Association of Uterine Leiomyoma with Type 2 Diabetes Mellitus in Young Women: A Population-Based Cohort Study (Diabetes Metab J 2024;48:1105-13)

Article information

Diabetes Metab J. 2024;48(6):1183-1184
Publication date (electronic) : 2024 November 21
doi : https://doi.org/10.4093/dmj.2024.0681
1Department of Obstetrics and Gynecology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea
2Department of Internal Medicine, CHA Bundang Medical Center, CHA University School of Medicine, Seongnam, Korea
3Department of Statistics and Actuarial Science, Soongsil University, Seoul, Korea
4Department of Internal Medicine, Kangbuk Samsung Hospital, Sungkyunkwan University School of Medicine, Seoul, Korea
Corresponding author: Cheol-Young Park https://orcid.org/0000-0002-9415-9965 Department of Internal Medicine, Kangbuk Samsung Hospital, Sungkyunkwan University School of Medicine, 29 Saemunan-ro, Jongno-gu, Seoul 03181, Korea E-mail: cydoctor@skku.edu

We appreciate the interest and thoughtful feedback of Qin and Xu on our study [1] suggesting the accelerated failure time (AFT) model. However, after careful consideration, we believe that the Cox proportional hazards model was the most appropriate choice for our research given its alignment with our objectives and the characteristics of our dataset.

Our study focused on mandatory health insurance enrollees in South Korea, utilizing health insurance claims data to assess the relative risk of diabetes. The primary goal was not to model the hazard function or survival time directly but to evaluate relative risks between groups. As such, we employed the Cox proportional hazards model, which is widely recognized for its ability to provide interpretable relative risk estimates in such contexts.

While the AFT model offers advantages in certain situations— particularly when the proportional hazards assumption does not hold or when dealing with various forms of censored data—it requires the assumption that survival times follow a specific parametric distribution [2,3]. In our case, survival data derived from health insurance claims exhibit high variability, and it is unlikely that they conform consistently to any particular parametric distribution [4]. The AFT model could lead to inaccurate estimates, potentially misrepresenting the true risk patterns in our data. The model assumes relatively constant risks over time, which may not be appropriate for capturing the time-varying risks observed in real-world data. On the other hand, the Cox model, which allows such time-varying risks, was better suited to accurately reflect the complex nature of the dataset.

Moreover, the AFT model is primarily designed to estimate survival times, but it provides limited insight into relative risk, which was the main focus of our study [5,6]. Our aim was not to predict the time of onset of diabetes but rather to compare the relative risks of diabetes occurrence across groups. The Cox model has long been considered the standard method for evaluating relative risk in survival analysis when the proportional hazards assumption holds, and it was a reliable and robust approach for the objectives of our study.

Additionally, while the AFT model is beneficial for handling a variety of censored data types, we did not encounter complex censoring in our dataset. Thanks to the nature of the South Korean health insurance system, interval censoring was not a significant issue, and our data were predominantly right-censored. This further justified our choice of the Cox model, which is well-suited for analyzing right-censored data and comparing relative risks. Furthermore, we extended our analysis using the Cox model to successfully conduct cause-specific hazard modeling, such as examining competing risks for mortality.

In conclusion, while we acknowledge the flexibility and potential utility of the AFT model in certain contexts, the Cox proportional hazards model was the more suitable choice for our study given its alignment with our research objectives and the nature of our data. The Cox model provided clear and reliable interpretations of the relative risk of diabetes onset, and it performed well under the proportional hazards assumption.

We greatly appreciate your thoughtful suggestions and hope this clarifies the rationale behind our methodological approach. Thank you again for your engagement and valuable input.

Notes

CONFLICTS OF INTEREST

No potential conflict of interest relevant to this article was reported.

References

1. Sung JH, Kim KS, Han K, Park CY. Association of uterine leiomyoma with type 2 diabetes mellitus in young women: a population-based cohort study. Diabetes Metab J 2024;48:1105–13.
2. Zare A, Hosseini M, Mahmoodi M, Mohammad K, Zeraati H, Holakouie Naieni K. A comparison between accelerated failure-time and cox proportional hazard models in analyzing the survival of gastric cancer patients. Iran J Public Health 2015;44:1095–102.
3. Orbe J, Ferreira E, Nunez-Anton V. Comparing proportional hazards and accelerated failure time models for survival analysis. Stat Med 2002;21:3493–510.
4. Bradburn MJ, Clark TG, Love SB, Altman DG. Survival analysis part II: multivariate data analysis--an introduction to concepts and methods. Br J Cancer 2003;89:431–6.
5. Patel K, Kay R, Rowell L. Comparing proportional hazards and accelerated failure time models: an application in influenza. Pharm Stat 2006;5:213–24.
6. Mardhiah K, Wan-Arfah N, Naing NN, Hassan MRA, Chan HK. Comparison of Cox proportional hazards model, Cox proportional hazards with time-varying coefficients model, and lognormal accelerated failure time model. Asian Pac J Trop Med 2022;15:128–34.

Article information Continued