Eyes

What research found here. What was shown to help comes first, then the risks reported, most common first.

Eyes

Strong several reviews agree · Moderate one review or consistent trials · Limited one or a few small studies · Untested no study in people, listed with what it rests on

Shown to help13

Hemoglobin A1c (HbA1c)Strong

Eye, kidney and nerve complications of diabetes

Shown in trials: intensive glucose control that lowered HbA1c about 0.9 points cut kidney events 20% and eye events 13% in type 2 diabetes.

The numbers and sources

Effect: In an individual-participant meta-analysis of 4 trials in type 2 diabetes, more intensive control left HbA1c 0.90 percentage points lower (95% CI 0.58 to 1.22) and reduced kidney events (HR 0.80, 95% CI 0.72 to 0.88) and eye events (HR 0.87, 95% CI 0.76 to 1.00), with no reduction in nerve events (HR 0.98, 95% CI 0.87 to 1.09) [11]. In type 1 diabetes, intensive therapy reduced the development of retinopathy by 76% (95% CI 62 to 85) and clinical neuropathy by 60% (95% CI 38 to 74) [12]. In an observational analysis of people with type 2 diabetes, each 1% lower updated mean HbA1c went with 37% fewer microvascular complications (95% CI 33 to 41), with no threshold [13]. In pooled data from nine studies, diabetes-specific retinopathy was rare below 6.0% and its threshold fell at 6.3 to 6.7% [14].
Population: 27,049 adults with type 2 diabetes in 4 trials, median follow-up 5.0 years [11]. 1,441 people with type 1 diabetes followed a mean 6.5 years [12]. 3,642 people with type 2 diabetes [13]. 28,010 people with HbA1c and retinal photographs [14].
Sources: [11], [12], [13], [14]
Caveats: The trials tested treatment strategies, not HbA1c itself. Intensive therapy raised severe hypoglycemia two to threefold in the type 1 diabetes trial [12].

SunlightStrongStudied in children

Myopia in children

Shown to cut new myopia in children by about 16% to 46% with more time outdoors, in three meta-analyses including several school trials.

Compared with control groups in trials and children spending less time outdoors in cohorts [5]; control groups in trials [6][7]; children in control schools who kept their usual activity [8]; children in control schools [9]

The numbers and sources

Effect: Outdoor time cut incident myopia in trials (RR 0.536, 95% CI 0.338 to 0.850) and cohorts (RR 0.574), and reduced myopic shift by 0.30 diopters over 3 years. In children already myopic, time outdoors was not related to progression [5]. Across 5 randomized trials, new myopia fell (RR 0.76, 95% CI 0.67 to 0.87), with no heterogeneity [6]. Across 7 randomized trials, myopia incidence fell (RR 0.84, 95% CI 0.78 to 0.91) and axial length grew 0.09 mm less [7]. An added 40-minute outdoor class each school day cut 3-year myopia incidence from 39.5% to 30.4% [8]. A recess-outdoors program slowed myopic shift in both myopic and non-myopic children over 1 year [9].
Population: 25 studies, 23 in children [5]. 3,014 children aged 6 to 12 in 5 trials, 9 to 36 months [6]. 9,437 children and adolescents in 7 trials [7]. 1,903 children aged about 6.6 in Guangzhou, China [8]. 693 first-graders in Taiwan [9].
Sources: [5], [6], [7], [8], [9]
Caveats: Most trials were in East Asia, where myopia is common. Outdoor time is daylight plus distance viewing, so sunlight is not the only possible cause. One trial found an effect at light levels as low as shade under trees [9]. The reviews disagree on children already myopic [5][9].

Ezetimibe and fenofibrateModerate

Fenofibrate: diabetic eye disease

Shown to slow diabetic retinopathy in people with early eye changes, 22.7% against 29.2% over 4 years in the largest eye trial, without changing vision.

Compared with placebo [12][13]; placebo or observation [14]

The numbers and sources

Effect: Referable retinopathy or maculopathy, or treatment for it, occurred in 22.7% on fenofibrate 145 mg and 29.2% on placebo (HR 0.73, 95% CI 0.58 to 0.91), with no effect on visual function, quality of life or acuity [12]. Pooling three trials, retinopathy progression fell (RR 0.77, 95% CI 0.68 to 0.88), 45 fewer per 1,000, and no trial found a benefit for vision [13]. An earlier Cochrane review found little to no difference in a mixed group (RR 0.86, 95% CI 0.60 to 1.25) but slower progression in people with retinopathy at the start (RR 0.21, 95% CI 0.06 to 0.71) [14]. In the main heart trial, laser treatment was needed by 3.6% on fenofibrate and 5.2% on placebo [8].
Population: 1,151 adults with diabetes and early retinal changes in Scotland, median 4 years [12]; 3,756 people in 3 trials [13]; 15,313 people with type 2 diabetes in 2 trials [14]; 9,795 adults [8].
Sources: [12], [13], [14], [8]
Caveats: Mostly type 2 diabetes. The earlier Cochrane review predates the largest eye trial [14].

Fasting glucoseModerate

Diabetic eye disease

Linked to diabetic eye disease from about 6.4 to 6.8 mmol/L (about 115 to 123 mg/dL) upward, in pooled data on 41,411 people.

The numbers and sources

Effect: In pooled data from nine studies, the prevalence of diabetes-specific retinopathy was low at fasting glucose below 6.0 mmol/L and rose above that level. The threshold fell at 6.4 to 6.8 mmol/L by one method and 6.6 mmol/L by another. The authors suggested the diagnostic level for fasting glucose could be lowered from 7.0 to 6.5 mmol/L [15].
Population: 44,623 people aged 20 to 79 with gradable retinal photographs in nine studies from five countries, 41,411 of them with fasting glucose [15].
Sources: [15]
Caveats: Cross-sectional. WHO states that its diabetes criteria mark a group with more premature death and more microvascular and cardiovascular complications, while noting the limits of the data behind them [5].

