Advertisement

Cardiovascular Outcome Risks in Patients With Erectile Dysfunction Co-Prescribed a Phosphodiesterase Type 5 Inhibitor (PDE5i) and a Nitrate: A Retrospective Observational Study Using Electronic Health Record Data in the United States

Open AccessPublished:August 10, 2021DOI:https://doi.org/10.1016/j.jsxm.2021.06.010

      ABSTRACT

      Background

      Phosphodiesterase type 5 inhibitors (PDE5i) are first-line therapy for erectile dysfunction (ED). Approximately 1–4% of PDE5i recipients co-possess nitrates, despite this combination potentially producing clinically significant hypotension. Real-world data in these patients and insights into prescriber rationales for co-prescription are limited.

      Aim

      This study investigated whether PDE5i and nitrate co-possession is associated with increased rates of cardiovascular (CV) outcomes.

      Methods

      Adult males with ED and PDE5i prescription and males with nitrate prescription were identified from a U.S. electronic health record database (2012–2016). Quantitative comparisons were made between patients with ED and co-possession (ED + PDE5i + nitrate), only nitrate possession (ED + nitrate and nitrate only [without ED]), and only PDE5i possession (ED + PDE5i).

      Outcomes

      We quantified incidence of CV outcomes in co-possession and comparator periods, calculating incidence rate ratios after propensity score matching. Prescriber rationales were derived by reviewing virtual patient records.

      Results

      Over 168,000 patients had ≥1 PDE5i prescription (∼241,000 possession periods); >480,000 patients had ≥1 nitrate prescription (∼486,000 possession periods); and 3,167 patients had 3,668 co-possession periods. Non-significantly different or lower rates of CV outcomes were observed for co-possession periods vs ED + nitrate and nitrate only periods. Most CV outcome rates were non-significantly different between co-possession and ED + PDE5i periods (myocardial infarction, hospitalized unstable angina and fainting were higher with co-possession). From qualitative assessment of patient records with co-possession, 131 of 252 (52%) documented discussion with a physician regarding co-possession; 69 of 131 (53%) warned or instructed on safely managing these contraindicated medications.

      Clinical Implications

      Findings from this real-world study indicate that co-possession of nitrate and PDE5i prescriptions is not associated with increased rates of CV outcomes, relative to possession of nitrates alone. Physicians should and often do discuss the risks of using both medications together with their patients.

      Strengths & Limitations

      Strengths of this study are the large size of the U.S. real-world patient cohort with data available for analysis, and our ability to utilize natural language processing to explore co-prescription rationales and patient-physician interactions. Limitations are the retrospective nature of the analysis and inability to establish whether recorded prescriptions were filled or the medication was consumed.

      Conclusion

      Co-exposure of PDE5i and nitrates should continue to be avoided; however, co-possession of PDE5i and nitrate prescriptions is not necessarily associated with increased CV risk. Co-possession can be successfully managed in suitable circumstances. Nunes AP, Seeger JD, Stewart A, et al. Cardiovascular Outcome Risks in Patients With Erectile Dysfunction Co-Prescribed a Phosphodiesterase Type 5 Inhibitor (PDE5i) and a Nitrate: A Retrospective Observational Study Using Electronic Health Record Data in the United States. J Sex Med 2021;18:1511–1523.

