Real-world landscape of drug-related ocular injuries: a retrospective pharmacovigilance study
Guangyao Li, Zhihui Song, Shuning Li, Jiawei Wang
TL;DR
This study identifies drugs linked to eye injuries using FDA data, highlighting high-risk medications and patient groups to improve safe drug use.
Contribution
The study detects new drug-ocular injury risk signals and characterizes their epidemiology using FAERS data.
Findings
Brolucizumab, macrogol 400, and cenegermin showed the highest risk for ocular injuries.
Blurred vision and visual impairment were the most common ocular injuries reported.
Most drugs showed early-onset adverse effects, emphasizing the need for close monitoring at treatment initiation.
Abstract
This study aimed to detect risk signals of drug-related ocular injuries, delineate their real-world epidemiological features, and provide evidence-based guidance for safe clinical use by utilizing the US Food and Drug Administration Adverse Event Reporting System (FAERS). Adverse event reports classified under “Eye disorders” (System Organ Class, SOC) were extracted from the FAERS database covering Q1 2004 to Q4 2024. Disproportionality analyses were employed to identify drug-ocular injury associations, and the time-to-onset (TTO) of adverse reactions was analyzed using Weibull distribution modeling. A total of 1,242,518 reports from 832,314 patients were included, with females accounting for 60.56% and elderly patients (≥65 years) for 21.63%. Serious outcomes comprised 62.80% of the reports. In total, 2,696 primary suspect drugs were identified, of which 359 met the signal detection…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1| Exposure group | Drug-related AEs | Non-drug-related AEs | Total |
|---|---|---|---|
| Drug | a | b | a + b |
| Non-drug | c | d | c + d |
| Total | a + c | b + d | N = a + b + c + d |
| Indicator | Number (%) |
|---|---|
| Gender | |
| Female (%) | 504,021 (60.56) |
| Male (%) | 251,228 (30.18) |
| Not Specified (%) | 77,065 (9.26) |
| Age | |
| <18 (%) | 29,961 (3.60) |
| 18–44 (%) | 113,971 (13.69) |
| 45–64 (%) | 186,712 (22.43) |
| ≥65 (%) | 180,008 (21.63) |
| NotSpecified (%) | 321,662 (38.65) |
| Age (Quantitative) | |
| N (Missing) | 510,652 (321,662) |
| Mean (SD) | 54.44 (20.21) |
| Median (Q1,Q3) | 57.00 (42.00.70.00) |
| Min,Max | 0.00,947.00 |
| Report year | |
| 2004 | 9605 (1.15) |
| 2005 | 12,298 (1.48) |
| 2006 | 13,522 (1.62) |
| 2007 | 13,681 (1.64) |
| 2008 | 15,162 (1.82) |
| 2009 | 16,755 (2.01) |
| 2010 | 25,465 (3.06) |
| 2011 | 26,470 (3.18) |
| 2012 | 29,281 (3.52) |
| 2013 | 35,555 (4.27) |
| 2014 | 35,595 (4.28) |
| 2015 | 50,057 (6.01) |
| 2016 | 48,343 (5.81) |
| 2017 | 52,526 (6.31) |
| 2018 | 59,067 (7.10) |
| 2019 | 61,143 (7.35) |
| 2020 | 61,923 (7.44) |
| 2021 | 60,668 (7.29) |
| 2022 | 62,655 (7.53) |
| 2023 | 64,832 (7.79) |
| 2024 | 77,711 (9.34) |
| Reporter | |
| Consumer (%) | 416,491 (50.04) |
| Lawyer (%) | 15,738 (1.89) |
| Not Specified (%) | 38,857 (4.67) |
| Other health-professional (%) | 80,203 (9.64) |
| Pharmacist (%) | 96,151 (11.55) |
| Physician (%) | 184,874 (22.21) |
| Outcome | |
| Life-threatening (%) | 20,708 (2.49) |
| Hospitalization - initial or prolonged (%) | 148,981 (17.90) |
| Disability (%) | 42,422 (5.10) |
| Death (%) | 17,839 (2.14) |
| Congenital Anomaly (%) | 4449 (0.53) |
| Required intervention to prevent permanent Impairment/Damage (%) | 5789 (0.70) |
| Other (%) | 425,819 (51.16) |
| Time-to-Onset | |
| 0-30d (%) | 129,323 (15.54) |
| 31-60d (%) | 14,124 (1.70) |
| 61-90d (%) | 8543 (1.03) |
| 91-120d (%) | 6232 (0.75) |
| 121-150d (%) | 4563 (0.55) |
| 151-180d (%) | 3844 (0.46) |
| 181-360d (%) | 15,217 (1.83) |
| 360d<(%) | 37,462 (4.50) |
| Missing or anomalous values (less than 0) (%) | 613,006 (73.65) |
| Time-to-onset (Quantitative) | |
| N (Missing) | 219,308 (613,006) |
| Mean (SD) | 272.63 (802.01) |
| Median (Q1,Q3) | 11.00 (0.00,161.00) |
| Min,Max | 0.00,73052.0 |
| ATC classifications | ATC code | Number of reports | ROR (95% CI) | PRR (95% CI) | IC (IC025) | EBGM (EBGM05) |
|---|---|---|---|---|---|---|
| Ntineoplastic and immunomodulating agents | L | 410,071 | 0.84 (0.84.0.85) | 0.85 (0.84.0.85) | −0.16 (-0.16) | 0.90 (0.89) |
| Sensory organs | S | 313,487 | 4.93 (4.91.4.95) | 4.60 (4.59.4.62) | 1.89 (1.88) | 3.69 (3.68) |
| Dermatologicals | D | 232,719 | 1.92 (1.92.1.93) | 1.89 (1.88.1.90) | 0.78 (0.78) | 1.72 (1.71) |
| Alimentary tract and metabolism | A | 219,439 | 1.39 (1.38.1.39) | 1.37 (1.37.1.38) | 0.39 (0.38) | 1.31 (1.30) |
| Nervous system | N | 177,608 | 0.71 (0.71.0.72) | 0.72 (0.71.0.72) | −0.40 (-0.41) | 0.76 (0.75) |
| Respiratory system | R | 154,599 | 1.78 (1.77.1.79) | 1.75 (1.74.1.76) | 0.73 (0.72) | 1.66 (1.65) |
| Cardiovascular system | C | 110,852 | 0.94 (0.93.0.94) | 0.94 (0.93.0.94) | −0.09 (-0.10) | 0.94 (0.94) |
