Missed specialty consultation and hospital admission in 72-hour emergency department return visits

Abstract

BACKGROUND:

Unscheduled 72-hour emergency department (ED) return visits are key quality indicators; however, aggregate return rates alone may mask modifiable clinical vulnerabilities. In high-performing EDs with low baseline return rates, the extent of preventable revisits and the impact of missed index-visit specialty consultations remain poorly defined.

OBJECTIVES:

This study aimed to classify return causes using standardized criteria, determine the proportion of potentially preventable return visits, and evaluate whether missed specialty consultations at the index visit are independently associated with hospital admission.

DESIGN:

Retrospective cohort study

SETTINGS:

Single-center, tertiary care university hospital emergency department.

METHODS:

This study evaluated all ED return visits occurring within 72 hours between April 1, 2024 and April 30, 2025. Two emergency physicians independently classified causes using predefined criteria (κ=0.81). The primary outcome was hospital admission at return, analyzed via multivariable logistic regression.

MAIN OUTCOME MEASURES:

Patient demographics (age, sex), triage zone at both visits, presenting complaints, final diagnoses, specialty consultation status, and disposition. Triage was zone-based: green (low acuity), yellow (moderate acuity), and red (high acuity).

SAMPLE SIZE:

1505 unplanned return visits among 81390 adult ED visits

RESULTS:

The 72-hour return rate was 1.85%. While most returns were disease-related (75.5%, n=1136), 24.5% (n=369) were potentially preventable. Missed consultation opportunities occurred in 14.0% of cases (n=210), with 75.2% (n=158) of these appearing in disease-related returns. Admission rates were significantly higher in the missed-consultation group (38.1% [80/210] vs. 8.0% [103/1295]; P<.001). After adjustment for disease severity (index-visit triage acuity and age), missed consultation showed the strongest association with hospital admission in the multivariable model (adjusted OR 6.84, 95% CI 4.70–9.95), followed by higher index-visit triage acuity (adjusted OR 4.47) and age ≥65 years (adjusted OR 2.32); the model showed good discrimination (c-statistic 0.84).

CONCLUSIONS:

One in four return ED visits was potentially preventable. Missed index-visit specialty consultations were independently associated with markedly higher admission risks and may warrant evaluation as quality improvement targets.

LIMITATIONS:

Retrospective, single-center design; the missed-consultation definition is not externally validated and is subject to information and hindsight bias; residual confounding cannot be excluded, precluding causal inference.

INTRODUCTION

Emergency department (ED) return visits within 72 hours are widely used as indicators of care quality, reflecting the adequacy of initial clinical assessment, treatment, and disposition decisions.^1-4 International benchmarks generally consider rates between 1% and 5% acceptable, with rates below 3% indicating optimal performance.⁵ However, aggregate return rates alone provide an incomplete picture: even high-performing EDs may harbor systematic, modifiable vulnerabilities that remain hidden within a small subset of revisits.⁶

Although a substantial body of literature has examined factors associated with ED return visits, comparatively little attention has been directed toward distinguishing potentially preventable returns from those resulting from unavoidable disease progression-a distinction critical for targeted quality improvement.^7-9 Among potentially modifiable factors, consultation decisions during initial ED encounters represent particularly highimpact clinical junctures.¹⁰ Emergency physicians must rapidly determine whether specialist input is likely to alter diagnosis, treatment, or disposition, balancing patient safety against the overuse of limited resources.¹¹

While prior studies have described ED consultation patterns and their association with outcomes,¹² few investigations have specifically examined patients who did not receive specialty consultation at the index visit but subsequently required consultation upon return. This gap is clinically relevant: when a patient is discharged without specialist input and returns within 72 hours requiring that very consultation, the sequence constitutes a potentially measurable quality signal. The cognitive processes underlying these decisions are complex, and biases such as anchoring and premature closure may contribute to consultation deferrals that later prove insufficient.^13,14

Most existing research has been conducted in settings with return rates exceeding 3%.^15,16 High-performing EDs face a different challenge: identifying preventable failures within a small, heterogeneous subset of revisits where quality signals are less conspicuous.^17,18 This context has received limited systematic attention.