      Key Words

      INTRODUCTION

      Erectile dysfunction (ED) is a common condition which is strongly positively correlated with age.
      • Saigal CS
      • Wessells H
      • Pace J
      • et al.
      Urologic diseases in America project. Predictors and prevalence of erectile dysfunction in a racially diverse population.
      As ED is predominantly a vascular disease and may even be an indicator for occult cardiovascular disease (CVD),
      • Solomon H
      • Man JW
      • Jackson G.
      Erectile dysfunction and the cardiovascular patient: Endothelial dysfunction is the common denominator.
      the cardiovascular (CV) conditions that nitrates are used to treat are frequently coincident with ED and often have shared etiology.
      • Nehra A.
      Erectile dysfunction and cardiovascular disease: Efficacy and safety of phosphodiesterase type 5 inhibitors in men with both conditions.
      Endothelial dysfunction is seen as the pathophysiological link between ED and CVD,
      • Solomon H
      • Man JW
      • Jackson G.
      Erectile dysfunction and the cardiovascular patient: Endothelial dysfunction is the common denominator.
      and ED has been suggested as a marker of CVD.
      • Diaconu CC
      • Manea M
      • Marcu DR
      • et al.
      The erectile dysfunction as a marker of cardiovascular disease: A review.
      Estimates from cross-sectional studies based on self-assessment of ED suggest that approximately 20% of men in the United States (U.S.) aged ≥20 years are affected.
      • Saigal CS
      • Wessells H
      • Pace J
      • et al.
      Urologic diseases in America project. Predictors and prevalence of erectile dysfunction in a racially diverse population.
      ,
      • Selvin E
      • Burnett AL
      • Platz EA.
      Prevalence and risk factors for erectile dysfunction in the US.
      Phosphodiesterase type 5 inhibitors (PDE5i) are the recommended first-line therapy for ED of all etiologies and severities.
      • Nehra A.
      Erectile dysfunction and cardiovascular disease: Efficacy and safety of phosphodiesterase type 5 inhibitors in men with both conditions.
      Male erectile function requires the interaction of psychological, hormonal, vascular, and neurological factors.
      • Corbin JD.
      Mechanisms of action of PDE5 inhibition in erectile dysfunction.
      ,
      • Sangiorgi G
      • Cereda A
      • Benedetto D
      • et al.
      Anatomy, pathophysiology, molecular mechanisms, and clinical management of erectile dysfunction in patients affected by coronary artery disease: A review.
      Spinal reflex and the L-arginine-nitric oxide (NO)-guanylyl cyclase-cyclic guanosine monophosphate (cGMP) pathway mediate smooth muscle relaxation that results in penile erection upon sexual stimulation. NO is produced from oxygen and L-arginine under the control of nitric oxide synthase (NOS). Sexual arousal stimulates neural pathways that result in the release of NO from nerves and endothelial cells directly into the penis. NO penetrates the cytoplasm of smooth muscle cells and binds to guanylyl cyclase. The interaction of NO with guanylyl cyclase causes a conformational change in this enzyme, which results in the catalytic production of 3′-5′-cGMP from guanosine 5′-triphosphate. cGMP is the intracellular trigger for penile erection by activating cGMP-dependent protein kinase, which in turn phosphorylates several proteins. These protein kinase interactions result in reduced intracellular calcium levels and a consequent relaxation of arterial and trabecular smooth muscle, leading to arterial dilatation, venous constriction, and the rigidity of penile erection. PDE5i inhibitors lower the activity of PDE5 by competing with cGMP thereby raising the level of cGMP. PDE5 inhibition is ineffective in the absence of stimulation of the NO pathway, therefore sexual arousal which stimulates this pathway in the penis is required.
      An established pharmacodynamic interaction between PDE5i and organic nitrates (any form) could potentially lead to clinically significant hypotension.
      • Schwartz BG
      • Kloner RA.
      Drug interactions with phosphodiesterase-5 inhibitors used for the treatment of erectile dysfunction or pulmonary hypertension.
      This is because both PDE5i and organic nitrates lead to increases in cGMP and relax smooth muscles.
      • Schwartz BG
      • Kloner RA.
      Drug interactions with phosphodiesterase-5 inhibitors used for the treatment of erectile dysfunction or pulmonary hypertension.
      Since they do so through different mechanisms, co-administration substantially increases cGMP levels and can result in pronounced systemic vasodilation and an augmented hypotensive response.
      • Kloner RA
      • Goggin P
      • Goldstein I
      • et al.
      A new perspective on the nitrate-phosphodiesterase type 5 inhibitor interaction.
      The magnitude and duration of this potentiation of hypotensive effect with co-administration of PDE5i and organic nitrates has been shown through results of randomized controlled trials in healthy men and men with stable angina.
      • Webb DJ
      • Freestone S
      • Allen MJ
      • et al.
      Sildenafil citrate and blood-pressure-lowering drugs: Results of drug interaction studies with an organic nitrate and a calcium antagonist.
      • Webb DJ
      • Muirhead GJ
      • Wulff M
      • et al.
      Sildenafil citrate potentiates the hypotensive effects of nitric oxide donor drugs in male patients with stable angina.
      • Kloner RA
      • Hutter AM
      • Emmick JT
      • et al.
      Time course of the interaction between tadalafil and nitrates.
      • Oliver JJ
      • Kerr DM
      • Webb DJ.
      Time-dependent interactions of the hypotensive effects of sildenafil citrate and sublingual glyceryl trinitrate.
      • Swearingen D
      • Nehra A
      • Morelos S
      • et al.
      Hemodynamic effect of avanafil and glyceryl trinitrate coadministration.
      Accordingly, concomitant use of organic nitrates is a labeled contraindication across the class of all PDE5is. However, despite the absolute contraindication for concomitant use of these medications, it is known that some ED patients who may be cardiovascularly stable and may be eligible for PDE5i are co-prescribed or co-possess drugs of these classes. A previous longitudinal prescription database study found that 1.4–2.1% of U.S. patients prescribed PDE5i receive concurrent nitrates.
      • Chang LL
      • Ma M
      • von Allmen H
      • et al.
      Co-possession of phosphodiesterase type-5 inhibitors (PDE5-I) with nitrates.
      The authors noted that co-possession of the 2 drug types was largely a result of prescribing by different providers, but that co-prescribing by the same physician was highest among cardiologists. A more recent study
      • Huri HZ
      • Ling CF
      • Razack AH.
      Drug-related problems in patients with erectile dysfunctions and multiple comorbidities.
      found similar results to Chang et al
      • Chang LL
      • Ma M
      • von Allmen H
      • et al.
      Co-possession of phosphodiesterase type-5 inhibitors (PDE5-I) with nitrates.
      of co-prescribing of PDE5i and nitrates: the percentage of nitrates co-prescribed with sildenafil (Viagra), vardenafil (Levitra), and tadalafil (Cialis) was 4.1% (n = 320), 1.07% (n = 93), and 1.25% (n = 80), respectively.
      Since these previous studies have highlighted that co-possession in current prescribing practice is rare, and patients with nitrate prescription(s) were excluded in original trials, it follows that little is known about the clinical profile of these patients. Furthermore, the effect of co-possession on the occurrence of hypotensive/CV outcomes, or the physician's rationale for co-prescribing of these medications, including what was discussed with the patient regarding the use of the 2 medications, merits further investigation. Typically, adverse effects of medications that do not result in medical services or are not otherwise coded in an electronic medical record (EMR)/claim have traditionally been difficult to study using real-world data. Use of a generalized natural language processing (NLP) system to extract and organize concepts from free text into semistructured fields allows identification of documented adverse events, including patient symptoms, that may not receive diagnostic codes.
      Using data derived from EMRs for a large cohort of patients from across the United States, the primary objective of the study was to quantitatively evaluate whether periods of PDE5i and nitrate prescription co-possession by patients with ED are associated with an increased incidence of hypotensive/CV outcomes relative to matched control cohorts. A qualitative analysis was also conducted to examine physician's rationale for co-prescription of a nitrate and a PDE5i among patients with ED.

      MATERIALS AND METHODS

       Study Design

      This was a retrospective, observational, cohort study utilizing electronic health record (EHR) data to identify male patients in the United States with PDE5i or nitrate prescriptions or administrations from January 1, 2012 to December 31, 2016. The EHR database is sourced from EMRs from over 92,000 providers and 195 hospitals, representing over 46 EMR-based provider/hospital networks across the United States. Patients may be followed between networks and over time. The primary aim was to investigate whether males with a diagnosis of ED and co-possession of PDE5i and nitrate medications have an increased incidence of hypotensive/CV outcomes compared with possession of a nitrate alone (with or without ED). Prescriptions of PDE5is and nitrates were identified from structured data fields corresponding to written prescriptions and medication orders. Because PDE5i and nitrate medications may be used intermittently, we define exposure as medication possession periods anchored to the prescription/order dates.
      Each medication possession period was preceded by a 12-month baseline period. The follow-up period, during which hypotensive/CV outcomes were assessed, commenced at the start of medication possession (prescription start or medication administration) and continued until the earliest of: the end of possession for the cohort entry medication (PDE5i and/or nitrate), exit from the observation period, or the end of the study. To account for changes in medication exposure over time for an individual patient, “person-time” was determined for each treatment period separately, with person-time subsequently pooled by exposure category (ie, PDE5i and nitrate, nitrate only, or PDE5i only). The incidence of hypotensive/CV outcomes was compared between the co-possession and control periods by calculating incidence rate ratios (IRRs) after propensity score matching. Qualitative analysis using NLP of electronic free-text clinical notes was used to assess the rationale for co-prescription of PDE5i and nitrate.

