Article Text

Diagnostic error among vulnerable populations presenting to the emergency department with cardiovascular and cerebrovascular or neurological symptoms: a systematic review
  1. Svetlana Herasevich1,
  2. Jalal Soleimani1,
  3. Chanyan Huang2,
  4. Yuliya Pinevich1,
  5. Yue Dong1,
  6. Brian W Pickering1,
  7. Mohammad H Murad3,
  8. Amelia K Barwise4,5
  1. 1 Department of Anesthesiology and Perioperative Medicine, Mayo Clinic, Rochester, Minnesota, USA
  2. 2 Department of Anaesthesiology, Sun Yat-sen University First Affiliated Hospital, Guangzhou, Guangdong, China
  3. 3 Center for Science of Healthcare Delivery, Division of Preventive Medicine, Mayo Clinic, Rochester, Minnesota, USA
  4. 4 Division of Pulmonary and Critical Care Medicine, Mayo Clinic, Rochester, Minnesota, USA
  5. 5 Bioethics Research Program, Mayo Clinic, Rochester, MN, USA
  1. Correspondence to Dr Svetlana Herasevich, Department of Anesthesiology and Perioperative Medicine, Mayo Clinic Rochester, Rochester, MN 55905, USA; Herasevich.Svetlana{at}mayo.edu

Abstract

Background Diagnostic error (DE) is a common problem in clinical practice, particularly in the emergency department (ED) setting. Among ED patients presenting with cardiovascular or cerebrovascular/neurological symptoms, a delay in diagnosis or failure to hospitalise may be most impactful in terms of adverse outcomes. Minorities and other vulnerable populations may be at higher risk of DE. We aimed to systematically review studies reporting the frequency and causes of DE in under-resourced patients presenting to the ED with cardiovascular or cerebrovascular/neurological symptoms.

Methods We searched EBM Reviews, Embase, Medline, Scopus and Web of Science from 2000 through 14 August 2022. Data were abstracted by two independent reviewers using a standardised form. The risk of bias (ROB) was assessed using the Newcastle-Ottawa Scale, and the certainty of evidence was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation approach.

Results Of the 7342 studies screened, we included 20 studies evaluating 7436,737 patients. Most studies were conducted in the USA, and one study was multicountry. 11 studies evaluated DE in patients with cerebrovascular/neurological symptoms, 8 studies with cardiovascular symptoms and 1 study examined both types of symptoms. 13 studies investigated missed diagnoses and 7 studies explored delayed diagnoses. There was significant clinical and methodological variability, including heterogeneity of DE definitions and predictor variable definitions as well as methods of DE assessment, study design and reporting.

Among the studies evaluating cardiovascular symptoms, black race was significantly associated with higher odds of DE in 4/6 studies evaluating missed acute myocardial infarction (AMI)/acute coronary syndrome (ACS) diagnosis compared with white race (OR from 1.18 (1.12–1.24) to 4.5 (1.8–11.8)). The association between other analysed factors (ethnicity, insurance and limited English proficiency) and DE in this domain varied from study to study and was inconclusive.

Among the studies evaluating DE in patients with cerebrovascular/neurological symptoms, no consistent association was found indicating higher or lower odds of DE. Although some studies showed significant differences, these were not consistently in the same direction.

The overall ROB was low for most included studies; however, the certainty of evidence was very low, mostly due to serious inconsistency in definitions and measurement approaches across studies.

Conclusions This systematic review demonstrated consistent increased odds of missed AMI/ACS diagnosis among black patients presenting to the ED compared with white patients in most studies. No consistent associations between demographic groups and DE related to cerebrovascular/neurological diagnoses were identified. More standardised approaches to study design, measurement of DE and outcomes assessment are needed to understand this problem among vulnerable populations.

Trial registration number The study protocol was registered in the International Prospective Register of Systematic Reviews PROSPERO 2020 CRD42020178885 and is available from: https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42020178885.

