This cohort study examines the risk for serous retinal detachment, retinal vascular occlusion, and ischemic optic neuropathy among men receiving phosphodiesterase type 5 inhibitors.
Key Points
Question
Is regular use of phosphodiesterase type 5 inhibitors (PDE5Is) associated with an increased risk for serous retinal detachment (SRD), retinal vascular occlusion (RVO), and ischemic optic neuropathy (ION) in older men?
Findings
This large cohort study of 213 033 men found an increase in the risk of SRD, RVO, and ION associated with the use of PDE5Is in those who regularly used these drugs.
Meaning
Results of this study suggest that individuals who regularly use PDE5Is should be cognizant of ocular adverse events associated with these drugs and alert their physicians if they experience any visual deficits.
Abstract
Importance
A number of case reports and small epidemiologic studies have quantified the risk of ocular adverse events associated with the use of phosphodiesterase type 5 inhibitors (PDE5Is). However, results have been conflicting, and epidemiologic data on the risk of serous retinal detachment (SRD) and retinal vascular occlusion (RVO) are not available.
Objective
To quantify the risk of SRD, RVO, and ischemic optic neuropathy (ION) associated with the use of PDE5Is.
Design, Setting, and Participants
This cohort study with a nested case-control analysis was performed using data obtained from the PharMetrics Plus database (IQVIA) from January 1, 2006, to December 31, 2020. Cohort members were followed up until the first diagnosis of SRD, RVO, or ION or termination of insurance coverage. For each case, 4 controls were matched by age and time of study entry using density-based sampling. Risk for regular users of PDE5Is was compared with that for nonusers, adjusting for potential confounding variables. Cases with diagnoses of SRD, RVO, and ION in the year before the cohort were excluded.
Main Outcomes and Measures
First diagnosis of SRD, RVO, or ION identified by International Classification of Diseases, Ninth Revision or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes. Adjusted incidence rate ratios (IRRs) with 95% CIs were calculated using conditional logistic regression, controlling for hypertension, coronary artery disease, smoking, and diabetes (for all 3 outcomes) as well as sleep apnea for the ION outcome.
Results
The cohort consisted of 213 033 men receiving PDE5Is, including sildenafil, tadalafil, vardenafil, and avanafil. The case-control analysis included a total of 1146 cases of SRD (278), RVO (628), and ION (240) and 4584 controls, and the mean (SD) age in both groups was 64.6 (13.3) years. Patients with SRD, RVO, and ION were more likely to have hypertension, diabetes, coronary artery disease, and sleep apnea. The adjusted IRR for the composite end points of any of the 3 outcomes was 1.85 (95% CI, 1.41-2.42; incidence, 15.5 cases per 10 000 person-years). The adjusted IRR for SRD, RVO, and ION as individual outcomes was 2.58 (95% CI, 1.55-4.30; incidence, 3.8 cases per 10 000 person-years), 1.44 (95% CI, 0.98-2.12; incidence, 8.5 cases per 10 000 person-years), and 2.02 (95% CI, 1.14-3.58; incidence, 3.2 cases per 10 000 person-years), respectively.
Conclusions and Relevance
Findings of this cohort study suggest that regular users of PDE5Is might have an increased risk for SRD, RVO, and ION. Regular users of PDE5Is need to be cognizant of ocular adverse events associated with these drugs and alert their physicians if they experience any visual deficits.
Introduction
Phosphodiesterase type 5 inhibitors (PDE5Is) are one of the most prescribed classes of medications in the US. Although sildenafil (Viagra) was the first drug introduced from this class, 3 other agents are currently available to men with erectile dysfunction, including vardenafil (Levitra), tadalafil (Cialis), and avanafil (Stendra).
Several ocular adverse events, including serous retinal detachment (SRD), retinal vascular occlusion (RVO), and ischemic optic neuropathy (ION), have been reported with PDE5I use.1 The evidence behind these adverse events is mostly in the form of case reports or small epidemiologic studies that produced imprecise estimates for these risks2,3; because of the rarity of these events, they could not be adequately studied in the original clinical trials of these drugs. A few case-crossover studies have reported an increased risk of ION associated with PDE5Is.4 However, by design, some of these studies could not quantify the incidence of ION, an important ocular adverse event that can lead to irreversible vision loss. Moreover, to our knowledge, no epidemiologic study to date has quantified the risk of SRD or RVO associated with PDE5I use. Thus, whether the observed cases of SRD or RVO with PDE5I use are potentially owing to confounding by disease severity (cardiovascular disease) or associated with PDE5I use is still unknown and must be further investigated because both these conditions can lead to serious visual deficits. Mechanistically, studies have found that PDE5Is may be associated with compromised perfusion of the optic nerve.5 Moreover, it is postulated that the choroid blood vessels can undergo smooth muscle relaxation through a cyclic guanosine monophosphate pathway that can lead to choroidal congestion.6
In 2020, there were approximately 20 million monthly prescriptions for PDE5Is in the US.7 Given the popularity of these drugs and the paucity of data on the magnitude of the risk of SRD, RVO, and ION, we sought to undertake a large cohort study to quantify the risk for SRD, RVO, and ION associated with use of PDE5Is.
