Predictive factors for longer length of stay in an emergency department: a prospective multicentre study evaluating the impact of age, patient's clinical acuity and complexity, and care pathways
- Enrique Casalino1,2,3,
- Mathias Wargon1,2,3,
- Anne Peroziello1,4,
- Christophe Choquet1,3,
- Christophe Leroy5,
- Sebastien Beaune6,
- Laurent Pereira1,2,3,
- Julien Bernard1,2,3,
- Jean-Claude Buzzi4
- 1Emergency Department, Assistance Publique-Hôpitaux de Paris (AP-HP), University Hospital Bichat-Claude Bernard, Paris, France
- 2Université Paris Diderot, Sorbonne Paris Cité, France
- 3Study Group for Efficiency and Quality of Emergency Departments and Non-Scheduled Activities Departments, Paris, France
- 4Medical Information Systems Program (PMSI), University Hospital Bichat-Claude Bernard, Paris, France
- 5Emergency Department, Assistance Publique-Hôpitaux de Paris (AP-HP), University Hospital Louis Mourier, Paris, France
- 6Emergency Department, Assistance Publique-Hôpitaux de Paris (AP-HP), University Hospital Beaujon, Paris, France
- Correspondence to Professor Enrique Casalino, Service d'Accueil des Urgences, Hôpital Bichat-Claude Bernard, 46 rue Henri Huchard, 75018 Paris, France;
- Received 13 November 2012
- Revised 23 January 2013
- Accepted 5 February 2013
- Published Online First 28 February 2013
Background It has been reported that emergency department length of stay (ED-LOS) for older patients is longer than average. Our objective was to determine the effect of age, patient's clinical acuity and complexity, and care pathways on ED-LOS and ED plus observation unit (EDOU) LOS (EDOU-LOS).
Methods This was a prospective, multicentre, observational study including all patients attending in 2011. Age groups were: I, <50; II, ≥50–64; III, ≥65–74; IV, ≥75–84; V, ≥85 years. Univariate and multivariate analyses were performed.
Results Of 125 478 attendances, 20 845(16.6%) were of patients aged ≥65 years. Multivariate analysis found significant predictors for ED-LOS (C-statistics 0.79, p<0.0000001) to be: arrival mode (ambulance, OR 1.13 (95% CI 1.08 to 1.18)); acuity level (level 4, OR 1.24 (95% CI 1.21 to 1.28); level 1–3, OR 1.54 (95% CI 1.5 to 1.59)); haematological examinations (OR 3.34 (95% CI 3.15 to 3.56)); intravenous treatment (OR 1.58 (95% CI 1.47 to 1.69)); monitoring of vital signs (OR 1.89 (95% CI 1.69 to 2.10)); x-ray examinations (OR 1.53 (95% CI 1.45 to 1.61)); CT/MRI/ultrasound (OR 2.60 (95% CI 2.39 to 2.82)); and specialist advice (OR 1.39 (95% CI 1.30 to 1.48)). For EDOU-LOS (C-statistics 0.81, p<0.0000001) we found: age group (II, OR 1.19 (95% CI 1.16 to 1.22); III, OR 1.42 (95% CI 1.38 to 1.46); IV, OR 1.69 (95% CI 1.65 to 1.74); V, 2.01 (95% CI 1.96 to 2.07)); acuity level (level 4, OR 1.31 (95% CI 1.27 to 1.35); level 1–3, OR 1.71 (95% CI 1.66 to 1.77)); haematological examinations (OR 7.81 (95% CI 7.23 to 8.43)); intravenous treatment (OR 1.95 (95% CI 1.8 to 2.12)); x-ray examinations (OR 1.95 (95% CI 1.85 to 2.06)); CT/MRI/ultrasound (OR 6.74 (95% CI 5.98 to 7.6)); specialist advice (OR 2.24 (95% CI 2.07 to 2.42)); admission to a medical or surgical ward (OR 0.61 (95% CI 0.54 to 0.68)); and transfer (OR 1.79 (95% CI 1.54 to 2.07)).