The present study addressed these gaps by systematically analyzing 72-hour ED return visits at a high-performing tertiary center. The study had three primary objectives: first, to classify return visit causes using standardized, hierarchy-based criteria; second, to determine the proportion attributable to potentially preventable causes; and third, to evaluate whether missed specialty consultation opportunities at the index visit are independently associated with hospital admission at the return visit. We hypothesized that, despite an optimal overall return rate, a meaningful proportion of returns would have potentially preventable causes and that missed consultation opportunities would be independently associated with hospital admission.

METHODS

Study design and setting

This retrospective cohort study evaluated all ED return visits occurring within 72 hours at Afyonkarahisar Health Sciences University, Health Practice and Research Center, Afyonkarahisar, Türkiye between April 1, 2024 and April 30, 2025. The ED operates continuously with board-certified emergency physicians and emergency medicine residents, with immediate access to all major subspecialty services. All major specialty consultation services were available on-site or on-call 24 hours a day during the study period, ensuring that missed consultation reflected a clinical decision rather than resource unavailability. The study was approved by the institutional Clinical Research Ethics Committee (approval number 2025/381, date 13 June 2025). Informed consent was waived owing to the retrospective, de-identified design, consistent with the Declaration of Helsinki.

Study population

All adult patients (≥18 years) who returned to the same ED within 72 hours of discharge from a prior ED visit were eligible. A 72-hour return visit was defined as any ED presentation occurring between 0 and 72 hours following discharge from an index visit. Patients were excluded if they were younger than 18 years at the time of the index visit or had incomplete medical records precluding reliable data extraction. Patients who left without being seen at either visit were also excluded. Only returns to the same institutional ED were captured; visits to other hospitals, outpatient clinics, or ambulatory care settings within 72 hours could not be systematically identified through the institutional electronic medical record (EMR).

Data collection

Return visits were identified via automated query of the institutional hospital information system (HIS/EMR), which flagged any adult patient presenting within 72 hours of a prior discharge from the same ED. Data were extracted retrospectively using a standardized abstraction form developed for this study and applied independently by two trained emergency medicine physicians. For each return visit, data from both the index and return encounters were collected: patient demographics (age, sex), triage zone at both visits, presenting complaints, final diagnoses, specialty consultation status, and disposition. Triage was zone-based: green (low acuity), yellow (moderate acuity), and red (high acuity).

Definitions

The following terms are used consistently throughout: (i) return visit—any unplanned ED presentation within 72 hours of discharge from an index visit at the same facility; (ii) hospital admission—admission to a general ward or intensive care unit (ICU) following the return visit; (iii) missed consultation opportunity—absence of specialty consultation at the index visit followed by formal consultation from the same specialty at the return visit, based on structured chart review; (iv) potentially preventable return—return visit classified as physician-related or patient-related per consensus review.

Classification of return visit causes

Each return visit was classified into one of three mutually exclusive categories using criteria adapted from established frameworks.^14,18 When multiple contributing factors were identified, a predefined hierarchy was applied: physician-related factors took precedence over patient-related factors, which took precedence over disease-related factors. This hierarchy was established a priori.

Physician-related factors: Returns attributable to deficiencies in the initial evaluation, management, or disposition, including inadequate diagnostic testing, inappropriate treatment, premature discharge, missed or delayed diagnosis, failure to obtain a clinically indicated specialty consultation, or failure to arrange appropriate follow-up. Representative examples include: a patient re-presenting with pneumonia after discharge with a diagnosis of urinary tract infection following inadequate radiographic evaluation; and a patient returning with acute coronary syndrome after initial discharge with non-specific chest pain without troponin measurement.

Patient-related factors: Returns attributable to patient behavior or circumstances, including non-adherence to discharge instructions, non-emergency re-presentations (e.g., wound dressings, prescription renewals), or inability to access recommended outpatient care due to logistical or socioeconomic barriers. Patient-related factors were determined through structured chart review of discharge documentation and patient-reported reasons where recorded.

Disease-related factors: Returns explained by the expected natural course of the presenting illness, including recurrence or progression of symptoms consistent with the index diagnosis, development of anticipated complications, or new symptoms unrelated to the initial presentation.

