Latest / Glaucoma, Vision & Longevity: Supplements & Science / Does Dietary Sugar Intake Increase Glaucoma Risk? A Systematic Review and Meta-analysis Blueprint

Transcript

Does Dietary Sugar Intake Increase Glaucoma Risk? A Systematic Review and Meta-analysis Blueprint · Glaucoma, Vision & Longevity: Supplements & Science

• 0:00
  SPEAKER_01: Does dietary sugar intake increase glaucoma risk?
• 0:03
  SPEAKER_01: A systematic review and meta-analysis blueprint introduction.
• 0:07
  SPEAKER_01: Whether dietary sugar contributes to glaucoma risk remains uncertain.
• 0:12
  SPEAKER_01: The available research is limited and does not consistently measure added sugar itself.
• 0:17
  SPEAKER_01: Some studies have examined total carbohydrate intake, low carbohydrate dietary patterns, processed sweets, or ultra-processed foods instead.
• 0:25
  SPEAKER_01: For example, a prospective Spanish cohort found that participants with the highest total carbohydrate intake had a higher risk of newly diagnosed glaucoma, although no individual carbohydrate food group was clearly responsible.
• 0:39
  SPEAKER_01: In contrast, a large analysis of three United States cohorts found no clear association between overall low carbohydrate dietary scores and primary open angle glaucoma.
• 0:50
  SPEAKER_01: Although a vegetable-based low carbohydrate pattern showed a possible association with a specific visual field loss subtype.
• 0:58
  SPEAKER_01: A small case control study also reported that patients with primary open-angle glaucoma consumed more sweets containing processed sugar than controls, but its retrospective design leaves substantial potential for recall bias and residual confounding.
• 1:14
  SPEAKER_01: A 2024 scoping review concluded that most glaucoma nutrition studies relied on food frequency questionnaires, and that only a small number of prospective studies had examined carbohydrate-related exposures.
• 1:27
  SPEAKER_01: This protocol is designed to determine whether total sugar, added sugar, freezer sugary beverages, total carbohydrate, glycemic load, or glycemic index are associated with incident primary open-angle glaucoma, change in intraocular pressure, retinal nerve fiber layer loss, or glaucoma progression.
• 1:46
  SPEAKER_01: The review should estimate associations carefully without implying that sugar causes glaucoma.
• 1:52
  SPEAKER_01: Dietary studies are vulnerable to measurement error, confounding, reverse causation, and differences in glaucoma diagnosis.
• 2:00
  SPEAKER_01: Review objective and research question.
• 2:03
  SPEAKER_01: Primary research question among adults is higher habitual intake of dietary sugar or sugar-related exposures associated with a greater risk of developing glaucoma or experiencing glaucoma-related structural or pressure changes.
• 2:17
  SPEAKER_01: Secondary research questions.
• 2:19
  SPEAKER_01: The review should determine whether associations differ between added sugar, total sugar, free sugar, and sugar-sweetened beverages, whether glycemic load or glycemic index shows a stronger relationship than sugar grams alone.
• 2:33
  SPEAKER_01: Whether associations differ according to baseline diabetes status, whether findings differ between primary open angle glaucoma, normal tension glaucoma, high tension glaucoma, exfoliative glaucoma, pigmentary glaucoma, and primary angle closure glaucoma.
• 2:49
  SPEAKER_01: Whether associations are stronger for glaucoma incidence or for disease progression.
• 2:54
  SPEAKER_01: How much results are affected by adjustment for diabetes, body mass index, blood pressure, and total energy intake.
• 3:02
  SPEAKER_01: Whether food frequency questionnaire measurement error may have weakened or distorted reported associations.
• 3:08
  SPEAKER_01: Prespecified causal estimates.
• 3:10
  SPEAKER_01: Because diabetes, body mass index, blood pressure, and total energy intake may play different roles in the causal pathway, the review should distinguish between two related but different questions.
• 3:22
  SPEAKER_01: Total dietary association.
• 3:24
  SPEAKER_01: Does a higher sugar exposure relate to glaucoma risk through all plausible pathways, including pathways involving weight gain, diabetes, blood pressure, and vascular or metabolic changes?
• 3:36
  SPEAKER_01: Metabolic independent association.
• 3:38
  SPEAKER_01: Does sugar remain associated with glaucoma after adjustment for diabetes, body mass index, and blood pressure?
• 3:46
  SPEAKER_01: The second estimate should not be described as the total effect of sugar.
• 3:50
  SPEAKER_01: It is better described as a conditional or metabolically adjusted association because adjustment may remove part of the pathway through which diet could influence eye health.
• 4:00
  SPEAKER_01: Eligibility criteria, population, include studies of human adults aged 18 years or older, general population samples, occupational or community-based cohorts, patients with ocular hypertension or established glaucoma, participants with or without diabetes.
