Article Text
PDF
Abstract
Background The goal of this study was to determine if ED surge and end-of-shift assessment of patients affect the extent of diagnostic tests, therapeutic interventions and accuracy of diagnosis prior to referral to internal medicine.
Methods This study was a health records review of consecutive patients referred to the internal medicine service with an ED diagnosis of heart failure, chronic obstructive pulmonary disease (COPD) or sepsis starting 1 December 2013 until 100 cases for each condition had been obtained. We developed a scoring system in consultation with emergency and internal medicine physicians to uniformly assess the completeness of treatments and investigations performed. These scores, expressed as percentage of possible points, were compared at high and low surge levels and at middle and end of shift at time of patient referral. End of shift was defined as 7:30–8:30, 15:30–16:30 and 23:30–00:30 as our shift changes occur at 8:00, 16:00 and 24:00. Rate of admission, diversion to other services and diagnosis disagreements were also assessed.
Results We included 308 patients (101 heart failure, 101 COPD, 106 sepsis) with a mean age of 74.7. Comparing middle of shift to end of shift, the mean scores were 91.9% versus 91.8% (difference 0.1% (95% CI −2.4 to 3.0)) for investigations and 73.0% versus 70.4% (difference 2.6% (95% CI −1.8 to 7.4)) for treatments. Comparing low to high surge times, the mean scores were 92.1% versus 91.7% (difference 0.4% (95% CI −1.2 to 2.4)) for investigations and 71.4% versus 73.6% (difference −2.2% (95% CI −5.6 to 1.3)) for treatments. We found low rates of diversion to alternate services (8.9% heart failure, 0% COPD, 6.6% sepsis) and low rates of diagnosis disagreement (4.0% heart failure, 10.9% COPD, 8.5% sepsis).
Conclusions We found no evidence that surge levels and end of shift impact the extent of investigations and treatments provided to patients diagnosed in the ED with heart failure, COPD or sepsis and referred to internal medicine.
- emergency department
- crowding
- heart failure
- COPD
- sepsis
Statistics from Altmetric.com
Key messages
What is already known on this subject
ED crowding has significant impact on patient outcomes and leads to delayed care. The reasons for this are not well defined. One possibility is that emergency physicians alter their practice in situations of high surge or crowding.
What this study adds
In this retrospective study of patients conducted by both emergency medicine and internal medicine investigators, we found no evidence to suggest patients referred for admission receive less extensive investigations or treatments at times of high surge. We also found no difference in the extent of investigations and treatments provided to patient referred at the middle versus end of shift.
Introduction
EDs are struggling with increasing numbers of patients, limited staffing and space as well as increasing volume of admitted patients ‘boarded’ in the ED. A number of studies have been published in the past decade that looked at ED crowding and its impact on patient outcomes. Generally, crowded EDs have been linked with poor outcomes, increased mortality,1 delayed analgesia,2–7delayed antibiotic administration8–10 and delayed treatment in myocardial infarcts.11 12 It is unclear what specific factors around ED crowding actually contribute to its association with these outcomes, although proposed theories suggest potentially suboptimal triage, medical workup and treatment due to patient flow pressures. To our knowledge, there has been no published literature evaluating the actual impact of ED crowding on the appropriateness of investigation and treatment for patients referred to in-patient services. Similarly, we have not found any studies analysing the potential relationship between end of shift and ultimate treatment and referral practices of this patient population.
