Clinicians' adherence versus non adherence to practice guidelines in the management of patients with sarcoma: a cost-effectiveness assessment in two European regions

Our starting point was 327 sarcoma patients aged ≥ 15 years (254 in Veneto and 73 in Rhône-Alpes) diagnosed over the relevant periods in the two regions. Absence of patient consent, care undertaken outside the participating regions or in private hospitals, and missing records (Table 1) reduced the number of patients included in the study to 219, 58 from Rhône-Alpes and 161 from Veneto. These patients were followed retrospectively for the three years after sarcoma diagnosis or until the date of death. All patients had histological confirmation of primary malignant sarcoma, with or without distant metastasis at initial diagnosis. With the exception of osteosarcoma, sarcomatoid carcinoma, mesothelioma, neuroblastoma, paraganglioma and mixed (epithelial and mesenchymal) tumours of the female genital tract, all histological subtypes were included. All patients in Rhône-Alpes had been diagnosed between March 2005 and February 2006 and treated at the University Hospital of Lyon and/or at the Léon Bérard Cancer Centre. All patients in Veneto had been diagnosed between January 2007 and December 2007 and treated in the public hospitals of the region. Patients were managed in accordance with the ethical principles for medical research involving human subjects described in the Declaration of Helsinki. The study received approval in France from the National Ethics Committee (N°904073) and the National Committee for Protection of Personal Data (N°05-1102), and from the Local Sanitary Agency of the Veneto Region and the Ethics Committee of the Azienda Ospedaliera di Padova (N°156/06/CE) in Italy. Each patient was required to give signed informed consent. Lack of informed consent and treatment of sarcoma outside the Veneto or Rhône-Alpes were exclusion criteria. Cases of relapsed disease were also excluded since CPGs for this setting are not available.

Hospital records were used to obtain data on the characteristics of patients (age, sex, comorbidities) and their sarcomas (visceral or soft tissue; superficial or deep tumour); localization in lower or upper limb, head-neck or trunk; histological subtype; major tumour diameter at imaging and surgery; and grade). Data were also obtained on resources utilization (using a micro-costing approach) at diagnosis and during surgery (primary and wide surgical resection), chemotherapy (drugs administered in hospital or outpatient facilities) and radiotherapy (sessions in hospital or outpatient facilities). We also obtained data covering relapse or metastasis during follow up, along with any subsequent need for surgical interventions, chemotherapy and/or radiotherapy. Information was collected on number of days of hospitalization, use of pathology resources such as micro-biopsy and cytology), use of imaging, and supportive treatments such as antibiotics. The date of relapse was used to calculate relapse free survival from diagnosis.

Physicians independent of the study (two from the Léon Bérard Cancer Centre and two from the University of Padua) assessed whether or not there had been compliance with CPGs in each management phase covered by such guidelines. Overall management was considered to have been compliant only when CPGs had been followed at all stages of diagnosis, treatment, and follow-up.

Costs were assessed for each patient and from the hospital's point of view for the period between diagnosis and the end of follow-up or death [18]. Sources of unit costs and prices are described in Table 2[19‐21]. Days of hospital admission were multiplied by the cost per day (taking the average of the 2009 costs in France and Italy). This covered the cost of personnel, medications (except for chemotherapy and blood transfusions), use of medical devices, laboratory tests, depreciation of equipment and overheads. Doses of chemotherapy, and number of transfusions, radiotherapy sessions, imaging procedures, biopsies and consultations were multiplied by their respective unit costs (again taking the average of 2009 prices in France and Italy). Discounting of 4% per year was applied. The mean costs were calculated for patients whose overall management had been compliant with CPGs and for patients in whom it had not.

Chi-square or t-tests were used, according to the type of data, to compare compliant (CPG+) and non compliant (CPG-) groups. Analysis of variance (ANOVA) was used to explore the costs of overall management according to histological subtype. Univariate survival analyses were performed using the Kaplan-Meier method and the log-rank test. The log-rank test was used to assess the effect of compliance with CPGs on survival after adjusting for confounding variables.

The incremental cost-effectiveness ratio (ICER), expressing the incremental cost per additional relapse-free year gained.

The uncertainty surrounding the ICER was captured by a probabilistic sensitivity analysis according to procedures established by the French National Authority for Health (HAS) [22]. Ten thousand replications were obtained by the non parametric bootstrap method. A graphical representation of the sampling uncertainty associated with the ICER on the cost-effectiveness plane is shown in Figure 1[23]. The four quadrants of the cost-effectiveness plane are as follows: northeast, compliance with CPGs more costly and more effective than non compliance; southeast, compliance less costly, more effective; northwest (more costly, less effective) and southwest (less costly, less effective). Confidence regions were assessed and are represented by ellipses.

The outer ellipse defines the confidence region for the mean cost difference and mean effect difference pair at the 95% level and the inner ellipse at the 50% level. The outer ellipse is equivalent to an acceptability region for an inference test at 5% significance level whose null hypothesis is the mean cost difference and mean effect difference pair being equal to (0, 0). Consequently, if the origin of the cost-effectiveness plane does not belong to the confidence ellipse, then both the management of patients with sarcoma (adherence versus non adherence to CPGs) are significantly different regarding costs and effects. For more robustness in the results, in addition to confidence ellipses, uncertainty around the ICER was taken into account by examining probabilities that it belonged to each of the quadrants of the cost-effectiveness plane.

Calculations were performed using STATA 11 and Gauss software version 9.0.