key: cord-0879042-607n5q6d
authors: Elmunzer, B. Joseph; Wolf, Bethany J.; Scheiman, James M.; Tierney, William M.; Taylor, Jason R.
title: Association between pre-admission acid suppressive medication exposure and severity of illness in patients hospitalized with COVID-19
date: 2020-11-12
journal: Gastroenterology
DOI: 10.1053/j.gastro.2020.11.007
sha: 84fa836b6812e0b9fa10f3b322b303471564313a
doc_id: 879042
cord_uid: 607n5q6d

nan

Clinical data from the time of symptom onset until discharge, death, or the end of the study period, including the use of PPI or H2RA within one month of admission, were manually abstracted through review of electronic health records. Data abstraction was performed by clinical research coordinators, medical students, internal medicine and gastroenterology trainees, as well as faculty gastroenterologists, depending on site.

Every site had a designated clinician investigator who oversaw and vouched for the integrity of data collection. Data quality was assured to the greatest extent possible using a multi-faceted strategy that has been described previously and included manual review of all incoming data by a dedicated data management specialist (Elmunzer, et al. Clin Gastroenterol Hepatol 2020) . The data collection form is provided in this appendix.

The association between acid suppressive medication use and mechanical ventilation or death was evaluated using two separate multivariable generalized linear regression models. PPI and H2RA use -the independent variables of interest -as well as age, race, and sex were included in both models. Lasso regression -an agnostic, datadriven approach to variable selection -was used to select the other covariates. Only laboratory values at admission were included as covariates in the models because inhospital laboratory values (highest/lowest) could have occurred after mechanical ventilation (a primary outcome) and could have been confounded by in-hospital treatments. To account for clustering by center, generalized estimating equation (GEE) models with a random center effect were employed.

To assess the validity of findings from the primary regression analysis, we performed a propensity score matched analysis of the association between PPI or H2RA use with mechanical ventilation or death. Propensity scores were estimated for PPI or H2RA use based on the predicted probability of exposure, estimated from generalized linear mixed models with either PPI/H2RA use as the outcome and all other patient factors (excluding mechanical ventilation and death) as predictors. A random center effect was employed in this model as well. Four-to-one control to treatment caliper matching with caliper = 0.2 was applied to the propensity scores to obtain matched patients by H2RA or PPI use.

Missing data were assumed to be missing at random and were imputed prior to model development. In the primary analysis, we excluded patients whose PPI and/or H2RA use within one month of admission was unknown. Sensitivity analyses were subsequently conducted to evaluate the validity of excluding patients with unknown PPI/H2RA status and of imputation. 1) Please complete the below data collection form (DCF) in REDCap at the time of discharge or death. Data will appear in the DMC19 database once entry and verification are complete.

2) We aim to capture inpatients with a confirmed COVID-19 diagnosis, regardless of whether they have digestive manifestations. After prevalence is defined in hospitalized patients, and as the numbers grow, we may focus on patients who are known to have GI manifestations and/or include outpatients.

3) Please make all efforts to collect data on the first 50-100 consecutive patients at your hospital or health system. 4) Eligible patients can and should be identified by any means necessary, which may include, but is not limited to, institutional laboratory records, data warehouse queries, electronic health record research subject identification tools/dashboards, and discussions with the infectious disease or critical care services, etc. You may elect to use the emergency ICD-10 code of U07.1 -2019-nCov acute respiratory disease -to help identify eligible patients. 5) Please triple check data for accuracy before submission. Although we are performing central data monitoring, we cannot verify incoming data against source documents, nor are we performing on-site monitoring visits. Therefore, the overall quality of the data is assured primarily at the site level. 6) Along the lines of #5, coordinators should confer with a clinician during data collection to ensure that clinical context is accounted for as much as possible in the interpretation of questions that involve an element of subjectivity.

7) All data fields should have affirmative, negative, and unknown options. Therefore, missing data will be assumed to be inadvertent and this will generate a query. 8) Please maintain a secure key at your site that allows patient identification on the basis of subject ID#. This may be used in the future for to collect data pertaining to long-term outcomes.

Age ( Yes -Mild to moderate respiratory compromise Yes -Severe respiratory compromise Yes -Congestive heart failure/cardiomyopathy Yes -Myocardial infarction No Unknown Endoscopic findings (please list abnormal findings in impression section of endoscopy report(s)) Histologic findings (please list abnormal findings in impression section of pathology report(s))

Prior to COVID-19, ideally when patient was healthy: White blood cells (WBC) Hemoglobin Platelets Aspartate aminotransferase (AST) Alanine aminotransferase (ALT) Alkaline phosphatase (ALK-phos) Total bilirubin International normalized ratio (INR) Albumin Factor V level Lipase

MSc (37)

Creatinine At time of hospital admission: White blood cells (WBC) Hemoglobin Platelets Aspartate aminotransferase (AST) Alanine aminotransferase (ALT) Alkaline phosphatase (ALK-phos) Gamma-Glutamyl Transferase (highest) U/L Total bilirubin Direct bilirubin International normalized ratio (INR) Albumin Factor V level Lipase -What is the upper limit of normal Amylase -What is the upper limit of normal Creatinine Highest or lowest level during illness: WBC (highest and lowest) Hemoglobin (lowest) Platelets (lowest) AST (highest) ALT (highest) ALK-phos (highest) Gamma-Glutamyl Transferase (highest) U/L Total Bilirubin (highest) Direct Bilirubin (highest) INR (highest) Albumin (lowest) Factor 5 (lowest) Lipase (highest) -What is the upper limit of normal Amylase (highest) -What is the upper limit of normal Creatinine (highest) Absolute lymphocyte count (lowest) C-Reactive Protein (CRP) (highest) Procalcitonin (highest) Troponin (highest) Ferritin (highest) ng/mL or ug/L Interleukin-6 (highest) pg/mL Duration between first onset of symptoms and highest AST (days) Duration between first onset of symptoms and highest ALT (days) Duration between first onset of symptoms and highest total bilirubin (days) Duration between first hospital day and highest AST (days) Duration between first hospital day and highest ALT (days) Duration between first hospital day and highest total bilirubin (days) Abnormal LFTs were Still close to max at discharge/death Improved but not resolved at discharge/death Resolved or close to resolved at discharge/death Not applicable Were the increased LFTs suspected to be due to a drug reaction (based on review of progress/consult notes)? If yes, histologic findings (please list abnormal findings in impression section of pathology report(s)).