While there is growing consensus on what ED crowding is conceptually, there is ongoing controversy on how to accurately measure it within the context of unique environments of individual institutions.13 The two most frequently cited overcrowding measurement tools in the literature (the National ED Overcrowding Study (NEDOCS) and the Emergency Department Work Index (EDWIN)) are limited in that they only provide computation of a single overall score14 15 that does not allow for an easy understanding of surge. Meanwhile, national organisations like the Canadian Association of Emergency Medicine16 17 recommend adopting an Input–Throughput–Output (ITO) conceptual model in the measurement of ED surge. In previous work, we described the development and implementation of a novel ED surge protocol that allows objective measures of crowding at various components of our ED based on the ITO model18 (figure 2). Following this protocol, every 2 hours a number of patient flow variables are used to assess the overall level of surge (0, no surge; 1, minor; 2, moderate; 3, major) at the different ITO components of the department, thus providing a robust real-time crowding measurement individualised for our institution. Since developing our surge score, the ICMED (International Crowding Measure in Emergency Departments) score has been validated.19 It uses the ITO model but also relies on some metrics like the percentage of patients leaving without being seen that makes it cumbersome to be measured in real time and, like EDWIN and NEDOCS, was validated against emergency physicians’ perception of crowding. It also did not perform as well when applied in Canadian ED.20 Nevertheless, our surge model includes the majority of metrics used in the ICMED and also incorporates other factors that specifically cater to our institutional environment.
Using this surge protocol data, we aimed to look at whether physician practice changes at times of high surge and at end of shift, when we are most pressed for time and efficiency. We looked specifically at patients referred for admission to internal medicine with one of three commonly encountered conditions: heart failure, chronic obstructive pulmonary disease (COPD) and sepsis. These were chosen because they are common presentations for which there are expected interventions and investigations. Since these patients were referred with the expectation that they would need to be admitted, we suspected there may be less extensive investigations and treatments provided prior to referral at times of high surge since emergency physicians may choose to focus their time on undifferentiated patients instead.
In order to assess completeness of investigations and treatments, we consulted with some of our internists and senior emergency physicians to develop a scoring system that reflected local expectations of what should be performed by the emergency physician prior to referral. This score is by definition locally derived and not externally validated; however, the reality of emergency medicine is that the practice varies by region and by centre, and expectations will differ between large academic centres and rural departments, as well as by region. We were trying to assess if by our centre’s standards emergency physicians performed the expected investigations and treatments as thoroughly when under time constraints at times of surge and at end of shift.
Methods
Study design, setting and subjects
This was a health records review of a consecutive sample of patients presenting to The Ottawa Hospital General and Civic campus that received an ED diagnosis of heart failure, COPD exacerbation or sepsis, were referred to the internal medicine service for admission and had surge data recorded. In our healthcare system, patients who are felt by the Emergency Physician (EP) to benefit from admission are referred to the admitting services (in this case internal medicine) who then do a consultation in the ED and decide whether or not to admit the patient. Data collection was started on 1 December 2013 until 50 cases were obtained from each campus and for each condition. The study was conducted at two EDs: the General and Civic campuses of The Ottawa Hospital, a large tertiary-care hospital. Patients were included if they were referred to internal medicine from the ED with a diagnosis of heart failure, COPD exacerbation or sepsis. Patients were excluded if there were no recorded surge data at the time of visit, if they were less than 18 years old, refused treatments or were palliated in the ED. The Ottawa Hospital Research Ethics Board approved the study and consent was not required.
Study protocol
We developed separate evaluation tools for investigations and treatments ordered by the ED physicians in consultation with senior emergency medicine and internal medicine physicians. For each condition evaluated (heart failure, COPD, sepsis), key items expected to be performed by the ED physician prior to referral to internal medicine were identified. We surveyed four senior ED physicians and four internal medicine physicians to attribute point values to each of these items, with the more important ones being given larger point values. This allowed us to develop a scoring sheet for investigations and treatments for each of the three conditions to enable comparison of scores at low versus high surge levels as well as at middle versus end of shift (figure 1A–C).
(A) Scoring sheet for heart failure. (B) Scoring sheet for COPD. (C) Scoring sheet for sepsis. ABG, Arterial Blood Gas; CBC, Complete Blood Count; CK, Creatinine Kinase; SBP, Systolic Blood Pressure; TNI, Troponin I; VBG, Venous Blood Gas.