       Data Source

      Optum's Pan-Therapeutic (PanTher) EHR database was used in this study. It is a patient-level research database that integrates multiple EMR data systems with medical claims, prescription, and practice management data. The database incorporates clinical data recorded during routine clinical practice, both for outpatients and inpatients. Data include medical records, laboratory results, and drug prescription (including number of refills) and administration data. The database contains EHR data of more than 70 million male and female patients from 2007 to 2016. The database also includes data derived from free-text clinical notes relating to medical encounters, from which a generalized NLP system is used to extract and organize concepts into semistructured fields. NLP-based data are available for over 40 million patients.

       Study Cohorts

      Exposure cohorts were developed based on patients’ periods of possession of the different medications, PDE5i and nitrate. The exposed cohort was defined as the period of time in which males with diagnosed ED had concomitant possession of PDE5i and nitrate medications; the “ED + PDE5i + nitrate” (co-possession) cohort. The comparator control cohorts for the primary analysis included: (i) ED + nitrate cohort: the period of time in which males with diagnosed ED had possession of a nitrate prescription or administration for a nitrate (without a PDE5i prescription); and (ii) Nitrate only cohort: the period of time in which males without an ED diagnosis had possession of a nitrate prescription or administration for a nitrate (without a PDE5i prescription). As a secondary analysis, a third comparator was identified: (iii) ED + PDE5i cohort: the period of time in which males with diagnosed ED had possession of a PDE5i prescription (without a nitrate prescription). This third comparator cohort, while of interest, is potentially prone to an intractable confounding by indication due to nitrates being prescribed for CV conditions that are associated with our CV outcomes of interest.
      The medication possession periods used to develop the described exposure cohorts represent the period of time during which a patient had a PDE5i and/or nitrate available (prescription/administration). As possession periods may have differed between patients with the same prescription, possession periods were empirically derived from days supplied by the prescription and the number of refills. Patients may have had multiple periods of PDE5i and/or nitrate possession and, therefore, contributed to multiple exposure groups.

       Cohort Eligibility Criteria

      Eligible patients were males over the age of 18 years with a PDE5i or nitrate prescription or administration with ≥1 year of encounter data within the EHR derived from a medical group contributing electronic free-text clinical notes. The first cohort entry date for a patient was the earliest PDE5i possession period and/or earliest nitrate possession period meeting the following criteria: (i) ≥1 year of medical encounter data prior to the prescription date; (ii) ≥1 medical encounter resulting in a clinical note within the prior 365 days; (iii) ≥1 outpatient encounter with an evaluation and management code within the previous 365 days; and (iv) ≥1 diagnosis of ED within the previous 365 days (for PDE5i initiation). Periods of medication possession following the cohort entry date were identified and characterized as PDE5i and nitrate co-possession, nitrate only, PDE5i only, or uncertain possession of either medication.

       Covariates

      Covariates were identified within the 12-month baseline period prior to each individual medication possession period and included demographics (eg, age, race/ethnicity), medical utilization (eg, insurance status, medical encounters), behavioral characteristics (alcohol and tobacco use), and comorbidities specified a priori (eg, hypotensive events, hypertension, peripheral vascular disease, angina). A complete list of covariates considered in the analyses is provided in Supplemental Table I. Additional comorbidities, procedures, and medications were assessed empirically from listings of the most common baseline diagnostic, procedural, and drug codes.

       Outcomes

      Hypotensive/CV outcomes were identified by diagnostic codes and NLP algorithms. Outcomes identified via diagnostic coding were: hypotension, syncope, myocardial infarction (MI), hospitalized unstable angina (or angina/chest pain), stroke and transient ischemic attack, ventricular arrhythmia/cardiac arrest, and death (CV-related and any cause). CV-related death was defined as a death with a diagnosis of MI, stroke, or sudden death on the date of death or within the preceding 30 days. A complete list of relevant diagnostic codes associated with these outcomes is provided in Supplemental Table II. Outcomes identified via NLP of the free-text clinical notes were: dizziness, fainting, hypotension, presyncope, and syncope. A list of medical terms associated with these outcomes is provided in Supplemental Table III.
      Hypotensive/CV outcomes, both code-based and derived from NLP, were identified from the day following the start of a medication possession period to the earliest of: (i) the date indicating exit from data capture; (ii) the end of the possession period; or (iii) the end of the study period. The first event within the possession period was treated as a censoring event separately for each type of hypotensive/CV outcome, and thus only one such event was counted per possession period. Outcomes were attributed to the exposure grouping at the time at which the outcome occurred. When quantifying incidence rates and IRRs, person-time was censored at the occurrence of the first outcome within a treatment period, separately for each type of outcome.
      The NLP algorithm included identification of mentions of outcomes, characterization of mentions as either supporting, neutral, or detracting, and characterization of sentiments and attributes. The validity of NLP algorithms was assessed by manual adjudication of a sample of the free-text notes, following which the algorithm criteria were evaluated and adapted to maximize the probability that the symptom/outcome identified in the notes was truly present. Further details of the NLP algorithm are provided in the Supplemental Methods.

       Data Analysis

       Descriptive Analyses

      Patient characteristics were summarized both pre- and postpropensity score matching for each of the cohorts. Descriptive statistics are presented for categorical variables as frequency and percentage, and for continuous variables as mean and standard deviation, or median and interquartile range.

       Propensity Score Matching

      Propensity score matching was performed to address the imbalance in treatment and patient characteristics that is inherent when comparing treatment groups using data from the real-world setting. Specific covariates encompassing demographics, medical utilization, behavioral characteristics, comorbidities, and medications (Supplemental Table I) were identified and used to develop a propensity score as a predicted probability of entering a possession period with ED + PDE5i + nitrate relative to a possession period with comparator exposure. Matching was performed in a 1:1 ratio using separate propensity score models for each of the separate comparison groups, such that each patient medication possession period in the ED + PDE5i + nitrate (co-possession) cohort was matched to an equivalent patient medication possession period in each of the 3 control cohorts (ED + nitrate, nitrate only, and ED + PDE5i).

