A qualitative study indicates that there is a positive selection bias towards favourable economic analysis of targeted therapies when these are funded by the manufacturer. At a time of increasing budgetary constraints and public scrutiny of the relationship between industry and the professions, we need a more mixed economy of funding for this field.
In terms of the history of medicine and health care, the 19th century may be regarded as the century of Public Health, clean water, sewerage and understanding the basis of infection; the 20th century might be regarded as the century of know-ledge, when systematic clinical and laboratory research yielded extraordinary insights into the mechanism of disease; we predict that the 21st century will be driven by value. Considering the spiralling costs of healthcare and an often confused approach to how we define value in a societal sense, and given the global financial crisis and the likelihood that for many nations the health budget will flatline, it is obvious that we need more data on the relative cost-effectiveness of innovative diagnostic or therapeutic agents if we are to make transparent and defensible judgements on their relative worth. This situation is set against a backdrop of increasing suspicion from policy and lawmakers and some patient groups that the relationship between practising clinicians and purveyors of these new technologies is not at sufficient arm’s length. In 2007, Djulbegovic et al., published a fascinating historical case study of the first conflicts of interest policy at the National Academy of Sciences. A fundamental debate in this case was whether one can simply declare a financial interest or whether one must also admit that this financial interest is a potential source of bias.
Now, a new study has been published by Valachis et al. that addresses this question in a different way. One of the characteristic points of the study is that the authors tried to investigate the role of manufacturers’ influence in various manifestations, such as the presence of any author affiliated with the manufacturer of the drug being assessed, or the presence of direct funding from the manufacturer for the health-economic study – as shown in previous studies – the role of funding and its bias in economic evaluation of drugs in oncology, and medical research in general. Of the 81 eligible studies that they identified, the authors found that economic analyses that were funded by pharmaceutical companies were more likely to report favourable qualitative cost estimates than those without an expressed funding association with these companies (28 out of 34 studies [82%] versus 21 of out of 47 studies [45%]; P=0.003). This phenomenon was seen to a similar degree for those studies that reported any financial relationship with the manufacturers, for example, author affiliation or author funding. Valachis et al. discuss the weaknesses inherent in their study with candour: the linkage between the eligible studies and their financial aspects depended solely on published details, as Valachis et al. made no effort to contact authors directly to further verify these data; there may have been a publication bias towards positive reports that might have skewed results; certain study criteria were poorly represented, so the database was rather small (for example, affiliation with manufacturers); and finally, their analysis was based on qualitative data. Nevertheless, Valachis et al. do seem to have demonstrated a consistent sponsorship bias towards the manufacturer of costly, targeted drugs with respect to economic analyses. It is concluded that the best way of dealing with perceptions of sponsorship bias is not increased rhetoric, but rather increased public funding for economic evaluation of medicines, thereby creating a true mixed economy for research funding in this field.
Does this sponsorship bias matter? If we are to adopt Michael Porter’s definition of value, then, yes it does.
“The way to deal with perceptions of sponsorship bias is not increased rhetoric, but increased public funding”
“Value in any field must be defined around the customer, not the supplier. Value must also be measured by outputs, not inputs. Hence it is patient health results that matter, not the volume of services delivered. But all outcomes are achieved at some cost. Therefore, the proper objective is … patient health outcomes relative to the total cost (inputs). Efficiency, as well as other objectives such as safety, is subsumed in the concept of value.”
Adoption of any new therapeutic agent in the current climate is likely to involve trade offs, comparing the value gained from the introduction of the targeted therapy relative to existing gold standards in cancer treatment, or, even more widely, comparing its value with that gained from hip replacements or cataract operations. The latter comparison might seem absurd, but within a finite health budget in which there is no ring-fencing of cancer funding, this could become an issue. So an economic evaluation of the new drug will have an often critical role in whether the drug is made available to cancer patients by governments or payers. If there are significant doubts about the veracity of the data, hanging over the analysis like the sword of Damocles, then this starts to undermine the validity of the data and even reduce the chances of a targeted therapy passing over whatever health-economic hurdles have been erected in its way.
So, is there a way to square this circle? In the same way that we now have mandatory listing of clinical trials to offset publication bias, one might establish a register of pharmacoeconomic studies; approaches might be made to journal editorial boards to lower their threshold for publishing negative studies; and payers could establish independently funded analytical units to give an entirely unbiased view of the economic case for acceptance or not of the agent under investigation. If the workings of these analytical units were utterly transparent and open to public review, then this would further enhance their credibility and relevance to citizens. Do we think that there is some methodical misrepresentation of results? Of course not; however, the paper by Valachis et al. is a timely warning of the subtle biases that can creep in unnoticed, and is perhaps doubly important given the wider economic challenges faced by all healthcare systems and, therefore, the increasing scrutiny that will be applied to all such economic analyses.
1. EG Campbell et al. (2007) A national survey of physician–industry relationships. NEJM 356:1742–50
2. B Djulbegovic, C Angelotta, KE Knox et al. (2007) The sound and the fury: financial conflicts of interest in oncology. JCO 25:3567–69
3. A Valachis, NP Polyzos, A, Nearchou et al. (2012) Financial relationships in economic analyses of targeted therapies in oncology. JCO 30:1316–20
4. S Jang, YK Chae, T Haddad et al. (2010) Conflict of interest in economic analyses of aromatase inhibitors in breast cancer: a systematic review. Breast Cancer Res Treat 121:273–279
5. M Barbieri, MF Drummond. (2010) Conflict of interest in industry-sponsored economic evaluations: real or imagined? Curr Oncol Rep 3:410–413
6. ME Porter. (2007) Defining and introducing value in health care. In: MB McClellan, JM McGinnis, EG Nabel et al (eds). Evidence-based medicine and the changing nature of health care: 2007 Institute of Medicine (IOM) annual meeting summary, 1st edn. National Academies Press, Washington DC, chapter 14, pp 161–171
7. R Sullivan et al. (2011) Delivering affordable cancer care in high-income countries. Lancet Oncol 12:933–980
8. US National Library of Medicine. (2012) http://www.clinicaltrials.gov
David Kerr, Nuffield Department of Clinical and Laboratory Sciences, University of Oxford, UK; Ahmed Elzawawy, Suez Canal University, Port Said, Egypt
This article was first published online in Nature Reviews Clinical Oncology on 1 May 2012, and is republished with permission. © 2012 Nature Publishing Group. doi:10.1038/nrclinonc.2012.75