Article Text

Types of diagnostic errors reported by paediatric emergency providers in a global paediatric emergency care research network
  1. Prashant Mahajan1,
  2. Joseph A Grubenhoff2,
  3. Jim Cranford3,
  4. Maala Bhatt4,
  5. James M Chamberlain5,
  6. Todd Chang6,
  7. Mark Lyttle7,
  8. Rianne Oostenbrink8,
  9. Damian Roland9,
  10. Richard M Ruddy10,
  11. Kathy N Shaw11,
  12. Robert Velasco Zuniga12,
  13. Apoorva Belle3,
  14. Nathan Kuppermann13,
  15. Hardeep Singh14
  1. 1Emergency Medicine and Paediatrics, University of Michigan, Ann Arbor, Michigan, USA
  2. 2Paediatric Emergency Medicine, University of Colorado Denver School of Medicine, Aurora, Colorado, USA
  3. 3Emergency Medicine, University of Michigan, Ann Arbor, Michigan, USA
  4. 4Paediatrics, University of Ottawa, Ottawa, Ontario, Canada
  5. 5Emergency Medicine, Children's National Medical Center, Washington, District of Columbia, USA
  6. 6Paediatric Emergency Medicine, Children's Hospital of Los Angeles, Los Angeles, California, USA
  7. 7Paediatric Emergency Medicine, University Hospitals Bristol NHS Foundation Trust, Bristol, UK
  8. 8Paediatric Emergency Medicine, Erasmus MC–Sophia Children's Hospital, Rotterdam, UK
  9. 9Paediatric Emergency Medicine, University of Leicester, Leicester, UK
  10. 10Paediatric Emergency Medicine, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, USA
  11. 11Paediatric Emergency Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
  12. 12Paediatric Emergency Medicine, Hospital Universitario Rio Hortega, Valladolid, Spain
  13. 13Emergency Medicine and Paediatrics, University of California Davis, Davis, California, USA
  14. 14Medicine - Health Services Research, Baylor College of Medicine, Houston, Texas, USA
  1. Correspondence to Dr Prashant Mahajan; pmahajan{at}med.umich.edu

Abstract

Background Diagnostic errors, reframed as missed opportunities for improving diagnosis (MOIDs), are poorly understood in the paediatric emergency department (ED) setting. We investigated the clinical experience, harm and contributing factors related to MOIDs reported by physicians working in paediatric EDs.

Methods We developed a web-based survey in which physicians participating in the international Paediatric Emergency Research Network representing five out of six WHO regions, described examples of MOIDs involving their own or a colleague’s patients. Respondents provided case summaries and answered questions regarding harm and factors contributing to the event.

Results Of 1594 physicians surveyed, 412 (25.8%) responded (mean age=43 years (SD=9.2), 42.0% female, mean years in practice=12 (SD=9.0)). Patient presentations involving MOIDs had common undifferentiated symptoms at initial presentation, including abdominal pain (21.1%), fever (17.2%) and vomiting (16.5%). Patients were discharged from the ED with commonly reported diagnoses, including acute gastroenteritis (16.7%), viral syndrome (10.2%) and constipation (7.0%). Most reported MOIDs (65%) were detected on ED return visits (46% within 24 hours and 76% within 72 hours). The most common reported MOID was appendicitis (11.4%), followed by brain tumour (4.4%), meningitis (4.4%) and non-accidental trauma (4.1%). More than half (59.1%) of the reported MOIDs involved the patient/parent–provider encounter (eg, misinterpreted/ignored history or an incomplete/inadequate physical examination). Types of MOIDs and contributing factors did not differ significantly between countries. More than half of patients had either moderate (48.7%) or major (10%) harm due to the MOID.

Conclusions An international cohort of paediatric ED physicians reported several MOIDs, often in children who presented to the ED with common undifferentiated symptoms. Many of these were related to patient/parent–provider interaction factors such as suboptimal history and physical examination. Physicians’ personal experiences offer an underexplored source for investigating and mitigating diagnostic errors in the paediatric ED.

