Abstract
Objective
To compare the cost effectiveness of general practitioners and nurse practitioners as first point of contact in primary care.
Design
Multicentre randomised controlled trial of patients requesting an appointment the same day.
Setting
20 general practices in England and Wales.
Participants
1716 patients were eligible for randomisation, of whom 1316 agreed to randomisation and 1303 subsequently attended the clinic. Data were available for analysis on 1292 patients (651 general practitioner consultations and 641 nurse practitioner consultations).
Main outcome measures
Consultation process (length of consultation, examinations, prescriptions, referrals), patient satisfaction, health status, return clinic visits over two weeks, and costs.
Results
Nurse practitioner consultations were significantly longer than those of the general practitioners (11.57 v 7.28 min; adjusted difference 4.20, 95% confidence interval 2.98 to 5.41), and nurses carried out more tests (8.7% v 5.6% of patients; odds ratio 1.66, 95% confidence interval 1.04 to 2.66) and asked patients to return more often (37.2% v 24.8%; 1.93, 1.36 to 2.73). There was no significant difference in patterns of prescribing or health status outcome for the two groups. Patients were more satisfied with nurse practitioner consultations (mean score 4.40 v 4.24 for general practitioners; adjusted difference 0.18, 0.092 to 0.257). This difference remained after consultation length was controlled for. There was no significant difference in health service costs (nurse practitioner £18.11 v general practitioner £20.70; adjusted difference £2.33, −£1.62 to £6.28).
Conclusions
The clinical care and health service costs of nurse practitioners and general practitioners were similar. If nurse practitioners were able to maintain the benefits while reducing their return consultation rate or shortening consultation times, they could be more cost effective than general practitioners.
Introduction
Although use of nurse practitioners is well developed in the United States, it is only in the past 10 years that they have become established in the United Kingdom. A nurse practitioner has been defined as “an advanced level clinical nurse who through extra education and training is able to practice autonomously, making clinical decisions and instigating treatment decisions based on those decisions, and is fully accountable for her own practice.”1 Models of nurse practitioner care have, however, developed in several different ways. In Britain, nurse practitioners working in general practice most commonly work as part of a team alongside general practitioners, and it is this model we have evaluated.
Nurse practitioners are increasingly used as points of first contact in primary care. The number of trained nurse practitioners is increasing as dedicated training programmes become more accessible. New government initiatives include nurses as front line providers for a national telephone advice service2 and for proposed new walk-in primary care clinics.3 Despite this, there have been few rigorous comparisons between doctors and nurses.Observational studies generally suggest that patients give positive reports of nurses in such roles.4,5 However, the only two randomised controlled trials comparing the cost effectiveness of nurses and doctors in first contact roles in primary care in the United States and Canada provide conflicting results.6,7 These studies were conducted on single sites with a small number of nurses. A recent meta-analysis commented on the limited evidence available to compare the cost effectiveness of doctors and nurses in primary care.8
The aim of this study was to compare the process, outcome, and costs of care given by general practitioners and nurse practitioners for patients requesting a same day appointment in 20 general practices. This group of patients was chosen because a high proportion would be likely to agree to randomisation as they would not have a strong preference for one practitioner who was already involved in their ongoing care.
Participants and methods
The study took place in 20 geographically dispersed practices in England and Wales. Table 1 shows the location, list size, and number of general practitioner partners in the practices recruited.Ethical approval was obtained for the 20 practices from local research ethics committees. Each practice employed a nurse who had completed a one or two year nurse practitioner training programme at diploma, BSc, or MSc level. The median length of time the nurses had been qualified as nurse practitioners was 3 (range 1-5) years and the median time as registered nurses was 22 (9-35) years. Each nurse practitioner had been seeing patients as first point of contact for at least two years.
Table 1.
