Medication errors cause appreciable morbidity and mortality in children. The objective was to determine the effectiveness of interventions to reduce pediatric medication errors, identify gaps in the literature, and perform meta-analyses on comparable studies.
Relevant studies were identified from searches of PubMed, Embase, Scopus, Web of Science, the Cochrane Library, and the Cumulative Index to Nursing Allied Health Literature and previous systematic reviews. Inclusion criteria were peer-reviewed original data in any language testing an intervention to reduce medication errors in children. Abstract and full-text article review were conducted by 2 independent authors with sequential data extraction.
A total of 274 full-text articles were reviewed and 63 were included. Only 1% of studies were conducted at community hospitals, 11% were conducted in ambulatory populations, 10% reported preventable adverse drug events, 10% examined administering errors, 3% examined dispensing errors, and none reported cost-effectiveness data, suggesting persistent research gaps. Variation existed in the methods, definitions, outcomes, and rate denominators for all studies; and many showed an appreciable risk of bias. Although 26 studies (41%) involved computerized provider order entry, a meta-analysis was not performed because of methodologic heterogeneity. Studies of computerized provider order entry with clinical decision support compared with studies without clinical decision support reported a 36% to 87% reduction in prescribing errors; studies of preprinted order sheets revealed a 27% to 82% reduction in prescribing errors.
Pediatric medication errors can be reduced, although our understanding of optimal interventions remains hampered. Research should focus on understudied areas, use standardized definitions and outcomes, and evaluate cost-effectiveness.
Comments
Author Response to: 'Are the designs of the included studies really as robust as suggested?'
Dear Editor,
We appreciate the comments by Maaskant et al.[1] regarding the studies included in our manuscript, 'Interventions to Reduce Pediatric Medication Errors: A Systematic Review.'[2] A goal of this systematic review was to be as inclusive as possible, as evidenced by our medication error definition, which was also necessarily broad. As with any systematic review, researchers face a tradeoff between broad definitions which allow multiple studies to be included and strict definitions which increase rigor but could exclude important research findings conducted in less than ideal settings. As we wrote in the discussion section, "We appreciate that not all quality improvement research projects can meet every metric regarding high-quality, bias-free studies as laid out by the Cochrane Effective Practice and Organization of Care Review Group guidelines.[3]"
We point readers to Figure 2, which describes the high risk of bias in many of the included studies, across all study types. This figure highlights, for example, that 88% of studies we labeled as interrupted time-series did not analyze data appropriately and that 63% of studies we labeled as controlled before-after studies did not adequately create comparable control study sites. Given this high risk of bias across studies, we wrote in our discussion, "The authors would recommend investigating each relevant study (Table 1) to clearly understand its applicability and context before drawing policy-level conclusions."
We believe the conclusions from this work would remain valid even if many of the included studies were relabeled as "before-after studies." Namely, "Despite a large increase in the number of published studies aiming to reduce pediatric medication errors since 2005, our knowledge of interventions to prevent pediatric medication errors remains hampered by non-uniform definitions, non-uniform data collection methodology, and non- uniform outcome reporting. The heterogeneity in current pediatric medication error intervention studies prevents wide generalizability of results and yields unclear guidance to hospitals on which interventions are best to adopt." We believe the central message of this work, that pediatric medication errors can be reduced but appreciable work remains to overcome current methodological shortcomings, is reinforced by the observations of Maaskant et al.[1]
References:
1.Maaskant JM, Soe A, Vermeulen H. Are the designs of the included studies really as robust as suggested?[E-letter], Pediatrics (September 12, 2014), http://pediatrics.aappublications.org/content/134/2/338/reply#pediatrics_el_63912. (accessed September 15, 2014).
2.Rinke ML, Bundy DG, Velasquez CA, Rao S, Zerhouni Y, Lobner K, et al. Interventions to reduce pediatric medication errors: a systematic review. Pediatrics. 2014;134(2):338-360.
3.McAuley L, Ramsay C, Mayhew A. Cochrane Effective Practice and Organization of Care Review Group: Data Collection Checklist. http://epoc.cochrane.org/sites/epoc.cochrane.org/files/uploads/datacollectionchecklist.pdf. Published 2002. Accessed June 26, 2013.
Conflict of Interest:
None declared
Are the designs of the included studies really as robust as suggested?
Dear editor,
With interest we read the systematic review on interventions to reduce pediatric medication errors by Rinke et al. (1). We commend the authors on their work to summarize the available evidence on this important subject. However we would like to express our concern on the methodology of the included studies in this review. There is no doubt systematic reviews are important for healthcare professionals, policy makers and researchers to get well informed on the current "state-of-the-art" on a certain subject. Therefore the quality of a systematic review is extremely important as those stakeholders might base their decisions on the presented results. The Cochrane Collaboration developed rigorous methodology and formulated valuable recommendations to ensure high reliable information. One of the recommendations is to include exclusively studies with a robust study design, i.e. randomized controlled trials (RCT), interrupted time series (ITS) and controlled before-after studies (CBA) (2). Rinke et al. included 63 studies of which the authors considered 52 ITS and eight CBA studies. As ITS and CBA are sparsely executed to study quality and safety interventions, we were surprised by these high numbers and decided to take a closer look at the included studies. Based on this closer investigation we disagree with the results as nearly all ITS and CBA studies do not meet the definitions and characteristics of the designs meant by the Cochrane Collaboration (2,3,4). Both in ITS and CBA studies, observations are made before and after an intervention. An ITS study uses multiple observations before and after an intervention and includes these time points in the analyzes, therewith adjusting the results for temporal trends. Segmented regression analysis is recommended for this type of study design and the statistics should present differences in level and trend (5). In a CBA study the observations before and after an intervention are performed in a study group and a comparable control population. Analyses compare the observations between the pre-intervention and post-intervention period and between the study and control group, e.g. by means of rate ratios. Well performed ITS and CBA studies strengthen the evidence that the observed changes are indeed related to the intervention, and are not the result of temporal trends or events that occurred contemporaneously. In most ITS studies included in the SR by Rinke et al. the observations are not analyzed as time series or do not include a control population, and therefore are simple before-after studies. Before-after studies are less robust, the results less credible, and thus should be removed from the included study list. Reconsidering this, the number of studies would be drastically reduced to two ITS studies and one CBA study only, when the right definitions of the study designs are applied. The results of the SR must be read with caution as the underlying studies are not identified and appraised accurately due to application of wrong definitions.
References
(1) Rinke ML, Bundy DG, Velasquez CA, Rao S, Zerhouni Y, Lobner K, Blanck JF, Miller MR. Interventions to reduce Paediatric Medication errors: a systematic review. Pediatrics, 2014;134(2):338-360.
(2) Higgins J, Green S. Cochrane Handbook for Systematic Reviews of Interventions version 5.1.0, updated March 2011.
(3) Eccles M, Grimshaw J, Campbell M, Ramsay C. Research designs for studies evaluating the effectiveness of change and improvement strategies. Quality and Safety in Health Care, 2003;12:47-52.
(4) Fan E, Laupacis A, Pronovost PJ, Guyatt GH, Needham DM. How to use an article about quality improvement. JAMA, 2010;304(20):2279-2287.
(5) Ramsay CR, Matowe L, Grilli R, Grimshaw JM, Thomas RE. Interrupted time series designs in health technology assessment: lessons from two systematic reviews of behavior change strategies. International Journal of Technology Assessment in Health Care, 2003;19(4):613-623.
Conflict of Interest:
None declared