Frequentist rules for regulatory approval of subgroups in phase III trials: A fresh look at an old problem.

BACKGROUND: The number of Phase III trials that include a biomarker in design and analysis has increased due to interest in personalised medicine. For genetic mutations and other predictive biomarkers, the trial sample comprises two subgroups, one of which, say B+ is known or suspected to achieve a larger treatment effect than the other B-. Despite treatment effect heterogeneity, trials often draw patients from both subgroups, since the lower responding B- subgroup may also gain benefit from the intervention. In this case, regulators/commissioners must decide what constitutes sufficient evidence to approve the drug in the B- population. METHODS AND RESULTS: Assuming trial analysis can be completed using generalised linear models, we define and evaluate three frequentist decision rules for approval. For rule one, the significance of the average treatment effect in B- should exceed a pre-defined minimum value, say ZB->L. For rule two, the data from the low-responding group B- should increase statistical significance. For rule three, the subgroup-treatment interaction should be non-significant, using type I error chosen to ensure that estimated difference between the two subgroup effects is acceptable. Rules are evaluated based on conditional power, given that there is an overall significant treatment effect. We show how different rules perform according to the distribution of patients across the two subgroups and when analyses include additional (stratification) covariates in the analysis, thereby conferring correlation between subgroup effects. CONCLUSIONS: When additional conditions are required for approval of a new treatment in a lower response subgroup, easily applied rules based on minimum effect sizes and relaxed interaction tests are available. Choice of rule is influenced by the proportion of patients sampled from the two subgroups but less so by the correlation between subgroup effects.

The contribution of the anaesthetist to risk-adjusted mortality after cardiac surgery.

It is widely accepted that the performance of the operating surgeon affects outcomes, and this has led to the publication of surgical results in the public domain. However, the effect of other members of the multidisciplinary team is unknown. We studied the effect of the anaesthetist on mortality after cardiac surgery by analysing data collected prospectively over ten years of consecutive cardiac surgical cases from ten UK centres. Casemix-adjusted outcomes were analysed in models that included random-effects for centre, surgeon and anaesthetist. All cardiac surgical operations for which the EuroSCORE model is appropriate were included, and the primary outcome was in-hospital death up to three months postoperatively. A total of 110 769 cardiac surgical procedures conducted between April 2002 and March 2012 were studied, which included 127 consultant surgeons and 190 consultant anaesthetists. The overwhelming factor associated with outcome was patient risk, accounting for 95.75% of the variation for in-hospital mortality. The impact of the surgeon was moderate (intra-class correlation coefficient 4.00% for mortality), and the impact of the anaesthetist was negligible (0.25%). There was no significant effect of anaesthetist volume above ten cases per year. We conclude that mortality after cardiac surgery is primarily determined by the patient, with small but significant differences between surgeons. Anaesthetists did not appear to affect mortality. These findings do not support public disclosure of cardiac anaesthetists' results, but substantially validate current UK cardiac anaesthetic training and practice. Further research is required to establish the potential effects of very low anaesthetic caseloads and the effect of cardiac anaesthetists on patient morbidity.

Clinical effectiveness and cost-effectiveness of endobronchial and endoscopic ultrasound relative to surgical staging in potentially resectable lung cancer: results from the ASTER randomised controlled trial.

