Article Text
Abstract
Aim To investigate the impact, in terms of hospital admission and investigations, of individual care plans for patients who frequently attend the emergency department (ED).
Method 32 patients who regularly attended the ED at St Thomas' Hospital were included in the study. After review of ED and hospital case records, an individual care plan was prepared for future attendances. The numbers of ED attendances, hospital admissions and investigations were collated from the electronic patient record system and compared for the 12 months prior to and 12 months after introduction of the care plan. Primary outcome measure was reduction in the number of hospital admissions (as a percentage of ED attendance). Secondary outcome measures were a reduction in the number of investigations and ED attendances.
Results In the 12 months prior to introduction of the individual care plans, the 32 patients accounted for 858 ED attendances and 209 admissions to hospital. In 12 months after introduction of the care plans, the number of ED attendances fell to 517, with only 77 hospital admissions. Median number of hospital admissions (as a percentage of ED attendances) fell from 18.8% to 7.1% (p=0.014) after introduction of the care plan. There were also reductions in median number of ED attendances (19 vs 5, p=0.001), median number of radiology tests (4 vs 1, p=0.001) and median number of blood tests (55 vs 12, p<0.001).
Conclusions Individual care plans for a carefully selected group of patients who frequently attend the emergency department can result in a decrease in the number of hospital admissions and number of investigations.
- Emergency department
- admission avoidance
- frequent attendance
- case management
Statistics from Altmetric.com
Introduction
Patients who regularly attend the emergency department (ED) on a frequent basis can present a number of significant challenges to ED staff. As well as placing an increased demand on staff in ED, they are more likely to consume resources by unnecessary investigations and potentially avoidable hospital admissions.
Many of the patients who attend frequently have complex medical conditions, and many have psychosocial problems including high incidences of homelessness and drug and alcohol abuse.1 2 The needs of this cohort of patients are often not met with staff tending to underestimate the severity of illness and failing to recognise the need for emergency treatment, although a previous study suggests that those who attend frequently may in fact be as sick as or sicker than patients who do not.3 Additionally these patients are often well known to the ED staff with the potential to base management decisions on staff's perceived knowledge of the patients' problems rather than documented fact. While this method can often be effective, it introduces an element of risk and can result in misleading inaccuracies being documented in the patients' records.
With individual case management, there is an opportunity to review a patient's history and produce an individualised care plan, thus ensuring accurately documented patient information and potentially decreasing inaccuracies in patients' perceived medical problems. In addition, there is also the potential to reduce the impact of these patients by reducing the number of inappropriate hospital admissions, repeat emergency department attendances and also the number of unnecessary investigations. Previous studies on case management of frequently presenting patients have reported mixed results, with some showing an increased and others a decreased impact with case management. Most of these studies have been conducted in healthcare settings outside the UK, and therefore may not be applicable to the population seen in the UK.
The aim of the study was to investigate the impact, in terms of a reduction in hospital admissions and a decreased number of ED investigations, of individual case management through the development of individual care plans on a selected group of patients who were known to attend ED on a frequent basis.
Methods
The study was carried out in the ED at St Thomas' Hospital between August 2005 and May 2008. St Thomas' Hospital is a large inner-city hospital located near the geographical centre of London. The ED sees approximately 120 000 new patient attendances per annum. The study was designed as a pragmatic prospective cohort study, using retrospective data for comparison. Patients were identified as suitable for case management by any member of the ED staff. There was no specified number or frequency of attendances required for inclusion into the study, merely the perception by a staff member of the need for case management. Only adult patients (over the age of 16 years) were included in the programme; otherwise, there were no exclusion criteria.
Individual written care plans were prepared for all patients included into the study. This involved extensive review by one of the ED consultants of the ED records, the inpatient hospital notes, and the results of investigations stored on the electronic patient record (EPR) system. Further information was gathered from the patient's general practitioner (GP), if they were registered, and any relevant specialist consultants or other practitioners. Where indicated, information and advice were obtained from the drug and alcohol liaison team, mental health team or social services. Additionally, attempts were made to contact other emergency departments in the area if the patient was known to attend other local hospitals.
Once the management plan was completed, a second ED consultant reviewed the plans to check for appropriateness and errors. A copy of the draft plan was sent to the patient's GP and any other relevant practitioners for further comments or clarifications. Once all relevant comments had been received and incorporated, the care plan was finalised. The finalised care plan was sent to the patient's GP and relevant hospital specialists, and both electronic and paper versions were made available for use in the ED. The majority of the patients were made aware that a care plan would be used for future attendances, but for a small number thought to be regularly seeking inappropriate admission to hospital, it was deemed justifiable to withhold the information contained in the care plan.
Once the care plan had been finalised, a warning flag was attached to the patient's record on the computerised ED record system which subsequently alerted staff to any future attendance of a patient with a management plan. When the patient attended, the care plan was consulted and used to guide patient management.