Methods
The PharMetrics Plus database (IQVIA) was used as the main data source for this cohort study and the nested case-control analysis. This private health claims database captures health information for more than 150 million enrollees in the US.8 The data include enrollee age, prescription drug information (drug name, dose, day supply), and physician diagnosis according to the International Classification of Diseases, Ninth Revision (ICD-9) or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes. Data on race and ethnicity were not collected. We had access to a random sample of this database from January 1, 2006, to December 31, 2020, that included 16 424 887 enrollees with at least 1 year of enrollment. Approval for the study was obtained by the University of British Columbia’s Clinical Research Ethics Board, which waived the requirement for informed consent because we used anonymized data from a health claims database. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort and case-control studies.
Cohort Description
Using the data, we created a cohort of new users of PDE5Is, including sildenafil, tadalafil, vardenafil, and avanafil, who did not use any of the drugs in the year before study entry. We excluded all cases of SRD, RVO, and ION in the year before study entry. This look-back period was used to identify and adjust for covariate variables. These new users were followed up for the following outcomes: diagnosis of SRD, RVO, or ION, or termination of health coverage, whichever came first.
Case and Control Definition
Within the cohort, we identified new cases of SRD, RVO, and ION (SRD ICD-9 codes 361.2, 362.4, 362.41, 362.42 and ICD-10 codes H35.711, H35.712, H35.713, H33.23, H33.21, H33.22; RVO ICD-9 codes 362.30, 362.31, 362.35 and ICD-10 code H34; and ION ICD-9 code 377.41 and ICD-10 codes H47.01, H47.011, H47.012). We defined cases as a composite of any of the 3 outcomes and performed 3 separate analyses, examining each outcome separately. We first identified all risk sets (all patients with corresponding controls who had survived up to the time of the case, also referred to as the index date). From the potential pool of controls, we randomly identified 4 patients as controls who had entered the cohort at the same time of the case, were the same age, and had been followed up to the case’s index date. Controls could become future cases and could be used more than once (used in more than 1 risk set). This incidence density sampling manner of control selection has been found to generate odds ratios that closely approximate the incidence rate ratio (IRR) in a cohort study.9
Statistical Analysis
We used descriptive statistics to examine demographic characteristics between cases and controls. For each outcome, we defined regular users of PDE5Is as those who had at least 1 prescription for a PDE5I every 3 months in the year before the index date. This definition was used because PDE5Is are often used transiently, and ensuring regular use of these drugs would minimize exposure misclassification. We compared regular users of PDE5Is with nonusers, defined as those who had not received a PDE5I prescription in the year before the index date. We computed adjusted IRRs using conditional logistic regression, controlling for the following potential confounding variables: hypertension, coronary artery disease, smoking, and diabetes (for all 3 outcomes). For the outcome of ION, we also controlled for sleep apnea because this condition is a risk factor for ION10 and erectile dysfunction.11 As a sensitivity analysis, we restricted the analysis to cases and controls without a diagnosis of hypertension, diabetes, and coronary artery disease, 3 main potential confounders for the 3 outcomes. All analyses were done using SAS, version 9.4 (SAS Institute, Inc).