  • Diagnostic errors
  • Patient safety
  • Emergency department

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

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 KNOWN ON THIS TOPIC

  • Diagnostic errors (DE) in hospital settings and especially in the emergency department (ED) are often under-recognised and under-reported but have impactful negative effects on outcomes. Minority populations and other marginalised groups such as those with a low socioeconomic status or without health insurance are likely to be at significantly higher risk of DE for multiple reasons. It is unknown which variables to indicate vulnerable groups are consistently associated with DE.

WHAT THIS STUDY ADDS

  • This systematic review highlights some of the patient characteristics associated with experiencing a DE when presenting to the ED with cardiovascular or cerebrovascular/neurological symptoms. Black race was associated with significantly higher odds of missed acute myocardial infarction and acute coronary syndrome in most included studies. Among patients presenting with cerebrovascular/neurological symptoms, no consistent associations between patient characteristics and DE were found across studies. This review highlights the heterogeneity of definitions, study designs, and measurement approaches to DE and outcomes reported.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE, OR POLICY

  • Given the identified heterogeneity we highlighted, further research is needed to support the identification and measurement of DE so robust comparisons and longitudinal quality improvement initiatives to address and mitigate DE can be conducted.

Introduction

Diagnostic error (DE) is defined by the National Academy of Medicine as a failure to establish an accurate and timely explanation of the patient’s health problem or communicate that explanation to the patient and in the health record.1 DE is a common problem in clinical practice, particularly in the emergency department (ED).2–4 It has been estimated that 40 000–80 000 deaths in the USA are attributed to DE annually,5 and major diagnostic discrepancies have been identified in 10%–20% of autopsy studies.6

DE in hospital settings and especially in the ED are often under-recognised and under-reported.6–8 A recent systematic review (SR) evaluated the existing evidence regarding harmful DE in hospitalised adult patients and reported the rate of harmful misdiagnosis as 0.7%.9 Even if DE does not cause direct patient harm, it can lead to unnecessary resource utilisation and increased healthcare costs.9 10 Studies have demonstrated it is not simply complicated diagnoses that are missed, but that common conditions and diseases may also be prone to DE.4 5 11 12

Disparities in health and healthcare are widely acknowledged to exist for minority groups in diverse settings.13–16 Minority groups and populations with a low socioeconomic status or without health insurance are likely to be at higher risk of DE for multiple reasons. These include limited access to healthcare, lack of trust in healthcare institutions, concerns about costs of care or necessity to divulge information they may be reluctant to share,17 quality of care in hospitals and clinics they attend, lower use of diagnostic tools, atypical presenting symptoms, or potential bias.18–22 There is a growing recognition that DEs among vulnerable populations deserve increased attention.13 However, while DE is a well described phenomenon in the general ED population,4 9 a SR evaluating DE among vulnerable populations is lacking. Given our recent work on DE in acute care settings,23–26 we were interested in exploring whether DE was consistently associated with specific demographic variables indicating vulnerable and minority populations. Identifying variables associated with an increased risk of DE may guide targeted interventions for DE prevention in the ED setting. Patients presenting to the ED with cardiovascular or cerebrovascular/neurological symptoms were of special interest. The delay in the diagnosis or failure to hospitalise may be most impactful in terms of adverse outcomes and long-term consequences in this large group of patients.

The objective of this SR was to evaluate the frequency and causes of DE in under-resourced patients (racial and ethnic minorities, those with limited English proficiency (LEP), migrant populations and patient groups qualifying for federal insurance) presenting to the ED with cardiovascular or cerebrovascular/neurological symptoms.

Methods

The results of the study were reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statements27 and Synthesis Without Meta-analysis (SWiM) guidelines (online supplemental appendix 1).28 No institutional review board approval was sought since this was secondary analysis of published work.

