ACADEMIC EMERGENCY MEDICINE 2011; 18:1349–1357 © 2011 by the Society for Academic Emergency Medicine
Objectives: The objective was to examine the effectiveness of triage nurse ordering (TNO) on mitigating the effect of emergency department (ED) overcrowding.
Methods: Electronic databases (Cochrane Central Register of Controlled Trials, MEDLINE, EMBASE, CINAHL, SCOPUS, Web of Science, HealthSTAR, Dissertation Abstracts, ABI/INFORM Global), controlled trial registry websites, conference proceedings, study references, experts in the field, and correspondence with authors were used to identify potentially relevant studies. Interventional studies in which TNO was used to influence ED overcrowding metrics (length of stay [LOS] and physician initial assessment [PIA]) were included in the review. Two reviewers independently assessed study eligibility and methodologic quality. Mean differences were calculated and reported with corresponding 95% confidence intervals (CIs).
Results: From more than 14,000 potentially relevant studies, 14 were included in the systematic review. Most were single-center ED studies; the overall quality was rated as weak, due to methodologic deficiencies and variable outcome reporting. TNO was associated with a 37-minute mean reduction (95% CI = −44.10 to −30.30 minutes) in the overall ED LOS in one randomized clinical trial (RCT); a 51-minute mean reduction (95% CI = −56.3 to −45.5 minutes) was observed in non-RCTs. When applied to injured subjects with suspected fractures, TNO interventions reduced ED LOS by 20 minutes (95% CI = −37.5 to −1.9 minutes) in three RCTs and by 18 minutes (95% CI = −23.2 to −13.2) in two non-RCTs. No significant reduction in PIA was observed in two RCTs.
Conclusions: Overall, TNO appears to be an effective intervention to reduce ED LOS, especially in injury and/or suspected fracture cases. The available evidence is limited by small numbers of studies, weak methodologic quality, and incomplete reporting. Future studies should focus on a better description of the contextual factors surrounding these interventions and exploring the impact of TNO on other indicators of productivity and satisfaction with health care delivery.
Overcrowding is one of the most important problems facing emergency departments (EDs) in many developed countries.1 The approaches to reducing waiting times, improving flow, and expediting disposition decisions have been a challenge for many emergency care providers and administrators; however, the multidimensional nature of ED overcrowding has made it difficult to design interventions that cover its multiple stressors. The input, throughput, and output model outlines a variety of operational processes that represent practical targets for operations research on ED overcrowding.2,3
One of the key causes of delays within the ED involves patient flow within the ED or throughput.4 This is the period from arrival in the ED (triage or registration time) to the point where decisions are made regarding patient disposition (admission or discharge). Interventions to improve throughput have been the focus of many publications on ED overcrowding;5 triage nurse ordering (TNO) is one of the proposed interventions. While TNO has been related to enhanced patient satisfaction, improved care, and increased teamwork, little is known about the effectiveness of this intervention in improving ED time metrics, quality of care, and cost of care.6
The purpose of this study was to examine the available evidence for TNO, document contextual issues associated with this intervention, and determine its effectiveness. Contextual exploration implies the examination of differences among studies in site of implementation (e.g., urban vs. rural; high volume vs. low volume) and the TNO approaches used (single issue vs. multiple issues [medical and injury]). We believe that these are important factors required to understand the effectiveness of the TNO intervention. Effectiveness implies influence on length of stay (LOS) metrics, left without being seen, medical errors, and satisfaction.
This was a systematic literature review.
An a priori protocol for the systematic review was used to define the search strategy, set the study selection criteria, outline quality assessment and data extraction procedures, and plan the analysis of the study results. While the current review involves TNO, other reviews in this funded program of research include triage liaison physicians,7 observation units, over-capacity protocols,8 and other interventions.
