The validity of self‐reported cancer in an Australian population study

Abstract Objective: The aim of this study is to determine the validity of self‐reported cancer data by comparing it to the Australian Cancer Database (ACD). Methods: Self‐reported data were obtained from the Australian Diabetes, Obesity and Lifestyle (AusDiab) study, which were then linked to the ACD up until 31 December 2010. Positive predictive value, negative predictive value, sensitivity and specificity were calculated. Cohen's kappa coefficient (ĸ) was also calculated to assess the agreement between self‐reported cancer and the ACD. Logistic regression was used to examine the determinants associated with false negative and false positive reporting. Results: The overall sensitivity of self‐report cancer was 71.1%, and sensitivities showed great variation by cancer site. Higher sensitivities were observed for breast (90.7%), bowel (77.8%) and prostate (77.1%) cancers, whereas the lowest sensitivity was observed for melanoma of the skin (36.9%). Similarly, the kappa coefficient analysis showed substantial agreement for self‐reported breast cancer (ĸ= 0.79) and moderate agreement for melanoma (ĸ= 0.45) against the ACD. Years since cancer diagnosis and older age were associated with false negative reporting and older age was associated with false positive reporting. Conclusions and implications: The use of self‐reported cancer to collect cancer outcomes has varying reliability, depending on cancer type and population. The findings presented here may assist medical researchers in making informed decisions when conducting research using self‐reported cancer data in Australia where the acquisition of registry data is not feasible