We conducted a multicenter, prospective, cross-sectional, observational and descriptive study focused on the identification and analysis of safety incidents in PEDs at the nationwide level in Spain.

Nine public hospitals of the National Health System of Spain, distributed across different autonomous communities, participated in the study. Participation was voluntary and supported by the Spanish Pediatric Emergency Research Group (RISeuP-SPERG) network, which facilitated the recruitment of centers by forwarding study information to pediatric emergency departments. Initially, 11 centers were expected to participate, but two withdrew due to the pandemic. All participating centers had PEDs that operated 24 h a day. Centers were included by convenience sampling, relying on the voluntary participation of professionals with previous training in patient safety and research experience. We recorded the number of health care professionals (pediatricians, nurses, assistants) on staff in each PED during the study period (Appendix B, Supplemental material 1). We calculated staff ratios per 1000 managed patients to allow comparisons between centers in relation to the European standard population of 2013 (Eurostat).11 To calculate the age-adjusted incidence rates (Appendix B, Supplemental material 2), we established the following age groups: <1 year, 1–4 years, 5–9 years, 10–14 years. The formula used for its calculation was: adjusted rate = Σ(Ri × Pi)/ΣPi, where Ri is the age-specific rate in group i and Pi the standard population in group i.

The study universe comprised all patients aged less than 14 years managed in the PEDs between April and June 2021. We identified 1102 patients, of who 49 did not complete follow-up, leaving a final sample of 1056 children. Of this total, 90 experienced at least one health care-related safety incident, with a total of 94 documented incidents, as four children experienced two incidents during the study period. We selected cases by opportunity sampling, stratifying by shift (morning, afternoon and night) and distributed randomly to ensure representativeness over time. We excluded patients who did not undergo any form of intervention or for whom it was not possible to obtain complete follow-up information.

There were two complementary phases in data collection: direct observation during the emergency care encounter, and follow-up after seven days, either by telephone or in person (if the patient was still hospitalized). This strategy made it possible to detect both incidents observed during the health care encounter, in real time, and those that manifested after discharge.

To collect the information, we used the version of the instrument developed from the data collection form of the ERIDA study12 validated and adapted for safety incident reporting in pediatric care settings.13 The questionnaire is structured in different sections to collect data on the sociodemographic characteristics of the patient, the characteristics of the health care encounter, the characteristics, clinical impact and degree of preventability of the incident, the contributing factors and whether or not the incident was documented in the health records.

The primary outcome was the incidence of safety incidents, defined as the proportion of patients who experienced at least one adverse event or incident that could have led to harm during the health care encounter or the follow-up:

We also collected data on clinical variables (reason for PED visit, triage level, interventions performed), organizational variables (shift, day of the week, patient origin), and incident-related variables: type, severity, preventability, need for follow-up care and documentation in health record.

Preventability was assessed independently by two researchers using the criteria of the trigger tool applied in the Canadian Paediatric Adverse Events Study1 and classifying each incidence as “nonpreventable” “potentially preventable” or “definitely preventable”. The operational definitions used to define incidents can be found in section 3 of the Supplemental material (Appendix B).

We performed a descriptive analysis calculating absolute and relative frequencies for qualitative variables and measures of central tendency and dispersion for quantitative variables. We calculated 95% confidence intervals for the main proportions. Since the objective of the study was strictly descriptive, we did not carry out any inferential analyses.

The statistical analysis was performed with the IBM SPSS Statistics software package, version 25 software (IBM Corp; Armonk, NY, USA), considering a percentage of missing data of less than 5% acceptable.

The study was approved by the Clinical Research Ethics Committee of the coordinating center (code CEIC-ARX-2021/04). We obtained informed consent from parents or legal guardians and guaranteed confidentiality in accordance with Organic Law 3/2018 on the Protection of Personal Data and the Guarantee of Digital Rights.

The mean (SD) age of the patients who experienced some form of health care-related incident was 3.7 (1.3) years, with an even sex distribution. The highest percentage of incidents occurred during the morning shift and most of the patients were self-referred. Table 1 summarizes the sociodemographic and health care encounter characteristics for patients who experienced safety incidents.

Of the 1054 patients included in the study, 90 (8.5%) experienced at least one safety incident during the emergency care visit or the subsequent follow-up. In most cases, only one incident was documented (95.6%), although some patients experienced two events (4.4%); there were no cases with more than two incidents. The overall incidence was 8.5% (95% CI, 6.0%–9.0%).