Two Search Strategies Employed in This Review
Sensitive Search. A comprehensive literature search was conducted in seven biomedical electronic databases: MEDLINE, EMBASE, EBM Reviews - Cochrane Central Register of Controlled Trials, HealthSTAR, Science Citation Index Expanded, Dissertation Abstracts, and ABI/INFORM Global. Due to the lack of standardized medical indexing for ED overcrowding, a wide variety of keywords were used to identify relevant literature (see Data Supplement S1 for a complete listing, available as supporting information in the online version of this paper).5 Clinical trial registries (ClinicalTrials.gov and controlled-trials.com) and Google Scholar Web search were also explored. The search strategy was identical to the one used in the Canadian Agency for Drugs and Technologies in Health (CADTH) report entitled “Interventions to Reduce Overcrowding in ED”5 and was updated for this study. The CADTH report involved literature searches from 1966 to 2005; this study supplemented the previous report by searching for citations between October 2004 and May 2009. Both searches included non-English languages and unpublished research. The overlapping was necessary to ensure that all the new literature indexed after the CADTH report was completed would be considered for inclusion.
Specific Search. A more specific literature search was conducted in January 2011 in four biomedical electronic databases: MEDLINE, EMBASE, SCOPUS, and CINAHL. This search focused on terms such as nurse/triage nurse AND test ordering, radiography, x-rays, and test requesting in the ED.
Hand searches were performed to identify abstracts presented to the following major scientific conferences between October 2004 and May 2009: the American College of Emergency Physicians, the Australasian College for Emergency Medicine, the Canadian Association of Emergency Physicians, the College of Emergency Medicine in the United Kingdom, and the Society for Academic Emergency Medicine. In addition, the references of identified articles were manually searched. Primary authors and experts in the field were contacted to identify additional published, unpublished, or ongoing studies. The search results from the CADTH report were merged with those of the updated searches, resulting in a comprehensive search strategy that identified potentially relevant studies published from 1966 to February 2011.
Eligible studies were primary research that assessed the effects of a TNO (but not nurse practitioners or floor nurses) to mitigate overcrowding in EDs serving adult (17 years or older) or mixed (child and adult) populations. Studies with one of the following designs were included: parallel or clustered randomized controlled trials (RCTs), controlled clinical trials (CCTs), prospective or retrospective analytical cohort studies, interrupted time series, case–control studies (C-C), and before–after (B-A) designs. Studies were required to report numeric data on at least one of the following outcomes: ED LOS (time in minutes from patient ED arrival to departure), physician initial assessment (PIA; time in minutes from patient ED arrival to physician assessment), and proportion of radiographs ordered by nurses. Non–primary research (e.g., editorials, commentaries, letters to the editor, narrative reviews, technology reports, and systematic reviews), studies conducted in pediatric EDs, multiple publications, and studies comparing two levels of the same intervention were excluded.
Four reviewers (XG, BHR, MO, CVR) independently screened titles and abstracts of studies identified by the literature search. The full-text versions of articles deemed potentially relevant, as well as those that reported insufficient information to determine eligibility, were independently reviewed by two of four reviewers (CVR, LW, BHR, MO). Any disagreements were resolved by consensus. Non-English literature was translated by foreign language reviewers (SMH and DS). Studies that met all inclusion criteria underwent quality assessment and data extraction.
A standard quality-rating tool developed by the Effective Public Health Practice Project (EPHPP)9 was used to appraise the methodologic quality of the studies. This tool is based on guidelines set out by Mulrow and Oxman10 and Jadad et al.11 and has accepted validity and reliability.12 The rating tool is based on six criteria: selection bias, study design, confounders, blinding, data collection methods, and withdrawals and dropouts. Each criterion is rated as “strong,”“moderate,” or “weak,” depending on information reported in the article. Once the ratings of characteristics are totaled, each study receives an overall assessment of strong, moderate, or weak quality. Two of four assessors (BR, LW, XG, CVR) independently assessed the quality of included studies. The kappa (κ) statistic was calculated to measure the level of agreement between reviewers.13 Finally, discrepancies were resolved by consensus.
