This study was a single-center, cohort study of patients who were hospitalized to undergo thoracic surgery at the Wuhan Union Hospital, a tertiary hospital located in Wuhan, China, from May 2022 to January 2023. Participants were recruited based on the following inclusion criteria: (1) they were at least 18 years old; (2) they had undergone lung surgery; and (3) they did not have a history of neurological or psychiatric disorders. The exclusion criteria included individuals who experienced critical conditions during the study period or those who did not complete follow-up assessments. In the context where the sample size exceeds 10 events per variable [13], and considering a potential 10% dropout rate, we would terminate the further enrollment of participants.

This study was a follow-up study based on earlier research [12] (Registration number: ChiCTR2200056651), and has been approved by the institutional ethics board of Wuhan Union Hospital of Tongji Medical College, Huazhong University of Science and Technology (No. 20220026‐2). All the participants provided written and oral informed consent. The original informed consent form, which was approved by the same ethics committee, explicitly allows for secondary analysis without the need for additional consent (No. 20200351). Participants were provided with free counseling services during the follow-up period as a form of compensation. All study data have been anonymized or deidentified to protect the privacy and confidentiality of the participants.

Before surgery and 6 months after surgery, we used the simplified Chinese version of the Fear of Pain-9 items (FOP-9) [14] to gather data on the FOP. The FOP-9 is a 5-point ordinal Likert scale where total scores range from 9 to 45, and a higher score signifies a greater level of FOP.

On the third day after surgery, the Physical Activity subscale of the Fear-Avoidance Beliefs Questionnaire (FABQ-PA) was used to measure the FAB. The FABQ-PA is a self-reported questionnaire comprising the first 5 questions of the original Fear-Avoidance Beliefs Questionnaire [15]. A higher score on this subscale signifies a greater tendency towards the FAB.

Psychological resilience was collected 1 month after surgery by using the Connor-Davidson Resilience Scale (CD-RISC) [16]. The CD-RISC comprises 25 items, each rated on a 5-point scale (0‐4), with higher scores reflecting greater psychological resilience.

Activity recovery was defined as the recovery of social activities and housework after discharge. Activity recovery was assessed at 6 months after surgery through questions such as, “Are your current social activities, like work and community activities, similar to what they were before the surgery?” and “Have your current family responsibilities returned to what they were before the surgery?” The answers were categorized as being lower or similar/higher than they were before surgery. Both the answers of “similar/higher” meant that the activity recovery was positive.

We used the adaptive least absolute shrinkage and selection operator (Lasso) with nested cross-validation to conduct variable selection, coefficient estimation, and performance prediction. All the statistics were completed using Python 3.12 (Python Software Foundation) and the scikit-learn library (developed by a community of open-source contributors, initially started by David Cournapeau). The process was as follows [17]: (1) Feature preprocessing: We conducted standardization on numeric features and one-hot encoding on categorical features. In order to avoid the problem of multicollinearity, we set the first one as the baseline and dropped it for each categorical variable, such as “Male” in sex, “Primary School” in education, “Presence” in chronic pain, and “Wedge Resection” in the surgery method. (2) Model training: We estimated the initial coefficients using Ridge regression with 5-fold cross-validation, and then conducted model training with the adaptive Lasso with 5-fold cross-validation after weighing the coefficients. (3) Model evaluation: In the 5-fold outer cross-validation, we selected the optimal model and calculated the average value of the model fitting parameters. (4) Variable estimation: We reported the regularization parameters and variable coefficients of the optimal model.

A total of 150 patients met the study’s inclusion criteria. However, 6 (4%) patients were excluded as they were lost to follow-up. A total of 144 participants were included in the final analysis, and the baseline characteristics are listed in Table 1. The study flow is shown in Figure 1.

The statistical analysis showed that the mean squared error was 14.273 (SD 3.254), indicating a reasonable prediction error range with room for improvement. The mean R^² value was 0.731 (SD 0.060), suggesting that the models could account for approximately 73.1% (SD 6%) of the variance in postoperative FOP and had a relatively good fit. The optimal alpha of the optimal model was 0.081. The coefficient of the optimal model revealed that the preoperative FOP, FAB, and pain intensity had significant positive effects on the postoperative FOP with coefficients of 8.620, 8.560, and 0.417, respectively. In contrast, psychological resilience and age had significant negative impacts, with coefficients of −5.822 and −2.853, respectively. Among categorical variables, patients with a university education had lower postoperative FOP compared to those with primary school education, while female patients had a higher postoperative FOP than male patients did. The surgical method and chronic pain had no significant effect on the postoperative FOP (Table 2).

