Screening tool to improve patient referral to acute surgical care from accident and emergency

Abstract

Introduction

This study aimed, first, to audit the appropriateness of surgical referrals to an acute surgical unit for urgent assessment and, second, to devise a screening tool for use in the emergency department to categorise patients into those who need an urgent surgical review and those who can be seen in an ambulatory setting within the next few days.

Methods

The first phase of the study was an audit of surgical referrals between 1 and 18 February 2020 to check the appropriateness of the surgical referral. In the second phase, a tool was designed to screen patients who did not require urgent surgical review and could be seen in the ambulatory clinic. A prospective questionnaire study was conducted from 1 February to 24 March 2020 with patients who were admitted to an acute surgical ward. Based on responses to the screening tool, patients were given the outcome of whether they can be discharged and seen in an ambulatory clinic. The accuracy of the screening tool outcome was assessed and compared with actual patient discharge outcomes by the surgical team evaluating patients’ electronic medical records.

Results

In the first audit of referrals to the acute surgical ward, 206 patients were referred to the acute surgical unit and seen by the senior surgeon. Of these, 142 (68.9%) were discharged on the same day with or without follow-up in the ambulatory surgical clinic. In the prospective questionnaire phase of the study, 98 patients completed the questionnaire. The most common presentation was abdominal pain (n=60) followed by urological symptoms (n=11), symptoms of hernia complication (n=10), abscess (n=7), testicular pain (n=2) and trauma (n=2). Of the patients discharged on the same day, 50% were given ambulatory care appointments and 50% were discharged with no further follow-up. The sensitivity and specificity of the screening tool were 100% and 60.7%, respectively; the overall accuracy was 88.4%.

Conclusion

A large proportion of patients who are referred to the acute surgical unit can be deferred and seen in the ambulatory clinic. The screening tool used for acute surgical referral had reasonable sensitivity and high specificity to screen patients who can be seen in ambulatory clinics. At the same time, it identified patients who were unwell and required urgent surgical admission.

Introduction

The unprecedented demand for healthcare services in the United Kingdom (UK) means that National Health Service (NHS) hospitals treat more people than ever before. With rising admissions and growing delays in discharging patients, hospitals are recording their highest-ever occupancy levels at any given time of the year for acute beds.¹ Performance targets against national waiting time standards have not been met for several years and continue to deteriorate, leaving many people waiting longer for the care they need. This puts patients seeking urgent healthcare at higher risk of becoming unwell, breaching the commitment to quality of care made to patients in the NHS Constitution.²

Surgical departments around the country have recognised a need to streamline the number of inappropriate referrals and divert nonurgent presentations to outpatient settings. In our surgical department at Southend University Hospital, implementation of a 7-day surgical ambulatory clinic offering early specialist intervention and same-day emergency care has reduced unnecessary surgical admissions by at least 20%.³ However, despite this initiative within our department, we recognise the need to assess and evaluate the quality of referrals to the on-call surgical team from general practitioners (GPs) and the emergency department. Many primary care physicians are not aware of the surgical referral pathways available or how to utilise them to seek optimal care for their surgical patients. This has resulted in unnecessary admissions, pressures on beds and high volumes of patients for the on-call team to see, often rendering them unable to adequately prioritise urgent care to critically unwell patients. A similar trend in referrals has been noted when patients are assessed by an accident and emergency doctor.³ Patients wait for a long time to be referred to the surgical specialty and then reviewed by the senior surgeon.³

The aim of this study is to develop a screening tool that can identify patients who do not require urgent surgical review and can be seen instead in a surgical ambulatory clinic the following week. The objective of this study is to reduce unnecessary surgical admissions to hospital and prioritise beds and urgent care for clinically unwell patients with the greatest need (Figure 1). We hypothesise that a significant number of patients who meet the criteria for surgical ambulatory clinic referral are unnecessarily admitted under surgery for an urgent surgical review without indication. It is particularly important to screen patients to the acute surgical ward and to improve the appropriate referral rate due to the COVID-19 crisis in which the risk of infection is high, and patients are more reluctant to undergo hospital admission and prefer clinic review.⁴

Figure 1 .

Proposed pathway for surgical referral from accident and emergency using a referral screening tool

Methods

Clinical data were collected for patients who were referred for assessment to the surgical department of Southend University Hospital, UK. Data collection was carried out over a period of 2 months (February to March 2020) and was completed before the COVID-19 crisis and the implementation of COVID-19 guidelines by the local hospital. All adult patients over the age of 16 referred to the acute surgical unit for general surgical signs and symptoms were included in the study. Common signs and symptoms were abdominal pain, symptomatic hernia, abscess on the trunk and back, lower gastrointestinal bleeding, haematuria, testicular pain and urinary retention. Patients presenting with signs and symptoms of bowel obstruction, unable to speak English or with language barriers, under the age of 16 or who did not have the capacity to consent were not included in the study. Patients admitted to the surgical department of Southend Hospital are referred by the hospital’s own emergency department or out-of-hours GPs and nurse practitioners working in the GP streaming service based at Southend Hospital. A smaller number of patients are also referred by community GPs, district nurses, community nurses and surgical outpatient clinics. Referrals are taken by surgical navigators, on-call senior house officers and on-call surgical registrars on weekdays during normal working hours of 8am to 6pm, and by the latter two only on weekends, bank holidays and out of hours.

