Development and Validation of a Service Quality Assessment System for Chronic Disease Management in Primary Healthcare

Summary

This study developed and validated a service quality assessment system for primary chronic disease management using Delphi- analytic hierarchy process (AHP) methods. Psychometric evaluation confirmed high reliability. A comparative analysis showed that the AHP-SERVQUAL framework has higher predictive power for patient adherence than the alternative models.

Abstract

Chronic non-communicable diseases require high-quality primary healthcare management, yet existing evaluation systems often fail to capture the multidimensional nature of patient-perceived service quality in chronic care settings. This study developed and validated a service quality assessment system (SERVQUAL) tailored for primary chronic disease management based on the SERVQUAL framework. A multi-stage methodology integrating the Delphi method and analytic hierarchy process (AHP) was employed to construct and weight the scale. The instrument was empirically validated using cross-sectional data from 433 patients across five primary healthcare institutions. Construct validity and reliability were evaluated through exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). Furthermore, the predictive performance of the AHP-SERVQUAL model was compared with that of alternative models (KANO, TOPSIS-RSR) for patient compliance, problem resolution rates, and complaint frequencies. Psychometric evaluation demonstrated that the adapted five-dimensional scale (tangibility, reliability, responsiveness, assurance, and empathy) possessed high internal consistency and structural validity. Application across the centers revealed significant variations in service quality, particularly in the assurance and empathy dimensions. Comparative modeling indicated that the AHP-SERVQUAL framework yielded higher predictive power for patient adherence and overall service quality than the alternative models. Additionally, the quality indices generated by this model were significantly correlated with higher problem-resolution rates and lower complaint frequencies. The AHP-SERVQUAL-based system offers a reliable and valid metric for evaluating chronic disease management in primary care. By accurately capturing patient-perceived quality, this instrument provides administrators with an evidence-based tool to target quality improvement initiatives.

Introduction

Chronic non-communicable diseases have become an extraordinarily serious public health issue around the world¹. According to recent data from the world health organization, more than 70% of global deaths are attributable to chronic diseases. With the acceleration of an aging population in China, the number of people with metabolic diseases such as hypertension and diabetes is increasing rapidly, putting greater pressure on society and the medical system^2,3. In response to this, primary healthcare institutions, as the frontlines of chronic disease management, have gradually become more prominently positioned in such roles. Specifically, primary healthcare institutions serve as the "gatekeepers" of the healthcare system, providing first-contact care, longitudinal health monitoring, and essential screenings for high-risk populations. Beyond basic clinical treatment, these institutions play a pivotal role in delivering personalized health education, facilitating behavioral interventions, and coordinating integrated care pathways between community settings and secondary/tertiary hospitals to ensure continuity of care.

At present, the management model is shifting from single quantity coverage to high-quality development⁴. Despite the strategic shift towards primary care in China, a critical bottleneck has emerged: traditional clinical control metrics (e.g., HbA1c or blood pressure readings) are inherently reductionist. While essential for assessing physiological status, these objective indicators fail to capture the psychosocial dimensions of continuous care, such as the quality of doctor-patient communication, the patient's daily barriers to treatment adherence, and the overall subjective well-being during long-term management^4,5. Consequently, relying solely on biomedical metrics obscures critical service delivery gaps that ultimately govern long-term health outcomes. There is an urgent need for an evaluation framework that transcends traditional clinical outcomes to capture the full service process through the lens of patient perception. This is particularly vital as primary healthcare transitions from a treatment-centric model to one focused on preventive, holistic health management⁶.

As a classical theory in health care services research, SERVQUAL has been widely applied to evaluate discrepancies between actual perceptions and expectations across several dimensions, including tangible components, reliability, staff response time, and assurance. This model is applicable to the management of chronic diseases in this context. First of all, chronic disease management has high continuity and service dependence, characteristics consistent with the reliability stress in the model. Secondly, chronic disease management includes not only clinical diagnosis and treatment but also long-term behavioral interventions and psychological care; therefore, empathy is an essential criterion for assessing the doctor-patient contractual relationship. Empirical research shows that when patients, as subjects receiving quality care from medical staff, are satisfied with the service, their behavior and treatment adherence for chronic diseases are more positively affected. Therefore, using this model to convert the abstract process of medical services into measurable indicators can identify the weak links in primary chronic disease services^6,7.