Surge data were obtained from the ED records. Surge levels are measured every 2 hours by care facilitators (senior emergency nurses working as department administrators) to reflect ED input, throughput and output (figure 2). Surge is graded for each component of ITO as 3 (high), 2 (moderate), 1 (mild) if one of the criteria described in figure 2 is met. If none of the criteria are met, the surge is labelled as 0. Surge levels of 0 or 1 were considered ‘low surge’ and levels of 2 or 3 were deemed ‘high surge’. End of shift was defined as 7:30–8:30, 15:30–16:30 and 23:30–00:30 as our shift changes occur at 8:00, 16:00 and 24:00. Middle of shift referred to patients assessed or referred outside of these times. Times of patient assessment and referral were obtained from the ED record of treatment or, if not recorded, from the nursing record. Charts were evaluated in equal numbers from each of the two campuses aiming for a total of 50 charts from each campus for each condition, basing our sample size on feasibility. Cases were excluded if there were no surge data available for that day.
The Ottawa Hospital ED surge protocol. EMS, Emergency Medical Services; WTBS, Waiting to be seen.
The primary outcomes were comparison of the investigation and treatment scores at time of patient referral presented as percentage of possible score at low versus high surge at time of referral and middle versus end of shift. Secondary outcomes were admission rates, redirection to alternate services by internal medicine and disagreements in primary diagnosis between the ED record of treatment and the internal medicine consult. Patient mortality and return visit to the ED at 30 days were also recorded. Scores were determined by the primary author after review of ED physician notes, nursing notes, consultations, laboratory results and imaging department results, all from the hospital electronic record system. Approximately 10% of charts were reviewed by a second investigator (EK) and differences in scores were resolved by discussion.
Data analysis
We calculated descriptive statistics for patient characteristics for each of the three groups: heart failure, COPD and sepsis. Using SAS software, comparisons were assessed by the appropriate univariate analyses according to the type of data: for nominal data, the χ2 test with continuity correction; for ordinal variables, the Mann-Whitney U test; and, for continuous variables, the unpaired two-tailed t-test, using pooled or separate variance estimates as appropriate.
Results
A total of 178 charts were assessed for patients with heart failure, 167 for COPD and 278 for sepsis (see online supplementary material). Charts were excluded mainly because of lacking surge data at the General campus for December 2013, March and April 2014. Surge data were measured during those times on paper charts; however, the boxes containing these specific months were misplaced prior to the information becoming digitalised. As a result, the groups in each campus represent a slightly different time period. Other reasons for exclusion include ED diagnosis that was not actually heart failure, COPD or sepsis, and cases not referred to the internal medicine service. The mean age was 78.3 (40–100) for patients with heart failure, 72.9 (39–98) for patients with COPD and 73.1 (28–99) for patients with sepsis and the percentage of women was 48.5%, 56.4% and 47.2%, respectively (table 1).
Supplementary file 1
Patient characteristics
As shown in table 2, admission rates were very high at 93.1% for heart failure, 91.1% for COPD and 96.2% for sepsis. Diagnosis disagreement occurred rarely (4% heart failure, 11% COPD and 8% sepsis) and few patients (9% heart failure, 0% COPD, 7% sepsis) were diverted to alternate services. There were only two cases where potentially significant diagnoses were missed and patients did not suffer negative outcomes in either case. Referral at the end of shift occurred for 21.8% of patients with heart failure, 16.8% of patients with COPD and 16.1% of patients with sepsis, which suggests that ‘batching’ of referrals at the end of shift was uncommon in our sample population. In addition, approximately half of the patients were referred at high surge (43.6% heart failure, 60.4% COPD, 52.8% sepsis).
