Linking population-based cohorts with cancer registries in LMIC: a case study and lessons learnt in India

• Authors: Aastha Aggarwal, Public Health Foundation of India; Ranganathan Rama, Cancer Institute-WIA; Preet K Dhillon, Public Health Foundation of India; Mohan Deepa, Madras Diabetes Research Foundation (ICMR Center for Advanced Research on Diabetes); Dimple Kondal, Emory University; Naveen Kaushik, Centre for Chronic Disease Control, Dwarka, Delhi, India; Dipika Bumb, Ramaiah International Centre for Public Health Innovations; Ravi Mehrotra, Centre for Health, Innovation and Policy, Noida, Uttar Pradesh, India; Betsy A Kohler, North American Association of Central Cancer Registries, Springfield, Illinois; Viswanathan Mohan, Madras Diabetes Research Foundation (ICMR Center for Advanced Research on Diabetes); Theresa Gillespie, Emory University; Alpa Patel, American Cancer Society; Swaminathan Rajaraman, Cancer Institute-WIA, Chennai, Tamil Nadu, India; Dorairaj Prabhakaran, Emory University; Kevin Ward, Emory University; Michael Goodman, Emory University
• Journal Title: BMJ OPEN
• Volume: Volume 13, Number 3
• Publisher: BMJ PUBLISHING GROUP | 2023-03-01, Pages e068644-e068644
• Type of Work: Article | Final Publisher PDF
• Publisher DOI: http://doi.org/10.1136/bmjopen-2022-068644
• Subjects: Health Sciences, Epidemiology; Health Sciences, Public Health; Health Sciences, Oncology
• Keywords: Science & Technology; Life Sciences & Biomedicine; Medicine, General & Internal; General & Internal Medicine; PUBLIC HEALTH; EPIDEMIOLOGY; ONCOLOGY; RECORD LINKAGE; SELF-REPORT; VALIDITY; HISTORY; RATES

Abstract

Objectives In resource-constrained settings, cancer epidemiology research typically relies on self-reported diagnoses. To test a more systematic alternative approach, we assessed the feasibility of linking a cohort with a cancer registry. Setting Data linkage was performed between a population-based cohort in Chennai, India, with a local population-based cancer registry. Participants Data set of Centre for Cardiometabolic Risk Reduction in South-Asia (CARRS) cohort participants (N=11 772) from Chennai was linked with the cancer registry data set for the period 1982-2015 (N=140 986). Methods and outcome measures Match∗Pro, a probabilistic record linkage software, was used for computerised linkages followed by manual review of high scoring records. The variables used for linkage included participant name, gender, age, address, Postal Index Number and father's and spouse's name. Registry records between 2010 and 2015 and between 1982 and 2015, respectively, represented incident and all (both incident and prevalent) cases. The extent of agreement between self-reports and registry-based ascertainment was expressed as the proportion of cases found in both data sets among cases identified independently in each source. Results There were 52 self-reported cancer cases among 11 772 cohort participants, but 5 cases were misreported. Of the remaining 47 eligible self-reported cases (incident and prevalent), 37 (79%) were confirmed by registry linkage. Among 29 self-reported incident cancers, 25 (86%) were found in the registry. Registry linkage also identified 24 previously not reported cancers; 12 of those were incident cases. The likelihood of linkage was higher in more recent years (2014-2015). Conclusions Although linkage variables in this study had limited discriminatory power in the absence of a unique identifier, an appreciable proportion of self-reported cases were confirmed in the registry via linkages. More importantly, the linkages also identified many previously unreported cases. These findings offer new insights that can inform future cancer surveillance and research in low-income and middle-income countries.