| Genito urinary system and sex hormones | G | 96,414 | 0.93 (0.93.0.94) | 0.94 (0.93.0.94) | −0.09 (-0.10) | 0.94 (0.93) |
| Musculo-skeletal system | M | 64,534 | 0.95 (0.94.0.96) | 0.95 (0.94.0.96) | −0.07 (-0.08) | 0.95 (0.95) |
| Antiinfectives for systemic use | J | 63,599 | 0.85 (0.84.0.86) | 0.85 (0.85.0.86) | −0.22 (-0.23) | 0.86 (0.85) |
| Blood and blood forming organs | B | 47,330 | 0.63 (0.62.0.63) | 0.63 (0.62.0.64) | −0.63 (-0.65) | 0.64 (0.64) |
| Systemic hormonal preparations, excl. Sex hormones and insulins | H | 43,274 | 0.91 (0.90.0.92) | 0.91 (0.90.0.92) | −0.13 (-0.14) | 0.92 (0.91) |
| Various | V | 31,951 | 0.91 (0.90.0.92) | 0.91 (0.90.0.92) | −0.13 (-0.15) | 0.91 (0.90) |
| Antiparasitic products, insecticides and repellents | P | 7350 | 1.64 (1.60.1.68) | 1.62 (1.58.1.66) | 0.69 (0.66) | 1.62 (1.58) |
| ATC code | Drug name | Number of reports | ROR (95% CI) | PRR (95% CI) | IC (IC025) | EBGM (EBGM05) |
|---|---|---|---|---|---|---|
| A16AX; D11AH; R01; R03DX | Dupilumab | 63,490 | 4.23 (4.20.4.27) | 3.96 (3.92.3.99) | 1.93 (1.92) | 3.80 (3.77) |
| L04AD; S01XA | Ciclosporin | 27,059 | 7.71 (7.61.7.81) | 6.72 (6.65.6.79) | 2.72 (2.70) | 6.59 (6.51) |
| S01LA | Ranibizumab | 22,247 | 30.13 (29.62.30.65) | 18.34 (18.15.18.53) | 4.17 (4.15) | 18.03 (17.72) |
| L01XX; S01LA | Aflibercept | 21,024 | 30.22 (29.69.30.76) | 18.37 (18.17.18.56) | 4.18 (4.15) | 18.07 (17.76) |
| D11AX; S01EE | Bimatoprost | 18,944 | 41.29 (40.48.42.12) | 21.83 (21.61.22.06) | 4.43 (4.40) | 21.52 (21.09) |
| L04AE | Fingolimod | 15,834 | 2.59 (2.55.2.63) | 2.50 (2.46.2.54) | 1.31 (1.29) | 2.48 (2.44) |
| A10AE | Insulin glargine | 15,013 | 2.95 (2.90.2.99) | 2.82 (2.78.2.87) | 1.49 (1.46) | 2.80 (2.76) |
| A10A; A10AB; A10AC; A10AD | Insulin lispro | 14,862 | 3.44 (3.38.3.50) | 3.26 (3.21.3.31) | 1.69 (1.67) | 3.23 (3.18) |
| A01AD; D11AA; D11AX; G04BD; M03AX; N02CX | Botulinum toxin type a | 14,551 | 3.84 (3.77.3.90) | 3.61 (3.55.3.66) | 1.84 (1.81) | 3.58 (3.52) |
| S01XA | Cenegermin | 13,454 | 60.19 (58.64.61.78) | 26.00 (25.72.26.29) | 4.69 (4.65) | 25.73 (25.07) |
| S01EE | Latanoprost | 13,152 | 21.68 (21.23.22.14) | 14.85 (14.65.15.06) | 3.88 (3.85) | 14.71 (14.40) |
| S01XA | Lifitegrast | 11,770 | 28.22 (27.57.28.88) | 17.58 (17.33.17.83) | 4.12 (4.09) | 17.42 (17.02) |
| L01FG; S01LA | Bevacizumab | 10,565 | 2.30 (2.26.2.35) | 2.24 (2.20.2.28) | 1.15 (1.13) | 2.23 (2.18) |
| L01CD | Docetaxel | 9924 | 2.57 (2.52.2.62) | 2.48 (2.43.2.53) | 1.30 (1.27) | 2.47 (2.42) |
| S01LA | Brolucizumab | 8409 | 132.15 (126.61,137.94) | 33.67 (33.31.34.04) | 5.06 (5.02) | 33.45 (32.05) |
| D11AX; S01EA; S01GA | Brimonidine | 7151 | 16.48 (16.03.16.93) | 12.25 (12.00.12.49) | 3.61 (3.57) | 12.18 (11.85) |
| C02KX; G04BE | Sildenafil | 6668 | 3.05 (2.97.3.12) | 2.91 (2.85.2.98) | 1.54 (1.50) | 2.90 (2.83) |
| A10A; A10AB; A10AC; A10AD; A10AE; A10AF | Insulin human | 6493 | 3.25 (3.17.3.33) | 3.09 (3.02.3.16) | 1.62 (1.59) | 3.08 (3.00) |
| N02CX; N03AX | Topiramate | 5726 | 3.23 (3.14.3.32) | 3.07 (3.00.3.15) | 1.62 (1.58) | 3.06 (2.98) |
| B01AB; C05BA; G04BX | Pentosan polysulfate | 5120 | 35.97 (34.66.37.32) | 20.19 (19.78.20.60) | 4.33 (4.28) | 20.11 (19.38) |
| A01AC; A07EA; C05AA; D07AA; H02AB; M02AX; R01AD; S01BA; S02BA; S03BA | Prednisolone | 5005 | 2.12 (2.07.2.19) | 2.07 (2.02.2.13) | 1.05 (1.01) | 2.07 (2.01) |
| C02KX; G04BE; G04CX | Tadalafil | 4586 | 2.61 (2.54.2.69) | 2.52 (2.45.2.59) | 1.33 (1.29) | 2.52 (2.44) |
| A06AD; S01XA | Macrogol 400 | 4515 | 117.96 (111.52,124.77) | 32.62 (32.13.33.12) | 5.02 (4.96) | 32.51 (30.73) |
| S01ED | Dorzolamide; timolol | 4408 | 34.24 (32.91.35.61) | 19.64 (19.21.20.08) | 4.29 (4.23) | 19.57 (18.81) |
| A01AC; C05AA; D07AB; H02AB; R01AD; R03BA; S01BA | Triamcinolone | 4276 | 4.77 (4.62.4.92) | 4.40 (4.27.4.52) | 2.13 (2.08) | 4.38 (4.25) |
| D06AX; J01MA; S01AE; S03AA | Moxifloxacin | 4042 | 2.88 (2.79.2.98) | 2.77 (2.69.2.85) | 1.47 (1.42) | 2.76 (2.68) |
| L01FX | Belantamab mafodotin | 3904 | 48.94 (46.75.51.23) | 23.61 (23.11.24.13) | 4.56 (4.49) | 23.54 (22.49) |
| C01BD | Amiodarone | 3552 | 2.08 (2.01.2.15) | 2.03 (1.97.2.10) | 1.02 (0.97) | 2.03 (1.96) |
| S01LA | Faricimab | 3523 | 26.73 (25.63.27.87) | 16.96 (16.52.17.41) | 4.08 (4.02) | 16.91 (16.22) |
| S01EE | Travoprost | 3519 | 25.19 (24.16.26.26) | 16.34 (15.92.16.78) | 4.03 (3.96) | 16.30 (15.63) |
| PT | Number of reports |
|---|---|
| Vision blurred | 121,517 |
| Visual impairment | 113,320 |
| Cataract | 51,826 |
| Eye pain | 46,937 |
| Eye irritation | 45,436 |
| Ocular hyperaemia | 39,886 |
| Dry eye | 39,479 |
| Blindness | 35,863 |
| Eye swelling | 32,907 |
| Visual acuity reduced | 31,890 |
| Eye disorder | 29,085 |
| Eye pruritus | 27,097 |
| Lacrimation increased | 25,699 |
| Diplopia | 23,172 |
