What was the impact of changing NICE suspected-cancer referral guidelines on time to diagnosis of ovarian cancer?

Talk Code: 
Sarah Price
Willie Hamilton, Ruben Mujica-Mota, Obi Ukoumunne, Yoryos Lyratzopoulos, Sal Stapley, Anne Spencer
Author institutions: 
University of Exeter, University College London


The 2005 NICE referral guidelines for suspected ovarian cancer were revised in 2011. It is important to know what impact this has had on the time to diagnosis after symptomatic presentation to healthcare.


This observational study was set in The Clinical Practice Research Datalink (CPRD), a dataset of UK prospective, primary care medical records. In a pre-post design, we studied two cohorts of women (≥18 years) with an incident diagnostic code for ovarian cancer made between 01/08/2008 and 31/21/2010 (“Pre”) or between 01/08/2011 and 31/12/2013 (“Post”). We identified CPRD codes recorded in the year before diagnosis for clinical features that might be caused by the undiagnosed ovarian cancer. We grouped women by their index cancer feature. The “Old NICE” group had index cancer features in the 2005 guidelines: bloating, abdominal/back pain, abdominal/pelvic mass, urinary symptoms, or constipation. “New NICE” women had index features introduced in 2011: early satiety, appetite loss, pelvic pain, weight loss, fatigue, change in bowel habit, raised Ca-125 or ascites. The outcome variable was diagnostic interval: the number of days from index cancer feature presentation to diagnosis. Quantile difference-in-difference regression estimated the change in median diagnostic interval attributable to guideline revision. Explanatory variables were cohort (Post/Pre), NICE grouping (New/Old) and an interaction between cohort and NICE grouping. Analyses adjusted for age and deprivation, and clustering within general practice. All analyses were carried out using Stata 15.


Ovarian cancer diagnostic codes were identified for 1,708 women (Pre: n=874; Post, n=834). Mean (SD) age at diagnosis was 65.5 (14.1) years. Of these women, 1,267 (74.2%) attended primary care with ≥1 clinical features of ovarian cancer before diagnosis, and were entered into analyses. The distribution of NICE grouping was similar in the Pre (“Old NICE”: n=484/636, 76.1%; “New NICE”: n=152/636, 23.9%) and Post (“Old NICE”: 464/631, 73.5%; “New NICE”: 167/631, 26.5%) cohorts. Abdominal pain was the most common index feature (Pre: 250/674, 37.1%, of all index features; Post: 234/689, 34.0%). Raised Ca125 accounted for 70/674 (10.4%) of all index features in the “Pre” period, and 108/689 (15.7%) in the “Post”. The unadjusted median diagnostic intervals, by NICE grouping and cohort, were: 62 days (“Old NICE”, “Pre”, n=484); 77 days (“Old NICE”, “Post”, n=464); 63 days (“New NICE”, “Pre”, n=152); and 50 days (“New NICE”, “Post”, n=167). The adjusted pre-to-post changes in median diagnostic interval in the “Old NICE” and “New NICE” groups were +11.6 days (95% confidence interval -1.5 to 24.7 days) and -14.5 (-33.1 to 4.2 days), respectively. The coefficient for the interaction term (cohort × NICE grouping) was -26.1 (-48.0 to -4.1) days (p=0.02).


Revising the NICE guidelines for referral for suspected ovarian cancer was associated with a reduction of diagnostic interval of about 1 month greater than expected from secular trends. We are now investigating the impact on patient outcomes, such as 1-year survival.

Submitted by: 
Sarah Price
Funding acknowledgement: 
Cancer Research UK