Depression symptoms, resilience and perceived disease uncertainty in Chinese glaucoma patients: a cross-lagged panel tracking study protocol

Abstract

Introduction

This study systematically examines the prevalence and inter-relationships of depressive symptoms, resilience and illness uncertainty in patients with primary angle-closure glaucoma (PACG), elucidating their dynamic, bidirectional mechanisms over time to inform the development of integrated biopsychosocial management strategies for PACG in clinical practice.

Methods and analysis

A prospective longitudinal study design will be employed. A total of 610 patients undergoing primary PACG surgery at the Ophthalmology Department of Chengde Medical College Affiliated Hospital will be recruited by convenience sampling between November 2025 and January 2027. Participants will be assessed at three time points—on admission (T1), 1 month after discharge (T2) and 3 months after discharge (T3)—using the Patient Health Questionnaire-9, Connor-Davidson Resilience Scale, and Mishel Uncertainty in Illness Scale to evaluate levels of depressive symptoms, resilience and illness uncertainty. Demographic and baseline clinical data will also be collected.

For statistical analyses, Harman’s single-factor test for common method bias and longitudinal measurement invariance analyses will first be conducted. Subsequently, Pearson correlation analysis will be performed to explore associations among variables. A cross-lagged panel model will then be constructed to estimate autoregressive paths, cross-lagged paths and contemporaneous covariances among the three variables. Finally, a parallel multiple mediation model will be established to examine the indirect effects of illness uncertainty on subsequent depressive symptoms through depressive symptoms and resilience as mediators, using bootstrapping. Covariates will be selected through univariate screening and multicollinearity diagnostics and included in all models; continuous covariates will be mean-centred before entering the models, while categorical variables will be dummy-coded.

Ethics and dissemination

This study has been approved by the Ethics Committee of Chengde Medical College Affiliated Hospital (Approval No.: CYFYLL2025447). Written informed consent will be obtained from all participants before enrolment. All study data will be anonymised and kept strictly confidential, with access restricted to authorised members of the research team. The findings will be submitted to peer-reviewed, open-access journals and presented at academic conferences to facilitate broad dissemination.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study employs a three-wave longitudinal design to examine the dynamic and reciprocal relationships among depressive symptoms, resilience and illness uncertainty in patients with primary angle-closure glaucoma, thereby addressing a gap in longitudinal evidence in this field.

• It integrates longitudinal measurement invariance testing, cross-lagged panel modelling and parallel multiple mediation modelling to simultaneously estimate autoregressive, cross-lagged and indirect effects, providing stronger evidence regarding temporal and reciprocal associations than cross-sectional designs.

• The single-centre design may result in a relatively homogeneous sample in terms of regional culture and healthcare-seeking behaviours, which may limit the generalisability of the findings.

• The 3-month follow-up period is relatively short, making it difficult to capture long-term lagged effects and elucidate the longer-term mechanisms underlying the observed relationships.

• All data are derived from self-report scales, which may introduce recall and reporting biases.

Introduction

Glaucoma is a progressive optic neuropathy characterised by irreversible apoptosis of retinal ganglion cells, optic disc cupping and visual field defects, ultimately leading to irreversible vision loss. It has become a major global public health challenge in the 21st century. Epidemiological data show that over 76 million people were affected worldwide in 2020, with projections estimating that this number will reach 111.8 million by 2040, nearly 60% of whom will reside in Asia, with China remaining one of the countries with the highest glaucoma burden globally.^1–4 Primary angle-closure glaucoma (PACG), caused by anatomical closure of the anterior chamber angle and a consequent rapid increase in intraocular pressure, carries a higher risk of blindness and therefore warrants urgent research attention.^{5 6} Global eye health initiatives have clearly emphasised the goal of achieving universal access to eye care services by 2030, eliminating public misconceptions and integrating vision loss prevention and treatment with patients’ psychological rehabilitation, underscoring the importance of a multidimensional approach to eye health management.²