Results
The cohort included 213 033 PDE5I users, which included 123 347 sildenafil users, 78 609 tadalafil users, 6604 vardenafil users, and 4473 avanafil users. The case-control analysis included a total of 1146 cases of SRD (278), RVO (628), and ION (240) and 4584 corresponding controls, and the mean (SD) age in both groups was 64.6 (13.3) years (Table 1 and Table 2). Patients with SRD, RVO, and ION were more likely than those in the control group to have hypertension, diabetes, cardiovascular disease, or sleep apnea (Table 1). The adjusted IRR for the composite end points of any of the 3 outcomes was 1.85 (95% CI, 1.41-2.42; incidence, 15.5 cases per 10 000 person-years). The adjusted IRR for SRD as individual outcome was 2.58 (95% CI, 1.55-4.30), and the incidence was 3.8 cases per 10 000 person-years. For RVO, the adjusted IRR was 1.44 (95% CI, 0.98-2.12), and the incidence was 8.5 cases per 10 000 person-years. The adjusted IRR for ION was 2.02 (95% CI, 1.14-3.58), and the incidence was 3.2 cases per 10 000 person-years. Both crude and adjusted IRRs for all 3 outcomes were similar in magnitude, signifying minimal confounding by measured confounders. When we restricted the primary analysis (with all 3 outcomes) to cases without hypertension, diabetes, or coronary artery disease, the IRR remained elevated at 2.12 (95% CI, 1.34-3.43).
Table 1. Characteristics of Cases and Controls for Serous Retinal Detachment, Retinal Vascular Occlusion, and Ischemic Optic Neuropathy.
| Characteristic | No. (%) | |
|---|---|---|
| Cases (n = 1146) | Controls (n = 4584) | |
| Age, mean (SD), y | 64.6 (13.3) | 64.6 (13.3) |
| Follow-up, mean (SD), y | 3.8 (3.5) | 3.8 (3.5) |
| Hypertension | 282 (24.6) | 408 (8.9) |
| Smoking | 112 (9.8) | 426 (9.3) |
| Diabetes | 437 (38.1) | 1196 (26.1) |
| Coronary artery diseases | 414 (36.1) | 1100 (24.0) |
| Sleep apnea | 178 (15.5) | 486 (10.6) |
Table 2. Crude and Adjusted Incidence Rate Ratios of Cases of Composite, SRD, RVO, and ION With Regular Use of PDE5I.
| Outcome | No. (%) | Incidence rate ratio (95% CI) | |
|---|---|---|---|
| Cases | Controls | ||
| Composite outcomes of SRD, RVO, and IONa,b | |||
| No. of individuals | 1146 | 4584 | |
| No use | 566 (49.4) | 2516 (54.9) | 1 [Reference] |
| Regular use | |||
| Crude | 256 (22.3) | 907 (19.8) | 1.88 (1.45-2.45) |
| Adjustedc | 256 (22.3) | 907 (19.8) | 1.85 (1.41-2.42) |
| SRDd | |||
| No. of individuals | 278 | 1112 | |
| No use | 83 (29.9) | 456 (41.0) | 1 [Reference] |
| Regular use | |||
| Crude | 84 (30.2) | 301 (27.1) | 2.60 (1.57-4.31) |
| Adjustedb | 84 (30.2) | 301 (27.1) | 2.58 (1.55-4.30) |
| RVOe | |||
| No. of individuals | 628 | 2512 | |
| No use | 367 (58.4) | 1542 (61.4) | 1 [Reference] |
| Regular use | |||
| Crude | 116 (18.5) | 434 (17.3) | 1.44 (0.99-2.10) |
| Adjustedf | 116 (18.5) | 434 (17.3) | 1.44 (0.98-2.12) |
| IONg | |||
| No. of individuals | 240 | 960 | |
| No use | 115 (47.9) | 518 (54.0) | 1 [Reference] |
| Regular use | |||
| Crude | 55 (22.9) | 172 (17.9) | 2.33 (1.36-3.98) |
| Adjustedc | 55 (22.9) | 172 (17.9) | 2.02 (1.14-3.58) |
Abbreviations: ION, ischemic optic neuropathy; PDE5I, phosphodiesterase type 5 inhibitor; RVO, retinal vascular occlusion; SRD, serous retinal detachment.
Composite outcomes indicates SRD or RVO or ION combined.
Incidence, 15.5 cases per 10 000 person-years.
Adjusted for covariates in Table 1.
Incidence, 3.8 cases per 10 000 person-years.
Incidence, 8.5 cases per 10 000 person-years.
Adjusted for all covariates in Table 1. Sleep apnea adjusted only for ION.
Incidence, 3.2 cases per 10 000 person-years.
A dose-response analysis comparing the risk among those taking 5 or more prescriptions compared with those taking fewer than 5 prescriptions showed a dose-response association for the overall analysis (IRR, 2.90; 95% CI, 1.15-3.81 vs IRR, 1.74; 95% CI, 1.10-6.77) (Table 3). The dose-response analysis of individual outcomes for those taking more than 5 prescriptions vs fewer than 5 prescriptions were IRRs of 1.90 (95% CI, 1.41-2.55) vs 1.73 (95% CI, 1.14-2.64) for SRD, 2.39 (95% CI, 1.38-4.14) vs 3.30 (95% CI, 1.48-7.38) for RVO, and 1.55 (95% CI, 1.00-2.40) vs 1.25 (95% CI, 0.70-2.21) for ION.