       Incidence Rates and Incidence Rate Ratios

      For each hypotensive/CV outcome of interest, the prematched incidence rate and IRR were first calculated from all periods in the ED + PDE5i + nitrate cohort relative to each control cohort. Thereafter, matched incidence rates and IRR for cohort comparisons were calculated from propensity score matched periods, with a final adjustment using a Poisson regression model. These final adjusted models included adjustment for age category and Charlson comorbidity index, and any covariates with P < .1 in the initial full model. Results presented herein are those of the matched adjusted models used to calculate IRRs relative to the matched control cohorts.

       Qualitative Analyses

      Qualitative summaries were generated from manual review of individual patient's virtual EHRs, which are large data tables that provide a chronological listing of relevant EHR information. These include both structured data and free-text summaries of the clinical notes derived from the use of a generalized NLP system. Patients were evaluated through this approach to identify whether there was a recorded discussion between the physician and patient during each period of co-possession of PDE5i and nitrate. Where this was the case, it was noted whether a recommendation to discontinue use of one or other of the medications was recorded. Patients were further categorized by occurrence or not of prior cardiac events.

      RESULTS

       Patient Disposition and Treatment Period Selection

      Approximately 1.8 million male patients satisfied the general eligibility criteria (Figure 1). After applying further eligibility criteria, 168,131 patients had ED and ≥1 PDE5i prescription, equating to 241,706 medication possession periods for the ED + PDE5i cohort, while 480,016 patients had ≥1 nitrate prescription (regardless of ED status), equating to 486,004 possession periods for the nitrate only cohort. A total of 3,167 patients had ED and ≥1 prescription for a PDE5i and a nitrate, equating to 3,668 medication possession periods for the ED + PDE5i + nitrate cohort. Table 1 provides nitrate prescription characteristics among PDE5i initiators (no PDE5i prescription preceding the eligible co-possession period). Approximately half of patients initiated tadalafil and just over a third initiated sildenafil. Overlapping possession of prescriptions for PDE5i and a nitrate occurred for approximately 2% of patients.
      Figure 1
      Figure 1Flow diagram of patient and medication possession period eligibility. ED = erectile dysfunction; PDE5i = phosphodiesterase type 5 inhibitor.
      Table 1Nitrate prescription characteristics among PDE5i initiators
      Total eligible PDE5i patients (N = 45,000)
      PDE5i product at initiation, n (%)
       Tadalafil (Cialis)23,343 (51.9)
       Sildenafil (Viagra)16,621 (36.9)
       Vardenafil (Levitra)4,287 (9.5)
       Avanafil (Stendra)749 (1.7)
      Nitrate timing relative to cohort entry, n (%)
       Most proximal baseline prescription
       >−6 months327 (0.7)
       −6 to −3 months265 (0.6)
       −3 to −1 months299 (0.7)
       −1 month to cohort entry482 (1.1)
       Most proximal follow-up prescription
       Cohort entry to 1 month232 (0.5)
       1–3 months258 (0.6)
       3–6 months214 (0.5)
       >6 months145 (0.3)
      Duration of PDE5i-nitrate overlap (total overlap in baseline and follow-up)
       None, n (%)44,720 (99.4)
       Any overlap, n (%)768 (1.7)
       Mean (SD), days103.2 (64.8)
       Median (IQR), days102 (62, 136)
      Percentages were calculated using all eligible PDE5i patients (N = 45,000) as the denominator, regardless of whether nitrate prescriptions were present during baseline/follow-up.
      IQR = interquartile range; PDE5i = phosphodiesterase type 5 inhibitor; SD = standard deviation.

       Cohort Characteristics

      The distribution of covariates in each cohort was assessed prior to (Supplemental Table IV) and following (Table 2 and Supplemental Table V) propensity score matching. Patients in the ED + PDE5i + nitrate (co-possession) cohort had multiple comorbidities and risk factors for CVD, with over 80% having hypertension, hyperlipidemia, and dyslipidemia (Table 2). Approximately 70% of patients also had 7 or more outpatient visits in the preceding year. Prior to matching and relative to both nitrate only cohorts (with and without ED), patients in the co-possession cohort were younger (less likely to be age 75 or older) and generally had similar or lower rates of comorbidities. Conversely, relative to the ED + PDE5i cohort, the co-possession cohort was older, with higher medical utilization and notably higher rates of comorbidities (Supplemental Table IV). Based on observed baseline covariate patterns, nitrate possession (irrespective of the particular cohort) was associated with increased baseline medical utilization and comorbidities compared with the ED + PDE5i cohort.
      Table 2Characteristics for patients from eligible medication possession periods, after propensity score matching between co-possession periods and comparator medication possession periods
      Co-possession match 1: N = 3,648Co-possession match 2: N = 3,652Co-possession match 3: N = 3,647
      ED + PDE5i + nitrate periodsED + nitrate periodsED + PDE5i + nitrate periodsNitrate only periodsED + PDE5i + nitrate periodsED + PDE5i periods
      Age group, years, %
       18–340.270.270.270.790.270.69
       35–442.301.922.303.182.304.33
       45–5414.8311.5714.8412.0514.8316.70
       55–6436.4030.6236.4227.2236.4433.07
       65–7434.2733.8334.2329.1934.2232.57
       75+11.9221.7911.9427.5711.9312.64
      Number of outpatient physician visits, %
       1–25.875.785.865.505.875.26
       3–411.6211.2911.6111.5811.6310.47
       5–612.1212.3112.1011.9912.1210.94
       7+70.0770.3470.1070.5970.0673.07
       None0.330.270.330.330.330.25
      Number of inpatient stays, %
       1–218.6718.6718.7318.8118.6718.23
       3–41.150.931.151.101.150.88
       5–60.300.250.300.160.300.27
       7+0.080.080.080.050.080.05
       None79.8080.0779.7479.8779.7980.56
      Charlson comorbidity index, %
       0–153.3253.2953.2951.4553.3653.19
       2–329.5230.0429.5731.1329.5627.64
       4–511.1811.8111.1712.0811.1111.49
       6–74.173.544.163.974.174.69
       8+1.811.321.811.371.812.99
      Baseline comorbidities, %
       Hypotensive events4.084.034.084.034.003.73
       Hypertension82.7082.8982.7282.1582.7084.89
       Angina (stable/unstable)11.0511.8411.0611.6610.948.34
       Heart failure13.5414.0413.5313.7513.4114.09
       Myocardial infarction12.7712.3112.7912.6812.6711.60
       Stroke4.394.174.384.744.394.69
       Arrhythmia25.6625.7425.6826.5125.6426.71
       Hyperlipidemia86.9887.3186.9787.2786.9588.43
       Dyslipidemia87.2887.6487.2787.4687.2588.73
      ED = erectile dysfunction; PDE5i = phosphodiesterase type 5 inhibitor.
      Propensity score matching was successful in matching almost all of the 3,668 co-possession (ED + PDE5i + nitrate) medication periods to a treatment period within each of the ED + nitrate (3,648; 99.5%) and the nitrate only (3,652; 99.6%) and ED + PDE5i (3,647; 99.4%) cohorts. Covariate distributions were closely balanced across cohorts after matching (Table 2 and Supplemental Table V); one variable for which balance was incompletely achieved was the age category of 75 years or older in both comparison cohorts.