Existing scholarship on quality assessment in primary care has predominantly relied on structural clinical indicators or broad patient satisfaction instruments, such as the primary care assessment tool (PCAT) or generic satisfaction scales⁸. While these tools provide essential benchmarks for systemic oversight, they often lack the methodological granularity needed to capture the iterative, process-based nature of chronic disease management. Recent attempts to apply the SERVQUAL framework in clinical settings have highlighted its diagnostic potential; however, many of these studies utilize unweighted models or generic dimensions that fail to incorporate context-specific elements—such as digital health touchpoints and the family doctor contractual relationship—that are central to modern primary care reforms^8,9. By systematically comparing a localized, AHP-weighted system against these conventional metrics, this study aims to demonstrate its superior predictive utility in capturing the nuances of patient-perceived quality, thereby offering a more robust methodological contribution to the field.

Although the SERVQUAL model has broad application prospects in primary chronic disease management, it cannot be used directly¹⁰. The traditional five dimensions generally focus more on a single service's successful completion and are less likely to consider the continuity of support for chronic diseases and patient-doctor collaboration as essential. Against the backdrop of digital health, primary chronic disease management has extended to online follow-up, wearable device monitoring, and community self-management groups. If no scene-based optimization is carried out, this assessment system cannot be extended to include aspects of digital after-care touchpoints or patient involvement in practice. Based on existing systematic reviews, the lack of Industry-specific general tools has limited the predictive value of these tools for prolonged patient conditions. In addition, the family doctor contract system introduced to China's primary care service provides a new interpretation of these services. Adding a long-term contractual relationship dimension to this generalized model will help address issues that cannot be covered in such models and enhance the scientificity of the research^8,9.

To address these gaps, this study develops and validates a localized service quality assessment framework specifically tailored for primary chronic disease management. While SERVQUAL has been widely adopted, its application in primary care often overlooks the longitudinal nature of family doctor contracts and the emerging digital touchpoints of community-based self-management^8,9. The novelty of this research lies in integrating these contextual dimensions—specifically, continuous care and self-management support—using a hybrid Delphi-AHP weighting scheme to enhance the model's granular sensitivity. It is hypothesized that this multidimensional, weighted approach provides superior predictive validity for patient outcomes compared to conventional models. The objective is to empirically test this framework across multiple centers, addressing the empirical question of whether a localized AHP-SERVQUAL model can more accurately explain variances in patient adherence and institutional efficiency than existing general-purpose tools¹¹.

Protocol

This study was conducted in strict accordance with the ethical principles outlined in the Declaration of Helsinki. The study protocol was reviewed and formally approved by the Institutional Review Board (IRB) of Chongqing Medical University (ethical approval no.: 2026127). Prior to enrollment, all participants were provided with a comprehensive explanation of the study’s objectives, procedures, and potential risks, and subsequently provided written informed consent. To safeguard patient privacy and ensure data confidentiality, all collected information was fully anonymized and de-identified before being accessed for statistical analysis. The complete list of survey instruments, expert consultation forms, and analytical software utilized in this study is detailed in the Table of Materials.

Study design and participant recruitment

A multi-stage methodological study, comprising a cross-sectional baseline survey and subsequent model-driven predictive simulations, was conducted across five primary healthcare institutions. Participants were recruited utilizing a stratified convenience sampling strategy during their routine follow-up visits. Eligibility criteria included: (1) adults aged 18 years or older; (2) medically diagnosed with at least one chronic condition (e.g., hypertension, type 2 diabetes) managed at the participating institution for a minimum of six months; and (3) possessing the cognitive ability to independently complete the evaluation. The surveys were administered face-to-face by trained clinical staff. To address missing data, Little's MCAR test was initially applied. Given that the missing data rate was low (<5%) and missing completely at random, multiple imputation techniques were employed to handle incomplete responses, ensuring the integrity of the dataset. To avoid overfitting and clarify the sample basis for each analytical stage, the fully imputed dataset (N = 433) was strictly partitioned. Specifically, 50% of the sample (n = 217) was utilized for exploratory factor analysis (EFA) to establish the factor structure. The remaining 50% (n = 216) was reserved for confirmatory factor analysis (CFA), structural equation modeling (SEM), and baseline establishment for institution-level quality profiling. The comparative model validation and simulated optimization scenarios were subsequently derived strictly from this second hold-out sample (n = 216) to prevent data leakage.