OBJECTIVE: To assess the clinical effectiveness and cost-effectiveness of endosonography (followed by surgical staging if endosonography was negative), compared with standard surgical staging alone, in patients with non-small cell lung cancer (NSCLC) who are otherwise candidates for surgery with curative intent. DESIGN: A prospective, international, open-label, randomised controlled study, with a trial-based economic analysis. SETTING: Four centres: Ghent University Hospital, Belgium; Leuven University Hospitals, Belgium; Leiden University Medical Centre, the Netherlands; and Papworth Hospital, UK. INCLUSION CRITERIA: known/suspected NSCLC, with suspected mediastinal lymph node involvement; otherwise eligible for surgery with curative intent; clinically fit for endosonography and surgery; and no evidence of metastatic disease. EXCLUSION CRITERIA: previous lung cancer treatment; concurrent malignancy; uncorrected coagulopathy; and not suitable for surgical staging. INTERVENTIONS: Study patients were randomised to either surgical staging alone (n = 118) or endosonography followed by surgical staging if endosonography was negative (n = 123). Endosonography diagnostic strategy used endoscopic ultrasound-guided fine-needle aspiration combined with endobronchial ultrasound-guided transbronchial needle aspiration, followed by surgical staging if these tests were negative. Patients with no evidence of mediastinal metastases or tumour invasion were referred for surgery with curative intent. If evidence of malignancy was found, patients were referred for chemoradiotherapy. MAIN OUTCOME MEASURES: The main clinical outcomes were sensitivity (positive diagnostic test/nodal involvement during any diagnostic test or thoracotomy) and negative predictive value (NPV) of each diagnostic strategy for the detection of N2/N3 metastases, unnecessary thoracotomy and complication rates. The primary economic outcome was cost-utility of the endosonography strategy compared with surgical staging alone, up to 6 months after randomisation, from a UK NHS perspective. RESULTS: Clinical and resource-use data were available for all 241 patients, and complete utilities were available for 144. Sensitivity for detecting N2/N3 metastases was 79% [41/52; 95% confidence interval (CI) 66% to 88%] for the surgical arm compared with 94% (62/66; 95% CI 85% to 98%) for the endosonography strategy (p = 0.02). Corresponding NPVs were 86% (66/77; 95% CI 76% to 92%) and 93% (57/61; 95% CI 84% to 97%; p = 0.26). There were 21/118 (18%) unnecessary thoracotomies in the surgical arm compared with 9/123 (7%) in the endosonography arm (p = 0.02). Complications occurred in 7/118 (6%) in the surgical arm and 6/123 (5%) in the endosonography arm (p = 0.78): one pneumothorax related to endosonography and 12 complications related to surgical staging. Patients in the endosonography arm had greater EQ-5D (European Quality of Life-5 Dimensions) utility at the end of staging (0.117; 95% CI 0.042 to 0.192; p = 0.003). There were no other significant differences in utility. The main difference in resource use was the number of thoracotomies: 66% patients in the surgical arm compared with 53% in the endosonography arm. Resource use was similar between the groups in all other items. The 6-month cost of the endosonography strategy was £9713 (95% CI £7209 to £13,307) per patient versus £10,459 (£7732 to £13,890) for the surgical arm, mean difference £746 (95% CI -£756 to £2494). The mean difference in quality-adjusted life-year was 0.015 (95% CI -0.023 to 0.052) in favour of endosonography, so this strategy was cheaper and more effective. CONCLUSIONS: Endosonography (followed by surgical staging if negative) had higher sensitivity and NPVs, resulted in fewer unnecessary thoracotomies and better quality of life during staging, and was slightly more effective and less expensive than surgical staging alone. Future work could investigate the need for confirmatory mediastinoscopy following negative endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) and endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA), the diagnostic accuracy of EUS-FNA or EBUS-TBNA separately and the delivery of both EUS-FNA or EBUS-TBNA by suitably trained chest physicians. TRIAL REGISTRATION: Current Controlled Trials ISRCTN 97311620. FUNDING: This project was funded by the NIHR Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 16, No. 18. See the HTA programme website for further project information.

The prevalence and association with transfusion, intensive care unit stay and mortality of pre-operative anaemia in a cohort of cardiac surgery patients.

Anaemia is increasingly prevalent in the United Kingdom. Despite recommendations to the contrary, many patients undergo cardiac surgery with undiagnosed and untreated anaemia. According to the World Health Organization definition, 1463/2688 (54.4%) patients undergoing cardiac surgery between 2008 and 2009 in our institution were anaemic. Compared with non-anaemic patients, anaemia was significantly associated with transfusion (791 (54.1%) vs 275 (22.4%), p < 0.001, OR (95% CI) 3.4 (2.8-4.1)), death (45 (3.1%) vs 13 (1.1%), p = 0.0005, OR 2.4 (1.2-4.5)), and prolonged ICU stay (287 (19.6%) vs 168 (13.7%) p < 0.001, OR 1.3 (1.0-1.6)). The prevalence of anaemia in this cohort is much greater than that previously reported. The cause of this excess is not clear. Pre-operative anaemia is a strong predictor of increased transfusion requirement, risk of ICU stay and death during cardiac surgery. The effect of increasing haemoglobin concentration therapeutically is not yet clear.

Expansion of the kidney donor pool by using cardiac death donors with prolonged time to cardiorespiratory arrest.

