LAUSR

Dear Editor, McCloskey and colleagues have raised concerns about the inclusion of the SOS in the meta-analysis of studies on fracture prevention by screening for high fracture risk [1]. We believe that the preselection of participants for bone densitometry is remarkable homogeneous between the three included studies. The risk factors in the SOS were similar to the main risk factors that were used by the FRAX with exception from omission of alcohol and smoking [2]. To underpin this, we have calculated the overlap between the population that was selected with the SOS criteria and the population that would have been selected if the FRAX was used: 96% of the participants with a FRAX score without BMD above the agedependent threshold were selected with the SOS criteria. The final advice for treatment with anti-osteoporosis medication in the SOS was based on similar criteria as used by SCOOP, namely (re)calculation of FRAX, using the risk factors and the T-score, using age-related cutoffs of estimated fracture risk. ROSE used T-score < −2.5 as cutoff [2–4]. However, some exceptions were included in the SOS for ethical reasons. Firstly, women with a treatment indication according to the Dutch guidelines, but without a treatment indication according to the calculated FRAX threshold, were advised treatment. As the SOS treatment protocol is wider than the Dutch guideline, this only concerned 36 out of 1417 women with a treatment indication. Secondly, it was decided that no patients with normal BMD (T-score < − 2.0) without vertebral fracture should be treated even if their risk estimation was above the threshold. This concerned 170 women. It is unlikely that those small numbers of exceptions with a relatively low fracture risk will have a substantial influence on the results. In addition to the T-score, both the SOS and ROSE performed vertebral fracture assessment for treatment advice. The assessment in the SOS was based on semi-quantitative analysis. Height measurements of the vertebra could be overruled in case of other diagnosis or the absence of signs of fractures. To minimize false positives, all images were evaluated by two trained independent observers that were blinded for each other’s judgement. If there was no consensus, the finale judgement was given by a third expert [2, 5]. Thus, the exact criteria for the advice for anti-osteoporotic drug were clearly specified in the SOS. Additionally, personalized advises, i.e. feedback to the general practitioner about falls, calcium take, and vitamin D status and important laboratory abnormalities, were provided. Moreover, we would like to remark that direct comparison of the baseline characteristics of the three studies is out of place. The characteristics do not represent the same population (preselected participant with ≥ 1 clinical risk factors in the SOS versus all participants in SCOOP and ROSE). If only the participants in the treatment group are compared, the characteristics with respect to the main risk factors, FRAX and falls are rather similar in SCOOP and SOS. The rather similar screening approach between the included studies, but remarkable, unexpected and unexplained large effect of screening on hip fractures within SCOOP were reason to perform the sensitivity analysis [6, 7]. Nevertheless, we think that the best estimate of the effect of screening on high fracture risk is the meta-analysis of the three studies combined. We hope that the next step will be to look into the details and differences of the three studies. We think that cooperation between the study groups to perform a per patient meta-analysis could help to gain more insights for optimizing future screening programs.