Article Text

Download PDFPDF

The impact of consultation on length of stay in tertiary care emergency departments
  1. Craig Brick1,
  2. Justin Lowes2,
  3. Lindsay Lovstrom2,
  4. Andrea Kokotilo2,
  5. Cristina Villa-Roel2,3,
  6. Patricia Lee4,
  7. Eddy Lang4,
  8. Brian H Rowe2,3
  1. 1School of Medicine, University College Cork, Cork, Ireland
  2. 2Department of Emergency Medicine, Faculty of Medicine & Dentistry, University of Alberta, Edmonton, Alberta, Canada
  3. 3School of Public Health, University of Alberta, Edmonton, Alberta, Canada
  4. 4Division of Emergency Medicine, Faculty of Medicine, University of Calgary, Calgary, Alberta, Canada
  1. Correspondence to Dr Brian H Rowe, Department of Emergency Medicine, University of Alberta, 1G1.42 Walter C. Mackenzie Centre, 8440 112 Street, Edmonton, Alberta, Canada T6G 2B7; brian.rowe{at}ualberta.ca

Abstract

Background Consultations in the emergency department (ED) are infrequently studied. This study quantifies the contribution of consultations to ED length of stay (LOS) and examines patient and consultation characteristics associated with prolonged ED LOS.

Methods Prospective cohort study of a convenience sample of shifts by volunteering emergency physicians (EP) at two urban tertiary care Canadian EDs. EPs completed standardised forms on all patients for whom a consultation was requested. Medical chart reviews and secondary analyses of administrative databases were also performed. Factors associated with longer LOS were determined through linear regression modelling.

Results 1180 patients received at least one consultation during study shifts and EPs completed data collection on 841 (71%) of these. Median patient age was 54 years, 53.3% were male, and 2.9% had documented dementia. Admitted patients receiving consultations had a longer overall LOS compared to discharged patients. Median time from triage to consultation request accounted for approximately 28% of the total median LOS in admitted patients compared to 46% for discharged patients. Consultation decision time accounted for 33% and 54% of the LOS for admitted and discharged patients, respectively. Linear regression modelling revealed that advanced age, longer latency between arrival and first consultation request, history of dementia and multiple consultations were significantly associated with longer LOS. Conversely, undergoing procedures while in the ED was associated with a shorter LOS.

Conclusions Consultation decision time contributes significantly to ED LOS. Further efforts are needed to validate these results in other ED settings and improve this aspect of ED throughput.

  • Emergency Department
  • Emergency Care Systems
  • Research, Epidemiology
View Full Text

Statistics from Altmetric.com

Introduction

Emergency department (ED) overcrowding occurs when the demand for services exceeds the ability to provide quality care within appropriate time frames, resulting in delays in time-sensitive diagnostics and interventions, dissatisfaction among patients and staff, and poor outcomes.1 A conceptual model of flow divides the overcrowding into three components: input, throughput and output.2 Benchmarks for ED discharge have been proposed,3 but achieving them will require multifaceted approaches, including improved consultation strategies and availability of inpatient beds.

Consultations occur frequently in emergency medicine,4 ,5 and involve an emergency physician (EP) requesting participation in a patient's care from another specialist.6 This process should result in one of the following outcomes: (1) admission to hospital; (2) completion of procedures or investigations; (3) discharge; or (4) another specialty consultation request.7 Consultation delays affect both throughput and output, and are of growing concern,8 given the severity of overcrowding in many EDs.9 Furthermore, many professional organisations have recommended the establishment of consultation time frames in order to improve patient flow.1

There has been limited research regarding the contribution of consultations to patient flow.6 There is evidence that improvements in consultation policies can alleviate ED overcrowding8 ,11 and several studies have examined consultations for ED patients.12–14 While specialty consultation has been associated with increased ED length of stay (LOS), there are few data regarding the relative contribution of consultations to overall stay.6 Recently, advice for consultations have been produced; however, these were neither evidence-based nor considered reasonable.10 This study examined the contribution of consultations to overall ED LOS and sought to determine patient characteristics and consultation factors influencing ED LOS. Our hypotheses were that consultation delays were an important component of overall ED LOS and that ED LOS would be substantial for any patient with multiple consultations.