Psychological comorbidities are highly prevalent among patients with glaucoma, with depressive symptoms being the most common psychiatric comorbidity. The Global Burden of Disease Study (2019) ranked depressive disorders first among mental health-related causes of disability-adjusted life-years globally and 13th among all disease categories.⁷ Data from the China Mental Health Survey indicate a lifetime prevalence of major depressive disorder among Chinese adults of 6.8%.^{8 9} Studies have reported a depressive symptom detection rate of 25.78% in glaucoma populations, which is significantly associated with age, sex, socioeconomic status, social support, sleep quality, severity of visual impairment and comorbid medical conditions.^10–12 Resilience, a core protective factor in psychological adaptation to chronic illness, has been shown to effectively reduce the risk of depression and buffer the adverse psychosocial impacts of disease.^13–16 Resilience levels are themselves influenced by marital status, occupation, social support and age.^17–19 Meanwhile, as glaucoma is an irreversible, progressive chronic eye disease, patients frequently experience illness uncertainty stemming from ambiguous diagnostic trajectories and an unpredictable long-term prognosis; this uncertainty is further modulated by socioeconomic status, educational attainment, marital status, comorbidities, sleep disturbances and perceived social support.^{20 21} Elevated illness uncertainty not only undermines treatment adherence and contributes to psychological maladjustment but may also exacerbate disease progression through impaired self-management behaviours and diminished health-related quality of life, thereby constituting a critical and indispensable dimension of comprehensive glaucoma care.^{22 23}

Most existing studies on this topic have used cross-sectional designs, which cannot elucidate the dynamic interactions or causal relationships among resilience, illness uncertainty and depressive symptoms.^24–28 Within the methodological framework of longitudinal data analysis, researchers face a range of model choices, each suited to addressing distinct research questions. The latent growth model is primarily used to characterise individual developmental trajectories and growth trends; however, it cannot reveal inter-relationships among multiple variables over time.²⁹ The autoregressive model examines the stability of a single variable across time points but cannot estimate cross-lagged effects between variables simultaneously.³⁰ In contrast, the cross-lagged panel model (CLPM) combines the strengths of both approaches: it simultaneously estimates autoregressive effects, synchronous correlations, and cross-lagged effects, thereby clarifying directional temporal associations and reciprocal influences among variables. This renders the CLPM especially suitable for investigating the dynamic psychological mechanisms underlying chronic disease progression.^31–33 In recent years, the random-intercept CLPM (RI-CLPM) has been applied in psychological research on chronic diseases to disentangle between-person and within-person effects.³⁴ However, its interpretive framework centres on within-person change, which differs fundamentally from the traditional CLPM’s focus on between-person temporal associations.^{35 36} This study aims to examine the temporal directions among resilience, illness uncertainty, and depressive symptoms at the between-person level. Given its methodological maturity, parsimony and broad empirical support, the traditional CLPM aligns most closely with our analytical objectives. Accordingly, we adopt a prospective three-wave longitudinal design, enrolling patients with PACG and administering assessments at admission (T1), 1 month postdischarge (T2) and 3 months postdischarge (T3). A CLPM will be estimated to systematically test dynamic reciprocal pathways among the three constructs, thereby addressing a critical gap in longitudinal research on PACG-related psychosocial adaptation and informing theory-driven psychological interventions for patients with glaucoma.

Study objectives

1. To examine the longitudinal trajectories of depressive symptoms, resilience and illness uncertainty in patients with glaucoma.

2. To test cross-lagged reciprocal associations among depressive symptoms, resilience and illness uncertainty.

3. To explore the longitudinal mediating effects of depressive symptoms and resilience in the association between illness uncertainty and subsequent depressive symptoms.

Research hypotheses and conceptual model

Hypothesis 1 (H1): Depressive symptoms, resilience and illness uncertainty are significantly intercorrelated at baseline in patients with glaucoma.

