Sir,

We thank the authors for their reflections (Irving and Holden, 2013). As they acknowledge, we did not endorse the notion that optimal care means correctly identifying and referring a patient with cancer during the first consultation (Lyratzopoulos et al, 2013). We question whether this notion is indeed ‘widely held’. In the United Kingdom, policy makers and cancer charities often use a binary measure of promptness of cancer diagnosis comprising one or two (compared with three or more) consultations, for example, in public reporting of NHS patient experience surveys (Department of Health, 2012). This is also the measure mainly profiled in our paper. The main conclusions of our paper were that the number of pre-referral consultations is a good surrogate marker of the primary care interval; and that there is large variation in respect of both measures between patients with different cancers.

We would agree in principle that some repeat consultations may be appropriate, for example, within an ‘active monitoring’ (or ‘safety-netting’) framework. The challenge for the primary care physician is to identify the earliest point at which the possibility of a cancer diagnosis would justify investigation or referral. Such decisions are informed by knowledge of cancer risk for given symptoms and by clinical practice guidelines (NICE, 2005; Hamilton, 2009). However, any decision to delay investigation or referral as part of ‘active monitoring’ must take into account patient preferences and the potential risk of disease progression. Recent UK research indicates that patients express a strong preference for investigation of symptoms likely to be due to cancer, even when the positive predictive value of such symptoms is known to be low (Hollinghurst et al, 2013). Therefore, doctors need to ensure that whenever deferral of investigation or referral is being considered, the associated uncertainties and risks are communicated and shared with patients. We note the authors’ work on the time efficiency principle, but because biological processes of disease progression vary between medical conditions, the importance of timeliness of diagnosis and treatment may well be different in the context of cancer (Chen et al, 2008).

The authors are correct in their assertion that cancer presentations in primary care are relatively rare, but we would wish to correct the illustration regarding multiple myeloma and stomach cancer provided by the authors, who have overinterpreted data in our paper. As we have previously alluded to, data from the National Audit of Cancer Diagnosis in Primary Care largely reflect the cancer ‘case-mix’ observed in population-based incidence statistics but cannot be used to estimate incidence, not least because data collection periods varied from 6 to 12 months between participating practices. Using Office for National Statistics data, the incidence per practice of multiple myeloma and stomach cancer is in the order of 1 case every 2 years and 1 case every 18 months, respectively.