  • emergency department
  • paediatrics
  • diagnostic errors

Data availability statement

No data are available.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

WHAT IS ALREADY KNOWN ON THIS TOPIC

  • Emergency care in paediatrics is challenging because of children’s unique physiology and difficulties with communication. Prior studies suggest that there are approximately 1.25 million instances of diagnostic errors in the USA among the 25 million annual paediatric emergency department visits.

WHAT THIS STUDY ADDS

  • In this survey study, multitudinous paediatric emergency physicians from a global research network shared cases of diagnostic errors underscoring their global burden, importance and shared contributory factors.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • The survey results suggest the need to develop systematic approaches to measuring diagnostic errors, which could then enable both local and wider global learning to reduce error rates.

Background

Errors in diagnosis are common in the emergency department (ED) because decisions are made under time constraints, under stress, often with inadequate information, involving illnesses that vary in severity or that have evolved insufficiently to allow for diagnostic certainty.1 Emergency care in children is particularly challenging because of a child’s unique physiology and difficulties with communication. The global burden of diagnostic errors in paediatric emergency care remains largely unknown, but conservative estimates of 5% frequency in ambulatory settings translates to approximately 1.25 million instances of diagnostic errors in the USA among the 25 million annual paediatric ED visits.2–4 These data underscore the urgency to investigate the causes of, and harms from, these diagnostic errors in the paediatric ED.

The causes of diagnostic errors are complex, multifactorial and influenced by patient, provider and system factors. Importantly, the ED provider is almost always at the centre of the diagnostic process; and, by some estimates, approximately 75% of errors have a cognitive component and approximately 80% involve a cognitive error during the patient/parent–provider encounter.5–8 Despite this, few studies investigating clinician perspectives on the causes of diagnostic errors exist in the literature and to our knowledge, there are no studies focused on the paediatric emergency setting.9 10

To facilitate data gathering of diagnostic errors from ED physicians, we pragmatically defined them as missed opportunities for improving diagnosis (MOIDs) regardless of patient harm.3 11–13 In the context of an ED visit, an MOID would be a case where sufficient data to suggest the final, correct diagnosis was present at the first ED/index visit or in which documented abnormal findings at the index visit should have prompted additional evaluation that would have revealed the correct, ultimate diagnosis. Most paediatric ED studies on MOID have been from the USA. To obtain an international perspective on diagnostic errors, we surveyed providers at 71 EDs across 6 countries participating in a large multicentre Paediatric Emergency Research Network (PERN; https://pern-global.com/) to solicit deidentified instances of diagnostic errors, their contributory factors and resultant patient harm.14

Methods

Patient and public involvement

Patients were not involved in this study.

Study design and survey development

We designed and administered a survey using the Qualtrics (Qualtrics, Provo, Utah, USA) online survey platform to paediatric ED providers to solicit specific examples of MOIDs. Respondents provided a briefcase summary of either their own or a colleague’s patient encounter and answered questions regarding resulting harm and factors contributing to the event. The survey (online supplemental file 1) was evaluated for construct and face validity and iteratively refined by 10 paediatric emergency medicine physicians at the University of Michigan. Survey questions were mainly closed-ended, although additional open-ended questions allowed participants to express opinions or to provide clarification of responses. The survey was administered from December 2018 to October 2019.

Supplemental material

Participants and survey distribution

Survey participation was voluntary, and respondents did not receive any incentives. The survey was distributed to ED physicians participating in the PERN, a consortium of global paediatric emergency care research networks.15 PERN is a collaborative network of seven paediatric emergency care research networks including: (A) Research in European Paediatric Emergency Medicine (REPEM), (B) Paediatric Emergency Care Applied Research Network (PECARN), (C) Pediatric Emergency Medicine Collaborative Research Committee of the American Academy of Pediatrics, (D) Paediatric Emergency Research Canada (PERC), (E) Paediatric Research in Emergency Departments International Collaborative, (F) Paediatric Emergency Research in the United Kingdom and Ireland (PERUKI) and (G) Red de Investigación de la Sociedad Española de Urgencias de Pediatría (RISeuP)/Spanish Paediatric Emergency Research Grou. Together, the 7 research networks manage >2 million paediatric emergency presentations per annum, in 71 hospitals, in 5 of the 6 WHO regions.14 Each participating network within PERN identified a lead investigator who was responsible for providing a list of sites and site champions from their network. Site champions completed a brief survey describing their ED characteristics and provided names and emails of potential participants at their site. These names and email IDs were used to create unique personal survey links that were emailed to each potential participant, which were used for measurement of response rate and to send a follow-up with requests to complete the partial responses. Each potential participant received up to two reminders to start the survey, and those with partial responses received a third reminder to complete the survey, if needed.