Two emergency medicine physicians independently reviewed and classified each return visit, each masked to the other's assessment and applying the pre-specified criteria described above. Because disposition is recorded within the same return-visit record, the reviewers could not be masked to the return disposition; this is an independent duplicate-review protocol rather than a fully outcome-blinded process, and the resulting potential for measurement bias is addressed in the Limitations. Inter-rater agreement was assessed using Cohen's kappa (κ=0.81, substantial agreement). Returns classified as physician- or patient-related were considered potentially preventable; disease-related returns were considered non-preventable.

The clinical relatedness of each index–return visit pair was additionally assessed as: related (same organ system and clinically compatible diagnoses), partially related (same organ system but distinct diagnosis suggesting diagnostic evolution), or unrelated (different organ system).

Missed consultation opportunity analysis

A prespecified sub-analysis examined patients who had no specialty consultation at the index visit but received formal consultation from the same specialty at the return visit within 72 hours. This pattern was operationally defined as a missed consultation opportunity and was determined independently by the same two reviewers applying pre-specified criteria. A consultation was considered to have occurred if a specialty service was formally documented in the EMR by the treating emergency physician. At Afyonkarahisar Health Sciences University, Health Practice and Research Center, specialty consultation is requested exclusively through the hospital information system (HIS), after which the consultant physically evaluates the patient; informal “curbside” or telephone exchanges are not part of the documented consultation workflow and do not substitute for a formal consultation. Undocumented informal exchanges therefore could not be captured, but are unlikely to have materially inflated the missed-consultation frequency given this workflow.

This definition mirrors quality review frameworks used in prior investigations.^14,18 We acknowledge that this operational definition is retrospective and has not been externally validated: the absence of consultation at the index visit may represent appropriate clinical judgment given available information at that time, with subsequent evolution reflecting natural disease progression rather than a quality failure. This inherent limitation is discussed in detail below.

Statistical analysis

All analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Continuous variables were assessed for normality using the Shapiro–Wilk test and are reported as mean (SD) or median (IQR) as appropriate. Categorical variables are reported as frequencies and percentages. Between-group comparisons used independent-samples t-tests or Mann–Whitney U tests for continuous variables and chi-square or Fisher's exact tests for categorical variables.

Multivariable logistic regression examined the independent association between missed consultation opportunity (primary exposure) and hospital admission (primary outcome). Covariates were selected a priori and included recognized markers of baseline disease severity: age (≥65 vs <65 years), sex, triage acuity at the index visit (yellow/red vs. green), and return timing (<24 hours vs. ≥24 hours). Adjusted odds ratios (ORs) with 95% confidence intervals (CIs) are reported.

Model discrimination was assessed with the c-statistic; calibration with the Hosmer–Lemeshow goodness-of-fit test; multicollinearity with variance inflation factors (VIF). A pre-planned sensitivity analysis additionally adjusted for return visit category (physician/patient/disease-related) to assess robustness. All tests were two-tailed; P<.05 was considered statistically significant.

RESULTS

Study population and return rate

During April 2024–April 2025, a total of 81390 adult ED visits were recorded, 1505 of which met the inclusion criteria as 72-hour return visits, corresponding to a return rate of 1.85%—within the lower range of international benchmarks (1–5%) and below the accepted optimal threshold of 3% (Figure 1).

Figure 1.

Study flowchart and hospital admission rates. Panel A: STROBE-compliant patient flow diagram showing 81390 total adult ED visits, identification of 1505 return visits meeting inclusion criteria (72-hour return rate 1.85%), classification into disease-related (n=1136; 75.5%), patient-related (n=334; 22.2%), and physician-related (n=35; 2.3%) categories, and identification of 210 patients (14.0%) with missed consultation opportunities. Panel B: Hospital admission rates comparing patients with missed consultation opportunity (38.1%; 80/210) versus those without (8.0%; 103/1295); χ²=150.89, P<.001. ED, emergency department; ICU, intensive care unit.

Patient characteristics

During the index visit, specialty consultation was performed in 319 patients (21.2%). At the return visit, this proportion increased to 402 patients (26.7%), consistent with greater clinical complexity at re-presentation. Mean age was 43 years (SD 19; range 18–92). Male patients constituted 52.2% (n=785) and female patients 47.8% (n=720). Return visits occurred within 0–24 hours in 38.9% (n=585), 24–48 hours in 38.9% (n=586), and 48–72 hours in 22.2% (n=334) (Table 1).