The primary outcome measure was the difference in the number of hospital admissions for the 12 months prior to and the 12 months after implementation of the individual care plan. As the number of ED attendances during these two time periods would clearly influence the number of hospital admissions, we compared the number of admissions as a percentage of the total ED attendances. Secondary outcome measures included the total number of investigations (radiological and pathology) and the total number of ED attendances between the two time periods.
Ethical approval for the study was not required, as the programme was designed as a service development to deal with a perceived area of need within our ED. Under the UK COREC (Central Office for Research Ethics Committees) guidelines, this study fulfils the criteria for service evaluation.
Statistical analysis
All data were extracted from the electronic patient record system and entered onto an Excel (Microsoft) spreadsheet. Due to the small sample size and both primary and secondary outcome measures being positively skewed, all data were analysed using non-parametric tests and results given as medians with interquartile ranges. As this was a paired sample study with data collected on the same patients (before and after management plan), we used Wilcoxon signed ranks to test the effect of the management plan in terms of the various outcome measures and held the assumption that all socio-demographic variables affected the pre- and post-test outcome measures equally.
Results
A total of 34 patients were included in the study and had individual care plans produced. Patients were incorporated into the study once they were identified by staff members and their individual care plan completed. During the recruitment period (August 2005 to May 2007), a total of 83 patients were identified as being potentially suitable for inclusion in the study. In eight of these cases, after review by an ED consultant, an individual care plan was deemed unlikely to be beneficial, and a further five patients with sickle cell disease already had comprehensive individual care plans developed by the sickle cell team. However, the main rate-limiting step in the process was the production of the individual care plan for the identified patients. This was extremely labour-intensive and occurred over a considerable time period (required for initial information gathering and subsequent review and revision of the plan). Consequently, not all patients identified as potentially suitable for an individual care plan could be included in the study time period. In total, 34 of the 70 patients potentially suitable for inclusion had a care plan completed during the study recruitment period.
Two of the 34 patients recruited died during the study period. Both of these were expected deaths, one patient had end-stage liver disease, the other had end-stage chronic obstructive pulmonary disease, and both management plans incorporated appropriate end of life care. As both of these patients died shortly after inclusion into the programme, neither has been included in the subsequent analysis. Of the remaining 32 patients, the majority of the patients were male, aged between 20 and 85 years; approximately two-thirds were registered with GP, and over half abused alcohol. The socio-demographic characteristics of the patients are outlined in table 1, which shows that 20 of the 32 patients studied (62.5%) were either living on the streets or in a homeless hostel.
Over the 12-month period prior to the introduction of individual care plans, the 32 patients accounted for a total of 858 ED attendances and 209 hospital admissions. In the 12 months following the introduction of the management plan, the same 32 patients accounted for 517 ED attendances and 77 hospital admissions. ED attendances were reduced for 27 of 32 patients (84.4%); hospital admissions were reduced for 31 of 32 patients (96.9%). The median number of ED attendances was significantly reduced (19 vs 5, p=0.001), as was the number of hospital admissions (4 vs 1, p<0.001).
As the number of ED attendances would clearly influence the hospital admission rate, our primary outcome measure was the percentage of hospital admission relative to the number of ED attendances. There was a significant reduction in the median percentage admission rate (18.8% vs 7.1%, p=0.014) in the 12 months following implementation of the individual care plans. The relative chance of admission to hospital after the implementation of the management plan was 0.61, demonstrating that implementation of the management plan reduced the chance of hospital admission by 39%.
Table 2 illustrates all of the outcome measures and the corresponding hypothesis test result. As well as a reduction in ED attendance and hospital admission, there was also a significant reduction in the number of investigations after implementation of the management plan, both in median number of radiology tests (4 vs 1, p=0.001) and in median number of pathology tests (55 vs 12, p<0.001). Overall, the results in table 2 confirm the effectiveness of the management plan to reduce the number of hospital admissions as well as the number of ED attendances, and investigations.
Discussion
Our study demonstrated a significant reduction in the impact of frequently attending patients by introduction of individual care plans. In particular, the chance of hospital admission was significantly reduced following instigation of the individual care plan. Additionally, there were reductions in the number of investigations and also the number of ED attendances.