Donation after Cardiac Death (DCD) is an increasingly important source of kidney transplants, but because of concerns of ischemic injury during the agonal phase, many centers abandon donation if cardiorespiratory arrest has not occurred within 1 h of controlled withdrawal of life-supporting treatment (WLST). We report the impact on donor numbers and transplant function using instead a minimum 'cut-off' time of 4 h. The agonal phase of 173 potential DCD donors was characterized according to the presence or absence of: acidemia; lactic acidosis; prolonged (>30 min) hypotension, hypoxia or oliguria, and the impact of these characteristics on 3- and 12-month transplant outcome evaluated by multivariable regression analysis. Of the 117 referrals who became donors, 27 (23.1%) arrested more than 1 h after WLST. Longer agonal-phase times were associated with greater donor instability, but surprisingly neither agonal-phase instability nor its duration influenced transplant outcome. In contrast, 3- and 12-month eGFR in the 190 transplanted kidneys was influenced independently by donor age, and 3-month eGFR by cold ischemic time. DCD kidney numbers are increased by 30%, without compromising transplant outcome, by lengthening the minimum waiting time after WLST from 1 to 4 h.

Disease-specific survival benefit of lung transplantation in adults: a national cohort study.

The lung transplantation candidate population is heterogeneous and survival benefit has not been established for all patient groups. UK data from a cohort of 1997 adult (aged > or = 16), first lung transplant candidates (listed July 1995 to July 2006, follow-up to December 2007) were analyzed by diagnosis, to assess mortality relative to continued listing. Donor lungs were primarily allocated according to local criteria. Diagnosis groups studied were cystic fibrosis (430), bronchiectasis (123), pulmonary hypertension (74), diffuse parenchymal lung disease (564), chronic obstructive pulmonary disease (COPD, 647) and other (159). The proportion of patients in each group who died while listed varied significantly (respectively 37%, 48%, 41%, 49%, 19%, 38%). All groups had an increased risk of death at transplant, which fell below waiting list risk of death within 4.3 months. Thereafter, the hazard ratio for death relative to listing ranged from 0.34 for cystic fibrosis to 0.64 for COPD (p < 0.05 all groups except pulmonary hypertension). Mortality reduction was greater after bilateral lung transplantation in pulmonary fibrosis patients (p = 0.049), but not in COPD patients. Transplantation appeared to improve survival for all groups. Differential waiting list and posttransplant mortality by diagnosis suggest further use and development of algorithms to inform lung allocation.

Clinical and cost-effectiveness analysis of an open label, single-centre, randomised trial of spinal cord stimulation (SCS) versus percutaneous myocardial laser revascularisation (PMR) in patients with refractory angina pectoris: The SPiRiT trial.

BACKGROUND: Patients with refractory angina have significant morbidity. This study aimed to compare two of the treatment options, Spinal Cord Stimulation (SCS) and Percutaneous Myocardial Laser Revascularisation (PMR) in terms of clinical outcomes and cost-effectiveness. METHODS: Eligible patients were randomised to PMR or SCS and followed up for exercise tolerance time (ETT), Canadian Cardiovascular Society (CCS) classification and the quality of life measures SF-36, Seattle Angina Questionnaire and the EuroQoL at 3, 12 and 24 months. Utilities were calculated using the EQ-5D and these and costs were compared between groups. The incremental cost-effectiveness ratio (ICER) per QALY for SCS compared to PMR was also calculated. RESULTS: At 24 months post-randomisation, patients that had SCS and PMR had similar ETT (mean difference 0.05, 95% CI -2.08, 2.18, p = 0.96) and there was no difference in CCS classification or quality of life outcomes. The difference in overall mean costs when comparing SCS to PMR was GBP5,520 (95% CI GBP1,966 to GBP8,613; p < 0.01) and the ICER of using SCS was GBP46,000 per QALY. CONCLUSION: Outcomes after SCS did not differ appreciably from those after PMR, with the former procedure being less cost-effective as currently applied. Larger studies could clarify which patients would most benefit from SCS, potentially increasing cost-effectiveness. TRIAL REGISTRATION: Current Controlled Trials ISRCTN09648950.

Flexible random-effects models using Bayesian semi-parametric models: applications to institutional comparisons.