InsulinModerate

Eye, kidney and nerve complications in type 1 diabetes

Shown to cut new eye disease by about 76%, kidney protein leakage by 39% to 54% and nerve damage by 60% with intensive insulin treatment in type 1 diabetes.

The numbers and sources

Effect: In people with type 1 diabetes and no eye disease, intensive insulin treatment lowered the risk of developing retinopathy by 76% (95% CI 62 to 85) against conventional treatment. In people with mild retinopathy it slowed progression by 54% (95% CI 39 to 66). Across both groups it lowered microalbuminuria by 39% (95% CI 21 to 52), albuminuria by 54% (95% CI 19 to 74) and clinical neuropathy by 60% (95% CI 38 to 74) [1]. A Cochrane review of 12 trials found retinopathy developed in 6.2% on intensive control against 23.2% on conventional control (RR 0.27, 95% CI 0.18 to 0.42), nephropathy in 16.3% against 28.4% (RR 0.56, 95% CI 0.46 to 0.68) and neuropathy in 4.9% against 13.9% (RR 0.35, 95% CI 0.23 to 0.53) [2].
Population: 1,441 people with type 1 diabetes, assigned to an insulin pump or three or more daily injections, or to one or two daily injections [1]; 2,230 patients in 12 trials followed for 1 to 6.5 years [2].
Sources: [1], [2]
Caveats: The trials compared more intensive with less intensive insulin treatment. Neither arm went without insulin. The largest trial dominates the review [2]. Once complications were present, the effect was weaker for eye disease and uncertain for kidney disease [2].

Red light therapyModerateStudied in children

Childhood nearsightedness

Shown to slow myopia progression in children by 0.68 diopters per 6 months with repeated low-level red light, at low certainty and with very high variation.

The numbers and sources

Effect: Across 8 studies pooled from 13 (1,857 children and adolescents), repeated low-level red light slowed refractive change by 0.68 D per 6 months (95% CI 0.38 to 0.97, I² 97.7%) and axial lengthening by 0.35 mm per 6 months (95% CI -0.51 to -0.19) compared with control [27].
Population: 1,857 children and adolescents, in 8 RCTs, 3 non-randomized trials and 2 cohort studies [27].
Sources: [27]
Caveats: The authors rate the evidence low certainty and call for 2-year trials [27]. This is a desktop laser instrument viewed with the eye, a different thing from a skin panel. See the retinal injury block under Risks.

Vitamin A (retinol) levelModerateStudied in children

Death in young children where deficiency is common

Shown in trials to cut deaths 12% in children aged 6 months to 5 years given vitamin A in countries where deficiency is common.

The numbers and sources

Effect: In a Cochrane review, vitamin A supplementation reduced deaths from any cause (RR 0.88, 95% CI 0.83 to 0.93, 19 trials, 1,202,382 children, high-certainty evidence) and deaths from diarrhea (RR 0.88, 0.79 to 0.98). It reduced night blindness (RR 0.32, 0.21 to 0.50), Bitot's spots (RR 0.42, 0.33 to 0.53) and vitamin A deficiency itself (RR 0.71, 0.65 to 0.78, 4 studies, 2,262 children). Deaths from measles and from respiratory disease did not differ [8].
Population: 47 studies with about 1,223,856 children in 19 countries, 30 of the studies in Asia. The mean age was about 33 months [8].
Sources: [8]
Caveats: The trials gave vitamin A to children in the community whatever their own serum retinol, so they show what supplementation did in populations with widespread deficiency. They did not test treating a child according to a measured level, and they do not describe adults or populations where deficiency is rare [8].

AstaxanthinLimitedStudied in children

Eye strain

One trial found 4 mg a day lowered eye strain scores 20% more than placebo in children with heavy screen time. Several authors worked for the maker.

The numbers and sources

Effect: In children with daily screen time of 4 hours or more, 4 mg a day for 84 days lowered computer vision syndrome questionnaire scores by 4.00 points against 1.72 on placebo, a 20% between-group difference. Visual fatigue scores were 27% lower than placebo. Visual acuity, near point of accommodation and the Schirmer tear test did not differ between groups [9].
Population: 64 children aged 10 to 14 with mild to moderate eye strain symptoms [9].
Sources: [9]
Caveats: One trial. Several authors were employed by Fuji Chemical Industries, which makes AstaReal astaxanthin, or its subsidiaries, and others by a contract research organization Fuji engaged [9].

C-peptideLimited

Eye disease and low blood sugar in type 1 diabetes

Limited evidence: in one type 1 diabetes trial cohort, higher stimulated C-peptide went with less retinopathy and hypoglycemia, and fasting C-peptide showed weaker links.

The numbers and sources

Effect: In the intensive treatment group of the Diabetes Control and Complications Trial, a 50% higher stimulated C-peptide at entry, such as 0.15 against 0.10 nmol/L, went with a 25% lower risk of sustained retinopathy, an 8.2% lower risk of hypoglycemia, a lower insulin dose and an HbA1c 0.07 percentage points lower. Outcomes changed in a nearly linear way as C-peptide fell to unmeasurable levels, with no threshold. Fasting C-peptide had weaker associations with outcomes than stimulated C-peptide [9].
Population: Participants in the Diabetes Control and Complications Trial with up to 5 years of type 1 diabetes at entry, followed for 7 years [9].
Sources: [9]
Caveats: The main findings are for C-peptide measured after a stimulus, a different quantity from the fasting value on this page. They are unadjusted, come from one trial cohort, and concern people with type 1 diabetes only. The abstract gives no participant count [9].