| Glaucoma | 17,281 |
| Photophobia | 16,290 |
| Conjunctivitis | 16,170 |
| Blindness unilateral | 12,616 |
| Eye haemorrhage | 12,204 |
| Seasonal allergy | 11,989 |
| Uveitis | 11,619 |
| Mydriasis | 11,475 |
| Madarosis | 11,201 |
| Eyelid oedema | 11,084 |
| Macular degeneration | 10,184 |
| Eye infection | 9857 |
| Eyelid ptosis | 8979 |
| Ocular discomfort | 8652 |
| Eye discharge | 8637 |
| Vitreous floaters | 8567 |
| Drug name | Preferred term (PT) name | Number of reports | ROR (95% CI) | PRR (95% CI) | IC (IC025) | EBGM (EBGM05) |
|---|---|---|---|---|---|---|
| Pentosan polysulfate | Pigmentary maculopathy | 426 | 431,861 (178794,1,043,119) | 415,645 (172148,1,003,557) | 12.24 (8.42) | 4822.85 (1996.71) |
| Moxifloxacin; triamcinolone | Haemorrhagic occlusive retinal vasculitis | 5 | 58,840.7 (22,613.3,153,106) | 56,905.1 (22,473.9,144,087) | 15.59 (1.30) | 49,215.4 (18,914.1) |
| Voretigene neparvovec | Foveal degeneration | 5 | 58,148.2 (20,828.9,162,333) | 57,305.5 (20,753.8,158,233) | 15.37 (1.23) | 42,225.4 (15,125.3) |
| Besifloxacin | Diffuse lamellar keratitis | 14 | 58,081.9 (26,315.2,128,196) | 57,419.2 (26,118.9,126,229) | 14.62 (2.97) | 25,265.0 (11,446.8) |
| Cyclopentolate; phenylephrine; tropicamide | Punctate keratitis | 7 | 42,432.9 (16,660.7,108,071) | 26,800.1 (14,811.4,48,492.7) | 14.70 (1.79) | 26,555.8 (10,426.8) |
| Aprotinin; calcium chloride; factor i (fibrinogen); factor xiii (fibrin stabilising factor); thrombin | Conjunctival granuloma | 3 | 35,637.7 (10,518.3,120,746) | 35,184.7 (10,525.7,117,613) | 14.90 (0.43) | 30,595.5 (9030.12) |
| Pentosan polysulfate | Hereditary retinal dystrophy | 7 | 34,170.3 (4203.63,277,762) | 34,149.2 (4201.71,277,546) | 12.06 (1.60) | 4269.53 (525.24) |
| Mitomycin | Conjunctival filtering bleb leak | 23 | 28,405.5 (14,124.0,57,127.6) | 28,231.7 (14,058.2,56,694.6) | 13.24 (3.82) | 9680.09 (4813.22) |
| Ocriplasmin | Ciliary zonular dehiscence | 5 | 24,407.1 (7060.92,84,366.3) | 24,353.4 (7055.02,84,065.9) | 13.57 (1.12) | 12,177.2 (3522.84) |
| Lotilaner | Demodex blepharitis | 4 | 24,080.8 (6461.17,89,748.8) | 24,028.5 (6457.39,89,412.0) | 13.70 (0.74) | 13,349.6 (3581.87) |
| Calcium chloride; magnesium chloride; potassium; sodium acetate; sodium chloride; sodium citrate | Toxic anterior segment syndrome | 411 | 22,935.6 (20,022.9,26,271.8) | 18,666.5 (16,544.2,21,061.0) | 13.39 (8.45) | 10,700.2 (9341.39) |
| Carbachol | Iris haemorrhage | 3 | 22,366.6 (6882.15,72,690.0) | 21,967.2 (6894.03,69,996.4) | 14.33 (0.48) | 20,594.3 (6336.82) |
| Tetracaine | Aqueous fibrin | 6 | 21,085.7 (8929.39,49,791.5) | 20,738.2 (8892.47,48,363.6) | 14.16 (1.63) | 18,249.7 (7728.40) |
| Calcium chloride; magnesium chloride; potassium; sodium acetate; sodium chloride; sodium citrate | Aqueous fibrin | 22 | 19,896.3 (11,365.2,34,831.2) | 19,698.1 (11,287.3,34,376.4) | 13.43 (3.79) | 11,031.4 (6301.36) |
| Pentosan polysulfate | Retinal pigmentation | 605 | 17,815.5 (15,041.0,21,101.9) | 16,865.5 (14,268.4,19,935.3) | 11.89 (8.87) | 3784.71 (3195.29) |
| Carmellose; hypromellose | Conjunctival granuloma | 3 | 14,934.4 (4425.18,50,401.7) | 14,854.3 (4426.50,49,847.3) | 13.66 (0.44) | 12,916.9 (3827.37) |
| Belantamab mafodotin | Corneal epithelial microcysts | 73 | 14,161.3 (9593.50,20,903.9) | 14,021.4 (9511.50,20,669.6) | 12.25 (5.76) | 4883.10 (3308.03) |
| Calcium chloride | Dellen | 4 | 13,989.4 (4505.99,43,431.9) | 13,947.1 (4503.93,43,189.5) | 13.35 (0.85) | 10,460.6 (3369.36) |
| Ascorbic acid; betacarotene; tocopherol | Chemical burns of eye | 5 | 13,585.2 (5424.59,34,022.6) | 12,556.1 (5369.46,29,361.6) | 13.59 (1.34) | 12,370.9 (4939.72) |
| Mitomycin | Blebitis | 18 | 13,320.4 (7040.65,25,201.3) | 13,256.6 (7018.21,25,040.2) | 12.77 (3.44) | 6977.64 (3688.11) |
| Boric acid; pilocarpine; thiamine | Corneal opacity | 4 | 13,136.3 (4317.72,39,965.9) | 10,217.3 (4297.19,24,293.5) | 13.31 (0.87) | 10,183.5 (3347.18) |
| Ivermectin | Onchocerciasis | 3 | 12,842.0 (1335.68,123,470) | 12,839.0 (1335.54,123,425) | 11.65 (0.06) | 3210.49 (333.92) |
| Amantadine | Opsoclonus | 7 | 12,731.1 (3291.60,49,240.3) | 12,722.3 (3290.26,49,192.6) | 11.90 (1.66) | 3817.38 (986.98) |
| Trypan blue | Corneal oedema | 4 | 12,048.2 (3707.30,39,154.9) | 8341.36 (3686.45,18,874.1) | 13.02 (0.82) | 8325.06 (2561.67) |
| Brolucizumab | Retinal perivascular sheathing | 66 | 11,718.2 (7528.59,18,239.4) | 11,649.2 (7490.05,18,117.8) | 11.76 (5.58) | 3470.67 (2229.79) |
| Carbachol | Anterior chamber fibrin | 4 | 11,639.4 (4245.94,31,907.3) | 11,362.3 (4242.23,30,432.8) | 13.42 (0.99) | 10,983.6 (4006.71) |
| Netarsudil | Blepharitis allergic | 9 | 11,566.9 (5291.32,25,285.5) | 11,516.4 (5280.83,25,114.8) | 12.98 (2.27) | 8061.77 (3687.88) |
