The Maladies of Marcel Proust. Doctors and Disease in His Life and Work

Perhaps the high points in the book are those examples of natural experiments wherein the temporal trends in a risk factor or an intervention strategy vary across subcategories and can be compared to cancer trends in the same subcategories. Particularly striking were the data presented by Hakama on cervical cancer: Since 1960 invasive cervical cancer has declined by a factor of two or more in the three Nordic countries which introduced extensive screening for carcinoma in situ, accompanied by a lesser decline in the fourth with moderate screening, and no decline in the fifth which lacked screening. The differences are so large and correspond so closely to the screening intervention that to account for them by invoking other factors is unrealistic. In the absence of the purity of a randomized trial of cervical cancer screening (previously deemed ethically unjustifiable), such data provide the single best answer to the efficacy of screening in preventing this cancer and have important implications regarding the natural history of cervical lesions viewed as premalignant. The generalizability of the experiment also has great appeal; it is conducted amongst those who are ultimately intended to benefit. At these moments, the book's positive view of the "chronoscopist" in the study of cancer cause and prevention rings most true.