Once patients had been seen in the acute surgical unit and their appropriate treatment started, data for the screening tool were collected by the surgical registrar or consultant who had assessed the patient. Patients were shown a paper copy of the screening tool and verbal consent was obtained from patients for their data to be used in the study. In an emergency setting, we considered verbal consent to be more appropriate and less disturbing for patients in a distressing situation. The information required to complete the screening tool was obtained from patients’ electronic medical records (EMRs). Approval to conduct the study was obtained from the department of clinical governance and audit, Southend University Hospital.

The screening tool was designed to be used by emergency department doctors and junior doctors. The questions in the screening tool are based on three components: history of presenting complaints, observations, blood and urine test (Appendix 1). Questions included physiological parameters such as blood pressure, pulse, breathing rate, temperature and biochemical values including haemoglobin, white cell count, creatinine, C-reactive protein and amylase. Urine dipstick analysis for infection and urinary beta-human chorionic gonadotropin were also a part of this section of the questionnaire. Information about presenting complaints, medical history and clinical observations required for the screening was obtained from the patient’s surgical clerk-in notes. EMRs were used to obtain the information about patients’ urine and blood tests. Information about their waiting time in the emergency department and surgical time to see the surgeon was also recorded. Patients included in the study who were discharged without being admitted to the surgical ward were followed up for 2 weeks to assess whether they had any unplanned readmission to the hospital and that those given an appointment in an ambulatory surgical clinic attended.

The screening tool output provided two outcomes, whether the patient should be admitted or whether they could be discharged and followed up in a surgical ambulatory clinic within 1 week. The screening tool is designed in such a way that answering ‘no’ to all the questions will automatically give the outcome of discharge and follow-up in a surgical ambulatory clinic.

Descriptive statistics were used to evaluate patient characteristics. Owing to the nonparametric distribution of the patient population, median values were used to record patients’ observations and to compare between groups. The Mann–Whitney U test was used for comparison of continuous variables and chi-square test was used to compare categorical variables.

Results

In total, 98 patients completed the questionnaire and were included in the prospective phase of the study. The average age of the study population was 57.8 and 50.5% of patients were male. Patients spent, on average, 3.1h (median 2.0, interquartile range (IQR) 3) in the emergency department before they were sent to the acute surgical ward. The average wait to see the senior surgeon in the surgical ward was 4.2h (median 3.0; IQR 3.5). The median duration of symptoms was 7 days (IQR 19). The most common presentation was abdominal pain (n=60) followed by urological symptoms (n=11), symptoms of hernia complication (n=10), abscess (n=7), testicular pain (n=2) and trauma (n=2). Patients were noted to have the following biochemical and haematological profile: median haemoglobin, 134g/l (IQR 20.7); median white blood cell count, 9.88×10⁹/l (IQR 6.6); median C-reactive protein, 12.9 (IQR 44); median lactate, 1.5 (IQR 1.3; and median creatinine, 74mg/dl (32). The most common diagnoses for patients referred to acute surgical ward were medical/gynaecological causes (n=11), nonspecific abdominal pain (n=10), cholecystitis/biliary colic (n=8), abscess (n=8), hernia (n=7) and acute appendicitis (n=6) (Figure 2).

Figure 2 .

Common diagnoses for patients referred to the acute surgical unit

The patient population was divided into two groups: those who were admitted (n=70, 72.2%) and those who were discharged with or without follow-up to the ambulatory clinic within a week (n=28, 27.8%). Of the patients discharged on the same day, 50% were given ambulatory care appointments and half were discharged with no further follow-up. The median age of the patients admitted and those who were discharged was 64.0 (IQR 38) and 46.5 (IQR 47) years, respectively (p=0.08). The proportion of males in the group admitted and the group who were discharged was 55.7% (n=39) and 35.7% (n=10), respectively (p=0.10). The proportion of females in the group admitted and those discharged was 44.3% (n=31) and 64.3% (n=18), respectively (p=0.11). The median number of hours spent in the emergency department for the group of patients who were admitted to the surgical ward and those that were discharged on the same day after being assessed by the surgeon were 2.0 (IQR 2.9) and 1.7 (IQR 5.1), respectively (p=0.54). The time spent in the acute surgical ward prior to seeing a senior surgeon for the group of patients who were admitted to the surgical ward and those that were discharged on the same day after being assessed by the surgeon were 3.0h (IQR 3) and 2.5h (IQR 5), respectively (p=0.96). The duration of symptoms for the patients admitted to the ward and those discharged on the same day by the surgical team was 7 days (IQR 19) and 7 days (IQR 19), respectively (p=0.90). Comparison of the biochemical markers between the two groups is illustrated in Table 1.