Table 1.

Baseline characteristics of 72-hour return visits.

Characteristic	n=1505
Age, years	43±19
Sex (%)
Female	720 (47.8)
Male	785 (52.2)
Timing of return visit (%)
0–24 hours	585 (38.9)
24–48 hours	586 (38.9)
48–72 hours	334 (22.2)
Triage acuity at index visit (%)
Green zone (low acuity)	996 (66.2)
Yellow zone (moderate acuity)	505 (33.6)
Red zone (high acuity)	4 (0.3)
Triage acuity at return visit (%)
Green zone (low acuity)	949 (63.1)
Yellow zone (moderate acuity)	545 (36.2)
Red zone (high acuity)	11 (0.7)
Worsening triage acuity between visits (%)	137 (9.1)

Note: Data presented as n (%) and mean±standard deviation (SD)

Triage and clinical characteristics

At the index visit, patients were triaged to the green zone in 66.2% (n=996), yellow zone in 33.6% (n=505), and red zone in 0.3% (n=4). At the return visit: green zone 63.1% (n=949), yellow zone 36.2% (n=545), red zone 0.7% (n=11) (Table 1). A total of 137 patients (9.1%) demonstrated worsening triage acuity between visits.

Classification of return visit causes

Disease-related factors were the primary cause in 75.5% (n=1136) of cases. Patient-related factors were identified in 22.2% (n=334), most commonly medication non-adherence (14.0%, n=211) and non-urgent re-presentation (7.1%, n=107). Physician-related factors were present in 2.3% (n=35), comprising missed diagnoses (2.0%, n=30) and treatment errors (0.3%, n=5). In total, 369 return visits (24.5%) were classified as potentially preventable (Table 2).

Table 2.

Classification and outcomes of 72-hour return visits (n=1505).

Characteristics	n (%)
Classification of return visits
Disease-related factors	1136 (75.5)
Patient-related factors	334 (22.2)
Medication non-adherence	211 (14.0)
Non-urgent re-presentation	107 (7.1)
Other patient-related	16 (1.1)
Physician-related factors	35 (2.3)
Missed diagnosis	30 (2.0)
Treatment error	5 (0.3)
Potentially preventable (total)	369 (24.5)
Clinical relatedness of index and return visitsa
Related (same system, compatible diagnosis)	948 (63.0)
Partially related (same system, different diagnosis)	197 (13.1)
Unrelated (different organ system)	360 (23.9)
Different diagnosis at return visit	516 (34.3)
Physician-related factors in diagnosis changes	20/516 (3.9)
Missed consultation opportunity — totalb	210 (14.0)
Among disease-related returns	158 (75.2)
Among patient-related returns	35 (16.7)
Among physician-related returns	17 (8.1)
Most common specialties consulted at return (missed-consultation group)
General surgery	30 (14.3)
Ophthalmology	29 (13.8)
Pulmonology	22 (10.5)
Internal medicine	18 (8.6)
Cardiology	13 (6.2)
Disposition at return visit
Discharged home	1165 (77.4)
Ward admission	152 (10.1)
ICU admission	31 (2.1)
Left against medical advice	141 (9.4)
Transferred to another facility	14 (0.9)
In-hospital mortality	2 (0.1)
Total hospital admission (ward + ICU)	183 (12.2)
Hospital admission by missed-consultation statusc
With missed consultation — admitted	80/210 (38.1)
Without missed consultation — admitted	103/1295 (8.0)

ICU, intensive care unit.

^aRelatedness assessed by structured review: related=same organ system and clinically compatible diagnoses; partially related=same organ system but distinct diagnosis; unrelated=different organ system.

^bDefined as absence of specialty consultation at the index visit followed by formal consultation from the same specialty at the return visit, as determined independently by two reviewers applying pre-specified criteria.

^cχ²=150.89, P<.001.

Diagnosis comparison revealed that 34.3% (n=516) of patients received a different diagnosis upon return. Among these, physician-related factors accounted for 20 cases (3.9%), including clinically significant missed diagnoses such as pneumonia (initially diagnosed as urinary tract infection) and acute coronary syndrome (initially classified as non-specific chest pain).