Hypothesis 2 (H2): Each of these three constructs at an earlier time point will significantly predict the other two constructs at the subsequent time point, after controlling for autoregressive stability and synchronous associations.

The conceptual framework of this study is presented in figure 1.

Figure 1. Conceptual framework.

Research methods and analysis

Research design and research setting

This prospective longitudinal cohort study was approved by the Ethics Committee of Chengde Medical College Affiliated Hospital on 7 November 2025 (Approval No.: CYEYLL2025447). The study will be conducted from November 2025 to January 2027 in the Department of Ophthalmology at Chengde Medical College Affiliated Hospital, Chengde, Hebei Province, China.

The study site is a grade A tertiary general hospital and serves as a provincial-level regional medical centre. It undertakes the majority of complex ophthalmic diagnosis and treatment for Chengde City and the surrounding areas. Chengde is located in northern Hebei Province; the hospital’s catchment area covers northern Hebei and adjacent regions of Inner Mongolia, including multiple prefecture-level cities, banners and counties. The patient population reflects a mixed urban–rural composition.

Chengde’s economic development level ranks mid-tier within Hebei Province. Among its subordinate counties and districts are both nationally designated poverty-stricken counties and economically developed districts. Consequently, the patient population exhibits considerable heterogeneity in income level, type of health insurance and healthcare access. The convenience sampling strategy in this study is expected to capture a broad spectrum of cases spanning urban and rural settings and diverse socioeconomic backgrounds, thereby enhancing sample representativeness and strengthening the external generalisability of the findings.

This study will be conducted in accordance with the Declaration of Helsinki. All participants will provide written informed consent prior to enrolment. The study protocol has been developed in accordance with the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) Checklist, and the final study will be reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement. The study flow chart is presented in figure 2.

Figure 2. Research flow chart. PACG, primary angle-closure glaucoma.

Follow-up schedule

The longitudinal assessment time points for this study were determined based on previous longitudinal studies in glaucoma populations and practical experience gained from our pilot study.

The European Glaucoma Society Terminology and Guidelines for Glaucoma recommend that follow-up frequency be individualised, with initial follow-up visits scheduled at 1 week, 1 month and 6 months.³⁷ In previous longitudinal studies of visual field damage progression in patients with primary open-angle glaucoma, researchers conducted follow-up assessments at multiple time points, including the day of discharge and 3 and 6 months postoperatively, to dynamically assess changes in disease status.³⁸ Another longitudinal study reported that the incidence of new depressive symptoms gradually declined over follow-up assessments conducted at baseline, 6 months, 12 months, 18 months and 24 months.³⁹ Our pilot study indicated that patient compliance was satisfactory during the first 3 months of follow-up. Based on these findings and practical considerations, we established three assessment time points for this study: at admission (T1), 1 month after discharge (T2) and 3 months after discharge (T3). During hospitalisation, participants will complete the questionnaires face-to-face in the ward, with assistance from trained research staff when necessary. After discharge, follow-up assessments will be conducted through telephone interviews, during which trained research staff will administer the questionnaires item by item and record participants’ responses.

To minimise loss to follow-up and reduce potential attrition bias, a multifaceted participant retention strategy will be implemented:

1. During the informed consent process, the study procedures and follow-up schedule will be explained in detail both verbally and in writing to ensure that participants have a clear and realistic understanding of the study requirements. The importance of their continued participation in maintaining the value of the study will also be emphasised.

2. At baseline, each participant will be asked to provide at least two independent contact methods, including a mobile phone number, landline telephone number, WeChat account and/or email address. Contact information for one family member will also be collected as a backup. Participants will also receive a reminder card listing their scheduled follow-up dates.

3. During the follow-up period, research staff will send standardised text message reminders and/or make telephone calls 3–5 days before each scheduled follow-up assessment to remind participants of the upcoming assessment.

