The headline finding is sobering for hospital administrators. When the researchers modeled the joint information from the three dimensions of the Nurse Innovative Behavior Scale, the largest group of nurses, 61.85 percent of the sample, or 261 individuals, landed in a profile the authors labeled a relative low-innovation level. A second group, 31.99 percent or 135 nurses, showed a relative moderate level, while only 6.16 percent, just 26 nurses, fell into the relative high-innovation profile. The authors are careful to stress that these labels are sample-relative descriptors rather than validated clinical or managerial cutoffs, meaning the categories describe where nurses stood within this particular dataset, not universal thresholds of innovation.

Latent profile analysis is a person-centered statistical method that differs from the more common practice of simply averaging questionnaire scores. Instead of treating every nurse as a point on a single continuous scale, LPA looks for subgroups of individuals who answer the scale’s dimensions in similar patterns. In this study, the observed indicators were the mean scores of the three dimensions of innovative behavior, and the researchers compared competing models using fit statistics such as the Bayesian information criterion. That technical detail matters, because the final interpretation of the study hinges on it: the data did not cleanly demand a three-type solution.

Indeed, the authors report that a continuous one-factor model actually achieved a lower BIC than the three-profile model, while a four-profile solution showed better information-criterion fit than the three-profile one. In plain terms, the statistical evidence suggests that innovative behavior among these nurses varies along a continuum, and the three-level stratification is best understood as a descriptive convenience rather than proof that nurses come in three discrete types. The team explicitly cautions that the three-profile solution should be interpreted as empirical stratification along a continuous dimension, not as a mutually exclusive latent structure. This kind of methodological transparency is relatively rare in person-centered nursing research, where profile solutions are often presented as if they revealed natural categories.

What did the profiles differ on? The three groups showed statistically significant differences in both total and dimensional scores of organizational justice and job embeddedness, with all comparisons reaching P values below 0.001. Organizational justice refers to employees’ perceptions of fairness in how decisions are made, how procedures are applied, and how people are treated, while job embeddedness captures the web of links, fit, and sacrifice that keeps people anchored in their jobs and communities. Both constructs have long been theorized as resources that could fuel discretionary effort, including the risky, unrewarded work of championing a new clinical practice.

To move beyond simple group comparisons, the researchers ran an exploratory multinomial logistic regression examining which factors predicted membership in the higher-innovation profiles relative to the low-innovation reference group. Organizational justice emerged as the most consistent and robust correlate. For each one-point increase in organizational justice scores, the odds of belonging to the moderate-innovation profile rather than the low-innovation profile rose by a factor of 1.081, with a 95 percent confidence interval of 1.044 to 1.120 and a P value below 0.001. The association was even stronger for the top profile: each one-point increase in perceived justice multiplied the odds of high-innovation membership by 1.285, with a confidence interval of 1.162 to 1.422.

Administrative position also mattered. Nurses who held no formal management role had roughly half the odds of belonging to the moderate-innovation profile compared with nurses at head nurse level or above, with an odds ratio of 0.497, a 95 percent confidence interval of 0.259 to 0.952, and a P value of 0.035. This finding hints at a familiar dynamic in hospital hierarchies: nurses with leadership titles may have more access to resources, project approval channels, and professional networks that make turning an idea into a implemented change feasible. Bedside nurses, however motivated, may simply lack the organizational leverage to push innovations through.

Perhaps the most striking negative results concern the factors that did not survive statistical adjustment. Job embeddedness, participation in research training, and monthly income were not significantly associated with profile membership once other variables were controlled. That job embeddedness dropped out is notable, because embeddedness is often assumed to be a driver of discretionary contributions like innovation. The result suggests that simply keeping nurses attached to their jobs is not the same as enabling them to change how care is delivered. Similarly, the null finding for research training challenges the assumption that formal research exposure automatically translates into bedside innovation, and the income result suggests that financial incentives alone may not unlock creative behavior in this population.

The authors are appropriately restrained about what these associations can and cannot mean. Because the study is cross-sectional, the between-profile comparisons are descriptive and cannot establish temporal or causal relationships. It is entirely possible that nurses who innovate come to perceive their organizations as fairer, rather than fairness breeding innovation, or that some unmeasured factor influences both. The regression estimates were also based on hard class assignment, meaning each nurse was treated as a definite member of one profile even though latent profile analysis inherently involves classification uncertainty. The high-innovation profile contained only 26 nurses, which limits the precision of estimates involving that group. Whether organizational justice can serve as a modifiable intervention target, the authors conclude, requires prospective or intervention-based evaluation rather than inference from a single snapshot.

Even with those caveats, the study carries a clear message for nursing management. In this sample, innovative behavior was predominantly at a relatively low level, and the factor most consistently linked to higher levels was something managers can plausibly influence: the perceived fairness of the organizational environment. The researchers suggest that a fair and supportive workplace may warrant evaluation as a potential management priority, and that profile stratification, while not a validated diagnostic tool, may prove useful for descriptive management and hypothesis generation. If subsequent intervention studies confirm that strengthening procedural and distributive justice raises the odds of nurses developing and implementing new ideas, the payoff could extend well beyond innovation metrics, since the quality and safety of patient care in hospitals ultimately depends on the people closest to the bedside being able to improve what they see every day.

Subject of Research: Latent profile analysis of innovative behavior and its associated factors among clinical nurses in China

Article Title: Innovative behavior among clinical nurses: empirical level stratification and associated factors in a cross-sectional study in Zhejiang Province, China

Article References: Cao, K., Lin, L., Yu, R., Zheng, H., & Chen, F. (2026). Innovative behavior among clinical nurses: empirical level stratification and associated factors in a cross-sectional study in Zhejiang Province, China. BMC Nursing. https://doi.org/10.1186/s12912-026-05474-2

Image Credits: AI Generated

DOI: 10.1186/s12912-026-05474-2

Keywords: clinical nurses, innovative behavior, latent profile analysis, organizational justice, job embeddedness, nursing management, cross-sectional study, Zhejiang Province, BMC Nursing, tertiary hospitals, nursing research, workplace fairness