Acute Myeloid Leukaemia Treatment in Adults in Norway Five Year Survival and Costs

Background: Acute myeloid leukaemia (AML) is a cancer that affects approximately 173 patients per year in Norway with a median age of 69 years at diagnosis. Many patients will not tolerate intensive chemotherapy with curative potential and such treatment is usually given to patients below 70 years. AML is characterized by an abnormal growth of the white blood cells in the bone marrow, which causes anaemia, thrombocytopenia and infections. Intensive treatment with curative potential requires hospitalisation over several months with the potential for life-threatening complications, such as opportunistic infections. About one-third of the patients are eligible for stem cell transplantation. Improving treatments strategies involves understanding the clinical pathways and identifying the associated costs. We conducted a retrospective analysis in Norway based on a similar study from the UK [1]. Similar survival and cost studies on AML have not been performed in Norway.

Objective: The aim of this study was to investigate the life expectancy and costs associated with treating AML in order to provide a representation of the Norwegian treatment regime by adapting a model developed for AML in the UK.

Methods: A probabilistic model combining a decision tree for the induction treatment and Markov models for remissions, relapses and transplantation was developed to conduct the study. The outcome was life expectancy and expected costs per patient in a 5-year perspective. Life expectancy and costs were estimated for two age groups (16-59 years and ≥60 years) and three different initial treatment regimens (Ara-C+Daunorubicin, Ara-C+Idarubicin and Other) and three responses (response, no response and early death). In the Markov models patients were distributed according to whether they achieved remission, had relapse of leukaemia, received transplantation or palliative care. Transition probabilities were estimated by means of Weibull regressions using individual level data from Oslo University Hospital - Rikshospitalet. The data set contained 307 patients diagnosed and treated between 2000 to 2014. Not all regressions were significant. Still, the insignificant regressions were included by the argument of a small cohort. Information on costs of treatment was partly based on individual level information for patient included in the sample (such as chemotherapy, blood usage, antibiotics and length of stay) and average cost estimates, such as transplantation and cost of care per day according to departments at the hospital.

Results: The total mean five year discounted cost and life expectancy was $ 199 828 and 31.87 months, ranging from $ 172 303 to $ 241 727 and 25.37 to 35.63 months per patient, respectively, depending on treatment regime and response. For young patients responding to treatment the mean five year discounted cost and life expectancy was $ 180 702 and 33.53 months, ranging from $ 137 824 to $ 220 896 and 22.14 to 38.77 months per patients. For elderly patients responding to treatment the mean cost and life expectancy was $ 120 711 and 18.77, ranging from $ 72 547 to $ 170 459 and 11.43 to 30.05. This implies heterogeneity in the AML population. The results indicate a better life expectancy and higher costs for young, compared to elderly patients. The model was validated by the use of face, internal, external and cross-validation methods. The results showed a fairly good fit with the empirical overall survival based on numbers from the Norwegian Cancer Registry, and the structure of the model corresponded well with the model developed for the UK.

Conclusions: AML life expectancy and costs vary according to clinical pathways and patient characteristics. Both survival and costs may be influenced by the inclusion of new molecular diagnostic tools and new treatment. This AML model may be used to evaluate treatments and enable policy makers to initiate informed decisions.

1. Wang, H.I., et al., Long-term medical costs and life expectancy of acute myeloid leukemia: a probabilistic decision model. Value Health, 2014. 17 (2): p. 205-14.