Interpersonal sensitivity and its associated factors among nursing students during the COVID-19 pandemic: a network analysis

Prior research has highlighted that interpersonal sensitivity is an individual trait characterized by an excessive awareness of others’ behavior and emotions, coupled with heightened sensitivity to perceived rejection or criticism [5]. Interpersonal sensitivity has been extensively studied in various populations, including cancer patients, individuals with mental health disorders, college students, and nurses [6‐9]. Previous research has consistently demonstrated that interpersonal sensitivity is associated with low self-esteem, negative self-concept, and increased psychological distress [10]. Individuals with higher interpersonal sensitivity are more prone to negative emotional reactions in social interactions, which can negatively impact their well-being and interpersonal relationships [11].

However, limited research has specifically focused on interpersonal sensitivity among nursing students, particularly in the context of the COVID-19 pandemic. Understanding the factors associated with interpersonal sensitivity in this specific population is crucial due to the unique stressors and challenges they face. Nursing students encounter a demanding educational curriculum, clinical training, and exposure to the complexities of patient care [12]. The additional stressors introduced by the pandemic, such as the existing constraints on learning resources and interpersonal interactions, further impact their psychological functioning and interpersonal sensitivity [13]. Therefore, it is essential to examine the factors associated with interpersonal sensitivity within this specific population in this period.

Cognitive-behavioral theory provides a relevant theoretical framework to understand interpersonal sensitivity among nursing students. According to the theory, individuals’ thoughts, emotions, and behaviors are interconnected, influencing their mental well-being [14]. Individuals high in interpersonal sensitivity may hold cognitive distortions, including negative self-perceptions, excessive self-criticism, and overestimation of others’ negative evaluations [15]. These distortions contribute to increased interpersonal sensitivity and subsequent negative emotional outcomes.

Additionally, the transactional stress model can be applied to explain how the unique stressors related to the COVID-19 pandemic affect nursing students’ interpersonal sensitivity. This model emphasizes the dynamic interaction between individuals and their environment [16]. This suggests that nursing students’ perceived stress, social support, self-esteem, and COVID-19-related fears may play significant roles in shaping their interpersonal sensitivity levels [17]. That is, increased stress, limited social support, low self-esteem, and heightened COVID-19-related fears may exacerbate interpersonal sensitivity among nursing students, leading to negative psychological consequences. To gain comprehensive insights into the nursing profession, it is imperative to investigate the significance of key indicators in elucidating the interpersonal sensitivity exhibited by nursing students. Additionally, analyzing the factors that are intricately associated with interpersonal sensitivity during the COVID-19 pandemic is essential.

Thus, this study aims to utilize network analysis to explore the interconnectedness of factors associated with interpersonal sensitivity among nursing students during the COVID-19 pandemic. Network analysis allows for the identification of central factors and specific inter-relationships within the network [18‐20] providing valuable insights into the key variables associated with interpersonal sensitivity. Additionally, this method enables a better understanding of the complex interplay between interpersonal sensitivity and other factors among nursing students during this challenging period. Enhancing the understanding of interpersonal sensitivity in this specific population can facilitate the promotion of their mental health and well-being, ensuring their ability to effectively cope with the challenges of the COVID-19 pandemic.

The current cross-sectional study was carried out at Jitang College of North China University of Science and Technology, located in Tangshan, Hebei, China. The study employed a single-centered design, with participants selected using a cluster sampling technique. Between November and December 2022, data were gathered via a self-report questionnaire. Due to the pandemic lockdown, students attended online classes during the survey. Trained investigators were responsible for delivering the questionnaire to the students. The Electronic “Questionnaire Star” tool (https://​www.​wjx.​cn/​) was employed for distributing questionnaires and gathering data from the participants. As a proficient online survey platform, “Questionnaire Star” exhibits several advantageous features, including efficiency, cost-effectiveness, ease of learning, and utilization. It has been previously utilized in various investigations about the COVID-19 pandemic [21, 22]. The students were provided with information regarding the study’s purpose and relevant considerations. Informed consent was obtained from all students, and data anonymization procedures were implemented to safeguard confidentiality. The inclusion criteria were as follows: (1) undergraduate nursing students currently enrolled at the university, (2) voluntary agreement to participate in the study, and (3) no prior involvement in similar studies.

The sample size was calculated using the formula commonly applied in cross-sectional studies: [n = (Z²_α/2pq)/d²] [23]. In this equation, n represents the required sample size; p denotes the prevalence rate of interpersonal sensitivity as reported in prior research; q is the complement of p, calculated as (1—p); Z_α/2 is set at 1.96, corresponding to a significance level of 0.05 for a two-tailed test; and d represents precision margin. According to earlier studies, the prevalence of interpersonal sensitivity among university students in China was found to be 58.1% [24]. As a result, a minimum sample size of 278 participants was required for the study. Given the availability of participants, this study ultimately enrolled 864 nursing students.

