Table 5 Summary of study design and significant findings from included papers, grouped by dependent variable.

Brandberg et al. [42]

Mixed model ANOVA

First admission, 2 and 12 months

HADS-D

Treatment modality (plaque/enucleation)

No between group differences were found when depression was treated as a continuum or when clinical cut-offs were used.

Brown et al. [32] ^a[linked sample]

Hierarchical logistic regression (controlling anxiety and 6 and 12 months) and mediation analysis.

6, 12 and 24 months

HADS-D

Worry about cancer recurrence

Anxiety

Depression

Visual and ocular symptoms, visual function problems

Post treatment symptoms, Functional problems, Chromosome 3 status (M3 versus others), Gender, Age

24 months

Block 1

• 6 month (OR = 12.06) and 12 month (OR = 9.94) depression or anxiety caseness

Block 2 (6 month)

• Headache (OR = 2.32)

• Functional problems (OR = 1.16)

• Worry about recurrent disease (OR = 0.71)

Block 3

No significant predictors

Non significant

Age, gender, ocular symptoms, visual impairment, M3

Brown et al. [35] ^a[linked sample]

Growth curve modelling to estimate intercepts and slopes. Intercepts and slopes predicted by prognostic variable (M3, D3, not tested, test failed)

6, 12, 24, 36, 48 and 60 months

HADS-D

Chromosome 3 status, Prognostic test

M3 patients had higher depression intercepts than D3 or not tested. No difference between slopes. Thus, M3 predicts consistently higher depression than D3 or not tested. D3 did not differ from not tested.

Brown et al. [36] ^a[linked sample]

Cross-lagged analysis examining mediation of relationship between concerns about symptoms and depression by worry about cancer recurrence

6, 12, 24, 36, 48 and 60 months

HADS-D

Anxiety

Depression

Worry about cancer recurrence

Concerns about symptoms and function (linear combination of Visual and ocular symptoms, visual function problems),

Chromosome 3 status (M3 versus others),

Age

Gender

Treatment

12 months

12 month depression predicted by female gender.

36 months+

36 and 48 month depression predicted by 24 and 36 month concerns and worry about recurrence.

In males 12 moth depression predicted 24 month concerns, which predicted 36 month depression.

In females, 24 month depression predicted 34 month concerns which then predicted 36 month concerns, which predicted 48-month depression

Non-significant

M3, age, treatment, anxiety

Damato et al. [37] ^a[linked sample]

Bootstrapped t-test for six month follow-up. Multilevel regression for linear trend over all observations

Linear regression

6 months and annual observation for 10 years

HADS-D

Treatment modality

6 months

Radiotherapy patients less likely to experience depression (obs. coeff. –0.408).

No analysis of treatment differences over time

Damato et al. [38] ^a[linked sample]

Linear regression predicting depression from treatment controlling treatment, pre-treatment visual acuity, eye laterality, TNM size, melanoma cytomorphology and tumour type. Linear regression predicting depression from age, sex, time between treatment and questionnaire, general health, marital status, employment status and social support.

6 months

HADS-D

Social Support

Self report QoL

Treatment modality

Gender

Age

Marital status

Employment status

Self-reported general health

6 months

Enucleation associated with depression (linear regression, 95% confidence interval [CI], 0.17–1.01).

Less depression in radiotherapy compared to enucleation (p = 0.007). (univariable t-test)

Multivariable analysis

Poor social support (p < 0.001).

Employment associated with less depression (p < 0.001).

Better self-reported general health associated with less depression (p < 0.001).

Non-significant

Age, gender, marital status

Hope-Stone et al. [34] ^a[linked sample]

Mixed model multivariate ANOVA

6, 12 and 24 months

HADS-D

Treatment modality

Prognosis (Monosomy 3)

Age

Gender

6, 12 and 24 months

Monosomy 3 patients more depressed over all three time points (F = 6.23, df 1259, p < 0.05).

Non-significant

No difference in depression for treatment, age or gender.

Melia et al. [45]

Intention to treat analysis. Markov calculation of transition probabilities between diagnostic categories across timepoints compared using log rank tests. Logistic regression used to estimate ORs at separate timepoints controlling baseline category and other covariates. Intention to treat analysis

6 weeks after surgery, 6, 12, 24, 36, 48 and 60 months

HADS-D

Treatment modality

Probability of diagnoses did not differ across treatments. Depression scores did not differ by treatment.

Lieb et al. [43]

Mixed modelling controlling age, sex and enucleation.

First admission, 6–12 weeks, 6 and 12 months.