Location, number of partners, and practice list size of recruited practices
| No of practices (n=20) | |
|---|---|
| Practice location: | |
| Inner city | 4 |
| Council estate | 2 |
| New town | 1 |
| Urban and rural | 3 |
| Urban | 10 |
| No of partners: | |
| 1 | 3 |
| 2-3 | 5 |
| 4-5 | 8 |
| >5 | 4 |
| Practice list size: | |
| 3000-5000 | 6 |
| -8000 | 4 |
| -10 000 | 4 |
| -12 000 | 4 |
| >12 000 | 2 |
Randomisation
In each practice, experimental sessions were booked when both the nurse practitioner and a general practitioner had appointments available for patients who asked to be seen on the same day. Patients were eligible for entry to the study if they requested an appointment the same day and were able to come to the experimental session. If these conditions were satisfied, the receptionist then asked patients whether they would agree to be randomised to see either a nurse practitioner or a general practitioner. A method of coded block randomisation was developed which meant that neither the receptionist nor the patient could determine the group to which a patient had been allocated at the time of booking. The coded blocks were generated from random number tables. The randomisation code was broken by one of the researchers at the start of each experimental session, at which point it became apparent which patient would see which practitioner. One of the researchers explained the study further to patients as they arrived for their appointment and informed consent was obtained. For drop-in clinics or where patients telephoned or called in after a session had started, the researchers randomised patients after they arrived in the surgery and had consented to enter the study. Randomisation continued until a minimum of 60 patients in each practice had been allocated to the clinician groups.
Patients were excluded from the study if one or more of the following criteria applied: patients who were temporarily resident or not yet registered with the practice, any patient with language or reading problems, any patient who was too ill, and unaccompanied children under 16 years of age.
Data collection
The general practitioners and nurse practitioners booked appointments at their normal intervals. For each consultation they recorded details of history, diagnosis, examination, tests carried out, prescriptions, and referrals. The time of each consultation, including interruptions, was recorded with an electronic time stamp. This included time taken by the nurse practitioners to get a prescription signed by a general practitioner. We extracted details of consultations in the following two weeks from the medical records.
Patients completed health status measures before the initial consultation and by post two weeks later (SF-369 for adults or the child health questionnaire for parents of children aged 5 to 1610). For children under 5 years, parents completed a brief health status questionnaire which had been developed for a previous study in general practice.11 After the consultation, patients completed the medical interview satisfaction scale12 or the paediatric version of this scale13 and the patient enablement instrument.14
We coded patients' diagnoses and prescriptions using Read codes. Data were double coded, double entered, and verified. For health status and satisfaction scales, scores were reported if 50% or more of items had been completed, and we used the method advised for the SF-36 scale to impute missing values.15 Costs of general practitioners' and nurse practitioners' time were taken from Netten et al using the actual grades on which the study nurses were employed.16 Costs of prescriptions were derived from the British National Formulary, and costs for investigations and referrals were supplied by the individual provider units associated with the practices.
Analysis
Because of potential correlation between the outcomes of patients treated by the same health professional, estimates of variation between health professionals may be over precise unless intracluster correlation is adjusted for.17 We adjusted outcome for the age and sex of the patients as these characteristics may also influence outcome. The statistical modelling used generalised estimating equations18in which the intracluster correlation is modelled by an exchangeable correlation structure. A logistic regression model was used for binary outcomes. Analyses were carried out with stata statistical software.19 As some of the cost data were highly skewed, estimates for costs were compared with estimates based on non-parametric clustered bootstrap to check the robustness of the analysis.20 Both estimates gave similar results and so only the direct estimates are presented.
Results
A total of 1716 patients were able to come to the experimental sessions and were informed about the study; 102 patients met one or more exclusion criteria and 298 declined to be randomised, leaving 1316 (76.7%) eligible patients who were randomised. Fifteen patients subsequently did not attend the appointment which they had booked. The figure gives further details of recruitment and response rates.