In terms of clinical relatedness, 948 return visits (63.0%) were classified as related to the index visit (same organ system and compatible diagnoses), 197 (13.1%) as partially related (same organ system, distinct diagnosis), and 360 (23.9%) as unrelated (different organ system) (Table 2).

Missed consultation opportunities

A total of 210 patients (14.0%) had a missed consultation opportunity. The distribution across return visit categories was: disease-related returns 158 (75.2%), patient-related returns 35 (16.7%), and physician-related returns 17 (8.1%) (Table 2). This distribution demonstrates that missed consultations were not confined to physician-error cases; the majority occurred in disease-related returns, suggesting that clinical evolution of the underlying illness was a frequent contributing mechanism.

Among patients with a missed consultation opportunity, 38.1% (n=80) required hospital admission at the return visit, compared with 8.0% (103/1295) among those without (χ²=150.89, P<.001) (Table 2; Figure 1). Notably, all 80 admitted patients in the missed-consultation group had received specialty consultation at the return visit, consistent with the consulting service having determined or guided the admission decision.

The most frequently consulted specialties at the return visit among patients with missed consultation were: general surgery (14.3%, n=30), ophthalmology (13.8%, n=29), pulmonology (10.5%, n=22), internal medicine (8.6%, n=18), and cardiology (6.2%, n=13).

Most common diagnoses

The five most common diagnoses at the index visit were upper respiratory tract infections (18.7%, n=281), soft tissue trauma (10.1%, n=152), myalgia (7.4%, n=112), non-specific pain (6.8%, n=102), and acute gastroenteritis (5.0%, n=75). Similar patterns predominated at the return visit.

Disposition outcomes

At the return visit, 77.4% (n=1165) were discharged, 10.1% (n=152) were admitted to a general ward, 2.1% (n=31) were admitted to the ICU, 9.4% (n=141) left against medical advice, 0.9% (n=14) were transferred, and 0.1% (n=2) died in-hospital. The overall hospital admission rate (ward + ICU) was 12.2% (n=183) (Table 2). Geriatric patients (≥65 years) had a substantially higher admission rate (75/275, 27.3%) than those aged 18–50 years (64/998, 6.4%) or 51–64 years (44/232, 19.0%).

Risk factors for hospital admission

Multivariable logistic regression adjusted for age, sex, index-visit triage acuity, and return timing revealed that missed consultation opportunity was independently associated with a 6.84-fold increase in the odds of hospital admission (adjusted OR 6.84, 95% CI 4.70–9.95, P<.001). Higher triage acuity at the index visit (adjusted OR 4.47, 95% CI 3.10–6.45, P<.001) and age ≥65 years (adjusted OR 2.32, 95% CI 1.60–3.38, P<.001) were additional independent predictors, whereas return within 24 hours (adjusted OR 1.38, 95% CI 0.97–1.96, P=.074) and sex were not statistically significant. The model showed good discrimination (c-statistic 0.84) and adequate calibration (Hosmer–Lemeshow P>.05), with no evidence of multicollinearity (VIF 1.00–1.11) (Table 3).

Table 3.

Multivariable logistic regression: Independent predictors of hospital admission at return visit (n=1505).

Variable	Adjusted OR	95% CI	P value
Missed consultation opportunitya	6.84	4.70–9.95	<.001
Higher triage acuity at index visit (yellow/red vs green)b	4.47	3.10–6.45	<.001
Age ≥65 years (vs <65)	2.32	1.60–3.38	<.001
Return within 24 hoursc	1.38	0.97–1.96	.074
Male sexd	1.11	0.78–1.57	.563
Model performance
C-statistic	0.84	—	—
Hosmer–Lemeshow test	—	—	>.05
Variance inflation factor range	1.00–1.11	—	—
Sensitivity analysis (+revisit category)e	6.38	4.36–9.33	<.001

OR, odds ratio; CI, confidence interval.

^aAbsence of specialty consultation at index visit followed by formal consultation from the same specialty at return visit.

^bReference category: green zone (low acuity).

^cReference category: return at ≥24 hours (24–72 hours).

^dReference category: female sex.