FructosamineLimited

Eye and kidney complications

One US cohort linked fructosamine above the 95th percentile to 50% more new chronic kidney disease, and higher levels to more retinopathy.

The numbers and sources

Effect: Fructosamine and glycated albumin were strongly associated with retinopathy (P below 0.0001 for trend). The adjusted hazard ratio for chronic kidney disease was 1.50 (95% CI 1.22 to 1.85) for fructosamine above the 95th percentile and 1.48 (1.20 to 1.83) for glycated albumin, compared with people without diabetes below the 75th percentile. Prediction of chronic kidney disease was nearly as strong as with HbA1c, with C statistics of 0.717, 0.717 and 0.726 [5].
Population: 11,348 adults without diabetes and 958 with diagnosed diabetes, followed for two decades [5].
Sources: [5]
Caveats: Observational, from one cohort. A laboratory guideline states that few published studies have convincingly shown a relationship between glycated protein levels and the chronic complications of diabetes [2].

Hyaluronic acid (oral)Limited

Dry eyes

One small trial found oral hyaluronic acid added to hyaluronic acid eye drops improved corneal staining and symptom scores more than the drops alone over 3 months.

The numbers and sources

Effect: In 54 adults with dry eye using hyaluronic acid eye drops, those also given 240 mg of oral hyaluronic acid a day had better corneal staining at 1 month (between-group p < 0.001) and better corneal staining and OSDI symptom scores at 3 months (p < 0.001 and p = 0.046). Symptom scores fell from 61.8 to 42.3 on the combination. Tear break-up time did not differ between groups [10].
Population: 54 adults with dry eye disease in Korea, 24 given oral hyaluronic acid and 30 drops alone [10].
Sources: [10]
Caveats: A pilot trial with no placebo capsule, so the control group knew it was not taking a supplement. Everyone also used hyaluronic acid drops [10]. It was publicly funded and the authors declared no conflicts [10].

SeleniumLimited

Mild Graves' eye disease

One trial found better quality of life and slower eye disease progression over 6 months in people with mild Graves' eye disease.

The numbers and sources

Effect: After 6 months, 100 µg twice a day improved quality of life, reduced eye involvement and slowed progression against placebo. The Clinical Activity Score fell more on selenium. Results held at 12 months [8].
Population: 159 people with mild Graves' orbitopathy [8].
Sources: [8]
Caveats: One trial. The abstract gives P values without effect sizes. Funded by the University of Pisa and the Italian ministry for education and research [8].

1 more

SulfonylureasLimited

Eye, kidney and other small-vessel complications

One trial found 25% fewer small-vessel complications over 10 years with intensive sulfonylurea or insulin treatment than with diet in newly diagnosed type 2 diabetes.

The numbers and sources

Effect: Against diet, intensive treatment with a sulfonylurea or insulin lowered any diabetes-related endpoint by 12% (95% CI 1 to 21) and microvascular endpoints by 25% (95% CI 7 to 40), including the need for retinal laser treatment. Results did not differ between chlorpropamide, glibenclamide and insulin. Diabetes-related death (10% lower, 95% CI -11 to 27) and death from any cause (6% lower, 95% CI -10 to 20) did not differ significantly [4].
Population: 3,867 newly diagnosed patients with type 2 diabetes, median age 54, followed over 10 years [4].
Sources: [4]
Caveats: The trial pooled sulfonylureas with insulin in its intensive group, and used chlorpropamide, glibenclamide and glipizide. It did not find fewer large-vessel events [4].
Why limited: This rests on one trial, of 3,867 people. Strength counts independent studies that agree, so one trial is limited at any size.

Reported risks32

LamotrigineModerateVery common

Dizziness, double vision, unsteadiness, nausea and sleepiness in epilepsy trials

Shown to raise dizziness, double vision, unsteadiness and nausea when added to other seizure drugs, with dizziness in 38% against 13% on placebo.

The numbers and sources

How common: Very common: more than 1 in 10 people.
Effect: In pooled add-on epilepsy trials, dizziness occurred in 38% on lamotrigine and 13% on placebo, headache 29% and 19%, double vision 28% and 7%, unsteadiness (ataxia) 22% and 6%, nausea 19% and 10%, blurred vision 16% and 5%, sleepiness 14% and 7%, and vomiting 9% and 4%. Dizziness, double vision, ataxia, blurred vision, nausea and vomiting rose with dose [4]. In a Cochrane review the risk ratios were 3.34 for ataxia, 1.76 for dizziness, 1.81 for nausea and 3.79 for double vision, with no difference in fatigue (RR 0.82) [7].
Population: 711 adults on lamotrigine and 419 on placebo, added to 1 to 3 other seizure drugs [4]; 14 trials with 1,806 participants [7].
Sources: [4], [7]
Caveats: These trials added lamotrigine to other seizure drugs, which raises these effects. Rates were lower in the bipolar trials, where it was taken alone [4].

TrazodoneModerateVery common

Dry mouth, blurred vision and other common side effects

Shown to raise dry mouth, blurred vision, swelling and fatigue against placebo, with dry mouth in 34% of outpatients against 20% on placebo.

The numbers and sources

How common: Very common: more than 1 in 10 people.
Effect: In outpatient depression trials: dry mouth 34% vs 20% on placebo, blurred vision 15% vs 4%, skin condition or edema 7% vs 1%, nasal congestion 6% vs 3% and diarrhea 5% vs 1%. In inpatient trials: dry mouth 15% vs 8%, fatigue 11% vs 4%, nausea or vomiting 10% vs 1% and incoordination 5% vs 0 [12]. In insomnia trials, decreased appetite was more frequent than in control groups (OR 2.81, 95% CI 1.14 to 6.92) [7].
Population: 157 outpatients and 142 inpatients on trazodone, 158 and 95 on placebo [12]; 466 participants in 11 insomnia RCTs [7].
Sources: [12], [7]
Caveats: Small, older trials at depression doses. Rates at the lower doses used for sleep may differ.

