Operational and communication determinants of patient satisfaction: A cross-sectional analysis framed by Donabedian’s model

INTRODUCTION

The purpose of this study is to identify determinants of outpatient satisfaction in a government sub-divisional hospital in West Bengal using the Donabedian framework.

Patient satisfaction is increasingly recognized as a critical indicator of healthcare quality and system responsiveness. Sub-divisional hospitals in India constitute essential secondary-level facilities, providing outpatient care to large catchment populations. Understanding patient experience in such facilities is central to improving service quality under the National Health Mission (NHM) and national quality assurance standards (NQAS).

Previous Indian studies have shown that factors such as doctor–patient communication, waiting time,^[1,2] pharmacy services, and environmental cleanliness strongly influence outpatient satisfaction.^[3,4] However, research specifically focused on sub-divisional hospitals remains limited. This study used the Donabedian framework to identify predictors of outpatient satisfaction in a sub-divisional hospital and to assess the internal consistency of service-quality domains aligned with NQAS. The objective of this study is to identify determinants of outpatient satisfaction in a sub-divisional hospital using Donabedian’s model.

MATERIALS AND METHODS

A cross-sectional descriptive study was conducted in the outpatient department (OPD) of a government sub-divisional hospital in West Bengal over a pre-specified 2-month period. The hospital functions as a secondary-level public facility serving urban and peri-urban communities. Consecutive adult patients aged 18 years or older who had completed their consultation were invited to participate. This sampling approach was selected to reflect routine patient flow and real-world operational conditions. After data cleaning and exclusion of incomplete responses, 273 complete cases were available for analysis.

Data were collected using a structured questionnaire adapted from validated Indian patient-satisfaction instruments and aligned explicitly with NQAS service-quality domains, including communication, patient experience with pharmacy, environmental cleanliness and physical amenities, and administrative processes. All items were scored on a 5-point Likert scale. The instrument captured sociodemographic characteristics (age, sex, education, marital status, occupation, and household income), along with detailed assessments of doctor–patient communication, opportunity to ask questions, clarity of explanations, understanding of medication instructions, pharmacy experience, cleanliness, availability of drinking water and toilets, signage, waiting-area comfort, and efficiency of registration processes. The questionnaire used in this study was adapted from previously validated Indian outpatient satisfaction instruments and structured to align with the NQAS domains (communication, environment, pharmacy, and administrative processes). Content relevance was reviewed by hospital administrative and clinical supervisors as part of routine quality-improvement processes. Reliability of multi-item domains was assessed using Cronbach’s alpha in the present dataset. As this was a minimal-risk, de-identified service-evaluation activity, external expert validation was not mandated under institutional or Indian Council of Medical Research - National Health Mission Guidelines (ICMR-NHM) guidelines. The full questionnaire is provided as Supplementary 1. The primary outcome variable, high satisfaction, was defined as selecting the maximum score (5) on the overall satisfaction item. The medication-understanding item (“I understand how and when to take my medicines”) was coded dichotomously for regression analyses (full understanding = 5 vs. less than full understanding <5). Education level was treated as an ordinal variable. Domain internal consistency was assessed using Cronbach’s alpha: communication (α = 0.547), environment (α = 0.468), pharmacy (α = 0.293, reflecting its two-item structure), and administrative process (not applicable, single-item domain). Lower alpha values are expected in short service-quality subscales and do not invalidate domain-level interpretability in patient-experience studies.

The study was conducted as part of routine, minimal-risk quality-improvement activities in accordance with institutional and NHM guidance. Hospital administration and clinical leadership provided verbal approval for the assessment, and no identifiable information was collected. As per the ICMR and NHM norms, de-identified service-evaluation activities that do not involve intervention or personal identifiers are eligible for ethics committee waiver. Verbal informed consent was obtained from all participants, and confidentiality was maintained throughout.

Descriptive statistics were used to summarize participant characteristics and item distributions. Bivariate associations between categorical variables and high satisfaction were examined using Chi-square tests with pairwise complete observations to reduce bias associated with listwise deletion. Logistic regression models were developed to identify independent predictors of high satisfaction. Because only 37 respondents fell into the not-highly-satisfied category, multivariable modeling followed established events-per-variable (EPV) guidance to minimize overfitting; simulation studies suggest that estimates become unstable when EPV <10.^[5] Accordingly, the primary adjusted model was intentionally parsimonious, including only three predictors selected on theoretical and empirical grounds: medication understanding, sex, and education. Sensitivity analyses were conducted using reduced predictor sets and penalized logistic regression (Firth correction and related bias-reduction approaches) to evaluate the stability of the estimates under small-sample and class-imbalance conditions. EPV considerations guided predictor selection. With 37 outcome events in the minority class, inclusion of more than three predictors risks overfitting (EPV ≈ 12.3 for three predictors). All analyses were conducted using Python (pandas, statsmodels) and R as needed. Regression results are reported as adjusted odds ratios (AOR) with 95% confidence intervals, and statistical significance was defined as P < 0.05.

