Portal de Periódicos da CAPES

Effect of clinician attention switching on workload and wrong-patient errors

2022; Elsevier BV; Volume: 129; Issue: 1 Linguagem: Inglês

10.1016/j.bja.2022.04.012

1471-6771

Sunny S. Lou, Seunghwan Kim, Derek Harford, Benjamin C. Warner, Philip Payne, Joanna Abraham, Thomas Kannampallil,

Patient Safety and Medication Errors

Editor—Clinical work environments, especially intensive care settings, are characterised by multiple competing demands, requiring clinicians to switch their attention between overlapping tasks. Previous time-and-motion studies have suggested that ICU clinicians perform ∼150 tasks per hour.1Carayon P. Wetterneck T.B. Alyousef B. et al.Impact of electronic health record technology on the work and workflow of physicians in the intensive care unit.Int J Med Inf. 2015; 84: 578-594Crossref PubMed Scopus (92) Google Scholar Given the limitations of human cognition, attention switching has been shown to be associated with decreased productivity and performance, increased cognitive burden, and errors in non-clinical settings.2Monsell S. Task switching.Trends Cogn Sci. 2003; 7: 134-140Abstract Full Text Full Text PDF PubMed Scopus (2373) Google Scholar However, the consequences of attention switching are less understood in clinical settings due to reliance on labour-intensive observational studies to measure attention switching. In this study, we developed a scalable metric for attention switching based on passively collected electronic health record (EHR) audit log data,3Adler-Milstein J. Adelman J.S. Tai-Seale M. Patel V.L. Dymek C. EHR audit logs: a new goldmine for health services research?.J Biomed Inform. 2020; 101: 103343Crossref PubMed Scopus (43) Google Scholar and used it to assess the downstream effects of attention switching in ICU clinicians. We hypothesised that a higher rate of attention switching is associated with more time spent using the EHR and more wrong-patient errors. This was retrospective, observational study conducted across four surgical ICUs encompassing 85 beds at Barnes-Jewish Hospital, a large academic medical centre in St. Louis, MO, USA. This study was approved by the institutional review board of Washington University (IRB# 202009032) with a waiver of informed consent. EHR audit logs (Epic Systems, Verona, WI, USA), which capture all EHR-based click activities of clinicians resulting in the display or modification of patient data, were extracted for all attending physicians and advanced practice providers (APPs; i.e. nurse practitioners and physician assistants) who worked at least one ICU shift in 2019. Audit logs record the timestamp, user, and patient for each EHR click event (Supplementary Figure S1); as such, they provide a comprehensive record of clinical activities, and have been used to describe clinical workflow and workload.3Adler-Milstein J. Adelman J.S. Tai-Seale M. Patel V.L. Dymek C. EHR audit logs: a new goldmine for health services research?.J Biomed Inform. 2020; 101: 103343Crossref PubMed Scopus (43) Google Scholar Audit logs were analysed at the level of each 12-h ICU shift. Audit logs for each shift were subdivided into sessions based on groups of consecutive activities separated by <5 min of inactivity, which were assumed to represent continuous EHR use.4Ouyang D. Chen J.H. Hom J. Chi J. Internal medicine resident computer usage: an electronic audit of an inpatient service.JAMA Intern Med. 2016; 176: 252-254Crossref PubMed Scopus (28) Google Scholar We defined attention switching as any transition from one patient's chart to another within a single session of EHR use (Supplementary Figure S1). The total time spent using the EHR per shift was computed as previously described,5Lou S.S. Liu H. Warner B.C. Harford D. Lu C. Kannampallil T. Predicting physician burnout using clinical activity logs: model performance and lessons learned.J Biomed Inform. 2022; 127: 104015Crossref PubMed Scopus (8) Google Scholar and was used as a measure of EHR-based workload. Wrong-patient errors were measured using the validated retract-and-reorder (RAR) decision rule,6Adelman J.S. Kalkut G.E. Schechter C.B. et al.Understanding and preventing wrong-patient electronic orders: a randomized controlled trial.J Am Med Inform Assoc. 2013; 20: 305-310Crossref PubMed Scopus (89) Google Scholar defined as any order that is placed by a clinician for a patient, cancelled within 10 min, and re-ordered by the same clinician on a different patient within the next 10 min. RAR events have a 76.2% positive predictive value for being true wrong-patient errors.6Adelman J.S. Kalkut G.E. Schechter C.B. et al.Understanding and preventing wrong-patient electronic orders: a randomized controlled trial.J Am Med Inform Assoc. 