Commentary: Analysis of case management service quality for PCI patients based on the SERVQUAL model: a single-center prospective study

In their article "Analysis of case management service quality for PCI patients based on the SERVQUAL model: a single-center prospective study", Wang et al. (1) analyze the quality of case management in patients undergoing percutaneous coronary intervention (PCI) using the SERVQUAL model and Importance-Performance Analysis (IPA). Although the topic is highly relevant for hospital management, the methodology employed contains conceptual and statistical problems that have been superseded in the specialized literature. The reliance on predefined dimensions, the redundancy of expectations, the imprecision in sampling, and particularly the fundamental misapplication of the IPA methodology significantly undermines the validity of the findings.Wang et al.'s (1) primary weakness is the use of a structured questionnaire based on SERVQUAL's 22 generic dimensions, which imposes the researcher's perspective over the patients. Recent research in cardiology services demonstrates that the use of associative maps and semantic priming techniques allows identification of which attributes patients genuinely associate with satisfaction in a free and spontaneous manner (2). The study of Consuegra-Sánchez et al. (2018) has shown that professional and personal care is the dominant attribute (mentioned by 97% of patients), while other factors such as food or waiting times have significantly different weights than those assumed in generic scales. In addition, forcing patients to evaluate dimensions such as tangibility through personal appearance of managers may generate artificial results, as overall service quality is a complex and multifaceted construct that does not always conform to the rigid SERVQUAL framework.Wang et al. ( 1) calculate quality as the difference between perceptions and expectations (P-E), a methodologically redundant practice. It has been demonstrated that when a patient evaluates service performance, they are already making an implicit comparison with prior expectations in their subjective judgment (3).Measures based exclusively on perception (such as SERVPERF) are superior in 37 construct validity and operational efficiency. Additionally, the use difference scores introduces 38 reliability problems, as expectations typically show low variability and positive bias, converting the 39 "negative gap" a statistical artefact of the measurement method rather than a reflection of an actual 40 service shortcoming (4). 41The analyzed article assumes that overall quality is a linear sum of attributes. However, overall 43 quality should be understood as a complex phenomenon, akin to a musical chord (5). While 44 individual attributes are the "notes", overall quality is the resulting "chord", whose perception is 45 qualitatively distinct from the sum of its parts (3). 46Attempting to deconstruct quality into predefined dimensions ignores the fact that the whole is more 47 than the sum of its notes. Moreover, halo effects contaminate the evaluation of specific attributes 48 once the patient has formed an overall judgment. This reduces the diagnostic value of measuring 22 49 separate items. A more robust approach would be to isolate the global perception of quality (typically 50 assessed with a single item) from the evaluation of specific operational attributes. 51Wang et al. (1) calculate their sample size based on the variance of expectations but ignore the 53 intrinsic imprecision of the measurement scale. It is a common error to confuse the variance of 54 interval scales with that of binomial proportions or to overlook how the scale range (1-5 or 1 to 5) 55 affects estimation precision (6). The use of the Measurement Scale Imprecision Factor (MSIF), an 56 index invariant to scale range, is recommended to allow robust comparison of imprecision across 57 different studies with Likert and other interval scales of different length (which is common in the 58 service quality literature); for example, the creators of the SERVQUAL model employed a seven-59 point scale (7). 60The IPA analysis presented in (1) contains a fundamental methodological error that deserves 62 particular attention. Although the authors cite Martilla and James (8), the originators of the IPA 63 technique, they misapply the method's core logic. In the original IPA framework, the horizontal axis 64 should represent attribute importance, not expectations, while the vertical axis represents actual 65 performance. The distinction is critical: importance reflects how much a feature matters to 66 consumers, whereas expectations reflect what consumers anticipate they will receive. 67Wang et al. ( 1) confound these concepts by using expectation scores on the horizontal axis as a 68 proxy for importance. This conflation, however, fundamentally distorts the interpretation of the IPA. 69Expectation levels and importance ratings are conceptually and empirically distinct constructs. A 70 patient may have low expectations for rapid service completion yet consider it highly important; 71 conversely, patients may expect extensive personalized care without attaching high importance to it. 72By substituting expectations for importance, the authors create a hybrid measure that obscures the 73 true structure of patient preferences and misguides quality improvement priorities. 74 Furthermore, the graphical representation lacks horizontal and vertical error bars. A graph showing 75 only mean points is misleading, as it cannot determine whether an attribute's location in a "priority 76 for improvement" quadrant is statistically significant or whether uncertainty causes overlap between 77 quadrants due to variance. When working with a difference function (P-E), the authors neglect error 78 propagation. error in the mean of perceptions and that of expectations combine, increasing 79 uncertainty in the IPA diagram's point (9). Failing to account for this propagation using 80 methods such as Taylor series approximations means that any management recommendations derived 81 from the graph lack statistical rigor. 82Although the study of (1) aims to improve case management in PCI, their reliance on outdated 84 models like SERVQUAL, coupled with a fundamentally flawed IPA implementation, severely 85 undermines the practical utility and validity of their conclusions. Future research should prioritize the 86 patient's voice through associative maps, employ direct perception measures, and include error bars 87 and uncertainty propagation in any importance-performance analysis, remaining faithful to the 88 original IPA methodology as established by its creators.