Detecting cancer earlier

The story behind the paper

Why we need process measures: to understand how we can better prioritise efforts to detect cancer earlier

A team of us have published a new paper today in the British Journal of Cancer: “Measures of promptness of cancer diagnosis in primary care: secondary analysis of national audit data on patients with 18 common and rarer cancers” (open access). Here is my “story behind the paper”:

Back in 1995 Jonathan Mant and his colleague Nick Hicks wrote a BMJ article with the intriguing title: “Detecting differences in quality of care: the sensitivity of measures of process and outcome in treating acute myocardial infarction.” You can read this frequently cited article here. In that paper Mant and Hicks demonstrate that process measures have important advantages compared to outcome measures, particularly when the outcome of interest is rare. Although both outcome and process measures of quality are needed, process measures have had bad press in recent years, which saw ‘outcomes’ making a big comeback (see this or that link for examples).

Let’s now revisit this enduring “processes vs. outcomes” debate in the field of health policies for cancer. For many years the only available ‘tools’ to measure success in cancer control were outcome measures (such as incidence, mortality and survival) which are difficult to interpret with confidence regarding the processes of care that have generated them. Survival, for example, is influenced by both treatment and stage at diagnosis, let alone many other patient and tumour factors. What aspect of cancer diagnosis or care to prioritise in order to improve survival is therefore difficult to judge without any additional information about the different time intervals from symptom onset to diagnosis; or about whether treatment was timely and comprehensive – in short without additional information about relevant process measures.

Although information about processes of cancer care has been traditionally hard to come by, things have started to change after the launch of the National Awareness and Early Diagnosis Initiative (NAEDI) in 2008. Several projects supporting NAEDI have generated much needed data on variation in the awareness and appraisal of cancer symptom among members of the general public (see examples here and here). Other projects linked to NAEDI have also helped to generate evidence about processes of care after patients with cancer first present (usually to their GP) with a cancer symptom. About a year ago a team of us published a paper in the Lancet Oncology (see here or here) showing wide variations in the number of times cancer patients had seen their GP with cancer symptoms before they were referred to a specialist for investigation. A rightful criticism of that paper was that it did not show whether the number of consultations was associated with the time interval from presentation to referral. In principle, patients who saw their GP many times could have had this experience during a short spell (e.g. three consultations in two weeks) or over several months.

We have now published evidence that helps to answer this criticism: We found that on average more pre-referral consultations do indeed mean more ‘delay’. Given how rare (and often non-specific) cancer presentations in primary care overall are, a remarkably large majority (82%) of cancer patients were referred after one (58%) or two (25%) consultations, after a median interval of 0 and 15 days. However, patients who had 3 or 4 consultations had intervals of 34 and 47 days, and those with 5 or more consultations 97 days. In other words, the number of pre-referral consultations and the time interval from presentation to referral are highly correlated. Patients with cancers who typically present with non-specific symptoms (such as multiple myeloma, lung and stomach cancer) were particularly likely to experience more pre-referral consultations. How can this information be useful to health policy?

First, this kind of research provides us with a ‘roadmap’ about where future investment in early diagnosis research is most needed. Patients with cancers that are harder-to-suspect (because their symptoms are not specific enough) will particularly benefit from better targeting of research efforts. We have a duty to advocate that these considerations are taken into account by research funders.

Second, although the sensitivity of medical consultation to suspect cancer appears to be far higher than what could be expected by the positive predictive values of different symptoms and comparable to several ‘objective’ tests, developing and evaluating initiatives to further support GPs to appropriately suspect cancer and refer patients for investigations (such as clinical decision support tools) is definitely worth while: if they work, they will help reduce both the number of pre-referral consultations and delays in investigations.

Much more research is needed until the challenges of early detection of cancer are overcome. But learning from variation in the processes involved in presentation and referral will be crucial in our efforts to improve outcomes.

Media coverage of this paper: BBC news, BMJ news, CRUK website


This entry was posted in Blog and tagged , . Group: . Bookmark the permalink. Both comments and trackbacks are currently closed.
  • The Cambridge Centre for Health Services Research (CCHSR) is a thriving collaboration between the University of Cambridge and RAND Europe. We aim to inform health policy and practice by conducting research and evaluation studies of organisation and delivery of healthcare, including safety, effectiveness, efficiency and patient experience.