Supplemental material

Data sources and search strategy

The literature search was done by a medical librarian using the concepts of DE and delay. The search strategies were created using a combination of keywords and standardised index terms. Searches were run in January 2020 and rerun in May 2021 and in August 2022 in Evidence-Based Medicine Reviews, Embase, Medline (including epub ahead of print, inprocess, and other non-indexed citations), Scopus and Web of Science. Results were limited to citations published from the year 2000 onwards and in English. The additional sources were identified using grey literature and reference mining searches.29 The full search strategy is provided in the online supplemental appendix 2.

Supplemental material

Study selection

We included studies that met all the following criteria: (1) Enrolled adult patients (≥18 years of age), (2) Participants presented to the ED with cerebrovascular/neurological, or cardiovascular symptoms (including studies where patients were admitted and studies where patients were discharged from the ED), (3) Study included information about differences in DE among different demographic groups (race, ethnicity, LEP, different types of insurance holders) or included demographics in the analysis of differences. We excluded (1) Case reports, (2) Case series, (3) Single abstracts and conference proceedings, (4) Editorial, commentary and review articles, (5) Non-English articles, (6) Simulation articles.

Titles and abstracts of all identified studies were independently reviewed by two reviewers (SH, AKB, JS, CH, YP, YD) using eligibility criteria. Studies included for full-text screening were also reviewed by pairs of independent reviewers.

Data extraction

Study details were abstracted by two independent reviewers (YD, JS, YP, CH) using a pilot-tested standardised data extraction form. Data extracted included: authors, publication date, country, duration, number of participating centres, study design, data sources, description and number of participants, including age, sex and race/ethnicity, outcomes assessed, description, rate and approach for definition of DE, demographic factors, differences in DE occurrence between demographic groups, statistical approach, and types of variables for which estimates were adjusted (tables 1 and 2, online supplemental tables 1 and 2). Additional reviewers (SH, AKB) assessed data extraction and resolved disagreements during all phases.

Supplemental material

Supplemental material

Table 1

Eligible studies evaluating DE in patients with cardiovascular symptoms/disease and participant characteristics

Table 2

Eligible studies evaluating DE in patients with cerebrovascular/neurological symptoms/disease and participant characteristics

Risk of bias/quality assessment

We assessed the risk of bias (ROB) using the Newcastle-Ottawa Scale for non-randomised studies (SH, AKB).30 This tool assesses the representativeness of the exposed cohort, ascertainment of exposure, assessment of outcome and adequacy of follow-up.

We evaluated the certainty of evidence using the Grading of Recommendations Assessment, Development, and Evaluation approach,31 32 which reflects certainty in the measure of association between DE and a risk factor across all available studies. Per standard grading, evaluation of observational studies was initially considered as low certainty of evidence. To assess potential modifying factors impacting the strength of evidence we evaluated: methodological limitations of included studies, precision, directness, consistency and publication bias.31 32

Outcome measures

We assessed the differences in occurrence of DE between different racial and ethnic groups, differently insured patients, and those with language barriers as prespecified outcomes.

Data synthesis and analysis

When possible, we extracted or calculated the ORs and corresponding 95% CIs for the difference in the DE rate (binary outcome). We planned to use DerSimonian and Laird random-effects method for meta-analysis of data. However, given the large heterogeneity of study designs, populations, definitions and approach to DE assessment in the included studies, the pooling of effect estimates was not justified and hence abandoned. Instead, we narratively synthesised the data from the included studies using SWiM guidelines.28 The eligible studies were grouped by the parameters which were considered the most likely sources of heterogeneity—types of clinical problem (cardiovascular vs cerebrovascular/neurological), outcome assessed (difference in DE occurrence between groups compared) and the approach for DE assessment.