InositolLimitedStudied in childrenVery common

Death in extremely preterm infants

One trial, stopped early, found more deaths in extremely preterm infants given myo-inositol, 18% against 11% on placebo.

The numbers and sources

How common: Very common: more than 1 in 10 people.
Effect: The trial stopped early. Death or severe retinopathy of prematurity occurred in 29% on myo-inositol and 21% on placebo (adjusted RR 1.41, 95% CI 1.08 to 1.83). Death before 55 weeks' postmenstrual age occurred in 18% and 11% (adjusted RR 1.66, 95% CI 1.14 to 2.43) [14].
Population: 638 infants born before 28 weeks' gestation at 18 US neonatal centers [14].
Sources: [14]
Caveats: One trial, stopped early. It used IV and enteral doses in newborns, not adult supplements [14].

InsulinLimitedVery common

Early worsening of eye disease when control tightens quickly

Small trials found retinopathy worsened in 34.7% after one year of intensive control against 14.9% on conventional control, before the long-term benefit appeared.

The numbers and sources

How common: Very common: more than 1 in 10 people.
Effect: In two trials, retinopathy had progressed after one year in 17 of 49 people on intensive glucose control (34.7%) against 7 of 47 on conventional control (14.9%) (RR 2.32, 95% CI 1.16 to 4.63). In trials lasting at least two years, intensive control lowered progression (23.2% against 38.7%) [2]. The labels state that rapid improvement in glucose control has been associated with a temporary change in vision, worsening of diabetic retinopathy and acute painful nerve symptoms [5][19].
Population: 96 participants with type 1 diabetes in two trials [2].
Sources: [2], [5], [19]
Caveats: Rated low quality evidence. The band above comes from 96 people who mostly had eye disease at the start [2].

Oral corticosteroidsLimitedVery common

Weight gain, cataracts and other effects of long-term use

Linked to weight gain in 70% of long-term users in one survey, with cataracts reported by 15% and fractures by 12%.

The numbers and sources

How common: Very common: more than 1 in 10 people.
Effect: Among long-term users surveyed by mail, more than 90% reported at least one adverse event and 55% found at least one very bothersome. Weight gain was reported by 70%, cataracts by 15% and fractures by 12%, and every effect rose with cumulative dose [29]. The label lists posterior subcapsular cataracts, glaucoma, fluid retention, muscle weakness and loss of muscle mass among adverse reactions [20].
Population: 2,446 long-term glucocorticoid users, mean dose 16 mg a day of prednisone equivalent, 38% of those mailed the survey [29].
Sources: [29], [20]
Caveats: Self-reported, with no comparison group, and 62% did not respond.

GabapentinoidsModerateCommon

Blurred vision

Shown to cause blurred vision in 7% on pregabalin against 2% on placebo, which resolved with continued dosing in most cases.

The numbers and sources

How common: Common: between 1 in 100 and 1 in 10 people.
Effect: In controlled trials, blurred vision was reported by 7% on pregabalin and 2% on placebo, and under 1% stopped treatment for vision-related events [7]. In trials of gabapentin for chronic low back pain, visual disturbances were more frequent than on placebo (RR 5.72, 95% CI 1.94 to 16.91) [10].
Population: Adults in controlled pregabalin trials [7]; 3 gabapentin trials in chronic low back pain [10].
Sources: [7], [10]
Caveats: The gabapentin estimate comes from small trials with a wide interval.

DulaglutideLimitedCommon

Diabetic eye disease complications

One large trial found diabetic retinopathy complications in 1.9% on dulaglutide against 1.5% on placebo, and 8.5% against 6.2% in people with prior retinopathy.

The numbers and sources

How common: Common: between 1 in 100 and 1 in 10 people.
Effect: Over a median 5.4 years, diabetic retinopathy complications occurred in 1.9% on dulaglutide 1.5 mg and 1.5% on placebo. In people with retinopathy at baseline the rates were 8.5% and 6.2%, and without it 1% in each group. The label notes that rapid improvement in glucose control has been linked to temporary worsening of retinopathy [1].
Population: 9,901 adults in REWIND, as reported in the label [1].
Sources: [1]
Caveats: A secondary outcome, reported without a confidence interval.
Why limited: This rests on one trial, of 9,901 people. Strength counts independent studies that agree, so one trial is limited at any size.

Letrozole and clomipheneLimitedCommon

Visual symptoms with clomiphene

The label warns of blurred vision, spots or flashes in 1.5% of trial patients, rising with dose and duration, sometimes prolonged and possibly irreversible.

The numbers and sources

How common: Common: between 1 in 100 and 1 in 10 people.
Effect: Visual symptoms, including blurred vision, lights, floaters, double vision and sensitivity to light, were reported by 1.5% of patients in the clinical trials. The label states these increase with total dose or length of treatment, are usually reversible, and have in some cases been prolonged or possibly irreversible. One patient on 200 mg daily had severe loss of visual acuity by day 10, which returned to normal 3 days after stopping. Postmarketing reports include cataract, optic neuritis, retinal hemorrhage and temporary or prolonged loss of vision [1].
Population: 8,029 patients in label trials and postmarketing reports [1].
Sources: [1]
Caveats: No placebo group, and postmarketing reports cannot give a rate or prove cause [1].

BisphosphonatesLimitedUncommon

Eye inflammation

Case reports and trial counts link bisphosphonates to rare eye inflammation such as uveitis, about 0.2% on zoledronic acid in trials. Cohort studies disagreed.

