The sample size was equal to 10 times the number of items being tested. There were 49 items in our questionnaire. Therefore, the calculation formula of sample size was n = (7 items + 16 items + 26 items) × 10 = 490, which meant that at least 490 subjects were required for this study. Considering an expected sample loss rate of 20%, the sample size needed be further expanded. Therefore, the final sample size required was n = 490 ÷ (1–20%) ≈612.5, and the final sample size required was 613 at least.

The study was conducted in accordance with the Declaration of Helsinki (32) and the guidelines of the Air Force Medical University. Participants were patients with confirmed COVID-19 recruited from a mobile cabin hospital during the outbreak of COVID-19 in Shanghai. All patients were informed that participation in this study was voluntary. They could withdraw from the study at any time for any reason, and the questionnaires were completed anonymously. The inclusion criteria were COVID-19 patients in a mobile cabin hospital. The exclusion criteria were those who had recently experienced major life events and those who were unwilling to participate in the survey. A total of 650 patients with COVID-19 were selected for the cross-sectional survey from April to May 2022. However, 18 subjects withdrew from the survey, 13 did not fill out the questionnaire completely, and 11 filled out the questionnaire with too much homogeneity, which was regarded as an invalid questionnaire. The final sample included 608 patients (270 men, 338 women) from the mobile cabin hospital, aged 11–76 years (M_age = 38.08, SD = 13.47).

Before beginning the formal investigation, we trained the researchers to ensure the quality of the research. With the consent of the mobile cabin hospital managers, researchers explained the purpose of the survey to the patients to obtain their approval and support prior to data collection, and the patients gave their consent to participate in the research. With the help of head nurses in the mobile cabin hospital, questionnaires were distributed to the patients. The questionnaire was collected on the spot. We eliminated invalid questionnaires that were not filled in completely and answered randomly. The method of double check was used to input the data of the valid questionnaires to ensure accuracy.

SPSS 26.0 statistical software and Mplus 8.3 were used for statistical analysis. Descriptive statistics (mean, standard deviation, frequency and percentage) were used to describe the sample's characteristics. The Chi-square test and variance analysis were used to screen statistically significant indicators. Logistic regression analysis was used to evaluate the influencing factors of potential categories. The data for security were entered into the latent profile analysis, with one class initially and additional classes added incrementally until a unique solution could not be determined with maximum likelihood methods. The latent profile analysis model evaluation indicators included the Akaike information criterion (AIC), Bayesian information criterion (BIC), sample-size-adjusted BIC (aBIC), bootstrapped likelihood ratio test (BLRT), Lo-Mendell-Rubin (LMR) adjusted likelihood ratio test, Vuong-Lo-Mendell-Rubin likelihood ratio test (VLMR), and entropy. A higher entropy value, which is an important indicator, indicates a more accurate classification of the model. The smaller the AIC, BIC, and aBIC values, the better the model fit. LMR, BLRT, and VLMR are often used in model comparison, and a significant P-value indicates that K model categories are better than K – 1 model categories. The best-fitting models were selected through comprehensive evaluation of the above indexes, and the security among patients infected with COVID-19 in the mobile cabin hospital was divided into different categories. Hierarchical regression analysis was used to explore the impact of psychological capital on the security of patients. All tests conducted were two-sided, with a significance level of 0.05.

Our research was in accordance with the ethical standards formulated in the Declaration of Helsinki (39) and was confirmed by the Fourth Military Medical University ethics committee approval (KY20224143-1). Informed consent was provided by the participants prior to their participation. The survey was anonymous, and the confidentiality of the information was assured.

The Harman single-factor method was used to test the common method deviation. The variance explained by the maximum factor variance was 27.8%, less than the critical value of 40% (40). The statistical test results showed that there was no significant common method bias in the measurement.

The best-fitting latent profile analysis was the three-class model (Table 1). The P-values of the LMR test (<0.001), VLMR test (<0.001), and BLRT test (<0.001) of the three-class model were the smallest, suggesting that this model was statistically significant at the α = 0.05 level. This model had the highest entropy value (0.920) and the lower AIC (28,730.713), BIC (29,021.784), and aBIC (28,812.250). Figure 1 shows the subtypes of patients' security (Classes 1, 2, and 3); the proportion of each type was 12.00, 49.51, and 38.49%. According to this model, they could be distinguished as having relatively low (Class 1), medium (Class 2), or high (Class 3) security. The correct classification probabilities mean the average latent class probabilities for the most likely latent class membership by latent class. The larger the proportion, the more accurate the result. In order to further verify the reliability of the classification results, the correct classification probabilities of the three categories of patients' safety were calculated to be 97.1, 96.6, and 95.7% respectively (Table 2).

Table 3 presents the security and its two dimensions of the three classes. The results of the analysis of variance of the total score and two dimensions of patient security in each group were statistically significant (F = 1,329.070, P < 0.001; F = 645.427, P < 0.001; F = 1,141.546, P < 0.001). Further least significant difference (LSD) analysis showed that Class 1 < Class 2 < Class 3 in terms of security and its two dimensions (P < 0.001). Univariate analysis showed that there were significant differences among the three groups in educational level (χ² = 9.956, P = 0.007), family income (χ² = 22.936, P = 0.001), psychological capital (F = 35.907, P < 0.001), and its four dimensions (F = 52.296, P < 0.001; F = 43.750, P < 0.001; F = 23.610, P < 0.001; F = 7.192, P = 0.001). There was no difference among the three groups in other demographic characteristics (Table 4).