Conclusions Whereas ED-LOS and EDOU-LOS seem to be directly related to patients’ acuity and complexity, notably the need for diagnostic and therapeutic interventions, only EDOU-LOS was significantly associated with age and proposed care pathways. We propose that EDOU-LOS measurement should be made in EDs with an OU.
Overcrowding negatively affects quality of care, patient satisfaction and satisfaction of emergency healthcare providers.1 A prolonged length of stay (LOS) in emergency departments (EDs) (ED-LOS) is associated with lower quality of care and worse outcomes, and ED-LOS is often associated with overcrowding.2 As ED-LOS is currently accepted as a quality indicator, some authors have studied ED and hospital census variables associated with ED-LOS.3 It has been previously reported that hospital characteristics,3 ,4 patient acuity,5 ,6 laboratory examinations,7 ,8 and advanced diagnostic imaging9–11 were associated with ED-LOS. It has also been reported that older patients have longer ED-LOS3 ,10–13 and longer LOS in the observation unit (OU)14 and that increasing age accounts for 18% of the observed reduced performance in EDs measured between 1994 and 2004.14 Visits to EDs have increased, and older people make up a large proportion of these patients.14 ,15 The number of visits by older patients to EDs is likely to increase over the coming years.16 It has been reported that, compared with younger patients, older patients present with higher acuity and more comorbidities, that they are likely to require more diagnostic procedures and treatments, and that they are more often admitted to hospital, which means they remain in the ED longer.10 ,17 ,18 Geriatric assessment in ED OUs is a useful tool for reducing re-attendance and hospitalisation rates.14 Because EDs have limited time available, the OU may provide the extra time needed for geriatric assessment19 and may increase efficiency.14 ED OUs have been proposed as an effective option for the most vulnerable geriatric patients.19 ,20
Previous studies that found longer ED-LOS for older patients than for younger patients did not evaluate other possible confounding factors, notably patients’ acuity, ED resource consumption, and final disposition decision. We hypothesised that age may be an independent factor for longer ED-LOS and ED plus OU LOS (EDOU-LOS), independently of the care pathway and the use of ED resources. Our objective was to determine the precise characteristics of clinical patients, how they are cared for in emergencies, and the predictive variables for longer LOS.
Study design and setting
This was a prospective, observational study that measured ED-LOS, OU-LOS and EDOU-LOS. The study was carried out in the ED of three university hospitals: Bichat-Claude Bernard (BCB), Beaujon (BJN) and Louis Mourier (LMR). All three only admit adult patients (ie, ≥15 years old). BCB is a 1000-bed hospital located in an urban area of Paris, whereas BJN and LMR (500 and 400 beds) are located in the Parisian suburbs. All are associated with the same medical faculty (Paris 7 Diderot) and the same hospital group (Hôpitaux Universitaires Paris Nord Val de Seine). Each hospital has an EDOU, an acute geriatric unit, and medical and surgical wards (MSW).
Our 12-month study was carried out between 1 January 2011 and 31 December 2011.
Selection of participants
All visits made to BCB, BJN and LMR EDs during the study period were included.
Methods of measurement
From electronic medical records, the time of arrival and departure from the ED or OU were recorded. The following variables were studied: age, gender, triage acuity level, and final disposition decisions from ED and OU. Triage acuity was measured using a five-point scale (level 1, resuscitation; level 2, emergent; level 3, urgent; level 4, less urgent; level 5, non-urgent). To describe the care pathways, we defined the final disposition from ED and OU.
