Cost effectiveness of community leg ulcer clinics: randomised controlled trial

Recruitment

The study was approved by appropriate ethics committees. Recruitment was from September 1994 to May 1995 in eight clinics (four urban, two suburban, two semirural) in four community trusts in Trent. The inclusion criteria were a venous ulcer below the knee to the foot that was present for at least 3 months and ability to travel to the clinic. Patients were assessed by clinic coordinators using Doppler ultrasound. Those with an ankle brachial pressure index <0.8 were excluded. The nurses all adhered to the standardised clinic protocol. A total of 233 patients provided written informed consent (fig 1). A random assignment schedule and serially numbered, sealed, opaque allocation envelopes were prepared in advance for each of the eight clinic sites.

Fig 1

Diagrammatic representation of sample size

• Download figure
• Open in new tab
• Download powerpoint

Intervention

The intervention was weekly treatment in a leg ulcer clinic. Clinic coordinators all completed a course on management of leg ulcers (ENB N18) and additional training days to practice applying four layer bandaging (supervised by experienced practitioners). Treatment adhered as closely as possible to the Charing Cross bandaging technique.9 After healing, patients were prescribed class 2 compression stockings and were reviewed by the clinic coordinators every 3 months. For the control group, as it was a pragmatic trial, nurses in the home provided the patients' usual treatment.

Outcomes and follow up

The primary clinical end point was time to complete healing of all ulcers within the 12 month follow up period. The nurses recorded the date of healing, defined as re-epithelialisation of all ulcers. Secondary end points were health status, time to first recurrence of a healed ulcer, and the number of weeks the patients were free from ulcers.

In the absence of disease specific measures,13 four measures of self perceived health status were used to cover all relevant dimensions of health: the 36 item short form health survey (SF-36),14 the EuroQol (EQ),15 the McGill short form pain questionnaire (SF-MPQ),16 and the Frenchay activities index.17 Questionnaires were posted 12 weeks and 12 months after recruitment to collect data on health status and satisfaction. Reminder questionnaires and telephone calls were used for non-responders or missing responses.

Economic evaluation and analysis

The economic analysis was undertaken from an NHS perspective. Total estimated NHS costs per patient were calculated by estimating the cost of treatment for leg ulcers(staff time, materials, transport, and overheads) and the cost of use of other health services (general practitioner contacts and hospital). For the control group data were collected for individual home visits to produce a mean cost per home visit. For the clinic group data on use of resources were available for each clinic session but not on an individual basis. The mean cost per attendance per patient was derived from the total cost per session.

Data on the use of health services related to care of leg ulcers were also collected by the postal follow up questionnaires. Data on resources not related to patients—for example, overheads—were collected from service managers. All resources were valued at 1995 prices. Unit costs were taken, when possible, from the local service providers or national data sources. 18 19 Because of the skewed distributions of the cost data, mean costs per patient for both groups were compared by permutation tests and confidence intervals were calculated by bootstrap methods,20 which work by resampling from the observed data and thus make no assumptions on distribution.

Statistical analysis

Sample sizes were calculated based on the expected difference in rate of healing between the groups after 12 weeks. To have an 80% chance of detecting as significant (at the 5% level) an increase in healing from 50% to 70%, 206 patients were required.

Fig 2

Healing times of initial leg ulcers by treatment group (solid line=control group, n=113; dotted line=clinic group; n=120). Log rank test statistic 4.90; df=1; P=0.03

• Download figure
• Open in new tab
• Download powerpoint

All the data analysis was by intention to treat. We used t tests or Mann-Whitney tests for continuous measurements, χ² tests for categorical data, the Kaplan-Meier method to calculate the time from recruitment for the initial leg ulcer to heal, and the log rank test to compare the healing times of both groups. Previous work suggested that age, ulcer area, ulcer duration, and history of deep vein involvement were important factors in predicting time to healing.21 Cox proportional hazards regression analysis was used to adjust healing times for these variables. 22 23 For each dimension of the four measures of health status, we calculated a summary measure of average health gain per month for all patients as the area under the response curve over the 12 months of follow up, measured with respect to the baseline score.24

Health status and satisfaction

For most dimensions of the SF-36 and EuroQol, health status deteriorated over time, with no difference between the groups. For three dimensions of the SF-MPQ there was evidence that the patients in the clinic group were more likely to experience a reduction in leg ulcer pain per month than control patients. Satisfaction was high in both groups, but there were no significant differences between the groups.26

Economic evaluation

Complete data on resource use were available for 66% (214/323) of clinics and 62% (2110/3429) of recorded home visits. From these data, the mean (SD) cost per clinic attendance was estimated at £29.90 (£14.18), while the mean (SD) cost per home visit was estimated at £10.60 (£3.79).

The annual treatment cost was estimated for each patient by using the mean cost per treatment, the mean frequency of treatment, and patient specific data on weeks with an active ulcer. The mean annual treatment costs were £804.03 and £681.04 for the clinic group and the control, respectively; a difference of £122.99 (£1.56 to £234.84). The control group reported greater use of general practitioner and hospital services (table 4). When these NHS costs were combined with treatment costs, treatment for the clinic group remained £14.51 more expensive (−£201.94 to £214.25).

One way sensitivity analysis27 was undertaken to investigate the effect that different assumptions might have on mean costs. Changes in assumptions relating to treatment costs and overheads in the control group did not significantly affect the magnitude of costs (see table 4). Weeks free from ulcers was the most appropriate measure available for the cost effectiveness analysis. The additional cost for the clinic group treatment (£14.51) for achieving the benefit of 5.9 ulcer free weeks gave an incremental cost effectiveness ratio of £2.46 (−£31.94 to £99.12) per ulcer free week.

Economic evaluation

We undertook a modelling exercise to make the cost results more applicable to the real world. Clinic coordinators indicated that it was not necessary to use G or H grade nurses for the service. Reanalysis of the data with E grade salary costs gave a mean cost per clinic attendance of £27.61 and an annual NHS cost of around £740 per patient. The restrictions on throughput imposed by the trial produced an artificially high mean cost per attendance at the start and end of the trial. We estimated that by increasing throughput to more than 10 patients the mean cost per attendance could be reduced to £24.87 and the annual NHS costs to around £670 per patient. Altering both grade mix and throughput produced a mean cost per attendance of £23.37 and mean annual NHS costs of around £630 per patient (compared with £681 for traditional home treatment). With either scenario, the clinic treatment with four layer bandaging would dominate the home based treatments, being both cheaper and more clinically effective.

The mean cost per clinic attendance (£29.92) was slightly more than the cost reported in Riverside (£22.10 at 1995-6 prices).11 The modelling work showed that small changes in the running of the clinics gave a similar cost (£23.37 per attendance). The mean cost per patient per year for clinic group patients in this trial (£804.03) was more than the cost identified in Riverside (£422 at 1995-6 prices). This difference could be explained by the healing rates in the Riverside study (80% of ulcers within 12 weeks).

Although the perspective of the economic evaluation was the NHS, a societal perspective would have included an account of patients' time and personal costs and loss of production due to treatment and poor health. The relative costs may have changed with this perspective as there were greater time costs for the patients in the clinic group.

This trial did not show the impressive healing effects reported in the early studies with four layer bandaging, 9 10 but overall the clinic healing rates improved on care provided by the district nursing service. The improved clinical outcomes were achieved at a small incremental cost to the NHS. If the conditions of nurse grade and throughput per session were optimised, the clinic treatment with four layer bandaging would be more clinically effective and less costly for community trusts than treatment in the home.