Review Reports - Real-World Adherence to Asthma and COPD Medications in Belgium: A Nationwide Analysis of Determinants Using Dispensing Data and Mixed-Effects Modeling

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

I have finished my review of the manuscript entitled "Real-World Adherence to Asthma and COPD Medications in Belgium: A Nationwide Analysis of Determinants Using Dispensing Data and Mixed-Effects Modelling"

It is about a nationwide pharmacoepidemiological analysis of adherence to asthma and COPD medications in Belgium using dispensing data from the Pharmanet database between 2020 and 2023

The topic is highly relevant for healthcare research because medication adherence remains a major challenge in chronic respiratory disease and chronic diseases in general, therefore adherence is esential for achieving optimal outcomes.

Nevertheless I have some concerns about the nterpretation of adherence estimates, this opens a door for possible misclassification bias. A model specification would be useful and also further explanatio of consistency between descriptive and sensitivity analyses. Tables still need clarity of some tables.

The study uses dispensation records to estimate medication adherence through CMA3. While this is a widely accepted proxy, it does not measure actual medication use. But what worries me more is that in discussion, authors tends to interpret adherence as real medication-taking behavior, which may overstate the conclusions.

Inference withous diagnostic codes avoids differentiation between asthma and COPD patients and it is a critical aspect because treatment patterns differ substantially between asthma and COPD and even if some medications are used for both diseases, adherence behavior may vary by disease severity and phenotype

Another relevant aspects is that adherence was calculated within an observation window (this may eliminate all records falling out of that window) starting 45 days after the first dispensation and requiring ≥3 dispensations. this is a strong limitation because it may introduce selection bias, because patients with fewer refills (possibly non-adherent patients) are excluded from the analysis and therefore never accounted for.

Odds Ratios are huge. Such high effect sizes may reflect confounding by disease severity or treatment indication, rather than true differences in adherence behavior. You may please include all goodness of fit tests and use adjusted and unadjusted models.

Year long and shorter follow-up windows exhibited larger sensitivity and it may be due to the shorter follow up period, but the manuscript should better explain this phenomenon. Authors should further expolaoin why the adherence nearly doubles in yearly analyses

Table 2 presents patient-class exposures, which may be confusing because the same patient appears multiple times. The 7unit for reference may not therefore be overestimated.

Once these observations are addressed, a new review would be adequate to establish if the manuscripts is improved enough.

Author Response

Dear Reviewer,
Please find attached the PDF file containing our detailed, point‑by‑point responses to your comments. We sincerely thank you for your careful review and constructive feedback, which have significantly improved the quality of our manuscript.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript addresses an important and timely topic, and the scale of the dataset and the use of mixed-effects modelling represent clear strengths of the study. So, congratulations to the authors. In my view only some aspects could benefit from further clarification and refinement.

1) While the results convincingly document low adherence and identify several predictors, it would be helpful if the authors could further clarify how this analysis advances current knowledge beyond confirming patterns already described in the literature.

2) The rationale for the selection of variables included in the mixed-effects model and the handling of potential confounders could be described more explicitly, which may help readers better understand the robustness of the findings.

3) Given that adherence was estimated using dispensing data rather than direct medication use, it may be useful to more clearly acknowledge the potential gap between medication possession and actual medication intake.

4) An interesting evidence is the very low adherence observed among children and adolescents. This finding might be further interpreted in light of the clinical and care complexity often associated with pediatric patients, whose care frequently requires higher levels of monitoring, coordination, and family involvement. Briefly acknowledging this aspect in the introduction or discussion may help better frame the implications of the findings. Relevant pediatric literature on determinants of hospitalization and care complexity of the recent years could be considered to support this point (PMID: 39803927).

5) The implications of these findings for healthcare systems and clinical practice, particularly regarding targeted adherence interventions for high-risk populations, could be discussed in greater depth.

Author Response

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript presents a comprehensive nationwide analysis of real-world adherence to asthma and COPD medications in Belgium, utilizing dispensing data from the Belgian Pharmanet database and mixed-effects modeling. I made the following comments to improve the manuscript’s quality.