Most incidents were detected during the health care encounter at the PED, with no documented cases of incidents occurring exclusively during follow-up. With regard to type, incidents were most frequently related to general care, followed by incidents related to the physical examination, diagnosis and treatment. Table 2 shows the distribution of incidents by frequency, timing and type.

Specific causal factors were identified in most of the analyzed safety incidents. The most frequent were those related to organizational and administrative aspects (23%), followed by communication failures (18%), human factors related to individual health care workers (17%) and teamwork issues (13%).

Less frequently, we identified factors related to clinical documentation, insufficient clinical knowledge, available material resources and care transition processes. There were also 13 causal factors that were present in a single case each, amounting to 13.8% of the total incidents.

Patients who experienced safety incidents were significantly younger than those who did not experience any (mean [SD], 3.7 [1.3] vs 4.2 [1.8] years; P = .018). There were no significant differences in sex, triage level, shift, or patient origin.

In approximately one-fifth of cases (20%), adequate classification of the causal factor was not possible due to incomplete records or a lack of standardized coding at the time of reporting.

We found statistically significant differences in the incidence of safety incidents among the participating hospitals (P =  .004), hinting at the potential influence of center-specific organizational, structural or cultural factors (Table 3). The variation between centers ranged from 1.6% to 15.5%, despite the fact that all of them applied the same inclusion criteria and data collection methodology. We did not find a significant correlation between the ratio of health care staff per 1000 patients (Appendix B, Supplemental material 1) and the incidence of adverse events (r = −0.18; P =  .642), which suggests that other organizational factors may have a greater impact than staffing levels.

Patients who experienced safety incidents were significantly younger than those who did not experience any (mean [SD], 3.7 [1.3] vs 4.2 [1.8] years; P =  .018). There were no significant differences between patients who did and did not experience incidents in terms of sex, triage level, shift or patient origin (Table 4). On the other hand, there were no significant differences in the incidence of incidents based on shift (P =  .345) or patient origin (P =  .445). These findings support the hypothesis that structural factors or those related to safety culture at the level of the organization may play a greater role than particular circumstances or conditions in care delivery.

Similarly, there were no significant differences when we compared the incidence of incidents based on shift (morning, afternoon or night). The clinical severity of the patient on arrival, established based on the triage level, was also not associated with an increased probability of experiencing a safety incident (P =  .796).

We observed a statistically significant association between the impact of the incident and its preventability (P = .003). There was a higher proportion of incidents considered to be preventable, especially of those with clear evidence of being preventable, among the incidents with greater clinical impact. Fig. 1 provides a graphical representation of this association.

Figure 1.

Distribution of preventability levels by incident impact.

There were no significant differences in pretriage times (P =  .284) or care time (P =  .670) between preventable and nonpreventable incidents. However, the total time to discharge was significantly shorter in the group of preventable incidents (P =  .014), which could reflect a lesser clinical complexity or faster resolution in this type of event. Fig. 2 shows the comparison between the two groups.

Figure 2.

Comparison of health care encounter times (pretriage, care and time to discharge) by preventability of the incident. ns: not significant. *P <  .05.

There are some limitations to this study that should be taken into account. First, its descriptive design and the lack of multivariate analysis precluded the establishment of causal relationships between variables. Second, although the data collection instrument had been validated, there may have been interobserver variability due to the collection of data in multiple centers, despite the prior training of the participating health care professionals. We can also not rule out the possibility of under-recording or reporting bias, especially in centers with a lower incidence, which could reflect differences in reporting culture rather than in the actual occurrence of incidents. Differences between centers in the catchment population and health care staff characteristics could affect the comparability of the results, although the adjustment based on the European standard population partially mitigates this limitation.

Future studies should delve deeper into the structural and cultural factors that explain the variability between centers and prospectively analyze the association between certain health care processes and the occurrence of incidents. It would be useful to integrate qualitative analyses to explore barriers to reporting and the organizational context. Likewise, the application of mixed methodologies and the use of digital surveillance tools could improve the early detection of adverse events. Finally, it would be interesting to evaluate the impact of specific patient safety interventions on the incidence rate of incidents in pediatric emergency departments.