligible geography or errors in survey entry, leaving 970 for the analysis. Survey 2 consisted of 1,000 eligible respondents. Demographics (Table 1) representativeness with AAD membership was confirmed (Table S2). Statistical significance was calculated using chi-square, difference-of-proportions, and two-tailed independent t-tests. Results: COVID-19 impact was material (Table 2). From the 3rd week in February to the 3rd week in March to the 3rd week in April, the average number of patients seen fell from 149.4 to 63.4 to 28.2(p<0.0001), practice days from 4.2 to 3.1 and then rose to 3.5(p<0.0001) and biopsies from 19.8 to 7.7 to 3.5(p<0.0001). Although by 3/16 there were only 24.5k cases nationally, the early-phase decrease in patient volume and office days suggests the magnitude of disease concern impact was greater than actual prevalence. Postponement of non-essential appointments increased from 35.5% to 79.4% to 95.6%(p<0.00001). In Survey 1, 66.3% of respondents estimated a >50% decrease in patient volume in the coming 2 weeks (18.9% completely closing practices) and, disturbingly, 47.2% of respondents in the 2nd survey estimated an additional 50%+ decrease in patient volume in the next 2 weeks. 54.6% (Survey 1) of postponed appointments were for >4 weeks with an additional 25.4% not rescheduled. Discussion: A greater negative impact was found in US “hotspot” regions (36% (Survey 1) and 34% (Survey 2) of respondents-Figure S1) for week 3/16-20 for practice days (3.0 hotspots vs. 3.3 non-hotspots) and patients seen (56.2 in hotspots vs. 70.0 in non-hotspots); and for week 4/13-18 (3.4 in hotspots vs 3.5 in non-hotspots) and patients seen (25.3 in hotspots vs 29.7 in non-hotspots). No significant differing telemedicine usage (39.5% hotspots vs 37.2% non-hotspots) or practice closure (21.0% hotspots vs 17.6% non-hotspots) was found in Survey 1 (March); however, a significant difference in telemedicine usage (54.5% hotspots vs 45.5% non-hotspots) and practice closure (25.4% hotspots vs 16.4% non-hotspots, when compared to a typical April week) was found in Survey 2 (April). Mean estimated telemedicine visits overall for the next 2 weeks was 37.8% (Survey 1) and 45.9% (Survey 2). Academic/University/Institutional dermatologists were significantly more likely to use telemedicine (Survey 1=57.1%, Survey 2=68.6%) than private practitioners (Survey 1=35.5%, Survey 2=46.2%). Telemedicine usage was less likely for dermatologists with >30 practice years (>30=32.4% vs 40.0%) and this trend continued in April with only 37.2% of more experienced dermatologists using telemendine. However, telemedicine usage does not have an impact on the deferred/postponed biopsies that had already occurred during the March (mean=10.7) or April (mean=7.9) weeks as well as those predicted to be subsequently postponed. Limitations include that this study reflects a “snapshot” which could materially change given the dynamically evolving situation. Estimations could have led to recall bias and the 10.1% response rate could have introduced sampling and non-response bias. Those with lower work volumes could have been more likely to have time to respond, but this bias was minimized by weekend-only data collection. However, the large sample size and representative distribution mitigate selection bias and standard statistical testing demonstrated significance. Conclusion: Our findings demonstrate the significant early impact of COVID-19 on US dermatologic care and can help better understand national trends. With an estimated 49.9 million annual US dermatology office visits, the 50%+ decrease in predicted visits could be devastating. Beyond telemedicine, other innovative approaches will need to be developed and implemented to help delivery of essential dermatology care during this crisis.