Improving early detection of cancer in primary care

The annual number of patients newly diagnosed with cancer in the Netherlands is rising and has recently exceeded 100,000, leading to 45,000 cancer related deaths each year. [1] Cancer survival rates in the Netherlands are low compared to other countries in Central and Northern Europe. [2] One of the reasons might be that in the Netherlands cancer is often detected in late stages of disease. As the majority of cancer patients present to primary care first, improving early detection of cancer by the GP could substantially contribute to improve cancer survival. [1,3] In the proposed study we aim to assess the extent of delay in diagnosis and treatment, and to identify determinants of diagnostic delay in the primary care setting, for the ten cancer types in which most gain is to be expected from delay reduction. These are cancer of the colon, lung, prostate, breast, esophagus, stomach, kidney, bladder, ovaries and for melanoma. These cancer types were selected because they are responsible for two-thirds of the cancer incidence and two-thirds of annual cancer related deaths in the Netherlands. In addition, these cancers have an unfavorable balance between stage distribution at detection and stage related survival rates, indicating substantial potential for improved survival through earlier diagnosis (Table 1, available as a separate file). For each of these cancer types the following questions will be answered: 1. What is the extent of delay in the different phases of cancer care? 2. What is the association between the extent of delay and the tumor stage distribution at diagnosis? 3. What are the determinants of delay in the diagnostic phase in primary care? We propose a retrospective cohort study in the Dutch Primary Care Database (DPCD), the joint routine care datasets of the 8 academic primary care registration networks with over 2 million primary care patients. Data of 3000 patients, 300 for each cancer type, will be extracted. The cancer diagnosis will be validated through linkage to the Netherlands Cancer Registry (NCR). The following data will be extracted from the patient files in the DPCD: first symptoms (kind and date), first cancer related visit to the GP (date), first referral for diagnosis by GP (date), date of histological diagnosis and date of treatment start, and data on possible determinants of delay for the primary care interval are retrieved manually from the individual patient files. Determinants of delay under study include characteristics of patients, disease, GP and consultation. The extent of delay is expressed as the median and mean duration of overall delay and the delay in the individual phases. Phases include; (1) the patient interval, (2) the primary care interval, (3) the diagnostic interval and (4) the treatment interval (Figure 1, available as a separate file). [4,5] The association between the extent of delay and the tumor stage distribution at diagnosis, is tested by comparing the mean and median duration of each interval for all tumor stages at diagnosis (stage I to IV). To define profiles of patients at risk of long delay in primary care, the possible determinants of delay in the longest (4th) quartile of delay will be compared to those in the first three quartiles. Multivariable logistic regression will be used and results expressed in OR, to be interpreted as the relative risk of specific determinants to cause delay. The required sample size is based on a pilot study on delay in diagnosis and treatment of colorectal cancer per formed by our research group. This pilot demonstrated that 197 patients, out of an initially selected 329 files, provided sufficient power to significantly identify determinants of delay in the different time intervals (Alpha = 0.05; Beta = 0.20). [5] Since the data linkage to the NCR and quality improvements of registrations counter most reasons for exclusion in the pilot study, we estimate that 300 initially selected patient files for each cancer suffice. [6] The results of this study will be distributed through scientific publications and presentations at national and international conferences. In addition the outcomes of the project will be transferred through guidelines and position papers for GPs, in educational materials for GPs, GP trainees and medical students, and via websites for health care professionals and patients.