       Hypotensive/CV Outcomes

      IRRs for hypotensive/CV outcomes were calculated for the ED + PDE5i + nitrate (co-possession) cohort relative to each of the matched control cohorts. These data, after propensity score matching and adjustment, are presented for hypotensive/CV outcomes as derived from diagnostic codes (Figure 2) and from NLP of EHR data (Figure 3). Calculated IRRs and 95% CIs relative to the ED + nitrate cohort indicated that the rates of hypotensive/CV outcomes were either not significantly different or were lower with co-possession, with a significantly lower rate of MI, hospitalized unstable angina, death from any cause, presyncope, and syncope. Similarly, the rates of hypotensive/CV outcomes in the co-possession cohort relative to the nitrate only cohort were not significantly different or were lower, and all IRRs were below 1.0. Calculated IRRs and 95% CIs indicated a significantly lower rate with co-possession for clinically recorded hypotension, ventricular arrhythmias/cardiac arrest, dizziness, and death from any cause (one death [not CV-related] over 1,115.5 person-years with co-possession vs 40 deaths over 1,871.6 person-years with nitrate alone; matched unadjusted analysis).
      Figure 2
      Figure 2Adjusted matched IRR for diagnosed hypotensive/CV outcomes in the ED + PDE5i + nitrate cohort relative to matched control cohorts. *IRR based on propensity matched analysis without residual confounding adjustment. CV = cardiovascular; ED = erectile dysfunction; HoTN = clinically recorded hypotension; IRR = incidence rate ratio; MI = myocardial infarction; PDE5i = phosphodiesterase type 5 inhibitor; TIA = transient ischemic attack; VA/CA = ventricular arrhythmia and cardiac arrest.
      Figure 3
      Figure 3Adjusted matched IRR for NLP-identified hypotensive/CV outcomes in the ED + PDE5i + nitrate cohort relative to matched control cohorts. *IRR based on propensity matched analysis without residual confounding adjustment. CV = cardiovascular; ED = erectile dysfunction; IRR = incidence rate ratio; NLP = natural language processing; PDE5i = phosphodiesterase type 5 inhibitor.
      In the secondary analysis comparing the co-possession cohort and the ED + PDE5i cohort, calculated IRRs were greater than one for most of the hypotensive/CV outcomes examined (Figures 2 and 3). The 95% CI for most outcomes (9 of 12) included 1.0, indicating no significant difference in the incidence of these outcomes between the 2 cohorts. Three outcomes, however, showed a statistically significantly higher rate in the co-possession cohort (MI, hospitalized unstable angina, and fainting). This may reflect the observation that the ED + PDE5i cohort was notably healthier from a CV viewpoint than the co-possession cohort (reflected by the indication for nitrate treatment in the latter).

       Qualitative Analysis of PDE5i and Nitrate Co-Possession

      Free-text clinical notes in the EHR database were analyzed using NLP. Overall, 131 (52%) of the 252 virtual EHRs that were reviewed documented that patients had held discussions with their physician regarding co-possession during their co-possession (PDE5i and nitrate) period (Figure 4). Of these, 69 (53%) patients were warned against using both medications together or were advised to discontinue one of the medications. The remaining 62 (47%) patients either had a note indicating infrequent use of one or other of the medications, or no further details of the discussions with their physician were recorded. The 69 who were directly warned against use by their physicians had a lower proportion of patients identified without events (55% vs 63%), and nearly twice as many (33% vs 16%) having a previous cardiac event, compared to patients who did not receive a direct warning. The majority (91; 75%) of the 121 patients who had no recorded discussion with their physician were prevalent users of one or other of the medications, and most (85; 70%) had no record of a previous cardiac event. For the remaining 36 patients, 9 (7%) had event-based co-possession and 9 (7%) had unstable angina, with the other 18 (15%) patients having a previous event prior to co-possession (Figure 4).
      Figure 4
      Figure 4Qualitative analysis of PDE5i and nitrate co-possession. Previous event (PE): includes patients with the occurrence of a diagnostic code for MI, cardiac arrest (ventricular arrhythmia), (congestive) heart failure, and/or stroke/TIA within the study period prior to combination possession. Event-based (EB): patients with no prior event, but had an event leading to the co-possession. Unstable angina (UA): includes patients with the presence of unstable angina at any point during the study period, without prior severe event. No event (NE): includes all patients who did not have a previous recorded cardiac event (as listed under PE) prior to co-possession or a diagnosis of UA during the study period. MI = myocardial infarction; PDE5i = phosphodiesterase type 5 inhibitor; TIA = transient ischemic attack.
      Among patients with PDE5i and nitrate co-possession, approximately 69% had at least one note in their clinical record mentioning both the PDE5i and nitrate. Among these patients, most PDE5i initiators (91% with as-needed nitrates, 100% with daily nitrates) received a warning or deferral in initiating the PDE5i (Table 3). Warnings were less common in those patients who were prevalent PDE5i users (61% with as-needed nitrates, 57% with daily nitrates). Example text extracts of warnings identified from the free-text clinical notes are presented in Table 3, categorized by level of PDE5i and nitrate use.
      Table 3Categorization and example text extracts identified from qualitative assessment of free-text clinical notes for ED patients with PDE5i and nitrate co-possession
      PDE5iNitrateWarningNumberExample text (modified for anonymity)
      InitiationPrevalent (as needed)Yes20 of 22Includes 2 observations of deferring initiation of PDE5i:

      "Patient requested medication to help treat erectile dysfunction. Levitra/Viagra/Cialis cannot be used with nitroglycerin. Awaiting consult from his cardiologist to approve PDE5i treatment"

      "We discussed drug interactions, particularly erectile medications with nitrates. He was warned not to take nitroglycerin with his medication"... "He understands these instructions and would like to use the medication pending cardiology."

