Introduction/Objective: If cancer care needs to be scientifically based, one would expect that priorities in basic science were set by their relevance to cancer care. We analysed whether the research effort per cancer site is in accordance with its priority for research. Thereto we use measures for (1) cancer research priority, and (2) cancer research output.

Methods: (1) The Cancer Research Priority (CaRP) is defined assuming a linear relationship between cancer incidence, mortality rate and the fraction of non-preventable deaths, using USA Cancer Statistics, 1997. CaRP is thus expressed as the number of inevitable US 'cancer deaths' in 1997. (2) The research effort per cancer site is assumed to be proportional the number of scientific articles in 1995 on each cancer site as indexed in Medline. The CaRP-per-Article (CaRP/A) was used as an indicator of research effort relative to the number of cancer deaths. The Cancer Research Adjustment Factor (CaRAF) was defined as the CaRP/A divided by the average CaRP/A for 'all cancer sites'. The CaRAF is a measure of the magnitude of imbalance between cancer research priority and research effort.

Results: The CaRP ranged from 1,000 (larynx) to 43,000 (colorectal) cancer deaths. The CaRP/A ranged from 0.25 (childhood) to 24 (pancreas) cancer-deaths per article, suggesting an almost 100-fold discrepancy in research effort. To meet the average research effort for all cancer sites (CaRAF=l), research efforts would need to be intensified > 50% for colorectal, pancreas, prostate, ovary and non-hodgkin lymphoma and could be diminished > 50% for childhood, Hodgkin's disease, melanoma, larynx, oral/pharynx and cervix uteri.

Discussion: The discrepancy between cancer research priorities and search output suggests that cancer research is unbalanced in addressing the relevant problems in clinical care and public health.