The numbers and sources

How common: Uncommon: between 1 in 1,000 and 1 in 100 people.
Effect: In the zoledronic acid osteoporosis trials, iritis, uveitis or episcleritis developed in less than 0.1% to 0.2% of treated patients and 0% to less than 0.1% on placebo, and 1.1% against 0% in the prevention trial [13]. A review found mostly case reports, some with recurrence on rechallenge, and three cohort studies with conflicting results [23].
Population: Zoledronic acid trial populations [13]; published case reports and cohort studies [23].
Sources: [13], [23]
Caveats: Trials were not powered for rare events [23].

TopiramateLimitedUncommon

Sudden nearsightedness and angle-closure glaucoma

Linked to angle-closure glaucoma, about 2.6 times the rate overall and 5 times under age 50 in one study, mostly within the first month.

The numbers and sources

How common: Uncommon: between 1 in 1,000 and 1 in 100 people.
Effect: The label reports acute myopia with secondary angle-closure glaucoma, usually within a month of starting, in adults and children, and visual field defects that mostly reversed after stopping [3]. In a US claims study, topiramate use was linked to angle-closure glaucoma (adjusted rate ratio 2.59, 95% CI 1.56 to 4.30), and 5.30 (95% CI 2.54 to 11.04) under age 50 [14]. In Taiwan, glaucoma was diagnosed in 0.36% of new users in the first month against 0.04% of non-users (adjusted HR 7.41), with no significant difference after the first month [15].
Population: Postmarketing reports [3]; a nested case-control study in a cohort of 6,110,723 people [14]; 1,956 new users and 15,648 matched non-users [15].
Sources: [3], [14], [15]
Caveats: Observational studies, which rely on diagnosis codes.

Alpha-blockersStrongHow often is not known

Floppy iris during cataract surgery

Linked to floppy iris syndrome during cataract surgery, most strongly for tamsulosin, at 31 times the odds in an updated meta-analysis. The labels warn of it.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: Tamsulosin carried the largest odds of intraoperative floppy iris syndrome, about 40-fold those of alfuzosin, the next highest. Alfuzosin and terazosin were also linked, and doxazosin reached significance in one analysis [22]. An updated meta-analysis found tamsulosin raised the odds 31-fold (OR 31.06, 95% CI 13.74 to 70.22) [23]. The tamsulosin label states the syndrome was seen in patients on or previously treated with alpha-1 blockers [4].
Population: 17 studies with 17,588 eyes [22]; 38 studies [23].
Sources: [22], [23], [4]
Caveats: Observational studies with wide intervals, and neither abstract gives the share of users affected.

IsotretinoinStrongHow often is not known

Dry lips, dry skin and other skin effects

Shown to double adverse events against control, more than half of them skin dryness. Dry lips, dry skin and dry eyes are among the most frequent effects on the label.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: Adverse events were twice as frequent on isotretinoin (751 events) as on control (388), and more than half were skin effects related to dryness. Stevens-Johnson syndrome, cheilitis, dry skin, acne flare and photophobia led to withdrawal in some [1]. Against antibiotics plus topical agents, minor adverse effects were 67% more frequent (RR 1.67) [4]. The label lists dry lips, dry skin, dry eye, nosebleeds, chapped lips, dermatitis and cheilitis among reactions occurring in 5% or more, and states cheilitis is dose related [3].
Population: 760 people in 11 trials [1]; 351 people in 2 trials [4]; label trials [3].
Sources: [1], [4], [3]
Caveats: The sources give event counts and a floor of 5%, not the share of people affected, so the band above is the lowest the label supports.

20 more

Screen timeStrongHow often is not known

Myopia in children

Linked to more myopia with more daily screen time in children and young people, in three meta-analyses, with odds rising about 21% per extra hour.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: Each extra hour of daily screen time was linked to 21% higher odds of myopia (OR 1.21, 95% CI 1.13 to 1.30). Odds rose from 1.05 at 1 hour a day to 1.97 at 4 hours, then more slowly [16]. In cohort studies, high against low screen time had an OR of 2.39 (95% CI 2.07 to 2.76), and each extra hour a day 1.07. Computers and TV were linked to myopia, smartphones were not [17]. Smart device time alone had an OR of 1.26 (95% CI 1.00 to 1.60), and 1.77 combined with computer use [18].
Population: 335,524 participants, mean age 9.3, in 45 studies [16]. 19 studies of children and adolescents [17]. People aged 3 months to 33 years in 33 studies, 11 pooled [18].
Sources: [16], [17], [18]
Caveats: Screen time was self-reported in every study, and screen use often replaces time outdoors, which itself is linked to myopia. Many studies were cross-sectional [18].

Inhaled corticosteroidsModerateHow often is not known

Cataracts and glaucoma

Linked to about 1.5 times the odds of cataract in a meta-analysis of four studies. The label reports glaucoma and cataracts with long-term use.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: Inhaled steroid use was associated with cataract (pooled OR 1.48, 95% CI 1.30 to 1.68), a number needed to harm of 16 [21]. The label reports glaucoma, raised eye pressure and cataracts with long-term use. In a 3-year COPD trial subset, new glaucoma was 2% on fluticasone-salmeterol, 5% on fluticasone alone and 2% on placebo, and cataract conclusions could not be drawn [15].
Population: 4 studies [21]; 658 people with COPD in an eye substudy [15].
Sources: [21], [15]
Caveats: The meta-analysis is small and from 2006 [21].

Vitamin AModerateHow often is not known

Eye irritation from topical vitamin A on or near the eye

Shown to raise eye pain, dryness and burning with topical vitamin A on or near the eye, mostly at prescription strength. Serious events did not differ.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: Against controls, topical vitamin A products applied to the eye surface or the skin around the eye raised the odds of eye pain (OR 6.37), dryness (OR 2.69), burning (OR 2.44) and any harm (OR 1.71). Serious adverse events did not significantly differ (OR 1.21). Prescription-strength products showed higher odds of harm, and nonprescription products were not significantly linked to more harm [13].
Population: 89 randomized and nonrandomized studies with 10,794 participants, 55 of them pooled [13].
Sources: [13]
Caveats: The authors rated the certainty very low to low. This covers products applied to the eye surface or the skin around it, not vitamin A taken by mouth [13].

