A simple, home-therapy algorithm to prevent hospitalisation for COVID-19 patients: A retrospective observational matched-cohort study

Background: Effective home treatment algorithms implemented based on a pathophysiologic and pharmacologic rationale to accelerate recovery and prevent hospitalisation of patients with early coronavirus disease 2019 (COVID-19) would have major implications for patients and health system. Methods: This academic, matched-cohort study compared outcomes of 90 consecutive consenting patients with mild COVID-19 treated at home by their family physicians between October 2020 and January 2021 in Northern and Central Italy, according to the proposed recommendation algorithm, with outcomes for 90 age-, sex-, and comorbidities-matched patients who received other therapeutic regimens. Primary outcome was time to resolution of major symptoms. Secondary outcomes included prevention of hospitalisation. Analyses were by intention-to-treat. Findings: All patients achieved complete remission. The median [IQR] time to resolution of major symptoms was 18 [14-23] days in the 'recommended schedule' cohort and 14 [7-30] days in the matched 'control' cohort (p = 0.033). Other symptoms persisted in a lower percentage of patients in the 'recommended' than in the 'control' cohort (23.3% versus 73.3%, respectively, p<0.0001) and for a shorter period (p = 0.0107). Two patients in the 'recommended' cohort were hospitalised compared to 13 (14.4%) controls (p = 0.0103). The prevention algorithm reduced the days and cumulative costs of hospitalisation by >90%. Interpretation: Implementation of an early home treatment algorithm failed to accelerate recovery from major symptoms of COVID-19, but reduced the risk of hospitalisation and related treatment costs. Given the study design, additional research would be required to consolidate the proposed treatment recommendations. Funding: Fondazione Cav.Lav. Carlo Pesenti (c) 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)