Abstract

The abstract is dense and includes excessive details, such as the number of patients eligible for adherence assessment and the specific adherence rates for each pharmacological class. While these details are important, they could be summarized more concisely to improve readability.

The abstract could better emphasize the most critical findings, such as the identification of high-risk groups (children, adolescents, and ICS users) and the implications for clinical practice.

Introduction

Focus on the most relevant statistics (on adherence) and reduce the emphasis on general prevalence and mortality data.

Expand on the consequences of poor adherence, such as reduced quality of life, increased exacerbations, and higher healthcare costs.

Emphasize how the study’s use of dispensing data and mixed-effects modelling provides novel insights into adherence patterns and determinants.

Method

The Methods section could benefit from clearer subheadings and a more structured flow (e.g., study design, participants, instruments, statistical analysis).

While the authors mention the removal of implausible values (e.g., negative reimbursement or copayment), they do not provide detailed criteria or examples of how these values were identified and excluded. More information on the data cleaning process would improve replicability.

While the use of mixed-effects models is described, the authors do not provide details on the random-effects structure, model diagnostics, or the assumptions made during the analysis. Including this information would improve clarity and transparency.

Explain how missing data were addressed, including whether imputation methods were used or if missing data were excluded entirely.

Results

Table 1: The percentages for Sex (Female: 52.5%, Male: 48.6%) do not add up to 100%. The total is 101.1%, which is incorrect. Ensure that the percentages for "Administration Route" sum to 100%. This suggests a rounding error or a miscalculation. Administration Route: Oral percentage should be corrected to 2.9% from 2,9%. The table does not provide the total number of patients or dispensations for reference. Including these totals would help verify the accuracy of the percentages and provide context for the data.

Line 179: “LABA+LAMA+ICS” is mentioned; however, in Table 1, it is mentioned as “ICS + LAMA + LABA”. The authors should be consistent.

Figure 1 is not needed as it repeats the data presented in Table 2.

Table 3: While the table states that all associations are statistically significant (p < 0.001), this information is only mentioned in the footnote. It would be helpful to include a column or marker in the table itself to highlight statistical significance.

Figure 2 is not needed as it repeats the data presented in Table 3.

Discussion

The authors mention guideline-concordant strategies (e.g., regimen simplification, digital tools). Provide specific examples of how these strategies can be applied in clinical practice, such as integrating digital adherence tools into routine care or designing educational programs for high-risk groups.

Expand the limitations section to discuss how dispensing data might overestimate or underestimate adherence, the implications of missing diagnostic coding, and the potential biases introduced by using WHO-defined DDDs for pediatric populations.

Compare the adherence rates to those reported in other countries or studies to provide context and highlight the significance of the findings.

Provide a more detailed analysis of behavioral barriers and discuss evidence-based strategies to address them, such as motivational interviewing or behavioral interventions.

The author briefly mentions socioeconomic factors but does not explore their role in adherence in detail. Discuss how factors like income, health literacy, and access to healthcare may influence adherence and suggest strategies to address these barriers.

Conclusion

The conclusion seems to overestimate the current findings. The authors can refine it as follows. The authors state that the study "provides a population-level evidence base to inform health system planning and resource allocation strategies" and "lays the groundwork for developing patient-fitted tools and strategies to improve long-term medication continuity." While the study identifies adherence patterns and high-risk groups, it does not provide direct evidence or test interventions that could improve adherence. The findings are descriptive and observational, and while they offer valuable insights, they do not directly demonstrate the effectiveness of specific strategies or tools.

Author Response

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Authors have significantly improved the manuscript. The submission now provides an explanation regarding higher adherence estimates in yearly analyses compared with the full observation window has been clarified and linked to shorter follow-up periods.

Also, authors added a clearer discussion of confounding by indication and disease severity, particularly for medication classes such as triple therapy and montelukast.

Even when the kimitation about dispensing based analysis is still present, in the revised manuscript authors now explain the eligibility criteria requiring ≥3 dispensations and a minimum observation window, and the authors now clarified how this affects adherence estimation.

the explanation that the diagnostic information was unavailable in the Pharmanet database justifies the analysis of patients treated with respiratory medications rather than confirmed disease categories.