AmitriptylineLimitedHow often is not known

Glaucoma, urinary retention and seizures

The label warns that amitriptyline's atropine-like action can trigger angle-closure glaucoma and urinary retention, and lists caution in people with seizures. It gives no rates.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: Pupil widening may trigger an angle-closure attack in people with narrow angles, and even average doses may precipitate an attack in people with angle-closure glaucoma. The label calls for caution in people with a history of seizures or urinary retention [13].
Population: Labeling statements [13].
Sources: [13]
Caveats: No rates are given.

BupropionLimitedHow often is not known

Angle-closure glaucoma

The label warns that pupil widening on antidepressants can trigger an angle-closure attack in people with untreated narrow angles. It gives no rate.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: The label states that the pupillary dilation that follows use of many antidepressants, including bupropion, may trigger an angle-closure attack in a person with anatomically narrow angles who has not had an iridectomy [8].
Population: Labeling statement [8].
Sources: [8]
Caveats: No rate and no trial data are given.

HomeopathyLimitedHow often is not known

Contaminated eye drops

FDA states that homeopathic eye drops it tested in 2019 were not sterile and had a dangerously high pH, which could cause eye infection or injury.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: In 2019 the FDA warned four companies making Puriton Eye Relief Drops. Multi-dose, preservative-free drops were made without any attempt at sterility, and samples were contaminated with Bacillus species or particulate matter and had a pH the FDA called dangerously high, which could lead to eye infection, glaucoma, corneal scarring and vision loss [2].
Population: FDA product testing [2].
Sources: [2]
Caveats: A finding about manufacturing of one product, not about homeopathy as a whole. FDA lists products for the eye and by injection among its higher-risk categories [2].

IsotretinoinLimitedHow often is not known

Eye and hearing effects

The label warns of reduced night vision that can persist, dry eyes, corneal opacities and hearing loss. It gives no rate.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: Decreased night vision has been reported, sometimes sudden and in some cases persisting after stopping. Corneal opacities resolved or were resolving 6 to 7 weeks after stopping. Dry eyes can make contact lenses hard to wear. Impaired hearing has been reported, sometimes persisting after treatment [3].
Population: Trial participants and postmarketing reports [3].
Sources: [3]
Caveats: No rates or comparison groups are given.

Light at nightLimitedHow often is not known

Side effects of blue-light filtering glasses

Limited evidence: trials reported infrequent side effects of blue-light filtering glasses, including headache, discomfort, lower mood and more depressive symptoms.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: Nine trials reported on device-related adverse events and three described any. Events with blue-light filtering lenses were infrequent and included increased depressive symptoms, headache, discomfort wearing the glasses and lower mood. Clear lenses brought occasional discomfort and raised mood [4].
Population: 333 participants in 9 randomized trials [4].
Sources: [4]
Caveats: Adverse events were not consistently reported, and the certainty was low [4].

PDE5 inhibitorsLimitedHow often is not known

Sudden vision or hearing loss

The label warns of sudden vision loss from NAION and sudden hearing loss. Case reports describe 54 people with hearing loss, and of 7 followed, 4 did not improve.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: The Cialis label warns that sudden loss of vision in one or both eyes could be non-arteritic anterior ischemic optic neuropathy (NAION), and notes higher risk with a prior NAION or a crowded optic disc. It also warns of sudden decrease or loss of hearing [7]. A systematic review of published cases found 54 people with moderate to profound sensorineural hearing loss while using PDE5 inhibitors, most often sildenafil, vardenafil or tadalafil. Of 7 with follow-up after stopping, 4 did not improve, 2 recovered fully and 1 partly [9].
Population: Postmarketing reports summarized in the label [7]; 54 patients aged 37 to 62 in 6 studies [9].
Sources: [7], [9]
Caveats: Postmarketing reports and case series cannot give a rate or establish cause. [9] had no control group.

SemaglutideLimitedHow often is not known

Optic nerve damage (NAION)

Linked to a higher rate of optic nerve damage in people with type 2 diabetes, about 1.4 more cases per 10,000 person-years than on SGLT-2 drugs.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: In Danish and Norwegian registries, people with type 2 diabetes starting semaglutide had a higher rate of non-arteritic anterior ischemic optic neuropathy than those starting SGLT-2 inhibitors (HR 2.81, 95% CI 1.67 to 4.75), an absolute difference of 1.41 more cases per 10,000 person-years [13].
Population: 61,377 semaglutide users with type 2 diabetes, 32 NAION events [13].
Sources: [13]
Caveats: An observational cohort with few events. The self-controlled analysis did not reach significance in either country [13].

SemaxGrey marketLimitedHow often is not known

Reported adverse events

Case reports: FDA's database held one report, eye pain and burning after semax nasal drops, with hospitalization. One voluntary report cannot show cause.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: FDA's adverse event database held one report for semax through December 3, 2025. A consumer reported eye pain and burning after using semax nasal drops in May 2024, with hospitalization, and the eye pain had not resolved when it was reported a year later. FDA found no case reports of adverse events in the published literature [3].
Population: Voluntary reports to FDA [3].
Sources: [3]
Caveats: A single voluntary report cannot establish cause or a rate. Reporting is voluntary and incomplete [3].

SSRIsLimitedHow often is not known

Angle-closure glaucoma

The label warns that pupil widening on antidepressants can trigger an angle-closure attack in people with untreated narrow angles. It gives no rate.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: The label states that the pupillary dilation that follows use of many antidepressants, including sertraline, may trigger an angle-closure attack in a person with anatomically narrow angles who has not had an iridectomy [24].
Population: Labeling statement [24].
Sources: [24]
Caveats: No rate and no trial data are given.