The security of patients with COVID-19 determined by latent profile analysis was used as the dependent variable, and Class 3 was used as the control group. Compared with Class 3, those with trouble falling asleep (OR = 6.572, P < 0.05), sleep quality as usual (OR = 3.693, P < 0.05), and lower tenacity (OR = 0.831, P < 0.05) were more likely to be grouped into Class 1. Those with a monthly per capita family income <3,000 (OR_class1 = 3.131, P < 0.05; OR_class2 = 2.383, P < 0.05), lower self-efficacy (OR_class1 =0.678, P < 0.001; OR_class2 = 0.783, P < 0.001), and lower hope (OR_class1 = 0.739, P < 0.001; OR_class2 = 0.868, P < 0.001) were more likely to be grouped into Classes 1 and 2 (Table 5).

Table 6 shows the results of hierarchical regression analysis with security as the dependent variable. The results showed that family income and sleep quality had entered the regression equation of the security of COVID-19 patients (F = 3.182, R² = 0.089, P < 0.001). On the basis of model 1, four dimensions of psychological capital were included in model 2 (F = 10.093, R² = 0.275, P < 0.001). Variance expansion factors were <10, indicating that there was no obvious collinearity among variables. Psychological capital was an important predictor of patients' security that could independently explain 18.70% of its variation (ΔR² = 0.187, P < 0.001).

The establishment of mobile cabin hospitals to treat a large number of patients with mild or moderate COVID-19 is an effective way to deal with the outbreak of the pandemic and block its spread at the social level (41). In the face of the special isolation environment in these hospitals, patients' security is threatened, which has an impact on their psychological health and disease treatment. However, there is a lack of research on the sense of security of this group, and limited studies have examined whether psychological capital might impact the sense of security of patients with COVID-19 infection in mobile cabin hospitals. Therefore, the purpose of this study was to clarify the subtypes of security of patients infected with COVID-19, to explore the influencing factors of different types of security, and to examine the relationship between patients' security and psychological capital. To our knowledge, this is the first research to study the security of patients with COVID-19 infection in mobile cabin hospitals.

Our study used latent profile analysis to classify patients' sense of security into three categories: low security (Class 1), moderate security (Class 2), and high security (Class 3). Among them, Class 2 had the largest number, accounting for 49.51%, which indicates that nearly half of the patients had moderate security. Although the number of patients in Class 1 was the lowest, accounting for 12%, this group needs the most attention.

More and more attentions have been paid to patients' sense of security in recent years, and it is an important aspect of patient-centered care (42). Previous studies have shown that patients with low sense of security were more worried about their economic situation (43). A longitudinal study showed that patients with high security always had good psychological health (44). The patient's sense of security includes many influencing factors, such as external environmental factors and self-related factors, such as economic and psychological conditions (45, 46). In our study, we explored the sense of security of patients with COVID-19 from different perspectives, which were showed in Tables 4, 5. However, the results in Tables 4, 5 were not consistent because that a multivariable logistic regression model was performed to include all independent variables.

Those whose per capita family income was <3,000 tend to have a low or medium sense of security. This is because under COVID-19, people in controlled zones are restricted from going to work, which has a great impact on people's income, especially in low-income families (47, 48). Such people have to worry about disease, but also their living expenses and loan repayment, so they are more likely to have a low sense of security. Furthermore, our results showed that the sleep quality of patients after diagnosed with COVID-19 also affected their security, and that the two are closely related. Our results are consistent with Hoyniak et al.'s (49) finding that emotional security is conducive to sleep. A cross-sectional study showed that COVID-19 is associated with changes in sleep schedule and in the quantity and quality of nighttime sleep because of the new stressors, altered roles, and uncertainties about health and economic security (50). Therefore, in the management and nursing of COVID-19 patients in mobile cabin hospitals, from the perspective of improving their security, we should focus on patients with low income and poor sleep quality.

Psychological capital belongs to the category of positive psychology, which emphasizes individual strength and enthusiasm (51, 52). Our study shows that COVID-19 patients with good psychological capital tend to have a high sense of security, and it is an important predictor of patients' security that can independently explain 18.70% of the variation in security, which is consistent with the results of Eweida et al. (28). In the hierarchical regression analysis of this study, tenacity was not significant in model 2. The reason for this may be that tenacity is a persistent quality that has few impacts on the sense of security. Research shows that individuals with low psychological capital have a low sense of security (53, 54). Psychological capital can buffer the uncertainty, stress, and anxiety of patients in mobile cabin hospitals, which can easily lead to an increase in security (55, 56). Therefore, managers of shelter hospitals can improve patients' security by improving their psychological capital level (28). Medical staff can guide patients to adjust their mentality and help them to adjust their negative emotions, which is conducive to the acquisition of a sense of security. According to Bandura's social cognitive theory, people who have subjective initiative can actively adapt to and change the environment (57). Research shows that mindfulness can improve the level of individual psychological capital (58). Therefore, mindfulness-based cognitive therapy can be conducted for patients in shelters to treat and alleviate their emotional and psychological problems, such as anxiety, depression, and impulsiveness, so as to improve their self-efficacy, hope, and optimism (59, 60). Individuals with a high level of psychological capital are more willing to take initiative and face difficulties and turning points, are more optimistic, confident, and hopeful, and can recover quickly even if they encounter setbacks (61).