Letter: non-alcoholic fatty liver disease is associated with a history of osteoporotic fractures but not with low bone mineral density-authors' reply

We would like to thank Dr. Bidhendi Yarandi for his interest in our recently published systematic review and meta-analysis of the association between non-alcoholic fatty liver disease (NAFLD) and bone mineral density or self-reported history of osteoporotic fractures in middle-aged and elderly individuals.1 Using the AMSTAR 2, which is a 16-item critical appraisal tool for systematic reviews that include randomised or non-randomised studies of health care interventions,2 Dr. Bidhendi Yarandi raised two methodological issues and concluded that the overall quality rating of our systematic review was “moderate”.3 Indeed, in his judgment, our systemic review scored 14 items out of 16 of the AMSTAR 2 measurement tool, while lost points from items 12 and 13, which were both related to the issue of the possible impact of risk of bias (RoB) on the results.3 According to the aforementioned AMSTAR 2 measurement tool (available on the website https://amstar.ca/docs/AMSTAR-2.php and rating the overall quality of systematic reviews as “high”, “moderate”, “low” or “critically low”),2 this means that our systematic review had “more than one weakness but no critical flaws, and that it may provide an accurate summary of the available studies that were included in the meta-analysis”,2 as also recognized by Dr. Bidhendi Yarandi in the conclusion of his letter. That said, we respecfully disagree with the methodological comment raised by Dr. Bidhendi Yarandi. We believe that selective reporting bias of studies was not a major concern in our analyses, as our comprehensive search have made it unlikely that any published report was missed, whilst visual inspection of the funnel plot and the Egger’s regression test did not reveal any significant publication bias (as shown in supplementary Figure S1 of our article).1 In addition, and more importantly, as specifically requested by the item 12 of the AMSTAR 2 measurement tool, we performed subgroup analyses to investigate the possible impact of RoB on summary estimates of effect, stratifing the included studies according to the Newcastle-Ottawa assessment scale (as shown in supplementary Figure S7 of our article).1 Finally, we also discussed in our manuscript both the overall quality of the included studies and the likely impact of RoB when interpreting the results of the meta-analysis (as specifically requested by the item 13 of the AMSTAR 2 measurement tool). Regarding the second methodological issue raised by Dr. Bidhendi Yarandi (i.e., “significant association between NAFLD and increased odds of osteoporotic fractures”), we agree with him that the results of the pooled analysis examining the effect of NAFLD on risk of self-reported history of osteoporotic fractures (as reported in Figure 3) could provoke some controversy due to the estimation of borderline confidence interval (n=2 Chinese studies; random-effects odds ratio 1.43, 95% CI 1.0-2.06, I2=55.1%). However, as clearly reported both in the figure and in the Results section, this risk was particularly increased among older Chinese men (random-effects OR 2.10, 95% CI 1.36-3.25; I2=0%).