HADS-D

Social Support

General distress

Anxiety

Depression

QoL: physical

QoL: mental

Perceived risk

Resilience

Fear of progression

Prognostic testing

Enucleation

Graduation

Age

Gender

12 months

Lower resilience: B (SE) = 0.074 (0.032), β = −0.220, (p = 0.023)

Higher fear of progression: B (SE) = 1.326 (0.508), β = 0.224, (p = 0.010)

Lower physical quality of life: B (SE) − 0.067 (0.029), β = −0.142, (p = 0.022)

Higher depression at baseline: B (SE) 0.418 (0.090), β = 0.432, (p < 0.001)

Non-significant

age, sex, graduation, social support, general distress, anxiety, QoL mental, perceived risk, enucleation monosomy group

Schuermeyer et al. [40]

Mixed model 2 sided and Bonferroni adjusted

Immediately pre-treatment, 3 and 12 months

HADS-D

Anxiety

Depression

Decision Regret

Prognostication group

Age

Gender

12 months

Baseline anxiety (coefficient estimate predicts depression [SE], 0.39 [0.06]; P < 0.001)

Baseline anxiety and depression interact to predicted an increase in depression.

No association between depression over time and decision regret score at baseline, results of prognostication test or entry into adjuvant therapy trial. Gender or prognostic group.

Brandberg et al. [42]

Mixed model ANOVA

First admission, 2 and 12 months

HADS-A

Treatment modality (plaque/enucleation)

No between group differences were found when depression was treated as a continuum or when clinical cut-offs were used.

Brown et al. [32] ^a[linked sample]

Pearson and point-biseral correlation. Hierarchical logistic regression (controlling anxiety and 6 and 12 months) and mediation analysis.

6, 12 and 24 months

HADS-A

Worry about recurrent disease

Anxiety

Depression

Visual and ocular symptoms, visual function problems,

Chromosome 3 status, Gender,

Age

24 months

Block 1

• 6 month (OR = 9.86) and 12 month (OR = 7.56) depression or anxiety caseness

Block 2 (6 month)

• Headache (OR = 1.77)

• Functional problems (OR = 1.12)

• Occular irritation (OR = 1.23)

Block 3 (12 month)

• Worry about recurrent disease (OR = 1.34)

WREC at 12 months mediated prediction of anxiety by 6-month symptoms and functional problems.

Non significant

Age, gender

Brown et al. [35] ^a[linked sample]

Growth curve modelling to estimate intercepts and slopes. Intercepts and slopes predicted by prognostic variable (M3, D3, not tested, test failed)

6, 12, 24, 36, 48 and 60 months

HADS-A

Chromosome 3 status, Prognostic test

M3 patients had higher anxiety intercepts than D3 or not tested. No difference between slopes. Interpretation is that M3 predicts consistently higher anxiety than D3 or not tested. D3 did not differ from not tested.

Brown et al. [36] ^a[linked sample]

Pearson correlations, cross-lagged analysis examining mediation of relationship between concerns about symptoms and depression by worry about cancer recurrence

6, 12, 24, 36, 48 and 60 months

HADS-A

Anxiety

Depression

Worry about cancer recurrence

Concerns about symptoms and function (linear combination of Visual and ocular symptoms, visual function problems),

Chromosome 3 status (M3 versus others), Gender,

Age

12 months

12 month anxiety predicted by 6 month concerns. and worry about recurrence. 24 month anxiety predicted by worry about recurrence.

36 months

36 month anxiety predicted by 24 month concerns and worry about recurrence.

60 months

60 month anxiety predicted by 48 month worry about recurrence. Prediction of 60 month anxiety by 48 month concerns may be mediated by 48 month worry.

In males 12-month anxiety predicted 24 month worry, which then predicted 36 month anxiety

Non-significant

M3, age, gender, anxiety, depression

Damato et al. [37] ^a[linked sample]

Bootstrapped t-test for six month follow-up. Multilevel regression for linear trend over all observations. Linear regression.

6 months and annual observation for 10 years

HADS-A

Treatment modality

No 6-month differences in anxiety. No analysis of treatment differences over time.

Damato et al. [38] ^a[linked sample]

Linear regression predicting anxiety from treatment controlling treatment, pretreatment visual acuity, eye laterality, TNM size, melanoma cytomorphology and tumour type. Linear regression predicting anxiety from age, sex, time between treatment and questionnaire, general health, marital status, employment status and social support.

6 months

HADS-A

Social Support

Self report QoL

Treatment modality

Gender

Age

Marital status

Employment status

Self-reported general health

6 months

Female sex (p < 0.001). SE = 0.19

Chromosome 3 loss (p = 0.039) SE = 0.36

Older age reduced anxiety - Younger age (p < 0.001). SE = 0.01

Poor social support (p < 0.001) SE = 0.07

Better general health reduced anxiety - Poor physical health (p < 0.001) SE = 0.09

Non-significant

No difference between treatment, employment status, marital status for anxiety.

Hope-Stone et al. [34] ^a[linked sample]

Mixed model multivariate ANOVA.