Definition of diagnostic error: Investigators from our group have successfully operationalised the measurement of diagnostic errors by viewing them as missed opportunities to make a correct or timely diagnosis regardless of patient harm.3 11–13 We provided the following example to assist the providers in describing an instance of a missed opportunity they may have experienced.

Briefcase summary: A 2-month-old male child presented to the ED with non-bilious vomiting for 1 week and failure to thrive. The physical examination revealed a normal appearing infant without obvious dehydration.

Diagnosis and disposition at first ED visit: Gastro-oesophageal reflux and discharged home.

Final, correct diagnosis: Patient came back with persistent vomiting 36 hours later, an ultrasound revealed a hypertrophic pyloric stenosis.

MOID: Although reflux as well as many other aetiologies can cause vomiting, hypertrophic pyloric stenosis should be a differential diagnosis at the initial visit and not obtaining or arranging for an ultrasound abdomen represents an MOID.

Outcomes: The primary outcome was to describe the frequency of conditions that could have been diagnosed at initial or index ED evaluation. The secondary outcomes were descriptions of the contributory factors and patient harm from MOIDs, with stratification and analysed by individual network. We created a categorisation system based on body systems (online supplemental table 1) for index and repeat ED visit complaints and diagnoses. For example, patients presenting with vomiting were categorised as having a gastrointestinal system condition while those presenting with headaches were categorised as having central nervous system conditions. If the patients were diagnosed with acute gastroenteritis at the index ED visit, they were categorised as having a gastrointestinal diagnosis. On repeat ED visit, if they were diagnosed with appendicitis, they were then categorised as having a gastrointestinal system final diagnosis. However, if the patient had a diagnosis of intracranial tumour, then they were categorised as having an oncological condition.

Contributory factors to MOIDs were categorised by adapting the diagnostic process dimensions described in the Safer Dx Framework.16 These dimensions are described in the survey (online supplemental file 1) and include the following categories: (A) patient-related factors (eg, delay in seeking care, language barriers, caregiver factors adversarial or incomplete history, left against medical advice), (B) patient/parent–provider encounter factors (eg, problems with patient history such as incomplete or misinterpreted history, problems with physical examination such as incomplete or misinterpreted examination findings, failure to review prior records, failure to order an indicated diagnostic test), (C) diagnostic tests (ordered but not performed or not interpreted appropriately), (D) follow-up/tracking factors related to tests and referrals, (E) consultations (not ordered, not available, contradictory recommendations, etc) and (F) miscellaneous factors (workload too high, lack of resources, institutional factors, etc).

Statistical analysis

We performed descriptive statistics including absolute and relative frequencies or means with SDs or medians with IQRs, as appropriate, to compare all survey question responses. We present summary statistics for the entire cohort and display differences stratified by individual research network. We used SAS V.9.4 (SAS Institute) for all quantitative analyses and report our results using elements from the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE), The SUrvey Reporting GuidelinE (SURGE) and Checklist for Reporting Results of Internet E-Surveys (CHERRIES) guidelines.17–20

Results

The survey was distributed to 1594 physicians across all participating EDs in PERN, of whom 412 (25.8%) responded. Mean age of respondents was 43 years (SD 9.3), and mean years in practice was 12.0 (SD 9.0); 58.0% identified as male. The response rate by participating network is given in figure 1. Providers worked a median 11.0 (IQR 6.0) shifts per month. Distribution by training included 67% PEM, 10.2% paediatrician, 9.2% PEM in training, 3.9% general EM, with the remainder of participants indicating that they received both PEM and general EM training. Six of the seven networks in PERN contributed to MOIDs cases, and the distribution of participants by network was PECARN (35.4%), PEM-CRC (30.1%), PERC (16.3%), PERUKI (9.0%), RISeuP (6.8%) and REPEM (2.4%).