Results

Study characteristics

The search strategy identified 9850 studies with 35 additional studies identified through manual searches of grey literature and searches of reference sections in studies included in the full-text review. Relevant review papers and SRs were included for full-text review and then excluded as a ‘wrong study design’ after reference mining.29 After removing duplicates, 7342 papers were screened, of which 7107 were removed following title and abstract screening and 215 following full-text review, resulting in a final set of 20 studies. The PRISMA diagram (figure 1) shows the study selection process, study phases and reasons for exclusion.

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.

Description of eligible studies and participant characteristics

A total of 20 studies with 7436 737 participants was included in the final analysis. The numbers of participants in each study ranged from 4933 to 3482 695,34 and the median number of participants was 7421. Further study and participant characteristics are shown in tables 1 and 2. Most studies were conducted in the USA2 11 12 33–48 and one study was multicountry (USA and Europe).49 Four studies were single-centre33 39 41 47 and 16 were multicentre studies.2 11 12 34–38 40 42–46 48 49 All studies used an observational design, including 2 prospective40 45 and 18 retrospective studies. The retrospective studies varied from electronic medical record (EMR) review to database analyses and national survey analysis. The study periods varied between 6 months40 and 11 years50 with a median of 5.75 years.

Eleven studies evaluated DE in a cohort of patients with cerebrovascular/neurological symptoms,2 12 33 34 37–40 42 44 47 eight studies evaluated DE in a cohort with cardiovascular symptoms11 35 36 41 43 45 46 49 and one study examined patients with both cerebrovascular and cardiovascular disorders.48

Participants’ mean age varied from 32.9 years to 77.9 years, and 62.3% of the total participants were female. Although not explicitly stated, race and ethnicity definitions were most likely based on self-reported data in the EMR. Some studies combined race/ethnicity into one variable and others separated race and ethnicity into separate variables. Outcome differences based on race and ethnicity were not the focus of most papers.

Reported outcome measures and definitions

The definitions of DE are outlined in online supplemental tables 1 and 2. Most studies defined DE as a potentially missed diagnosis of a serious cardiovascular or cerebrovascular/neurological condition. The majority of studies measured DE as the proportion of patients with subsequent inpatient admissions or ED visits for a diagnosis of interest after discharge from the ED with symptoms indicative of this diagnosis.2 11 12 34 36 37 43 46 48 For DE assessment some studies used a look-forward approach and analysed the proportion of ED patients discharged from the ED with symptoms indicative of a serious diagnosis and then hospitalised or seen in the ED with the condition of interest,34 37 43 while other studies used a look-back approach and analysed the cohort of patients hospitalised with the diagnosis of interest and assessed which of them were previously seen in the ED.2 11 12 36 48 Sharp et al used both approaches.46

Kuruvilla et al 39 used a look-back approach and defined misdiagnosis as cases (either admitted to the hospital or discharged from the ED) that were given a non-stroke diagnosis in the ED and were subsequently identified as having a stroke. Two other studies used a look-forward approach. One, by Pope et al,45 defined DE as the mistaken ED discharge of patients with acute myocardial infarction (AMI) (failure to hospitalise). The other study42 defined DE as a missed stroke diagnosis in the ED among patients admitted to the hospital.

Other studies assessed DE as a prolonged time interval between either presentation to the ED and diagnosis, or first physician evaluation, or principal diagnostic test.33 35 38 40 41 44 49 Wang et al 47 did not separate delay before and after ED arrival and mostly evaluated the reasons for DE.

No eligible studies evaluating overdiagnosis were identified. The diverse approaches to defining DE within the included studies are notable.

ROB/quality appraisal

The overall ROB was low for most studies. In some studies ROB was moderate or high, mostly due to potential confounding and limited representativeness of the study population (online supplemental table 3).33 35 39 41 47 49

Supplemental material

The certainty of evidence was very low, mostly due to serious inconsistency across studies, indirectness in the outcome measure and some variability in the exposure measure. Some studies evaluating patients presenting with cerebrovascular/neurological symptoms also had methodological limitations (online supplemental table 4).

