Dear Editor,

We thank Drs. Cure-Cure and Cure [1] for their interest in our recently published article [2]. We agree that well-designed cross-sectional studies are early, reliable, and cost-effective sources of information. Meta-analysis of high-quality cross-sectional studies should be encouraged. This analysis will facilitate the presentation of valuable and reliable evidence combined with findings of prospective studies.

In our previous meta-analyses, we searched for prospective and non-prospective (e.g., cross-sectional and case control) studies reporting the parity-related risk of osteoporotic fracture (OF) at any skeletal location, such as spine, wrist, and hip. Subgroup analysis based on non-prospective reports indicated that parous women exhibited a lower OF risk than nulliparous women, with the corresponding pooled OR of 0.725 (95 % confidence interval (CI) = 0.614–0.836, I 2 = 57.3 %, n = 19), 0.803 (95 % CI = 0.730–0.876, I 2 = 0.0 %, n = 15), and 0.612 (95 % CI = 0.490–0.733, I 2 = 47.1 %, n = 14) for at least 1, 1–2, and 3 or more parities, respectively. When the hip fracture risk was exclusively considered, the respective counterpart values were 0.797 (95 % CI = 0.717–0.877, I 2 = 0.0 %, n = 12), 0.835 (95 % CI = 0.725–0.944, I 2 = 0.0 %, n = 10), and 0.726 (95 % CI = 0.610–0.843, I 2 = 0.0 %, n = 10).

We hypothesize that the specific explanatory variable of interest (i.e., parity) contributes to the concurrent findings from prospective and non-prospective studies. Based on Hill’s criteria for epidemiologic causal inference, time sequence of cause and effect is legitimate. That’s the major limitation of cross-sectional study. Generally, deliveries occur before menopause, and women often suffer OF after menopause. Therefore, it’s reasonable to assume that most OFs among women occur after delivery. That’s why cross-sectional studies may also present reliable information on parity-related OF risk.