STUDY OBJECTIVES

• To evaluate the effectiveness of triage and direct transportation of patients to MIUs by ambulance crews in terms of:

  • – Ambulance performance

  • – Patient satisfaction

  • – Clinical safety

• To describe factors reported by crews as encouraging or inhibiting MIU use

DESIGN

The main study was designed as a cluster randomised controlled trial. After training, for a one year period ambulance crews were asked to use protocols to transport patients who met specific criteria to an MIU during randomly selected intervention weeks (totalling six months). During all other weeks such patients were to be treated according to normal practice, forming the control group. Standard practice was to transport patients who consented to travel to A&E. At the time of the trial, crews in the two services did not normally have protocols to take patients to MIUs, although various initiatives had been and were underway before the study and during the study in other areas of the services.

RESEARCH SETTING AND METHODS

The study was conducted with ambulance crews (n = 55) at five ambulance stations in the London and Surrey Ambulance Services (LAS and SAS), with three participating MIUs in the catchment areas of those stations. The subjects were 999 patients, attended by participating ambulance crews, who fell within an agreed list of dispatch criteria that were judged to cover all potential cases of minor injury, and who were assessed as matching protocols for transportation to an MIU by the study paramedic, on retrospective review of patient report forms completed by the attending crew. This paramedic was blinded to the group to which the week in which the case occurred was allocated.

The study protocol was written to be acceptable to all the three participating MIUs (see box). Patients were unaware of the study and were allocated to either the intervention or control group according to the date of their call. Subjects were followed up through ambulance service, hospital and/or MIU records, and by postal questionnaire. The patient questionnaire used was based on the UK version of the Parkside Emergency Room Quality of Care Monitor.¹⁵ Three versions were produced for patients conveyed to A&E, MIU, or not conveyed. All versions contained items about reasons for the 999 call and satisfaction with the ambulance service. For patients conveyed to A&E or MIU, additional information about satisfaction with care received in the unit, preparation for returning home, time spent in the unit, and transport home were also requested.

All patients taken to MIU and transferred on to the A&E department were reviewed by two A&E consultants. From prehospital, A&E, and MIU records, opinion was sought on whether:

• the patient was within the protocols for transportation to MIU

• the crew decision to take the patient to MIU was appropriate

• the transfer on to A&E was appropriate, and why

• the condition or outcome of the patient was adversely affected by their initial conveyance to MIU rather than A&E

Patients who were treated and discharged from the MIU were not further reviewed as we were not attempting to evaluate care at the MIU.

All participating ambulance crews and MIU nurses attended 1 of 14 project training sessions, covering an overview of the research background and methods; the MIU project in detail; and issues the staff thought were important. Throughout the study, close contact was maintained with crews and local managers, through Project Steering Group meetings, site visits, and study newsletters.

In addition, to explore patterns of MIU use, semi-structured interviews were undertaken with crews (n = 15). A content analysis approach was taken to identify themes emerging from the interviews.

Ethical approval for the study was granted by the Multi Centre Research Ethics Committee and by 10 local research ethics committees responsible for the A&E and MIU sites that took part in the study.

RESULTS

A total of 409 patients were recruited during intervention weeks and 425 during control weeks. The proportion of patients meeting the criteria for conveyance to MIU who were taken there was low, and in addition, the proportion of patients taken to MIU during control and intervention weeks was almost the same (8.7% compared with 10.0% respectively) (table 1). This picture was similar in London and Surrey, in terms of overall use of MIUs and in terms of compliance with the research protocol. Patterns of conveyance were strikingly different, however, in that a significant proportion of patients were left at scene in London (31.4%) whereas in Surrey nearly all patients were conveyed to A&E or MIU (98.9%).

Of the 834 patients recruited into the study, 43 had insufficient details for follow up. Thus, a total of 791 follow up packs were sent and of these 559 responses were received, 473 resulting in a completed questionnaire This gave an overall response rate of 70.7%, and a completion rate of 59.8%.