Supplemental material

Studies evaluating DE in patients with cardiovascular symptoms

Nine studies evaluated DE in cardiovascular patients.11 35 36 41 43 45 46 48 49 The main findings from the analysis of this subset are that black race is significantly associated with higher odds of DE in most studies evaluating missed AMI/acute coronary syndrome (ACS) diagnosis compared with white (adjusted OR (aOR) from 1.18 (1.12–1.24) to 4.5 (1.8–11.8)). The association of other analysed demographic factors with DE varied from study to study and is inconclusive (online supplemental table 1).

Race and ethnicity: All nine included studies explored the association between race/ethnicity and DE. Most studies combined race and ethnicity into one domain. One study (Chang et al)36 reported race and ethnicity as separate groups. To facilitate comparison of this study with others, we calculated the OR for the combined race/ethnicity group with non-Hispanic white as a reference. Harris et al 49 did not provide a clear definition and reported white and non-white groups.

DE as a potentially missed diagnosis of AMI or ACS was evaluated in six studies.11 36 43 45 46 48 Among these, black race was found to be associated with higher odds of missed AMI/ACS in four studies, using either a look-back11 36 48 or look-forward45 approach (aOR from 1.18 (1.12–1.24) to 4.5 (1.8–11.8)). Sharp et al used both approaches46 and reported higher odds of DE among black patients using a look-back approach (OR 1.30 (1.06–1.60)) and lower odds using a look-forward approach (OR 0.70 (0.53–0.92)). One study43 using a look-forward approach did not demonstrate a significant association between black race and DE (aOR 0.92 (0.82–1.04)). The association between other racial/ethnic groups and missed diagnosis was disparate (online supplemental table 1). Thus, Hispanic ethnicity was associated with lower odds of missed AMI/ACS in two out of six studies (aOR 0.89 (0.79–0.99) and 0.91 (0.85–0.97)),43 48 three out of six studies did not demonstrate any significant difference (aOR from 0.49 (0.03–8.16) to 1.19 (p=0.32)),11 45 46 but one out of six reported higher odds of a missed diagnosis among Hispanics36 (OR 1.57 (1.36–1.81)). One study evaluating AMI additionally reported lower odds of a possible missed diagnosis of other cardiovascular conditions among Hispanics (aOR 0.76 (0.73–0.80)) and no difference among black patients (aOR 1.01 (0.97–1.05)).43 Waxman et al 48 additionally reported higher odds of a missed diagnosis of abdominal aortic aneurysm (AAA) among black patients (aOR 1.35 (1.08–1.70)), and lower odds of a missed diagnosis of aortic dissection (AD) among Asian patients (aOR 0.65 (0.46–0.93)).

Diagnostic delay was evaluated in three studies.35 41 49 Banco et al reported a prolonged wait time to see the ED provider in the non-white group.35 No difference in time to diagnosis of AD among different racial or ethnic groups was observed in two other studies.41 49

Type of medical insurance: The association between type of medical insurance and occurrence of DE was described in five of nine studies.11 36 41 43 48 Medicaid insurance eligibility was associated with higher odds of DE compared with private insurance in one study (aOR 1.45 (1.28–1.65)),43 and no difference was found in two studies (aOR 1.12 (p=0.39) and 53% vs 51% (p=0.84)).11 41 Having dual Medicare/Medicaid eligibility was associated with higher odds of DE compared with Medicare alone in one study (aOR from 1.23 (1.05–1.42) for ruptured AAA to 1.40 (1.35–1.45) for AMI).48 Using Medicare insurance was associated with lower odds of DE in one study (aOR 0.80 (p=0.04))11 and no difference in another (aOR 1.12 (1.00–1.26)).43 No significant difference was found for self-payers in one study11 (aOR 0.87 (p=0.28)) while another study reported higher odds of DE in this group (aOR 1.17 (1.03–1.34)).43 Chang et al 36 reported no difference in adjusted rate of DE in differently insured patients. In summary, the evidence about insurance is not clear.

