Anxiety levels were assessed using the state subscale of the State-Trait Anxiety Inventory (STAI-S). Studies have confirmed the multidimensional factor structure of the Chinese version of the scale and demonstrated good psychometric properties, including adequate convergent and discriminant validity. In this sample, the STAI-S demonstrated excellent internal consistency (Cronbach's α = .95). The STAI-S consists of 20 items that assess anxiety-related feelings, thoughts, and behaviors at the time of assessment. Each item is rated on a 4-point Likert scale ranging from 1 ("not at all") to 4 ("very much"). Items 1, 2, 5, 8, 10, 11, 15, 16, 19, and 20 are reverse-scored. The total score ranges from 20 to 80, with scores between 20 and 39 indicating mild anxiety, 40 to 59 indicating moderate anxiety, and 60 to 80 indicating severe anxiety.

Perceived job stress was measured using the 14-item Work Pressure Inventory developed by Huang et al. (2017), which has demonstrated good internal consistency. The scale comprises three dimensions: low self-development, workload, and job characteristics. Each item is rated on a 5-point Likert scale ranging from 1 ("strongly disagree") to 5 ("strongly agree"), with total scores ranging from 14 to 70. Higher scores indicate greater perceived occupational stress. In the original validation study, the Cronbach's α coefficients for the three subscales were 0.81, 0.73, and 0.77, respectively. In the present study, the Work Pressure Inventory showed good overall internal consistency (Cronbach's α = .84).

Occupational burnout was assessed using the Chinese version of the Copenhagen Burnout Inventory (CBI), which has demonstrated good psychometric properties in Taiwanese populations. The scale consists of four subscales: personal burnout, work-related burnout, client-related burnout, and overcommitment to work. Each item is rated on a five-point frequency scale: "always" (100), "often" (75), "sometimes" (50), "rarely" (25), and "never" (0). Subscale scores are calculated as the average of the items within each domain, ranging from 0 to 100, with higher scores indicating more severe occupational burnout. The original validation study reported Cronbach's α values above 0.84 across all subscales. In the present study, the scale demonstrated excellent internal consistency (Cronbach's α = .95).

Emotional distress was measured using the Distress Thermometer (DT), a single self-report screening instrument with a score range of 0 (no distress) to 10 (extreme distress), with higher scores indicating greater emotional distress. The DT demonstrated good psychometric properties in validation studies, with sensitivities ranging from 0.50 to 1.00 (median = 0.83) and specificities ranging from 0.36 to 0.98 (median = 0.68).

This study used a DT cutoff score of ≥3 as the inclusion criterion. Other studies have shown that the optimal DT cutoff score varies across settings, typically ranging from 3 to 5, while thresholds of ≥4 or ≥5 are commonly used in clinical practice. The use of a score of ≥3 in this study was intended to maximize sensitivity and minimize the risk of underidentifying caregivers considered at high risk for psychological distress.

Total of all reporting groups

160 subjects filled in the proportion of each group according to gender

Use percentages to look at the distribution of male and female students in the control and intervention groups.

Race and Ethnicity were not collected from any participant.

The 160 subjects were mainly clinical nurses at Kaohsiung Veterans General Hospital in Taiwan.

All 160 subjects were admitted in Taiwan, so the number of people can be used to represent it.

Psychological distress was measured using the Brief Symptom Rating Scale-5 (BSRS-5), a validated screening tool for general psychological distress. The BSRS-5 assesses the subjective severity of the following symptoms: (1) anxiety, (2) depression, (3) hostility, (4) low self-esteem, and (5) insomnia. Each symptom was scored on a 5-point Likert scale, ranging from 0 ("not at all") to 4 ("extremely"), with a total score ranging from 0 to 20 points. Higher scores indicate more severe psychological distress.

Studies have shown that a total score of 3-4 is the optimal threshold for identifying clinically relevant distress based on receiver operating characteristic (ROC) curve analysis. The BSRS-5 showed high accuracy (AUC = 0.92) and good sensitivity (0.83) and specificity (0.86). Therefore, this study used a BSRS-5 total score ≥4 as one of the inclusion criteria to ensure that participants with at least mild psychological distress were included in the study.

