Randomised Controlled Trial

To compare the effectiveness o f care between specialist nurses and senior house officers (SHOs), an open randomised controlled trial (RCT) was performed. Patients were randomised to receive care from a specialist nurse or a senior house officer and those who attended their appointment had their clinical records audited using the audit form described below.

M easuring effectiveness of care

Diagnosing and treating sexually transmitted infections at the Mortimer Market Centre is standardised through the use o f an in-house ‘Clinic Guide’. For example, a woman presenting with a vaginal discharge should have a defined set o f investigations performed to reach a diagnosis. Once a diagnosis has been determined, treatment and instructions should be given according to the Clinic Guide. The clinic also has a casenote proforma for documenting care processes. This proforma prompts a series of questions to ensure a consistent and standard level of reporting within the patient casenotes.

In this study, effectiveness of care was defined as managing care according to the Clinic Guide. To assess and measure effectiveness of care, an audit form (Appendix 3) was developed and piloted to audit the completion of the casenote proforma and adherence to the Clinic Guide. The audit form listed thirty key variables in addition to a section to record the patient’s date o f birth, ethnic origin and presenting condition. From the audit form, a scoring system was devised to provide an overall unitary index score, out of a total of 100 points, for each set o f clinical casenotes. Objective criteria were established on how to judge each variable before allocating a score between zero and one. For example, a score of zero was given if details o f the sexual history were not recorded, 0.5 if there was incomplete data and a score of one if a complete sexual

history was recorded according to the clinic proforma. In instances when no information was required, for example when the patient did not require an examination, a non- applicable option was available. The unitary index score was calculated for each set of patient casenotes and then converted to a percentage for analysis. The main variables associated with the outcome of specialist nurse and SHO decision making (diagnostic test request, preliminary diagnosis and treatment provided) were then analysed independently.

Since two different models of care were essentially being compared, documentation by staff nurses supporting the SHOs and documentation from doctors supporting the specialist nurses was also counted. Casenotes in which appropriate management was difficult to determine were discussed with the lead consultant physician.

Score reliability

The reviewer was not blinded to the identity o f the practitioners. However, fifty percent (112/224) of clinical records were randomly selected and reviewed for score reliability by a second person (lead consultant physician). Pearson’s correlation coefficient was used to measure the strength of the association between the final unitary index scores of the reviewers. For each of the thirty items used to create the unitary index score, the proportion of cases where the two reviewers agreed as to whether the item was recorded or not was then computed to determine those items where discrepancy between reviewers was greatest.

RCT sam ple size determ ination

Dimond (1994) states that nurses need to consider the legal aspects o f role expansion when adopting roles that are normally performed by doctors. Patients are ‘entitled to the

standard of care which would be expected of the skilled practitioner exercising and professing to have that special skill’ (p.65). Therefore, nurses undertaking activities normally performed by doctors are expected to meet the standards of care that would be required from a doctor were the doctor to perform the tasks. These principles were key to determining the sample size and outcome measure comparisons used in this study.

The objective of many trials is to show that one treatment is significantly better than another, or that two treatments are equally effective (equivalence). The principal objective of this study, and that used in determining its power, was to establish that care provided by specialist nurses was not inferior to care provided by SHOs. Senior house officers (SHOs) were chosen as controls as this was the minimum level o f care expected of the specialist nurses.

The sample size for the clinical record audit was based on a sample of (November 1998) case notes (n=23) that were reviewed utilising the audit form. The notes reviewed had a mean score of 85.5 (s.d. 7.5). Based on this we accepted specialist nurses functioning at a level at least as good as SHOs, if we were 95% certain that their true mean score was no more than five points less than the SHOs. The necessary sample size depended on how specialist nurses were expected to perform relative to SHOs. Therefore, a sample of 130 in each arm was calculated with 90% power to detect whether nurses were in truth three points less than SHOs on average, using two tailed tests.

RCT sampling strategy

Patients were randomised to see a specialist nurse or a doctor when they telephoned for an appointment. SHO cases were selected from the doctor group at a later date, as reception staff were unable to determine between SHOs and other medical staff grades at the time of randomisation.

A computer program was used to generate three simple random sequences in order to allocate patients to specialist nurse:doctor in ratios of 1:3, 1:4 and 1:5 with block sizes of eight, ten and twelve. The different ratios were generated to account for the daily changes in appointment availability. For example, if there were not enough specialist nurse appointments available, the ratio could be changed to account for this. The telephone reception staff were provided with a simple table listing the randomisation allocations, with a space next to each allocation to enter the patient clinic reference number as the appointment was booked. Although concealing treatment allocations in randomised trials seeks to eliminate selection bias (Altman and Schulz, 2001), the reception staff were not blinded to the randomisation allocations for pragmatic reasons. The practicalities o f managing and motivating the reception staff to use a masked system (e.g. a sealed envelope approach) may have led to poor enrolment in addition to service disruption. However, to reduce any selection bias, the time o f appointment booking was recorded next to each allocation and regular supervision of the reception staff performing the randomisation ensured that no favouring of group allocation occurred. Reception staff do not perform any triage of patients and therefore it was not in their interests to influence the order of enrolment.

Not all patients seeing specialist nurses and SHOs during the study time were randomised. Some patients attended the service without an appointment and there were times when the telephone reception staff were too busy to carry out the randomisation. However, all staff at clinic level (i.e. doctors and specialist nurses) were blinded to those patients who had been randomised.

Patients were not consented at the time of randomisation. All patients were informed o f the study when they presented at the clinic and were given an opportunity to opt out o f the randomisation if unhappy with their allocated practitioner. Post­ randomisation pre-intervention consent served as a solution to the issues of consenting

patients over the telephone. For further discussion relating to the randomisation process, see Section 5.4.3 regarding ethical issues. A summary o f the randomisation process is presented below in Figure 3.

Figure 3 - Randomisation procedure

Patient telephones for appointment

Randomised by receptionist from a fixed randomisation schedule

A specialist nurse (SpN) B senior house officer (SHO)

Name of practitioner given to patient

Patient arrives at clinic and given study information sheet - able to opt out and see a different practitioner at this stage

A2 objects & see SHO A1 see SpN B1 see SHO B2 objects & see SpN

Intent to treat analysis i.e. A1A2 compared to B1B2

RCT Analysis

Following the randomisation period, patients who were randomised to specialist nurse and SHO groups were checked for attendance. Those who attended their appointment had their casenotes reviewed. Data were extracted and entered onto the audit form as described above. Following this, data from the audit forms were single-entered and analysed using SPSS. An intention to treat analysis was used. For instance, if a patient was randomised to see a specialist nurse but was referred to a doctor in the course o f the appointment, the data were analysed within the specialist nurse arm. The ^ test, and Fisher’s Exact test where appropriate, were used to compare the specialist nurse and SHO groups with respect to rates o f attendance and the individual variables. Non- parametric Mann-Whitney IJ tests were used to compare the unitary index scores between specialist nurse and SHO groups.