There were no significant differences between responders and non-responders for age, sex, study group, or between patients taken to MIU and A&E. However, there were significant differences in the proportion of responders between MIU patients and patients that were not conveyed (81.1% versus 59.8% respectively, χ² = 9.04, df = 1, p = 0.003), and between A&E patients and patients that were not conveyed (71.2% versus 59.8% respectively, χ² = 5.39, df = 1, p = 0.027). Patients attended by SAS were more likely to respond than those attended by LAS (77.4% versus 62.2% respectively, χ² = 21.87, df = 1, p = 0.000).

As a result of low compliance with the study design, outcomes were compared for patients taken to MIU with those taken to A&E, rather than by study group as originally planned. Comparisons were adjusted for factors found to influence patient destination.

The biggest influence on choice of patient destination was distance from the incident to where the patient was taken. With this taken into account the distance to the nearest other type of facility, the time of day (in or out of hours), whether the patient had a head injury, the patient’s sex, and the service involved (LAS/SAS) all continued to significantly influence choice of destination but there was no evidence that the randomisation schedule affected the destination (table 2).

These factors were used to adjust the case mix when comparing outcomes (tables 3 and 4). Ambulance service job cycle time, time to treatment, and total time in unit were shorter for patients taken to MIU than for patients taken to A&E (−7.8, 95% CI:−11.5 to −4.1); (−84.6, 95% CI:−108.1 to −61.1); (−222.7, 95%CI:−331.9 to −123.5) respectively. MIU patients were 7.2 times as likely to rate their overall care as excellent as A&E patients (95% CI:1.99 to 25.8).

Seven patients were taken to MIU and subsequently transferred by ambulance to A&E during the study. Although the medical reviewers agreed that in three of these cases the patient did not meet the protocol for conveyance to MIU, in none of the cases did they judge that the condition or outcome of these patients suffered adversely as a result of initial conveyance to MIU. However, they did highlight three difficulties associated with implementing protocols to enable ambulance crews to convey patients to alternative destinations other than A&E. Firstly, patients may choose to attend MIU or refuse A&E despite A&E being the appropriate destination. This may prove difficult for crews to negotiate. Secondly, injuries that apparently meet protocols for transportation to MIU may turn out to have gone to MIU inappropriately after further investigation. Finally, crews may miss injuries or be misled by patients about the nature of their injuries (table 5).

The first level of analysis of crew interview transcripts indicated 26 categories related to crews deciding not to convey to MIU, and 18 categories related to crews deciding to convey to MIU. The second level of analysis combined similar categories to reveal a smaller number of themes. At this level, nine factors hindered crew conveyance to MIU, and seven factors encouraged conveyance to MIU. Operational factors such as distance to MIU compared with A&E, uncertainty about patients being acceptable to the MIU, opening times of MIUs, and restrictive MIU protocols were frequently mentioned as factors for not transporting patients to MIU. In addition, patient factors such as age, underlying medical condition, and choice were also mentioned as factors for not transporting to MIU. Despite low use, some factors were mentioned that crews reported might lead to transportation to MIU: location, perceived positive differences in service delivery at MIU compared with A&E, patient needs and choice, reduced job-cycle times, and improved handover. The study design of randomised intervention weeks seemed to cause confusion about when to transport patients to MIU (table 6).

DISCUSSION

Key findings

In both study sites the participating crews used the MIUs for only a small proportion of patients who appeared to be suitable for MIU treatment. The most important factor influencing the decision to take a patient to MIU was its location in relation to the incident. When the MIUs were used, the ambulance service and patients benefited in terms of job cycle time, waiting time, and satisfaction with care.