After the pre-test questionnaire assessment (baseline week 0), the subjects (control group/intervention group) will be followed up with weekly questionnaires for eight weeks (weeks 1 to 8) after the start of the program, of which the Mood Thermometer (Brief Symptoms Rating Scale, BSRS-5) questionnaire will obtain data for a total of nine weeks (baseline week 0 and follow-up weeks 1 to 8) and conduct data analysis.

This scale assesses an individual's ability to adapt and recover from stress and adversity. Each item is rated on a 5-point Likert scale, ranging from 1 ("strongly disagree") to 5 ("strongly agree"), with a total score ranging from 10 to 50, with higher scores indicating greater resilience.

Initial validation studies demonstrated strong psychometric properties, including good model fit in confirmatory factor analysis (GFI = 0.973) and excellent internal consistency (Cronbach's α = .91).

All instruments used in this study were authorized by their original developers and have been psychometrically validated in previous studies. All scales demonstrated good to excellent internal consistency, with Cronbach's α values ranging from 0.84 to 0.95.

After the pre-test questionnaire assessment (baseline week 0), the subjects (control group/intervention group) will be followed up with weekly questionnaires for eight weeks (weeks 1 to 8) after the start of the program, of which the Resilience questionnaire will obtain data for a total of nine weeks (baseline week 0 and follow-up weeks 1 to 8) and conduct data analysis.

Anxiety levels were assessed using the state subscale of the State-Trait Anxiety Inventory (STAI-S). Studies have confirmed the multidimensional factor structure of the Chinese version of the scale and demonstrated good psychometric properties, including adequate convergent and discriminant validity. In this sample, the STAI-S demonstrated excellent internal consistency (Cronbach's α = .95). The STAI-S consists of 20 items that assess anxiety-related feelings, thoughts, and behaviors at the time of assessment. Each item is rated on a 4-point Likert scale ranging from 1 ("not at all") to 4 ("very much"). Items 1, 2, 5, 8, 10, 11, 15, 16, 19, and 20 are reverse-scored. The total score ranges from 20 to 80, with scores between 20 and 39 indicating mild anxiety, 40 to 59 indicating moderate anxiety, and 60 to 80 indicating severe anxiety.

After the pre-test questionnaire assessment (baseline week 0), the subjects (control group/intervention group) will be followed up with weekly questionnaires for eight weeks (weeks 1 to 8) after the start of the program. The anxiety questionnaire will obtain data for a total of nine weeks (baseline week 0 and follow-up weeks 1 to 8) and conduct data analysis.

Perceived job stress was measured using the 14-item Work Pressure Inventory developed by Huang et al. (2017), which has demonstrated good internal consistency. The scale comprises three dimensions: low self-development, workload, and job characteristics. Each item is rated on a 5-point Likert scale ranging from 1 ("strongly disagree") to 5 ("strongly agree"), with total scores ranging from 14 to 70. Higher scores indicate greater perceived occupational stress. In the original validation study, the Cronbach's α coefficients for the three subscales were 0.81, 0.73, and 0.77, respectively. In the present study, the Work Pressure Inventory showed good overall internal consistency (Cronbach's α = .84).

After the pre-test questionnaire assessment (baseline week 0), the subjects (control group/intervention group) will be followed up with weekly questionnaires for eight weeks (weeks 1 to 8) after the start of the program. The Nurse Stress questionnaire will obtain data for a total of nine weeks (baseline week 0 and follow-up weeks 1 to 8) and conduct data analysis.

Occupational burnout was assessed using the Chinese version of the Copenhagen Burnout Inventory (CBI), which has demonstrated good psychometric properties in Taiwanese populations. The scale consists of four subscales: personal burnout, work-related burnout, client-related burnout, and overcommitment to work. Each item is rated on a five-point frequency scale: "always" (100), "often" (75), "sometimes" (50), "rarely" (25), and "never" (0). Subscale scores are calculated as the average of the items within each domain, ranging from 0 to 100, with higher scores indicating more severe occupational burnout. The original validation study reported Cronbach's α values above 0.84 across all subscales. In the present study, the scale demonstrated excellent internal consistency (Cronbach's α = .95).