Information regarding the study design and methods (e.g., year, country, type of publication, study duration, number of participating centers), intervention characteristics (e.g., nurse experience and training received), and comparison groups and outcomes of interest were extracted using a pretested data extraction form. Finally, information was collected on study conclusions, as reported by the authors of the primary studies. Two reviewers (BHR, XG) independently conducted data extraction. When standard deviations (SDs) were not reported in the included articles, they were estimated based on the reported interquartile ranges (IQRs) using a standard formula [(IQR upper level – IQR lower level)/1.35]. Attempts were made to communicate with investigators for clarification or additional data.
Characteristics of the included studies were summarized using descriptive statistics. Evidence tables were constructed to report information on each article’s source, study design, study population, treatment groups, and outcomes. Analyses were focused on ED LOS, PIA, and proportion of radiographs ordered by triage nurses.
Outcomes from individual studies were presented as mean differences with 95% confidence intervals (CIs) for continuous variables and risk ratios (RRs) with 95% CIs for dichotomous variables. The small number of studies included in this review precluded generating pooled estimates for some outcomes. Analyses were divided by study design (RCTs vs. other designs), and subgroup analyses by injury/suspected fracture status were completed for ED LOS.
Statistical significance was set at a p value of less than 0.05. All data were entered into Review Manager (RevMan, Version 5.0. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2008).
The sensitive search resulted in the identification of 14,446 potentially relevant citations, from which 3,615 studies clearly addressed the topic of ED overcrowding. After screening study titles and abstracts, 354 full manuscripts were retrieved for further examination. The specific search resulted in the identification of 325 potentially relevant citations, from which 54 studies clearly addressed the topic of TNO. The application of the selection criteria to these potentially relevant manuscripts resulted in 14 studies included (two unique to the sensitive search, eight unique to the specific search, and four identified by both searches). Figure 1 outlines the study selection flow for the review. The complete list of references of excluded studies is available from the authors upon request.
Study Characteristics. Of the 14 included studies, 12 were journal publications,14–25 one was an abstract,26 and one was a thesis report.27 Most of them were single-center ED studies, and one was a multicenter study16 (Table 1). Three were RCTs, one was a CCT, two were retrospective cohort studies, three were prospective cohort studies, two were C-C studies, and three were B-A studies.
|First Author (Year)||Location||Sample||Intervention Period||Study Design||TNP Intervention||Comparison|
|Bliss (1971)25||United States||100||Unknown||Retrospective cohort study||Triage nurse initiated distal limb x-rays only.||ED physician initiated x-ray|
|Stiell (1993)26||Canada||1,180||5 months||B-A||Triage nurse initiated foot/ankle x-rays only.||ED physician initiated x-ray|
|Lee (1996)17||Hong Kong||1,633||3 months||Prospective cohort study||Triage nurse initiated x-rays.||ED physician initiated x-ray|
|Thurston (1996)16||United Kingdom||1,833||NR||RCT||Triage nurse initiated x-rays.||ED physician initiated x-ray|
|Parris (1997)15||Australia||175||3.25 months||CCT||Triage nurse initiated x-rays.||ED physician initiated x-ray|
|Ching (1999)24||Singapore||276||3 months||C-C||Triage nurse initiated limb/skull x-rays using a standardized form.||ED physician initiated x-ray|
|Lindley-Jones (2000)14||United Kingdom||675||2 weeks||RCT||Triage nurse initiated x-rays.||ED physician/ENP initiated x-ray|
|Winn (2001)27||United States||40||2 months||Retrospective cohort study||Triage nurse initiated diagnostic test (e.g., x-ray, blood test, urinalysis, ECG).||ED physician initiated diagnostic test|
|Cheung (2002)18||Canada||250||NR||B-A||Triage nurse initiated x-ray and blood test.||ED physician initiated x-ray and blood test|
|Australia||1,806||12 months||Prospective cohort study||Triage nurse initiated isolated distal upper or lower limb x-rays only.||ED physician initiated x-ray|
|Fan (2006)19||Canada||130||3 months||RCT||Triage nurse initiated x-rays.||ED physician initiated x-ray|
|Pedersen (2009)23||Denmark||106||NR||Prospective cohort study||Triage nurse initiated x-rays on low-energy injuries.||ED physician initiated x-ray|
|Rosmulder (2010)22||Netherlands||704||22 days||B-A||Triage nurse initiated foot/ankle x-rays only.||ED physician initiated x-ray|
|Retezar (2011)21||United States||15,188||2 years||Retrospective nested C-C study||Triage nurse requested investigations for chest pain, shortness of breath, abdominal pain, or genitourinary complaints.||ED physician initiated investigations.|
|Totals (n = 14)||Three Canada; Three United States||100–15,188||2 weeks–2 years||Variable||Triage nurse.||ED physician|
Quality. Based on the EPHPP tool,9 the quality of the 14 studies was rated as weak due to methodologic deficiencies and poor outcome reporting (Figure 2). Before the consensus process, the interrater agreement was good on the overall quality assessment (κ = 0.66).
