Highlights
- Latent Class Analysis (LCA) identified four distinct clinical phenotypes among Chinese individuals with type 2 diabetes (T2D) who developed incident chronic kidney disease (CKD).
- The highest healthcare costs were observed in the young-onset T2D group (Class 1), with mean costs reaching US$6087 per-patient-per-year (PPPY).
- A middle-aged group with low baseline comorbidities (Class 3) experienced a significant cost surge during CKD onset, largely driven by outpatient and psychiatric care.
- Incident CKD occurred at a rate of 26.29 per 1000 person-years, emphasizing the urgent need for targeted early intervention strategies.
Background
Type 2 diabetes mellitus (T2D) is a global epidemic, and chronic kidney disease (CKD) remains its most prevalent and economically burdensome complication. In Asia, and specifically within the high-density urban environment of Hong Kong, the prevalence of T2D has risen alongside an aging population and changing lifestyles. CKD in the context of diabetes not only significantly increases cardiovascular risk but also accelerates the progression toward end-stage kidney disease (ESKD), necessitating costly interventions such as dialysis and transplantation.
The healthcare system in Hong Kong, dominated by the public-sector Hospital Authority (HA) which provides approximately 90% of inpatient care, offers a unique opportunity for longitudinal economic analysis. Despite the availability of structured care, the economic impact of CKD transition is not uniform. Traditional risk stratification often fails to capture the heterogeneity of T2D patients. Identifying specific “phenotypes” or “latent classes” of patients who are likely to incur higher costs during the transition to CKD is critical for health policy and the implementation of precision medicine. This study by Du et al. utilizes the Hong Kong Diabetes Register (HKDR) to apply advanced statistical modeling—Latent Class Analysis (LCA)—to unravel these complex cost trajectories.
Key Content
Methodological Framework: The Hong Kong Diabetes Register (HKDR)
The study analyzed data from 2886 individuals with T2D and incident CKD selected from the prospective HKDR cohort between 2007 and 2019. The HKDR is a world-class registry that has tracked the clinical outcomes of thousands of patients since its inception in 1995. For this specific analysis, participants were required to have complete data for 42 baseline attributes, ensuring a robust foundation for LCA.
LCA is a person-centered statistical technique that identifies unobservable subgroups within a population based on shared characteristics. In this study, 14 variables—including age at onset, metabolic markers (HbA1c, blood pressure, lipids), and renal function (eGFR, ACR)—were used to classify participants. Following the identification of these classes, a hierarchical generalized linear mixed model (HGLMM) was employed to evaluate longitudinal healthcare costs, accounting for the repeated measures within individuals over 109,784 person-years of follow-up.
Characterization of the Four Latent Classes
The LCA revealed four distinct classes, each representing a unique clinical and economic profile:
- Class 1 (Young-Onset, High Intensity): Comprising 18.3% of the cohort, these individuals had an average onset age of 44.4 years. Despite their youth, they exhibited moderate comorbidities (25.6% high Elixhauser Comorbidity Index [ECI]) and heavy medication use (90.2% on ≥3 medications). This group incurred the highest PPPY cost (US$6087).
- Class 2 (Old-Age Onset): Making up 21.2% of the cohort, these patients were older at onset (66.9 years) with moderate comorbidities. Their costs were relatively lower (US$3822 PPPY), reflecting a more traditional geriatric diabetes management profile.
- Class 3 (Middle-Aged, Low Comorbidity): This was the largest group (33.9%), with an onset age of 54.2 years. At baseline, they had few comorbidities and low medication intensity. However, they exhibited a sharp increase in costs (US$4260 PPPY) upon the development of CKD.
- Class 4 (Middle-Aged, Moderate Comorbidity): 26.5% of the cohort, similar in age to Class 3 (54.1 years) but with significantly higher baseline medication use (98.9%) and moderate comorbidities. Their cost was US$3923 PPPY.
Cost Drivers and Temporal Trajectories
The analysis showed that the year of CKD onset is a critical financial inflection point. Class 1 and Class 3 showed the most dramatic increases during this period. In Class 1 (young-onset), the high costs were distributed across multiple sectors, including inpatient stays, specialized outpatient clinics, and emergency services. This suggests a systemic multi-organ impact of early-onset diabetes that manifests acutely when renal function begins to decline.
In contrast, the cost drivers for Class 3 were surprisingly focused. Despite having a “healthier” baseline, their post-CKD costs were largely attributed to outpatient services and, notably, psychiatric care. This suggests that the psychological burden of a new CKD diagnosis in a previously low-risk middle-aged individual may be a significant, yet often overlooked, contributor to healthcare utilization.
Expert Commentary
The findings by Du et al. provide a profound shift in how we view the economic burden of diabetic complications. Traditionally, clinical focus and resource allocation have been directed toward the elderly or those with the highest number of comorbidities. However, this study underscores that the greatest economic “shock” to the healthcare system comes from the young-onset population and a specific subset of middle-aged patients who were previously considered low-risk.
The high cost of Class 1 (young-onset) aligns with the growing body of evidence that young-onset T2D is a more aggressive metabolic phenotype, characterized by rapid beta-cell decline and early development of complications. From a health policy perspective, this justifies aggressive early intervention and the use of newer, more expensive therapies (like SGLT2 inhibitors and GLP-1 receptor agonists) in younger patients to prevent the catastrophic costs associated with early CKD transition.
Perhaps the most intriguing finding is the high psychiatric and outpatient cost in Class 3. This highlights a critical gap in our current management of CKD: the lack of integrated psychosocial support. When middle-aged, working-class individuals—who may have perceived their diabetes as well-controlled—suddenly face the reality of CKD, the mental health impact is substantial. Integrating psychiatric screening and support into renal clinics could potentially mitigate some of these costs and improve overall patient quality of life.
One limitation of the study is its focus on the public healthcare sector in Hong Kong. While the HA covers the vast majority of chronic disease care, it may not fully capture the costs of medications purchased in the private sector or the indirect costs (loss of productivity), which are likely to be even higher in the younger cohorts of Class 1 and Class 3.
Conclusion
This latent class trajectory analysis demonstrates that the economic burden of CKD in T2D is highly heterogeneous. Individuals with young-onset T2D represent the highest-cost group, while middle-aged individuals with low initial comorbidity profiles experience significant surges in outpatient and psychiatric costs upon CKD development. These insights advocate for a more nuanced, phenotype-driven approach to diabetes management. Future research should focus on whether early intensive metabolic control and integrated psychosocial care can flatten these cost trajectories and improve long-term renal outcomes in these high-risk clusters.
References
- Du Y, Zhang M, Li AQY, et al. Differential healthcare costs in individuals with type 2 diabetes and incident chronic kidney disease in Hong Kong: a latent class trajectory analysis. Diabetologia. 2026. PMID: 41848900.
- Chan JCN, Lim LL, Luk AOY, et al. The Hong Kong Diabetes Register: 25 years of research and healthcare evolution. Diabetes Obes Metab. 2020. PMID: 32363780.
- Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998. PMID: 9420519.