4. After completion of each questionnaire, participants will be provided with a brief standardised health education leaflet or verbal summary containing general eye-care recommendations and stress-management tips. These materials will be non-diagnostic and non-interventional and will be provided to enhance participants’ engagement and perceived value of continued participation.

5. If a participant misses a scheduled follow-up, the research team will proactively contact the participant within 1 week using all available contact methods and will attempt to arrange a make-up assessment. Up to three contact attempts will be made at different times. If a participant explicitly withdraws from the study or remains unreachable after three attempts, the date of the last contact, the known reason for loss to follow-up, and the contact status will be documented in detail. All reasons for failure to participate in subsequent assessments due to loss to follow-up will be categorised and summarised to facilitate assessment of potential attrition bias.

Patient eligibility criteria

Inclusion criteria

1. Age ≥18 years.

2. Diagnosis of PACG according to the Chinese Glaucoma Guidelines (2020).⁴⁰

3. Undergoing primary surgical intervention for angle-closure glaucoma.

4. Being alert and oriented, with intact communication and comprehension abilities.

5. Providing written informed consent and voluntarily agreeing to participate in this study.

Exclusion criteria

1. History of psychiatric disorders or cognitive impairment.

2. Presence of concomitant ocular diseases, such as cataract or retinopathy.

3. Withdrawal from the study, loss to follow-up or incomplete outcome data.

Follow-up assessments

Face-to-face assessments will be conducted during hospitalisation. After discharge, follow-up assessments will be conducted through telephone interviews, during which trained research staff will administer the questionnaires item by item and record participants’ responses.

Data collection and quality control

All patient screening, baseline assessments, multitime-point follow-ups and data collection will be performed by a uniformly trained research team in strict accordance with standardised protocols. A three-tier quality control procedure will be implemented throughout the study to ensure scientific rigour and data reliability. Completed original questionnaires will be sequentially coded. Questionnaire data will be entered and independently verified by two reviewers. If any discrepancies arise, the original questionnaires will be consulted to verify the accuracy of the entered data.

Demographic and clinical data

A general information questionnaire will be developed by the research team based on a literature review and expert panel discussions, in accordance with the study objectives. The questionnaire will consist of two sections:

1. Demographic characteristics: sex, body mass index, smoking status, alcohol consumption, age, educational level, marital status, occupation, monthly household income, long-term place of residence and living arrangement.

2. Clinical data: affected eye(s), visual acuity of the affected eye(s), intraocular pressure of the affected eye(s), sleep quality, self-care ability, comorbidities and complications.

Nine-item patient health questionnaire

The Patient Health Questionnaire-9 (PHQ-9) was developed by Kroenke et al.⁴¹ It is a widely used clinical instrument for assessing the severity of depressive symptoms. The scale consists of nine items based on the depressive symptom diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, covering common depressive symptoms. Respondents will rate each item based on their experiences during the previous 2 weeks using a 4-point Likert scale, with each item scored from 0 to 3. The total score is calculated by summing all item scores, ranging from 0 to 27. Scores of 0–4 indicate no depressive symptoms; 5–9 indicate mild depressive symptoms; 10–14 indicate moderate depressive symptoms and 15–27 indicate severe depressive symptoms.

Connor-davidson resilience scale

The Connor-Davidson Resilience Scale (CD-RISC) is a psychometric instrument used to assess individual resilience. It was originally developed by Connor and Davidson.⁴² The Chinese version of the scale was culturally adapted and validated by Yu and Zhang.⁴³ The CD-RISC is designed to measure an individual’s ability to adapt to and recover from life stress, reflecting the capacity to maintain psychological well-being and a positive outlook when facing adversity or challenges. The scale comprises 25 items across three dimensions: competence, trust in instincts, and acceptance of change (also described as personal competence, tolerance of negative affect and strengthening effects of stress). Each item is rated on a 5-point Likert scale ranging from 0 to 4. The total score ranges from 0 to 100, with higher scores indicating greater psychological resilience.