• 4:18
  SPEAKER_01: Studies involving children may be included in a separate exploratory analysis if sufficient evidence exists, but they should not be pooled with adult studies.
• 4:27
  SPEAKER_01: Eligible study designs include prospective cohort studies, longitudinal population-based studies, nested case control studies within prospective cohorts, case control studies in which dietary exposure was assessed for a period before glaucoma diagnosis or progression, retrospective analyses only if dietary exposure clearly preceded the outcome, and the study provides usable effect estimates.
• 4:52
  SPEAKER_01: Prospective studies should be the main evidence base for incidence and progression.
• 4:57
  SPEAKER_01: Case control studies should be synthesized separately because their exposure measurement is more vulnerable to recall and selection bias.
• 5:05
  SPEAKER_01: Required exposure timing.
• 5:07
  SPEAKER_01: The exposure must be measured before glaucoma diagnosis, before the progression period being studied, or repeatedly during follow-up in a way that permits a time-ordered analysis.
• 5:21
  SPEAKER_01: Studies measuring diet and glaucoma at the same visit should be excluded from the primary quantitative synthesis because they cannot establish whether dietary exposure preceded the disease.
• 5:32
  SPEAKER_01: They may be retained for narrative context.
• 5:34
  SPEAKER_01: Required outcome information.
• 5:36
  SPEAKER_01: Studies must report at least one of the following incident glaucoma, change in intraocular pressure, retinal nerve fiber layer loss, glaucoma progression, visual field progression, optic nerve or neuroretinal rim progression.
• 5:50
  SPEAKER_01: Studies must provide a relative risk, hazard ratio, odds ratio, mean difference, annualized rate of change, or enough data to calculate one, exclude, exclude.
• 5:59
  SPEAKER_01: Animal or laboratory studies, ecological studies, case reports and case series, studies of blood glucose or glycated hemoglobin without a dietary exposure, studies focused only on diabetic retinopathy, studies that do not distinguish glaucoma from nonspecific visual impairment, studies in which the dietary exposure is a nutritional supplement rather than food or beverage intake.
• 6:23
  SPEAKER_01: Duplicate reports from the same cohort unless they provide non-overlapping exposure or outcome data.
• 6:29
  SPEAKER_01: If several publications use the same cohort, select the report with the longest follow-up or the most complete outcome ascertainment.
• 6:37
  SPEAKER_01: Additional reports may be used for subtype or progression analyses without counting participants twice.
• 6:43
  SPEAKER_01: Exposure definitions.
• 6:46
  SPEAKER_01: Sugar-related exposures must not be treated as interchangeable.
• 6:50
  SPEAKER_01: The protocol should extract each exposure separately and avoid pooling them unless the definitions are sufficiently comparable.
• 6:58
  SPEAKER_01: Exposure classification table.
• 7:00
  SPEAKER_01: For the full table, please open this article on VisualFieldTest.com.
• 7:04
  SPEAKER_01: The United States Food and Drug Administration distinguishes total sugars from added sugars.
• 7:10
  SPEAKER_01: Total sugars include sugars naturally present in foods such as fruit and milk, whereas added sugars are incorporated during processing or preparation.
• 7:19
  SPEAKER_01: The World Health Organization defines free sugars more broadly to include added sugars and sugars naturally present in honey, syrups, fruit juice, and fruit juice concentrate while excluding the sugars naturally present in intact fruit, vegetables, and milk.
• 7:36
  SPEAKER_01: Sugar-sweetened beverages.
• 7:38
  SPEAKER_01: Sugar-sweetened beverages should be analyzed separately from solid foods because they are often consumed quickly, may produce a high glycemic exposure, usually provide few or no grams of fiber, may be associated with total energy intake and weight gain, can be measured more reliably in some questionnaires than total added sugar.
• 7:59
  SPEAKER_01: Diet beverages should not be combined with sugar-sweetened beverages.
• 8:02
  SPEAKER_01: They should be analyzed separately as an exploratory exposure because people who choose artificially sweetened drinks may differ systematically from people who do not.
• 8:12
  SPEAKER_01: 100% fruit juice should also be analyzed separately.
• 8:16
  SPEAKER_01: It should not automatically be classified as equivalent to either a sugar-sweetened beverage or intact fruit.
• 8:23
  SPEAKER_01: Glycemic load.
• 8:24
  SPEAKER_01: For studies providing sufficient information, glycemic load should be calculated or extracted as.
• 8:31
  SPEAKER_01: Glycemic load equals glycemic index, grams of available carbohydrate per serving, to 100.
• 8:38
  SPEAKER_01: Glycemic load reflects both the quality and quantity of carbohydrate.
• 8:42
  SPEAKER_01: Glycemic index reflects the relative glucose response to a carbohydrate-containing food, whereas glycemic load also accounts for the amount consumed.