After small edits of typos, manuscript is ready for acceptance, congratulations!

Author Response

Dear Reviewer,

Thank you very much for your positive assessment of our revised manuscript. We are pleased that the clarifications regarding adherence estimates, confounding by indication, dispensing‑based limitations, and the absence of diagnostic information were satisfactory.

We also appreciate your final suggestions, which have now been implemented.
Thank you for recommending the manuscript for acceptance.

Reviewer 2 Report

Comments and Suggestions for Authors

Congratulations to the authors for the substantial improvements made to the manuscript; the revised version is methodologically stronger, more transparent, and represents a solid scientific contribution.

Author Response

Dear Reviewer,

Thank you very much for your positive and encouraging feedback. We are grateful that you found the revised manuscript methodologically stronger and more transparent. We truly appreciate your supportive assessment.

Reviewer 3 Report

Comments and Suggestions for Authors I appreciate the author's effort in revising the manuscript; however, a few comments yet to be addressed and needs appropriate revision in the manuscript. In Table 1, the percentages for "Sex" (Female: 52.5%, Male: 48.6%) do not sum to 100%. This discrepancy should be corrected or explained (e.g., rounding errors). The author uses "LABA+LAMA+ICS" and "ICS+LAMA+LABA" interchangeably. While the authors mentioned correcting this inconsistency, it is important to ensure that the correction has been applied consistently throughout the manuscript. There are some formatting inconsistencies, such as using commas instead of decimal points (e.g., "2,9%" instead of "2.9%"). Ensure that all numerical values are formatted consistently. Both Figure 1 and Table 2 present similar data. While the authors justified retaining Figure 1 for visual clarity, it may still be worth considering whether this repetition is necessary or if one of the elements could be removed.

Author Response

Dear reviewer,

We thank you once again for the opportunity to submit a further revised version of our manuscript entitled “Real-World Adherence to Asthma and COPD Medications in Belgium: A Nationwide Analysis of Determinants Using Dispensing Data and Mixed-Effects Modelling”. We appreciate the continued time and effort you and the reviewers have dedicated to evaluating our work.

We are grateful for the insightful and constructive comments provided throughout the review process, which have significantly strengthened the manuscript. In this latest revision, we have carefully addressed all remaining concerns and incorporated the suggested changes.

Please find below our detailed, point‑by‑point responses to each comment. All page numbers refer to the revised manuscript. Additions in the manuscript appear in red.

__________________________________________________________________________________

Comment 1: [I appreciate the author's effort in revising the manuscript; however, a few comments yet to be addressed and needs appropriate revision in the manuscript. In Table 1, the percentages for "Sex" (Female: 52.5%, Male: 48.6%) do not sum to 100%. This discrepancy should be corrected or explained (e.g., rounding errors). The author uses "LABA+LAMA+ICS" and "ICS+LAMA+LABA" interchangeably. While the authors mentioned correcting this inconsistency, it is important to ensure that the correction has been applied consistently throughout the manuscript. There are some formatting inconsistencies, such as using commas instead of decimal points (e.g., "2,9%" instead of "2.9%"). Ensure that all numerical values are formatted consistently. Both Figure 1 and Table 2 present similar data. While the authors justified retaining Figure 1 for visual clarity, it may still be worth considering whether this repetition is necessary or if one of the elements could be removed.”

Thank you very much for your constructive feedback. We carefully reviewed the manuscript again and addressed all remaining inconsistencies:

Sex proportions in Table 1: The discrepancy in percentages has been corrected.
Consistency in terminology (LABA+LAMA+ICS vs. ICS+LAMA+LABA): We thoroughly checked the entire manuscript and ensured that the terminology is now used consistently throughout.
Numeric formatting: All decimal separators have been harmonized, and the manuscript now consistently uses decimal points instead of commas.
Overlap between Figure 1 and Table 2: Following your suggestion, we have removed Figure 2 from the manuscript to avoid redundancy and improve overall clarity.

We appreciate your careful review and believe these revisions significantly strengthen the manuscript.