      The remainder (18) recommended abstaining from taking together and/or noted lack of nitrate use, for example:

      "He does carry nitroglycerin but has not used it in the past 10 years"..."Prescription provided for PDE5i, and he was warned about possible interaction with nitroglycerin"

      "We discussed the possible dangerous and potentially fatal side effect of if nitro is taken within 24 hours PDE5i. Do not take PDE5i within 24 hours of Nitro."
      No2 of 22No indication of discussion related to co-use of PDE5i and nitrate
      InitiationPrevalent (daily)Yes6 of 6Includes 2 observations in which the daily nitrate was terminated or switched to as-needed use to accommodate the request for PDE5i:

      "Imdur has been stopped "..."Since the patient is now off Imdur, patient may now try PDE5i"

      "Previous chest pain may have been musculoskeletal. Will trial a period without daily nitrate"..."The patient cautioned not to use nitroglycerin with PDE5i."

      Includes 2 observations in which the PDE5i initiation was delayed:

      "Patient reports taking PDE5i after abstaining from Imdur for a day. He asks about stopping Imdur completely"..."I am reluctant to recommend for him to stop Imdur at this time and advised he discuss the issue with cardiology"

      "Pharmacy called with concerns about patient with script for isosorbide and PDE5i. Patient needs to be informed that he cannot use PDE5i with isosorbide mononitrate. He definitely needs isosorbide and was called to inform he cannot use PDE5i with this med"

      The remaining 2 observations included general guidance to avoid taking the drugs together.
      PrevalentPrevalent (as needed)Yes25 of 41Includes 2 observations where PDE5i was terminated:

      "He has had his most recent PDE5i prescription filled since he was told by his physician not to take nitroglycerin. He wants to be able to take sublingual nitroglycerin if needed and has decided that he no longer needs the PDE5i."

      "Patient has recent episodes of exertional chest pain. He is prescribed nitroglycerin but has not taken it. He is also prescribed a PDE5i, and we discussed the risks of using PDE5i with nitro. I advise patient to stop taking PDE5i until the chest pain is better managed."

      The remainder (23) recommended abstaining from taking together and or noted lack of nitrate use, for example:

      "He has not required any nitroglycerin in the past year"..." He was shocked by the cost of the PDE5i DRUG1 and requested a switch to PDE5i DRUG2. He is also considering trying to get it from a Canadian pharmacy. No contraindication."

      "We did talk about the risks of using nitroglycerin with PDE5i. He has never used the drugs at the same time"
      No16 of 41No indication of discussion related to co-use of PDE5i and nitrate
      PrevalentPrevalent (daily)Yes4 of 7Includes 2 observations where PDE5i was terminated or rarely used:

      "Do not take daily nitrate with PDE5i. Stop taking thePDE5i until further notice."

      "Patient reported that he has been taking a PDE5i. This is not on his medication list. Informed patient that the combination of PDE5i with daily nitrate can be dangerous. Patient reported that they won't be taking their PDE5i in the near future."

      Includes 2 observations with recommendations to abstain from nitrate prior to taking PDE5i:

      "He may take PDE5i if he holds Imdur for 2–3 days. He is aware he cannot take PDE5i with Imdur."

      "Discussed that these medications could never be used within 2-days of each other. Patient understands risks"
      No3 of 7No indication of discussion related to co-use of PDE5i and nitrate
      PrevalentEmergencyYes0 of 4No indication of discussion related to co-use of PDE5i and nitrate
      No4 of 4No indication of discussion related to co-use of PDE5i and nitrate
      ED = erectile dysfunction; PDE5i = phosphodiesterase type 5 inhibitor.