LEP (need for interpreter) was not found to be associated with higher odds of a missed diagnosis of AMI in one study (7.5% in both groups).46 It is not clear whether patients used an interpreter during their admission.

Studies evaluating DE in patients with cerebrovascular/neurological symptoms

Eleven studies evaluated DE in patients with cerebrovascular/neurological symptoms including stroke, subarachnoid haemorrhage (SAH) or serious neurological disease.2 12 34 37–40 42 44 47 48 One study assessed patients with multiple sclerosis (MS) who were previously seen in ED with symptoms indicative of demyelinating disease and not correctly diagnosed.33 Overall, we did not find consistent associations between race/ethnicity and insurance status with occurrence of DE in this subset of the studies (online supplemental table 2).

Race and ethnicity: All 12 studies examined the association between race/ethnicity and occurrence of DE. Most studies combined race and ethnicity into one domain. Three studies did not provide a clear definition and reported white and non-white groups.2 39 42

DE as a potentially missed diagnosis of stroke or SAH was evaluated in 5/12 studies.2 12 39 42 48 Four more studies evaluated delayed diagnosis of stroke/SAH in the ED.38 40 44 47 Among these nine studies, two studies reported higher odds of DE in black patients (aOR 1.68 (1.13–2.48) and 1.14 (1.06–1.19))12 44 and five studies did not observe any difference (aOR from 0.70 (0.36–1.36) to 2.14 (0.98–4.67)).2 38–40 42 One more study found black race to be associated with a higher risk of missed stroke (aOR 1.09 (1.07–1.11)), but not SAH (aOR 1.00 (0.90–1.12)).48 Among five studies evaluating Hispanic ethnicity, two out of five demonstrated higher odds of DE among Hispanics (aOR 1.18 (1.03–1.35) and 7.95 (1.51, 41.6)),12 40 one study reported lower odds (aOR 0.93 (0.88–0.99)),42 and one out of five did not find any difference (aOR 1.01 (0.41–2.49)).38 Waxman et al reported lower odds of missed stroke (aOR 0.90 (0.88–0.93)), and no difference in the SAH group (aOR 0.91 (0.79–1.05)), among Hispanics. We did not observe any patterns in the results related to the approach of DE assessment (look-forward approach, look-back approach or evaluation of delayed diagnosis). Wang et al 47 did not differentiate prehospital and in-hospital delay, but found race/ethnicity was not significantly associated with delay.

Dubosh et al 34 and Fahimi et al 37 evaluated potentially missed diagnosis of serious neurological conditions in the ED. Both studies used a look-forward approach, however, the association between race/ethnicity and DE differed in these studies. One study reported higher odds of DE in the black racial group (aOR 1.68 (1.13–2.48)) and no difference among Hispanics (aOR 0.82 (0.62–1.07)),37 while another study demonstrated lower odds of DE in both black and Hispanic groups (aOR 0.92 (0.87–0.98) and 0.76 (0.72–0.81)).34 One more study33 did not report any difference in diagnostic delay in MS diagnosis among racial and ethnic groups. More details about the associations between race/ethnicity and DE are provided in online supplemental table 2.