Methods

Study design

A prospective cohort study was carried out on a sample of patients seen during the shifts of volunteering EPs at two urban Canadian tertiary care teaching hospitals between July and September 2010; a total of five shifts were selected for each EP involving day, evening, weekend and night shifts.

Setting

The study was conducted at two hospitals in Edmonton, Alberta (University of Alberta hospital (UAH) and Royal Alexandra Hospital (RAH)). Both hospitals are academic regional referral centres and provide a large number of specialist services. Together their EDs manage over 130 000 patient visits per year.

Data collection

Data on demographic factors, severity measured by the Canadian Triage and Acuity Scale (CTAS) score, presenting complaint, mode of arrival and first physician involved (eg, EP, resident, etc) were collected from medical records and the electronic ED information system (EDIS) using a standardised form.

In both centres, ED physicians initiate a consultation through a central paging system which generally is answered by a resident physician. The time the consultant is paged, as well as the time when contact occurs, was electronically recorded in the EDIS; however, consultant arrival is not routinely recorded. While disposition time was not recorded either, it was estimated by obtaining the time of the in-patient admission or discharge order from the EDIS.

The times of ED arrival/triage, bed location (time first seen by a nurse), assessment (initial assessment by a triage liaison physician (TLP) or EP)15 and consultation request were documented. No information was collected on the level of training of the responding resident physician. The time of a bed request (for admitted patients) or ED discharge was also noted. Information on the EP preliminary diagnosis, any history of dementia, investigations prior to consultation, procedures in the ED and the consultant's final diagnosis were also collected.

EPs completed standardised forms on all patients over 17 years of age for whom a consultation was requested from any service, with the exception of non-clinical consultations (radiology and social services), services having automatic ambulance activation (eg, acute stroke/myocardial infarction) and ED-based services (eg, trauma team). The service, priority and mechanism of consultation were documented for a maximum of two consultations per patient. EPs then rated their perception of ‘delayed’ consultations and were asked to elaborate on the perceived cause(s).

Outcomes

The primary study outcome was the proportion of LOS attributable to the consultation process; subanalyses based on admission status, age distribution and single/multiple consultations were also performed. Total ED LOS was subdivided into intervals: ‘bed location’ (time from triage to time of ED bed location), ‘MD [medical doctor] assessment’ (time from ED bed location to time first seen by an EP), ‘MD decision time’ (time from first seen by an EP to time of first consultation request), ‘consult time’ (time from first consultation request to time of disposition decision, which represents the time the bed was requested for admitted patients or the time of ED discharge for discharged patients) and ‘emergency inpatient time’ (time from bed request to time of ED departure for admitted patients only).

Statistical considerations

Analyses

Data were analysed using Stata Release 11 (StataCorp LP, College Station, Texas, USA). Proportions are reported for categorical data; continuous data are reported as medians with IQRs. Mann –Whitney and χ2 tests were used for comparison of medians and proportions, respectively. The influence of individual patient characteristics and consultation factors on LOS was analysed through a multiple linear regression model. No cluster effects by study site or physician were considered due to similarities between the EDs in terms of characteristics, resources and specialist care provided,12 as well as among physicians’ practice.

For the regression model, LOS was the dependent variable (defined as the time from triage to admission decision for admitted patients and time from triage to discharge for discharged patients); statistically significant and clinically relevant independent associations were included: age (<65 vs ≥65 years), sex, arrival mode (eg, ambulance vs other mode of arrival), severity (CTAS level 3 vs other CTAS levels), dementia, time between arrival and consultation request, imaging prior to consultation, procedures while in the ED (eg, joint/fracture reduction, cardioversion) and having two or more consultations. A stepwise forward selection procedure was followed to select important predictors; interactions between age and documentation of dementia, and having two or more consultations were tested. Two-tailed results with p<0.05 were considered statistically significant.

