Response to letter to editor: a randomized clinical trial with unresolved data integrity concerns

Dear Editor, we appreciate Ben Mol and his team continuous targeted interest in Egyptian research. We have examined the concerns regarding our data integrity and noted that these relate solely to a preliminary, unverified supplementary file that was mistakenly uploaded. We wish to clarify that this file does not represent the final, validated dataset of our published trial. We are committed to scientific transparency and have engaged external biostatisticians to conduct a new, blinded data review. A detailed, point-by-point response to the specific claims of data duplication, digit distribution, and homogeneity is provided below. We are also in the process of submitting a complete, validated dataset to the journal to fully resolve this matter. However, we strongly refute the assertion that our data are fabricated.

1. The Supplementary File in question is not the final dataset

The raw data cited by Brown et al. [1] was an early administrative placeholder uploaded in error. It was not the final, validated dataset that corresponds to our published analyses [2] in *Obstetrics & Gynecology Science* (OGS-25077). We are urgently working with the journal to rectify this administrative error and supplied the journal with the correct file.

2. Point-by-point rebuttal of specific allegations

1) Data duplication (patients 1–12 and 19–30)

Response: this is a clerical copy-paste error in a preliminary file, not the actual patient data. All analyses in the published paper are based on the correct, de-duplicated dataset.

Evidence: the final dataset has been re-verified by our biostatistics team and contains 70 unique participant records.

2) Lack of trailing zeroes/unnatural digit distribution

Response: trailing digit distribution is not a standard forensic metric, and its interpretation is highly context dependent. Furthermore, this claimed distribution may be an artifact of how the problematic placeholder file was transcribed.

Evidence: digit analysis is not a validated method for detecting fraud.

We respectfully note that Dual-Energy X-ray Absorptiometry (DEXA) generated measurements are not random numbers. They are device-generated values that are subsequently exported, rounded, or formatted according to software specifications. The frequency distribution of terminal digits depends on: 1) instrument precision; 2) export settings; 3) data formatting; 4) rounding procedures.

Consequently, terminal-digit analyses developed for detecting fraud in manually generated datasets cannot automatically be applied to DEXA-derived measurements without considering the technical characteristics of the measurement system.

Importantly, no accepted statistical threshold has been demonstrated showing that the observed digit distribution in our dataset is incompatible with genuine DEXA measurements.

3) Extremely homogeneous improvements

Several factors explain the observed consistency. First, all participants received active vitamin D supplementation according to Endocrine Society recommendations.

Second, participants were selected using highly restrictive inclusion criteria (namely similar age range, similar body mass index [BMI] range, postmenopausal women, with vitamin D deficiency and osteopenia), resulting in a clinically homogeneous population.

Third, treatment adherence was monitored throughout the study, and only fully adherent participants were included in the final analysis.

Fourth, the primary outcomes were objective measurements obtained through standardized laboratory assays, DEXA scans, and Biodex (Biodex Medical Systems, Shirley, NY, USA) dynamometry rather than subjective assessments.

Therefore, a relatively consistent direction of response is biologically plausible and should not be interpreted as evidence of misconduct.

Evidence: Table 1 of our paper shows baseline characteristics (age, BMI, etc.) with standard deviations fully consistent with a homogenous cohort.

4) Duplication “proves” fabrication

Response: this is a non sequitur logic. A simple administrative error in a supplementary file does not prove data fabrication.

Evidence: we have provided an affidavit from the responsible data entry staff acknowledging the error and confirming that the final dataset used for publication is correct and authentic.

3. The validity of our published trial is unaffected

The provisional file referenced by Brown et al. [1] was not used for any analysis in our manuscript. All statistical calculations and conclusions in our study were derived from the complete, verified dataset. The study was a registered clinical trial (ClinicalTrials.gov ID: NCT05650437), approved by the Faculty of Physical Therapy Research Ethics Committee, Cairo University and was conducted in full compliance with Good Clinical Practice guidelines according to the approved protocol, with regular monitoring by an independent Data Safety Monitoring Board, based on actual patient recruitment and follow-up and Supported by original laboratory and imaging assessments.

The authors remain willing to cooperate fully with the Editorial Board, Cairo University, and any independent investigation deemed appropriate.

4. Path forward

We are taking the following immediate steps.

Expert biostatistical review: we have commissioned an independent forensic data analysis by external biostatisticians, who have already verified that the final dataset does not contain the anomalies alleged. Their full report will be submitted to the journal.

Correction of Supplementary File: we will work with the journal to replace the placeholder file with the correct, validated dataset.

We welcome a transparent investigation by the journal and remain confident that a thorough review will vindicate the integrity of our work. In the meantime, we consider the provocative claims of data fabrication by Brown et al. [1] to be unsubstantiated and premature. We stand by the scientific validity of our findings.

Conclusion

The concerns raised in the Letter are based primarily on interpretations of numerical patterns rather than direct evidence of fabrication. While statistical anomalies may warrant examination, they do not constitute proof of misconduct.

We respectfully maintain that the available evidence does not support the conclusion that the dataset was fabricated, nor does it justify retraction of the article. We welcome independent institutional review and remain committed to complete transparency regarding the conduct and reporting of this study.

We also noted that one of the authors has several retracted articles, the last one resulted from double publication of the same article [3].