2013; 20: 305-310Crossref PubMed Scopus (89) Google Scholar A linear mixed-effect model was used to evaluate the relationship between the average rate of attention switching and total EHR time per shift, controlling for sex, patient load, and repeated measures per clinician. Separate models were constructed for attending physicians and APPs. A Poisson regression was used to determine the relationship between the rate of attention switching and wrong-patient errors, adjusting for patient load and order volume. During the study period, 24 attending physicians and 71 APPs worked 11 627 shifts with 62 367 h of EHR use and cared for 27 165 patients. Although ICU clinicians use the EHR for several hours each shift (median 2.4 h for attendings, 4.2 for APPs), their EHR usage was fragmented across numerous sessions of continuous use (median 17 sessions per 12-h shift for attendings, 22 for APPs), each typically lasting ∼10 min (Supplementary Table S1). Within each session, attention switching between patients occurred frequently (median 4.4 times per session for attendings, 3.0 for APPs), resulting in a median of 8.5 attention switches per 100 EHR actions for attending physicians, and 6.3 for APPs. After adjusting for sex and patient load, we found that an increased rate of attention switching was associated with increased total EHR time per shift for APPs but not for attending physicians (Table 1). For APPs, an increase in attention switching rate from 4.0 (25th percentile) to 8.6 (75th percentile) switches per 100 EHR actions increased the total EHR time per shift by 0.28 h (95% confidence interval [CI], 0.24–0.32; P<0.001).Table 1Multivariable model for total EHR time as a function of attention switching rate. Linear mixed-effects models were used to examine the relationship between the rate of attention switching, patient load, and total EHR time, controlling for repeated measures per clinician and gender. Estimated effect sizes for continuous variables are scaled by a 25th to 75th percentile change in each fixed effect. Separate models were constructed for attending physicians and APPs. APPs, advanced practice providers; CI, confidence interval; EHR, electronic health record.VariableScaled parameter estimate (95% CI)P-valueAttending physicianPatient load0.41 (0.32 to 0.49)<0.001∗∗∗Switch rate per 100 EHR actions0.04 (−0.02 to 0.10)0.156Gender (=male)−0.94 (−2.01 to 0.06)0.078Advanced practice providerPatient load2.56 (2.50 to 2.63)<0.001∗∗∗Switch rate per 100 EHR actions0.28 (0.24 to 0.32)<0.001∗∗∗Gender (=male)−0.29 (−0.88 to 0.31)0.348∗∗∗ : p < 0.001. Open table in a new tab ∗∗∗ : p < 0.001. APPs engaged in 340 305 ordering sessions and made 116 wrong-patient errors over the study period. After adjusting for patient load and order volume, an increase in the rate of attention switching from 4.0 (25th percentile) to 8.6 (75th percentile) switches per 100 EHR actions increased the risk for wrong-patient errors by 28% (rate ratio=1.28; 95% CI, 1.04–1.55; P=0.014) (Supplementary Table S2). This analysis was not conducted for attending physicians because of their limited role in placing orders. In summary, we found that EHR work was highly fragmented, with frequent attention switching between different patients. For APPs, who are front-line clinicians directly responsible for patient care, a high rate of attention switching was associated with downstream effects such as an increase in total EHR time, which has wellness implications given prior literature relating EHR use to burnout.7Yan Q. Jiang Z. Harbin Z. Tolbert P.H. Davies M.G. Exploring the relationship between electronic health records and provider burnout: a systematic review.J Am Med Inform Assoc. 2021; 28: 1009-1021Crossref PubMed Scopus (31) Google Scholar More strikingly, attention switching was also associated with an increased risk for wrong-patient errors, consistent with prior literature on the effects of task-related interruptions.8Douglas H.E. Raban M.Z. Walter S.R. Westbrook J.I. Improving our understanding of multi-tasking in healthcare: drawing together the cognitive psychology and healthcare literature.Appl Ergon. 2017; 59: 45-55Crossref PubMed Scopus (52) Google Scholar This is one of the first studies of the downstream effects of attention switching for ICU clinicians. Given the systemic complexities of clinical workflow, attention switching is likely unavoidable. However, a better understanding of the frequency and contexts in which attention switching occurs can potentially influence the design of the EHR and clinical workflows to reduce cognitive burden, with benefits for clinician workload, wellness, and patient safety.9Melnick E.R. Sinsky C.A. Krumholz H.M. Implementing measurement science for electronic health record use.JAMA. 2021; 325: 2149-2150Crossref PubMed Scopus (18) Google Scholar To this end, we developed a metric for attention switching using passively collected EHR audit log data, which enables measurement at scale across diverse clinical settings. However, our metric has limitations. For example it does not capture all attention switches (i.e. those outside the EHR or within the same patient are not captured), and the cause of each observed attention switch is unknown. Nevertheless, this work is an important first step in the systematic characterisation of clinical attention switching behaviours. As this was a single-centre study, further work is necessary to establish the generalisability of these findings. The Washington University/BJC HealthCare Big Ideas Healthcare Innovation Award. US National Institutes of Health grant (5T32GM108539-07 to SSL.).