Table 1 .

Comparison of haematological and biochemical markers between the two groups

	Median (interquartile range)
	Haemoglobin (g/dl)	White blood cell count (×109)	C-reactive protein	Creatinine (mg/dl)
Patients admitted to the surgical ward	132 (87–176)	10.25 (4.8–27.1)	16 (1–399)	73 (39–863)
Patients discharged on the same day by the surgical team	135 (82–170)	7.9 (3.4–19.7)	9 (1–84)	80 (46–187)
p-value	0.96	0.03	0.01	0.51

EMRs were assessed for the patients who were discharged with or without surgical ambulatory clinic review. None had unplanned readmission to the hospital in the following 1 week. These patients were diagnosed with the following conditions: nonspecific abdominal pain (n=9), biliary colic (n=4), diverticulosis/uncomplicated diverticulitis (n=3), symptomatic reducible inguinal hernia (n=3), urinary tract infection (n=3), urinary stone (n=2), urinary retention (n=1), mild acute pancreatitis (n=1), lower gastrointestinal bleeding (n=1) and ovarian cancer (n=1).

The sensitivity of the screening tool identified those patients who required urgent hospital admission after being assessed by the surgical team. The specificity of the screening tool identified those patients who did not require urgent surgical review and could have been seen in the surgical ambulatory clinic in the following week. The sensitivity and specificity of the screening tool was 100% and 60.7%, respectively. The overall accuracy of the screening tool was 88.4%.

Discussion

Just under 30% of patients who were reviewed by the senior surgical team on the acute surgical wards were discharged the same day with a follow-up in the ambulatory clinic, demonstrating that many surgical patients can be seen in an ambulatory surgical clinic. Furthermore, the median time of 2h that surgical patients waited in the emergency department and the further median of 3h spent awaiting a surgical review on the acute surgical ward illustrate the need for an alternate pathway that can triage these patients safely to ambulatory care.

The screening tool had reasonable sensitivity to identify patients who were well enough to be seen in the ambulatory surgical clinic. It did not miss any patients who were unwell and required acute surgical admission for further treatment, proving its safety and high specificity. Overall accuracy on the receiver operating characteristic curve was 0.8. This means that it can be used to screen patients who can be discharged safely from the emergency department without waiting for a surgical bed and review by the busy surgical team. Considering the high number of daily referrals from the emergency department to the surgical unit, this tool has a lot of potential to reduce the load on emergency departments and on-call surgical teams, enabling medical staff to prioritise urgent clinical care to those who truly require it. Many patients who are admitted to the acute surgical ward go on to stay for even longer than our study found, because they often remain after their initial review for a senior review before they are sent home. It has been estimated that the cost of an NHS bed around £400 per day, and this reinforces a need to selectively streamline and aptly triage referrals from primary care and the emergency department.⁵

Many protocol-driven artificial intelligence tools, symptom checkers and e-referral systems have already been employed to facilitate more efficient and accurate referring and triaging across the primary–secondary provider interface. These are often developed with the aid of technology and continuously improved with support from physicians in the relevant specialties. The artificial intelligence-powered Babylon Triage and Diagnostic System is one such example of online symptom-checking software that was designed to aid primary care referrals. In one study, the Babylon Triage and Diagnostic System was able to produce differential diagnoses with precision and recall comparable with that of doctors, in some cases exceeding human level performance.⁶ The University Hospitals Birmingham NHS Foundation Trust is considering whether a modified version of this tool could be used to ‘triage out’ avoidable attendances that present at emergency departments.⁷ A study in Belfast showed that an electronic triage system enabled the osteoporosis service to actively manage referral demand, supporting alternative management pathways through use of virtual advice and direct to investigation services.⁸ In Halifax, Canada, the Spinal Problem E-Referral system (SPER) which offers protocol-based decision support to streamline and standardise the referral process from primary care physicians has passed a pilot usability study and been deemed to be easy to use by physicians. In operation, the SPER system identifies critical cases and triages them for specialist referral, whereas for noncritical cases SPER system provides clinical guideline-based recommendations to help the primary care physician effectively manage the patient.⁹ Self-referral has also been associated with higher quality of care in the case of cancer patients, as identified by a study in America.¹⁰

Despite successful and promising results, e-triage via questionnaire-based technologies such as apps and online web portals has not yet translated into clinical practice on a wider scale. In this study, the focus has been on determining the usefulness and accuracy of a symptom-checking triage tool to minimise the number of inappropriate referrals to the acute surgical ward to reduce load on emergency and surgical departments.

