Context Pharmacoeconomic analyses are being used increasingly as the basis for
reimbursement of the costs of new drugs. Reports of these analyses are often
published in peer-reviewed journals. However, the analyses are complex and
difficult to evaluate.
Objective To describe the nature of problems encountered in the evaluation and
interpretation of pharmacoeconomic analyses used as a basis for reimbursement
Data Sources All major submissions to the Department of Health and Aged Care (DHAC)
by the pharmaceutical industry for funding made under the Australian Pharmaceutical
Benefits Scheme. Specifically, the DHAC's database of submissions that were
received between January 1994 and December 1997 were reviewed.
Study Selection Of a total of 326 submissions, 218 had serious problems of interpretation
and were included in the analysis. The nature of the serious problems reviewed
were classified as estimates of comparative clinical efficacy, comparator
issues, modeling issues, and calculation errors.
Data Extraction All submissions in the DHAC's database were reviewed and data were extracted
if both the DHAC evaluators and technical subcommittee considered problems
to have a significant bearing on the decisions of the parent committee.
Data Synthesis Of a total of 326 submissions, 218 (67%) had significant problems and
31 had more than 1 problem. Of the 249 problems identified, 154 (62%) related
to uncertainty in the estimates of comparative clinical efficacy, and 71 (28.5%)
related to modeling issues, which included clinical assumptions or cost estimates,
used in the construction of the economic models. There were 15 instances of
disagreement over the choice of comparator, and serious calculation errors
were found on 9 occasions. Overall, 159 problems (64%) were considered to
Conclusions Significant problems were identified in these pharmacoeconomic analyses.
The intensive evaluation process used in the Australian Pharmaceutical Benefits
Scheme allowed for identification and correction of pharmacoecomomic analysis
problems, but the resources that are required may be beyond the capacity of
many organizations, including peer-reviewed journals.