| Besifloxacin | Corneal flap complication | 3 | 11,306.4 (3186.61,40,116.4) | 11,278.8 (3186.70,39,919.3) | 13.14 (0.38) | 9023.22 (2543.11) |
| Netarsudil | Cornea verticillata | 59 | 11,256.3 (8290.49,15,283.1) | 10,933.9 (8103.57,14,752.9) | 12.92 (5.47) | 7771.96 (5724.20) |
| Belantamab mafodotin | Keratopathy | 771 | 10,819.2 (9696.23,12,072.2) | 9690.38 (8729.35,10,757.2) | 12.04 (9.22) | 4225.48 (3786.90) |
| Durg name | Number of cases | TTO (days) Median (IQR) | Scale parameter α | 95% CI | Shape parameter β | 95% CI | Failure type |
|---|---|---|---|---|---|---|---|
| Dupilumab | 4862 | 29.00 (0.00,150.00) | 156.34 | 148.32–164.81 | 0.68 | 0.66–0.69 | Early failure |
| Fingolimod | 3910 | 37.50 (0.00,199.00) | 201.44 | 188.65–215.10 | 0.59 | 0.57–0.60 | Early failure |
| Aflibercept | 3215 | 63.00 (1.00,435.00) | 280.61 | 261.22–301.43 | 0.58 | 0.56–0.60 | Early failure |
| Botulinum toxin type a | 3154 | 1.00 (0.00.6.00) | 14.17 | 12.81–15.67 | 0.49 | 0.47–0.50 | Early failure |
| Ranibizumab | 2892 | 67.00 (2.00,273.00) | 200.1 | 187.00–214.11 | 0.63 | 0.61–0.65 | Early failure |
| Docetaxel | 2540 | 63.00 (18.50,265.00) | 152.58 | 144.17–161.47 | 0.76 | 0.73–0.78 | Early failure |
| Ciclosporin | 1857 | 0.00 (0.00.16.00) | 69.56 | 59.19–81.74 | 0.44 | 0.42–0.46 | Early failure |
| Bevacizumab | 1617 | 7.00 (0.00,107.00) | 90.57 | 80.58–101.79 | 0.52 | 0.50–0.54 | Early failure |
| Cenegermin | 1513 | 4.00 (1.00.22.00) | 28.42 | 25.17–32.10 | 0.51 | 0.49–0.53 | Early failure |
| Bimatoprost | 1408 | 1.00 (0.00.30.00) | 38.84 | 33.30–45.29 | 0.45 | 0.43–0.47 | Early failure |
| Brolucizumab | 1302 | 35.00 (8.00.86.00) | 78.77 | 73.17–84.80 | 0.85 | 0.81–0.89 | Early failure |
| Latanoprost | 1203 | 61.00 (0.00,731.00) | 559.18 | 493.57–633.51 | 0.57 | 0.54–0.60 | Early failure |
| Lifitegrast | 873 | 0.00 (0.00.4.00) | 29.05 | 23.30–36.22 | 0.52 | 0.49–0.57 | Early failure |
| Naphazoline; pheniramine | 824 | 0.00 (0.00.0.00) | 38.4 | 23.23–63.48 | 0.35 | 0.31–0.39 | Early failure |
| Topiramate | 801 | 9.00 (3.00.30.00) | 53.35 | 44.71–63.66 | 0.46 | 0.44–0.48 | Early failure |
| Tadalafil | 772 | 7.00 (0.00,161.00) | 135.16 | 111.33–164.09 | 0.46 | 0.43–0.49 | Early failure |
| Sildenafil | 764 | 30.00 (0.00,609.50) | 452.6 | 383.74–533.82 | 0.55 | 0.51–0.59 | Early failure |
| Insulin glargine | 757 | 181.00 (2.00,762.00) | 560.8 | 490.52–641.16 | 0.63 | 0.59–0.67 | Early failure |
| Moxifloxacin | 752 | 0.00 (0.00.4.00) | 16.34 | 13.02–20.50 | 0.48 | 0.45–0.52 | Early failure |
| Insulin lispro | 626 | 366.00 (3.00,1795.00) | 1215.49 | 1054.52–1401.03 | 0.65 | 0.60–0.70 | Early failure |
| Brimonidine | 597 | 0.00 (0.00.5.00) | 76.74 | 55.70–105.71 | 0.41 | 0.37–0.45 | Early failure |
| Faricimab | 560 | 50.00 (1.00,142.50) | 118.86 | 105.73–133.63 | 0.84 | 0.78–0.91 | Early failure |
| Insulin human | 532 | 1445.50 (17.00,4017.00) | 2763.25 | 2439.08–3130.51 | 0.79 | 0.73–0.86 | Early failure |
| Belantamab mafodotin | 414 | 35.00 (19.00.59.00) | 62.78 | 56.01–70.38 | 0.93 | 0.87–1.00 | Random failure |
| Triamcinolone | 372 | 1.00 (0.00.6.00) | 30.72 | 21.42–44.06 | 0.38 | 0.35–0.42 | Early failure |
| Prednisolone | 367 | 2.00 (0.00.14.00) | 47.75 | 33.85–67.36 | 0.41 | 0.37–0.45 | Early failure |
| Pentosan polysulfate | 359 | 269.00 (0.00,3344.00) | 3313.91 | 2883.14–3809.03 | 1.03 | 0.91–1.16 | Random failure |
| Macrogol 400 | 265 | 0.00 (0.00.1.00) | 6.72 | 4.36–10.37 | 0.56 | 0.48–0.64 | Early failure |
| Travoprost | 240 | 1.00 (0.00.54.00) | 71.06 | 48.80–103.48 | 0.45 | 0.40–0.50 | Early failure |
| Dorzolamide; timolol | 238 | 1.00 (0.00.70.00) | 169.97 | 111.47–259.18 | 0.43 | 0.38–0.49 | Early failure |
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsDrug-Induced Ocular Toxicity · Pharmacovigilance and Adverse Drug Reactions · Retinal Diseases and Treatments
Introduction
1
Adverse drug reactions (ADRs) represent a global public health concern threatening patient safety, contributing significantly to hospitalization and mortality rates (Alhawass et al., 2014; Giardina et al., 2018). Among ADRs, drug-related ocular disorders warrant particular attention due to their insidious symptoms and potential for permanent visual impairment (Slattery et al., 2014). The unique anatomical and physiological characteristics of ocular tissues, such as the high permeability of the blood-ocular barrier, render them susceptible to the effects of both systemic and topical medications (Slattery et al., 2014). Clinical manifestations include dry eye syndrome, glaucoma, and retinopathies. For instance, antiepileptic drugs can induce retinopathy and optic nerve damage, immune checkpoint inhibitors may trigger uveitis and keratopathy, and systemic medications (e.g., antidepressants) are also associated with an increased risk of dry eye syndrome (Han et al., 2017; Hamed, 2019; Liu et al., 2020).