Random effects models are used in many applications in medical statistics, including meta-analysis, cluster randomized trials and comparisons of health care providers. This paper provides a tutorial on the practical implementation of a flexible random effects model based on methodology developed in Bayesian non-parametrics literature, and implemented in freely available software. The approach is applied to the problem of hospital comparisons using routine performance data, and among other benefits provides a diagnostic to detect clusters of providers with unusual results, thus avoiding problems caused by masking in traditional parametric approaches. By providing code for Winbugs we hope that the model can be used by applied statisticians working in a wide variety of applications.

Double-blind, randomised, controlled trial assessing controller medications in asthma.

BACKGROUND: The motive behind conducting this study was to determine if better control of asthma can be achieved by adding a second controller medication and to assess its use to reduce the dose of inhaled steroids. OBJECTIVES: The study aimed to determine whether either oral sustained-release theophylline or montelukast added to inhaled steroids improved clinical symptoms and pulmonary function test parameters when compared to high-dose steroids alone. METHODS: Ninety patients with incompletely controlled asthma were allocated, in a randomised, double-blind fashion, to one of three treatment groups: group A: double dose of inhaled budesonide (400 microg b.i.d.), group B: 400 mg oral sustained-release theophylline plus budesonide (200 microg b.i.d.) and group C: 10 mg montelukast plus budesonide (200 microg b.i.d.). The primary endpoints were forced expiratory volume in 1 s (FEV(1)) and mean morning peak expiratory flow rate (PEFR). RESULTS: All three groups had improved FEV(1) and PEFR at 8 weeks (p < 0.001). Group C increased their PEFR by 18.7 l/min (95% confidence interval, CI, 12.4-25.1) more than group A and by 19.8 l/min (95% CI 13.4-26.1) more than group B (both p = 0.001). Similarly, group C had a 114 ml (95% CI 45-183 ml) greater improvement in FEV(1) than group A and a 95 ml (95% CI 26-164 ml) greater improvement than group B (both p = 0.01). CONCLUSIONS: Addition of montelukast to budesonide is safe and results in greater improvement in pulmonary function test parameters than high-dose budesonide treatment or addition of theophylline.

Octogenarians undergoing cardiac surgery outlive their peers: a case for early referral.

OBJECTIVE: To examine short and long term outcomes of octogenarians having heart operations and to analyse the interaction between patient and treatment factors. METHODS: Multivariate analysis of prospectively collected data and a survival comparison with an age and sex matched national population. The outcomes were base in-hospital mortality, risk stratified by logistic EuroSCORE (European system for cardiac operative risk evaluation), and long term survival. RESULTS: 12,461 consecutive patients (706 over 80 years) operated on between 1996 and 2003 in a regional UK unit were studied. Octogenarians more often had impaired ventricular function, pulmonary hypertension, and valve operations. They also included a higher proportion of women, had a higher serum creatinine concentration, and had a trend towards more unstable angina. Younger patients had a higher prevalence of previous cardiac operation, previous myocardial infarction, and diabetes. The in-hospital mortality rate was 3.9% for all patients (EuroSCORE predicted 6.1%, p < 0.001) and 9.8% for octogenarians (predicted 14.1%, p = 0.002). Long bypass time and non-elective surgery increased the risk of death above EuroSCORE prediction in both groups. A greater proportion of octogenarians stayed in intensive care more than 24 hours (37% v 23%, p < 0.001). Long term survival was significantly better in the study patients than in a general population with the same age-sex distribution (survival rate at five years 82.1% v 55.9%, p < 0.001). CONCLUSIONS: Cardiac surgery in a UK population of octogenarians produced excellent results. Elective referrals should be encouraged in all age groups.

Bayesian evidence synthesis to extrapolate survival estimates in cost-effectiveness studies.

This paper is concerned with survival extrapolation that represents an integral part of cost-effectiveness analysis. In the absence of long-term survival estimates from randomized clinical trials or meta-analysis we show how age-sex matched U.K. population data can additionally be used to estimate survival patterns. We adopt a Bayesian approach and we synthesize evidence from different sources such as patient registries, U.K. population statistics and meta-analyses. We also present methodology for Bayesian analysis of the additive hazards model and we show how to apply the techniques using freely available software. The methods are illustrated using data from a cohort of cardiac arrhythmia patients.

Comparison of shuttle walk with measured peak oxygen consumption in patients with operable lung cancer.