The Ethics Review Committee of Jitang College, North China University of Science and Technology, granted ethical approval (JTXY-2022–002) for this study. The research strictly adhered to the principles outlined in the Declaration of Helsinki and followed the Measures for Ethical Review of Biomedical Research Involving Human Beings.

The study employed a combination of self-designed questionnaires and standardized questionnaires.

The Patient Health Questionnaire-9 (PHQ-9) was applied to examine participants’ depression symptoms [30]. Each item is rated on a 4-point Likert scale, ranging from 0 (never) to 3 (nearly every day). Total scores on the PHQ-9 range from 0 to 27, with higher scores reflecting higher levels of depression symptoms. The Cronbach’s α of the PHQ-9 was 0.930 in this study.

The Generalized Anxiety Disorder-7 (GAD-7) was employed to assess participants’ depression symptoms [31]. Each item is rated on a 4-point Likert scale, ranging from 0 (never) to 3 (nearly every day). Total scores on the GAD-7 range from 0 to 21, with higher scores indicating higher levels of anxiety symptoms. The Cronbach’s α of the GAD-7 was 0.952 in this study.

The World Health Organization Well-Being Index-5 (WHO-5) was utilized to evaluate participants’ psychological well-being [32]. Each item is rated on a 6-point Likert scale, ranging from 0 (never before) to 5 (all times). Total scores on the WHO-5 range from 0 to 25, with higher scores indicating elevated levels of well-being. The Cronbach’s α for the WHO-5 was 0.959 in this study.

The Rosenberg Self-Esteem Scale (RSES) was used to measure participants’ self-esteem [33]. The RSES is composed of 10 items. Each item is rated on a 4-point Likert scale, ranging from 1 (strongly disagree) to 4 (strongly agree). Items 1, 3, 5, 7, and 9 are reverse scored. Total scores on the RSES range from 0 to 40, with higher scores signifying heightened levels of self-esteem. The Cronbach’s α for the RSES was 0.753 in this study.

The Three-Item Loneliness Scale (T-ILS) was employed to assess participants’ feelings of loneliness [34]. Each item is rated on a 3-point Likert scale, ranging from 1 (never) to 3 (often). Total scores on the T-ILS range from 3 to 9, with higher scores reflecting greater levels of loneliness. The Cronbach’s α for the T-ILS was 0.866 in this study.

The Perceived Social Support Scale (PSSS) was implemented to examine participants’ perceived social support [35]. The PSSS is composed of 3 items. Each item is rated on a 7-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree). Total scores on the PSSS range from 3 to 21, with higher scores suggesting greater perceived social support levels. The Cronbach’s α for the PSSS was 0.935 in this study.

The Family Communication Scale-10 (FCS-10) was used to appraise participants’ communication within the family [36]. Each item is rated on a 5-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree). Total scores on the FCS-10 range from 10 to 50, with higher scores denoting greater levels of family communication. The Cronbach’s α for the FCS-10 was 0.976 in this study.

First, the normality of continuous variables was assessed through the Kolmogorov–Smirnov test. A visual analysis of Q-Q plots revealed that the distribution of continuous variables exhibited a close approximation to normality. Descriptive statistics for continuous variables were presented as the mean and standard deviation (SD), and categorical variables were represented as numbers and percentages. Second, a univariate linear regression model was applied to investigate the association between the study variables and interpersonal sensitivity. Third, the variables demonstrating statistical significance (P value < 0.05) in the univariate linear regression model were chosen in the study and subsequently subjected to a multivariable linear regression model. Finally, the estimation of the network model (i.e., interpersonal sensitivity measure and interpersonal sensitivity exhibited statistically significant associations with variables in the multivariable linear regression model) was conducted employing the R packages “bootnet” and “qgraph” [37]. The first network model was carried out to assess measures of interpersonal sensitivity. The second network model was performed to investigate the associations between interpersonal sensitivity and statistically significant variables identified in the multivariable linear regression model.

In the network model, every variable was assigned the role of a “node”, while the interconnections among individual variables were regarded as “edges” [38]. The thickness of an edge within the network diagram represents the strength of the association between nodes, where thicker edges indicate stronger associations. For the present model, the default setting for the gamma hyperparameter was specified as 0.5, and the “EBICglasso” method was adopted as the default approach [39]. Thus, edge coefficients with smaller values were constrained to zero, leading to the identification of an optimal network.