Medical insurance: The association between medical insurance type and occurrence of DE was described in seven studies.12 33 34 37 38 40 48 The included studies used different groupings of insurance types and demonstrated inconsistent results. For example, Medicare use, when compared with private insurance, was associated with lower odds of DE in one study (aOR 0.66 (0.55–0.80)),12 40 higher odds in two studies (aOR 1.54 (1.08–2.17) and 3.03 (1.27–7.14)),37 38 and no difference was observed in one study (aOR 1.46 (0.72–2.97)).38 Dubosh et al 34 reported higher odds of DE in the back pain group (aOR 1.28 (1.14–1.45)) and no difference in the headache group (aOR 1.05 (0.98–1.12)) among those with Medicare. Medicaid eligibility was associated with lower odds of DE compared with private insurance in one study (aOR 0.70 (0.58–0.84)),12 higher odds compared with Medicare in another (aOR 4.46 (1.45, 13.7)),40 and did not show any difference in three others (aOR from 0.96 (0.83–1.11) to 1.26 (0.85, 1.87) vs Medicare, 0.59 (0.17–2.11) vs private insurance).34 37 38 Waxman et al 48 reported higher odds of missed diagnosis of stroke and SAH for patients with Medicare and Medicaid dual eligibility compared with Medicare alone (aOR 1.40 (1.35–1.45) and 1.29 (1.18–1.41)). Farber et al 33 did not report any significant differences in MS diagnosis delay between different types of insurance.

Discussion

In this SR we evaluated differences in DE among minority and potentially under-resourced populations presenting to the ED with cardiovascular or cerebrovascular/neurological symptoms. These populations included different racial and ethnic groups, differently insured patients and those with language barriers. Due to the heterogeneity of included studies a meta-analysis was deemed inappropriate. The most remarkable finding was that black patients presenting with cardiovascular symptoms had significantly higher odds of a missed AMI/ACS diagnosis in most studies included in this SR. Other demographic factors analysed did not consistently show an association in the same direction. Among the studies evaluating DE in patients with cerebrovascular/neurological symptoms, no consistent association between race/ethnicity and insurance status with occurrence of DE was observed.

We believe we identified disparities in AMI/ACS conditions because there were a larger number of multisite studies and a more homogeneous definition of DE. It is also possible that AMI/ACS symptoms may have been more easily recognised by clinicians in some demographic groups and not in other groups. Additionally, for AMI/ACS cognitive errors account for DE rather than systems issues.4 Increased heterogeneity in DE definitions and measurement approaches among the cerebrovascular/neurological conditions which included five studies about missed diagnosis and four for a diagnostic delay, may explain the lack of consistent findings. These errors are likely due to atypical presentations and other factors that are beyond cognitive clinical errors. There were also fewer multisite studies in the cerebrovascular/neurological conditions, therefore the enrolled numbers may have been too small to demonstrate differences.

Our findings are in line with a recent AHRQ report that highlighted cardiovascular and cerebrovascular/neurological conditions among the most frequently misdiagnosed conditions in the ED and underlined the potential causes of DE in the ED.4 However, our SR is unique as we investigated the role of demographic factors that may influence diagnosis. Although there are several epidemiological-type studies examining the prevalence of stroke and treatment among racial and ethnic minorities, few studies have explored this issue from a DE perspective.50 Faigle et al 51 examined racial disparities in thrombolysis use among patients who had a stroke and noted that for minority men, the odds of receiving thrombolysis were significantly lower in minority hospitals compared with white hospitals.

Evidence also suggests that even when minority groups receive an accurate and timely diagnosis, there may be delays in initiation of treatment. For example, a two-centre study of about 500 patients by Bell et al demonstrated that time to first ECG and per cent of patients receiving cardiac catheterisation was longer and lower among black patients presenting to the ED.52 Although not directly linked to DE, it suggests treatment/therapeutic management for this population may have been delayed. Reasons for this are unclear and may reflect implicit bias.52 53

Our study focused on diagnostic assessment after arrival in the ED. We did not account for triaging prearrival in the ED or factors external to the ED. We understand that there are several other contributors that can delay diagnosis including access to ED care impeded by transportation and other social determinants of health and even recognition of symptom significance.54 55 For example, several studies have shown that socioeconomic status can influence routine ED use more strongly than race and ethnicity.47 56 However, studies examining the association between socioeconomic status and DE in the ED are lacking.