^ePrimary model additionally adjusted for return visit category (physician/patient/disease-related); the modest attenuation in effect size from 6.84 to 6.38 is expected given partial correlation between missed consultation and return visit category. C-statistic >0.70 indicates good discrimination; Hosmer–Lemeshow P>.05 indicates good calibration; VIF <2.0 indicates no multicollinearity.

In the pre-planned sensitivity analysis additionally adjusting for return visit category (physician/patient/disease-related), the association between missed consultation and hospital admission remained highly significant and virtually unchanged (adjusted OR 6.38, 95% CI 4.36–9.33, P<.001), confirming the robustness of the primary finding and indicating that it is not explained by return visit category.

DISCUSSION

Principal findings

This study analyzed 72-hour ED return visits at a tertiary academic center and identified three principal findings. First, despite an overall return rate of 1.85%—well below the 3% optimal threshold—nearly one quarter of returns were classified as potentially preventable. Second, 14.0% of all revisiting patients had a missed consultation opportunity, and this group exhibited dramatically higher hospital admission rates. Third, missed consultation opportunity showed the strongest association with hospital admission in the multivariable model, conferring a more than sixfold increase in adjusted odds—surpassing both index-visit triage acuity and older age, established severity markers in the literature.

Return rate and preventability

The 72-hour return rate of 1.85% is substantially lower than the 3.2–5.5% range typically reported in the literature.^15,16,19 Yet, as demonstrated by Grabinski et al,⁶ structured analysis can uncover actionable quality improvement opportunities even in high-performing departments. In our cohort, approximately one quarter of returns were classified as potentially preventable—consistent with findings from comparable high-performing centers^9,14—with patient-related factors predominating over physician-related factors by nearly tenfold. The observation that more than one-third of patients received a different diagnosis upon return underscores the diagnostic vulnerability inherent in emergency settings.^20,21

Missed consultation opportunities as a quality indicator

The most striking finding is the strong and independent association between missed consultation opportunities and hospital admission. Patients who had no specialty consultation at the index visit but required consultation upon return experienced an admission rate of 38.1%—nearly five times that of other return patients (8.0%). This association persisted after multivariable adjustment, with an adjusted OR of 6.84, and remained robust in the sensitivity analysis including return visit category (adjusted OR 6.38).

Importantly, the distribution of missed consultations across return visit categories—75.2% occurring in disease-related returns, 16.7% in patient-related, and only 8.1% in physician-related—provides critical context. These findings indicate that most missed consultations did not represent straightforward physician errors, but rather reflected cases where clinical evolution of the underlying disease made specialist involvement necessary only at the time of return. This nuance is essential when interpreting the association with adverse outcomes.

Several mechanisms may explain the overall association. Patients requiring specialty input often present with diagnostic complexity exceeding the scope of isolated emergency evaluation. Consultation deferral may reflect cognitive biases such as anchoring or premature closure.¹³ Systems-level factors—perceived consultant availability, institutional consultation norms, and time pressures—may further contribute. Prior systematic work by Voaklander et al¹² demonstrated that consultation-focused interventions can improve both efficiency and care quality, lending external plausibility to our findings. Few prior studies have operationalized missed consultation opportunity as a measurable quality indicator with quantified clinical consequences; this study extends that framework to a high-performing setting.

Clinical and policy implications

These findings suggest that consultation decision-making may represent an important area for quality improvement in emergency care. These findings should not be interpreted as establishing a causal relationship between consultation decisions and admission outcomes; rather, missed consultation may function as a clinical marker identifying patients whose diagnostic complexity exceeded the scope of the initial evaluation. Targeted review of patients discharged without specialty consultation may enable early identification of those at elevated admission risk. Development of standardized consultation thresholds or clinical decision support tools²² could reduce unwarranted variation and support emergency physicians in complex decision-making. Incorporating missed consultation opportunities as a routine quality metric may provide a more sensitive signal of care gaps than aggregate return rate data alone. Notably, our findings do not establish that increasing specialty consultation at the index visit would reduce subsequent admissions; this possibility requires prospective evaluation.