We ascertained the occurrence of the following diagnostic or therapeutic interventions during the stay in the ED: vital-sign monitoring, alarms, haematological tests, biochemical tests, toxicology tests, bacterial–parasitological tests, blood-group analyses, ECG, x-ray examinations, CT/MRI/ultrasound, blood transfusion, oral treatments, drug-aerosol inhalation, electrolytes added to intravenous lines, intravenous lines, intravenous treatment, other treatments, splints/casts or plaster, specialist advice, bladder catheter or gastric tube. For each patient, we then calculated the number of tests performed and divided the interventions into three groups: a nurse score, a medical score, and a total score summed from the nurse and medical scores. The medical score included x-ray examinations, CT/MRI/ultrasound, blood transfusions, splints/casts or plaster, specialist advice, bladder catheter and gastric tube. All others constituted the nurse score. We assumed that interventions defined patients’ ED resource consumption, and that acuity level and ED resource consumption indicates patients’ clinical complexity.
Data were extracted from the computerised ED system (Urqual; McKesson, Paris, France) and analysed using Statistica 10 software.
All datasets were completely anonymous and did not contain any identifiable personal health information. The dataset is currently being used by the ED as a quality and performance measure as part of an on-going emergency activity and performance evaluation approved by the Assistance Publique-Hôpitaux de Paris committees on research and information.
LOS was the main outcome measured. We defined it as follows.
ED-LOS: number of minutes between the time the patient was identified in the ED and the time the patient left the ED.
OU-LOS: for patients admitted to the OU, the number of minutes between when the patient left the ED and the time the patient left the OU.
EDOU-LOS: for all patients, the number of minutes between the time the patient was identified in the ED and the time the patient left the ED (from ED if not admitted or from OU if admitted to OU).
We divided the data into five age groups: ≥15–49, ≥50–64, ≥65–74, ≥75–84 and ≥85 years old. Categories for continuous LOS measurements and treatment groups were defined by median values. Streaming was implemented after evaluation by the triage nurse, dividing patients into three process streams: level 1–3, level 4 and level 5.
Analysis of primary data
Statistical significances between observed differences in categorical outcomes were assessed using the χ2 or Fisher exact test, as appropriate. ORs and 95% CI were calculated as indicated. For interval data, the non-parametric Kruskal–Wallis test or the Mann–Whitney U test was used. We used the Cochran–Armitage test for trend21 to order the effects of age on utilisation of ED resources and to asses if it varied linearly as a function of age group.
All statistical tests were two-tailed, and p<0.01 was considered significant.
We performed multivariate logistic regression to investigate the predictive value of the variables in combination, and their independent effects to test for differences between ED-LOS and EDOU-LOS groups. The presence of confounding factors was assessed empirically by entering potential covariates into a logistic regression model, one at a time, and by comparing the adjusted and unadjusted ORs. Logistic regression models included covariates that altered unadjusted ORs by at least 10%. Our analyses included forward stepwise selection, forced entry of all potential risk factors and backward elimination. Our final model included only those variables that were statistically significant in one or more of these approaches, to reduce the chance of identifying spurious associations.22 Model fit was determined using Hosmer–Lemeshow statistics for homoscedasticity and the C-statistic to reflect overall fit.23
Data for 125 478 attendances over 365 days were analysed (BCB, 71 135 patients (56.7%); LMR, 25 628 patients (22.9%); BJN, 28 715 patients (20.4%)). As presented in table 1, 20 845patients (16.6%) were aged ≥65 years, 13 211patients (10.5%) were aged ≥75 years, and 6081 patients (4.9%) were aged ≥85 years.
Table 1 shows the principal characteristics of the study population regarding continuous variables as a function of age group and ED and OU final disposition decisions. Table 2 shows comparisons between baseline categorical variables and age categories.
A total of 29 820 patients (23.8%) were admitted from ED to OU, to MSW or transferred. Final overall admission rate (patients admitted to MSW plus transferred patients) directly from the ED or after an OU stay was 15.4% (19 378patients). A total of 163 patients died in the ED or OU (0.8%).