In recent years, the US Food and Drug Administration Adverse Event Reporting System (FAERS) database has emerged as a crucial tool for investigating drug-related ocular injuries (Ali et al., 2025; Chen et al., 2025; Chen et al., 2024; Huang et al., 2025; Le et al., 2024; Li et al., 2025a; Li et al., 2025b; Li et al., 2025c; Wu et al., 2025a; Wu et al., 2024a). Studies utilizing FAERS have revealed that ocular adverse events (AEs) associated with certain drugs (e.g., cataracts, lens opacities with cariprazine) were not documented in their prescribing information (Zhu et al., 2024). Dupilumab may induce ocular degenerative changes via the IL-4/IL-13 pathway (Bridgewood et al., 2022), and dexamethasone implants have been linked to severe complications such as corneal decompensation (Zhao et al., 2025). FAERS, with its large-scale data, real-time updates, and standardized analytical methods, offers advantages over traditional clinical studies in detecting rare or delayed signals (Wong et al., 2015; Hu et al., 2024; Jiang et al., 2024). However, existing research primarily focuses on individual drugs, specific drug classes (e.g., systemic medications), or particular ocular AEs, lacking a systematic, panoramic investigation encompassing all drug categories. For instance, while recent studies have provided valuable insights into ocular adverse reactions associated with systemic drugs using the FAERS database (Wu et al., 2025b), a comprehensive real-world profile of drug-related ocular injury across all drug categories—particularly regarding local ophthalmic agents—remains elusive. This gap hinders a holistic understanding of the risk landscape and the development of stratified monitoring strategies.
The objective of our study is to identify and analyze reports of drug-related ocular injury adverse events within the FAERS database over the past 2 decades. We aim to delineate the clinical characteristics of drug-related ocular injury in the real world, identify high-risk drugs associated with ocular damage, define the types of ocular injuries and their temporal patterns, and ultimately enhance the safety of pharmacotherapy.
Methods
2
Data source
2.1
Data for this study were sourced from the FAERS raw database downloaded from the official US FDA website (https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html; FAERS Quarterly Data Extract Files). Publicly available since Q1 2004 and updated quarterly, we downloaded ASCII data packages covering Q1 2004 to Q4 2024. Each quarterly file contained seven data types: DEMO (patient demographic and administrative information; one record per event report), REAC (adverse events coded using Medical Dictionary for Regulatory Activities [MedDRA] terms), DRUG (drug/biologic information for reported medications), OUTC (patient outcomes for the event), RPSR (report sources), THER (drug therapy start and end dates), and INDI (indications for use/diagnoses coded using MedDRA terms). The breadth of FAERS data provides a robust basis for drug safety signal detection (Li et al., 2024).
Data cleaning
2.2
Data cleaning strictly followed FDA guidance documents. The deduplication process involved selecting the PRIMARYID, CASEID, and FDA_DT fields from the DEMO table. Reports were sorted by CASEID, FDA_DT, and PRIMARYID. For reports sharing the same CASEID, the report with the most recent FDA_DT was retained. If CASEID and FDA_DT were identical, the report with the largest PRIMARYID was kept (Zhao and Tao, 2024; Yin et al., 2022). Starting from Q1 2019, each quarterly data package included a deletion report list; reports listed therein were excluded after deduplication. From the FAERS database, we obtained data from 22,375,298 patient reports. Applying FDA deduplication rules, 3,761,306 duplicate reports from the same patients were removed, resulting in 18,613,992 unique patients and 55,357,463 AE reports.
Data standardization
2.3
AE names in FAERS are described using Preferred Terms (PTs) from MedDRA. As MedDRA is updated biannually (March and September), potentially altering PTs and their corresponding System Organ Classes (SOCs), we standardized FAERS PTs using the latest MedDRA dictionary version (MedDRA v27.1) and mapped them to the current SOCs for subsequent analysis. After extracting drug names from the FAERS database, we systematically mapped and standardized drug trade names, aliases, and generic names by referencing the World Health Organization Drug Dictionary (September 2024 version), converting them to International Nonproprietary Names for subsequent analysis. The core purpose of this process was to consolidate reports corresponding to the same active ingredient, thereby ensuring the accuracy and consistency of drug identification. The Anatomical Therapeutic Chemical (ATC) classification codes for drugs were annotated according to the World Health Organization ATC index.
Identification of drug-related ocular injuries
2.4
In this study, the adverse drug events were derived from the MedDRA v27.1, (http://www.meddra.org/). MedDRA PTs serve as unique identifiers for specific medical concepts (e.g., symptoms, signs, disease diagnoses). The terminology is hierarchically organized into five levels, grouping concepts via High-Level Terms (HLTs), High-Level Group Terms (HLGTs), and PTs. Ultimately, HLGTs are classified into SOCs based on etiology, manifestation site, or purpose, with each PT assigned a primary SOC. Utilizing this framework, we focused our analysis on PTs classified under the primary SOC of 'Eye disorders' in MedDRA v27.1. All PTs primarily classified under this SOC were included in the analysis. The complete list of specific PTs used is provided in Supplementary Table S1. We extracted AE reports involving ocular injury where a drug was designated as the “Primary Suspect” (PS). These reports were characterized by sex, age, time to onset, reporter type, reporting country, and outcome.
Statistical analysis
2.5
To assess associations between drugs and ocular injury AEs, we employed four established signal detection algorithms (Table 1; Supplementary Table S2): Reporting Odds Ratio (ROR), Proportional Reporting Ratio (PRR), Bayesian Confidence Propagation Neural Network (BCPNN), and Multi-item Gamma Poisson Shrinker (MGPS). Signal identification criteria were: (1) ROR: ≥3 reports and lower limit of 95% confidence interval (CI) >1; (2) PRR: ≥3 reports and lower limit of 95% CI >1; (3) BCPNN: Information Component (IC) lower bound (IC025) >0; and (4) MGPS: Empirical Bayes Geometric Mean lower bound (EBGM05) >2 and number of reports (a) >0. Meeting the threshold in all four methods indicated a potential drug-event association (Chen et al., 2024; Wu et al., 2025a; Wu et al., 2024a). Each method has specific strengths: ROR corrects for under-reporting bias, while PRR enhances specificity by comparing the reporting rate for the drug-event pair against others. BCPNN, based on Bayesian principles, integrates data from diverse sources and supports cross-validation, enhancing signal robustness. MGPS excels at identifying rare signals and managing sparse data. By combining these methods, we maximized detection coverage and validated findings from multiple perspectives, strengthening the assessment of drug-ocular injury associations. SAS software (version 9.4) was used for database processing and statistical analysis. Statistical significance was set at p < 0.05.
Time-to-onset analysis
2.6
Time-to-onset (TTO) was defined as the interval between the AE occurrence date (EVENT_DT in DEMO file) and the drug start date (START_DT in THER file). Reports with inaccurate date entries, missing specific dates, or errors (e.g., EVENT_DT before START_DT) were excluded. TTO was characterized using medians, interquartile ranges (IQR), and Weibull shape parameters (WSP) (Mazhar et al., 2021). The Weibull distribution shape is described by scale (α) and shape (β) parameters. AE hazard decreasing over time characterizes an early failure type (β < 1 and 95% CI < 1). A constant AE hazard over time characterizes a random failure type (β ≈ 1 and 95% CI includes 1). An increasing AE hazard over time characterizes a wear-out failure type (β > 1 and 95% CI > 1).