However, any quality improvement initiative in this area must carefully weigh resource implications. Systematic expansion of consultation volume may prolong ED length of stay, place additional demands on specialist services, and increase costs²³—factors that are particularly relevant in resource-constrained settings. Interventions should therefore be prospectively evaluated with attention to both clinical benefit and operational feasibility before broad implementation. Importantly, these findings should not be interpreted as advocating routine or indiscriminate consultation for all patients presenting with diagnostic uncertainty; rather, they highlight a specific subset—those who return within 72 hours requiring specialist input—as a target for structured quality review.

Several limitations merit consideration. First, the retrospective design is subject to documentation-dependent information bias. Undocumented verbal or informal telephone consultations at the index visit would be misclassified as missed opportunities, potentially inflating the frequency of this phenomenon. In addition, because chart abstraction necessarily revealed the return disposition, reviewers could not be masked to the outcome, a further potential source of measurement bias.

Second, and most critically, the operational definition of missed consultation opportunity is inherently retrospective, has not been externally validated, and is subject to hindsight bias. As evidenced by the finding that 75.2% of missed consultations occurred in disease-related returns, many cases likely reflect clinically appropriate management at the index visit followed by unpredictable disease progression rather than a genuine quality failure. Two independent reviewers applying predefined criteria with substantial agreement (κ=0.81) mitigated but cannot eliminate this bias. Future prospective studies should employ blinded adjudication panels with pre-specified consultation criteria.

Third, consultation availability varies by time of day and day of week; we were unable to systematically determine what proportion of missed consultations occurred during overnight or weekend shifts when specialist access is typically more limited. This represents an uncontrolled source of confounding. Fourth, the single-center design limits generalizability; consultation culture, specialty availability, and ED organization differ substantially across healthcare systems. In particular, our findings may not generalize to settings lacking 24-hour on-site or on-call specialty availability, where a missed consultation may reflect resource constraints rather than a discretionary clinical decision. Fifth, only same-facility return visits were captured; returns to other hospitals or outpatient settings within 72 hours could not be identified, potentially underestimating the true return rate and altering cohort composition. Sixth, patients who left without being seen were excluded; a recent study reports an 18% seven-day return rate in this subgroup.¹ Seventh, comorbidity burden was not systematically captured and may represent residual confounding—including confounding by indication, whereby greater illness severity can simultaneously prompt specialty consultation and independently increase admission risk—that the model could not fully account for. Finally, causal inference cannot be established from this observational design. This study has several strengths that should be weighed against its limitations. Consecutive return visits were examined through structured chart abstraction by two independent reviewers using predefined criteria, with substantial inter-rater agreement (κ=0.81). The analysis distinguished disease-related returns from encounters that met an a priori definition of missed consultation opportunity and related consultation patterns to subsequent disposition, yielding clinically granular information on a sparsely described quality gap. Complete electronic records and specialty consultation documentation at a single center further allowed consistent capture of documented consultative activity at the index visit.

CONCLUSIONS

In an ED achieving optimal aggregate return visit metrics, nearly one quarter of 72-hour returns were attributable to potentially preventable causes. Missed specialty consultation opportunities at the index visit were independently and strongly associated with hospital admission, with a nearly fivefold higher admission rate (38.1% vs. 8.0%) and an adjusted OR of 6.84 that remained robust on sensitivity analysis. Importantly, most missed consultations occurred in disease-related returns, highlighting the role of unpredictable clinical evolution rather than physician error alone. These findings suggest that consultation decision-making may represent an important focus for future quality improvement research in emergency care. Prospective studies with blinded outcome adjudication are needed to establish causality and evaluate whether consultation-focused interventions can reduce preventable admissions. The results should be regarded as hypothesis-generating and do not demonstrate that expanding index-visit consultation would, by itself, lower admissions.

Funding Statement

None.

AUTHORS CONTRIBUTIONS

AE: Conception, design, data collection, analysis, manuscript writing. MS: Design, supervision, data collection, materials. UR: Literature review, analysis and interpretation. All authors read and approved the final manuscript.

DATA AVAILABILITY

The data are available from the corresponding author upon reasonable request.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE

The study was approved by the institutional Clinical Research Ethics Committee (approval number 2025/381, date 13 June 2025). Informed consent was waived owing to the retrospective, de-identified design, consistent with the Declaration of Helsinki.