      DISCUSSION

      Concomitant use of all PDE5is (class labeling) for ED and nitrates for CV morbidity is contraindicated due to pharmacodynamic interactions potentially leading to the development of hypotension.
      • Schwartz BG
      • Kloner RA.
      Drug interactions with phosphodiesterase-5 inhibitors used for the treatment of erectile dysfunction or pulmonary hypertension.
      ,
      • Kloner RA
      • Goggin P
      • Goldstein I
      • et al.
      A new perspective on the nitrate-phosphodiesterase type 5 inhibitor interaction.
      As rates of co-possession are accordingly low, evidence is scarce regarding hypotensive/CV outcomes in those patients who do concurrently possess both PDE5i and a nitrate.
      In this retrospective, observational study using prescription data from a large, U.S.-wide EHR database, we found that patients with ED under physician's care who were provided access to prescription co-possession of PDE5i and nitrates did not appear to have an increased rate of hypotensive/CV outcomes relative to those using prescription nitrates alone (with or without ED). Rather, our analyses, albeit based on prescription data, suggest that the incidence of outcomes such as MI and hospitalized unstable angina was in fact lower than with prescribed nitrates alone. Qualitative analysis of free-text clinical notes indicated that discussions between patients and physicians do occur during the medication co-possession period, especially prior to providing co-possession for the first time. These important discussions between physicians and patients, combined with very careful patient-by-patient analysis by physicians prior to prescribing to patients with ED who are cardiovascularly stable, may be at least partly responsible for the trends observed in this study.
      Notwithstanding the contraindication for their concomitant use, previous studies have estimated that between 1% and 4% of patients prescribed PDE5i also co-possessed nitrates.
      • Chang LL
      • Ma M
      • von Allmen H
      • et al.
      Co-possession of phosphodiesterase type-5 inhibitors (PDE5-I) with nitrates.
      ,
      • Huri HZ
      • Ling CF
      • Razack AH.
      Drug-related problems in patients with erectile dysfunctions and multiple comorbidities.
      This low level of co-prescribing has seemingly altered little over the past 2 decades.
      • Williams D
      • Feely J
      Initial uptake in use of sildenafil in general practice.
      We found a similar, low level of co-possession in our study, where approximately 2% of patients had overlapping possession of prescriptions for PDE5i and a nitrate. Co-possession among patients with ED was not associated with a greater incidence of hypotensive/CV outcomes relative to patients with prescription nitrates alone. Indeed, relative to both the ED + nitrate and nitrate only cohorts, rates of all examined hypotensive/CV outcomes (identified by diagnostic code or NLP) were either not significantly different or were significantly lower in the ED + PDE5i + nitrate (co-possession) cohort. Those outcomes with a significantly lower incidence included MI (IRR 0.68) and hospitalized unstable angina (IRR 0.35) (both vs ED + nitrate) and clinically recorded hypotension (IRR 0.67) and ventricular arrhythmia/cardiac arrest (IRR 0.48) (both vs nitrate only). All-cause death was also significantly lower vs both nitrate cohorts.
      To our knowledge, no prior literature describes CV outcomes in patients with ED co-prescribed PDE5i and nitrates in a real-world setting. Our novel findings suggest that, in the real-world prescription environment, patients with ED are able to successfully manage their co-possession of PDE5i and nitrates without increasing their risk of hypotensive/CV outcomes.
      While mindful of the potential limitations of retrospective, observational prescription-based data, the finding of a significantly lower incidence of CV outcomes such as MI and ventricular arrhythmia/cardiac arrest with co-possession relative to nitrates alone is interesting. One possible explanation could be that generally healthier individuals, ie, those with lower CV comorbidity, may be more likely to seek and receive a prescription for PDE5i. However, the potential impact of differences in underlying comorbidities was reduced following propensity score matching of the co-possession and comparator cohorts.
      Another possible element of the explanation could be related to the reported cardioprotective effect of PDE5i, through mechanisms including vasodilation and direct myocardial actions.
      • Kukreja RC
      • Salloum FM
      • Das A.
      Role of cGMP signaling and phosphodiesterase-5 inhibitors in cardioprotection.
      ,
      • Hutchings DC
      • Anderson SG
      • Caldwell JL
      • et al.
      Phosphodiesterase-5 inhibitors and the heart: compound cardioprotection?.
      As previously outlined, there is a strong association between ED and CVD, and ED is an early sign of CVD, indicating that men with ED should be evaluated and treated in order to reduce CV risk.
      • Guo W
      • Liao C
      • Zou Y
      • et al.
      Erectile dysfunction and risk of clinical cardiovascular events: A meta-analysis of seven cohort studies.
      • Miner M
      • Parish SJ
      • Billups KL
      • et al.
      Erectile dysfunction and subclinical cardiovascular disease.
      • Zhao B
      • Hong Z
      • Wei Y
      • et al.
      Erectile dysfunction predicts cardiovascular events as an independent risk factor: a systematic review and meta-analysis.
      Furthermore, several studies have suggested that the treatment with PDE5is may reduce CVD risk.
      • Andersson DP
      • Trolle Lagerros Y
      • Grotta A
      • et al.
      Association between treatment for erectile dysfunction and death or cardiovascular outcomes after myocardial infarction.
      • Tzoumas N
      • Farrah TE
      • Dhaun N
      • et al.
      Established and emerging therapeutic uses of PDE type 5 inhibitors in cardiovascular disease.
      • Andersson DP
      • Landucci L
      • Lagerros YT
      • et al.
      Association of phosphodiesterase-5 inhibitors versus alprostadil with survival in men with coronary artery disease.
      It has been suggested that prescribing PDE5i for ED to a man with CVD and no contraindications may be a benefit for CV outcomes.
      • Maas R
      • Rodionov RN.
      Phosphodiesterase-5 inhibitors and survival in men with coronary artery disease.
      However, it is important to emphasize that there is no approved indication for any PDE5i for the reduction of CVD risk.
      Beyond preclinical studies supporting reduced myocardial infarct size and ventricular arrhythmia with PDE5i,
      • Hutchings DC
      • Anderson SG
      • Caldwell JL
      • et al.
      Phosphodiesterase-5 inhibitors and the heart: compound cardioprotection?.
      evidence from nationwide cohort studies shows that PDE5i users have a reduced risk of mortality and hospitalization for heart failure following a first MI
      • Andersson DP
      • Trolle Lagerros Y
      • Grotta A
      • et al.
      Association between treatment for erectile dysfunction and death or cardiovascular outcomes after myocardial infarction.
      and a reduced risk of heart failure and MI relative to the general male population.
      • Vestergaard N
      • Søgaard P
      • Torp-Pedersen AK
      Relationship between treatment of erectile dysfunction and future risk of cardiovascular disease: A nationwide cohort study.
      Well-designed clinical studies are warranted to gain a deeper understanding of the mechanisms underlying the observations we report.
      The data from this real-world retrospective study do not provide evidence of an additive effect on hypotensive/CV outcome risk when co-possessing PDE5i and a nitrate ie, outcome risk exacerbated by both the PDE5i exposure and nitrate exposure. In fact, adding possession of PDE5i to a nitrate (co-possession) resulted in outcome rates that were comparable with (or lower than) those with a nitrate alone. Conversely, rates of these outcomes during co-possession periods tended to be higher than those observed with PDE5i alone, with a statistically higher incidence apparent for 3 outcomes (MI, hospitalized unstable angina, and fainting). Whether this represents a genuine increase associated with co-possession or more reflects that the ED + PDE5i cohort was healthier from a CV perspective (ie, no clinical indication for prescription of a nitrate) and that matching could not fully account for this, is unclear. Considered together, the data suggest that—under the careful supervision of a prescriber—co-possession of PDE5i and nitrate might not confer any increased hypotensive/CV risk beyond that attributable to possession of a nitrate. It would not be appropriate to draw conclusions from these retrospective study data on the safety of co-administration of the 2 medications. Co-prescription may be safe if undertaken in a manner directed by the prescribing physician, supporting managed use with essential risk mitigation to minimize risks of hypotensive/CV events and guide appropriate behavior. Well-designed prospective clinical studies in large, suitable populations are required to better understand the risks and benefits and underpin any changes to current clinical practice recommendations.
      Our qualitative analysis findings provide some insight into the prescriber rationale in cases of co-possession of PDE5i and nitrates. For patients who rarely or never need to take their nitrate medication, or those who need to take their nitrates during a predictable time, providers generally advised their patients to avoid taking the PDE5i and nitrate together, often indicating not within 24 hours of each other. This suggests that physicians might identify a subset of patients with nitrate prescriptions who they view as capable of safely managing co-possession of PDE5i and nitrate prescriptions, and this may explain the lack of increased risk observed in the quantitative analysis carried out in this study. The qualitative analysis highlights there is a need to gain a fuller understanding of physician rationale when co-prescribing of PDE5i and nitrates occurs, and physicians’ free-text notes provide a valuable source of information. It should be noted that the American Heart Association statement on sexual activity and CVD indicates that PDE5i are effective for the treatment of ED in patients with stable CVD, and recommends that physicians review the need for continued nitrate therapy or consider an alternate therapy in patients who have undergone complete revascularization requesting a PDE5i.
      • Levine GN
      • Steinke EE
      • Bakaeen FG
      • et al.
      Sexual activity and cardiovascular disease: A scientific statement from the American Heart Association.
      This study is the first to use NLP of free-text notes in an EHR database to evaluate PDE5i and nitrate co-possession and hypotensive/CV outcomes in a population with ED. We used a combination of structured diagnostic codes and NLP of the free-text clinical notes to identify outcomes. NLP enables the identification of less serious outcomes, such as dizziness that may not receive a diagnostic code. The algorithm rules for using NLP were developed with the intention of maximizing the probability that the symptom/outcome was truly present, if identified as an event. Nevertheless, that a formal evaluation of validity was not performed is a potential limitation. The possibility of shifts in clinical practice concerning the use of PDE5i and nitrate medications over the 5-year study period was considered but any such changes are not thought to have significantly impacted on our findings.
      Our study does have some significant limitations. While the use of EHR data is valuable for assessing clinical healthcare outcomes and treatment patterns, an inherent limitation common to all EHR databases is that data are retrospective and collected primarily for the purposes of patient management. Also, we based our analyses on prescription data, which represent written prescriptions for a medication but do not necessarily indicate that a prescription was filled, or the medication consumed (and if so, taken as prescribed), and provide limited insight into clinical follow-up. Furthermore, any medications, diagnoses and procedures received outside of the provider/hospital networks captured by the EHR database used herein were not included.
      We specifically designed our study to try and account for differences in underlying disease characteristics. To this end, we used propensity score matching and additional adjustments by regression modeling, the results of which resulted in closely matched cohorts. However, this approach to patient-level and cohort-level matching is only able to control for identified covariates that were included in the analysis. As such, there remains a possibility that covariates not considered may confound the results. Another potential source of confounding was the variability in the use of both PDE5i and nitrates. We took an innovative approach to address this by defining medication usage based on real-world prescribing habits and identifying “uncertain periods” during which medication possession could not be confirmed. Any outcomes occurring during such periods were therefore excluded from analysis to help mitigate this potential confounding.