• 8:51
  SPEAKER_01: Total carbohydrate, glycemic load, and added sugar should not be pooled into a single exposure category.
• 8:57
  SPEAKER_01: A diet may contain a large amount of carbohydrate from whole grains or legumes, but relatively little added sugar.
• 9:04
  SPEAKER_01: Exposure timing.
• 9:05
  SPEAKER_01: Extract whether exposure was measured at baseline only, as a cumulative average of repeated questionnaires, as the most recent diet measure, as a change from baseline, using a lagged exposure, such as diet measured two or more years before diagnosis.
• 9:21
  SPEAKER_01: Repeated and cumulative dietary measures should be preferred for chronic outcomes when available because a single questionnaire may not represent long-term dietary habits.
• 9:32
  SPEAKER_01: Outcome definitions.
• 9:33
  SPEAKER_01: 4.1.
• 9:34
  SPEAKER_01: Incident primary open angle glaucoma.
• 9:37
  SPEAKER_01: The primary incidence outcome should be newly diagnosed primary open angle glaucoma among participants free of glaucoma at baseline.
• 9:45
  SPEAKER_01: The preferred case definition should include an open anterior chamber angle, characteristic optic nerve or neuroretinal rim damage, corresponding visual field loss when available, absence of a secondary cause of glaucoma.
• 9:59
  SPEAKER_01: The American Academy of Ophthalmology describes primary open angle glaucoma as a chronic, progressive optic neuropathy with an open angle, characteristic optic nerve damage, and no secondary cause of elevated eye pressure.
• 10:13
  SPEAKER_01: Studies that use only self-reported glaucoma should be included in a sensitivity analysis, not automatically treated as equivalent to studies using ophthalmic examination, visual field testing, optic nerve imaging, or medical record confirmation.
• 10:28
  SPEAKER_01: 4.2 Intraocular pressure change.
• 10:31
  SPEAKER_01: Intraocular pressure should be analyzed as a separate continuous outcome rather than as a substitute for glaucoma.
• 10:37
  SPEAKER_01: Extract.
• 10:38
  SPEAKER_01: Baseline intraocular pressure, follow-up intraocular pressure, mean change from baseline, annualized change when repeated measurements are available.
• 10:47
  SPEAKER_01: Measurement method, time of day, use of pressure lowering treatment, whether both eyes or only one eye were analyzed.
• 10:55
  SPEAKER_01: The primary analysis should use the mean difference in millimeters of mercury.
• 11:00
  SPEAKER_01: If studies report slopes, the preferred outcome should be the difference in annual intraocular pressure change.
• 11:07
  SPEAKER_01: If the study reports baseline and follow-up standard deviations, but not the standard deviation of the change, the review team may reconstruct it using an assumed baseline to follow-up correlation.
• 11:18
  SPEAKER_01: Sensitivity analyses should repeat the calculation using several plausible correlations.
• 11:25
  SPEAKER_01: 4.3.
• 11:26
  SPEAKER_01: Retinal nerve fiber layer loss.
• 11:28
  SPEAKER_01: Retinal nerve fiber layer loss should be defined using optical coherence tomography or another validated structural imaging method.
• 11:36
  SPEAKER_01: Preferred measures include mean retinal nerve fiber layer change in micrometers, annualized retinal nerve fiber layer thinning, sector-specific thinning, especially inferior or superior sectors, event-based progression determined by the study's validated software, trend-based progression based on repeated imaging.
• 11:55
  SPEAKER_01: The primary structural analysis should use micrometers per year where possible.
• 12:00
  SPEAKER_01: Studies using substantially different imaging technologies may require a standardized mean difference or separate subgroup analysis.
• 12:09
  SPEAKER_01: The European Glaucoma Society identifies optic nerve, retinal nerve fiber layer, macular, and visual field assessment as complementary parts of glaucoma evaluation and monitoring.
• 12:19
  SPEAKER_01: 4.4 Secondary Outcomes.
• 12:22
  SPEAKER_01: Secondary outcomes should include glaucoma progression defined by visual field worsening, change in visual field mean deviation, optic disc core, neuroretinal rim progression, conversion from ocular hypertension to glaucoma, any open angle glaucoma, normal tension glaucoma, high tension primary open angle glaucoma, exfoliative glaucoma, pigmentary glaucoma, primary angle closure glaucoma, primary open angle glaucoma, and primary angle closure glaucoma should not be combined in the primary analysis because their anatomy, mechanisms, and clinical course differ.
• 12:55
  SPEAKER_01: Eye level data.
• 12:56
  SPEAKER_01: If a study reports separate outcomes for both eyes, participants must not be counted as two independent individuals.
• 13:03
  SPEAKER_01: Extract whether the analysis accounted for correlation between eyes.
• 13:07
  SPEAKER_01: If not, use a participant level estimate where available or perform a sensitivity analysis using the study's specified eye selection method.