Conceptual framework and indicator development

The construction of the evaluation index system began by mapping the five core SERVQUAL dimensions—tangibility, reliability, responsiveness, assurance, and empathy—to the specific workflow of primary chronic disease management. This process involved integrating concrete service touchpoints, such as medical record creation, pharmaceutical distribution, and longitudinal follow-up protocols. To ensure relevance for the target demographic, the primary care information system was analyzed to assess its accessibility for elderly patients, followed by iterative refinements to the items. Specific reliability metrics focused on the accuracy of follow-ups and staff commitment, while responsiveness was operationalized as the timeliness of responses to patient inquiries (see Table 1 for operational definitions). The overall framework, including the interrelationships between dimensions, is conceptualized in Figures 1 and 2.

Delphi expert consultation and content validity

A two-round Delphi method was employed to refine the initial item pool. Experts were purposively recruited based on a multi-dimensional set of eligibility criteria to ensure professional authority and methodological rigor. Candidates were invited to the panel if they met the following requirements: (1) holding a senior professional title (e.g., Associate Professor, Chief Physician, or Nursing Director); (2) possessing a minimum of ten years of clinical, managerial, or research experience specifically focused on primary healthcare or chronic disease management; and (3) demonstrating a recognized track record in health service evaluation or public health policy development. The final panel of fifteen experts represented a cross-disciplinary cohort, including primary care physicians (n = 5), nursing managers (n = 4), public health researchers (n = 4), and health policy administrators (n = 2). This diverse composition and the rigorous selection criteria directly ensured the reproducibility and credibility of the subsequent content validity and AHP weighting procedures. Experts evaluated each item's relevance and clarity using a 4-point Likert scale (1 = not relevant, 4 = highly relevant). Consensus was defined a priori as ≥80% agreement among experts rating an item as 3 or 4. Items with an Item-Level Content Validity Index (I-CVI) ≥ 0.78 were retained (Table 2). Disagreements between rounds were resolved through iterative anonymous feedback; experts were provided with aggregated scores from the previous round and permitted to adjust their ratings. Items failing to reach consensus after the second round were either substantially revised based on qualitative expert feedback or eliminated, culminating in the final scale-level content validity index (S-CVI)^12,13.

Analytic hierarchy process (AHP) and scoring mechanism

To determine the relative weights of the evaluation indicators, the AHP method was utilized. A purposive sub-sample of nine senior experts from the Delphi panel completed a 9-point fundamental scale to construct pairwise comparison matrices for all dimensions and sub-dimensions. The geometric mean method was used to aggregate the experts' judgments. A Consistency ratio (CR) was calculated for each matrix; only matrices with a CR of were considered acceptable, ensuring logical consistency in expert judgments. Final global weights (W_i) were derived by multiplying the local weights of sub-dimensions by the weights of their parent dimensions. The final service quality score (SQ) for each patient was calculated using the formula:

SQ = ∑(W_i x P_i) (1)

where P_i represents the patient's perceived performance score for the item i. Notably, this study adopted a perception-only (P) scoring method—consistent with the SERVPERF paradigm—rather than the traditional gap-based (P-E) approach. This methodological choice, measured on a 7-point Likert scale (ranging from 1 = "Strongly Disagree" to 7 = "Strongly Agree"), was implemented to enhance the model's predictive validity and minimize the measurement instability often associated with quantifying patient expectations in longitudinal chronic care settings.

