In the UK, routine ambulance service management consists of the transportation of patients to the nearest emergency department (ED) unless they refuse to travel. In England in 2006, 73% of responses made by the ambulance service resulted in patients being transported to hospital.1 Emergency medical service (EMS) staff do not currently routinely divert patients to different hospitals or other care providers except in a limited proportion of cases. This is different to current US practice, where patients judged to have certain clinical conditions (eg, major trauma) may be transported to an appropriate facility that is not necessarily the closest. In addition, when a hospital in the United States becomes overcrowded, EMS staff are put on “diversion” with the aim of reducing hospital crowding by matching clinical need with the facilities of receiving units.2

At the end of 2006 the Department of Health published two documents proposing changes to how emergency services are configured in England and Wales.3 4 This included the disposition of acutely unwell patients to hospitals that may not be the nearest, but may provide the facilities that it is judged the patient requires, but also the transfer of patients to less acute facilities with no ability to admit such as minor injury units. These changes mean that EMS staff will take on more clinical responsibility and decision-making for prehospital patient triage and disposition. The purpose of this study was to survey EMS staff regarding the appropriate use of their service and determine whether they can accurately predict which transported patients will require admission to hospital. There have been no previous UK studies on this area.

METHODS

Setting

This prospective study was undertaken at an adult ED of a major urban UK teaching hospital. The ED has approximately 90 000 new patient attendances per year and of these approximately 30 000 are ambulance transports. The ED serves a population of 513 000.5 The ED and hospital is served by ambulance service staff from the Yorkshire Ambulance Service (YAS). Between April 2006 and March 2007 there were 69 162 ambulance transports in the catchment area for this ED, of which 50 924 were emergency responses.

Participants

All EMS staff attending the ED with a patient during “data collecting sessions” between 7 May and 5 June 2007 were eligible for participation in the study. On arrival in the ED, a researcher approached the attending member of each ambulance crew. The aims of the study were explained, their anonymity was reassured and they were invited to complete the brief study questionnaire about the patient they had just transported to the ED. Crews who were transporting a patient who had seen a general practitioner or an emergency care practitioner before transport were excluded. EMS staff not wishing to participate were free not to do so.

During the study period only one EMS staff member refused to complete a questionnaire, citing a personal preference not to do so.

Data collection

The questionnaire asked the EMS staff to predict whether their patient would be discharged from the ED or admitted to hospital. The questionnaire asked EMS staff to: state whether they felt attendance at the ED by emergency ambulance was appropriate; predict whether patients would be admitted to hospital or discharged from the ED.

Additional information recorded included date and time of arrival in the ED, level of the ambulance service staff member completing the questionnaire (eg, paramedic, emergency medical technician) and years of experience in the ambulance service.

The questionnaire was piloted over one 4-h session to ensure EMS staff comprehension and the relevance of the questions.

The start time of the data collection “sessions” were allocated and systematically rotated through 24 h over a one-month period, ensuring that any inherent influences due to day and time of the week were accounted for. Each “session” lasted 4–6 h and there were 24 “sessions” over the whole study period. During each session, all EMS staff were approached and invited to complete a questionnaire in relation to the patient they were transporting. Questionnaires were administered by two medically qualified researchers.

The hospital computer system was used to record actual ED diagnosis (trauma or non-trauma) and patient disposal from the ED. No clinical notes were reviewed.

Sample size

If we assume that approximately 50% of the cases that the EMS staff transport to the ED are admitted to hospital, then in order to estimate this proportion with ±5% (ie, 95% CI 45% to 55%) would require 385 patients. With half of these patients admitted to hospital and the other half discharged then we can estimate the sensitivity and specificity of the EMS staff predictions of patient disposition (compared with the actual ED disposition) with a precision of ±0.065 assuming a sensitivity or specificity of 0.70.

Data analysis

Study data were entered into SPSS version 12.0 and descriptive data were used to calculate sensitivities, specificities, positive predictive values (PPV) and negative predictive values (NPV) with 95% CI. Sensitivities and specificities were compared between different groups (eg, trauma and non-trauma patients) using χ² tests.

A separate record was kept of how many questionnaires each EMS staff member had completed. This study was conducted as a “service evaluation” and permission for it was granted by YAS (south area) and staff at the ED.

RESULTS

There were 2553 emergency transports to the ED during the study period, and of these, 397 EMS staff were approached for the study (15.6%). One EMS staff member refused to participate, citing a personal preference not to do so. The EMS staff completed 396 out of 397 questionnaires (15.5% completion rate). The average age of all patients transported to the ED during the study period was 56.1 years (SD 24.0, range 14–110) and in the study group was 55.5 years (SD 23.6, range 16–97). Baseline data, ambulance transports, actual disposal, diagnosis and completing EMS staff are shown in table 1.

Questionnaires were completed by 149 different EMS staff. The mean number of questionnaires completed by each EMS member was 2.7 (SD 2.0, range 1–8). When asked about the need to attend the ED, it was the opinion of two-thirds of EMS staff that the patient transported required an emergency ambulance (n  =  263, 66.4%), in the remaining one third (n  =  133, 33.6%) the EMS staff felt that either the attendance or the method of transport was inappropriate (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 EMS staff opinion about level of care required by transported patients. ED, emergency department; EMS, emergency medical service.

Table 2 demonstrates agreement between EMS staff prediction of the need for admission and the actual disposition at the hospital. Sensitivity, specificity, PPV and NPV were calculated with 95% CI for: all predictions of disposal; by diagnosis (non-trauma versus trauma); by role (paramedics versus ambulance technicians).