6, 12, 24 months

HADS-A

Treatment modality

Prognosis (Monosomy 3)

Age

Gender

24 months (6, 12 and 24 months)

Younger patients compared to older (F = 13.21, df 1259, p < 0.01)

Female sex compared to male (F = 19.75, df 1259, p < 0.01)

Non-significant

No difference in anxiety between treatment, and monosomy 3.

Lieb et al. [43]

Mixed modelling controlling age, sex and enucleation.

Hospital admission, discharge, 6–12 weeks, and 6 and 12 months.

HADS-A

Fear of progression

Social Support

General distress

Anxiety

Depression

QoL: physical

QoL: mental

Perceived risk

Resilience

Fear of progression

Prognostic testing

Enucleation

Graduation

Age

Gender

12 months

Baseline younger age: B (SE) = −0.050 (.023), β = −0.147, (p = 0.031)

Baseline higher fear of progression: B (SE) = 1.744 (0.524), β = 0.298, (p = 0.001)

Bassline higher anxiety: B (SE) = 0.212 (0.102), β = 0.215 (p = 0.039)

Baseline lower physical QoL: B (SE) = − 0.061 (0.030), β = −0.131 (p = 0.042)

Baseline lower mental QoL: B (SE) = −0.061 (0.030), β = −0.178 (p = 0.043)

Non-significant

Graduation, resilience, social support, general distress, depression, perceived risk, enucleation and monosomy group non-significant.

No significant results for fear of progression between groups.

Neither the mean score for anxiety

(x¯ = 5.38, SE = 3.84) reached the cut off for a clinically relevant condition at t4.

Melia et al. [45]

Intention to treat analysis. Markov calculation of transition probabilities between diagnostic categories across timepoints compared using log rank tests. Logistic regression used to estimate ORs at separate timepoints controlling baseline category and other covariates. Intention to treat analysis.

6 weeks after surgery, 6, 12, 24, 36, 48 and 60 months

HADS-A

Treatment modality

Anxiety scores did not differ by treatment.

Schuermeyer et al. [40]

Paired t-test

Mixed model 2 sided and Bonferroni adjusted

Immediately pre treatment, 3 and 12 months

HADS-A

Anxiety

Depression

Decision Regret

Prognostication group

Age

Gender

Age interacted with anxiety, age predicted a decrease with older age (coefficient estimate

[SE], −0.06 [0.02]; P < 0.001).

No association between anxiety over time and decision regret score at baseline, results of prognostication test or entry into adjuvant therapy trial.

Van Beek et al. [44]

Multilevel hierarchical regression.

Second admission, 2, 6, 12, 24, 36 and 48 months

STAI (state trait anxiety)

Treatment modality

Metastases/No metastases

36 months+

Patients with metastatic disease greater anxiety at 36 (p = 0.026, d = 0.59) and 48 months (p = 0.022, d = 0.81).

Non-significant

Treatment modality

Brandberg et al. [42]

Mixed model ANOVA

First admission, 2 and 12 months

IES

Treatment modality (plaque/enucleation)

No significant predictors.

Lieb et al. [43]

Mixed modelling controlling age, sex and enucleation.

Hospital admission, discharge, 6–12 weeks, and 6 and 12 months.

Distress Thermometer

Social Support

General distress

Anxiety

Depression

QoL: physical

QoL: mental

Perceived risk

Resilience

Fear of progression

Prognostic testing

Enucleation

Graduation

Age

Gender

12 months

Higher level of distress at baseline B (SE) 0.452 (0.105), β = 0.450, p < 0.001

Non-significant:

age, sex, graduation, resilience, social support, fear of progression, depression, anxiety, QoL physical and mental, perceived risk, enucleation, monosomy group

Van Beek et al. [44]

Multilevel hierarchical regression.

Cox regression analysis.

Second admission, 2, 6, 12, 24, 36 and 48 months

IES

Treatment modality

Metastases/No metastases

Patients with mets scored slightly higher on intrusion at baseline p = 0.037 than those without. No sig effects in follow ups.

No difference between enucleated and fSRT patients for IES scores.

Brandberg et al. [42]

Mixed model ANOVA.

First admission, 2 and 12 months

EORTC QLQ-C30

Treatment modality (plaque/enucleation)

No significant differences between plaque and enucleation on QoL.

Brown et al. [35] ^a[linked sample]

Growth curve modelling to estimate intercepts and slopes. Intercepts and slopes predicted by prognostic variable (M3, D3, not tested, test failed)

6, 12, 24, 36, 48 and 60 months

EORTCQLQ C30 (WREC)

Chromosome 3 status

Age, Sex

M3 patients had higher worry about cancer recurrence intercepts than D3 or not tested. No difference between slopes. Thus, M3 predicts consistently higher worry about cancer recurrence than D3 or not tested. D3 did not differ from not tested.