RESULTS

A total of 273 respondents were included. High satisfaction was reported by 236 participants (86.45%). Communication-related items, pharmacy interactions, and environmental cleanliness were strongly associated with overall satisfaction in bivariate analysis. The sociodemographic characteristics of the study participants are summarized in Table 1.

Multivariable logistic regression identified understanding medication instructions as the strongest predictor of high satisfaction (AOR 17.43; 95% confidence interval [CI] 6.02– 50.43; P < 0.001). Higher educational attainment showed a negative association with satisfaction (AOR 0.46; 95% CI 0.295–0.723; P < 0.001). Gender was not significantly associated with satisfaction (AOR 1.14; 95% CI 0.51– 2.54; P = 0.744).

Domain reliability analysis revealed acceptable internal consistency for communication (α = 0.547), moderate consistency for environment (α = 0.468), low consistency for pharmacy due to a two-item structure (α = 0.293), and alpha not applicable for administrative processes (single item).

A total of 273 respondents were included. Age was recorded as discrete categories; therefore, the median age was 45 years (interquartile range 35–55). The sample included 158 females (57.9%) and 115 males (42.1%). Most of the participants had primary or secondary education, and the majority reported household monthly incomes below ₹20,000. Overall satisfaction with outpatient services was high: 236 of 273 respondents (86.45%) selected the highest possible satisfaction rating, while 37 participants (13.6%) fell into the not-highly satisfied category.

In bivariate analyses, education level showed a significant association with high satisfaction (P < 0.001), with higher educational attainment associated with lower odds of reporting the highest rating.^[6] Bivariate associations between outpatient satisfaction and independent variables are presented in Table 2. Gender demonstrated no significant association with satisfaction status. Communication-related items, particularly a clear understanding of how and when to take prescribed medication, showed the strongest unadjusted relationships with satisfaction. Items such as willingness to return to the facility, intention to recommend the facility to others, and clarity of the doctor’s explanation of the illness exhibited the lowest univariate P-values (all P < 1 × 10^-5).

A parsimonious multivariable logistic regression model was constructed using high satisfaction as the dependent variable. Because the minority outcome class consisted of 37 patients, the model was restricted to three predictors to satisfy events-per-variable assumptions. After adjusting for sex and education level, understanding medication instructions remained a strong and independent predictor of high satisfaction (adjusted odds ratio [OR] 4.38, 95% CI consistent with model estimates, P = 0.0014). Higher education level continued to show a significant negative association with satisfaction (adjusted OR 0.57 per category increase, P = 0.028). Gender was not significantly associated with the outcome (adjusted OR 1.16, P = 0.739). Adjusted odds ratios from the multivariable logistic regression analysis are shown in Table 3.

Model coefficients showed acceptable numerical stability given the sample and EPV constraints. Sensitivity analyses using reduced predictor sets and penalized logistic regression (Firth correction) produced directionally consistent estimates: understanding medication instructions remained [Table 4] the strongest predictor (penalized OR 6.63), and the inverse effect of education persisted (penalized OR 0.485). These analyses confirmed the robustness of the principal findings. Communication-related processes, particularly clarity of medication instructions, emerged as the most influential determinants of high outpatient satisfaction in this sub-divisional hospital setting.

This study has several important limitations. The distribution of the outcome variable was highly skewed toward high satisfaction, with only 37 participants in the not-highly-satisfied category. This imbalance restricted the number of covariates that could be included in multivariable logistic regression models under established events-per-variable criteria, thereby reducing the ability to detect modest associations. Although sensitivity analyses using reduced predictor sets and penalized logistic regression demonstrated directionally consistent effects, some degree of residual model instability cannot be excluded. The study was conducted in a single sub-divisional hospital, which may limit generalizability to other settings, particularly higher-level facilities or private-sector OPDs. Consecutive sampling, although operationally appropriate for quality-assessment purposes, may introduce selection bias by oversampling patients attending during specific time windows. All responses were self-reported and may be influenced by recall bias or social-desirability tendencies, particularly in public-sector environments. Finally, the activity was undertaken as a de-identified, minimal-risk quality-improvement initiative with verbal administrative approval, consistent with ICMR and NHM guidance; however, the absence of formal written ethics-committee documentation may be regarded as a procedural limitation. Larger, multi-facility studies with more balanced outcome distributions are recommended to enhance generalizability, improve statistical power, and permit evaluation of additional covariates.

DISCUSSION

In this cross-sectional assessment of outpatient satisfaction in a government sub-divisional hospital in West Bengal, overall satisfaction levels were high, consistent with prior Indian evidence from public-sector OPD settings where ratings frequently cluster at the upper end of the scale due to cultural tendencies toward positive reporting, short consultation expectations, and hierarchical clinician - patient relationships. Despite the ceiling effect, the study identified several specific service components that exert strong and independent effects on satisfaction. Communication quality - particularly patients’ understanding of how and when to take their medications - emerged as the most important determinant in both the primary adjusted regression model and the penalized sensitivity model, consistent with previous findings on provider communication and patient satisfaction in outpatient settings.^[7] This finding is highly concordant with existing literature from India and similar LMIC settings, where clarity of medication instructions consistently predicts satisfaction, treatment adherence, and follow-up compliance. Communication remains a central pillar of the NQAS, and the current study reinforces that meaningful patient comprehension - rather than mere information delivery - is a critical quality-of-care indicator.