In the study group of 396 patients, 187 patients (47.2%) were admitted to hospital from the ED and 209 (52.8%) were discharged. EMS staff predicted that 46% (182/396) of patients would require admission to hospital. Of these, 26.4% (48/182) were subsequently discharged from the ED. The EMS staff predicted that 54% (214/396) of patients would be discharged from the ED. Of these, 24.8% (53/214) were admitted to hospital.

Table 2 shows the agreement between EMS staff prediction of the need for admission and the actual disposition to hospital (sensitivity 71.7% (134/187); 95% CI 65% to 78%). The ability to predict patients who would be leaving the ED was better (specificity 77.0% (161/209); 95% CI 71% to 82%).

EMS staff were significantly better at predicting admission in non-trauma cases compared with trauma cases (sensitivity 75.9% versus 57.1%, difference 18.9%; 95% CI 2.2% to 35.4%; χ²  =  5.53 (df  =  1), p = 0.02). However, their ability to predict patients who would be discharged was significantly better in trauma cases than non-trauma cases (specificity 86.6% versus 67.6%, difference 19.0%; 95% CI 7.0% to 31.0%; χ²  =  7.91 (df  =  1), p = 0.005).

When comparing the disposal predictions of paramedics with emergency medical technicians, there was no significant difference between the ability to predict admission (sensitivity 76.8% versus 69.5%, difference 7.3%; 95% CI −6.3% to 20.1%; χ²  =  1.03 (df  =  1), p = 0.31). However, when predicting those who would be discharged, technicians were significantly more accurate than paramedics (specificity 83.8% versus 64.4%, difference 19.4%; 95% CI 6.8% to 32.1%; χ²  =  10.1 (df  =  1), p = 0.001).

DISCUSSION

Our study has shown that EMS staff have a reasonable level of accuracy when predicting patient disposition following ED attendance. EMS staff were slightly better at predicting discharge than admission. The ability to predict admission accurately was significantly better in non-trauma than trauma cases but the ability to predict discharge was significantly better in trauma than non-trauma cases. There was no significant difference in the ability between paramedics and technicians to predict admission correctly but technicians were significantly more accurate than paramedics at predicting discharge.

The Department of Health plans to reconfigure acute services in England and Wales3 4 suggest that patient care for certain conditions will be improved by transfer to a more appropriate facility. These changes represent a move away from the traditional transportation role of the ambulance service to an increased triage role, signposting patients to the most appropriate place for their needs. This role requires skills in assessment and decision-making for ambulance staff, along with greater clinical responsibility and accountability. There has been no previous work in the UK looking at EMS staff ability to predict disposition of the patient they are transporting. Our study has shown that with standard UK ambulance service training, EMS staff already have some skills in predicting disposition. Our findings are in line with similar previous studies from the USA, evaluating EMS staff ability to predict disposition.6^–8 Levine et al,6 when comparing predicted admission with actual admission, showed an overall sensitivity of 62% (95% CI 54 to 68) and specificity of 89% (95% CI 86 to 91). The suggestion that EMS staff appear to be better at predicting discharge is replicated by Price et al,7 who demonstrated a sensitivity of 82.6% (95%CI 77.2 to 87.0) and a specificity of 76.6% (95% CI 69.1 to 82.8).

The difference in sensitivity between predicting disposition in non-trauma compared with trauma patients may be explained by the availability of near-patient tests. EMS staff have the facility to assess non-trauma patients clinically using vital sign measurement and cardiac monitoring and this may account for the increased sensitivity, whereas in trauma cases, radiology is often required before accurately being able to predict disposition. Levine et al6 and Price et al7 found improved predictions when EMS staff were dealing with trauma cases, but were unable to account for these findings. Richards et al8 concluded that the predictions of disposition improved if patients had complaints such as shortness of breath or chest pain, but their study excluded patients with other acute medical and traumatic conditions requiring immediate action on arrival in the ED.

The study also found that technicians were significantly better at predicting patients who would be discharged from the ED. It may be that the greater clinical knowledge and skills of paramedics have encouraged them to make more cautious decisions about disposal.

Limitations

This study has several limitations. All the data in the study were collected from a single centre. In view of the relatively small proportion of ambulance transports sampled (15.5%), bias could have been reduced by increasing the number of “data collecting sessions”. The study was sufficiently powered to detect significant results. The sample size required was exceeded (n  =  385 versus n  =  396), therefore sampling more EMS staff would have added little to our results. We had excellent participation from EMS staff during our “data collecting sessions” of data collection, reducing the possibility of selection bias.

In an attempt to maintain confidentiality and increase participation in the study, the researchers did not identify the EMS staff on each questionnaire, but on a separate tally sheet. We were therefore not able to link individual questionnaires with named EMS staff and so could not account for any effect that “clustering” by EMS staff may have had on the results. Some EMS staff completed more than one questionnaire and it is not clear whether their predictions were independent events or were influenced by previous questionnaire completion.

Some bias may have been introduced if EMS staff opinions were affected by conversations with other EMS staff, triage nurses and doctors. In order to minimise this, we interviewed only the attending EMS staff member. The ability of staff to triage to the correct facility was not specifically measured in this study. This would require further evaluation.

CONCLUSIONS

Staff in one UK ambulance service have shown reasonable accuracy when predicting the likelihood of admission of the patient they transport. Results would seem to be in line with previous work from the USA. Further work is needed to support this conclusion and identify further training that would be required in order to increase the ability of ambulance service staff to triage patients safely to an alternative destination of care and the education and training required to do so.

FUTURE DIRECTIONS

Future research should be directed at whether additional training or guidelines would improve the accuracy of EMS staff decision-making when considering prehospital patient triage and disposition. The impact of this research would be important in measuring the reduction of ambulance journeys and ED attendances and should therefore be considered alongside any reconfiguration of emergency services in the UK.