Analytical approach:
A decision-analytic Markov model was used to simulate a patient's expected transitions through relevant health states after receiving one of the interventions. Progression rates through the model states were informed by published data. The authors’ stated the perspective of the study was from the Australian healthcare system. The time horizon was 20 years.
Effectiveness data:
The main clinical outcomes included axillary recurrence, arm lymphoedema and overall survival. A literature review was undertaken. Six randomised trials that compared SLNB and ALND were identified as relevant as they reported one or more the above clinical endpoints. The model relied mostly on two randomised controlled trials (see Other Publications of Related Interest), the Australian Sentinel Node Axillary Clearance (SNAC) trial and the NSABP-32 trial. NSABP-32 provided the risk of recurrence and involved 5,611 recruited patients. SNAC trial data were extracted for proportions of patients who developed arm lymphoedema, failure rates of SLNB and SLNB false negative findings.
Monetary benefit and utility valuations:
Utilities were measured on a breast cancer population using the standard gamble approach (Karnon J et al. 2009, see Other Publications of Related Interest). Assumptions were made for disutility of moderate-to-severe arm lymphoedema following ALND or SLNB.
Measure of benefit:
The measure of benefit used was quality-adjusted life years (QALYs). Future benefits were discounted at a rate of 5% in line with Australian Government recommendations.
Cost data:
Direct medical costs included those related to the ALND and SLNB surgical procedures, hospitalisation, lymphoscintigraphy and lymphotropic dye injection, initial surgery, adjuvant therapies and long-term management of breast cancer (such as outpatient visits and pharmaceutical costs). Resources were valued using the Australian Revised Diagnosis Related Group cost weights adjusted for length of stay data from the SNAC trial data, Medicare and Pharmaceutical Benefits Schedules. Prices were presented in Australian dollars ($) for 2008 and discounted by 5% per annum.
Analysis of uncertainty:
The model parameters on the probabilities of treatment, outcomes and utility combinations were examined with one-way and multi-way sensitivity analyses. Two-way sensitivity analyses were presented in diagrams. The study undertook analyses considering high-risk women (risk of metastases of 50%). A willingness-to-pay threshold of $50,000 was used for the analyses.