Table 3 shows the diagnostic and therapeutic interventions used in the ED as a function of age group. Cochran–Armitage tests found a significant increasing linear trend from younger to older groups for the following variables (p<0.01): arrival by ambulance; acuity level 1–3; ED disposition decision, admission to OU; ED disposition decision, admission to MSW; OU disposition decision, transfer; ED+OU disposition decision, transfer; ED+OU disposition decision, admission to MSW; monitoring of vital signs; haematological examinations; biochemical tests; bacteriological–parasitological tests; blood-group analyses; ECG; x-ray examinations; CT/MRI/ultrasound; intravenous lines; intravenous treatments; other treatments; specialist advice; bladder catheter or gastric tube. We found a significant decreasing linear trend from younger to older groups for the following variables (p<0.01): arrival by own means; acuity level 5; acuity level 4; ED final disposition decision, not to admit; OU final disposition decision, not to admit; ED+OU final disposition, not to admit.
Predictive factors for longer ED-LOS and EDOU-LOS
Table 4 presents the results of unadjusted analyses for the association between the ED-LOS and EDOU-LOS categorical variables (median value) and the patients’ principal characteristics and care pathways.
Table 5 presents the association between the ED-LOS and EDOU-LOS categorical variables versus each of the diagnostic and therapeutic interventions used in the ED.
Adjusted logistic regression predictive factors associated with longer ED-LOS and EDOU-LOS
Table 6 presents the results of the final model of multiple regression analysis for ED-LOS (C-statistics 0.791 (95% CI 0.788 to 0.794)) and EDOU-LOS (C-statistics 0.812 (95% CI 0.81 to 0.814)). ED-LOS remained significantly associated with arrival mode, acuity level, vital-sign monitoring, haematological tests, intravenous treatment, x-ray examinations, CT/MRI/ultrasound and specialist advice. EDOU-LOS remained significantly associated with age group, acuity level, haematological examinations, intravenous treatment, x-ray examinations, CT/MRI/ultrasound, specialist advice, admission to MSW and transfer.
We found that patients aged ≥65 years and patients aged >75 years comprised 16.7% and 10.5%, respectively, of the ED population. Patients aged ≥75 years represent 8–9.5% of the general population in the Paris metropolitan area.24 Our results also indicate the greater clinical acuity and complexity of older patients. It can be concluded that different clinical features are linked with older patients: mode of arrival, triage acuity level, and diagnostic and therapeutic emergency treatments. We found a significantly increasing trend from younger to older patient groups for those who arrived by ambulance. Our results are concordant with previously reported data, confirming that older adults were two to four times more likely to arrive at the ED in an ambulance than younger adults.18 Similarly, we found that the percentage of triage levels 1–3, which indicate more acute conditions, was greater for older patients, whereas younger patients had more trivial acuity conditions. We also found that the number of diagnostic and therapeutic interventions increased progressively from younger to older patients. It has been previously reported that older patients have more acute conditions17 and require more emergency resources and more biological and x-ray tests than younger patients.11 ,18
Our results suggest that age affects the clinical care pathway. We found that, in our hospitals trust, the OU was a referral service for seniors, with a significantly increasing trend from younger to older groups, even in patients aged over 65 years. Our data are consistent with other recent reports on these units.25 A comprehensive geriatric assessment takes at least 30–45 min and thus cannot be given to every older person in the ED, although should, optimally, be given to identify frailty in aged patients.14 ,19 ,20 ,25
We found in an unadjusted analysis that ED-LOS, OU-LOS and EDOU-LOS differed significantly with age group, increasing progressively from younger to older patients. We also found that there was an association between the patient's acuity level and clinical complexity and ED-LOS and EDOU-LOS. In addition, the clinical care pathway had an impact on ED-LOS and EDOU-LOS, which were significantly shorter for non-admitted than admitted patients, and longer for transferred patients.
Thus, we found that, individually, age, final disposition decision and ED resource consumption were associated with a longer ED-LOS. We also found that age was associated with ED resource consumption and with some care pathways. Therefore, our study suggests that some relationships may exist between these time periods and patient age, clinical acuity and complexity, and some clinical care pathways. Previous studies on ED-LOS for older patients did not evaluate other patient characteristics that may have affected ED-LOS.