Table 2 shows the demographic characteristics of the patients and the main diagnoses. Two hundred (15.5%) patients were aged between 5 and 15, and 224 (17.3%) were children under 5. The commonest presenting problem was upper respiratory illness (36.8% of all consultations).
Table 2.
Demographic information and the five most common diagnoses. Values are numbers (percentages) of patients
| Total (n=1292) | Nurse practitioners (n=641) | General practitioners (n=651) | |
|---|---|---|---|
| Age (years): | |||
| >16 | 866 (67.0) | 414 (64.6) | 452 (69.4) |
| 5-15 | 200 (15.5) | 114 (17.8) | 86 (13.2) |
| <5 | 224 (17.3) | 112 (17.5) | 112 (17.2) |
| Sex: | |||
| Male | 547 (42) | 269 (42) | 278 (43) |
| Female | 743 (58) | 371 (58) | 372 (57) |
| Diagnosis*: | |||
| Upper respiratory tract infection | 475 (36.8) | 236 (36.8) | 239 (36.7) |
| Viral illness | 147 (11.4) | 81 (12.6) | 66 (10.1) |
| No specific diagnosis | 142 (11.0) | 76 (11.9) | 66 (10.1) |
| Minor injuries | 119 (9.2) | 70 (11.0) | 49 (7.5) |
| Eye and ear conditions | 98 (7.6) | 45 (7.0) | 53 (8.1) |
362 (28%) of patients had more than one diagnosis.
The nurse practitioners spent a mean of 11.57 minutes face to face with patients compared with 7.28 minutes by general practitioners (table 3). In addition, the nurses spent a mean of 1.33 minutes per patient in getting prescriptions signed. Table 4 shows that there was no significant difference in the percentage of patients who had a physical examination (nurse practitioners 88.1% v general practitioners 85.7%). Nurse practitioners issued fewer prescriptions than general practitioners, but the difference was not significant (391 (61.0%) v 421 (64.6%); odds ratio 0.88, 95% confidence interval 0.66 to 1.17). Nurses ordered more tests and investigations than general practitioners (56 (8.7%) v 37 (5.6%); 1.66, 1.04 to 2.66). In particular, the nurse practitioners carried out more tests associated with opportunistic screening such as urine testing and cervical screening. Nurse practitioners were also significantlymore likely to ask patients to return (37.2% v 24.8%; 1.93, 1.36 to 2.73). In81 (12.6%)consultations the nurse discussed the patient with a doctor, and in 26 (4.1%) consultations the patient was seen by the doctor.
Table 3.
Differences in care given at initial consultation (values adjusted for age, sex, and intracluster correlation)
| Nurse practitioners
|
General practitioners
|
Adjusted mean difference (95% CI) | P value | Intracluster correlation | ||||
|---|---|---|---|---|---|---|---|---|
| No of patients | Mean (SD) | No of patients | Mean (SD) | |||||
| Total consultation time (min) | 639 | 12.90 (6.28) | 639 | 7.49 (4.75) | 5.46 (4.16 to 6.78) | <0.001 | 0.19 | |
| Face to face consultation time* (min) | 639 | 11.57 (5.79) | 651 | 7.28 (4.80) | 4.20 (2.98 to 5.41) | <0.001 | 0.17 | |
| Mean number of physical examinations per patient | 612 | 2.28 (1.55) | 635 | 1.95 (1.57) | 0.19 (−0.03 to 0.71) | 0.072 | 0.19 | |
| Mean number of return visits | 638 | 0.49 (0.79) | 651 | 0.36 (0.66) | 0.14 (0.05 to 0.22) | 0.002 | 0.0† | |
Face to face consultation time = total consultation time minus time to have prescription signed or time to sign a prescription.
Negative estimate of intracluster correlation coefficients.
Table 4.