• 13:15
  SPEAKER_01: Search strategy databases.
• 13:17
  SPEAKER_01: Search from database inception to the date of the final search in Medline through PubMed, MBase Web of Science, Core Collection, Scopus, Cinehal, Cochrane Library, PsychInfo, Global Health.
• 13:29
  SPEAKER_01: Also search Clinicaltrials.gov, World Health Organization, International Clinical Trials Registry Platform, ProQuest Dissertations and Theses, Conference Proceedings from Major Ophthalmology and Nutrition Meetings, Reference Lists of all included studies, cited by records of key studies and reviews, search concepts.
• 13:49
  SPEAKER_01: The search should combine three broad concept groups: sugar and carbohydrate terms, sugar-added sugar-free sugar, total sugar, sucrose, glucose, fructose, sweetened beverage, soft drink, soda drink, sports drink, energy drink, processed sugar dessert, confectionary carbohydrate glycemic load, glycemic index, dietary intake, food frequency, questionnaire glaucoma terms, glaucoma, primary open angle glaucoma, open angle glaucoma, normal tension glaucoma, high tension glaucoma, exfoliative glaucoma, pigmentary glaucoma, primary angle closure glaucoma, ocular hypertension, intraocular pressure, retinal, nerve fiber layer, optic nerve visual field, glaucoma progressions, study design terms.
• 14:27
  SPEAKER_01: Search filters should be avoided initially because they may exclude irrelevant cohort or case control studies.
• 14:33
  SPEAKER_01: Design should be determined during screening.
• 14:36
  SPEAKER_01: No language restrictions should be applied.
• 14:38
  SPEAKER_01: Non-English articles should be translated when feasible.
• 14:41
  SPEAKER_01: The protocol should be registered in an appropriate registry before screening begins and reported according to the preferred reporting items for systematic review and meta-analysis protocols.
• 14:53
  SPEAKER_01: The completed review should follow the preferred reporting items for systematic reviews and meta-analyses and the meta-analysis of observational studies in epidemiology reporting guidance.
• 15:05
  SPEAKER_01: Study selection and data extraction.
• 15:07
  SPEAKER_01: Screening Two reviewers should independently screen titles and abstracts, review potentially eligible full texts, record exclusion reasons, resolve disagreements by discussion or a third reviewer.
• 15:20
  SPEAKER_01: The selection process should be presented in a study flow diagram.
• 15:24
  SPEAKER_01: Data extraction.
• 15:25
  SPEAKER_02: Extract the following information.
• 18:27
  SPEAKER_01: Sugar as a percentage of energy may represent dietary composition rather than absolute sugar intake.
• 18:34
  SPEAKER_01: For sugar-sweetened beverages, energy adjustment may approximate replacing other calories with beverage calories rather than simply adding a beverage to the diet.
• 18:44
  SPEAKER_01: Therefore, do not require adjustment for diabetes when the aim is to capture all pathways from dietary sugar to glaucoma.
• 19:16
  SPEAKER_01: Metabolic adjusted model include baseline diabetes status and when available, diabetes duration, treatment, and glycemic control.
• 19:24
  SPEAKER_01: Stratified model, analyze participants with and without diabetes separately.
• 19:29
  SPEAKER_01: Diabetes should not be treated as a simple nuisance covariate.
• 19:32
  SPEAKER_01: Previous meta-analyses have found an association between diabetes and glaucoma risk, although individual studies have reported conflicting results.
• 19:41
  SPEAKER_01: Body mass index.
• 19:43
  SPEAKER_01: For the total association analysis, body mass index should not automatically be required because it may lie on the pathway between high-energy dietary intake and glaucoma-related metabolic changes.
• 19:55
  SPEAKER_01: For the metabolically adjusted analysis, prefer continuous body mass index over broad categories.
• 20:01
  SPEAKER_01: Use nonlinear modeling when individual participant data are available.
• 20:06
  SPEAKER_01: Record whether body weight was measured or self-reported.
• 20:09
  SPEAKER_01: Distinguish baseline body mass index from weight change during follow-up.
• 20:13
  SPEAKER_01: Evidence linking body mass index with intraocular pressure and glaucoma risk is heterogeneous.
• 20:19
  SPEAKER_01: So body mass index should be treated as an important but potentially time-dependent covariate rather than an automatic confounder.
• 20:27
  SPEAKER_01: Blood pressure extract, systolic blood pressure, diastolic blood pressure, hypertension diagnosis, antihypertensive treatment, pulse pressure if reported.
• 20:37
  SPEAKER_01: Blood pressure measurement method.
• 20:40
  SPEAKER_01: Use blood pressure in the metabolically adjusted analysis.
• 20:43
  SPEAKER_01: Do not adjust for blood pressure in the total association model if it is measured after the dietary exposure or if the study clearly treats it as a possible intermediate pathway.