Our study has several strengths. The search strategy was developed and deployed by an experienced librarian with regular input from the study team to refine the approach. We updated our search in August 2022 during the study to ensure no recent publications were missed. The PRISMA guidelines were followed during all phases of the study. We grouped the eligible studies by the parameters which were considered the most likely sources of heterogeneity—types of clinical problem, outcome assessed, DE definitions and the approach for DE assessment following the SWiM guidelines when writing the narrative synthesis.

Limitations of the study are as follows. The most significant limitation was a substantial heterogeneity among the included studies in demographic variables and DE definitions and measurement approaches, which precluded us from doing a meta-analysis.

Specifically, racial and ethnic groups were sometimes combined and examined as one variable and other studies followed the widely recognised census bureau category definition which separates race (white, black or African American, Asian/Pacific Islander, American Indian/Alaska Native/Native, American mixed race) and ethnicity (Hispanic/non-Hispanic).57 We would also like to reiterate that this nomenclature is evolving and race is a social construct and potentially used in these studies to reflect social determinants of health and its intersectionality with structural racism and not as a marker of genetic differences.58 Insurance categories were combined differently across studies making it difficult to make inferences about the role of insurance and potentially socioeconomic status in DE.

We noted different time cut-offs for categorisation of DE, and approaches to assessment varied from EMR or database review to national survey analysis. Missing data and lack of adjustment might also potentially affect the results in some of the included studies. These challenges created conditions that prevented us from combining large numbers of studies in the quantitative synthesis and drawing more robust conclusions.59

The certainty of evidence was very low due to substantial inconsistency, indirectness in the outcome measure, variability in the exposure measure, and methodological limitations including high risk of confounding in some studies.

It is essential to note that disparities in care and potentially DE may also be impacted by hospital-level factors not directly assessed in this work.16 60 61 For example, lack of availability of diagnostic testing may expose those living in the catchment areas of under-resourced hospitals to delays in diagnosis.62

Concerns about definitions, approaches to measurement and quantifying DE have been noted previously and have made the study of DE challenging and quality improvement initiatives to address this issue difficult to implement.23 Our SR highlights this concern.63–66 The use of a look-forward or a look-back approach may both be helpful and may depend on available data and research infrastructure. AHRQ has issued further guidance to support research and practice in understanding the issues of DE.67 It is also worth noting that patients that re-present to the ED with consistent symptoms may be admitted the second time and those events might be documented as a DE although they may simply reflect that the threshold for admitting those who re-present is lower than at the initial presentation. Additionally, challenges with relying on EMR review to measure DE are a recognised issue but often the methodological approach most accessible to investigators.26

Conclusions

This SR demonstrated consistent increased odds of missed AMI/ACS diagnosis among black patients presenting to the ED compared with white patients in most studies. We did not identify consistent associations in the same direction for other factors or for cerebrovascular/neurological diagnoses. This difference in findings is likely to be explained by the large heterogeneity in definitions and approaches to measurement of DE and variability in demographic variables categorisation found in the present study. The certainty of evidence was very low, and inferences should be interpreted with caution due to inconsistency across studies. Further research is needed to support the identification and measurement of DE so robust comparisons and longitudinal quality improvement initiatives to address and mitigate DE can be conducted. Furthermore, given the widespread issues around healthcare disparities that are noted in the literature we expect a focus on vulnerable population in the realm of DE to be a priority and a consideration for policy makers interested in improving health equity.68

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

Ethics statements

Patient consent for publication

Ethics approval

Not applicable.

References

Supplementary materials

Footnotes

  • Twitter @dongyue

  • Contributors SH, AKB and MHM contributed to the design of the study. SH, JS, CH, JP, YD and AKB screened articles for inclusion and abstracted data from included studies. SH, AKB, BWP and MHM contributed to analysis, and interpretation of the data. SH and AKB drafted the manuscript. All authors (SH, JS, CH, JP, YD, MHM, BWP and AKB) critically revised the article for important intellectual content and approved the final version of the manuscript.

  • 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.

  • 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.

Linked Articles