On first receiving this book it occurred to me that the Chris Williams and his co-contributors were staging a last ditch effort to achieve the impossible—i.e. to make a general textbook in Oncology a relevant and useful tool in everyday practice. In rapidly changing fields like Oncology, general textbooks usually lapse at once into the "out-of-date" category. The early signs were promising with an accompanying CD inside the front cover (tantalising the reader with the possibility of accessing the content on a laptop computer) and the promise that regular updates would be posted at www.evidbasedoncology.com to incorporate new evidence when it becomes available.

The opening three chapters on "Appraising clinical literature in cancer", "Finding the best evidence for cancer care," and "Understanding the concepts behind health economics" were masterly summaries of their topics and by this stage the appetite was fully whetted and ready for more. Unfortunately what followed became steadily indigestible largely due to the age-old problem of contributions from multiple authors. Not surprising given that there were 94 contributors and 11 associate editors!

The authors were asked to use systematic searches of the literature and summarise their findings. Conclusions for each review question were graded according to the strength of evidence underlying that conclusion. It is unclear whether the authors were given a set of important clinical questions to address for a particular topic or were allowed to generate their own questions. I suspect the latter as the questions ranged from very broad (How is the diagnosis made? - Pancreatic Cancer) to very specific (Is a superficial groin node dissection as effective as a femoroinguinal groin node dissection? - Vulval Cancer). This in itself led to some unevenness between chapters.

There were large variations amongst the chapter authors as to whether they used the suggested grading system (ranging from not at all to religious zeal), or included sources of evidence used or provided succinct conclusions, implications for practice or identified gaps in the evidence.

One of the motivations for producing an "evidence-based" textbook (utilising systematic review evidence where available) was to avoid the pitfalls of traditional reviews and the inherent biases of authors. Unfortunately there is plenty of evidence of the traditionalists active in this publication. Too often, authors contributed information re-litigating the pathology, epidemiology and clinical manifestations of a cancer which don't help answer any of the questions which clinicians have about managing patients. The chapter on "Screening for breast cancer" had 57 references of which at least 13 were by the chapter's author. The huge controversy engendered by the Cochrane review on mammographic breast screening was dismissed in one sentence by reference to the author's own work.

The chapter on cutaneous melanoma provided a further example that authors were using evidence selectively. The effectiveness of adjuvant interferon after melanoma is an ongoing controversy but the authors ignored a systematic review (J Clin Oncol 2002;20:1818-1825) on the topic which concluded “there is still no proven standard adjuvant therapy for high-risk cutaneous melanoma.” The textbook chapter authors concluded that “interferons used as adjuvant treatments can benefit some patient groups.”

The final frustration came when the book’s CD wouldn’t run on my laptop computer (admittedly it did run on another desktop computer) and the www.evidbasedoncology.com website was accessed to find that 18 months after publication there were no chapter updates posted.

So will this distinctive purple text at arm’s reach from my desk be used in preference to a few mouse-clicks at the computer to find evidence to help me make clinical decisions about patients? The answer is not clear at this stage but my money is probably on the mouse.