• HEPATOCELLULAR CARCINOMA
• OBESITY
• EPIDEMIOLOGY

We read with great interest the recent article by Hagström et al, 1 in which they used register data from more than 1.2 million Swedish men enlisted for conscription between 1969 and 1996 in order to investigate the relationship between body mass index (BMI) in late adolescence and future liver diseases. During a follow-up of more than 34 million person-years, 251 cases of hepatocellular carcinoma (HCC) were identified from the Swedish Cancer Register (SCR). In this study, the authors used a sample survey from the year 19982 for reference to indicate the high register completeness from the SCR, which has captured around 96% of all diagnosed tumours. However, another study, also from Sweden, which was just recently published on Hepatology, indicates that up to 37%–45% of HCCs in Sweden were never reported to the SCR, and this percentage has increased slightly over time.3 If calculated according to this rate, at least 90–110 cases in this study by Hagström et al would possibly have been under-reported by the SCR. Nevertheless, the missing cases are substantial and their impact on the outcome of the study is not to be ignored.

In this study, the authors assessed the impact of late adolescence BMI on the risk of HCC and divided it into BMI <18.5, 18.5≤BMI <22.5, 22.5≤BMI <25, 25≤BMI <30 and BMI ≥30, with HCC occurrence of 31, 138, 47, 25 and 10 in each group, respectively. Because this is a block analysis between up to five groups, even if there is only a single-digit deviation (ie, no more than 10 cases) in each group, it is sufficient to challenge the statistical significance of the current results. Are the 90–110 missing cases, consistent to the currently enrolled 251 cases, normally distributed according to the different BMI groups? The considerable amount of missing cases and the high degree of uncertainty may constitute considerable challenges in the analyses and interpretation of results, and potentially weaken the validity and robustness of results.4 We are very concerned about the large number of missing cases and their statistical impact on the results. With regard to the issue of HCC incidence, this study is more like a sample survey than an overall survey, and the reliability of its conclusions warrants careful interpretation and further validation.

In summary, clarification regarding the above-mentioned omission would greatly solidify the conclusions of their study.