Analytic Summary of a Peer-Reviewed IMRAD Article
Title
Association between Body-Mass Index and Risk of Death in More Than 1 Million Asians
​
Authors
Wei Zheng, M.D., Ph.D., Dale F. McLerran, M.S., Betsy Rolland, M.L.I.S., Xianglan Zhang, M.D., M.P.H., Manami Inoue, M.D., Ph.D., Keitaro Matsuo, M.D., Ph.D., Jiang He, M.D., Ph.D., Prakash Chandra Gupta, Sc.D., Kunnambath Ramadas, M.D., Shoichiro Tsugane, M.D., Ph.D., Fujiko Irie, M.D., Ph.D., Akiko Tamakoshi, M.D., Ph.D., Yu-Tang Gao, M.D., Renwei Wang, M.D., Xiao-Ou Shu, M.D., Ph.D., Ichiro Tsuji, M.D., Ph.D., Shinichi Kuriyama, M.D., Hideo Tanaka, M.D., Ph.D., Hiroshi Satoh, M.D., Ph.D., Chien-Jen Chen, Sc.D., Jian-Min Yuan, M.D., Ph.D., Keun-Young Yoo, M.D., Ph.D., Habibul Ahsan, M.D., Wen-Harn Pan, Ph.D., Dongfeng Gu, M.D., Ph.D., Mangesh Suryakant Pednekar, Ph.D., Catherine Sauvaget, M.D., Ph.D., Shizuka Sasazuki, M.D., Ph.D., Toshimi Sairenchi, Ph.D., Gong Yang, M.D., M.P.H., Yong-Bing Xiang, M.D., M.Ph., Masato Nagai, M.Sc., Takeshi Suzuki, M.D., Ph.D., Yoshikazu Nishino, M.D., Ph.D., San-Lin You, Ph.D., Woon-Puay Koh, M.B., B.S., Ph.D., Sue K. Park, M.D., Ph.D., Yu Chen, Ph.D., Chen-Yang Shen, Ph.D., Mark Thornquist, Ph.D., Ziding Feng, Ph.D., Daehee Kang, M.D., Ph.D., Paolo Boffetta, M.D., M.P.H., and John D. Potter, M.D., Ph.D.
​
Introduction
There have been many studies on BMI and related health concerns, but they were mainly based on populations of European origin. The relationship between BMI and overall risk of death remains unclear in Asian populations. Studies have shown that Asians generally have a higher percentage body fat for a given BMI and they have elevated health-related risks at a relatively lower level of BMI. Because of these observations, it has been suggested to have lower BMI cutoff points for overweight and obese for Asian populations.
​
Methods
A total of nineteen cohorts were included in the study. All the cohorts have accrued at least 5 years of follow up data and a minimum of 10,000 participants. All the participating cohorts were required to have available baseline data on BMI, age, sex, and cigarette smoking status and follow-up data on deaths of any cause. Additional data were collected on selected baseline illnesses and cause-specific deaths. There was a total of 1,141,609 subjects.
The association between BMI and the risk of death was analyzed with the use of Cox proportional-hazards regression model. There were 10 BMI levels with increments of 2.5 BMI units. There was additional analysis done to adjust for variables of cigarette smoking status and status regarding known baseline conditions (cancer, stroke, etc). Analyses was performed separately on data from the Indian and Bangladeshi populations and East Asian populations.
​
Results
The average BMI for the study population was 19.8 to 23.7. Over the average follow-up period of 9.2 years, approximately 120,700 cohorts died. Cardiovascular diseases accounted for 35.7% of deaths, cancer accounted for 29.9% of deaths, and other causes accounted for 34.3% of deaths. There was an elevated risk of death from any cause in the lowest BMI group with a factor of 2.0 to 2.8. In groups with a BMI higher than the reference range, the hazard ratios for death from any cause were elevated in East Asian populations but not in Indian and Bangladeshi populations. A U-shape association is seen between BMI and the risk of death from specific causes in East Asian populations but not in Indian and Bangladeshi populations. There was no elevated risk of death from any of the specific causes in high-BMI groups of Indian and Bangladeshis.
​
Discussion
In East Asians, both a low BMI and a high BMI were associated with an increased risk of death from any cause or cause-specific death, resulting in an overall U-shape association. In Indian and Bangladeshi populations, there is an elevated risk of death only among those with low BMI. Over the past 10 years, several large cohort studies have also evaluated the association between BMI and risk of death. In general, these studies have shown that the lowest risk of death is associated with a BMI in the range of 23 to 27, regardless of study population. The finding that the optimal weight range is similar in many studies of different populations argue strongly against the use of race- or ethnicity-specific cutoff points to define overweight and obesity.
​
Is the topic of the paper somewhat original?
Yes. There have been similar studies, but they were mainly based populations of European descent. This study used an Asian-only population, particularly East Asian, Indian, and Bangladeshi.
​
Were enough data obtained to reach valid conclusions?
Yes. There was a total of 1,141,609 subjects. Each cohort in the study had at least 10,000 participants.
​
Are the results consistent with those of other studies?
Yes. Over the past 10 years, several large cohort studies have also evaluated the association between BMI and risk of death. In general, these studies have shown that the lowest risk of death is associated with a BMI in the range of 23 to 27, regardless of study population. The study with Asian populations is consistent with other studies of the same type (BMI and risk of death).
​
Do the study’s findings have practical importance, regardless of whether they have statistical significance?
Yes, the study determined if there is a need for an Asian-specific BMI cutoff point for overweight and obese.
​
How are outliers handled in the data?
Data that are not included in the sample include missing data, participants who have a BMI of over 50, and participants who are under 18.
​
Who sponsored the study?
There were no identifiable sponsors in the study.
​
Do the Results section and the Methods section match?
Yes, the Results and Methods sections match. In the Methods section, it is specified that there will be 8 BMI levels. In the Results section, the graphs show that there are 8 BMI levels (shown below). The study specifies that they will be using Cox proportional-hazards regression models, which is shown in the graphs. In the y-axis, they indicate the hazard ratio, and on the x-axis, they indicate the BMI levels. For the tables provided in the Results section, the numbers add up to the indicated sample size in the Methods section. Data is split according to the Methods section (smokers and nonsmokers, known illnesses, etc).
​
Article Source: http://www.nejm.org/doi/full/10.1056/NEJMoa1010679