Thiazide diureticsLimitedHow often is not known

Sudden eye pain and blurred vision

The label warns that hydrochlorothiazide can cause sudden nearsightedness and acute angle-closure glaucoma, usually within hours to weeks of starting. It gives no rate.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: The label states that hydrochlorothiazide, a sulfonamide, can cause an idiosyncratic reaction resulting in acute transient myopia and acute angle-closure glaucoma, with sudden loss of visual sharpness or eye pain, typically within hours to weeks of starting [14].
Population: People taking hydrochlorothiazide, as described in the label [14].
Sources: [14]
Caveats: No rate or comparison group is given.

Tranexamic acidLimitedHow often is not known

Eye problems and retinal clots

The label reports retinal vein and artery occlusion, impaired color vision and ligneous conjunctivitis with tranexamic acid, from postmarketing and published reports.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: The label states retinal venous and arterial occlusion have been reported, and that ligneous conjunctivitis resolved after the drug was stopped. Postmarketing reports include impaired color vision and other visual disturbances [4].
Population: Postmarketing reports summarized in the label [4].
Sources: [4]
Caveats: Reports without a denominator. No rate is available.

TrazodoneLimitedHow often is not known

Angle-closure glaucoma

The label warns that pupil widening on antidepressants can trigger an angle-closure attack in people with untreated narrow angles. It gives no rate.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: The label states that the pupillary dilation that follows use of many antidepressants, including trazodone, may trigger an angle-closure attack in a person with anatomically narrow angles who has not had an iridectomy [12].
Population: Labeling statement [12].
Sources: [12]
Caveats: No rate and no trial data are given.

YogaLimitedHow often is not known

Serious harms in case reports

Case reports: 76 published cases of harm linked to yoga, most often after headstand, shoulder stand, lotus position or forceful breathing. One person died.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: Of 76 cases in 35 case reports and 2 case series, 35.5% affected the musculoskeletal system, 18.4% the nervous system and 11.8% the eyes. Ten had preconditions, mainly glaucoma and osteopenia. Fifteen recovered fully, 9 partially, 1 not at all, and 1 died [18].
Population: 76 published cases [18].
Sources: [18]
Caveats: Case reports give no rate and outcomes were not reported for every case [18].

YogaLimitedHow often is not known

Raised eye pressure in head-down poses

One small study found eye pressure rose within a minute in four head-down poses, from 17 to about 28 mm Hg in downward dog, returning to baseline within 2 minutes.

The numbers and sources

How common: How often is not known: the sources give no rate.
Effect: All four poses raised intraocular pressure (P < 0.01). In downward facing dog it rose from 17 to 28 mm Hg in people with glaucoma and from 17 to 29 mm Hg in people without it. The rise did not differ between groups, and pressure returned to baseline within 2 minutes of sitting up [14].
Population: 10 people with primary open-angle glaucoma and 10 without [14].
Sources: [14]
Caveats: Twenty people and a short exposure. The authors state it is unknown whether these rises speed glaucoma progression [14].

Hyperbaric oxygen therapyModerate

FDA warnings on unproven uses

FDA states hyperbaric devices are not proven to cure cancer, Lyme disease, autism or Alzheimer's disease, and that explosions and fires have occurred in unreviewed chambers at unaccredited facilities.

The numbers and sources

Effect: FDA's consumer update lists the risks as ear and sinus pain, middle ear injuries including eardrum rupture, temporary vision changes, and rarely lung collapse. It says high oxygen concentrations pose a risk of fire, that explosions and fires have occurred in chambers not reviewed by FDA at unaccredited facilities, and that people have used zippered altitude-sickness bags to build homemade devices that pose risks of fire and suffocation. It names the Undersea and Hyperbaric Medical Society as the accrediting body [1].
Population: US consumers [1].
Sources: [1]
Caveats: The update carries a content date of July 26, 2021, and is cited from an archived copy because the page was no longer on FDA's site when checked in November 2026. FDA also states that serious complications are rare when chambers are used for cleared conditions.

Whole-body cryotherapyModerate

FDA consumer warning

FDA states it has not cleared or approved any whole-body cryotherapy device for medical treatment, and names asphyxiation, frostbite, burns and eye injury as hazards.

The numbers and sources

Effect: In a consumer update dated July 5, 2016, FDA said it does not have evidence that whole-body cryotherapy treats conditions such as Alzheimer's, fibromyalgia, migraines, rheumatoid arthritis, multiple sclerosis, stress, anxiety or chronic pain. It listed potential hazards including asphyxiation, especially when liquid nitrogen is used for cooling, because nitrogen vapor in a closed room lowers oxygen and can cause loss of consciousness, along with frostbite, burns and eye injury. It also voiced concern about people choosing cryotherapy in place of established treatment [1].
Population: US consumers [1].
Sources: [1], [25]
Caveats: The update dates from 2016 and is cited here from an archived copy, since the page was no longer on FDA's site when checked in November 2026. A later safety review argues that some reports behind regulator concerns involved partial-body nitrogen cabins, not whole-body chambers [25].

Red light therapyLimited

Retinal injury from red lasers aimed at the eye

Case reports: one young patient had reversible vision loss and retinal changes after repeated low-level red light for myopia, and two such devices exceeded laser exposure limits in testing.