The estimation of three commonly utilized node centrality indices was performed using the centrality plot function provided by the “qgraph” package [39]. Node strength captures the extent of direct connectivity of a node within the network and is determined by summing the weights of the edges connecting that particular node to other nodes in the network. Closeness reflects the indirect connectivity of a node within the network, computed as the reciprocal sum of the shortest path lengths between that node and all other nodes in the network. Betweenness represents the indirect connectivity of a node within the network, quantified by the count of times the node lies on the shortest path connecting two other nodes in the network. Strength centrality is prioritized due to its proportionality to the extent to which a specific node uniquely accounts for the variance in the nodes it is connected to. The centrality indices underwent standardization to obtain z scores.

To evaluate the accuracy and stability of the observed network model, two analyses were performed utilizing the “bootnet” package in R [37]. The first analysis involved estimating the stability of node centrality (specifically the strength index) using a case-drop bootstrap procedure with 1000 iterations. The second analysis focused on assessing the confidence intervals (CIs) of the edge weights using a nonparametric bootstrap procedure consisting of 1000 iterations. The stability of node strength is visually presented and evaluated using the correlation stability coefficient (CS-C), necessitating a value greater than 0.25 and ideally exceeding 0.50 [37]. Regarding bootstrapped CIs for edge weights, the presence of larger CIs indicates lower precision in estimating edges, whereas narrower CIs are indicative of a more reliable network [40].

In addition, network property differences, including node strengths and edge weights, were examined through bootstrapped difference tests [37]. The test employed 95% CIs to determine the statistical significance of differences between two edge weights or two node centrality indices.

This study comprised a sample of 864 nursing students, consisting of 180 (20.83%) males and 684 (79.17%) females. The mean scores for interpersonal sensitivity, psychological resilience, self-efficacy, well-being, self-esteem, loneliness, perceived social support, and family communication among nursing students were 43.68 (SD 9.41), 22.44 (SD 6.70), 10.62 (SD 2.20), 16.19 (SD 5.52), 27.54 (SD 3.75), 5.25 (SD 1.52), 15.29 (SD 3.71), and 37.04 (SD 7.97) points, respectively (Table 1). The mean score range of IPSM items was from 2.30 to 3.69 (Table 2).

The univariate linear regression analysis demonstrated that extraversion (β = −0.63), agreeableness (β = −0.56), conscientiousness (β = −1.06), neuroticism (β = 1.93), openness (β = 0.50), depression symptoms (β = 0.71), anxiety symptoms (β = 1.02), well-being (β = −0.77), self-esteem (β = −0.65), loneliness (β = 2.90), perceived social support (β = 0.30), and family communication (β = −0.08) had significant associations with interpersonal sensitivity (Table 3). The results of the multivariate linear regression analysis indicated that openness (β = 0.67), anxiety symptoms (β = 0.45), well-being (β = −0.63), loneliness (β = 1.55), and perceived social support (β = 0.66) were significantly associated with interpersonal sensitivity (Table 4).

The network of interpersonal sensitivity measures was shown in Fig. 1. The edge IPSM9-IPSM10 showed the strongest association. IPSM9 had the highest node strength in the network among nursing students, followed by IPSM2 and IPSM4 (Fig. 2). IPSM9 had the highest predictability in the network (Table 2).

The network of variables (i.e., interpersonal sensitivity, openness, anxiety symptoms, well-being, loneliness, and perceived social support) was shown in Fig. 3. Positive connections were found between interpersonal sensitivity and the variables of openness, anxiety symptoms, loneliness, and perceived social support. Negative connections were observed between interpersonal sensitivity and well-being (Fig. 3). Interpersonal sensitivity had the highest node strength in the network among nursing students, followed by anxiety symptoms and loneliness (Supplementary Table 1) (Supplementary Fig. 1).

In Supplementary Fig. 2 and Supplementary Fig. 3, the case-dropping bootstrap procedure (n = 1000) showed that the CS-Cs of node strength were 0.75, which revealed that 75% of samples could be dropped. The results of the nonparametric bootstrap procedure (n = 1000) indicated statistically significant differences in most comparisons involving edge weights and node strength (Supplementary Figs. 4–7). In addition, the narrow bootstrapped 95% CIs suggested the trustworthiness of the edges (Supplementary Fig. 8 and Supplementary Fig. 9).

The limitations of this study necessitate mention. First, the findings of our study may have limited generalizability due to the cross-sectional design and the inclusion of only one center. Second, our study’s inability to establish causality among variables should be noted. Future research endeavors should prioritize longitudinal studies to explore dynamic networks and identify the specific variables that adversely influence interpersonal sensitivity.