Differences in actions taken during consultation with nurse and general practitioners (values adjusted for age, sex, and intracluster correlation)
| Action | Nurse practitioner
|
General practitioner
|
Odds ratio (95% CI) | P value | Intracluster correlation | |||
|---|---|---|---|---|---|---|---|---|
| No of patients | % | No of patients | % | |||||
| Physical examination | 590/640 | 88.1 | 572/649 | 92.2 | 1.76 (0.90 to 3.42) | 0.097 | 0.10 | |
| Prescription given | 391/641 | 61.0 | 421/651 | 64.7 | 0.88 (0.66 to 1.17) | 0.375 | 0.03 | |
| Antibiotic prescribed | 195/641 | 30.4 | 207/651 | 31.7 | 0.94 (0.76 to 1.17) | 0.576 | 0.0* | |
| Investigation carried out | 56/641 | 8.7 | 37/651 | 5.6 | 1.66 (1.04 to 2.66) | 0.033 | 0.01 | |
| Hospital referral | 11/641 | 1.7 | 25/651 | 3.8 | 0.50 (0.16 to 1.63) | 0.250 | 0.08 | |
| Asked to return | 236/634 | 37.2 | 161/648 | 24.8 | 1.93 (1.36 to 2.73) | <0.001 | 0.05 | |
| Actually returned | 224/634 | 35.3 | 184/647 | 28.4 | 1.42 (1.18 to 1.71) | <0.001 | 0.0* | |
Negative estimate of intracluster correlation coefficient.
The satisfaction questionnaires (table 5) showed patients were more satisfied after consultations with nurses. Scores were significantly higher for the adult medical interview satisfaction scale scores and all its subscales and for the paediatric medical interview satisfaction scale scores in children and two of its subscales. There were no significant differences in enablement scores between the groups. The differences in satisfaction scores were still significant when the scores were additionally controlled for the length of face to face contact (mean difference 0.16, 95% confidence interval 0.08 to 0.24).
Table 5.
Differences in satisfaction after consultation (adjusted for age, sex, time, and intracluster correlation)
| Nurse practitioner
|
General practitioner
|
Adjusted mean difference (95% CI) | P value | Intracluster correlation | ||||
|---|---|---|---|---|---|---|---|---|
| Mean (SD) score | No of patients | Mean (SD) score | No of patients | |||||
| Adults | ||||||||
| Medical interview satisfaction scale | 4.40 (0.46) | 388 | 4.24 (0.52) | 390 | 0.18 (0.09 to 0.26) | <0.001 | 0.04 | |
| Communication | 4.35 (0.54) | 370 | 4.21 (0.60) | 384 | 0.13 (0.06 to 0.21) | 0.001 | 0.0* | |
| Distress relief | 4.43 (0.47) | 390 | 4.26 (0.57) | 400 | 0.19 (0.08 to 0.29) | 0.001 | 0.01 | |
| Clinician behaviour | 4.44 (0.49) | 375 | 4.22 (0.57) | 369 | 0.23 (0.15 to 0.32) | <0.001 | 0.2 | |
| Enablement score | 4.92 (3.62) | 335 | 4.43 (3.65) | 361 | 0.65 (−1.50 to 0.19) | 0.13 | 0.14 | |
| Children | ||||||||
| Medical interview satisfaction scale† | 4.39 (0.46) | 220 | 4.17 (0.57) | 181 | 0.23 (0.12 to 0.34) | <0.001 | 0.3 | |
| Communication with parent | 4.58 (0.51) | 223 | 4.48 (0.65) | 190 | 0.07 (−0.26 to 0.16) | 0.159 | 0.00 | |
| Communication with child | 4.16 (0.63) | 176 | 3.67 (0.77) | 147 | 0.47 (0.29 to 0.67) | <0.001 | 0.11 | |
| Distress relief | 4.41 (0.53) | 222 | 4.21 (0.64) | 186 | 0.21 (0.08 to 0.34) | 0.002 | 0.04 | |
| Adherence intent | 4.47 (0.53) | 218 | 4.44 (0.53) | 185 | −0.01 (−0.10 to 0.08) | 0.817 | 0.0* | |
Negative estimate of intracluster correlation coefficient.