Nurse Training. From the 14 studies, details of the nurses and their training were extracted (Table 2). Most studies designated nurses as triage staff; however, several employed senior/experienced nurses only. Training also varied; however, test ordering was initiated after brief training programs (1-hour lectures) in most cases.
|First Author (Year)||Experience of Triage Nurses||Type of Training|
|Bliss (1971)25||Experienced triage nurses||Triage nurses attended a lecture by a staff orthopedic surgeon. Signs and symptoms of orthopedic fractures were given, as well as instructions for filling out x-ray requisitions.|
|Stiell (1993)26||Not specified||Not specified|
|Lee (1996)17||Not specified||Triage nurses received a 1-hour training session on the inclusion/exclusion criteria for x-rays requesting as set out in the study protocol. This training took place 1 month before the study.|
|Thurston (1996)16||Not specified||Not specified.|
|Parris (1997)15||Experienced ED nurses||Triage nurses underwent a training program in the skills of triage.|
|Ching (1999)24||Triage nurses with >1 yr of experience.||Triage nurses attended a standardized instructional course on examination skills, inclusion and exclusion criteria conducted within the ED. They were also instructed on the criteria for x-rays requisitions as in the pre-set protocol used in the study.|
|Lindley-Jones (2000)14||Not specified||Not specified.|
|Winn (2001)27||Triage nurses, 1–12 yr of experience.||Not specified.|
|Cheung (2002)18||Not specified||Triage nurses participated in a 4-hour workshop to introduce the concept of “advanced triage.” This was followed by supervised practice sessions at the triage desk.|
|Fry (2001)20||Not specified||Triage nurses participated in an education program that included a 20-minute video developed for the study and a 1-hour lecture conducted by an ED staff and a clinical nurse consultant.|
|Fan (2006)19||Not specified||Triage nurses received a 1-hour training session on the study protocol and the use of the Ottawa Ankle Rules and 2 weeks to clarify misconceptions with the study investigators.|
|Pedersen (2009)23||Not specified||Not specified.|
|Rosmulder (2010)22||Not specified||Not specified.|
|Retezar (2011)21||Not specified||Triage nurses were instructed to use their clinical judgment with regard to x-ray ordering. They were also instructed to execute their orders only when patients could not be placed in a treatment room within a reasonable period.|
ED LOS. Most included studies reported outcome data on ED LOS.14–19,21,22,24,26,27 One RCT showed a significant reduction in ED LOS when comp0aring TNO interventions to emergency physician (EP) and emergency nurse practitioner (ENP) x-ray ordering (mean difference = −37.2 minutes; 95% CI = −44.1 to −30.3 minutes); consistent results were observed when pooling the results from three non-RCTs (mean difference = −51 minutes; 95% CI = −56.3 to −45.5 minutes; Figure 3). For injured patients in whom a fracture was suspected, three RCTs indicated that TNO interventions produced a significant reduction in ED LOS (pooled mean difference = −19.7 minutes; 95% CI = −37.5 to −1.9 minutes); however, this result is tempered by high heterogeneity (I2 = 92%). A lower yet statistically significant difference (pooled mean difference = −18.2 minutes; 95% CI = −23.2 to −13.2 minutes) was observed in five non-RCTs. In contrast, a nonsignificant improvement was observed in two RCTs that included patients without suspicion of injury or fracture (pooled mean difference = 0.93 minutes; 95% CI = −5.44 to 7.31 minutes). Details for non-RCT designs are illustrated in Figure 4.