Table 3. Dose-Response Analysis for the Composite Analysis and Individual Outcomes of SRD, RVO, and ION Associated With PDE5I Use.
| Outcome and No. of PDE5I prescriptions | Adjusted IRR (95% CI)a |
|---|---|
| Overall | |
| <5 | 1.74 (1.10-6.77) |
| ≥5 | 2.90 (1.15-3.81) |
| SRD | |
| <5 | 1.73 (1.14-2.64) |
| ≥5 | 1.90 (1.41-2.55) |
| RVO | |
| <5 | 3.30 (1.48-7.38) |
| ≥5 | 2.39 (1.38-4.14) |
| ION | |
| <5 | 1.25 (0.70-2.21) |
| ≥5 | 1.55 (1.00-2.40) |
Abbreviations: ION, ischemic optic neuropathy; IRR, incidence rate ratio; PDE5I, phosphodiesterase type 5 inhibitor; RVO, retinal vascular occlusion; SRD, serous retinal detachment.
Compared with nonuse of PDE5Is.
Discussion
To our knowledge, this is the largest epidemiologic cohort study that has examined 3 important ocular outcomes associated with use of PDE5Is, one of the most prescribed classes of medications. We found a combined increase in the risk of SRD, RVO, and ION associated with the use of PDE5Is in older men. For each individual outcome, use of PDE5Is was independently associated with an increase in risk. The risk for RVO was elevated but less precise, meaning wider CIs, possibly owing to a smaller number of cases.
To date, data on the 3 ocular outcomes associated with PDE5I use have been mostly published in case reports or case series, which on their own cannot show a causal association. A small epidemiologic study also found an association between PDE5I use and the risk for ION.3 A warning for the risk of ION is currently listed in the monographs of sildenafil,12 tadalafil,13 vardenafil,14 and avanafil.15 Retinal vascular occlusion is stated as a potential adverse event without mentioning its frequency for sildenafil,12 tadalafil,13 and vardenafil14 but is not stated for avanafil.15 There is no mention of the specific risk of SRD with any of the 4 drugs.
There is a paucity of evidence regarding the pathogenesis of PDE5I-induced SRD, RVO, and ION. With both RVO and ION, it is postulated that they might be caused by systemic hypotension associated with PDE5I use, leading to reduced perfusion and thus a hypercoagulable state and areas of blood stasis.16 In addition, because PDE5Is are usually taken at night, they can further accentuate normal physiologic nocturnal hypotension, which may lead to ischemia.5 It is believed that PDE5Is also may decrease the perfusion of the optic nerve head.5 Another hypothesis is that the role of PDE5Is in increasing choroidal blood flow, ischemia, and congestion of the retinal blood vessels may precipitate an SRD; however, further research is needed to elucidate these mechanisms.17
Strengths and Limitations
This study has strengths. The main strength is the large sample size that allowed risk quantification for 3 ocular outcomes. Moreover, we made every attempt to mitigate confounding by disease severity using a cohort composed of new PDE5I users as well as restriction, matching, and statistical adjustment. Defining regular use of PDE5I also ensured continuity of use and minimized exposure misclassification given the transient nature of these drugs. Another strength of our study is risk assessment for the 3 outcomes, which minimizes confounding by severity of erectile dysfunction because a homogenous cohort of all PDE5I users was analyzed.
This study also has limitations. The main limitation was that we had data only on drug dispensation and could not verify actual intake. Also, the outcome definitions were captured using ICD-9 and ICD-10 coding because we did not have access to patients’ medical records. Moreover, because of the small number of events, we could not examine the risk for the study outcomes associated with each individual PDE5I separately. In addition, as with all pharmacoepidemiologic studies that use health claims data, there is a possibility of measurement error for the covariates adjusted for. However, this is unlikely to alter the study results because there was minimal difference between crude and adjusted effect sizes.
Conclusions
The results of this cohort study confirm the findings from other studies regarding the risk of ION associated with PDE5I use but also quantifies the risk of SRD and RVO associated with use of these drugs, which has not been previously assessed. The findings suggest individuals who regularly use PDE5Is need to be cognizant of ocular adverse events associated with these drugs and alert their physicians if they experience any visual deficits. Future studies, especially those that can adequately address risk of these outcomes associated with different PDE5Is, are needed to confirm these findings.
