Response to the Letter to the Editor Regarding “Knowledge and Attitudes Toward Attention-Deficit/Hyperactivity Disorder Among Physicians and Senior Medical Students in Jordan” [Response to Letter]

Dear editor

We thank Majeed, Chowdhury, and Pandit for their thoughtful engagement with our article.¹ Their Letter raises methodological considerations concerning scoring, variable selection, ordinal regression, subgroup precision, interpretation of the knowledge–attitude relationship, web-survey recruitment, and theoretical framing.² Several issues overlap with previous correspondence and our published Response.³ ˒⁴ We address both the overlapping and the new points below.

Rationale for the Three-Point Scoring System

The three-point scoring system was predefined to distinguish three educationally meaningful response states: correct knowledge (+1), acknowledged uncertainty (“not sure” = 0), and incorrect knowledge or misconception (−1). This distinction is relevant in health-professions education because acknowledged uncertainty and an incorrect belief are not necessarily equivalent educational states. Related assessment approaches have used positive credit for correct responses, no penalty for acknowledged uncertainty, and negative marking for incorrect responses to distinguish lack of knowledge from incorrect knowledge and reduce random guessing.⁵ ˒⁶ Similar three-state scoring logic has also been applied in health-professions education research.⁷

For completeness, the predefined knowledge thresholds, already reported in our previous Response, were >12 for good knowledge, 1–12 for fair knowledge, and ≤0 for poor knowledge;⁴ these allow readers to verify the reported 55.2%/37.6%/7.2% distribution. They were abbreviated in the article as “predefined cutoffs”. Importantly, the Letter’s statement that the system “effectively treats a confidently wrong answer the same as an uncertain one” is inconsistent with the scoring used: an incorrect response received −1, whereas “not sure” received 0.²

Variable Selection and Multivariable Modeling

We acknowledge that variable-selection strategies have limitations and that bivariate significance should not substitute for theory-based confounder selection when causal inference is the objective.⁸ ˒⁹ Our analyses were exploratory and explanatory rather than causal or predictive. Bivariate screening was used pragmatically, and the adjusted estimates should be interpreted as statistical associations within this cross-sectional sample, not as causal effects or coefficients from a clinical prediction model. The models were nonetheless reported with the relevant quality checks: adjusted odds ratios with 95% confidence intervals, model-fit statistics, and a formal test of the proportional-odds assumption.¹ A theory-driven sensitivity analysis is essential for a confirmatory causal or prediction model; for the present exploratory design, it is a valuable extension for future work rather than a requirement, and its absence does not constitute a methodological error.

Ordinal Regression and Pediatric Rotation

The significant test of parallel lines was explicitly evaluated and reported in the original article (χ²(4)=20.726, p<0.001), indicating that the proportional-odds assumption was not fully supported; the student-model estimates were therefore identified as requiring cautious interpretation.¹ Ordinal logistic regression was selected because knowledge was classified into three ordered levels—poor, fair, and good.¹⁰ Partial proportional-odds and generalized ordered-logit models are alternatives that retain this ordinal structure when proportional odds are not satisfied,¹¹ and are appropriate for future confirmatory analyses. Accordingly, the adjusted common odds ratio should not be interpreted as a uniform effect across all knowledge thresholds.

The Letter also appears to attribute the main pediatric-rotation finding to the 23 students who were currently in rotation.² The statistically significant adjusted comparison, however, concerned students who had completed pediatric rotation vs those who had not completed it. The completed-rotation group included 144 students, of whom 110 (76.4%) had good ADHD-related knowledge, compared with 32 of 123 (26.0%) who had not completed the rotation (p<0.001; Cramér’s V=0.36).¹ In the multivariable model, completed vs not completed rotation was significant (OR=3.370, 95% CI 1.075–10.561; p=0.037), whereas currently in rotation vs not completed was not (OR=3.422, 95% CI 0.839–13.956; p=0.086).¹ Nevertheless, the cross-sectional design and violated proportional-odds assumption require caution, and we regard the association as hypothesis-generating rather than causal.

Physician Subgroup and Parenthood

The association between having children and knowledge had a wide confidence interval (OR=3.223, 95% CI 1.003–10.363; p=0.049) and should be interpreted cautiously.¹ In the crude analysis, physicians with children had a lower proportion of good knowledge than those without children (18/52, 34.6% vs 127/211, 60.2%; p=0.004; Cramér’s V=0.21).¹ This reversal between the crude distribution and the adjusted estimate likely reflects instability arising from the relatively small subgroup, overlap with marital status, and adjustment for other physician characteristics.¹² This was explicitly acknowledged in the published Discussion, which noted that the small subgroup of participants with children “may have contributed to instability between crude subgroup proportions and adjusted regression estimates”.¹ Parental status should therefore be regarded as an exploratory adjusted association rather than a definitive independent predictor, and the physician interpretation rests on the broader pattern of contextual and experiential factors rather than on this single estimate.