The model performed excellently after analysis, with an average accuracy of 0.820 (SD 0.024) and an average area under the curve of 0.926 (SD 0.044), indicating that the model had a good predictive ability and stability for activity recovery. Psychological resilience, FAB, and pain intensity were significant factors that affected the activity recovery, with coefficients of 1.185, −0.966, and −0.170, respectively. Both university education and segmental resection significantly positively affected activity recovery compared to baseline, with coefficients of 0.283 and 0.042, respectively. Age, sex, chronic pain, and preoperative FOP had no significant effect on activity recovery (Table 3).

This study delved into the role of psychological resilience in perioperative FOP and avoidance behavior among surgical patients, particularly those undergoing lung surgery. The findings contribute novel insights to the existing literature, suggesting that psychological resilience may serve as a crucial factor in breaking the vicious cycle of FOP and facilitating recovery.

Our study employed adaptive Lasso regression under nested cross-validation to analyze the factors influencing postoperative FOP, a common phenomenon experienced by patients following surgery [18]. The developed model demonstrated robust performance, with a high proportion of variance explained in postoperative FOP. Among the factors, psychological resilience emerged as the most significant inhibitory factor influencing postoperative FOP. Preconceived biases regarding postoperative activities and pain can lead to the development of the FOP [19,20]. We observed that some patients strictly adhered to the doctor’s orders despite potential pain, while others rejected and avoided them. This disparity can be attributed to the varying levels of psychological resilience among individuals. Resilient individuals are more likely to accept and face pain, actively adjust their pain concepts, and gradually expose themselves to pain stimuli [21,22]. For instance, they may perceive postoperative pain as a normal part of the recovery process and actively communicate with medical staff to learn pain management strategies, gradually adapting to and overcoming the FOP. Conversely, individuals with lower psychological resilience may avoid stressful situations, including postoperative activities, leading to a vicious cycle of pain, fear, and avoidance [9,10,23].

The adaptive Lasso logistic regression exhibited higher accuracy and area under the curve values with a lower standard deviation, indicating that the model has high accuracy and stability in predicting postoperative activity recovery. The fact that the coefficient of psychological resilience is the largest revealed that psychological resilience had a stronger influence on postoperative activity recovery than other factors. This suggests that psychological resilience primarily enables individuals to initiate activities that may cause pain despite the presence of FOP, serving as a positive factor in overcoming fear. Patients with higher psychological resilience tend to have better ability to resume their activities postoperatively [24,25], and our research has also confirmed this. They may be more actively involved in social interactions and resume daily housework, thereby promoting physical function recovery and improving psychological well-being, resulting in a virtuous cycle. This may be because patients with high psychological resilience are better able to cope with negative emotions, have more reasonable cognitive expectations of the rehabilitation process, and can better overcome the fear of activities, thus promoting the recovery of postoperative activities [26,27].

This study has several limitations. First, psychological resilience was measured postoperatively, which may introduce bias due to the impact of surgery. Second, although both models demonstrated certain predictive abilities, there is room for improvement in the R^² values. Third, the generalizability of the findings is limited, as this study specifically focused on patients undergoing lung surgery, and the results may not be applicable to all surgical patients. Therefore, conducting multicenter studies with a bigger sample size would further validate the robustness of the results.

This study revealed the significant role of psychological resilience in postsurgical pain management, particularly in reducing the FOP and promoting recovery of activities. Individuals with higher psychological resilience are more likely to effectively cope with postsurgical pain and reduce avoidance behaviors. These findings underscore the importance of assessing and potentially enhancing psychological resilience during the perioperative period, especially for those with higher levels of FOP. By fostering resilience and promoting correct pain cognition, clinical practitioners can empower patients to overcome the FOP, engage in rehabilitation activities, and achieve better postoperative outcomes.