This retrospective single-center observational study was conducted at the Pediatric Emergency Department (PED) of the Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milan, Italy. The study covered the period between January 1st, 2022, and December 31st, 2022. In this hospital, blood laboratory tests, specialist consultations, and instrumental examinations (X-ray and CT scans) are available 24 h a day, every day. On the other hand, primary care pediatricians are only available on weekdays, at various times, to visit children outside the hospital. Therefore, to avoid any potential selection bias due to the different availability of primary care pediatricians, only visits occurring on Saturdays, Sundays, and public holidays were included in this study. At the study site, there are only two shifts on Saturdays, Sundays, and public holidays: a day shift from 8:30 a.m. to 8:30 p.m. and a night shift from 8 p.m. to 8 a.m. During these shifts, the staff consists of one emergency pediatrician, two pediatric residents, two nurses, and one social and health worker.

Data on the following services were collected: blood laboratory tests, specialist consultations, and instrumental examinations (X-ray and CT scans). These ancillary diagnostic procedures are requested through the department’s online system, making it possible to determine the time of each request. Additional examinations (e.g. ultrasounds) were excluded because they are not always available during the 24 h. Data on the urgency code (white, green, yellow and red) assessed during triage and the physician on duty (indicated by an anonymous code) during the shift were also collected. All data were collected by an automated extraction program to avoid reporting errors or missing data. Visits and physicians on duty were collected using anonymized codes.

Since decision making is typically constant over an 8-h shift [9], work shifts were divided into two major groups, corresponding to the first 8 h and the last 4 h of the shift. For each patient, we considered three binary variables: request for i. blood laboratory tests; ii. X-ray/CT scans; and iii. specialist consultations. Additionally, we assessed a variable that scored 1 if at least one of these procedures was requested. These variables were summarized using frequencies and percentages.

The Chi-square test was used to evaluate differences in diagnostic service requests during the two shift periods. Subsequently, mixed-effects logistic regression models were employed to investigate the probability of requesting ancillary diagnostic procedures in the last 4 h of the shift (independent variable), adjusted for the total number of patients during the whole shift and urgency code of the patient. The physician on duty was considered as random effect. Results were presented using odds ratios and their respective 95% confidence intervals. A two-tailed p value < 0.05 was considered significant. R software was used for the analyses.