References
- 1.Li AS, Pomeranz HD. Food and Drug Administration adverse event reports of retinal vascular occlusions associated with phosphodiesterase type 5 inhibitor use. J Neuroophthalmol. 2016;36(4):480-481. doi: 10.1097/WNO.0000000000000450 [DOI] [PubMed] [Google Scholar]
- 2.Margo CE, French DD. Ischemic optic neuropathy in male veterans prescribed phosphodiesterase-5 inhibitors. Am J Ophthalmol. 2007;143(3):538-539. doi: 10.1016/j.ajo.2006.10.006 [DOI] [PubMed] [Google Scholar]
- 3.Nathoo NA, Etminan M, Mikelberg FS. Association between phosphodiesterase-5 inhibitors and nonarteritic anterior ischemic optic neuropathy. J Neuroophthalmol. 2015;35(1):12-15. doi: 10.1097/WNO.0000000000000186 [DOI] [PubMed] [Google Scholar]
- 4.Campbell UB, Walker AM, Gaffney M, et al. Acute nonarteritic anterior ischemic optic neuropathy and exposure to phosphodiesterase type 5 inhibitors. J Sex Med. 2015;12(1):139-151. doi: 10.1111/jsm.12726 [DOI] [PubMed] [Google Scholar]
- 5.Kerr NM, Danesh-Meyer HV. Phosphodiesterase inhibitors and the eye. Clin Exp Ophthalmol. 2009;37(5):514-523. doi: 10.1111/j.1442-9071.2009.02070.x [DOI] [PubMed] [Google Scholar]
- 6.Breazzano MP, Coleman DJ, Chen RWS, Chang S, Daly S, Tsang SH. Prospective impact of sildenafil on chronic central serous chorioretinopathy: PISCES trial. Ophthalmol Retina. 2020;4(11):1119-1123. doi: 10.1016/j.oret.2020.06.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hernandez I, Gul Z, Gellad WF, Davies BJ. Marked increase in sales of erectile dysfunction medication during COVID-19. J Gen Intern Med. 2021;36(9):2912-2914. doi: 10.1007/s11606-021-06968-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.IQVIA PharMectrics Plus . Fact sheet. IQVIA; 2020. Accessed November 7, 2021. https://www.greygreenmedia.com/app/uploads/sites/2/2020/09/IQVIA-PharMetrics-Plus-US-Fact-Sheet.pdf
- 9.Richardson DB. An incidence density sampling program for nested case-control analyses. Occup Environ Med. 2004;61(12):e59. doi: 10.1136/oem.2004.014472 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Mojon DS, Hedges TR III, Ehrenberg B, et al. Association between sleep apnea syndrome and nonarteritic anterior ischemic optic neuropathy. Arch Ophthalmol. 2002;120(5):601-605. doi: 10.1001/archopht.120.5.601 [DOI] [PubMed] [Google Scholar]
- 11.Chen CM, Tsai MJ, Wei PJ, et al. Erectile dysfunction in patients with sleep apnea—a nationwide population-based study. PLoS One. 2015;10(7):e0132510. doi: 10.1371/journal.pone.0132510 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sildenafil. Product monograph. Pharmascience Inc; 2018. Accessed February 7, 2022. https://pdf.hres.ca/dpd_pm/00048380.PDF
- 13.Tadalafil. Product monograph. Eli Lilly Canada Inc; 2020. Accessed February 7, 2022. https://pi.lilly.com/ca/cialis-ca-pm.pdf
- 14.Vardenafil. Product monograph. Bayer Inc; 2020. Accessed February 7, 2022. https://pdf.hres.ca/dpd_pm/00055199.PDF
- 15.Avanafil. Product monograph. VIVUS, Inc; 2012. Accessed February 7, 2022. https://www.accessdata.fda.gov/drugsatfda_docs/label/2012/202276s000lbl.pdf
- 16.Pinto LM, Morekar S, Mahashur AA. Central retinal vein occlusion in a patient after being commenced on sildenafil citrate for pulmonary arterial hypertension. Indian J Chest Dis Allied Sci. 2009;51(4):249-251. [PubMed] [Google Scholar]
- 17.Marmor MF, Kessler R. Sildenafil (Viagra) and ophthalmology. Surv Ophthalmol. 1999;44(2):153-162. doi: 10.1016/S0039-6257(99)00079-X [DOI] [PubMed] [Google Scholar]