Sample size

This was a convenience sample of consultations and efforts were made to provide narrow CIs; a minimum sample size of 818 patients with individual data were needed to detect a significant difference with an α error of 5% and a β error of 20%. From previous research, we determined that we would require approximately 200 shifts to obtain 1000 consultations.12

Ethics

The protocol for this study was approved by the Health Research Ethics Board of the University of Alberta. Patients did not provide consent; however, all EPs participating in the study provided consent.

Results

Overall, 7460 consultations were recorded during the study period; 1180 (16%) occurred during study-assigned EP shifts and EPs completed 71% of the consultation forms (figure 1). Forms could not be recovered for the remaining 339 (29%) consultations; however, patient information and time stamps were collected at the end of the study from patient charts, chart review and the EDIS. The median number of study forms per physician that were not completed was 5 (IQR 3–10).

Figure 1

Data collection of consultation information during the study period. ED, emergency department; EP, emergency physician.

Patient characteristics

Our sample was obtained from a total of 21 956 adult patients who visited the UAH and RAH EDs during the study period (52.1% males, median patient age 45 years (IQR 29–61), CTAS 3: 49.2%). Characteristics of the patients requiring a consultation during the study period are described in tables 1 and 2. Of the 1180 patients receiving consultations, 663 (56.2%) were admitted, 513 (43.5%) were discharged, six (0.5%) required prolonged observation in the ED and four (0.3%) died in the ED. Admitted patients were older (77 vs 58 years of age; p<0.001), had more severe presentations (CTAS 1/2: 47.5% vs 36.6%; p<0.001) and were more likely to have documented dementia (4.4% vs 1.0%; p<0.001) than discharged patients. Elderly (≥65 years old; 379, 32.1%) and younger patients had a similar CTAS distribution; however , elderly patients were more likely to have documented dementia (8.2% vs 0.4%; p<0.001).

Table 1

Characteristics of patients (N=1180) during the study period

Table 2

Investigations, procedures and admissions in the study population based on triage level

Consultation characteristics

Of the 1180 consultations requested, 883 (74.8%) were originally seen by an EP, 162 (13.7%) by a senior emergency resident, 65 (5.5%) by a junior emergency resident and 70 (5.9%) by a medical student; 6.4% of the patients requiring consultations were seen by a TLP.

Of the 841 EP-documented consultations, 729 (86.7%) patients were reviewed by a single consulting service, while 92 (10.9%) required two or more consultants. Patients requiring two or more consulting services were similar in age (55 years), sex, dementia status and illness severity. The number of consulting services could not be determined for 20 patients (2.4%).

The most common initial services consulted were general internal medicine (GIM, n=107, 12.7%) and cardiology (n=98, 11.6%). Of the 92 patients requiring additional consultations, the most common second service consulted was GIM (n=16, 17.4%) followed by orthopaedics and general surgery (n=10, 10.9% each).

The most common reason stated for all consultations was ‘transfer of care for admission’ (n=439, 52.2% (first consultation) and n=42, 45.6% (second consultation)). Other reasons for consultation included requests for special investigations or treatment (109, 12.9%) and urgent therapy or opinion (96, 11.4%). The most common mechanism of consultation was to page the resident physician directly in both the first (n=562, 66.8%) and second (n=50, 54.3%) consultations.

Consultation delays

Delays in consultation were reported in 170 (20.2%) of initial consultations with the most common reasons being ‘busy service’ (n=47, 27.6%). Of the 92 patients requiring an additional consultation, fewer (13, 14.1%) experienced delays in the consultation process. Patients with multiple consultations were more severely ill at presentation (CTAS 1/2 categories 51% vs 42%); however, age (56 vs 55 years) and sex (61% vs 51% males) were similar.