• 20:55
  SPEAKER_01: Blood pressure may also relate to intraocular pressure and optic nerve perfusion, so studies should be grouped according to whether they adjusted for hypertension, measured blood pressure continuously, or did not adjust for it.
• 21:08
  SPEAKER_01: Adjustment set meta-analysis.
• 21:11
  SPEAKER_01: If enough studies are available, conduct separate pooled analyses for minimal demographic adjustment, core lifestyle, and ocular risk factor adjustment, adjustment including total energy, adjustment including diabetes, body mass index, and blood pressure, models adjusting for possible mediators such as baseline intraocular pressure.
• 21:31
  SPEAKER_01: These models should not be combined indiscriminately.
• 21:35
  SPEAKER_01: Differences in adjustment sets may explain part of the observed heterogeneity.
• 21:39
  SPEAKER_01: Addressing misclassification from food frequency questionnaires.
• 21:44
  SPEAKER_01: Dietary misclassification is likely to be one of the largest limitations in this literature.
• 21:50
  SPEAKER_01: Food frequency questionnaires may miss brand specific sugar content, changes in recipes over time, portion size differences, sugar added during cooking.
• 22:00
  SPEAKER_01: Sugary beverages consumed outside the home, seasonal or episodic intake, under reporting of sweets and high-calorie foods.
• 22:07
  SPEAKER_01: A 2024 scoping review found that food frequency questionnaires were the dominant dietary assessment method in glaucoma nutrition research.
• 22:16
  SPEAKER_01: Planned methods.
• 22:17
  SPEAKER_01: Extract validation information.
• 22:19
  SPEAKER_01: For each study, record whether the questionnaire was validated against, multiple 24-hour dietary recalls, food diaries, weighed food records, dietary biomarkers, a second dietary questionnaire.
• 22:32
  SPEAKER_01: Validated and non-validated instruments should be analyzed separately.
• 22:35
  SPEAKER_01: Prefer repeated dietary measurements.
• 22:38
  SPEAKER_01: When repeated data are available, use cumulative average intake, update exposure over time.
• 22:44
  SPEAKER_01: Average multiple measurements.
• 22:47
  SPEAKER_01: Analyze changes in intake separately from usual intake.
• 22:50
  SPEAKER_01: A single baseline questionnaire should be treated as a higher risk exposure measurement, particularly for studies with long follow-up.
• 22:56
  SPEAKER_01: Use lagged analyses.
• 22:59
  SPEAKER_01: Where data allow, conduct analyses in which diet is lagged by at least two years.
• 23:04
  SPEAKER_01: Longer lags, such as five years, should be examined when possible.
• 23:08
  SPEAKER_01: This can reduce reverse causation, such as dietary changes after early visual impairment, diabetes diagnosis, or a glaucoma-related medical visit.
• 23:17
  SPEAKER_01: Use regression calibration.
• 23:20
  SPEAKER_01: If an internal validation sample is available, use regression calibration to estimate usual intake from the food frequency questionnaire using repeated recalls or food records as reference measures.
• 23:31
  SPEAKER_01: Measurement error research supports using validation data and regression calibration when available because dietary measurement error can attenuate or distort diet disease associations.
• 23:43
  SPEAKER_01: Use probabilistic sensitivity analysis.
• 23:46
  SPEAKER_01: If validation data are unavailable, conduct quantitative bias analyses using plausible values for sensitivity and specificity of high versus low sugar classification, correlation between reported and usual intake, differential recall error in case control studies, underreporting of sugar-sweetened beverages.
• 24:06
  SPEAKER_01: The analysis should not assume that food frequency questionnaire error is always non-differential.
• 24:12
  SPEAKER_01: In case control studies, participants with glaucoma may remember or report their past dietary habits differently from controls.
• 24:19
  SPEAKER_01: Avoid unsupported conversions.
• 24:21
  SPEAKER_01: Do not convert servings into grams, sweets into added sugar, beverage frequency into milliliters.
• 24:28
  SPEAKER_01: Total carbohydrate into sugar unless the original study provides a valid serving size or composition estimate.
• 24:35
  SPEAKER_01: If absolute dose conversion is not possible, use a study-specific standardized contrast or a per standard deviation analysis rather than creating a falsely precise common dose.
• 24:46
  SPEAKER_01: Use dietary biomarkers cautiously.
• 24:48
  SPEAKER_01: Biomarkers may help assess short-term sugar intake, but may not represent habitual long-term intake.
• 24:55
  SPEAKER_01: Studies using biomarkers should therefore be analyzed separately from questionnaire-based studies unless the measurement period and biological meaning are comparable.
• 25:03
  SPEAKER_01: Statistical synthesis plan.
• 25:06
  SPEAKER_01: Effect measures.
• 25:07
  SPEAKER_01: Analyze study designs separately in the primary analysis.