Mishel uncertainty in illness scale for adults

The Mishel Uncertainty in Illness Scale (MUIS) was originally developed by Mishel.⁴⁴ It was subsequently introduced into China and culturally adapted by Xu and Huang,⁴⁵ and later revised and revalidated by Ye et al.⁴⁶ The scale consists of 20 items covering three dimensions: ambiguity, complexity and unpredictability. Each item is rated on a 5-point Likert scale ranging from 1 to 5. After reverse scoring of five items, the total score ranges from 20 to 100, with higher scores indicating higher levels of illness uncertainty. The study schedule is presented in table 1.

Table 1. Research timeline.

Item	Study period
	T1: On the day of admission	T2: 1 month after discharge	T3: 3 months after discharge
Enrolment
Eligibility screening	√
Informed consent	√
Data collection
Baseline characteristics	√
Depressive symptoms (PHQ-9）	√	√	√
Resilience (CD-RISC）	√	√	√
Disease uncertainty (MUIS）	√	√	√

CD-RISC, Connor-Davidson Resilience Scale; MUIS, Mishel Uncertainty in Illness Scale for Adult; PHQ-9, 9-item Patient Health Questionnaire.

Sample size estimation

An a priori statistical power analysis will be conducted using the R package semPower.⁴⁷ Based on empirical benchmarks for cross-lagged effect sizes reported by Orth et al,⁴⁸ the expected standardised effect size will be set at a medium level (standardised path coefficient β=0.07). The df for the hypothesis test will be set to 2. Specifically, the two cross-lagged paths, X₁→Y₂ and X₂→Y₃, will be simultaneously constrained to zero.

The significance level will be set at α=0.05, and the statistical power will be set at 1−β=0.80. The results indicated that a sample of 549 participants would be required to detect this effect size. This number will then be adjusted for an anticipated 10% dropout rate. Accordingly, the target sample size will be determined as 610 participants to ensure that at least 549 valid cases remain after data cleaning for model estimation and hypothesis testing.

If the actual dropout rate unexpectedly exceeds 10% during study implementation, the research team will initiate supplementary recruitment in a timely manner to maintain a final valid sample size at or above the threshold required by the power analysis.

Statistical analysis

Data will be double-entered using Excel. All statistical analyses will be performed using R software (V.4.5.3). The summary of the research variables is presented in table 2.

Table 2. Summary of research variables.

Measuring tools	Variables
General Demographic Questionnaire	Demographic data: gender, age, body mass index, smoking status, drinking status, education level, marital status, occupation, family monthly income, long-term residence location, living arrangement; Clinical data: affected eye, visual acuity on the affected side, intraocular pressure on the affected side, sleep quality, self-care ability, comorbidities, complications
PHQ-9	Depressive symptoms
CD-RISC	Resilience
MUIS	Disease uncertainty

CD-RISC, Connor-Davidson Resilience Scale; MUIS, Mishel Uncertainty in Illness Scale; PHQ-9, 9-item Patient Health Questionnaire.

Missing data

Given the three-wave longitudinal design of this study, the missing data mechanism will first be examined using Little’s MCAR test. If the proportion of missing data for each variable is below 5%, complete-case analysis will be performed. If the proportion of missing data ranges from 5% to 10%, multiple imputation will be adopted as the primary analytical approach. The robustness of the findings will be assessed through sensitivity analyses using complete cases. If the proportion of missing data for a single variable exceeds 10%, the research team will consider excluding that variable or removing its corresponding paths from the model based on the extent and pattern of missingness.

Descriptive analysis

Descriptive analyses will be performed using R software. The normality of continuous variables will be assessed using the Shapiro-Wilk test in combination with Q-Q plots. Normally distributed continuous variables will be presented as mean±SD, whereas non-normally distributed continuous variables will be presented as median and IQR (M (Q1, Q3)). Categorical variables will be presented as frequencies and percentages (n (%)). All statistical tests will be two-sided, and a p<0.05 will be considered statistically significant.