Length of stay

Admitted patients stayed longer than discharged patients (14.4 h vs 9.5 h; p<0.001); however, consult time contributed 54.4% to ED LOS for discharged patients and 32.7% for admitted patients (4.9 h, IQR 2.6–10.0 vs 3.5 h, IQR 2.1–5.9; p<0.001; figure 2). LOS was longer for older than younger patients (15.5 h vs 10.3 h; p<0.001), with ‘MD decision time’ being the only interval to show statistically significant differences (2.6 h vs 2.1 h; p<0.001). Finally, LOS was longer for patients requiring multiple compared to a single consultation (15.9 h vs 11.1 h; p=0.011). While illness severity at presentation affects both the timing and number of consultations in univariate analyses (table 2), severity as measured by CTAS did not contribute significantly to the model (median LOS: CTAS 1/2: 11:30 h; CTAS 3: 12:00 h, CTAS 4/5: 10:25 h).

Figure 2

Intervals constituting overall emergency department stay by admission status, number of consultations (1 vs 2 or more) and age (<65 vs ≥65 years). ED, emergency department; EIP, emergency in-patient; MD, medical doctor.

In multivariable modelling, LOS was significantly influenced by time from arrival to consultation request (p<0.001), age≥65 years (p<0.001), having ED procedures (p=0.034) and having multiple consultations (p<0.001); the interaction between dementia and having multiple consultations was significant (p<0.001). The final regression model yielded an R2 value of 0.201 (p<0.0001). Removal of factors with coefficient t values with significance levels <0.05 did not suggest the presence of confounding.

Discussion

Patient flow through acute care is closely linked to health outcomes and throughput issues have been the focus of much research attention.16 While ED consultation has been shown to increase overall LOS,12 this is one of the first studies to quantify the contribution of consultations to patient flow, and to identify patient and consultation characteristics associated with delays. Consult time was found to account for 33% of LOS for admitted patients and 54% for discharged patients. This disparity likely reflects higher acuity patients receiving more timely consultations and clearer decision-making than patients with lower acuity. Conversely, patients with primary mental health presentations had less frequent admissions and more consultation delays prior to discharge. These findings are consistent with previous studies reporting longer ED consult times among lower acuity patients.11 Importantly, admitted patients in this study faced an additional 6.7 h wait in the ED before admission inpatient ward occurred, therefore reducing the relative contribution of other time intervals including ‘consult time’ to LOS. This issue of ‘access block’ reflects system-wide pressures but may also create an environment of defeatism conducive to greater consultation delays (‘What's the rush, there is no place for the patient to go anyway’).

The most common reason for consultation being ‘transfer of care for admission’ is consistent with a systematic review identifying high proportions of patients receiving consultations and requiring admission among urban tertiary care EDs.6 While improved consultation efficiency could reduce LOS, interventions in this process have yielded widely disparate results.6 More research regarding the efficacy of interventions aimed at improving consultation efficiency is needed.

The most common reasons for consultation delay suggest that contingency strategies for overwhelmed consulting services and improved communication between EPs and consultant staff may play a role in improving patient flow. Multiple attempts are often required to reach a consulting service,17 a significant proportion of consultations are reported as difficult by EPs,12 and the proportion of consultations perceived as ‘difficult’ increases during periods of high patient volume.12 It is important to note the bidirectional nature of the consultation process, and more research is therefore needed to explore the opinions of consulting services as to the nature of consulting delays. Two key interventions that have been shown to be effective in reducing consult times involved an enhanced emphasis on consultation timing,8 and the involvement of senior doctors in the consulting service regarding patient disposition.11

The advanced modelling allowed us to estimate the impact of patient and consultation characteristics on patient delays. Each hour delay from ED arrival to consultation request was associated with a 1.4 h increase in LOS; approximately one-third of all consultations involved elderly patients (≥65 years old), who spent 5.1 h longer in the ED; patients with documented dementia who required multiple consultations experienced an extra 7.9 h in the ED; finally, patients receiving ED procedures spent 4.3 h less while patients requiring multiple consultations spent an extra 6.8 h in the ED.