• 25:10
  SPEAKER_01: Prospective cohorts, hazard ratios, risk ratios, or incidence rate ratios, case control studies, odds ratios, intraocular pressure, mean differences, retinal nerve fiber layer, mean annualized change, progression events, hazard ratios, or risk ratios.
• 25:27
  SPEAKER_01: Do not automatically pool odds ratios with hazard ratios.
• 25:31
  SPEAKER_01: An exploratory combined analysis may be considered only when the case control study uses risk-set sampling or the outcome is sufficiently uncommon and the estimates are clinically comparable.
• 25:42
  SPEAKER_01: Primary contrasts.
• 25:44
  SPEAKER_01: The main categorical comparison should be highest versus lowest exposure category.
• 25:49
  SPEAKER_01: The primary continuous contrasts should be per 10 grams per day of total added or free sugar, per 250 milliliters per day of sugar sweetened beverages, per 10 glycemic load units per day, per 10 glycemic index units, if data permit, per 10 grams per day of total carbohydrate.
• 26:07
  SPEAKER_01: These units should be changed only when they are not supported by the available data.
• 26:11
  SPEAKER_01: Random effects meta-analysis.
• 26:13
  SPEAKER_01: Use a random effects model because studies will differ in population, dietary instruments, glaucoma ascertainment, follow-up time adjustment sets, exposure definitions, treatment patterns, geographic setting.
• 26:27
  SPEAKER_01: Report.
• 26:28
  SPEAKER_01: Pooled effect estimate 95% confidence interval between study variants, statistical heterogeneity, 95% prediction interval.
• 26:38
  SPEAKER_01: A prediction interval is particularly useful because it shows the range in which the association from a future comparable study might fall.
• 26:46
  SPEAKER_01: Cochrane Guidance recommends careful interpretation of heterogeneity and use of prediction intervals when random effects models are used.
• 26:54
  SPEAKER_01: Continuous intraocular pressure outcomes.
• 26:57
  SPEAKER_01: For intraocular pressure, use mean differences when measurements are in millimeters of mercury.
• 27:02
  SPEAKER_01: Use annualized change when follow-up intervals differ.
• 27:06
  SPEAKER_01: Adjust for baseline intraocular pressure where the primary study did so.
• 27:10
  SPEAKER_01: Separate untreated participants from participants receiving pressure-lowering medication when possible.
• 27:16
  SPEAKER_01: Account for treatment changes during follow-up.
• 27:19
  SPEAKER_01: Retinal nerve fiber layer outcomes.
• 27:22
  SPEAKER_01: For retinal nerve fiber layer outcomes, prefer annualized thinning rates.
• 27:27
  SPEAKER_01: Separate global retinal nerve fiber layer thickness from sectoral measures.
• 27:31
  SPEAKER_01: Record optical coherence tomography device and software version.
• 27:36
  SPEAKER_01: Exclude scans with inadequate quality where the study provides that information.
• 27:40
  SPEAKER_01: Analyze eyes using participant level or clustered methods.
• 27:44
  SPEAKER_01: If studies use incompatible measurement scales, use a standardized mean difference only when the underlying construct is genuinely comparable.
• 27:53
  SPEAKER_01: Dose response modeling.
• 27:55
  SPEAKER_01: Linear dose response analysis.
• 27:57
  SPEAKER_01: For studies reporting at least three exposure categories with dose levels, use a generalized least squares trend method based on the Greenland and Longnecker approach.
• 28:07
  SPEAKER_01: This method accounts for the fact that estimates from several exposure categories within one study are correlated.
• 28:13
  SPEAKER_01: For each study, extract the effect estimate and confidence interval for each exposure category.
• 28:19
  SPEAKER_01: Assign the reported median or mean exposure to each category.
• 28:23
  SPEAKER_01: Use category midpoints when medians are unavailable.
• 28:26
  SPEAKER_01: Handle open-ended categories using prespecified midpoint rules.
• 28:30
  SPEAKER_01: Estimate the linear trend per unit increase.
• 28:33
  SPEAKER_01: Pool the study's specific slopes using a random effects model.
• 28:37
  SPEAKER_01: Nonlinear dose response analysis.
• 28:39
  SPEAKER_01: If at least five studies provide usable dose categories, fit restricted cubic spline models to test for linear relationships, threshold effects, J-shaped or U-shaped relationships, steeper risk at high intake levels, possible differences between low and moderate intake.
• 28:57
  SPEAKER_01: The dose response meta-analysis should use a pre-specified knot strategy, such as the 10th, 50th, and 90th percentiles of the exposure distribution.
• 29:06
  SPEAKER_01: Alternative knot placements should be examined in sensitivity analyses.
• 29:10
  SPEAKER_01: Restricted cubic spline dose response methods are established for modeling nonlinear associations in meta-analysis.
• 29:17
  SPEAKER_01: Comparator and substitution analyses.