BACKGROUND: The relationship between the shuttle walk test and peak oxygen consumption in patients with lung cancer has not previously been reported. A study was undertaken to examine this relationship in patients referred for lung cancer surgery to test the hypothesis that the shuttle walk test would be useful in this clinical setting. METHODS: 125 consecutive patients with potentially operable lung cancer were prospectively recruited. Each performed same day shuttle walking and treadmill walking tests. RESULTS: Shuttle walk distances ranged from 104 m to 1020 m and peak oxygen consumption ranged from 9 to 35 ml/kg/min. The shuttle walk distance significantly correlated with peak oxygen consumption (r = 0.67, p<0.001). All 55 patients who achieved more than 400 m on the shuttle test had a peak oxygen consumption of at least 15 ml/kg/min. Seventy of 125 patients failed to achieve 400 m on the shuttle walk test; in 22 of these the peak oxygen consumption was less than 15 ml/kg/min. Nine of 17 patients who achieved less than 250 m had a peak oxygen consumption of more than 15 ml/kg/min. CONCLUSION: The shuttle walk is a useful exercise test to assess potentially operable lung cancer patients with borderline lung function. However, it tends to underestimate exercise capacity at the lower range compared with peak oxygen consumption. Our data suggest that patients achieving 400 m on the shuttle walk test do not require formal measurement of oxygen consumption. In patients failing to achieve this distance we recommend assessment of peak oxygen consumption, particularly in those unable to walk 250 m, because a considerable proportion would still qualify for surgery as they had an acceptable peak oxygen consumption.

Extended ganciclovir prophylaxis in lung transplantation.

BACKGROUND: Positive cytomegaloviral status of the donor or of the recipient adversely affects survival and enhances the development of bronchiolitis obliterans syndrome (BOS) in lung transplant recipients. The role of ganciclovir prophylaxis in cytomegalovirus infection in respect to obliterative bronchiolitis or to BOS development is not known. METHODS: From the Papworth transplant database, we identified 146 patients who received organs from cytomegalovirus-positive donors. We classified patients into 3 groups as follows: Group 1 consisted of 42 patients who underwent transplantation between 1990 and 1992 when no prophylaxis was given; Group 2 consisted of 49 patients who underwent transplantation between 1992 and 1995 when 4 weeks of IV ganciclovir was given as prophylaxis; and Group 3 consisted of 55 patients who underwent transplantation between 1995 and 1998 when cytomegalovirus prophylaxis consisted of IV (1 week) followed by oral ganciclovir for a total of 3 months. Donor management, recipient management during and after surgery, and pharmacotherapy were uniform during the study period. We used survival and regression methods to compare these groups, adjusting for the transplantation type (single lung, double lung, or heart-lung) and for HLA typing. RESULTS: We found a significant difference among all 3 groups in numbers of cytomegaloviral disease episodes in the 1st year after transplantation. The number of rejection episodes in the 3 groups during the 1st post-transplant year gradually decreased from Group 1 to Group 3. We identified no statistically significant benefit in the time to BOS occurrence or in actuarial survival. CONCLUSION: Extended prophylaxis with IV and oral ganciclovir practically abolishes cytomegaloviral disease and is related to a decreased incidence of rejection episodes. However, ganciclovir prophylaxis is not related to a decreased incidence or progression of BOS or survival.

Inflammatory related changes in bone mineral content in adults with cystic fibrosis.

BACKGROUND: Proinflammatory cytokines stimulate osteoclast activity and this could lead to increased bone resorption in patients with cystic fibrosis. The aim of this study was to determine whether markers of systemic inflammation are related to changes in bone mineral content (BMC) in adults with cystic fibrosis. METHODS: Total body BMC was assessed by dual energy x ray absorptiometry in 100 patients (54 male) of mean (SD) age 25.6 (7.1) years and forced expiratory volume in 1 second (FEV(1)) 61.8 (24.1)% predicted on recruitment to the study and 1 year later. Blood was also taken at these time points to measure markers of systemic inflammation. RESULTS: After 1 year BMC had reduced by 16.1 (62.1) g, p = 0.01; (0.6 (2.8)%). The change in BMC was related to mean levels of interleukin (IL)-6 (r(s) = -0.39, p<0.001) and C reactive protein (r(s) = -0.34, p = 0.002), intravenous antibiotic use (r(s) = -0.27, p = 0.006) and oral corticosteroid use (r(s) = -0.20, p = 0.045). Urinary markers of osteoclast activity were also related to IL-6 (r(s) = 0.27, p = 0.02). Multiple linear regression revealed that IL-6 (coefficient -2.2 (95% CI -3.4 to -1.0) per pg/ml, p = 0.001), colonisation with Burkholderia cepacia (coefficient -46.8 (95% CI -75.5 to -18.1), p = 0.002), and annual change in BMI (coefficient 15.4 (95% CI 3.6 to 27.2) per kg/m(2), p = 0.011) were independently significant predictors of annual change in BMC. CONCLUSIONS: These data suggest a pathophysiological mechanism by which chronic pulmonary infection results in bone loss in patients with cystic fibrosis.