Paediatric version.
In the two weeks after the initial consultation, patients who had seen a nurse practitioner were more likely to make a return visit to the clinic (mean number of returns 0.49 v 0.36). These return visits were mainly to general practitioners because there were more general practitioners than nurses in the practices and many of the nurses worked part time. There were no differences in health status at the end of two weeks.
Table 6 shows health service costs. These include the basic salary costs of each health professional plus the costs of prescriptions, tests, referrals, and the cost of return consultations in the following two weeks. Since return consultations were not timed, we estimated that they lasted an average of 7 minutes for general practitioner consultations and 11.5 minutes for nurse practitioner consultations. There was no significant difference in the cost of care given by the nurse practitioners and the general practitioners. Further details of costings are available from us on request.
Table 6.
Difference in cost of consultation with nurse and general practitioners based on salary costs. Data are adjusted for age, sex, and intracluster correlation
| Nurse practitioner
|
General practitioner
|
Adjusted mean difference (95% CI) | P value | Intracluster correlation | ||||
|---|---|---|---|---|---|---|---|---|
| Mean (SD) cost (£) | Range | Mean (SD) cost (£) | Range | |||||
| Initial consultation: | ||||||||
| Total time | 11.71 (25.23) | 0.66-297.1 | 14.14 (29.62) | 0.78-246.5 | 2.17 (−1.18 to 5.51) | 0.204 | 0.009 | |
| Face to face time* | 11.29 (25.18) | 0.66-297.1 | 14.11 (29.63) | 0.79-246.5 | 2.58 (−0.73 to 5.89) | 0.127 | 0.009 | |
| Return consultation | 6.40 (21.20) | 0.0-219.3 | 6.56 (22.85) | 0.0-217.3 | -0.03 (−2.25 to 2.20) | 0.98 | 0.008 | |
| Total costs: | ||||||||
| Total time | 18.11 (33.43) | 0.66-2971 | 20.70 (33.43) | 0.78-300.6 | 2.33 (−1.62 to 6.28) | 0.247 | 0.0† | |
| Face to face time* | 17.69 (33.41) | 0.66-297.1 | 20.68 (33.41) | 0.78-300.6 | 2.73 (−1.20 to 6.66) | 0.173 | 0.0† | |
Face to face time = total consultation time minus time to get a prescription signed or time to sign a prescription.
Negative estimate of intracluster correlation coefficient.
Discussion
We have evaluated care given by nurse practitioners working as part of primary care teams alongside general practitioners. Our results do not therefore relate to nurse practitioners who are working independently. It is often assumed thatwhen nurses substitute for doctors, the same service is provided.21 However, the British literature suggests that the combination of nursing and medical skills provides a more comprehensive and flexible service for patients than that provided by doctors.22 Our study provides limited support for this theory.
In many respects the behaviour of the nurses was similar to that of general practitioners, but some important differences existed. Nurse practitioners spent more time with patients and were more likely to ask patients to return. There were no differences in health outcome, although the study did not have sufficient power to detect a difference in rare serious events.
The differencesin working stylesbetween nurse practitioners and general practitioners are shown by the number and types of tests ordered and the numbers of patients who were asked to return to surgery. Nurse practitioners carried out more opportunistic screening. This was also found in a comparative study in the United States.7
In Britain only two groups of nurses (district nurses and health visitors) are able to prescribe drugs, and then from a limited list of items. During training nurse practitioners do an extensive pharmacology module with supporting modules in pathophysiology and disease management, and they argue that they are able to use the same range of drugs as doctors.23 We found that the nurse practitioners had similar prescribing behaviour to the general practitioners. As they had been qualified for some time, unlike nurse practitioners in other British studies,4,22,24 and were experienced nurses, this finding is not unexpected. These nurses were working in teams alongside general practitioners and consistent prescribing behaviour should, in theory, be adopted by all practice staff. Indeed, some practices had developed specific prescribing protocols for both general practitioners and nurse practitioners.
