Inclusion of patients lost to follow-up in “Kaplan-Meier” cumulative survival curves

Dear Editor, With great interest, we read the article titled “Survival pattern in cervical cancer patients in North West India: A tertiary care center study” by Yadav et al. published in the latest issue of this journal.[1] We congratulate the authors on conducting this much needed study on a very important health problem in India, that is, Cancer Cervix. The study enrolled 520 cervical cancer patients in North western state of India, Rajasthan with 5-year observed survival rates being 84.0%, 73.8%, 47.7%, and 13.5% for stages I, II, III, and IV, respectively. The overall survival rates at 1, 3, and 5 years were 93.07%, 72.3%, and 54.9%, respectively, with reported median survival of 60 months. We read the methodology and results section in detail and observed few findings that need mention here. First, we noted that the survival analysis was done using the “Kaplan-Meier” method. However, those patients who were lost to follow-up (n = 130) were excluded from the study. Assuming that these patients were present for atleast some portion of the study period, their inclusion may add value to the survival rates. Instead of exclusion, such participants could have been included in the “censored” data as and when they were lost to follow-up.[2] If exact time of events such as “loss to follow-up” or “death” were not known, another method of estimating cumulative survival, “actuarial (life table) method,” could have been used. Unlike, Kaplan-Meier method which requires an exact time of event (producing a step-like graph), actuarial method requires events taking place within a period (producing a smooth graph).[3] The second observation which confuses us as readers is that on seeing the graph, censoring was done only at the end of the follow-up period. It is not clear whether all surviving patients completed the follow-up period till 5 years here without any further withdrawal or loss to follow-up. Most probably both these observations are related to each other and the root cause is exclusion of lost to follow-up cases from the study. Third, the median survival in the study that has been derived is 60 months, which is equal to the total follow-up period of the study. By definition, median survival is the length of duration at which 50% of the study participants are surviving. Here, 54.9% patients (more than 50%) were surviving at the end of 60 months, which probably means that the study did not reach its median survival duration. The survival data of various malignancies collected from various cancer registries serve as a representative of the entire country. However, the coverage of population-based cancer registries in India is less than 10% of the entire population with even poorer in rural areas (0.1%).[4] Hence, studies similar to that conducted by Yadav et al. are important to understand the situation of the region in a better manner. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.