Figure 1

MOIDs study recruitment flow chart. MOIDs, missed opportunities for improving diagnosis; PERN, Paediatric Emergency Research Network; PECARN, Paediatric Emergency Care Applied Research Network; PEM-CRC, Paediatric Emergency Medicine Collaborative Research Committee; PERC, Paediatric Emergency Research Canada; PERUKI, Paediatric Emergency Research in the United Kingdom and Ireland; RISeuP/SPERG, Red de Investigación de la Sociedad Española de Urgencias de Pediatría/Spanish Paediatric Emergency Research Group; REPEM, Research in European Paediatric Emergency Medicine.

Most (62.3%) of the MOIDs involved the reporting clinicians themselves either in isolation or as a part of the clinical care team; while 10.7% were observations of care in which they were not directly involved. The median time in hours to discovery of the MOIDs was 36.0 hours (IQR 52.0); 75.8% were recognised within 72 hours and 87.8% were detected within 7 days. Almost two-thirds reported that the correct diagnosis was made on a repeat ED visit; 6.8% and 2.4% stated the correct diagnosis was made in the in-patient or intensive care unit setting, respectively. Approximately, 1% also mentioned that the correct diagnosis was made at autopsy.

Most patients with MOIDs presented initially with common undifferentiated symptoms, such as abdominal pain (21.1%), fever (17.2%), vomiting (16.5%) and headache (8.5%%) (table 1). Abdominal pain was either the most common or a very common presenting symptom when MOIDs were stratified by individual research network. Acute gastroenteritis (16.7%), viral syndrome (10.2%), constipation (7.0%) and migraine (4.1%) were the top four discharge diagnoses at the index ED visit, and this pattern was largely maintained when stratified by individual research networks (table 2). Participants provided instances of missed appendicitis as the most common MOID overall, in the total dataset and when stratified by network (table 3). Brain tumour, meningitis, non-accidental trauma, pneumonia and intussusception were the other common MOIDs reported by participants.

Table 1

Ten most frequently reported presenting symptoms at index ED visit in entire cohort and by network

Table 2

Ten most frequent discharge diagnosis at index ED visit in entire cohort and by network

Table 3

Ten most frequently reported MOIDs in the entire cohort and by network

Difficulties arising from patient/parent–provider interactions were the most frequently cited contributing factor (59.1%) in the total dataset and in all except two networks. Incomplete history-taking or performance of physical examination was identified as a factor within the patient/parent–provider encounter in all but two networks (online supplemental table 2)

More than half of the reported MOIDs led to either moderate (48.7%) or major (10.0%) harm as a result of the MOIDs, while 35% had mild harm and 6.4% reported no harm to the patient (online supplemental figure 1).

Discussion

In this study, a large number of paediatric emergency physicians from a global research network shared cases of diagnostic errors underscoring their global burden, importance and shared contributory factors. Types of errors and contributing factors were common across international regions, suggesting areas for targeted interventions to reduce diagnostic errors globally. Physicians’ personal experiences offer an underexplored source for investigating such errors and may provide valuable qualitative information into the cognitive processes and systems factors contributing to MOIDs.

Our study builds on a previous US-based study involving a convenience sample of 310 physicians, including general internists, medical specialists and emergency physicians, who voluntarily reported 583 cases of diagnostic errors.9 We intentionally focused on obtaining perspectives of paediatric ED providers and leveraged a pre-existing international cohort of research networks to achieve a broader understanding of the challenges faced in this unique practice environment. Contributory factors and level of harm were assessed by questions based on newer frameworks (National Academies of Science, Engineering, and Medicine and Safer Dx framework),2 16 but we found similar proportions of harm to patients compared with that study. Although the cases were self-reported in both studies and direct comparison between the two studies is not possible, only 10.0% of the reported cases had major harm, compared with 28% in the prior study.

