30 years of pharmaceutical cost-utility analyses: growth, diversity and methodological improvement.

To review and critically evaluate published cost-utility analyses (CUAs) pertaining to pharmaceuticals for the past 3 decades. We examined data from the Tufts Medical Center Cost-Effectiveness Analysis Registry (www.cearegistry.org), which contains detailed information on English-language CUAs and their ratios (in $US, year 2008 values) published in peer-reviewed journals. We summarized study features using descriptive statistics for articles published from 1976 to 2006. Changes in study methodology over time were analysed by trend test. Analysis of ratios was restricted to those published from 2000 to 2006 from studies that correctly discounted future costs and benefits. Factors associated with having a favourable value (defined to be more than the median for all included ratios) were identified by logistic regression. Of 1393 CUAs published through 2006, 640 (45.9%) pertained to pharmaceuticals. The proportion of CUAs that focussed on pharmaceuticals increased from 34% for the period 1990-5 to 47% for the period 2001-5. Investigations with a US perspective accounted for 51% of all CUAs, although this proportion has decreased over time. The UK perspective investigations accounted for nearly 16% of all studies, and this portion has increased over time. About 24% of all CUAs were sponsored by industry, 48% were sponsored by non-industry sources, and 28% did not disclose their funding. Adherence to good methodological practices is roughly similar for studies with industry and non-industry sponsorship. Adherence to these practices has increased over time. Among the 1969 ratios meeting our inclusion criteria, the median value was $US22 000 per QALY. Logistic regression revealed that, while controlling for the intervention category (e.g. pharmaceutical, medical device, screening), ratios were more likely to be favourable if they were from studies sponsored by a pharmaceutical or device manufacturer (OR 1.53; 95% CI 1.07, 2.19). Ratios for pharmaceutical CUAs were less favourable than other ratios while controlling for sponsorship (OR 0.66; 95% CI 0.44, 0.98). The number of published pharmaceutical CUAs has grown steadily and accounts for almost half of all published CUAs. Adherence to good methodological practices does not appear to differ by study sponsor. Ratios from industry-sponsored studies are more favourable than other ratios. The results highlight that there are many opportunities for efficient healthcare investment, among pharmaceutical and non-pharmaceutical interventions, just as there are many investments that are inefficient.