PIA. Three studies reported outcome data on PIA.14,16,17 Two RCTs indicated a nonsignificant reduction in PIA when comparing TNO interventions to EP/ENP x-ray ordering (pooled mean difference = −3.00 minutes; 95% CI = −6.99 to 0.99 minutes); one prospective cohort study showed a 10 min reduction on PIA (Figure 5).
Proportion of Radiographs Ordered. Two studies reported data on the proportion of radiographs ordered by triage nurses compared to EPs.14,16 Pooled analysis of two RCTs indicated that the risk of ordering radiographs was the same when comparing TNO interventions to EP/ENP x-ray ordering (RR = 0.98; 95% CI = 0.83 to 1.15); the risk of obtaining positive results was also the same when comparing these two groups (RR = 1.03; 95% CI = 0.85 to 1.23).
This systematic review summarizes the best available evidence on the operational issues and effectiveness of TNO in reducing ED metrics such as ED LOS, PIA, and the proportion of radiographs ordered by triage nurses. To our knowledge, this is the first review of this topic with a comprehensive search of related articles, detailed data collection, and reproducible methods. Overall, the review demonstrated that the TNO intervention was associated with a 37-minute reduction in the overall ED LOS, as reported in one of the included RCTs.14 A higher and statistically significant difference (51-minute reduction) was observed when analyzing other study designs.18,21,27 For TNO interventions applied to injured subjects suspected of having a fracture, ED LOS was reduced by 19 minutes in three RCTs;14,16,19 a similar, statistically significant difference (18-minute reduction) was observed in five non-RCTs.15,17,22,24,26 No statistically significant differences were identified when TNO interventions were applied to subjects without injury or a suspected fracture in the highest quality study designs.14–16 Only two RCTs reported data on PIA, and while a reduction was observed, it failed to reach statistical significance;14,16 poor outcome reporting from other study designs precluded PIA data pooling.17 Finally, our results suggested that TNO interventions have no significant effect on x-ray ordering or results, from both the clinical and the statistical perspectives.
Differences were observed regarding the nature of the TNO interventions. For example, studies differed somewhat with respect to study populations, and that may have influenced the fact that some included radiograph ordering only,14–17 while other studies included additional diagnostic test requesting such as blood tests, urinalysis, and electrocardiogram.27 The experience level of triage nurses in ordering radiologic examinations varied among studies, from having no previous training to receiving special training as part of the intervention or having extensive experience ordering radiographs. Recently, ordering agreement for diagnostic tests between triage nurses and physicians has been summarized, and many concerns have been put to rest.28 While the research conducted on this specific aspect of the TNO intervention is not methodologically strong, good training and the use of accepted protocols seem to improve health-related outcomes associated with some of the disadvantages identified by previous studies (e.g., nurses requesting more or inappropriate radiographs and missing important initial findings).16,29 Although few studies reported on the time and costs of implementing this strategy, most training appeared to be limited. Regardless of the study design, the studies included in this review showed an important reduction of the overall ED LOS; in the future, a more comprehensive research design addressing cost-effectiveness and quality of care issues could identify nursing investigation protocols applicable to similar ED settings.
In most of the studies, the comparison group consisted of EPs; in one study the comparison group included EPs or ENPs.14 The level of training and experience of the comparison group was not reported in detail in most of the included studies. Variations in this training and experience (e.g., junior vs. senior physicians) and in the process of attending patients (e.g., orders approved, or not, by consulting physicians) could explain some of the results observed in our outcomes of interest. Finally, the description of other important factors surrounding these interventions, such as the previous existence of hospital TNO protocols for specific cases (e.g., patients with limb injuries), the availability of 24-hour radiology service to ED patients, and the engagement of multidisciplinary groups and ED champions during the early stages of the research process, was not consistently reported across the studies. A careful description of such factors (facilitators vs. potential barriers) may help to decide whether research results can be translated into practice. Because research has the potential to improve our understanding of the implementation of complex health care interventions such as TNO, future research should focus on these reporting issues.30 Alternatively, well-designed implementation research may further inform those who wish to employ this approach.
