A model for analysing the cost of autologous peripheral blood progenitor cell (PBPC) transplantation

Data from autologous peripheral blood progenitor cell (PBPC) transplant recipients were used for cost analysis and modelling so as to link the main intervention procedures and clinical events to resource use and costs. This cohort consisted of 64 patients from 4 to 62 years old at transplantation (mean, 36.9 years) who underwent a first transplant between August 1994 and May 1997. The main indications for transplantation were non-Hodgkin's lymphomas (47%), multiple myeloma (30%) and Hodgkin's lymphomas (15%). The course of a patient during the whole transplant procedure was modelled using a Markov chain of six states of health: (1) mobilisation and recovery of PBPC; (2) post-mobilisation phase; (3) conditioning and transplant; (4) critical haematological reconstitution; (5) non-critical haematological reconstitution; (6) death. The probability of transition between the different health states, together with the estimated costs, were the input for the Markov model. The model also managed transition probabilities depending both on the current health state and on various demographic, clinical and procedure-related covariates unique to the patient. The expected time spent in each clinical state and the expected total cost were, therefore, estimated. This analysis gave an actual total cost per transplanted patient of $26,600 (95% range: $24,700 to $43,500) while mean duration was 197 days. The expenses for in-hospital stay accounted for 80% of the costs. Both the probability of staying in the different states, and the consequent cost were dependent on the number of CD34-positive cells collected, the phase and the type of the disease, the subset of patients (either children or adults), and the post-transplant G-CSF prophylaxis. The sensitivity of the estimates to alternative assumptions was studied, and the method of comparing alternative future scenarios by the model was explored.