This study provides among the first quantitative evidence from Bangladesh on SDM in cancer care, revealing a gap between patients' high preference for SDM and their low rate of actual participation. We identified younger age, formal education, and employment as significant predictors of involvement, and found a positive association between SDM participation and patient satisfaction.

This disparity between preference and practice is not unique but may be especially pronounced in Bangladesh, related to systemic pressures, cultural norms, and patient-level barriers inherent to a resource-constrained setting (Human Resources for Health 2022). This gap reflects the severe limitations of the country’s healthcare system, where the ideal of patient-centered communication is constrained by a physician-to-patient ratio of only 0.67 per 1000, an average consultation time of 48 seconds (Irving et al. 2017), and overwhelming patient volumes, as seen in institutions like NICRH (National Institute of Cancer Research Hospital (NICRH) 2025). In such an environment, the default mode of interaction reverts to a traditional, time-efficient paternalistic model (Resnicow et al. 2022), a finding is consistent with evidence from other LMICs (Salek et al. 2023; Ozdemir et al. 2021), where systemic, cultural and structural barriers often prevent the implementation of patient-centered care.

The strong correlation between formal education and SDM participation may serve as a proxy for health literacy in Bangladesh. In a country where 43% of patients with chronic diseases have inadequate health literacy, those with less education often face a significant power imbalance in clinical encounters (Salwa et al. 2024). Formal education provides patients with a repertoire of skills, confidence, and verbal competencies that may be more likely to facilitate navigation of patient-physician dynamics (Shahid et al. 2022; Ryman et al. 2024). This highlights that patient empowerment in Bangladesh is fundamentally a health literacy challenge, which clinicians can address by avoiding medical jargon, using “teach-back” approach (Shersher et al. 2021) and fostering an environment that encourages participation. Simply providing more information may be insufficient; interventions must be tailored to be understandable and actionable across diverse literacy levels.

A striking finding is the marked decline in SDM participation with increasing age signals more than a simple demographic trend, pointing to an ongoing cultural shift in patient expectations in Bangladesh. This suggests that younger generations, with greater exposure to global information and modern concepts of patient rights, appear to be challenging the traditional paternalistic model of healthcare (Cecconi et al. 2025). In contrast, older patients may adhere to long-standing cultural norms that view the physician as an unequivocal authority figure whose judgment should not be questioned (Schulman-Green et al. 2006) and often expect a more paternalistic relationship with their physicians, which can adversely influence their engagement in SDM processes (Pel-Littel et al. 2021). Physicians must navigate vastly different expectations simultaneously, suggesting that a one-size-fits-all approach (Hawley and Jagsi 2015) to SDM may be inadequate.

Employment status was associated with SDM, with working patients having over four times the odds of participation. In the Bangladeshi context, formal employment may serve as a proxy for more than just socioeconomic status, reflecting personal agency, social capital, and the confidence to advocate for oneself (Uddin 2018). Non-working patients, a group that in this context predominantly includes women engaged in domestic roles, may face multifaceted barriers such as financial dependency, possibly lower social standing, and limited experience in formal negotiations, which collectively restrict their ability to engage in complex treatment decisions (Hossain et al. 2020; Masoom 2024), highlighting a significant disparity in social empowerment. Therefore, achieving truly patient-centered cancer care in Bangladesh requires targeted interventions to empower non-working patients, ensuring that all individuals, regardless of their employment status, have an equitable opportunity to participate in their own healthcare.

Finally, the findings on patient satisfaction highlight that patients who preferred either fully autonomous (patient-controlled) or fully passive roles (physician-controlled) were overwhelmingly dissatisfied, suggesting that such preferences may reflect flaws in the healthcare system rather than true desires (Kehl et al. 2015). Patients seeking full control may feel this way due to limited access to information and a paternalistic care environment, while those preferring no control may be reacting to rushed and impersonal interactions that leave them feeling ignored (Williams et al. 2020). These patterns indicate that the key driver of satisfaction is not the extreme of autonomy or passivity, but the practice of SDM, where open communication, mutual respect, and collaboration shape care (Hienen et al. 2025). The goal in a setting like Bangladesh, therefore, is not only to acknowledge stated preferences but to foster an environment where patients naturally feel empowered to engage in shared treatment roles.

These findings point to several actionable policy implications for the Bangladeshi Ministry of Health and cancer care institutions. First, these data support the need for culturally tailored SDM communication training in postgraduate medical curricula and continuing medical education for oncologists. These modules should focus on practical skills for low-literacy contexts, such as the 'teach-back' method, along with strategies to effectively engage patients across different generations. Second, our data on patient preferences for information support a call for the development and dissemination of Bengali-language decision aids. These should be prioritized for the most common cancers identified in our sample, notably digestive and female reproductive cancers, and should incorporate the formats patients most desire: simple written text and lists of questions to ask their physician.

This study’s conclusions must be interpreted in light of several significant limitations. First, and most critically, are the methodological constraints related to the measurements presented. The primary outcome, patient participation in SDM was assessed using a single, unvalidated binary question. This approach threatens the construct validity of our findings, as it oversimplifies the complex, multi-step process of SDM and severely limits the comparability of our results with the international literature that relies on validated, multi-item scales. Similarly, health literacy, a central construct in our interpretation, was not directly measured; education was used as a proxy. This prevents a direct test of our hypothesis that health literacy mediates the relationship between education and SDM participation. Furthermore, the bespoke questionnaire lacked formal psychometric validation; however, it was informed by existing literature and pilot-tested to ensure clarity and appropriateness for the study population.

Second, the study design limits the generalizability and causal interpretation of the results. The use of a convenience sample from a single, national tertiary cancer hospital in the capital city, which means the findings may not be representative of the broader population of patients with cancer in Bangladesh, particularly those treated in rural, district-level, or private facilities who may face even greater barriers to care. The cross-sectional design also prevents any causal inferences from being drawn between the identified predictors and SDM participation. Furthermore, as data on patients who declined to participate were not collected, we cannot assess the potential for non-response bias.

Third, our statistical analysis has weaknesses that affect the precision of the findings. The wide confidence interval for the effect of employment indicates a degree of statistical imprecision, likely due to the small number of employed participants in our sample. Additionally, the dichotomization of the 5-point satisfaction scale, while creating a clinically interpretable outcome, resulted in a loss of data granularity and statistical power. Reliance on self-reported data also introduces the potential for recall and social desirability biases.

Finally, the exclusion of severely ill patients, which may have underrepresented a subgroup facing greater challenges in SDM. Similarly, by grouping homemakers, students, and retirees with unemployed individuals, our categorization of employment status may have limited the accuracy of socioeconomic interpretations of the results. Thus, the combined effect of possible measurement error, non-probability sampling, and exclusion criteria on internal and external validity should nuance the interpretation of our findings, and underscore the need for further, more generalizable study of health services in Bangladesh.