The epidemiology of drug-related hospital admissions in paediatrics - a systematic review

Abstract

Background: Despite previous efforts, medication safety in paediatrics remains a major concern. To inform improvement strategies and further research especially in outpatient care, we systematically reviewed the literature on the frequency and nature of drug-related hospital admissions in children.

Methods: Searches covered Embase, Medline, Web of Science, grey literature sources and relevant article citations. Studies reporting epidemiological data on paediatric drug-related hospital admissions published between 01/2000 and 01/2024 were eligible. Study identification, data extraction, and critical appraisal were conducted independently in duplicate using templates based on the 'Joanna Briggs Institute' recommendations.

Results: The review included data from 45 studies reporting > 24,000 hospitalisations for adverse drug events (ADEs) or adverse drug reactions (ADRs). Due to different reference groups, a total of 52 relative frequency values were provided. We stratified these results by study characteristics. As a percentage of inpatients, the highest frequency of drug-related hospitalisation was found with 'intensive ADE monitoring', ranging from 3.1% to 5.8% (5 values), whereas with 'routine ADE monitoring', it ranged from 0.2% to 1.0% (3 values). The relative frequencies of 'ADR-related hospitalisations' ranged from 0.2% to 6.9% for 'intensive monitoring' (23 values) and from 0.04% to 3.8% for 'routine monitoring' (8 values). Per emergency department visits, five relative frequency values ranged from 0.1% to 3.8% in studies with 'intensive ADE monitoring', while all other eight values were ≤ 0.1%. Heterogeneity prevented pooled estimates. Studies rarely reported on the nature of the problems, or studies with broader objectives lacked disaggregated data. Limited data indicated that one in three (median) drug-related admissions could have been prevented, especially by more attentive prescribing. Besides polypharmacy and oncological therapy, no other risk factors could be clearly identified. Insufficient information and a high risk of bias, especially in retrospective and routine observational studies, hampered the assessment.

Conclusion: Given the high frequency of drug-related hospitalisations, medication safety in paediatrics needs to be further improved. As routine identification appears unreliable, clinical awareness needs to be raised. To gain more profound insights especially for generating improvement strategies, we have to address under-reporting and methodological issues in future research.

Trial registration: PROSPERO (CRD42021296986).

Keywords: Adverse drug events; Adverse drug reactions; Children; Drug-related hospitalisations; Drug-related problems; Hospital admissions; Hospitalisation; Medication safety; Paediatric.

Fig. 1

Flow diagram based on “PRISMA 2020 flow diagram for new systematic reviews which included searches of databases, registers and other sources” [22]. Abstracts for which no reports could be retrieved were also assessed for eligibility. Therefore, in addition to reports, eligible abstracts were also included. This additional information is shown in blue

Fig. 2

Overview of the heterogeneity found in the 45 studies included

Fig. 3

Summary of the critical appraisal. Review authors’ judgements about each critical appraisal domain presented as percentages across all included studies. Critical appraisal was conducted on 45 included studies using a checklist that covered domains on 3 general criteria and 4 risk of bias criteria. The results, ‘low concerns’, ‘some concerns’, and ‘high concerns about issues or risks of bias’, have been pooled based on equal weighting of each study for each domain. Critical appraisal domains: G1: General applicability, G2: General reporting issues, G3: General issues in precision, R1: Risk of selection bias, R2: Risk of information bias due to attrition, R3: Risk of information bias due to issues with methods and measurements, R4: Risk of information bias due to issues with data analysis, Overall: R1-4: Overall risk of bias appraisal

Fig. 4

Frequency outcomes of drug-related hospitalisations, visualised by forest plots, accompanied by critical appraisal. Frequencies of DRHs as a proportion of inpatient stays in studies investigating ‘adverse drug events’ (A1) or ‘adverse drug reactions’ (A2). Frequencies of DRHs as a proportion of emergency department visits in studies investigating ‘adverse drug events’ (B1) or ‘adverse drug reactions’ (B2). Frequencies and 95% confidence intervals (Clopper-Pearson) were calculated from the data presented in the studies. It was refrained from reporting pooled estimates due to heterogeneity (based on visual inspection, statistical investigation of heterogeneity (I2 > 80%), too obviously different study designs and settings), and a high risk of bias, especially in routine studies. Abbreviations: No. Number of the frequency value, DRH Drug-related hospitalisation, CI Confidence interval, ED Emergency department. Stratification of the frequency values: No. 1–5, 40–44: frequency values of studies with a broader concept of ‘adverse drug events’, ‘with intensive monitoring’ (yellow background). No. 6–8, 45–49: frequency values of studies with a broader concept of ‘adverse drug events’, ‘based on routine monitoring’ (orange background). No. 9–31, 50: frequency values of studies with a narrower concept of ‘adverse drug reactions’, ‘with intensive monitoring’ (green background). No. 32–39, 51–52: frequency values of studies with a narrower concept of ‘adverse drug reactions’, ‘based on routine monitoring’ (blue background). Footnotes: Number¹: just abstract available, Number²: study identified via citation search, Number³: study without oncology patients. Critical appraisal: Critical appraisal domains: G1: General applicability, G2: General reporting issues, G3: General issues in precision, R1: Risk of selection bias, R2: Risk of information bias due to attrition, R3: Risk of information bias due to issues with methods and measurements, R4: Risk of information bias due to issues with data analysis, Overall: R1-4: Overall risk of bias appraisal

Fig. 5

Frequency of drug-related hospitalisations for the different groups, i.e., considering the different concepts of drug-relating problems (ADE vs. ADR), the monitoring method (intensive vs. routine monitoring), and the reference populations (inpatients vs. emergency department (ED) visits, and if applicable focused populations (only admissions with medication, acute admissions, admissions following emergency department visits, or acute admissions with medication). The frequency [%] of drug-related hospital admissions per group is displayed as median, range and interquartile range if there are more than 3 frequency values in the group. The background colours indicate: yellow – studies with a broader concept of ‘adverse drug events’, ‘with intensive monitoring’; orange – studies with a broader concept of ‘adverse drug events’, ‘based on routine monitoring’; green – studies a narrower concept of ‘adverse drug reactions’, ‘with intensive monitoring’; blue – studies with a narrower concept of ‘adverse drug reactions’, ‘based on routine monitoring’. ¹ Here, one outlier was excluded

Fig. 6

Detailed pattern of drugs implicated in drug-related hospital admissions. ATC codes (1st or 2nd level) were assigned to the drugs (drug classes) identified in nine studies and their relative frequencies were listed. Abbreviations: ATC code Anatomical Therapeutic Chemical (ATC) Classification, https://www.whocc.no/atc_ddd_index/, NR Not reported, Socio-Demographic Index, considered year = End of data collection, “Global Burden of Disease Study 2019” [93]