By using multiple logistic regression analysis to identify the independent predictive value of age, clinical acuity and clinical complexity on ED-LOS and EDOU-LOS, we found that longer ED-LOS was strongly associated with arrival mode, acuity level, haematological tests, intravenous treatment, vital-sign monitoring, x-ray examinations, CT/MRI/ultrasound and specialist advice. In addition, we found a strong predictive value for longer EDOU-LOS for age group, acuity level, haematological tests, intravenous treatment, x-ray examinations, CT/MRI/ultrasound, specialist advice, and admission to MSW or transfer. Our results indicate that ED-LOS and EDOU-LOS remain closely related to patient acuity and to diagnostic and therapeutic complexity, but other variables, such as age, the need for specific care pathways and decisions regarding transfer and MSW admission, are only powerful predictors for EDOU-LOS. It has been widely reported that older patients spend a longer time in EDs.11 ,14 ,17 ,18 ,25 However, even though we found in unadjusted analyses that ED-LOS was significantly longer in older groups than in younger ones, using multivariate analysis, we found that age was not a significant predictor of longer ED-LOS.
This study shows that admission to the OU was the preferred pathway for older patients and those with more complex clinical conditions. In France, the admission of acute older patients to geriatric units is often based on particular characteristics of older patients, which require geriatric assessment before admission, and so admission to the OU is the likely consequence. Similarly, we found that a decision to transfer was a significant predictor of longer EDOU-LOS. Patients are transferred from the ED or OU mostly because of a lack of facilities or beds in the hospital. In addition, a significant number of geriatric beds are located outside of our hospitals, to which the ED may be able to transfer some of the older patients. The decision to transfer also requires a complicated and time-consuming process. In contrast, our study shows that the decision to admit to a hospital MSW is usually done as a precautionary measure and is thus associated with a shorter EDOU-LOS.
The main variables associated with ED-LOS and EDOU-LOS were found to be blood testing (haematological examinations) and advanced imaging (CT/MRI/ultrasound scans), which reflect consumption of resources. It has been reported that completion of these tests is associated with a longer ED-LOS.7–11 The complex process of decision making, ordering and actualisation of some diagnostic and therapeutic interventions may explain the increase in some time intervals.
Our study has some limitations. First, it was conducted in three university EDs, one urban and the other two suburban, located in the Paris metropolitan area. Second, we did not evaluate other potential indicators of quality-of-care or ED performance, and focused only on ED-LOS and EDOU-LOS. Third, we did not evaluate other parameters of ED, such as overcrowding or the impact of the number of older patients on overcrowding. Fourth, we did not assess the reasons for the consultation or emergency diagnosis made. Fifth, we did not assess some hospital census variables previously described as affecting ED-LOS. Consequently, these results may not be generalisable to other settings. Finally, the cultural characteristics of the ED providers, notably the use of OUs as the preferred places for final decisions to be made about the care of older patients in the ED, may have had an impact on analysis results.
Our study shows that ED-LOS and EDOU-LOS are not associated with the same patient characteristics and clinical care pathways. ED-LOS is principally associated with clinical acuity and the need for diagnosis and therapeutic interventions, whereas EDOU-LOS is associated with the same features but also with age and proposed care pathways. There are several possible explanations: older patients are more complex and the clinical decision-making process can be longer than in younger patients; gaining access to advanced imaging can take longer for older patients; and admission to OUs, even if necessary to optimise evaluation and reach the best final disposition decision, can lengthen the overall time spent in the ED.
This is the first study of its kind and provides new insights into the relative effect of age on ED-LOS and EDOU-LOS, and provides more information about EDOU-LOS by defining this new time period. We believe that, in EDs with an OU, EDOU-LOS should be measured to complete interpretation of ED-LOS. Our results suggest that new strategies to reduce ED-LOS and EDOU-LOS should include providing suitable advice about blood testing and advanced imaging.
Contributors All authors planned, conducted and reported the work described in the article. EC is responsible for the overall content and is the guarantor.
Competing interests None.
Ethics approval Assistance Publique Hôpitaux de Paris.
Provenance and peer review Not commissioned; externally peer reviewed.