The outcomes of interest involved 5 indications of clinical complications in COVID-19 associated patients: hospitalization, maximum hospital length of stay (LOS), moderate ventilation, invasive ventilation, and death. Maximum LOS was created by calculating the difference in days between the start and end dates of each patient encounter and taking the maximum difference per patient. Maximum LOS was split further into two sub-groups: 1) those staying less than one day and 2) those staying one day or greater. Hospitalization was a binary indication (yes/no) and flagged “yes” for those patients that had a max LOS of one day or greater. Ventilation was a binary indication (yes/no) of whether a patient ever had a diagnosis, procedure, or encounter result indication that included reliance on either a 1) moderate ventilator (i.e., CPAP or BiPAP) or 2) invasive ventilator (i.e., intubation or tracheostomy). Death was a binary indication (yes/no) of whether a patient died at discharge or any time thereafter during the time of data collection (through June, 2020).

The primary predictor of interest was pregnancy status. This was a binary indicator (yes/no) of whether a patient had any ICD-9/10 diagnosis code of pregnancy no more than 7 and a half months before a qualifying COVID-19 “associated” encounter. This time frame ensured that the pregnancy was generally timely with the COVID-19 “associated” encounter.

The key clinical factor for our secondary aim was patient comorbidity, which was expressed via the Elixhauser comorbidity index (ECI) [19]. This index is analogous to the Charlson comorbidity index (CCI) [20], in that it measures patient comorbidity by calculating a risk-of-death score from each qualifying condition a given patient may have. The scores, from each condition, are then summed and weighted to provide a total score of mortality risk for the patient. Conditions for the ECI are defined by ICD-10 diagnosis codes. The ECI, however, uses a slightly different set of pre-existing conditions as well as a different weighting algorithm. The weighting of this score is performed using the Agency for Health Care Research and Quality (AHRQ) weighting methodology [21]. The ECI provides categorization of scores of less than 0, 0, 1–4, and 5 or higher [22] [23].

Demographic characteristics considered as potential confounders included age in years, race and ethnicity (non-Hispanic (NH) American Indian/Alaskan Native (AI/AN), NH Asian/Pacific Islander (API), NH Black/African American (Black), NH White, NH other/mixed/unknown race, Hispanic or Latino), insurance status (private, government/miscellaneous, Medicaid, Medicare, self-pay, missing), and 1-digit zip-code region. 1-digit zip-codes were grouped into four regions (northeast, southeast, midwest, west) for descriptive presentation of the data, but were left in their original form for inferential modeling. Additional clinical factors considered as potential confounders were binary (yes/no) indications of COVID-19 medication use: Hydroxychloroquine, Remdesivir, Decadron and Prednisone, Aspirin and Plavix, and anticoagulants. These were also compiled from Multum medication records. An additional clinical factor was a binary indication of whether the patient had a history (defined by ICD 9/10 diagnosis code) of gestational diabetes (DM).

Note: The content above has been extracted from a research article, so it may not display correctly.



Q&A
Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.



We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.