Education level showed a significant negative association with satisfaction, with higher-educated patients less likely to report being “highly satisfied.”^[8] Similar trends have been observed in multiple Indian patient-satisfaction studies, which attribute this pattern to higher expectations, greater awareness of service standards, and lower tolerance for structural or interpersonal deficiencies among more educated users.^[4,9] This underscores the principle that satisfaction is partly expectation-dependent, with higher socioeconomic groups demonstrating more critical evaluations of public health facilities. The implication for administrators is clear: Perceived acceptability of services cannot be judged solely from aggregate satisfaction rates; stratified feedback, especially from higher-literacy groups, provides more sensitive insight into service gaps.

The role of physician explanation - specifically, clearly explaining the illness - showed strong, independent predictive value in the primary multivariable model and remained influential in penalized models including multiple correlated communication and process variables. This strengthens the inference that communication quality is not merely correlated with but contributes substantially to satisfaction. Within NQAS and other Indian quality frameworks, “provider - patient communication” is a functional domainer that hospitals often struggle to operationalize due to caseload pressures, time constraints, and lack of formal communication training. Our findings suggest that even minimal investments in communication-focused micro-interventions (e.g., structured explanation protocols, visual counselling aids, or brief medication-use education at pharmacy counters) may produce disproportionately large improvements in perceived quality.

Process and environment indicators, including cleanliness, waiting-area comfort, and being seen without excessive waiting, showed moderate effect sizes in the penalized sensitivity model, suggesting that although clinical communication is the strongest satisfaction driver, operational experience meaningfully contributes to patient perceptions. Cleanliness and waiting times have repeatedly been identified as key determinants of public facility satisfaction in India.^[10] The regularized model, which accounts for collinearity across process variables, supports the robustness of these associations. Notably, the extremely large penalized estimate for “being informed about follow-up/referral” should be interpreted cautiously due to wide uncertainty and expected penalization behavior; nonetheless, it highlights a critical gap in continuity-of-care communication that deserves administrative attention.

The results of the sensitivity analysis reinforce the credibility of the primary findings. When ten service-quality domains were included simultaneously using L1 regularization, the strongest retained predictors continued to be communication- and process-oriented measures, with medication understanding remaining one of the most robust determinants across all analytic approaches. This triangulation between conventional EPV-appropriate modeling and penalized high-dimensional modeling adds confidence that the observed associations are not artifacts of model specification.

Although satisfaction scores were high overall, the study reveals that targeted communication improvements, especially ensuring consistent medication instruction clarity and structured doctor explanation, may be significantly more efficient levers for quality improvement than broad, diffuse interventions. From an administrative and systems perspective, these findings emphasize the need to embed communication skills training, standard patient-education protocols, and post-consultation counselling mechanisms (e.g., pharmacist-led reinforcement) within outpatient workflows. Given resource constraints in government facilities, such streamlined strategies align well with NHM’s quality-improvement philosophy of low-cost, high-yield interventions.

The findings should also be understood in the context of the study’s limitations, including the skewed outcome distribution, single-center setting, and reliance on consecutive sampling. Nevertheless, the internal consistency analysis demonstrated acceptable reliability for multi-item domains, and the triangulated regression approach - encompassing parsimonious adjusted modeling, sensitivity analyses, and penalized regression - provides methodological robustness and enhances confidence in the study’s conclusions.

In summary, the study demonstrates that outpatient satisfaction in this public-sector facility is driven predominantly by communication quality, especially medication understanding and physician explanation, with supportive contributions from environmental cleanliness and efficient patient flow. These determinants are actionable, measurable, and aligned with national quality standards, indicating clear opportunities for targeted quality-improvement initiatives. Future studies across multiple facilities and with more balanced satisfaction distributions would deepen generalizability, but the present findings provide an evidence-based roadmap for enhancing patient-centered care in similar government outpatient settings.

CONCLUSION

This study demonstrates that outpatient satisfaction in a government sub-divisional hospital is generally high; however, specific communication and process elements strongly influence whether patients report the highest level of satisfaction. Clear explanation of illness and complete understanding of medication instructions were the most consistent and robust predictors across both parsimonious adjusted models and penalized sensitivity analyses. Environmental cleanliness, waiting-area comfort, and being seen without excessive delay also contributed meaningfully to satisfaction. These findings underscore that patient experience is shaped less by structural inputs alone and more by communication quality, interpersonal interactions, and streamlined service processes. The single-center design and skewed outcome distribution limit generalizability and statistical power; nonetheless, congruence between EPV-compliant multivariable models and penalized sensitivity analyses supports internal robustness. The study reinforces that patient satisfaction is not a superficial metric but reflects underlying processes and communication quality that can be systematically improved within public health facilities.