Referral outcomes and mortality
Please see online supplementary material for detailed average scores for investigations and treatments for each diagnosis at high and low surge as well as end and middle of shift. Mean scores are presented for surge levels at ED input, throughput and output at time of initial assessment and referral. There were no significant differences in scores found in either investigations or treatment for patients referred at the middle or end of shift, or patients referred at times of high versus low surge. Table 3 presents the averaged score (in percentage) at time of referral of all three conditions (heart failure, COPD and sepsis) for investigations as well as treatments at high compared with low surge. The average investigations score was 91.7% at high surge and 92.1% at low surge (difference 0.4% (95% CI −1.2 to 2.4)) and the average treatment score was 73.6% at high surge and 71.4% at low surge (difference −2.2% (95% CI −5.6 to 1.3)). Similarly in table 3, the average score (in percentage) of all three conditions is presented for both investigations and treatments at middle compared with end of shift. The average investigations score was 91.9% for middle of shift and 91.8% for end of shift referrals (difference 0.1% (95% CI −2.4 to 3.0)) and the average treatment score was 73.0% for middle of shift and 70.4% for end of shift referrals (difference 2.6% (95%CI −1.8 to 7.4)).
Averaged scores at time of referral for all three conditions at high versus low surge and at middle versus end of shift
Discussion
In this study, we found that there was no significant difference between the extent of investigations and treatments performed in the ED and the end versus middle of shift, or surge level. Intuitively, we would have expected to see lower scores at times of high surge and end of shift. It seems logical that emergency physicians would tend to carry out less extensive testing or treatments prior to referral when pressured with flow, bed blocking and time constraints. We found no evidence of this, which may be in part because our experienced nurses carry out more investigations independently when they notice the department is overwhelmed. It may also be simply that our study looked at conditions that have a well-standardised approach to investigations and treatments and rely mainly on cognitive ‘type one processing’21 for diagnosis and management, which is based on pattern recognition and is therefore considerably more efficient and less affected by time constraints. It would be interesting to see if obscure or rare diagnoses would be managed differently at end of shift or high surge. It is also important to note that this study was carried out in two large academic referral centres and the results may not extend to smaller centres where surge may be less frequent and less severe but also much more difficult to mitigate due to limited resources.
We found that regardless of surge levels, there were very low rates of diversion of patients and very few disagreements in diagnosis.
Limitations
This was a health records observational study in two large academic EDs. Our point scoring system allowed us to use a standardised tool to compare a large number of charts to assess for trends, but it does not allow for individual variations in cases and does not assess the appropriateness of care. While not every intervention would have been appropriate for every case, the scoring scheme did represent what would usually be expected for the majority of patients diagnosed with these conditions. There are no gold standards established for evaluating our practice, and it is likely that the expectations would vary between institutions. We designed our scoring tool to reflect the accepted practice pattern in our hospital.
Our sample size was small, with slightly more than 100 patients recruited for each of the three conditions assessed. Furthermore, once the scoring scheme was developed, the charts were abstracted only by emergency physicians.
The greatest limitation in the study came from the lack of surge data at one of the campuses for several months during our study period due to the loss of paper surge record boxes prior to electronic entry of the data. Because of this, the two campus populations represent patients presenting at different times of the year and for the affected campus the patient sample was not continuous. Regardless, we do not see this as a threat to the validity of our findings since the expected practice does not vary significantly within a few months.
Conclusions
We were unable to show evidence that surge levels and end of shift impact the extent of investigations and treatments provided to patients diagnosed in the ED with heart failure, COPD or sepsis and referred to internal medicine. Admission rates for the patients referred were above 90%, and there were very few diagnosis disagreements or diversion to alternate service by internal medicine. We believe this supports the emergency physician’s ability to adapt to time and surge constraints, particularly in the context of well-defined and commonly encountered conditions.
References
Footnotes
Contributors EK and VC conceived the idea. IGS designed the study and oversaw statistical analysis and management of data. LB assisted in the design of the study. VC performed data collection and analysis. All authors supervised the conduct of the trial and data collection, drafted the manuscript and/or contributed to its revision, and approved the final version. IGS had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Funding We acknowledge funding from the Department of Emergency Medicine.
Competing interests IGS holds a Distinguished Professorship and University Health Research Chair from the University of Ottawa.
Patient consent Not required.
Ethics approval Ottawa Health Sciences Network Ethics Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.