Before the formal analyses, the potential severity of common method bias will be assessed. Given that all data will be collected using self-report scales, common method bias may exist. Harman’s single-factor test will therefore be conducted using the psych package in R. An unrotated principal component analysis will be performed on all scale items, and factors with eigenvalues greater than 1 will be extracted. The proportion of variance explained by the first unrotated factor will then be examined. If the variance explained by the first factor is ≤40% and no single factor accounts for the majority of variance, common method bias will be considered unlikely to be substantial. Conversely, if these criteria are not met, the presence of potentially serious common method bias will be indicated.

Longitudinal invariance testing

Longitudinal measurement invariance will be tested separately for the PHQ-9, CD-RISC and MUIS to determine whether the measurement properties of each latent construct remain consistent across time points, thereby ensuring the validity of subsequent longitudinal comparisons. The R package lavaan will be used to construct single-factor confirmatory factor analysis models. Robust maximum likelihood estimation will be applied. Three nested models—configural, metric and scalar invariance models—will be tested sequentially, and their fit indices will be compared. Changes in model fit will be evaluated using Comparative Fit Index (ΔCFI) and root mean square error of approximation (ΔRMSEA). Measurement invariance will be considered established if ΔCFI ≤0.010 and ΔRMSEA ≤0.015.

Covariate adjustment

To control for potential confounding effects, all demographic and clinical variables collected using the general information questionnaire will be statistically screened to identify covariates for inclusion in the final models. Sex and age will be included directly in the final models. For the remaining variables, univariable analyses will first be performed. Pearson correlation analysis will be used for continuous variables to assess their associations with scores on the three core scales (PHQ-9, CD-RISC and MUIS) at each of the three time points. Independent-samples t tests or one-way analysis of variance will be used for binary and nominal categorical variables, respectively, to compare scale scores between groups at each time point. Variables with p<0.10 in the univariable analyses will be retained for subsequent modelling. Second, multicollinearity among the continuous variables retained after the initial screening will be assessed by calculating the variance inflation factor (VIF). A VIF ≥10 will be considered indicative of severe multicollinearity; in such cases, one of the collinear variables will be excluded based on clinical relevance. Categorical covariates will be dummy coded, whereas continuous covariate will be centred before being entered into the models. The covariates retained after these procedures will be incorporated into the CLPM and longitudinal mediation model for parameter estimation.

Correlation analysis

The ‘Hmisc’ package in R will be used to compute the correlation matrix among the study variables. To explore the patterns of associations among PHQ-9, CD-RISC and MUIS scores across the three time points, scatterplot matrices will be generated to assess linear trends. Pearson or Spearman correlation analyses will then be performed, as appropriate. Significance testing for correlation coefficients will be two-sided. Both the correlation coefficient matrix and the corresponding p value matrix will be generated using the rcorr function.

Cross-lagged panel analysis: A structural equation model will be constructed using the ‘lavaan’ package in R. Missing data will be handled using full information maximum likelihood estimation. First, a CLPM will be developed to simultaneously estimate autoregressive paths, cross-lagged paths, and contemporaneous correlations among PHQ-9, CD-RISC and MUIS. Modification indices will be consulted to identify potential additional paths, which will only be added when theoretically justified. Models will be estimated using maximum likelihood estimation. Model fit will be evaluated using the χ² statistic, CFI, Tucker-Lewis index (TLI), RMSEA, and standardised root mean square residual (SRMR). Acceptable model fit will be defined as CFI ≥0.90, TLI ≥0.90, RMSEA ≤0.08, and SRMR ≤0.08. The full model will be compared with a more parsimonious model using the χ² difference test, and the simpler model will be selected if it provides an adequate fit. All path coefficients will be reported as standardised coefficients (β), unstandardised coefficients (B), 95% CI and corresponding significance levels.

