A case control study of clinical factors associated with ovarian cancer in a younger population with high rates of migration, deprivation and ethnic diversity.

Problem

To identify the risk factors and symptoms associated with ovarian cancer within a multiethnic population in East London and determine if this is consistent with current guidance.

Approach

Design and setting-Case control study using primary care data from 204 general practices in four multiethnic Clinical Commissioning Group (CCG) areas in East London.Method-We identified 632 women diagnosed with ovarian cancer in the sample population between January 2010 and December 2020, matched with 3159 female controls of the same age and regional primary care commissioning group. Primary care records from each case and matched control were analysed for associated risk factors and symptoms established from national clinical guidelines. Each control was allocated a pseudodiagnosis date equal to the date of diagnosis of her matched case. We had data from primary care records on symptoms up to three years prior to diagnosis, demographic factors, including ethnicity and clinical factors such as personal medical history, lifestyle and family history of ovarian cancer. Univariate and multiple logistic regression analysis (MVR) were used to estimate the effect of the various factors on likelihood of ovarian cancer. A final model was developed using backward stepwise regression.

Findings

The age-specific incidence of ovarian cancer per 1,000 women between the ages of 35-89 years, was lower in our study population when compared with England and Wales. Significantly increased ovarian cancer risk were observed for those underweight, smokers past and current, white ethnicity, family history of breast or ovarian cancer, past history of non-ovarian malignancy and history of menstrual disorders. Between 12-36 months prior to ovarian cancer diagnosis, those with the following symptoms were at significantly greater risk: abdominal distension (OR = 7.100, CI 4.267-11.815, P<0.001); abdominal/pelvic pain or discomfort (OR= 5.158, 95% CI 4.041-6.584; P<0.001), change in bowel habits (OR=2.775, 95%CI 1.978-3.892; P<0.001), bowel obstruction (OR=8.1; 95%CI 1.301-51.078; P<0.05); nausea (OR=2.701; 95%CI 1.076-6.784; P<0.05), postmenopausal bleeding (OR = 5.608; 95% CI 2.673-11.763; P<0.001), shortness of breath (OR=2.092; 95%CI=1.426-3.071; P<0.001) and weight loss (OR=3.037, 95% CI 1.321-6.980, P<0.05). These symptoms remained significant over 36-months prior to diagnosis, with the exception of bowel obstruction, nausea and shortness of breath. In the multivariant regression analysis, abdominal distension, abdominal pain, change in bowel habits and postmenopausal bleeding remained significant though these clinical factors were less predictive of ovarian cancer in our study population compared with the rest of England and Wales.

Consequences

Conclusion We identified comparable diagnostic clinical risk factors in our multiethnic study population that mirror findings in predominantly white populations, although with a less significant predictive capacity for ovarian cancer. The variation is more attributable to age-specific distinctions within our study population than differences in race or ethnicity. This indicates the necessity of incorporating demographic characteristics when applying clinical guidance to assess the risk of disease.