This is the first study to adapt, modify, and validate the CSI-N-3 through a rigorous, multiphase process. Psychometric evaluation based on the CTT and IRT showed that the 15-item CSI-N-3 with a 2-factor solution is a reliable and valid self-report measure for assessing COVID-19 stigma for nurses. The factor analytic strategies used in CTT shared the same factor structure model with the original scale, the HSI-N , including subscales of nurses stigmatizing patients and nurses being stigmatized.
In addition to the construct validity of the CSI-N, as supported by CFA, the convergent validity of the scale was also supported, as there were significant negative correlations with self-reported physical health, psychological health, and social support levels. Similar to other infectious diseases such as HIV, SARS, and MERS [9, 20], our findings showed that COVID-19 stigma adversely affects the physical and mental health of frontline nurses, although the r-value was low. The low r-value was simply shows that these constructs were significantly correlated but different in the individual constructs . Besides, the Cronbach's α was more than 0.6, indicating that the CSI-N-3 had satisfactory internal consistency and reliability .
Using IRT analysis, we have provided information about items in the CSI-N-3 that expand on traditional CTT methods [35–36]. Our data support that the ordered threshold in the category probability curves, which means that the category rating scale of the CSI-N-3 worked well and that nurses could use the scale to differentiate the four levels of item difficulty [30, 36]. The combination of a good person-separation index (> 2) and person reliability (> 0.8) suggests that the CSI-N-3 has acceptable measurement precision and is sensitive to distinguishing both high and low levels of social support among frontline nurses .
Regarding the TIF, when represented graphically, high TIF values are associated with low standard errors of measurement and can thus indicate precision . The most precise information provided by the TIF for the CSI-N-3 displays the precise and reliable measure of the low to middle levels of the CSI-N-3. Furthermore, IRT measures also allow for the estimation of the equivalence of item calibrations across different samples and contexts . In our study, we examined how 15 items may have been used differently, based on the nurses' professional titles and working places. The DIF findings showed that there were no professional titles and working place differences in the item difficulty, which further support the stability and validity of the CSI-N-3.
The score of CSI-N-3 reflects the level of COVID-19 stigma perpetrated or experienced by nurses; however, we found that the mean score of CSI-N-3 (2.80 ± 3.73) appears to suggest a major floor effect; that is, the level of nurses stigmatizing patients or being stigmatized was not as high as the level of nurses who worked with people living with HIV (8.74 ± 9.31;  and MERS-CoV . This finding might be explained by the cultural differences between China and South Africa. As the original study was conducted in 2008, after effective interventions to decrease stigma in healthcare institutions and nursing educations in these years, the external stigma might be decreased toward infectious diseases.
Under the influence of Confucian culture, most Chinese nurses have manifested a sense of work responsibility, dedication to patient care, personal sacrifice, and professional collegiality during the pandemic [3, 5]).
The milder forms of stigma were mainly reflected in terms of nurses being stigmatized, such as being subject to labeled as COVID-19, gossip, and as being infected and contagious. The possible explanation is that the general population, especially neighbors, routinely misunderstood nurses as a threat to the safety of others and as “disease-carriers” , and thus they faced avoidance by the community due to this fear . Furthermore, item 3 (A nurse who kept her distance when talking to a COVID-19 patient) got the highest score, i.e., was most often endorsed. During the early months of the COVID-19 pandemic, personal protective equipment (PPEs) for nurses was in short supply, but nurses knew that droplet, contact, and aerosol transmissions were the main perceived infection routes of COVID-19, and thus they avoided close contact with patients when communicating with them to protect themselves. On the other hand, even with sufficient PPE, nurses showed a certain degree of fear and stigma toward COVID-19 patients.
Nevertheless, the total level of stigma was low among the nurses, and nurses were unaware that their physical distancing behaviors may have biased their provision of care  and exacerbated avoidance, mistreatment, and stigma toward COVID-19 patients .
nciding with similar studies [40, 41]), this study found that social support is negatively associated with COVID-19–associated stigma among nurses. This result suggests that social support is an effective coping strategy that can alleviate stigma. As Gardner and Moallef  suggested: support from the media and community for nurses “stalwart heroism and sacrifice” contributed to their positive experience and to less stigma . As Liu et al.  indicated, multiple support systems, including the hospitals, colleagues, families, friends, and society, can help frontline nurses minimize the stigma associated with caring for COVID-19 patients. With logistical support from their hospital and peer support and encouragement among colleagues (e.g., the sharing of workplace experiences), frontline nurses had a sense of safety and felt less stigma . But, clearly, in light of the relatively small explained variance in the regression model, further exploration of other factors that might have been included is encouraged, and including the complexity of factors that affect COVID-19 stigma for nurses is suggested.
This study has several limitations. First, this sample came from one of the premier infectious disease hospitals in Shanghai, China; therefore, it might limit the generalizability of the findings to other Chinese-speaking regions. Second, the low magnitude correlations between stigma and physical health, psychological health, and social support might be due to the three single-item physical health, psychological health, and social support measures used in this study not adequately assessing these constructs. Thus, valid and reliable scales that are available in Chinese to assess nurses’ physical health, psychological health, and social support are needed to further assess the construct validity of the scale. Third, the Chinese government is encouraging all the healthcare providers actively engaging COVID care in all the social and mass media. Since we recruited from the infectious institution in Shanghai, nurses might not have been willing to share their “true” feeling as the survey link came from their work place. A longitudinal study is recommended to see if nurses will be more forthcoming in their answers, and to compare current and future answers to see if the passage of time and the fading of the national attention to COVID-19 will affect their responses.
Furthermore, the non-significant relationship between physical and psychological health and nurses’ reported stigma may be related to measurement issues. Third, some psychometric characteristics of the CSI-N-3 could be assessed further, such as test-retest reliability, and the responsivity or sensitivity of the scale. Therefore, future longitudinal or experimental studies are warranted. Third, the sample size for IRT analysis was relatively small, despite the lack of consensus on the optimal sample size. A further refinement of the scale based on testing a larger representative sample may produce more stable parameter estimates and robust results.