Patterns of repeated diagnostic testing for COVID-19 in relation to patient characteristics and outcomes

Abstract

Background: Whilst the COVID-19 diagnostic test has a high false-negative rate, not everyone initially negative is re-tested. Michigan Medicine, a primary regional centre, provided an ideal setting for studying testing patterns during the first wave of the pandemic.

Objectives: To identify the characteristics of patients who underwent repeated testing for COVID-19 and determine if repeated testing was associated with downstream outcomes amongst positive cases.

Methods: Characteristics, test results, and health outcomes for patients presenting for a COVID-19 diagnostic test were collected. We examined whether patient characteristics differed with repeated testing and estimated a false-negative rate for the test. We then studied repeated testing patterns in patients with severe COVID-19-related outcomes.

Results: Patient age, sex, body mass index, neighbourhood poverty levels, pre-existing type 2 diabetes, circulatory, kidney, and liver diseases, and cough, fever/chills, and pain symptoms 14 days prior to a first test were associated with repeated testing. Amongst patients with a positive result, age (OR: 1.17; 95% CI: (1.05, 1.34)) and pre-existing kidney diseases (OR: 2.26; 95% CI: (1.41, 3.68)) remained significant. Hospitalization (OR: 7.88; 95% CI: (5.15, 12.26)) and ICU-level care (OR: 6.93; 95% CI: (4.44, 10.92)) were associated with repeated testing. The estimated false-negative rate was 23.8% (95% CI: (19.5%, 28.5%)).

Conclusions: Whilst most patients were tested once and received a negative result, a meaningful subset underwent multiple rounds of testing. These results shed light on testing patterns and have important implications for understanding the variation of repeated testing results within and between patients.

Keywords: COVID-19; diagnostic testing; false-negative rate.

Fig. 1

Patterns of COVID‐19 testing at Michigan Medicine. (a) Number of tests by day, stratified by test result for the 15 920 patients presenting to Michigan Medicine before 4 June 2020. Between 10 March and 4 June, 19 540 tests were administered. (b) Proportion of patients tested once versus more than once amongst for patient outcomes of increasing severity: negative test result, positive test result, hospitalization, intensive care unit admission, and death.

Fig. 2

Characterization of repeated testing patterns. (a) Unique sequences of testing results for the 19,540 tests ordered between 10 March and 4 June, administered to the 15 920 patients presenting to Michigan Medicine before 4 June 2020. (b) Distribution of the number of tests administered and (c) average wait time for these tests, corresponding to each unique test result sequence in (a). The solid vertical line in (c) is the overall mean wait time in days. SD: standard deviation. (d) Test results for the five patients who underwent at least 10 tests. The horizontal axis represents the test date (first test was administered on ‘Day 0’ for each patient), and the vertical axis represents each patient ID.

Fig. 3

Associations between repeated testing and patient characteristics. Odds ratios and 95% confidence intervals are reported for each characteristic, fully adjusting for all other demographic and clinical characteristics in a logistic regression model. OR: odds ratio; CI: confidence interval; L: lower confidence limit; U: upper confidence limit; NSES: neighbourhood socio‐economic status. ‘All Tested’ and ‘P All’ refer to the full study cohort of n = 15 920 patients, whilst ‘Positive’ and ‘P Positive’ refer to the 1167 patients with a positive COVID‐19 test in their medical history.