Psychometric validation and structural equation modeling

The randomly split dataset was subjected to psychometric evaluation. For the first half of the data, EFA was conducted using principal axis factoring with promax oblique rotation, as the SERVQUAL dimensions were hypothesized to be correlated. Factors were retained based on eigenvalues >1 and visual inspection of the scree plot. Items with factor loadings <0.40 or significant cross-loadings were excluded. The second half of the data was used for CFA to confirm the factor structure. Maximum likelihood (ML) estimation was employed for the CFA and subsequent SEM analyses. Model fit was evaluated using stringent thresholds established in the psychometric literature, which indicate an acceptable balance between model parsimony and data fit: Comparative fit index (CFI) > 0.90, Tucker-Lewis index (TLI) > 0.90, root mean square error of approximation (RMSEA) < 0.08, and standardized root mean square residual (SRMR) > 0.08.

Predictive model comparison plan

To assess the comparative effectiveness of different evaluative frameworks, a predictive modeling analysis was conducted. To establish the methodological robustness of the proposed framework, the basic SERVQUAL model was benchmarked against three distinct analytical approaches: the proposed AHP-SERVQUAL, the KANO model, and the TOPSIS-RSR (Technique for Order Preference by Similarity to Ideal Solution combined with Rank-Sum Ratio) method. The KANO model was selected as a comparator because it categorizes service attributes into non-linear satisfaction drivers (e.g., basic needs versus attractive qualities), offering a theoretical contrast to the linear assumptions of SERVQUAL. Conversely, TOPSIS-RSR is a rigorous multi-criteria decision-making (MCDM) tool widely utilized for comprehensive institutional ranking, thereby serving as a robust methodological control for our AHP-based scoring system. To execute the comparison, each of the four models was independently applied to the hold-out validation dataset (n = 216), incorporating identical context-specific parameters (such as digital follow-up and self-management support). The evaluative process subsequently compared how effectively each model's generated quality indices correlated with external performance metrics (complaint frequency and issue resolution rates) and their predictive power (R²) for patient compliance. The predictive accuracy of each model was quantified using R² values generated through linear regression and SEM. To ensure model parsimony and avoid over-fitting, information criteria, specifically the Akaike information criterion (AIC) and Bayesian information criterion (BIC), were integrated into the selection process. Finally, cross-validation was executed to evaluate the robustness of the selected models across the five participating institutions. The comparative metrics for problem resolution, complaint frequency, and overall predictive power (R²) are shown in Figures 3 and 4.

Results

Study design and participant recruitment

A total of 433 cases involving individuals who had visited any of the five primary healthcare institutions in different areas, both urban and rural, were included. The participants' ages ranged from 30 to 85 years, with an average age of 61.2 years (standard deviation: 10.0). The gender distribution was 50.4% male and 49.6% female. Approximately 56.8% of the participants had hypertension, about 50.0% had type 2 diabetes mellitus (T2DM),...

Discussion

This study proposes a refined methodological framework for evaluating primary chronic disease management by integrating the SERVQUAL model with a Delphi-AHP weighting scheme. Beyond establishing a standard assessment system, our findings demonstrate that the inclusion of context-specific dimensions—such as longitudinal care continuity and digital touchpoints—addresses critical gaps in existing primary care evaluation tools, thereby providing a more granular lens for assessing patient-centered service quality<...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
AHP Pairwise Comparison Matrix Template	This study (Repository / Supplementary Material)	Saaty’s 1-to-9 fundamental scale	Determination of relative indicator weights
Delphi Expert Consultation Forms	This study (Repository / Supplementary Material)	Two-round consensus protocol	Indicator screening and content validity evaluation
IBM SPSS AMOS	IBM Corp., Armonk, NY, USA	Version 24.0	CFA, SEM validation, and predictive modeling
IBM SPSS Statistics	IBM Corp., Armonk, NY, USA	Version 26.0	Data management, MICE imputation, and EFA
Localized AHP-SERVQUAL Questionnaire	This study (Repository / Supplementary Material)	Final 5-dimension, 15-item version	Assessment of patient-perceived service quality