Study limitations

An important limitation to the interpretation of findings from this study was the failure of the study design. The crews in both participating services seemed to take several factors into account when deciding where to convey a patient, but not the randomisation schedule. Patients were equally likely to be taken to an MIU during the control weeks, when, according to the study protocol, patients were only to be taken to the local A&E, according to standard practice. This lack of compliance meant that the analysis plans had to be amended. The planned comparisons between the intervention and control groups simply would not be able to demonstrate any costs or benefits associated with the new treatment protocols allowing crews to triage and transport patients directly to MIU.

In this study, the research team were very familiar with the ambulance service setting. Because of experience on earlier trials with ambulance personnel, pre-trial training was provided for participating crews. This training focused on research methods—and the need for a control group—so that comparisons could be made on the study outcomes. In addition, extensive consultation was undertaken at the outset of the study with local crews and managers in London and Surrey. There was no indication of dissent at any stage. Practically, it proved difficult to find reliable systems of reminding crews whether the week was one in which their MIU protocols could be used or not. However, this level of non-compliance cannot be explained by some forgetfulness. Crews simply seemed not to take into account the randomisation schedule when deciding upon the destination of any individual patient.

Response rates were not high, although they were probably as high as can be expected in a study of this group of patients. Nevertheless, the patients who did not respond may not have held the same views as those that did, and this should be taken into account when interpreting results. Response rates did not differ between the MIU and A&E groups and seem unlikely therefore, to affect the comparisons made between satisfaction levels in the two groups.

General discussion

As is common in evaluative studies, the intervention being assessed was introduced alongside the evaluation. This permitted no “bedding down” or familiarisation period and although use of MIUs during the study was found to be low this could change over time. Many change management issues were raised by crews during the qualitative interviews—such as lack of confidence in MIU acceptance of patients that they transport there—which might be resolved over time.

Despite overall low use of the MIUs, a minority of patients were inappropriately taken to MIU, resulting in subsequent transfer to A&E. Although in this study, none of these were judged to have suffered an adverse outcome as a result of the original mis-triage, the sample size was small and this possibility cannot be excluded if the policy were to become routine. Crews may need more clinical training to reduce the likelihood of patients being inappropriately conveyed to MIU.

The non-compliance we have found in this study mirrors that reported before in previous work.⁷¹⁶¹⁷ Partial randomisation of paramedic crews to appropriate cases was planned in Nicholl’s 1998 evaluation of paramedic skills for serious trauma, although in the event only 185 were randomly allocated to paramedic or technician response. Of these, only 16 met the study inclusion criteria and could be included in the total sample of 2045 for analysis of outcomes.¹⁶ A further study by the same team, published in 2000, used a study design of two protocols for intravenous fluids in serious trauma, to which appropriate patients should have been randomly allocated. Again, compliance was low, particularly in the cases of most interest, those with serious injury, or in a serious condition on scene.¹⁷ A trial of telephone advice for non-serious 999 callers also suffered from low compliance on the part of call takers who did not pass over all calls falling within study inclusion criteria to the nurse and paramedic advisors, but selected calls they felt might benefit from advice.⁷ The dearth of randomised controlled trials in prehospital care research has been lamented,^18–20 but our experience in several studies has been that compliance with research protocols can be low in this area of research, and study design should be considered carefully at the planning stages of any study.

Conclusions

Study design needs to be carefully considered in prehospital care research. It is not always apparent that personnel will not comply fully with study protocols, but experience suggests that randomisation is difficult to achieve and the lower the reliance on participants to decide when to apply protocols, the more robust the study will be.

Although MIU use was lower than anticipated, benefits associated with their use have been identified to the ambulance service in terms of reduced distances travelled and reduced job-cycle times, and to patients in terms of reduced waiting time for treatment, overall time spent at the unit, and increased satisfaction.

After introduction of protocols that allow crews to use alternative receiving units many other factors may impede practice change and may need to be addressed before potential benefits are fully realised.

On balance, MIUs should be used for appropriate patients within locally defined catchment areas based on location of nearest alternative A&E department.