Letter to the editor in response to “Glycemic control is worse in rural compared to urban type 2 diabetes in Bangladesh, irrespective of food security status”

Abstract

Dear Editor,

We read the recent article titled “Glycemic control is worse in rural compared to urban type 2 diabetes in Bangladesh, irrespective of food security” status¹ with great interest. This study represents the first systematic exploration in Bangladesh (a lower-middle-income country, LMIC) of the associations between food insecurity (FIS), place of residence (rural/urban disparity), and glycemic control in type 2 diabetes, filling a significant data gap for developing countries in this field. Given the distinct challenges in healthcare resource allocation, social structure, and food security in LMICs compared to the high-income countries where previous similar research has largely focused, this study provides crucial evidence for global diabetes health equity. However, key methodological limitations necessitate clarification to ensure robust interpretation of the findings.

We have significant methodological concerns regarding the interpretation of the key finding—the independent effect of rural residence. The authors adjusted only for age, sex, and BMI in their regression models. Yet, table 1 clearly demonstrates systemic socioeconomic disadvantages in the rural group: only 3.9% had higher education vs 24.0% in the urban group (P < 0.001); median annual household income was $2,336 in rural areas vs $5,607 in urban areas (P < 0.001); and only 7.9% of the rural group were professionals compared to 28.3% in the urban group (P < 0.001). Compounding this, as mentioned in the Introduction, rural residents needed to travel long distances to urban specialty clinics for care. These factors—education, income, occupation, and healthcare access—directly impact diabetes management capabilities, such as medication affordability, health literacy, and frequency of follow-up visits. Consequently, had variables like education, income, and occupation been included in the model, the reported effect size attributed to residence alone (β = 1.4, P < 0.001) might have been substantially attenuated or even disappeared. While the authors claim rural residence is an independent factor, the failure to control for these critical confounders casts doubt on the credibility of this conclusion, and we strongly recommend re-running the regression models incorporating these socioeconomic variables to verify if the residence effect remains significant.

Moreover, significant sample selection bias is a concern. The study exclusively recruited patients from specialty diabetes clinics (BADAS clinics), yet the Introduction notes that primary healthcare resources are severely lacking in rural Bangladesh. This recruitment strategy likely introduces a critical bias: rural patients able to attend urban specialty clinics probably represent a subgroup with relatively better economic and/or educational status, as they can bear the associated travel costs and time commitments. This suggests the study may underestimate the true glycemic levels of the broader rural diabetic population, as the most impoverished and/or severely ill rural patients are likely unable to access these clinics and are thus excluded from the sample. Therefore, this sample cannot be considered representative of the rural diabetic population in Bangladesh. This limitation must be explicitly stated in the manuscript, clarifying that the conclusions may only apply to “diabetic patients able to access specialty care” and not to the entire rural patient population.

In summary, while this study importantly highlights the grim reality of poorer glycemic control among rural diabetic patients in Bangladesh, its conclusions must be interpreted with considerable caution. The potential overestimation of the “independent” effect of residence due to inadequate control for key socioeconomic confounders, combined with the inability of the specialty clinic-based sample to represent the most vulnerable rural patients in resource-poor settings, limits the generalizability of the findings. Future research needs to incorporate multi-level social determinants of health and expand recruitment to primary healthcare settings to avoid selection bias, enabling the accurate identification of intervention targets to effectively advance health equity.

The authors declare no conflict of interest.

Approval of the research protocol: N/A.

Informed consent: N/A.

Approval date of registry and the registration no. of the study/trial: N/A.

Animal Studies: N/A.