Effect of introduction of integrated out of hours care in England: observational study

Introduction

The 2000 review of out of hours care in England recommended that patients calling their general practitioner out of hours should be automatically diverted to NHS Direct for initial assessment by telephone (figure).1 One telephone call would lead to effective and timely advice and where necessary a face to face consultation. A national “exemplar programme” was established to enable providers to gain experience of developing integrated services.2 We measured service integration achieved in the programme and its effect on the wider health system.

Integrated model of out of hours care. Source: Department of Health, 2000

• Download figure
• Open in new tab
• Download powerpoint

Methods

We carried out an observational before and after study of demand, activity, and trends in use of other health services in 34 English general practice cooperatives with NHS Direct partners (“exemplars”). Of 34 cooperatives, four functioned as a communications “hub,” making 31 distinct sites. We selected newly integrating exemplars for in-depth study rather than those with existing arrangements with NHS Direct. Further selection was made on the basis of a patient population of 200 000 or more, availability of computerised records of calls, and readiness to proceed. Eight exemplars met these criteria, from which we selected four “case exemplars” to obtain variety in the approach to integration and geographical location.

We defined integration as NHS Direct handling calls and giving advice to a proportion of people, redirecting others as appropriate. We defined single call access as automatic diversion of a call to NHS Direct from the practice or cooperative number. Nineteen cooperatives with more than 100 000 patients were invited to take part as controls. Eleven (58%) agreed and 10 provided data. All had out of hours contact rates within the range described in a previous study of representative English cooperatives.3 We requested data on the extent of integration and single call access from all exemplars. Within the area covered by each case exemplar and control cooperative we collected data for a year before and after integration on out of hours calls to NHS Direct and cooperatives; new attendances at emergency departments, minor injuries units, and NHS walk-in centres; and 999 ambulance transports to hospital. Patterns of data collection in participating NHS organisations meant that data could not be restricted to out of hours periods for all sources.

We examined the effects of introduction of case exemplars on use of other services by fitting linear regression models to data reporting the number of contacts per month with each immediate care service. These methods were adapted from those used in the evaluation of the first three NHS Direct pilot sites.4 Integration was achieved at different times for each of the four case exemplars, necessitating separate analysis for each. We used the Durbin-Watson test to detect first order autocorrelation in the residuals of the regression models. Systematic monthly variation in service use was removed by fitting “month” as a fixed effect within each model, with “log monthly count of service contacts” treated as an independent effect. We tested the effect of integration of each case exemplar site by fitting it as a step function (before = 0; after = 1).

Results are reported as regression coefficients with standard errors and P values, and the percentage difference between observed activity after the integration of the exemplar site and predicted activity based on data before integration (with 95% confidence intervals) for the use of each service. We created a value for total calls by summing 999 ambulance calls, new attendances at emergency departments, calls to general practice cooperatives and, where appropriate, attendances at minor injuries units (but not emergency hospital admissions) and tested for a change in the linear trend in total calls. We analysed pooled control data for each of the case exemplar integration dates to determine whether any estimated changes associated with integration of an exemplar occurred in the control sites. Although none of our case exemplar sites contained an NHS walk-in centre, several of the control sites did: for analysis we combined these data with those for the minor injuries units. All analyses were carried out in SPSS 12.5.

Results

Twenty one of 31 sites (68%) integrated all out of hours call management during the study period, but in only nine (29%), all patients achieved access with a single call. Only one case exemplar (02) carried out nurse telephone triage before integration, thus allowing direct comparison: cooperative nurses managed 2622/6687 calls with telephone advice (39%) before integration compared with 2092/7086 (30%) by NHS Direct during the three months of maximum integration (P < 0.0001). None of the other three case exemplars achieved 30%: exemplar 07, 3354/19 555 (17%); exemplar 29, 3582/18 606 (19%); exemplar 30, 2813/12 226 (23%).

A small but significant downturn in overall demand for care was seen in two case exemplars (07 and 29) in the year after integration (table 1), but this was also seen in pooled data from the control sites (table 2). The number of emergency ambulance transports increased in three case exemplars after integration, reaching statistical significance in two. In each case this was accompanied by a significant reduction in the number of calls to the integrated service.

Discussion

Observation of the first year of integration of out of hours care in England suggests there are limited efficiencies to be gained from routing all incoming calls through NHS Direct if the workload of general practice providers is insufficiently reduced. This study had a before and after design, with the limitations that imposes, but also included data on 10 control sites to account for secular trends. Most exemplars implemented the integrated model of care and gained valuable experience of building local partnerships. However, most patients still needed to make at least two telephone calls to contact NHS Direct, and then had to wait to be called back by a nurse. NHS Direct completed fewer calls with telephone advice in the case exemplars than expected,5 reflecting the extent of integration achieved.

Calls to 999 ambulance services increase annually, but we found a further significant upward change in that trend after integration. Decision support software used by NHS Direct may have influenced triage end points, and this may be amenable to change. Alternatively the increase in demand for 999 ambulances may have been a transfer effect, with patients contacting the ambulance service rather than waiting for their return telephone call. NHS Direct may not have the capacity to manage all out of hours demand in the way the 2000 review of out of hours care in England envisaged.1