This study has provided an evaluation of the appropriateness of surgical referrals for same-day surgical reviews at Southend University Hospital, giving us a clear indication as to the high percentage of patients who have been found to be referred inappropriately. It has also suggested an underuse of the available outpatient referral pathways for semi-urgent and nonurgent surgical patients. Patients have responded favourably to the questionnaires and filling them out has proved to be quick and easy. An app or web-based surgical triage tool based on the screening tool used in this study can be used by primary care physicians and triage nurses in the emergency department. In most emergency departments, blood and urine tests are conducted on the patients by the triage team. These patients must then wait for further review by emergency department medical staff who later refer the patients to the surgical unit, enhancing the time spent in the emergency department and hospital. This has a lot of potential to save time and money for the NHS.

There are certain limitations of the study that must be considered for the interpretation of the results. The study used a newly formulated questionnaire which has not been validated. The plan was to validate the questionnaire during the second cycle of this audit. However, it was not completed owing to the second wave of COVID-19 in the UK and we aim to finish the audit loop once the COVID-19 crisis is over. It was not possible to find a validated questionnaire encompassing all surgical presentations for the purpose of the study. The data were collected from a single centre, which could lead to potential selection bias. Furthermore, the study evaluated referrals from both GPs and the emergency department, and therefore it has not been possible to distinguish between the rates of inappropriate referrals from both domains. It could also be argued that at present the screening tool does not take into consideration previous surgical admissions for similar or related problems. Whether this should change the threshold for admission and urgent surgical review is something that needs to be investigated with senior specialists in primary care and surgical fields. Lastly, use of the screening tool, may potentially lead to more load on ambulatory clinic services in the department. There will be a few patients who may not require ambulatory clinic follow-up. In this case, the screening tool doctor can decide whether the patient will require ambulatory clinic review or can be discharged completely with safety netting. Patients attending the ambulatory clinic with abdominal pain, may require further imaging with ultrasound or computed tomography (CT) scan. Most of the clinics, in our experience, have slots allocated to them for ultrasound by ultrasonographers, however, if the patient requires a CT scan, that has to be discussed with the radiology team by the surgeon carrying out the clinic. Hence, the use of screening tool will require agreed support from the radiology department.

The concept behind this screening tool is to make it simple to follow and to create one referral tool for the whole specialty. In our experience, we have found that in our local trust and nationally, there are several management guidelines and referral pathways that medical staff in the emergency department find cumbersome. These guidelines and pathways are not only hard to find in the system or on the local trust website, but are also difficult to follow with numerous ‘arrows’ in various directions. This tool is an initial ‘one size fits all’ approach but has the potential to reduce surgical admissions. Further assessment and validation of the tool will be helpful for its pragmatic use.

Further studies with a large data set, set in an emergency department and used by the triage team are required to fully evaluate use of the tool in a pragmatic way. The screening tool is used for the common presentations for acute surgical emergencies. It does not include patients presenting with trauma and injuries, early postoperative complications, testicular pain and acute vascular symptoms. Because the risk of missing acute and life-threatening diagnosis with these symptoms is high, we currently suggest these patients be seen by the surgical team on an urgent basis. Similarly, acute appendicitis commonly presents in male patients with right iliac fossa pain, hence, we consider these patients should be urgently assessed by the surgical team. Once the screening tool has been evaluated further in a further cohort of patients, it can be modified to also assess other presentations. Further research is also required to modify the screening tool so that it can predict imaging modality for the patient before they are seen in the ambulatory clinic, making it more efficient. If the patient is seen by the emergency department doctor with abdominal pain and the doctor has clinical concerns despite the outcome of discharge by the screening tool, the emergency department doctor should certainly refer the patient to the surgical team. Hence, the use of the screening tool should be for emergency department doctors only.

Conclusion

In conclusion, the screening tool is safe to use and has the potential to reduce the number of patients referred to the acute surgical team and divert them to a surgical ambulatory clinic. The sensitivity of the screening tool for identifying patients who require hospital admissions is 100% indicating that it is very safe to use. What makes it different from other pathways for managing various medical conditions is that it is simple to use and can be applicable to many common surgical presentations. This screening tool can be used to support decisions made by primary care and emergency department physicians about whether patients presenting with surgical symptoms need an urgent surgical review in hospital or can be swiftly discharged with an ambulatory clinic appointment. More studies are needed to ensure the safety of the questions used in the study as well as an evaluation of the different technological platforms that could best support patients and physicians using this tool.

Data availability

Data from the study is stored within Southend University Hospital. It can be obtained upon request to the corresponding author.

Conflict of interest

There were no conflicts of interest in the execution of this study.

Funding

No funding was obtained to run the study. AM is supported by the NIHR Applied Research Collaboration NW London.