Longitudinal mediation analysis

To examine the underlying mechanism through which illness uncertainty influences depressive symptoms, a parallel multiple mediation model will be constructed. T1 MUIS will be specified as the independent variable, T3 PHQ-9 as the dependent variable, and T2 PHQ-9, T2 CD-RISC and T2 MUIS as parallel mediators. Three indirect pathways will be estimated simultaneously: (1) T1_MUIS → T2_PHQ-9 → T3_PHQ-9; (2) T1_MUIS → T2_CD-RISC → T3_PHQ-9; and (3) T1_MUIS → T2_MUIS → T3_PHQ-9. The model will estimate the path coefficients from the independent variable to each mediator (a1, a2 and a3), the path coefficients from each mediator to the dependent variable (b1, b2 and b3), and the direct effect of the independent variable on the dependent variable (c). The significance of indirect effects will be tested using the bootstrap method with 5000 resamples. An indirect effect will be considered statistically significant if its 95% CI does not include zero. All path coefficients will be reported as unstandardised estimates (B), with corresponding 95% CIs and p values.

Network analysis

Based on the nine items of the PHQ-9 and the three-dimensional scores of the CD-RISC and MUIS at T1, network construction and analysis will be performed using the R packages ‘qgraph’, ‘bootnet’, ‘networktools’ and ‘ggplot2’. The EBICglasso method will be used to estimate a partial correlation network through a penalised likelihood function, with the regularisation parameter λ automatically selected based on the EBIC. The network will be visualised using the Fruchterman-Reingold spring layout to illustrate clustering structures. Positive edges will be represented by grey solid lines, whereas negative edges will be represented by red dashed lines. Only edges with an absolute weight ≥0.1 will be displayed.

For centrality estimation, node strength, expected influence, closeness centrality and betweenness centrality will be calculated. In addition, bridge strength, bridge expected influence, bridge closeness and bridge betweenness centrality will be calculated to assess the cross-community influence of each node within the modular network. To evaluate network accuracy and stability, 95% CIs for edge weights will be estimated using non-parametric bootstrapping with 5000 resamples to assess the precision of edge estimates. Furthermore, a case-dropping bootstrap procedure will be conducted, with 10%–50% of cases randomly removed. The correlation between node strength rankings derived from the reduced samples and those from the full sample will be calculated to quantify the stability of centrality indices.

Patient and public involvement

No formal patient or public involvement activities were conducted during the design phase of this study. The research questions, outcome measures and assessment instruments were primarily determined based on previous clinical observations, literature reviews and relevant policy guidelines. During the pilot study, participants’ subjective experiences regarding the comprehensibility of the scale items, response time and follow-up arrangements were collected through oral interviews. No significant barriers were identified from the feedback, confirming the clinical acceptability of the formal study protocol. During the formal data collection phase, an open-ended comment section and experience evaluation items will be included at the end of the questionnaire. Investigators will also record any questions or difficulties raised by respondents during questionnaire completion.

Ethics and dissemination

This study was approved by the Ethics Committee of Chengde Medical College Affiliated Hospital (Approval No.: CYFYLL2025447). All participants provided written informed consent before formal enrolment. The informed consent form clearly described the study objectives, procedures, potential risks and benefits, the voluntary nature of participation, the right to withdraw and measures for ensuring data confidentiality.

For patients with severe visual impairment who were unable to independently complete the written consent form, a researcher read the consent form aloud item by item. After obtaining oral consent from the participant, an independent witness signed the form to confirm the consent process. Participants had the right to withdraw from the study at any time without providing a reason, and withdrawal did not affect their subsequent clinical care.

All original questionnaires and data were anonymised through coding. The linkage file connecting participant names with identification codes was encrypted and stored separately from the study data by designated personnel. Electronic data were stored on password-protected computers, with access restricted to core members of the research team. Study results will be presented only in aggregated form on publication, without any personally identifiable information.