The numbers and sources

Effect: A systematic review of 20 studies (2,380 participants aged 3 to 18) found two case reports describing the same patient, whose reduced visual acuity and retinal scan abnormalities resolved 4 months after stopping. No permanent vision loss was reported, temporary afterimages were the most common symptom, and the incidence of side effects was 0.088 per 100 patient-years (95% CI 0.02 to 0.50) [37]. A 2023 case report describes retinal damage after repeated low-level red-light laser exposure [38]. Bench testing of two red light instruments for myopia found that 3 minutes of continuous viewing approached or surpassed the maximum permissible exposure for the retina [39].
Population: 1,436 children treated with repeated low-level red light for a median of 9 months [37]. One patient [38]. Two devices [39].
Sources: [37], [38], [39]
Caveats: This concerns instruments that send laser light into the eye. The longest follow-up was 24 months [37]. No comparable study of eye exposure from skin panels or masks was located.

Mixed results2

Light at nightModerate

Blue-light filtering glasses in the evening

Mixed results: three reviews of blue-light filtering glasses found sleep improved in some trials and not others, with no change in objectively measured sleep against clear lenses.

Compared with non-blue-light filtering lenses [4]; visually identical clear-lens glasses worn at the same time [5]; control conditions the abstract does not describe [6]

The numbers and sources

Effect: Six trials gave inconsistent findings on sleep quality, three showing improvement and three no difference, with very low certainty of evidence [4]. In three double-blind crossover trials, the glasses did not change time to fall asleep (-4.86 minutes, 95% CI -20.23 to 10.52), total sleep time (8.75 minutes, 95% CI -35.31 to 52.82), sleep efficiency or wake after sleep onset, measured by wrist actigraphy [5]. A review of 16 randomized trials and other reports found substantial evidence of shorter time to fall asleep in people with sleep disorders, jet lag or variable shift schedules [6].
Population: 148 participants in 6 trials of sleep, within 17 trials overall [4]. 49 adults in 3 crossover trials [5]. 453 patients in 16 randomized trials, plus uncontrolled studies and case reports [6].
Sources: [4], [5], [6]
Caveats: Trials were small and short, and most were at high risk of bias [4]. The positive review was not a meta-analysis and included uncontrolled studies [6]. Populations ranged from healthy adults to people with insomnia, mania and depression.

Not shown to help3

Red light therapyModerate

Vision in dry macular degeneration

Not shown to improve vision in dry age-related macular degeneration at 12 months in a Cochrane review of 2 small trials.

The numbers and sources

Effect: Corrected visual acuity at 12 months did not differ meaningfully between photobiomodulation and sham or no treatment (mean difference 0.02 logMAR, 95% CI -0.02 to 0.05, 2 RCTs, 90 eyes). Conversion to the exudative form did not differ (risk ratio 0.97, 95% CI 0.17 to 5.44) [34].
Population: 90 participants with non-exudative age-related macular degeneration in 2 single-center RCTs [34].
Sources: [34]
Caveats: Low to very low certainty. One trial was unmasked, and investigators in the other reported conflicts of interest [34]. Searches ran to May 2020.

Screen timeModerate

Blue-light filtering glasses

Not shown to ease eye strain from computer use against ordinary lenses in short trials, with effects on sleep uncertain, in a Cochrane review of 17 small trials.

Compared with non-blue-light filtering lenses [8]

The numbers and sources

Effect: Visual fatigue scores may not differ at under 1 week (mean difference 9.76, 95% CI -33.95 to 53.47, 120 participants, low certainty). Visual acuity probably did not differ (MD 0.00 logMAR). Of six trials of sleep, three found better sleep scores and three found no difference (very low certainty). Reported side effects included headache, discomfort and lower mood [8].
Population: Adults in 17 randomized trials of 5 to 156 participants, followed under 1 day to 5 weeks [8].
Sources: [8]
Caveats: No trial had a low risk of bias in every domain, and no meta-analysis was possible [8].

Vitamin AModerateStudied in pregnancy

Pregnancy outcomes

Not shown to lower maternal or newborn death or preterm birth in pregnant women. Night blindness and anemia in mothers were lower.

The numbers and sources

Effect: Vitamin A alone did not change maternal death (RR 0.88, 95% CI 0.65 to 1.20), perinatal death (RR 1.01, 95% CI 0.95 to 1.07) or preterm birth (RR 0.98, 95% CI 0.94 to 1.01). Maternal night blindness (RR 0.79, 95% CI 0.64 to 0.98) and maternal anemia (RR 0.64, 95% CI 0.43 to 0.94) were lower [2].
Population: 19 trials with over 310,000 women, mostly in Nepal, Ghana and Bangladesh [2].
Sources: [2]
Caveats: Baseline vitamin A status differed between populations. The reduction in maternal infection (RR 0.45, 95% CI 0.20 to 0.99) came from low quality evidence [2].

Not studied in people1

KPVGrey marketUntested

Wound healing (eye and mouth)

Not studied in people. KPV eye drops sped healing of scraped corneas in rabbits, and a gel containing KPV aided healing of mouth ulcers in rats.

No study in people. This rests on animal studies, lab studies.

The numbers and sources

Effect: After the corneal surface was scraped off in both eyes, rabbits given KPV eye drops had smaller remaining defects than those given saline. At 60 hours, 8 of 8 corneas treated with KPV or with a nitric oxide donor had fully healed over, against none given placebo. Blocking nitric oxide production inhibited the effect of KPV, and in cultured rabbit corneal cells KPV raised cell viability [15]. In rats with chemotherapy-induced mouth ulcers, a hydrogel carrying KPV and EGCG, an antibacterial compound, improved food intake, body weight recovery and tissue repair, and lowered IL-1beta and TNF-alpha [16].
Population: Rabbits, and rabbit corneal cells in the lab [15]. Rats with chemotherapy-induced oral mucositis [16].
Sources: [15], [16]
Caveats: The gel also contained EGCG, which has its own antibacterial activity, so its effect cannot be assigned to KPV alone [16]. Scraped corneas and mouth ulcers in animals do not show what happens to wounds in people.