Patient satisfaction is an important component of nearly all studies looking at the role of nurse practitioners, and patients generally report high levels of satisfaction with nurse practitioner care.4,5,22,24,25 Increased satisfaction has been linked with longer consultations, and nurse practitioners have been shown to spend longer with patients than general practitioners. However, the differences in satisfaction remained in our study after we had controlled for differences in consultation time.
The health service costs of consultation with nurse practitioners were 12.5% lower than those for general practitioners, but this difference was not significant. However, a larger study with greater power to detect cost differences is needed. We were unable to do power calculations for cost before the study because none of the British studies of nurse practitioners have compared cost of consultations for general practitioners and nurse practitioners.4,5,22,24In the United States studies have shown conflicting results.25 Nurses were paid less than the general practitioners, but they took longer to see patients and more of their patients returned for further consultations. This reduced the overall difference in consultation costs. If lifetime training costs were included the general practitioner costs would be higher.
Conclusion
Our results relate to patients requesting a same day appointment in general practice and cannot necessarily be generalised to other situations. Overall, the clinical care and outcome were similar for nurse practitioners and general practitioners. Patients who requested a same day appointment were satisfied with nurse practitioner consultations. If nurse practitioners were able to work in different ways—for example, to shorten their consultation times (which our results suggest will not alter higher patient satisfaction with nurse practitioners) or reduce their return consultation rate—they could be more cost effective than general practitioners for this group of patients.
What is already known on this topic
Nurse practitioners are increasingly used as point of first contact in primary care
Observational studies suggest patients react positively to use of nurse practitioners
What this study adds
Patients were more satisfied with nurse practitioner consultations even after their longer consultation times were allowed for
There were few differences in clinical care and no difference in clinical outcome between nurse practitioners and general practitioners
Health service costs were not significantly different between nurses and general practitioners
Figure.
Flow chart tracking patients through study
Acknowledgments
We thank all the patients who participated in the study and the administrative, nursing, and medical staff in the 20 practices recruited for the study.
Footnotes
Funding: Wellcome Trust, project number 045384.
Competing interests: None declared.
References
- 1.Royal College of Nursing. Nurse practitioners in primary health care—role definition. London: RCN; 1989. [Google Scholar]
- 2.Department of Health. New opportunities for NHS Direct. London: DoH; 1999. . (Press release 1999/0227.) [Google Scholar]
- 3.Department of Health. Up to £30 million to develop 20 NHS fast access walk-in centres. London: DoH; 1999. . (Press release 1999/0226.) [Google Scholar]
- 4.South Thames Regional Health Authority. Evaluation of nurse practitioner pilot projects: summary report. London: Touche Roche Management Consultants, STRATA; 1994. [Google Scholar]
- 5.University of Newcastle Upon Tyne, Centre for Health Services Research. Evaluation of nurse practitioners in general practice in Northumberland: the EROS projects 1 and 2. Newcastle Upon Tyne: CHSR; 1998. [Google Scholar]
- 6.Spitzer WO, Sackett DL, Sibley JC, Roberts RS, Gent M, Kergin DJ, et al. The Burlington randomised controlled trial of the nurse practitioner. N Engl J Med. 1974;290:251–256. doi: 10.1056/NEJM197401312900506. [DOI] [PubMed] [Google Scholar]