Results
3
Patient baseline characteristics and ocular injury adverse events
3.1
A total of 832,314 patients experiencing ocular injury AEs were included, involving 1,242,518 reports. Gender distribution showed a significantly higher proportion of females (60.56%, n = 504,021) compared to males (30.18%, n = 251,228), with sex unspecified in 9.26% (n = 77,065). The median age was 57.00 years (IQR 42.00–70.00). Patients aged ≥65 years accounted for 21.63% (n = 180,008), while age was unspecified in the largest proportion (38.65%, n = 321,662) (Table 2).
The majority of reports originated from the United States (62.91%, n = 523,595), followed by Canada (5.27%, n = 43,857) and the United Kingdom (4.37%, n = 36,371). Regarding AE outcomes, 62.80% were classified as serious, including hospitalization (17.90%, n = 148,981), disability (5.10%, n = 42,422), death (2.14%, n = 17,839), and life-threatening events (2.49%, n = 20,708), congenital anomaly (0.53%, n = 4449), required intervention to prevent permanent impairment/damage (0.70%, n = 5789), and others (51.16%, n = 425,819) (Table 2).
The TTO analysis revealed a median TTO of 11.00 days (IQR 0.00, 161.00) and a mean of 272.63 days (SD = 802.01). However, 73.65% of TTO data were excluded due to missing or anomalous values, leaving only 26.35% valid data. Among valid data, 15.54% of events occurred within 0–30 days post-drug initiation, while only 4.50% occurred after 360 days (Table 2).
Drug signal detection
3.2
The reported ocular injury AEs involved 2,696 unique “Primary Suspect” drugs, of which 359 met the signal detection criteria. The full list of these drug-induced ocular injury-related positive signal drugs and their corresponding ATC codes is provided in Supplementary Table S3. Analysis by ATC classification (Table 3) revealed that drugs for sensory organs (ATC: S class) exhibited the highest signal strength (ROR = 4.93, 95% CI 4.91–4.95; PRR = 4.60, 95% CI 4.59–4.62), indicating the strongest association with ocular injury. Dermatologicals (ATC: D class), alimentary tract and metabolism (ATC: A class), respiratory system (ATC: R class), and antiparasitic products, insecticides and repellents (ATC: P class) also showed strong associations with ocular injury AEs (ROR>1, PRR>1, and IC025 >0). Antineoplastic and immunomodulating agents (ATC: L class) had the highest report count (410,071 reports) but a low ROR value (0.84, 95% CI 0.84–0.85), below the overall signal threshold (Table 3).
Figure 1 and Table 4 displays the top 30 drugs with the most eye disorder reports. The top 30 drugs cover 14 ATC classes, including sensory organs (14 drugs), alimentary tract and metabolism (8 drugs), dermatologicals (7 drugs), and antineoplastic and immunomodulating agents (6 drugs). The top five drugs by report count are dupilumab (63,490 reports), ciclosporin (27,059 reports), ranibizumab (22,247 reports), aflibercept (21,024 reports), and bimatoprost (18,944 reports).Among specific drugs, brolucizumab (ROR = 132.15, 95% CI 126.61–137.94), macrogol 400 (ROR = 117.96, 95% CI 111.52–124.77), and cenegermin (ROR = 60.19, 95% CI 58.64–61.78) ranked highest in signal strength.
Top 30 primary suspect drugs for eye disorders (sorted by number of reports in Descending order).
Distribution of ocular injury types
3.3
Top 30 PTs for ocular injuries are shown in Table 5. The most common PTs for ocular injuries were vision blurred (121,517 reports), visual impairment (113,320 reports), cataract (51,826 reports), eye pain (46,937 reports), and eye irritation (45,436 reports). Ocular hyperaemia (39,886 reports), dry eye (39,479 reports) and blindness (35,863 reports) were also frequently reported.
Table 6 lists the top 30 drug-PT pairs by signal strength (ROR value) from the drug-ocular injury signal detection. Pentosan polysulfate showed the strongest association with pigmentary maculopathy (ROR = 431,861, 95% CI 178,794–1,043,119), suggesting an extremely high risk. Moxifloxacin; triamcinolone was strongly associated with haemorrhagic occlusive retinal vasculitis (ROR = 58,840.7), and voretigene neparvovec with foveal degeneration (ROR = 58,148.2), indicating significant associations with specific retinal or corneal pathologies.
Time-to-onset (TTO) analysis of adverse events
3.4
The TTO analysis based on the Weibull distribution is summarized in Table 7. The vast majority of drugs belonged to the Early Failure Type (β < 1 and 95% CI excludes 1), such as dupilumab (β = 0.68, 95% CI 0.66–0.69) and ciclosporin (β = 0.44, 95% CI 0.42–0.46), indicating the highest risk during the initial treatment period. Botulinum toxin type A had a median TTO of only 1 day, and ciclosporin had a median TTO of 0 days, suggesting some events occurred on the day of administration. The Random Failure Type (β ≈ 1 and 95% CI includes 1) included only belantamab mafodotin (β = 0.93, 95% CI 0.87–1.00) and pentosan polysulfate (β = 1.03, 95% CI 0.91–1.16). The former showed no significant time-dependence in risk, while the latter slightly exceeded one but the CI included 1, warranting cautious interpretation regarding long-term risk.
Discussion
4
By analyzing 1,242,518 reports of ocular injury AEs in the FAERS database from 2004 to 2024, this study delineates the epidemiological characteristics, high-risk drugs, and temporal risk patterns of drug-related ocular injury, providing crucial evidence-based insights for clinical medication safety. Compared to the recent study by Wu et al., which focused solely on ocular adverse reactions caused by systemic medications (Wu et al., 2025b), this study is the first to conduct a systematic panoramic analysis of all drug categories (including both topical and systemic agents) in FAERS. We not only confirmed the risks associated with some systemic drugs but also discovered that local ophthalmic agents (ATC: S class) exhibit the highest signal strength, thereby expanding our understanding of the risk spectrum for drug-induced ocular injury. Furthermore, through TTO analysis, this study quantified evidence of how risk changes over time, providing more refined guidance for clinical monitoring timing.
Patient demographics and risk factors
4.1
The study included a total of 832,314 patients, with a significantly higher proportion of females (60.56%) compared to males (30.18%). This finding is consistent with the results of multiple drug adverse reaction monitoring studies (Mao et al., 2024; Wu et al., 2024b). This phenomenon may be influenced by multiple factors. At the physiological level, sex differences can affect drug-metabolizing enzyme activity; females often exhibit higher plasma drug concentrations and prolonged elimination times, making them more susceptible to reaching toxicity thresholds (Zucker and Prendergast, 2020). Furthermore, the regulatory role of estrogen on the ocular surface may render female ocular tissues more sensitive to certain drugs, e.g., antihistamines, beta-blockers (Grasso et al., 2021). At the behavioral level, females generally demonstrate greater willingness and engagement in seeking healthcare, which may increase the likelihood of detecting and reporting adverse events (Watson et al., 2019).