Performance status score: do patients and their oncologists agree?

Oncologists traditionally assess their patients' ECOG performance status (PS), and few studies have evaluated the accuracy of these assessments. In this study, 101 patients attending a rapid access clinic at Papworth Hospital with a diagnosis of lung cancer were asked to assess their own ECOG PS score on a scale between 0 and 4. Patients' scores were compared to the PS assessment of them made by their oncologists. Of 98 patients with primary non-small-cell lung cancer (NSCLC) and small-cell lung cancer (SCLC), weighted kappa statistics showed PS score agreement between patient and oncologist of 0.45. Both patient- and oncologist-assessed scores reflected survival duration (in NSCLC and SCLC) as well as disease stage (in NSCLC), with oncologist-assessed scores being only marginally more predictive of survival. There was no sex difference in patient assessment of PS scores, but oncologists scored female patients more pessimistically than males. This study showed that, with few exceptions, patients and oncologists assessed PS scores similarly. Although oncologists should continue to score PS objectively, it may benefit their clinical practice to involve their patients in these assessments.

The use of hospital admission data as a measure of outcome in clinical studies of heart failure.

AIM: To describe and evaluate the methods currently used to summarize, depict, and analyse hospitalization data in clinical studies of patients with heart failure. METHODS AND RESULTS: Studies published during 1999 and 2000 were reviewed if they reported hospitalization data collected during the follow-up of a defined group of adults who had been diagnosed as having heart failure. Of 130 studies identified, the majority considered only the first of any hospital admissions for each individual. When presented, statistical analyses in 70% of papers, and graphics in 75% of papers, were based purely on the first hospitalization. A minority of papers based analyses on the number of admissions (23%), or the amount of time spent in hospital by patients (14%) during the follow-up period. CONCLUSIONS: Data on the hospitalizations undergone by patients with heart failure are potentially informative about the disease burden to individuals and health services. The widespread practice of focussing a statistical analysis on the first admission a patient undergoes discards those aspects of hospitalization that are informative about disease burden: multiple admissions and length of stay. Statistical methods that incorporate these aspects appeared in a handful of papers, and the more widespread adoption of these is desirable.

A randomised controlled crossover trial of nurse practitioner versus doctor led outpatient care in a bronchiectasis clinic.

BACKGROUND: With the decrease in junior doctor hours, the advent of specialist registrars, and the availability of highly trained and experienced nursing personnel, the service needs of patients with chronic respiratory diseases attending routine outpatient clinics may be better provided by appropriately trained nurse practitioners. METHODS: A randomised controlled crossover trial was used to compare nurse practitioner led care with doctor led care in a bronchiectasis outpatient clinic. Eighty patients were recruited and randomised to receive 1 year of nurse led care and 1 year of doctor led care in random order. Patients were followed up for 2 years to ensure patient safety and acceptability and to assess differences in lung function. Outcome measures were forced expiratory volume in 1 second (FEV(1)), 12 minute walk test, health related quality of life, and resource use. RESULTS: The mean difference in FEV(1) was 0.2% predicted (95% confidence interval -1.6 to 2.0%, p=0.83). There were no significant differences in the other clinical or health related quality of life measures. Nurse led care resulted in significantly increased resource use compared with doctor led care (mean difference pound 1497, 95% confidence interval pound 688 to pound 2674, p<0.001), a large part of which resulted from the number and duration of hospital admissions. The mean difference in resource use was greater in the first year ( pound 2625) than in the second year ( pound 411). CONCLUSIONS: Nurse practitioner led care for stable patients within a chronic chest clinic is safe and is as effective as doctor led care, but may use more resources.