Interventions:
The interventions were appropriately selected to compare the actual and the optimal levels of treatment adherence. The authors included the four most commonly prescribed osteoporosis treatments in Sweden.
Effectiveness/benefits:
No systematic review was reported to identify the clinical inputs, but the epidemiological data were appropriately selected from Swedish registries. The treatment effect was from clinical trials and the natural history of disease was from a large prospective study. These were appropriate study designs for each model parameter. Most of the clinical data were incorporated in a published model. QALYs were an appropriate benefit measure, and they capture the impact of the disease on survival and quality of life, both of which are relevant for patients with osteoporosis. Limited information was provided on the sources used to derive the utility values and on the instruments used to elicit them; it was not clear whose values (patients, experts, or the general population) were used.
Costs:
All costs associated with osteoporosis were included irrespective of who paid them, which was in accordance with the societal perspective, as stated by the authors. The sources are likely to have reflected the Swedish context, but were not fully described. The total cost of each fracture was reported, but the unit costs and resource quantities were not reported separately limiting the reproducibility and the transparency of the analysis. In general, the costs were not varied in the sensitivity analysis. Other details, such as the price year and currency conversion, were provided.
Analysis and results:
The results were extensively presented and were reported for men and women separately, as well as for age subgroups. The uncertainty was partly investigated in sensitivity analyses that considered variations in a few inputs. The authors stated that no discounting was applied to costs and benefits as the study focused on the assessment of the annual burden of disease. Little information on the decision model was provided, as it was based on a published model. The authors stated that the burden of disease might have been underestimated as they focused only on persistence of treatment and not incomplete adherence. The analysis focused on the Sweden, with a high incidence of osteoporotic fractures, and cannot be easily transferred to other settings.
Concluding remarks:
Some key assumptions were made, but the methods were valid and the authors’ conclusions appear to be robust.