• 29:19
  SPEAKER_01: Nutrition research is inherently comparative.
• 29:22
  SPEAKER_01: Reducing sugar may mean replacing it with water, unsweetened beverages, milk, whole fruit, refined starch, whole grains, fat, protein, alcohol.
• 29:32
  SPEAKER_01: If a study reports substitution estimates, these should be analyzed separately rather than pooled with ordinary high versus low comparisons.
• 29:39
  SPEAKER_01: A finding that sugar-sweetened beverages are associated with glaucoma risk may therefore reflect the effect of the beverage itself, the calories it contributes, or what it replaces.
• 29:50
  SPEAKER_01: Prespecified subgroup analyses, primary subgroup analyses, baseline diabetes status, conduct separate analyses for participants with diabetes, participants without diabetes, mixed populations, studies adjusting for diabetes, studies not adjusting for diabetes, test for interaction when possible.
• 30:10
  SPEAKER_01: The main interpretation should focus on the size and precision of the subgroup estimates rather than on whether one subgroup reaches statistical significance and another does not.
• 30:20
  SPEAKER_01: Glaucoma subtype.
• 30:22
  SPEAKER_01: Analyze separately, primary open angle glaucoma, normal tension glaucoma, high tension primary open angle glaucoma, exfoliative glaucoma, pigmentary glaucoma, primary angle closure glaucoma, unspecified glaucoma.
• 30:36
  SPEAKER_01: Do not combine primary open angle and primary angle closure glaucoma in the main analysis.
• 30:41
  SPEAKER_01: Additional subgroup analyses, where sufficient studies exist, examine incidence versus progression, validated versus non-validated dietary instruments, repeated versus baseline-only dietary measurement, clinically confirmed versus self-reported glaucoma, optical coherence, tomography versus visual field outcomes, short versus long follow-up, older versus younger populations, sex race, and ethnicity, geographic region, baseline intraocular pressure, use versus non-use of pressure lowering treatment, high versus low risk of bias, energy adjusted versus non-energy adjusted estimates, meta-regression should generally be reserved for analyses with enough independent studies to support stable estimates.
• 31:20
  SPEAKER_01: Subgroup findings should be considered exploratory when the number of studies is small.
• 31:25
  SPEAKER_01: Heterogeneity, small study effects, and publication bias.
• 31:29
  SPEAKER_01: Assess heterogeneity using clinical comparison of study populations, comparison of exposure definitions, comparison of outcome ascertainment between study variants, the inconsistency statistic, prediction intervals, leave-one-out analyses, influence diagnostics.
• 31:46
  SPEAKER_01: If at least 10 independent studies contribute to a pooled analysis, assess small study effects using funnel plots, eggger type regression, selection model or weight function methods when appropriate.
• 31:58
  SPEAKER_01: Trim and fill procedures should be considered exploratory rather than corrective because they rely on strong assumptions.
• 32:05
  SPEAKER_01: Gray literature searching, trial registry searches, and contact with investigators are important because small or null studies may be less likely to appear in journals.
• 32:16
  SPEAKER_01: Risk of bias assessment, prospective and longitudinal studies.
• 32:21
  SPEAKER_01: Use the risk of bias in non-randomized studies of exposures framework for follow-up studies.
• 32:26
  SPEAKER_01: It evaluates domains including confounding exposure measurement, selection of participants, post-exposure interventions, missing data, outcome measurement, selective reporting.
• 32:38
  SPEAKER_01: The tool is designed for observational studies of exposures and emphasizes judging the specific result being synthesized rather than assigning a simple overall numeric score.
• 32:49
  SPEAKER_01: Case control studies.
• 32:50
  SPEAKER_01: Use the Joanna Briggs Institute case control checklist, which addresses whether cases and controls came from comparable populations, whether matching was appropriate, whether exposure was measured similarly, whether outcome definitions were valid, whether confounding was addressed, whether the exposure window was long enough, whether the statistical analysis was appropriate.
• 33:12
  SPEAKER_01: Risk of bias judgments should be completed independently by two reviewers.
• 33:18
  SPEAKER_01: The review should not simply exclude every study judged to have some concern.
• 33:22
  SPEAKER_01: Instead, risk of bias should inform sensitivity analyses, evidence grading, interpretation of contradictory findings, weight given to self-reported or poorly validated outcomes, certainty of the evidence, rate certainty separately for sugar intake and incident primary open angle glaucoma, sugar-sweetened beverages and incident glaucoma, glycemic load and incident glaucoma, sugar intake and intraocular pressure, sugar intake and retinal nerve fiber layer loss, sugar intake and glaucoma progression.
• 33:55
  SPEAKER_01: Use the grading of recommendations assessment, development, and evaluation approach for risk of bias, inconsistency, indirectness, imprecision, publication bias.