After the pre-test questionnaire assessment (baseline week 0), the subjects (control group/intervention group) will be followed up with weekly questionnaires for eight weeks (weeks 1 to 8) after the start of the program. The workplace fatigue questionnaire will obtain data for a total of nine weeks (baseline week 0 and follow-up weeks 1 to 8) and conduct data analysis.

Emotional distress was measured using the Distress Thermometer (DT), a single self-report screening instrument with a score range of 0 (no distress) to 10 (extreme distress), with higher scores indicating greater emotional distress. The DT demonstrated good psychometric properties in validation studies, with sensitivities ranging from 0.50 to 1.00 (median = 0.83) and specificities ranging from 0.36 to 0.98 (median = 0.68).

This study used a DT cutoff score of ≥3 as the inclusion criterion. Other studies have shown that the optimal DT cutoff score varies across settings, typically ranging from 3 to 5, while thresholds of ≥4 or ≥5 are commonly used in clinical practice. The use of a score of ≥3 in this study was intended to maximize sensitivity and minimize the risk of underidentifying caregivers considered at high risk for psychological distress.

After the pre-test questionnaire assessment (baseline week 0), the subjects (control group/intervention group) will be followed up with weekly questionnaires for eight weeks (weeks 1 to 8) after the start of the program. A total of nine weeks (baseline week 0 and follow-up weeks 1 to 8) of Distress Thermometer questionnaire data will be obtained and analyzed.

From enrollment until end of follow-up, up to 8 weeks.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 2.

This is the second week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 3.

This is the third week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 4.

This is the fourth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 5.

This is the fifth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 6.

This is the sixth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 7.

This is the seventh week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 8.

This is the eighth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 2.

This is the second week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 3.

This is the third week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 4.

This is the fourth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 5.

This is the fifth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 6.

This is the sixth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 7.

This is the seventh week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 8.

This is the eighth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 2.

This is the second week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 3.

This is the third week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 4.

This is the fourth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 5.

This is the fifth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 6.

This is the sixth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 7.

This is the seventh week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 8.

This is the eighth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 2.

This is the second week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 3.

This is the third week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 4.

This is the fourth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 5.

This is the fifth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 6.

This is the sixth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 7.

This is the seventh week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 8.

This is the eighth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 2.

This is the second week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 3.

This is the third week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 4.

This is the fourth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 5.

This is the fifth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 6.

This is the sixth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 7.

This is the seventh week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 8.

This is the eighth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 2.

This is the second week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 3.

This is the third week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized Estimating Equations were used to analyze repeated measures data and to assess the interaction effect between group and time. Potential interfering factors such as age and years of work experience were controlled during the analysis, and the effect size (Cohen's d) and its 95% confidence interval were calculated to assist in interpreting the clinical significance. Statistical analysis was performed using SPSS version XX or R version XX, and the significance level was set at p < 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 4.

This is the fourth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 5.

This is the fifth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 6.

This is the sixth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 7.

This is the seventh week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.

Generalized estimating equations (GEE) were chosen over repeated-measures analysis of variance or generalized linear mixed models because they provide estimates of the population mean, are robust to correlated error specifications, can accommodate missing data under the MAR/MCAR assumptions, and do not require sphericity. All tests were two-tailed, with α = 0.05.

Repeated-measures GEE was used to examine the group × time interaction between week 0 and week 8.

This is the eighth week of tracking data. A P value < 0.05 indicated a statistically significant difference.

A generalized estimating equation (GEE) model included group (intervention group vs. control group), week (categorical: 0-8), group × week interaction, and age as covariates. Robust (sandwich) standard errors were used. Regression coefficients (B), standard errors, Wald chi-squared statistics, 95% confidence intervals, and p-values for group × week comparisons are reported.