The decrease in the overall number of ED attendances between the two time periods was larger than we expected, as the individual care plan did not specifically aim to address the issue of attendance in ED, and indeed its use was only activated by the patients' attendance in ED. There are several possible explanations for why a reduction of this magnitude occurred. First, the individual care plan was sent to each patient's GP with an accompanying letter suggesting that a similar process might be suitable in the primary care setting. It is therefore possible that a number of the patients were case-managed in primary care, and specific measures were taken to reduce the number of ED attendances. Indeed, six of the patients included in the study had multidisciplinary case-management meetings, which were also attended by one of the ED consultants, following receipt by the GP of the care plan. One of these six patients died during the study, and of the remaining five, three had a reduction, and two had an increase in the number of ED attendances. Overall, the five patients had a slight decrease in the total number of attendances from 191 to 166. Although this alone would therefore not account for the large overall reduction in ED attendances, it is possible that more patients were case-managed in the primary care setting, and we were simply unaware. A randomised controlled study in a Swedish population, however, showed no reduction in ED visits for frequently attending patients (four or more visits per year) when information on the attendances was shared with primary care physicians.4
Another possibility for the reduction in the ED attendances found in our study is that some of the patients in the study were attending other departments in London more frequently instead of St Thomas', and consequently their overall number of ED attendances (to any department) was either the same or higher. London has a large number of hospitals with an ED within a relatively small geographical area; the next nearest department to St Thomas' being less than 3 miles away.
Previous studies have shown differing results for case management of frequently attending ED patients.3 5–10 While some have shown a subsequent decrease in number of ED attendances, others have shown little or no difference. These studies are not identical in their design, however, and were done within different healthcare systems.
Okin described a 12-month study on 53 patients in the USA where each was case managed by a trained social worker and showed a significant decrease in median number of ED visits (15 vs 9, p<0.01) and a reduction in overall costs.5 Sixty-seven per cent of the study group were homeless, and many of the interventions were aimed towards engaging the patient with primary care services and improving utilisation of outpatient clinics, which would therefore account for the large reduction in ED usage. The study showed no difference in the number of hospital admissions despite the reduction in ED attendance. The findings from this pilot study were further confirmed by another study by the same authors in 20086 that followed patients over a 2-year period. The study was designed as a randomised controlled trial but only included patients with psychosocial problems. Again, patients in the treatment group were case-managed by a social worker, and the primary outcome measures were psychosocial function (at 6-monthly intervals over the 2-year-period) and hospital service utilisation. Although the group that was case-managed had fewer hospital admissions, this did not achieve statistical significance.
Lee and Davenport described a study of nurse-led case management on 50 patients in the USA, with outcome measures over a 5-month period before and after implementation of case management.7 Again, the majority of interventions from the case management were either arranging appropriate primary care or reducing access to prescription medication, and they reported a slight decrease (7.4%) in the total number of ED visits in the study group. However, one study on case management in Australia actually showed an increase in ED attendances following the introduction of case management, although this did not reach statistical significance.8
A recent study by Skinner et al in a large inner-city ED in the UK showed a significant reduction in ED attendance with case management.9 Although the level of reduction was similar to that in our study, the authors reported a greater reduction in ED attendance from those patients who were not case-managed. Only 14 of the total of 36 case-managed patients in this study had an individual care plan produced as part of their case management.
The effect of individual care plans on ED attendance was studied by Pope on 24 patients in a US healthcare system.10 After introduction of individual care plan, the median number of attendances fell from 26.5 to 6.5. Patients were included only if they had attended more than a few times and had one of either chronic medical condition, complex medical condition, drug-seeking behaviour or violent/abusive behaviour. The committee was chaired by a social worker and met for 1 h a month with four new patients reviewed each month.
There are several limitations to our study. The study was designed to use each patient as their own control, and there is therefore a possibility of overestimating the effect of the case management. This is particularly true in a study about frequent users of the ED, as previous studies have noted that the number of attendances in this patient group tends to decrease gradually over time, even without any formal intervention (regression to the mean).11 However, the majority of previous studies on this group of patients have used a similar design. Additionally, we looked retrospectively at the 12 months prior to initiation of the management plan to obtain the comparative data. Although this introduces a potential source of bias, this is limited by the fact that all the data points were objective and obtained in a prospective manner.
Most previous studies on frequently attending ED patients have been done in healthcare systems outside the UK. Our study population may be relatively unique and potentially unrepresentative of other ED in the UK or elsewhere in the world. The geographical location of St Thomas' hospital in central London means that there were a large number of homeless persons (either living on the street or in homeless hostels) included in our study population (62.5%). In addition, there are also a relatively large number of emergency departments within a relatively small geographical area to which the patients could attend.
Finally, the cohort size was small and a convenience sample. The programme was designed as a pragmatic solution to a perceived problem in our ED. We felt that it was important to allow the ED staff to identify individuals who might benefit from a care plan, as they were best-placed to do so. Although a larger number of patients were identified as potentially suitable for inclusion into the study than were included, the rate-limiting step in the process was the actual preparation of the individual care plan. Due to the necessity of producing an accurate and robust care plan, this was a very labour-intensive process and thereby limited the number that could be included in the study. Overall, the cohort was therefore carefully selected as most likely to be helped by an individual care plan positively biasing the results.
In conclusion, our study demonstrates that the utilisation of an individual care plan for a carefully selected group of patients who frequently attend the emergency department can reduce the rate of hospital admissions and of unnecessary ED investigations for this challenging cohort of patients.
References
Footnotes
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- Primary survey