Following publication of the academic paper, the full text will be made openly available to participants and the public. The study findings will also be submitted to an open-access journal for dissemination to maximise the societal impact of the research. These dissemination activities will be conducted in strict accordance with ethical standards, ensuring that the information shared is accurate and truthful while avoiding unnecessary anxiety or misunderstanding.

Discussion

This study will adopt a prospective longitudinal design. Patients undergoing first-time PACG surgery will be enrolled as participants. Repeated measurements will be conducted at three time points: on admission (T1), 1 month after discharge (T2) and 3 months after discharge (T3). A CLPM will be used to systematically examine the levels of depressive symptoms, resilience and illness uncertainty among patients with PACG, as well as the dynamic reciprocal relationships among these three variables over time. This study aims to address the gap in longitudinal research in this field.

The study design is scientifically robust. The study has been approved by the ethics committee and will be reported in accordance with the STROBE guidelines. Standardised scales with established reliability and validity will be used for outcome assessment. A three-tier quality control procedure, double data entry and standardised approaches for handling missing data will be implemented to ensure data reliability and the validity of statistical analyses. The sample size was calculated based on statistical formulae and adjusted for the expected dropout rate, resulting in a target sample size of 610 participants. This sample size is expected to provide sufficient statistical power for the planned cross-lagged analyses.

The association between depressive symptoms and illness uncertainty has been systematically examined across various disease populations and public health settings. Illness uncertainty has consistently been found to be positively associated with depressive symptoms. Furthermore, baseline illness uncertainty has been shown to significantly predict depressive symptom levels at the 3-month follow-up.^{24 49 50} Multiple cross-sectional studies have demonstrated an interactive relationship between depressive symptoms and resilience.^51–53 Using longitudinal designs, including CLPMs and RI-CLPMs, existing studies have demonstrated a complex bidirectional association between depressive symptoms and resilience, with variations observed across different populations.^{54 55} However, research examining the relationship between resilience and illness uncertainty remains limited.

This study is expected to clarify the longitudinal trajectories of depressive symptoms, resilience and illness uncertainty in patients with PACG. It will identify the longitudinal mediating effects among the three variables and reveal their synchronous correlations and cross-lagged predictive effects across time. Furthermore, the study will test the protective role of resilience in the association between illness uncertainty and depressive symptoms, thereby providing a scientific basis for understanding the development and progression of psychological comorbidities in glaucoma patients. In addition to the CLPM, this protocol also prespecifies network analysis as a complementary analytical approach. The two analytical strategies are highly complementary methodologically across multiple dimensions. Specifically, they differ in the level of analysis (total scores vs individual items), the temporal dimension (cross-lagged predictive effects across time vs cross-sectional associations) and the focus for clinical intervention (variable-level intervention targets vs symptom-level intervention nodes). Cross-validation of the results from both methods may help identify key psychological nodes and sensitive time windows in PACG patients, thereby enhancing the robustness and precision of statistical inference. The findings will provide theoretical support for developing integrated mind–body management strategies for PACG patients in clinical practice. They will assist clinical staff in the early identification of individuals at high risk for depression. Through targeted interventions, clinicians can enhance patients’ resilience and reduce illness uncertainty, thereby decreasing the incidence of depressive symptoms and improving postoperative psychological adaptation and overall recovery outcomes. This aligns with the Global Eye Health Initiative’s emphasis on multidimensional management.

This study has several limitations. First, the single-centre sampling may result in a sample that is highly homogeneous in terms of regional cultural characteristics and healthcare-seeking behaviours, which limits the external validity of the findings. Second, the 3 months postoperative follow-up period is relatively short, making it difficult to capture the long-term lagged effects among depressive symptoms, resilience and illness uncertainty and thus insufficient to fully elucidate the long-term developmental mechanisms underlying these variables. Finally, all data are derived from self-report scales, which may introduce recall bias and reporting bias. Future research could adopt multicentre designs with longer follow-up periods and incorporate structured clinical interviews or observer-rated assessments to more robustly validate the clinical benefits of integrating psychological rehabilitation into glaucoma management.