Nearly two-thirds (62.3%) of respondents in this study reported MOIDs in which they were personally involved, and ~93% of reported cases involved some degree of patient harm with 58.7% classified as major/moderate. This proportion is lower than that reported (82.3%) in a previous multicentre study of US-based academic paediatric providers but higher than an Irish study of paediatricians and paediatric residents (44.1%). This likely represents the fact that we did not limit respondents to describing only cases in which they were involved. Nonetheless, these figures underscore that MOIDs contribute significantly to patient harm.21–23

Diagnostic safety experts cite the need for feedback and calibration as prerequisites to improve diagnostic expertise.24 The characteristics of ED care assures that paediatric ED providers are unlikely to learn of their MOIDs unless there is significant associated harm. A potential solution to this problem could include better feedback processes for patients with high-risk conditions (eg, abdominal pain and headache), such as follow-up telephone calls to patients after ED discharge or critical review of cases with return visits to the ED. Among the cases described by respondents, the subsequent final correct diagnoses included many life-threatening conditions that paediatric ED providers are well trained to detect, and know must not be missed (myocarditis, bacterial meningitis, appendicitis and child abuse). However, these were not detected despite the presence of clinical information suggesting evidence existing that should have led to a correct diagnosis. This mirrors a study by Okafor et al examining cases of diagnostic errors reported by general emergency providers in which the most common missed cases included must-not-miss diagnoses typical of that practice setting: sepsis, myocardial infarction, fractures, vascular injuries and stroke.8 These cases are likely to come to the clinician’s attention through departmental or institutional safety surveillance mechanisms. However, clinicians often do not report diagnostic errors or may not report errors with minor harm,25 which suggests the need for alternative mechanisms to supplement error reporting. About 75% of MOIDs became apparent within 72 hours and almost 65% were discovered during a subsequent ED encounter, suggesting that automated return visit evaluation could be potential quality improvement tools for identifying diagnostic errors and alerting clinicians for the need to review their decision-making at a prior encounter.26 27

Our study has several limitations. Despite soliciting broad, international participation, the overall survey response rate was low and suffers from possible reporting bias, recall bias (providers may remember only those cases which had significant outcomes) and availability bias. We mitigated some of these limitations by soliciting only deidentified instances of diagnostic errors and framing cases as learning opportunities. Additionally, we provided an example of an MOID, which allowed the providers to model their own cases accordingly. We comprehensively tested the survey for face, construct and content validity before soliciting participation to mitigate survey design issues. Our response rate was in range of other surveys of paediatricians regarding diagnostic errors.10 22 Another limitation was our inability to verify the accuracy of the contributory factors or level of harm. It is likely that retrospective recall of diagnostic errors results in inaccurate assignment of contributory causes in some cases. However, it was reassuring to note that most cases submitted included those experienced by the reporting clinician themselves, which allowed them to introspect about the MOIDs and reflects what likely happened in the ED. Finally, the participation of respondents varied across the research networks ranging from 35.4% to 2.4% with a higher participation by paediatric ED providers in North America.

Although our multinational survey was not designed to measure the epidemiology or frequency of diagnostic errors, it nevertheless reveals that diagnostic errors lead to substantial patient harm in paediatric EDs globally. Use of an MOIDS approach was relatively easy for conveying the concept of diagnostic errors. Our findings also reveal the commonality of issues across EDs such as breakdown in patient/parent–provider interactions especially issues with history-taking and physical examination. Thus, interventions aimed at improving information gathering and synthesis during paediatric ED encounters may have substantial potential to reduce patient harm globally. The survey suggests the need to develop systematic approaches to measuring diagnostic errors, which could then enable both local and wider global learning to reduce error rates.

Data availability statement

No data are available.

Ethics statements

Patient consent for publication

Ethics approval

The study was determined not to be human subjects research by the University of Michigan Institutional Review Board (HUM00128709).

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Twitter @MahajanPemDoc

  • Correction notice This article has been corrected since it was first published. Author name Richard M Rudy is changed to Richard M Ruddy.

  • Contributors PM conceived and designed the study including developing the survey instrument, analysing the results, drafting the manuscript. PM is also the guarantor of this article. PM, JAG, JC, MB, JMC, TC, ML, RO, DR, RMR, KNS, RVZ, AB, NK and HS helped review the survey instrument, discussed the results and assisted in editing the manuscript. PM, JC and AB had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.