The Knowledge–Attitude Gap

The term “knowledge–attitude gap” was not intended to deny an association between knowledge and attitude; the article explicitly reported that higher knowledge was associated with more positive attitudes (χ²=18.29, p<0.001).¹ Rather, the term describes the incomplete translation of knowledge into consistently positive attitudes: relatively favorable knowledge coexisted with negative attitudes in 47.2% of participants and with persistent misconceptions.¹ The divergence in predictors was presented as consistent with this interpretation, not as proof that knowledge fails to shape attitude. Among students, year of study and pediatric and psychiatry rotations were associated with knowledge but not with attitude, and no participant characteristic independently predicted attitudes.¹ We agree with Majeed et al that unmeasured factors may influence both outcomes;² the Discussion already identified personal beliefs, cultural interpretations, clinical experiences, stigma, and the hidden curriculum as plausible influences.¹ The appropriate interpretation is therefore that knowledge and attitudes are related but not interchangeable; causal pathways require longitudinal or interventional investigation.

Convenience Sampling, Self-Selection, and CHERRIES

CHERRIES provides a useful framework for reporting internet-based surveys.¹³ The article reported its core elements, including design, eligibility, recruitment channels, voluntary participation, electronic consent, anonymity, questionnaire structure, psychometrics, scoring, analysis, and limitations.¹ The elements not reported were portal-dependent web-analytics metrics. Because recruitment occurred through multiple overlapping university, professional, and social-media channels, the number of invited individuals could not be established, and unique visits, view, participation, and completion rates, and duplicate-entry tracking could not be reliably obtained.¹ We acknowledge this as a reporting limitation. The article also acknowledged that convenience sampling may have introduced selection bias and limited representativeness.¹ We agree that self-selection may affect attitudes more than factual knowledge and that its direction and magnitude cannot be determined from these data. Accordingly, the 52.8% positive and 47.2% negative attitude distribution describes this sample rather than national prevalence; this qualifies generalizability but does not invalidate the association-based findings.

Theoretical Framework and Individual Attitude Items

The study was designed as an applied cross-sectional medical-education investigation, not as a formal theory-testing study of attitude formation. The composite attitude score was therefore an applied summary of overall attitude direction rather than a multidimensional latent construct. We agree that frameworks such as the Theory of Planned Behavior, which distinguishes attitudes, subjective norms, and perceived behavioral control, potentially combined with stigma theory, could strengthen future theory-driven research.¹⁴ The article already recognized that attitudes may reflect personal beliefs, cultural interpretations, clinical experiences, stigma, and the hidden curriculum, and it called for longitudinal and interventional studies addressing both knowledge and professional attitudes.¹ Such frameworks therefore represent a valuable next step rather than a requirement for the validity of the present findings.

The interpretation of individual items was deliberately differentiated. The article reported that 20.8% considered ADHD a “fashionable” disorder, 29.8% believed it may be used as an excuse for poor behavior, and 50.6% perceived parental reluctance to administer medication.¹ Beliefs questioning the legitimacy of ADHD were interpreted as attitudes with potential consequences for referral and counseling, whereas perceived parental reluctance was explicitly not treated as a simple misconception, because it may reflect awareness of real-world barriers such as adverse-effect concerns, stigma, and treatment acceptability.¹ We agree that closed-ended items cannot determine the reasoning underlying each response and that cognitive interviewing or qualitative approaches would add depth in future studies;¹⁵ their absence, however, does not invalidate the closed-ended findings, which identify educationally relevant attitudinal signals.

Conclusion

The scoring framework explicitly distinguished uncertainty from misconception; the proportional-odds limitation was reported and constrains interpretation of the student model; the significant pediatric-rotation comparison concerned completed rather than current rotation; and the parental-status estimate is exploratory, as the original Discussion acknowledged. The knowledge–attitude gap is a descriptive educational interpretation, not a claim of independence or causality. These clarifications do not alter the study’s findings or conclusions; they define their scope and support future longitudinal, theory-informed investigation.

Disclosure

The authors report no conflicts of interest in this communication.