- 7.Diers D, Hamman A, Molde S. Complexity of ambulatory care: nurse practitioner and physician caseloads. Nurs Res. 1986;35:310–314. [PubMed] [Google Scholar]
- 8.Brown SA, Grimes D. A meta-analysis of nurse practitioners and nurse midwives in primary care. Nurs Res. 1995;44:332–338. [PubMed] [Google Scholar]
- 9.Brazier JE, Harper R, Jones N, O'Cathain A, Thomas K, Usherwood T, et al. Validating the SF-36 health survey questionnaire: new outcome measure for primary care. BMJ. 1992;305:160–164. doi: 10.1136/bmj.305.6846.160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Landgraf J, Maunsell E, Speechley KN, Bullinger M, Campbell S, Ware J. Canadian, French, German and UK versions of the child health questionnaire: methodology and preliminary item scaling results. Qual Life Res. 1998;7:433–445. doi: 10.1023/a:1008810004694. [DOI] [PubMed] [Google Scholar]
- 11.McKinley RK, Cragg DK, Hastings AM, French DP, Manku-Scott TK, Campbell S, et al. Comparison of out of hours care provided by patients' own general practitioners and commercial deputising services: a randomised controlled trial. 2. Outcome of care. BMJ. 1997;314:190–193. doi: 10.1136/bmj.314.7075.190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wolf MH, Putnam SM, James SA, Stiles WB. The medical interview satisfaction scale: development of a scale to measure patient perceptions of physician behaviour. J Behav Med. 1978;1:391–401. doi: 10.1007/BF00846695. [DOI] [PubMed] [Google Scholar]
- 13.Lewis C, Scott D, Pantell R, Wolf M. Patient satisfaction with children's medical care: development, field test and validation of a questionnaire. Med Care. 1986;24:209–215. doi: 10.1097/00005650-198603000-00003. [DOI] [PubMed] [Google Scholar]
- 14.Howie JGR, Heaney D, Maxwell M, Walker JJ. A comparison of the patient enablement instrument (PEI) against two established satisfaction scales as an outcome measure of primary care consultations. Fam Pract. 1998;15:165–171. doi: 10.1093/fampra/15.2.165. [DOI] [PubMed] [Google Scholar]
- 15.Ware J. SF-36 health survey: manual and interpretation guide. Boston: Health Institute, New England Medical Center; 1993. [Google Scholar]
- 16.Netten A, Knight J, Dennett J, Cooley R, Slight A. A ready reckoner for staff costs in the NHS. Volume 1. Estimated unit costs. Canterbury: Personal Social Services Research Unit, University of Kent; 1998. [Google Scholar]
- 17.Roberts C. The implication of variation in outcome between health professionals for the design and analysis of randomised controlled trials. Stat Med (in press). [DOI] [PubMed]
- 18.Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics. 1986;42:121–130. [PubMed] [Google Scholar]
- 19.Stata Corporation. Stata statistical software, release 5.0. College Station, TX: Stata Corporation; 1997. [Google Scholar]
- 20.Barber JA, Thompson SG. Analysis and interpretation of cost data in randomised controlled trials: review of published studies. BMJ. 1998;317:1195–1200. doi: 10.1136/bmj.317.7167.1195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Richardson G, Maynard A. Fewer doctors? More nurses? A review of the knowledge base of doctor-nurse substitution. York: Centre for Health Economics, University of York; 1995. [Google Scholar]
- 22.NHS Executive. Nurse practitioner evaluation project: final report. Uxbridge: Coopers and Lybrand; 1996. [Google Scholar]
- 23.Mayes M. A study of prescribing patterns in the community. Nursing Standard. 1996;10(29):34–37. doi: 10.7748/ns1996.04.10.29.34.c6220. [DOI] [PubMed] [Google Scholar]
- 24.Reveley S. The role of the triage nurse practitioner in general medical practice: an analysis of the role. J Advan Nurs. 1998;28:584–591. doi: 10.1046/j.1365-2648.1998.00685.x. [DOI] [PubMed] [Google Scholar]
- 25.Office of Technology Assessment. Nurse practitioners, physician assistants and certified midwives: a policy analysis. Washington DC: US Government Printing Office; 1986. . (Health technology case study 37, OTA-HCS-37.) [Google Scholar]