While these factors require validation, they do identify areas of particular concern and potential targets for future research. These features may disproportionately impact patient flow, and specific strategies designed to address these factors may be useful in mitigating delays. On the other hand, requiring a procedure likely reflects the implementation of a definitive management plan followed by a relatively straightforward disposition decision, and is associated with faster throughput.

Limitations

There are several limitations to this study. First, it was not possible to randomise EP shifts due to rapidly changing EP schedules and our convenience sample may not be representative of the larger population. Second, the inclusion of four daytime/evening shifts and one night shift for each participating EP may have resulted in an over-representation of daytime consultations, which involve a different set of issues than night time consultations. Times of presentation, however, were very similar to those in a previous study conducted at the same sites.12 Third, missing data were common; however, strategies to mitigate this issue were employed: in 33% of the study forms inconsistent timestamps were validated with chart review and electronic tracking systems available at both sites. In the case of a missing timestamp (eg, bed requisition time), the time intervals calculated using that timestamp (e.g., consult time and emergency inpatient time) could not be included (<10% of the total timestamp data). Fourth, all but one of the 66 physicians approached for the study agreed to participate, making the possibility of volunteer bias remote. Despite high participation, we recognise that ED physicians differ in the completeness of work-up and timing of a consultation request, and this was not fully investigated in this study. Physician self-report was used to define ‘delayed’ consultation, which may vary from physician to physician and from the consultant's perspective. There is a possible Hawthorne effect since EPs knew their shift was included in the study; however, the EPs were unaware of the hypothesis of the study. Fifth, consultations with radiology and social services were excluded, with a focus instead on strictly clinical consultations; further research is needed to evaluate the contribution of non-clinical consultations to LOS in the ED. Sixth, the staff-to-resident physician consultation model used in the study centres may not be generalisable to other centres, which may use different consultation approaches (eg, staff-to-staff). In addition, while the centres studied require a ‘consultation’ in order to admit a patient, in other centres an EP can admit a patient without the input of another specialist. Canadian EPs are discouraged by their national organisation from writing admitting orders.18

Conclusion

Consultations are integral to the practice of emergency medicine; however, they contribute significantly to overall LOS for all patients. Being elderly, having dementia and requiring evaluation by multiple consulting services were strongly associated with prolonged LOS. Understanding these patient subgroups provides opportunities to explore ways of improving patient flow within the ED. Innovative and cooperative strategies involving consulting services and EPs aimed at reducing consult times need to be investigated as part of any comprehensive strategy to mitigate ED overcrowding.

Acknowledgments

The research team thank Diane Milette for her secretarial support. The study team would also like to thank the physicians at both sites who participated in the study.

References

View Abstract

Footnotes

  • Data from this study were presented at the Canadian Association of Emergency Physicians Annual Scientific Meeting, St. John's, Newfoundland, Canada, 4–8 June 2011.

  • Funding The Emergency Medicine Research Group (EMeRG) in the Department of Emergency Medicine at the University of Alberta provided funding for this study. CV-R is supported by a graduate student award from the Canadian Institutes of Health Research (CIHR) in partnership with the Knowledge Translation branch. BHR is supported by the CIHR as Tier 1 Canada Research Chair in Evidence-based Emergency Medicine from the Government of Canada (Ottawa, ON).

  • Contributors All authors contributed to the project and take responsibility for its contents. CB contributed to the study protocol, data collection and interpretation, and manuscript preparation and revision; JL and LL contributed to data extraction, cleaning and interpretation; AK contributed to the study protocol and ethics approval; CV-R coordinated contributors to the study, conducted statistical analyses, and contributed to data interpretation and manuscript revision; PL and EL contributed to the study design/protocol and manuscript revision; BHR conceived the project, contributed to the study protocol, obtained ethics approval, assisted with data collection, analysis and interpretation, manuscript preparation and revision, and was responsible for study supervision/funding.

  • Competing interests None.

  • Ethics approval The Health Research Ethics Board of the University of Alberta approved this study.

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

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.