• 34:06
  SPEAKER_01: A dose response gradient should increase certainty only when it is credible, consistent across studies, not easily explained by confounding, and supported by sufficiently direct evidence.
• 34:17
  SPEAKER_01: Current guidance specifically cautions against automatically upgrading evidence simply because a dose response pattern is statistically significant.
• 34:25
  SPEAKER_01: Planned sensitivity analyses.
• 34:27
  SPEAKER_01: Repeat the primary analyses after, removing studies at thigh risk of bias, removing self-reported glaucoma outcomes, removing nonvalidated food frequency questionnaires, removing case control studies, removing studies with baseline-only dietary measurement, applying at Who-year or 5-year exposure lag, excluding studies with substantial cohort to overlap, using only physician-confirmed or objectively examined glaucoma, using only studies adjusted for total energy intake, using only studies adjusted for diabetes body mass index and blood pressure, excluding studies that adjusted for likely mediators, using alternative midpoint assumptions for open-ended exposure categories, using alternative baseline to follow-up correlations for intraocular pressure change, comparing fixed effect and random effects models, excluding influential studies one at a time, comparing absolute sugar units with within study standardized units, a null pool dissociation should not be interpreted as proof of no biological relationship if dietary measurement error is substantial.
• 35:28
  SPEAKER_01: Anticipated limitations.
• 35:29
  SPEAKER_01: The review will likely face several limitations.
• 35:32
  SPEAKER_01: Few studies may measure added sugar directly.
• 35:35
  SPEAKER_01: Total carbohydrate may be reported more often than added sugar.
• 35:39
  SPEAKER_01: Food frequency questionnaires may misclassify usual intake.
• 35:42
  SPEAKER_01: Sugar intake may be a marker of the overall dietary pattern or ultra-processed food intake.
• 35:48
  SPEAKER_01: Diabetes, body mass index, and blood pressure may be both confounders and mediators.
• 35:54
  SPEAKER_01: Case control studies may be affected by recall bias.
• 35:57
  SPEAKER_01: Self-reported glaucoma may miss undiagnosed disease or include incorrect diagnoses.
• 36:03
  SPEAKER_01: Glaucoma progression may be influenced strongly by treatment, adherence, baseline severity, and follow-up frequency.
• 36:10
  SPEAKER_01: Sugar-sweetened beverage associations may depend on the replacement food or beverage.
• 36:15
  SPEAKER_01: Different glaucoma subtypes may have different biological relationships with metabolic exposures.
• 36:21
  SPEAKER_01: The 2024 Perspective Ultra Processed Food Study illustrates why these issues matter.
• 36:27
  SPEAKER_01: Its models adjusted for total energy intake, body mass index, diabetes, hypertension, cardiovascular disease, smoking, physical activity, and several dietary variables, but the authors still acknowledge the possibility of residual confounding and limitations in food frequency questionnaire classification.
• 36:46
  SPEAKER_01: Conclusion.
• 36:47
  SPEAKER_01: A rigorous review of dietary sugar and glaucoma should treat this as a family of related but distinct exposure questions, not as one simple comparison between high sugar and low sugar.
• 36:57
  SPEAKER_01: The strongest protocol would separate total sugar, added sugar, free sugar, sugary beverages, total carbohydrate, glycemic load, and glycemic index.
• 37:07
  SPEAKER_01: Analyze perspective and case control evidence separately.
• 37:11
  SPEAKER_01: Treat primary open-angle glaucoma incidence, intraocular pressure change, and retinal nerve fiber layer loss as distinct outcomes.
• 37:19
  SPEAKER_01: Pre-specify total association and metabolically adjusted models.
• 37:24
  SPEAKER_01: Report the effects of adjustment for diabetes, body mass index, blood pressure, and total energy intake.
• 37:30
  SPEAKER_01: Address food frequency questionnaire error using repeated measures, validation studies, regression calibration, lagged exposure, and quantitative bias analysis.
• 37:40
  SPEAKER_01: Use linear and nonlinear dose response models.
• 37:43
  SPEAKER_01: Conduct subgroup analyses by baseline diabetes status and glaucoma subtype.
• 37:49
  SPEAKER_01: Grade certainty without overstating observational evidence.
• 37:52
  SPEAKER_01: The final review should determine whether a consistent association exists, whether it is independent of metabolic factors, and whether it is strong enough to justify future prospective studies using repeated dietary assessments.
• 38:05
  SPEAKER_01: Objective glaucoma examinations, optical coherence tomography, visual field testing, and validated biomarkers of sugar exposure.
• 38:14
  SPEAKER_01: All links to sources are available in the text version of this article.
• 38:18
  SPEAKER_01: You can find the full article at Visual Fieldtest.com.
• 38:22
  SPEAKER_01: Thanks for listening.
• 38:23
  SPEAKER_01: To check your visual field, click the link at the bottom of this article or visit visualfieldtest.com.