Implementing wait-time reductions under Ontario government benchmarks (Pay-for-Results): a Cluster Randomized Trial of the Effect of a Physician-Nurse Supplementary Triage Assistance team (MDRNSTAT) on emergency department patient wait times

Emergency department (ED) crowding occurs when the demand for emergency services exceeds the ability to provide care in a reasonable amount of time [1]. The main cause of crowding is access block [1–3], where no ward bed is available for patients assessed by the ED requiring admission. Consequently, patients requesting ED care accumulate in the waiting room. In Ontario, discharged patients with longer ED length-of-stay (EDLOS) had higher subsequent hospital admission rates and short-term mortality [4]. Crowding increases both mortality and costs [4–9]. This problem is international [2, 5, 10, 11] in scope and impacts [1, 2, 5, 10–13] governments, [10, 14] insurers [7], hospitals, health care workers [11], and patients [4].

In Ontario, there are 163 emergency departments with 5.25 million visits per year [14]. In November 2007, the Ontario Ministry of Health and Long-Term Care declared that reducing EDLOS was a government priority. It established provincial targets, public reporting, and a number of policy initiatives including a Pay-for Results program with financial incentives to reduce EDLOS for targeted hospitals. [15] Our hospital, Sunnybrook Health Sciences Center (Sunnybrook), was one of the targeted hospitals.

The goal of Ontario’s Pay for Results program was for individual hospitals to improve their EDLOS targets by 10% from the previous year (2008) until > =90% of all ED patients reach the set targets [15]. For Sunnybrook, the 2009–10 Pay for Results target was to have > =38% of admitted and > =71% of discharged high acuity patients (Canadian Triage Acuity Scale [CTAS] 1–3) with EDLOS < =8 hrs. For discharged low acuity patients (CTAS 4–5), the Pay for Results target was to have > = 75% of patients with an EDLOS of < =4 hrs. The 2009–10 90^th percentile target for physician initial assessment time was set at 5:24, with the future goal of reaching the ideal target of 3:48 [15].

Despite widespread concerns, there are few rigorous evaluations of interventions to reduce EDLOS. The usual triage process has a nurse taking a brief assessment of non-critical patients arriving to the ED and assigning a triage score representing the urgency for emergency physician assessment. Patients who are critically ill (cardiac arrest) bypass triage, and enter a resuscitation room for immediate management. When ED stretchers are full, the rest of the patients, even those with high acuity scores, remain in the waiting room for lengthy periods of time before seeing an emergency physician.

Triage by a nurse alone is intended to prioritize care for the most severely ill patients by assigning acuity scores (CTAS). It is known that correlation of acuity scoring between a triage nurse and physician is poor [16]; therefore, physicians do not triage. The MDRNSTAT (Physician (MD)-Nurse (RN) Supplementary Team At Triage) does not triage. Its role is to interact and start interventions on patients without an ED care space after interacting with the triage nurse. At least 10 studies testing the impact of adding a physician to the existing nurse at triage have shown some benefit [17–26] but only one of these was a randomized trial [20]. We modified an Australian model of physician-triage and report the evaluation of a MDRNSTAT.

The primary objective of this study is to examine the impact and limitations of adding 6.5 hours of MDRNSTAT on EDLOS among non-consulted, discharged patients seen by the emergency physician.

This was a pragmatic study. Because the only changes were the added MDRNSTAT resource, its benefits could be applied to other hospitals similar to Sunnybrook.

This study had a number of limitations. It is a single-center study, and one of two randomized control trials [20]. To increase its generalizability from academic tertiary level centers, a multi-center trial is recommended. Coverage on weekdays from 8:00–14:30 meant that patients arriving on weekends and evenings could not benefit. Focusing MDRNSTAT shifts during times of crowding or adaptively engaging MDRNSTAT when bed blockage occurs may benefit from future study.

Because physicians and nurses entered time data, measurement bias was possible. This study was not blinded, but all health care workers were under the same Pay for Performance program. Additionally, the health care workers who volunteered to be MDRNSTAT also worked in the department as control or part of the intervention on different days. Consequently, all physicians would have the same pressure to discharge patients within targets. However, wait times could have been shortened because the MDRNSTAT wanted to discharge patients quicker, or faster physicians/nurses volunteered to be the MDRNSTAT. The differences between the MDRNSTAT and the control could be attributed to patient selection; however, time gains were observed with the combined group as well.

It could be argued that the benefits of this trial arose from the extra staff, and would have occurred had they been deployed in the ED rather than at triage. However, Ontario hospitals run lean with fewer ward beds per capita than many US settings [12, 31] creating access block. Consequently, ED admissions can occupy ED stretchers for a median 10 hours creating a bottleneck of physical space for assessment. Under these conditions, added staff capacity would be optimized at triage since patients do not have to wait for an available ED stretcher in order to see a physician. To further explore this issue, future studies could compare a control physician-nurse team in the department to the MDRNSTAT.

We only reviewed Sunnybrook hospital’s database for harm. Patients returning to a different hospital, or dying outside of the hospital would have been missed. However, distribution should have been similar in both arms given randomization. This suggests no detectable harm, but it is important to bear in mind that this study was not powered to detect mortality differences.

For patients managed solely by the ED, MDRNSTAT significantly reduced EDLOS and physician initial assessment time. The MDRNSTAT initiated actions (laboratory, diagnostic imaging, and consultation request times) that shortened emergency physician decisions and patient delays. Because waiting times to disposition were reduced, left-without-being-seen rates declined, resulting in greater patient throughput.

For patients whose EDLOS was dependent upon factors external to the ED (i.e. consultation), the gains with shortened physician initial assessment and request times were diluted by the time required to complete these external requests by non-ED hospital staff.

Serious illnesses seen in ED’s, such as ischemia, have time-dependant treatments [32]. Skilled, timely care [10, 33] is crucial; hence the focus of performance indicators on waiting times [34]. However, rules focusing solely on wait-times, without thoughtful improvement in care processes may compromise quality of care [35] and increase costs [36]. The 4-hrs to disposition rule in the UK was estimated to cost an extra £820 million without consistent health improvement [36]. None of 750 patients assessed in this prospective, randomized-controlled trial experienced harm according to our measures; suggesting that this method to speed critical decisions does not compromise quality of care.

Emergency departments have successfully implemented fast-track areas. [37, 38]. However, the MDRNSTAT can see mixed acuity patients, whereas the fast-track population will be mainly ambulatory, lower acuity patients. While fast-track might be useful in community hospitals, Sunnybrook has few (27%) CTAS 4–5 patients compared to the provincial average (48.2%) [39].

Some have argued that it is inappropriate to have the “the ED compensate(s) for deficiencies in the larger organization by subordinating its own needs and priorities” [40]. The ED is squeezed between rising demand for emergency access to beds and increasing block as hospitals beds are closed or occupied by more lucrative, elective patients. MDRNSTAT cannot address systemic emergency department [41] or health economic problems [42]. Thus, we support the conclusion of multiple studies that emphasize the need for “throughput-output” solutions, such as hospital bed availability [11, 43, 44], long-term care, diagnostic imaging, and associated funding. However, the MDRNSTAT may provide an additional solution to those requiring government co-operation or hospital resources [8, 45–47].

The MDRNSTAT was effective in decreasing EDLOS, physician initial assessment time and left-without-being-seen rates of discharged patients requiring ED services in an urban, academic tertiary-level hospital, without compromising the quality of patient care. Given that the risk of admission and death for ED patients is associated with increasing EDLOS [4], the time gains associated with MDRNSTAT could improve patient outcomes. We are currently analyzing data on the economic value of the MDRNSTAT program.