A systematic review of methodological considerations in time to diagnosis and treatment in colorectal cancer research

It is well established that the timely, early diagnosis of cancer is important for improving a range of outcomes for people with cancer [1] . Many differences in outcomes can be accounted for by the stage of cancer at the time of diagnosis [2] . For this reason, early diagnosis and the intervals before diagnosis and treatment are critical factors in improving outcomes and reducing mortality. This holds for colorectal cancer, which is the fourth most commonly diagnosed cancer worldwide [3] . Despite many jurisdictions implementing screening programmes to diagnose colorectal cancer earlier, many countries have poor uptake rates, and most cancers are diagnosed once people develop symptoms [45] . Consequently, research focusing on the timeliness of events in the period before diagnosis and initial treatment in colorectal cancer is of importance.

Previous attempts to synthesise the evidence on the role of varying lengths of time to diagnosis and treatment with cancer outcomes, such as survival, have noted that studies in this area are heterogenous and often have poor methodological quality, which leads to difficulty combining results [12678 . While scoping reviews of specific methodological issues [2] , and design and reporting guidelines for research on the early diagnosis of cancer (the Aarhus Statement [8] ) exist, no systematic review of the methodological approaches taken when conducting studies of time to cancer diagnosis and treatment has been performed. A systematic review of the methodological approaches in a research area allows for a comprehensive assessment of potential issues in the design and conduct of research, in order to improve the strength of the evidence base and find possible areas of improvement.

This review considers three key methodological features of importance in this research area. Firstly, methodological issues including the choice and treatment of covariates in statistical analysis 91011121314 . Secondly, the potential for bias [15] , including confounding by indication 16171819 , immortal time bias [20] and the appropriate use of disease stage. Finally, the choice and justification of intervals used. Given that the period before diagnosis and initial treatment is a diverse period of activity and events, this final feature is important to assess how intervals align to theoretical frameworks.

None of these methodological issues have been comprehensively assessed across the entire body of work in the early diagnosis literature, despite advances in reporting and design guidelines such as the Aarhus statement, and noticeable methodological advancements [18] . Therefore, the objective of this review was to report on the methodological features of studies that assess the role of an interval in the lead up to diagnosis and treatment on any health-related outcome, in colorectal cancer. Colorectal cancer was chosen as it is common, occurs in both men and women, and has been the focus of much research in the early diagnosis field. Additionally, previous research indicates the importance of early diagnosis [1] and has highlighted considerations such as the wait-time paradox, where there is a paradoxical increase seen in mortality associated with shorter lengths of time [1819] . This suggests that methodological considerations are important in the design and conduct of research in this area, as well as likely generalisable to research in other cancer types.