It is rare to find experimental designs in operations research conducted in ED settings; however, from the 14 studies identified by our search, three were RCTs and one was a CCT. While results from observational studies (e.g., B-A studies and analytical cohort studies) should not be dismissed, they need to be interpreted with caution, because they are more vulnerable to bias than higher-quality designs.31 It is imperative that operations researchers use more robust designs and methods and improve outcome reporting. Moreover, future research needs to include other outcomes such as patient and staff satisfaction, patient outcomes, and costs.
Currently, there are no standard guidelines for TNO interventions. TNO interventions have been employed during the process of triage in an effort to ensure that appropriate diagnostic tests are undertaken at the earliest possible time according to the severity of individual patients presenting to the ED. The promising effect of reducing the overall LOS, and particularly in patients with suspicion of injury or fracture, suggests that TNO in the ED should be considered as an alternative to mitigate the effects of ED overcrowding. Because effectiveness is often driven by local needs and champions, in settings where injury is common TNO seems a reasonable consideration; perhaps this intervention would be less successful in tertiary care referral settings where minor injuries are less commonly encountered.19
Finally, although rarely examined in the literature, the success of TNO interventions appears to be cost-effective. For example, for moderate reductions in LOS, the implementation of a TNO intervention can be achieved using existing triage nurses and brief additional staff training. Especially if applied to the patients with suspected fracture, this strategy has the potential to improve nursing, patient, and physician satisfaction without any obvious safety concerns.
Some limitations of this review should be acknowledged. First, only a small number of studies that specifically assessed the effect of a TNO intervention on mitigating ED overcrowding were identified. We believe that scope and selection bias were limited by the comprehensive search and inclusion of unpublished and non-English language studies in this review; however, the possibility of missing unpublished and/or negative studies still remains. Due to the small number of studies, the effect of publication bias in the review could not be assessed.32 Evidence suggests that publication bias is less pervasive in the ED literature;33 however, negative trials are less likely to be published and more likely to be excluded from a review of this nature. Second, included studies had variable research methodologies and incomplete reporting, and this limits the confidence one has in the results. Third, missing data were common in these studies; however, efforts to contact the investigators made it possible to add some original data, and appropriate and widely accepted imputation techniques were used to ensure valid inference.34 Fourth, due to insufficient data, anticipated subgroup comparisons and sensitivity analyses were not always possible; however, subgroup analyses based on injury or suspected fracture were possible. Fifth, the possibility of ordering a test and having a patient leave prior to review by a physician is real in the ED setting, and this is a problem for both the patient (if the results are positive) and the physician or their department/hospital (who may be medically and legally responsible for the outcome). This outcome was not documented in the included studies. Finally, all abstracts and manuscripts were screened by at least two independent reviewers using standardized eligibility criteria in an effort to decrease the likelihood of selection bias.
The evidence regarding triage nurse ordering interventions is derived from a small number of studies with variable research methodologies and variable outcome reporting; however, triage nurse ordering appears to be an effective intervention to reduce ED length of stay, especially in injured patients suspected of having a fracture. In addition, this success can likely be achieved using triage nurses and little additional staff training. Future studies must focus on a more comprehensive description of the contextual factors surrounding these interventions and on exploring the effect of triage nurse ordering on other indicators of quality and crowding with health care delivery.
The authors thank the corresponding authors, Drs. Thurston and Lee, for their responding to our request on additional data and clarification. The authors are grateful to Donna Ciliska and Donna Fitzpatrick-Lewis for their explanation on the EPHPP quality assessment tool and to Diana Satanovsky-Feldman and Siri Margrete Holm for their assistance in translation.