Comment on “Risk analysis of metformin use in prostate cancer: a national population-based study”

To the editor,

One study performed by Dr. Jo and colleague reported that the risk of prostate cancer would decrease with increasing duration of metformin use [1]. Some points are presented for discussion. First, the flow chart of Dr. Jo et al.’s study showed that there were 45372 people in the non-metformin group (43.12%), but the result section showed that the incidence of urological cancer was 59844 in the non-metformin group (56.88%). Readers are confused by these numbers and words. The spelling errors are possible, but they are not caused by the journal. A correction notice is suggested. Second, the result section of Dr. Jo et al.’s study showed that the metformin group had an increased risk of developing prostate cancer than the non-metformin group (HR = 1.141, 95% CI = 1.127–1.156) [1]. Based on this number, metformin use would increase the risk of prostate cancer among people with type 2 diabetes mellitus. Then metformin is harmful and it should be avoided. However, the conclusion section showed that metformin use correlates with a reduced risk of prostate cancer and metformin use could be a new strategy in the prevention of prostate cancer. Readers are confused by the conclusion statement. Readers hope to know whether the HR is correct or not and what variables are adjusted during analysis? Third, the incidence rate of prostate cancer was 16981.7 per 100000 person-years among the metformin group and 18300.33 per 100000 person-years among the non-metformin group in Dr. Jo et al.’s study [1]. The absolute risk reduction of prostate cancer was 1319 per 100000 person-years in people with metformin use compared to non-use of metformin. Based on this number, metformin use would decrease the risk of prostate cancer among people with type 2 diabetes mellitus. Then the subsequent analysis on treatment duration of metformin would make sense. The research results would be reasonable that the longer the cumulative use of metformin, the lower the HR of prostate cancer risk. Fourth, the non-metformin group was defined as people who had never been prescribed with metformin during the study time or people with a metformin treatment period < 180 days. That is, people without the diagnosis of diabetes mellitus were also included in the study. Theoretically, the comparison group should focus on people with type 2 diabetes mellitus that did not take metformin. In fact, the baseline characteristics are different between people with and without diabetes mellitus [2]. The study results might be confounded by the selection bias found in Dr. Jo et al.’s study. Fifth, the flow chart of Dr. Jo et al.’s study showed that people diagnosed with prostate cancer between 2012 and 2014 were selected in this cohort study. So it did not need to observe the new development of prostate cancer. How did the authors measure the incidence rate of prostate cancer? Sixth, Table 5 of Dr. Jo et al.’s study showed that the total person-years in the non-metformin group were 30115.444, but Table 4 showed that the total person-years in the non-metformin group were 247930. Readers hope to know why the total person-years were different. Seventh, the method statement in abstract section showed that subjects diagnosed with prostate cancer from 2012 to 2014 were excluded. Is this statement correct? Overall, the purpose of this letter is to discuss scientific results with the authors, not criticize their research. However, some numbers and words in Dr. Jo et al.’s study are not presented correctly. Readers may interpret the results in the wrong way.

Authors’ contributions

Shih-Wei Lai contributed to the conception of the study, initiated the draft of the study, and approved the final draft. Bing-Fang Hwang, Chiu-Shong Liu and Kuan-Fu Liao interpreted the data.

Disclosure statement

The authors disclose no conflicts of interest.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.