Back

Cohort profile: A multicenter evaluation of clinical decision rules applied to emergency department triage of patients presenting with acute respiratory infection or infectious diarrhea

Berthelot, S.; Boissinot, M.; Bergeron, M. G.; Vachon, M.-L.; Trottier, S.; Huletsky, A.; Gilca, R.; Guertin, J. R.; Tremblay, C.; Longtin, Y.; Afilalo, M.; Mercier, E.; Dube, E.; Simonyan, D.; Singbo, N. M. U.; Bluteau, A.

2023-10-13 emergency medicine
10.1101/2023.10.12.23296964 medRxiv
Show abstract

PurposeEmergency department (ED) patients suffering from acute respiratory infection or infectious diarrhea often present with self-limiting conditions. The study objective was to evaluate the performance of triage clinical decision rules consisting of a rapid molecular test and a self-administered patient questionnaire to identify ED patients who can self-treat at home without consulting an emergency physician. This article describes the profile of the cohorts recruited. ParticipantsParticipants were prospectively recruited in 4 EDs in Quebec City and Montreal, Canada, from February 2022 through March 2023. Participants were aged [≥]18 years, had an acute respiratory infection and/or acute infectious diarrhea, and had received a Canadian Triage and Acuity Scale score between 3 (urgent) and 5 (non-urgent). Participants were asked to complete a self-administered risk stratification questionnaire after triage and to follow usual ED care afterward. Nasopharyngeal and/or rectal swabs were collected and frozen for subsequent testing on a rapid molecular testing device. Data were obtained during the recruitment visit, during a follow-up phone call 7 days later and from medical records. The primary outcome to be predicted by the clinical decision rules was an aggregation of hospitalization, return visit and mortality at 7 days. Findings to dateWe recruited 1,391 participants, 62.3% of whom were women, 80.7% were aged under 60, 78.2% had no comorbidities, 76.5% presented with an acute respiratory infection, 17.8% with an acute infectious diarrhea and 5.7% with both. Hospitalization and return visits incidence proportions at 7 days were respectively 10.8% and 13.1% for respiratory infections and 14.1% and 16.5% for infectious diarrhea. No death was recorded. Future plansThe data gathered from these cohorts will enable us to test, refine, derive, and validate clinical decision rules used to help ED triage nurses offer the most suitable care to patients presenting with acute respiratory infections or infectious diarrhea. Strengths and limitationsOur study has both strengths and limitations. Among the strengths: O_LIThe cohorts were recruited from 4 different EDs and reached the target sample size for acute respiratory infections and acute infectious diarrhea. C_LIO_LIThe potential economic impact of the clinical decision rules will be assessed from the perspective of both the health system and the patient. C_LI The main limitations are the following. O_LICohorts were recruited by convenience sampling and may not be representative of the entire ED population. C_LIO_LIThe patient self-administered questionnaires used in this study were derived from systematic reviews and rapid prototyping, but not according to the methodological standards recommended for the derivation of clinical decision rules. However, the study dataset was built to enable rules to be refined and if necessary, new rules to be derived and internally validated. C_LIO_LIWe recorded a 12.9% loss of participants at the 7-day follow-up phone call. However, the primary outcome measures (return visits, admissions and deaths) will be obtained from provincial administrative databases. These reliable data will enable us to overcome this limitation for future projects to refine and validate robust triage clinical decision rules. C_LI

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.