The impact of rapid diagnostic testing on hospital administrative coding accuracy for influenza

Abstract

Background: Hospital administrative coding may underestimate the true incidence of influenza-associated hospitalisation. Earlier availability of test results could lead to improved accuracy of administrative coding.

Methods: In this study we evaluated International Classification of Diseases 10 (ICD-10) coding for influenza (with [J09-J10] or without [J11] virus identified) in adult inpatients who underwent testing in the year prior, compared to those in the 2.5 years after, the introduction of rapid PCR testing in 2017. Other factors associated with influenza coding were evaluated using logistic regression. Discharge summaries were audited to assess the impact of documentation and result availability on coding accuracy.

Results: Influenza was confirmed by laboratory testing in 862 of 5755 (15%) patients tested after rapid PCR introduction compared with 170 of 926 (18%) prior. Following the introduction of rapid testing there was a significant increase in patients allocated J09 or J10 ICD-10 codes (768 of 860 [89%] vs 107 of 140 [79%], P = 0.001). On multivariable analysis, factors independently associated with correct coding were rapid PCR testing (aOR 4.36 95% CI [2.75-6.90]) and increasing length of stay (aOR 1.01, 95% CI [1.00-1.01]). Correctly coded patients were more likely to have documentation of influenza in their discharge summaries (95 of 101 [89%] vs 11 of 101 [10%], P < 0.001) and less likely to have pending results at discharge (8 of 101 [8%] vs 65 of 101 [61%], P < 0.001).

Conclusion: The introduction of rapid PCR testing for influenza was associated with more accurate hospital coding. One possible explanation is faster test turnaround leading to improvement in clinical documentation.

Keywords: Epidemiology; Hospital coding; Influenza; Molecular diagnostics.