Reply: The utilization of accurate body mass index classification is imperative for grouping based on BMI.

Sir,

We read with interest the comments to our paper (Hernáez ) by Drs. Liang and Fan (Liang and Fan, 2022). They are correct that the BMI thresholds used in our article did not correspond to the standard values suggested by the World Health Organization (WHO). We have acknowledged this in the discussion of the study: ‘Fifth, we were unable to use standard BMI categories for underweight (WHO threshold <18.5 kg/m2), as this only included 2.69% and 0.14% of the female and male participants, respectively’. We decided to group all participants with BMI values below 20.0 kg/m2 to obtain more robust odds ratios in the stratified analyses. For example, in Mendelian randomization, a 1 standard deviation increase in the genetically predicted BMI values was linked to 14% less odds of subfertility in all women with BMI values <20 kg/m2 (odds ratio [OR] 0.86, 95% confidence interval [CI] 0.76–0.98). When we split this group to underweight women (<18.5 kg/m2) and women with low-normal BMI values (18.5–19.9 kg/m2), the direction and magnitude of the associations remained constant (18.5–19.9 kg/m2: OR 0.86, 95% CI 0.75–0.99; <18.5 kg/m2: OR 0.84, 95% CI 0.58–1.22), although the 95% CIs were wider due to the more limited sample size in each group. Something similar happened in men (<20 kg/m2: OR 0.81, 95% CI 0.47–1.41; 18.5–19.9 kg/m2: OR 0.84, 95% CI 0.43–1.63; <18.5 kg/m2: OR 0.69, 95% CI 0.19–2.43). In any case, the J-shaped association curves suggest that there is an increase in the odds of subfertility in participants at the lower end of BMI distribution (Staley and Burgess, 2017). The ‘low-normal’ BMI category has also been used in previous studies following our same methodology (Sun ) and several other epidemiological works (Flegal ). Therefore, our conclusions are robust, even though this decision prevents us from establishing associations between clinically defined underweight and subfertility. Consequently, we have not used the term ‘underweight’ in the description of our results in the text; we systematically used ‘participants with BMI values below 20.0 kg/m2’ to describe this group and indicated in the limitations of the study that further studies involving larger populations in the lower end of the BMI distribution are warranted.

Conflict of interest

The authors have no conflicts of interest to declare.