Brown et al. [36] ^a[linked sample]

Pearson correlations, cross-lagged analysis examining mediation of relationship between concerns about symptoms and depression by worry about cancer recurrence

6, 12, 24, 36, 48 and 60 months

EORTCQLQ C30 (WREC)

Anxiety

Depression

Worry about cancer recurrence

Concerns about symptoms and function (linear combination of Visual and ocular symptoms, visual function problems),

Chromosome 3 status (M3 versus others), Gender,

Age

Younger age predicted worry about recurrence.

M3 (M3 patients vs all others) predicted worry about recurrence.

Damato et al. [37]^a[linked sample]

Bootstrapped t-test for six month follow-up. Multilevel regression for linear trend over all observationsLinear regression.

6 months and annual observation for 10 years

EORTC QLQ-OPT30

FACT-G

Treatment modality

6 Months

Radiotherapy was associated with fears of local recurrence (obs. coeff., 0.291)

Radiotherapy less likely to experience worries about metastasis (obs. coeff., –0.164)

Patients who had radiotherapy were less likely to suffer from loss of emotional well-being (obs. coeff., 0.518).

60 months+

Enucleation late loss of emotional well being (obs. coeff., linear: –0.181; quadratic: –0.011)

Damato et al. [38] ^a[linked sample]

Linear regression predicting QoL from treatment controlling treatment, pre-treatment visual acuity, eye laterality, TNM size, melanoma cytomorphology and tumour type. Linear regression predicting QoL from age, sex, time between treatment and questionnaire, general health, marital status, employment status and social support.

6 months

EORTC QLQ-OPT30

FACT-G

Self-reported QoL

Social support

Treatment modality

Gender

Age

Marital status

Employment status

Self-reported general health

6 months

Radiotherapy experienced less worries about local recurrence (p < 0.001, OR = 0.58), metastasis (p < 0.001, OR = 1.49), health (p < 0.001, OR = 1.42, (p < 0.001, OR = 1.72) losing eye and appearance (p < 0.001, OR = 2.90)

Radiotherapy better emotional well-being (p = 0.012, SE = 0.22). (t-test)

Female sex better emotional wellbeing (p < 0.001, SE = 0.20).

Older ager increased emotional wellbeing (p < 0.001, SE = 0.01).

Poor social support showed lower emotional well-being (p < 0.001, SE = 0.08).

Being employed was associated with better emotional well-being (p = 0.002, SE = 0.24).

Better self-reported general health was associated with better emotional wellbeing (p < 0.001, SE = 0.09)

Non-significant

FACT-G and treatment type

Hope-Stone et al. [33] ^a[linked sample]

Mixed model ANOVA, controlling chromosome 3 status, age, sex, health problems, logMAR (fellow eye), tumour thickness and diameter. MANOVA

6, 12 and 24 months

EORTC QLQ-OPT30 (WREC)

Treatment modality

No significant difference for WREC.

Hope-Stone et al. [34] ^a[linked sample]

Mixed model multivariate ANOVA

6, 12 and 24 months

FACT-G

Treatment modality

Prognosis (Monosomy 3)

Age

Gender

No significant predictors.

Lieb et al. [43]

Mixed modelling controlling age, sex and enucleation.

Hospital admission, discharge, 6–12 weeks, and 6 and 12 months.

SF-12

Fear of progression

General distress

Anxiety

Depression

QoL: physical

QoL: mental

Perceived risk

Resilience

Fear of progression

Social Support

Prognostic testing

Enucleation

Graduation

Age

Gender

No prediction of QoL mental or fear of progression.

Klingenstein [41]

Mixed model ANOVA

First admission, 12 and 24 months.

SF-12v2

Glaucoma

Best Corrected Visual Acuity

Gender

Overall

Mental health improved (p = 0.023) after radiotherapy.

Mental health increased significantly (p = 0.0003) in the no glaucoma group

12 months

After 1 year mental health was significantly higher within the male subgroup (p = 0.042) but not in 2 years.

24 months

Mental health was higher compared with no glaucoma patients at the second follow up (2 years) (p = 0.02).

Non-significant

No differences between role emotional between glaucoma groups or male vs. female group.

No difference between BCVA and mental health or role emotional.

Melia et al. [45]

Generalised linear model.

6 weeks after surgery, 6, 12, 24, 36, 48 and 60 months

SF-36

Treatment modality

6 months

Brachytherapy increase in mental health and emotional role function scales compared to enucleation (p < 0.05) however difference between both not significant.

No change for SF-36 over follow up.

Van Beek et al. [44]

Multilevel linear regression.

Second admission, 2, 6, 12, 36, 48 months

EORTC QLQ C-30

Treatment modality

Metastases/No metastases

36 months

Patients with metastases showed lower emotional functioning after 36 months. (score 83.4; 95% CI, 78.4–88.4, p = 0.022, Student’s t test)

Non-significant

Treatment modality