Study of effect of delays on ovarian cancer was weak

Editor—Kirwan et al concluded that delays in referral or diagnosis did not adversely affect the survival of patients with ovarian cancer.¹ Their analysis of the relation with survival, however, includes stage of diagnosis as a covariate, so if delays adversely affect survival by producing a more advanced stage of disease at presentation (which seems an obvious possibility) this study would not show it.

In other diseases, such as breast cancer, some analyses show that, after adjustment for stage, a longer symptom history relates to an improved prognosis; this is understandable as for a given stage at presentation a longer symptomatic phase may indicate a less aggressive cancer.²,³ Analysis of length of symptom history with regard to survival is complex, and long follow up is needed to allow adjustment for lead time effects. Analysis should not adjust for variables such as stage, which may be intermediary in the causal pathway.⁴

Kirwan et al's study is also weak because it depends solely on general practitioners' records, with no attempt at validation, and is of a small sample. Surprisingly, the authors can “see no other way to obtain the relevant data” than to rely on general practitioners' records. The issue is little different to that of obtaining reasonably consistent and accurate data about aetiological factors, which is done by using well designed structured questionnaires and trained interviewers who are independent of the staff providing clinical management.

While of course needing more resources, investigations of symptom history and presentation can be carried out on representative series of patients, perhaps in combination with studies of aetiological factors.⁵