In addition, the median age was 57.00 years (IQR 42.00–70.00), with elderly patients (≥65 years) comprising 21.63% of the cohort. Consistent with this, Wu et al. reported a mean age of 56.2 years (median: 60 years, IQR: 44–72) in drug-related keratitis patients, highlighting elevated risks in middle-aged and elderly populations (Wu et al., 2024b; Kozaru et al., 2023). These findings underscore age as a critical risk factor, likely driven by polypharmacy and age-related declines in hepatic/renal function, which alter drug metabolism in this group.
Sensory organ drugs: direct toxicity and mechanisms
4.2
Drugs for sensory organs (S class) exhibited the highest signal strength (ROR = 4.93), primarily due to their localized administration and pharmacological targets. As topical or intravitreal agents, they achieve high ocular tissue concentrations, bypassing systemic metabolism and directly exposing delicate ocular structures to drug effects (Koutsoviti et al., 2021). For example, the intravitreal injection drug brolucizumab (ROR = 132.15), used for wet age-related macular degeneration, showed high signal strength, necessitating vigilance for severe complications like retinal vasculitis and vitreous hemorrhage (Xiong et al., 2024). Similarly, topically applied prostaglandin analogs such as bimatoprost (ROR = 41.29) may induce conjunctival hyperemia, blepharitis, and eyelid lesions due to their long-term action on ocular surface tissues and impact on meibomian gland function (Manabe et al., 2020). The study by Williams et al. indicated that among patients using bimatoprost eye drops for long-term (4 years) treatment of ocular conditions, the incidence of blepharitis ranged from 2.6% to 5.1% (Williams et al., 2008). Furthermore, Smith et al. observed in a clinical study on eyelash growth that after daily application of 0.03% bimatoprost solution to the eyelid margin for 4 months, the incidence of adverse events such as eyelid lesions, conjunctival hyperemia, and dry eye all exceeded 2% (Smith et al., 2012). This heightened risk profile underscores that ocular ADRs predominantly stem from direct tissue exposure, the eye’s limited capacity for drug clearance, and immune privilege complexities. Vigilance remains critical, particularly for off-label uses altering drug-tissue contact.
Heterogeneity of antineoplastic and immunomodulating agents
4.3
Antineoplastic and immunomodulating agents (ATC: L class) had the highest number of reports (410,071 cases), yet the overall signal strength was low (ROR = 0.84). The high report count coupled with a low signal strength in ATC class L is a classic example of the dilution effect. This phenomenon arises from the massive patient population and the overshadowing of ocular symptoms by life-threatening systemic toxicities (e.g., myelosuppression) (Tanaka et al., 2022). Furthermore, the pharmacological heterogeneity within this class—ranging from highly specific ADCs to less toxic targeted therapies—further dilutes the average signal strength (Cui et al., 2025).
Traditional chemotherapeutic agents such as docetaxel (9,924 reports) exert cytotoxic effects by inhibiting tubulin polymerization, which may indiscriminately damage rapidly dividing epithelial cells on the ocular surface, leading to dry eye or corneal epithelial defects (Souza Monteiro de Araujo et al., 2022). In contrast, biologics such as the anti-CD38 antibody-drug conjugate (ADC) belantamab mafodotin (3,904 reports, ROR = 48.94) have a more specific toxicity mechanism: the drug binds to the CD38 antigen on corneal epithelial cells, internalizes, and releases its cytotoxic payload, resulting in the formation of corneal microcysts. This represents the ocular toxicity subtype with the strongest signal within ATC class L. (Mao et al., 2024; Farooq et al., 2020; Lu et al., 2023) This antigen-mediated targeted toxicity differs from the non-selective cytotoxicity of small-molecule drugs. Fingolimod (15,834 reports), another type of biologic acting as a sphingosine receptor modulator, may interfere with lymphocyte migration across the blood-ocular barrier, thereby inducing uveitis (Posarelli et al., 2011). It is also important to note that cancer patients often receive multi-drug combination therapy. Reported ocular damage may not be attributable to a single agent but rather the result of cumulative toxicity or interactions from multiple drugs, or it may be masked by the underlying disease or other more severe systemic toxicities. The spontaneous reporting nature of FAERS makes it challenging to control for this confounding effect of concomitant medications, highlighting the need for high-quality prospective studies to further explore this issue (Mao et al., 2024).
Dermatological drugs: local and systemic mechanisms
4.4
Dermatological drugs (ATC: D class) ranked third in this study with 232,719 reports and moderate signal strength (ROR = 1.92). Their association with ocular injury mainly stems from the dual mechanisms of local exposure and systemic immunomodulation. Botulinum toxin type A (14,551 cases) is commonly used to treat blepharospasm, hemifacial spasm, and various neuromuscular disorders of the head and neck. However, this treatment itself may cause blepharoptosis (drooping of the upper eyelid) and reduced blinking, leading to corneal epithelial defects and corneal ulcers (Nestor et al., 2021). Biologics such as dupilumab (63,490 cases), when used to treat atopic diseases, can precisely inhibit IL-4/IL-13 to improve symptoms such as asthma, but may cause ocular surface diseases (such as palpebral conjunctivitis, dry eye, etc.) and cytokine imbalance leading to abnormal function of ocular surface goblet cells due to immune response deviation (Bridgewood et al., 2022; Reji et al., 2022).
Common ocular injuries: clinical features and bias
4.5
Blurred vision (121,517 cases), visual impairment (113,320 cases), and cataract (51,826 cases) were the most common types of ocular injuries. Blurred vision and visual impairment, as the most common complaints of ADRs, may be because patients are more likely to perceive and report subjective experiences such as “inability to see clearly” rather than organic lesions requiring professional diagnosis (Karrer et al., 2019; Wu et al., 2025c). In contrast, cataract, as a clear clinical diagnosis, its high report volume reflects the direct association between drug exposure and organic damage. Long-term use of drugs such as glucocorticoids (e.g., prednisolone) and miotics (e.g., pilocarpine) can cause posterior subcapsular opacities by interfering with lens metabolism or inducing oxidative stress (Blomquist, 2011; Syed et al., 2021). These lesions require diagnosis by ophthalmic slit-lamp examination, so they are more likely to be actively reported by healthcare personnel. It is worth noting that the drug-relatedness of cataracts has a dose-time dependence, and regular ophthalmic screening should be implemented for long-term medication users (Al-Namaeh, 2024). This distinction between subjective symptoms and objective diagnoses is further highlighted by the composition of our data sources. Furthermore, consumer reports accounted for 50.04% of cases in this study, which may introduce reporting bias. Consumers tend to report easily perceptible subjective symptoms, whereas healthcare professionals are more likely to report based on objective diagnostic findings. This difference in reporting sources, along with factors such as patient education level and healthcare-seeking frequency, collectively influences the reporting frequency and patterns of different Preferred Terms. Future studies employing stratified analysis by reporter type could better elucidate this bias.
Time-to-onset patterns and monitoring
4.6
The TTO analysis revealed that most drugs exhibited an “early failure type” pattern (β < 1), such as dupilumab (β = 0.68) and ranibizumab (β = 0.63), indicating the highest risk within the first month post-initiation. This may relate to acute inflammatory responses or immune activation triggered by the drugs (Reji et al., 2022). Real-world studies have shown that ocular adverse events (such as conjunctivitis) associated with dupilumab in the treatment of atopic dermatitis primarily occur during the early treatment phase, especially within the first 4 months (Touhouche et al., 2021). Retrospective analysis revealed that 85.4% of severe intraocular inflammation events occurred within 5 days after ranibizumab injection (Souied et al., 2016). Only belantamab mafodotin (β = 0.93) and pentosan polysulfate (β = 1.03) showed a random failure pattern, suggesting sustained monitoring is required as risk lacks significant time-dependence (Farooq et al., 2020; Crivaro et al., 2019).
