Cost analysis of nurse telephone consultation in out of hours primary care: evidence from a randomised controlled trial

Initial set up costs for the intervention (for example, furniture, computer and software supplies) were met by bodies funding the trial and were administered through the university finance system. Data on capital costs were obtained from university records and these costs were annuitised using a discount rate of 6% as recommended by the NHS Management Executive^w1 and assuming a useful life of three or five years as appropriate. Data on operating costs were collected from the participating cooperative. Data on length of hospital stay were taken from the trial database. Costs per inpatient day were calculated using national data,^w2 and data on surgery attendance within three days were extrapolated from a parallel trial of overnight calls.^w3 Savings were calculated for differences in outcome during the trial year (emergency hospital admission, home visits by a general practitioner, and surgery attendances within three days) using point estimates and their 95% confidence limits. In calculating the size of the original trial, we used a one sided calculation as we were primarily interested in safety and establishing that the intervention was no worse than current care in generating adverse events. We could have examined differences between rates but did not as this was inappropriate in the context of an equivalence trial. It was clear from visual examination of the data, however, that there was a reduction in hospital admissions in the intervention arm, and we justify a secondary analysis of these data for the purpose of the economic evaluation. Savings associated with reduced emergency admission to hospital are attributed to NHS secondary care budgets, and differentiated from savings accruing to general practice, which are reported separately. Additionally, values for surgery attendances within three days of a call are extrapolated from a small dataset, and will show wide confidence intervals.

Costs of the programme

Six part time senior nurses (1.4 whole time equivalent) were employed at mid point G grade, without extra duty payments for unsocial hours. Employers "on costs" and VAT are included, although these transfer payments would normally be excluded if a societal rather than an NHS perspective were assumed. An education programme comprising 100 hours taught time was provided before the trial, and the costs of this are annuitised at 6% over three years. The cost of continuing professional education for the nurses is included by inflating salary costs by 2%.^w1 A project nurse was employed (H grade 0.5 whole time equivalent) to assist with nurse education and to support the day to day management of the trial. One of us (VL) provided 10 days of lecturer time. The manager of the cooperative contributed to service developmentC the decision support software (tas) had to be linked to the existing call management system, and intermittent technical support was required during the year.

Costs of computer hardware and software are taken from invoices and include VAT at 17.5%. These are annuitised at 6% over three years to reflect the limited currency of information technology. Costs of furnishings and of a digital tape recorder used to record all calls are annuitised at 6% over five years.

Savings for the NHS

Length of stay for all adults and for children under the age of 16 years was analysed for each arm of the trial using dates of admission and discharge. We used national data for cost per inpatient day^w2 to calculate the cost of emergency admission. This cost had two components, an accident and emergency cost and a generic inpatient day or paediatric inpatient day cost. Around 56% of adults were admitted through accident and emergency, as general practitioners were able to make direct admission arrangements with some hospital wards including those for elderly care, coronary care, gynaecology, and obstetrics. During the last three months of the trial a medical admissions unit opened, receiving all patients aged 16 years and over that were medical emergencies. We have therefore proportionately reduced the unit cost of accident and emergency admission of £296 to 56% of its value, giving a value of £166 per admission. General practitioners were also able to admit children directly to the paediatric ward and in the control arm; 16 of 99 (16%) children received accident and emergency care. In the intervention arm fewer children were admitted to hospital, but a greater proportion of those admitted were assessed in accident and emergency (29 of 71: 41%). Savings from reduced admission of children in the intervention arm were therefore attenuated by the increased costs of accident and emergency care. Nurses were not able to admit patients directly to hospital, and the most reasonable explanation for this difference seems to be that a nurse would recommend that a child be taken to hospital rather than wait until a general practitioner was available to make a direct referral. Costs for emergency admissions may vary in other settings where a greater proportion of patients are admitted through accident and emergency. The analysis of savings for the NHS is based on average costs of admissions to hospital, on the assumption that resources freed up by the programme would be employed for other patients at a cost effectiveness ratio similar to those of other widely accepted hospital based treatments or, alternatively, that these would be redeployed in the long run to other forms of care.

Savings for general practitioners

Nurses managed 50% of calls, but did they simply postpone patient contact with a general practitioner? With over 14 000 calls managed during the year it would have been impractical to follow up each case to see how many patients consulted their general practitioner the next day, but the question of whether patients attended surgery is pertinent. An estimate of the likely rate of surgery attendance within three days of a call was derived from a separate, parallel trial of nurse telephone consultation at night, undertaken over four weeks.^w3 Savings are calculated using Netten’s unit cost of £14 per consultation^w2 and assuming that the same proportion of attendance at surgery within three days of a call would have taken place with the evening and weekend service. The uncertainty around this assumption is addressed in the sensitivity analysis. We cannot assume, however, that a reduction in follow up visits would lead to a reduction in fees per capita, even in the long run. The reduction may lead to a less heavy workload for general practitioners or to shorter waiting times for appointments, but it is unlikely to have any financial consequences. Of greater consequence for the NHS is the likelihood that surgery attendance within three days of a call may result in a prescription for drugs or diagnostic tests. The savings we report per consultation saved are more likely to be a reflection of the opportunity cost of the general practitioner’s time.

We calculated reduced travel costs associated with home visits using a unit cost for travel of £4.18 per home visit by a general practitioner.^w2 This assumes use of own vehicle as a proxy for use of a cooperative lease car and driver. Similarly, these savings for general practitioners do not translate into savings for the NHS.

w1 National Health Service Management Executive. Cost allocation general principles and approach for 1993/4. Leeds: NHSME, 1994.

w2 Netten A, Dennett J, Knight J. Unit costs of health and social care. University of Kent at Canterbury: Personal Social Services Research Unit, 1998.

w3 Thompson F, George S, Lattimer V, Smith H, Moore M, Turnbull J, et al. Overnight calls in primary care: randomised controlled trial of management using nurse telephone consultation. BMJ 1999;319:1408.