Study strengths and limitations
4.7
This study possesses two significant strengths. Firstly, in terms of data scale and methodology, it included over 1.2 million reports and employed four signal detection algorithms (ROR, PRR, BCPNN, MGPS) for cross-validation, ensuring the robustness of signals. Combining Weibull distribution analysis provided quantitative evidence of time-risk associations for pharmacovigilance. Secondly, regarding clinical guidance value, identifying high-risk categories (such as Sensory Organ drugs and biologics) through ATC classification and specific drug signals indicates that baseline ocular assessments and intensive early monitoring should be implemented for patients using these drugs.
However, the study also has some limitations. First, there are issues of data bias and incompleteness. As a spontaneous reporting system, FAERS has inherent limitations, including incomplete reporting and missing demographic information. In this study, gender was unspecified in 9.26% of cases, and age information was missing in 38.65% of cases. This may compromise an accurate assessment of risk characteristics in specific populations. Furthermore, time-to-event data were severely incomplete (73.65% missing TTO data), which limits the precise depiction of temporal patterns in drug-related ocular injuries, particularly hindering effective evaluation of late-onset risks associated with long-term medication use, such as cumulative chronic toxicity. Therefore, the TTO analysis based on the Weibull model in this paper primarily reflects the characteristics of cases with clearly documented time records, and its results, especially those concerning long-term risks, should be interpreted with caution. Future studies should integrate more comprehensive data sources to validate the findings. Second, the depth of mechanistic analysis is insufficient. Although high-risk drugs and temporal patterns were identified, this study lacks exploration into molecular mechanisms, such as blood-ocular barrier permeability and genetic susceptibility, which require further validation through basic experiments. Third, this study did not analyze the effects of dosage and combination therapy. Dosage information in FAERS reports is often recorded irregularly with a high missing rate, and it is impossible to accurately obtain patients' complete medication histories. Consequently, it is difficult to systematically analyze the impact of drug dosage, treatment duration, or concomitant medications on ocular injury, which may lead to the omission of important risk factors. Fourth, there is a limitation concerning the geographical representativeness of the data. The majority (62.91%) of the FAERS data utilized in this study originated from the United States. Consequently, the results may be influenced by local medication practices, healthcare reporting systems, and population genetic backgrounds. Therefore, caution should be exercised when extrapolating the conclusions to other ethnicities or regions. Future pharmacovigilance studies should integrate data from other regulatory agencies, such as those in the European Union and Japan, to verify the consistency of signals across different populations. Finally, constrained by the overall research objectives and the existing analytical framework of this study, stratified analyses based on reporter type, reporting country, or gender were not performed. We plan to prioritize stratified analysis in future research focusing on specific categories of drug-related ocular diseases, such as drug-related optic neuropathy.
Clinical risk management strategies
4.8
In clinical practice, several key strategies are recommended to mitigate risks. First, high-risk populations-including females, elderly patients, and those with underlying ocular diseases (e.g., diabetic retinopathy)-should undergo baseline ocular assessments before initiating relevant medications. Second, for drugs with “early failure type” patterns (e.g., dupilumab, ciclosporin), intensive monitoring of visual acuity, intraocular pressure, and fundus examinations during the first month of treatment is essential, while annual fundus screening should be incorporated for patients on long-term therapies (e.g., pentosan polysulfate). Lastly, patient education on proper topical eye drop administration is critical to minimize corneal irritation. Before initiating antineoplastic therapy, ocular contraindications should be assessed, and co-administration with other potentially eye-toxic drugs should be avoided.
Conclusion
5
This large-scale pharmacovigilance study delineates the real-world landscape of drug-related ocular injury, confirming the early treatment phase as the highest risk period and highlighting the need for focused attention on female and elderly patients. The findings provide evidence-based support for updating drug labeling, informing regulatory decisions, and guiding clinical monitoring strategies, ultimately contributing to enhanced medication safety. Future research should integrate mechanistic studies and prospective cohort analyses to further elucidate causal relationships in drug-related ocular toxicity, enabling more precise personalized pharmacotherapy.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Al-Namaeh M. (2024). Systemic medications and their ocular side effects. Cureus 16, e 74976. 10.7759/cureus.74976 39744265 PMC 11692015 · doi ↗ · pubmed ↗
- 2Alhawassi T. M. Krass I. Bajorek B. V. Pont L. G. (2014). A systematic review of the prevalence and risk factors for adverse drug reactions in the elderly in the acute care setting. Clin. Interv. Aging 9, 2079–2086. 10.2147/CIA.S 71178 25489239 PMC 4257024 · doi ↗ · pubmed ↗
- 3Ali A. Downes D. G. Eleiwa T. K. Hassan A. K. El Sheikh R. H. Chauhan M. Z. (2025). Drug-induced tubulointerstitial nephritis and uveitis syndrome in a nationwide surveillance. Eye (Lond) 39, 2036–2041. 10.1038/s 41433-025-03808-z 40240507 PMC 12209433 · doi ↗ · pubmed ↗
- 4Blomquist P. H. (2011). Ocular complications of systemic medications. Am. J. Med. Sci. 342, 62–69. 10.1097/MAJ.0b 013e 3181 f 06b 21 21139494 · doi ↗ · pubmed ↗
- 5Bridgewood C. Wittmann M. Macleod T. Watad A. Newton D. Bhan K. (2022). T helper 2 IL-4/IL-13 dual blockade with dupilumab is linked to some emergent T helper 17‒Type diseases, including seronegative arthritis and enthesitis/enthesopathy, but not to humoral autoimmune diseases. J. Invest Dermatol 142, 2660–2667. 10.1016/j.jid.2022.03.013 35395222 · doi ↗ · pubmed ↗
- 6Chen X. D. Xiao K. H. Zhou C. B. (2024). Drug-induced retinal vein occlusion: a disproportionality analysis from the FDA adverse event reporting system (2004-2023). Front. Pharmacol. 15, 1480269. 10.3389/fphar.2024.1480269 39734405 PMC 11671269 · doi ↗ · pubmed ↗
- 7Chen X. Wu S. Wang S. Yu C. Guo Z. Huang S. (2025). Real world pharmacovigilance assessment of drug related macular degeneration risks. Sci. Rep. 15, 1220. 10.1038/s 41598-024-84679-4 39774257 PMC 11707227 · doi ↗ · pubmed ↗
- 8Crivaro A. N. Mucci J. M. Bondar C. M. Ormazabal M. E. Ceci R. Simonaro C. (2019). Efficacy of pentosan polysulfate in in vitro models of lysosomal storage disorders: Fabry and Gaucher Disease. P Lo S One 14 (3752), e 0217780. 10.1371/journal.pone.0217780 31150494 PMC 6544267 · doi ↗ · pubmed ↗