      CONCLUSION

      The findings of this retrospective observational cohort study using EHR data from the United States suggest that patients under the current prescription environment and under continued physician's care who have ED and co-possession of PDE5i and nitrate prescriptions do not have an increased rate of hypotensive/CV outcomes compared with patients with possession of a nitrate only (with or without ED). A number of factors may help explain this finding, including selective prescribing of the combination of PDE5i and nitrate to healthier CV/ED patients. Qualitative analyses suggest that physicians and patients with ED and CV comorbidities have detailed discussions around risks of concomitant use of both PDE5i and a nitrate medication and about how to mitigate those risks, particularly when prescribing these drugs concomitantly for the first time. The analysis also suggests that providers make clinically relevant decisions according to their perceptions of the patients’ CV health status and ability to co-manage these medications.

      STATEMENT OF AUTHORSHIP

      Conceptualization, A.P.N., J.D.S., A.S., A.G. and T.M.; Methodology, A.P.N., J.D.S., A.S., A.G. and T.M.; Software, A.P.N., J.D.S.; Validation, A.P.N., J.D.S.; Formal Analysis, A.P.N., J.D.S., A.S., A.G. and T.M.; Investigation, A.P.N., J.D.S., A.S., A.G. and T.M.; Resource, A.P.N., J.D.S., A.S. and A.G.; Data Visualization, A.P.N., J.D.S., A.S., A.G. and T.M.; Supervision, A.P.N., J.D.S. and A.S.; Project Admin, A.P.N., A.S. and T.M.; Acquisition of funding, A.S., A.G. and T.M.; Writing-review & Editing, A.P.N., J.D.S., A.S., A.G. and T.M.

      Appendix. Supplementary materials

      REFERENCES

        • Saigal CS
        • Wessells H
        • Pace J
        • et al.
        Urologic diseases in America project. Predictors and prevalence of erectile dysfunction in a racially diverse population.
        Arch Intern Med. 2006; 166: 207-212https://doi.org/10.1001/archinte.166.2.207
        • Solomon H
        • Man JW
        • Jackson G.
        Erectile dysfunction and the cardiovascular patient: Endothelial dysfunction is the common denominator.
        Heart. 2003; 89: 251-253https://doi.org/10.1136/heart.89.3.251
        • Nehra A.
        Erectile dysfunction and cardiovascular disease: Efficacy and safety of phosphodiesterase type 5 inhibitors in men with both conditions.
        Mayo Clin Proc. 2009; 84: 139-148https://doi.org/10.1016/S0025-6196(11)60822-7
        • Diaconu CC
        • Manea M
        • Marcu DR
        • et al.
        The erectile dysfunction as a marker of cardiovascular disease: A review.
        Acta Cardiologica. 2020; 75: 286-292https://doi.org/10.1080/00015385.2019.1590498
        • Selvin E
        • Burnett AL
        • Platz EA.
        Prevalence and risk factors for erectile dysfunction in the US.
        Am J Med. 2007; 120: 151-157https://doi.org/10.1016/j.amjmed.2006.06.010
        • Corbin JD.
        Mechanisms of action of PDE5 inhibition in erectile dysfunction.
        Int J Impot Res. 2004; 16: S4-S7https://doi.org/10.1038/sj.ijir.3901205
        • Sangiorgi G
        • Cereda A
        • Benedetto D
        • et al.
        Anatomy, pathophysiology, molecular mechanisms, and clinical management of erectile dysfunction in patients affected by coronary artery disease: A review.
        Biomedicines. 2021; 9: 432https://doi.org/10.3390/biomedicines9040432
        • Schwartz BG
        • Kloner RA.
        Drug interactions with phosphodiesterase-5 inhibitors used for the treatment of erectile dysfunction or pulmonary hypertension.
        Circulation. 2010; 122: 88-95https://doi.org/10.1161/CIRCULATIONAHA.110.944603
        • Kloner RA
        • Goggin P
        • Goldstein I
        • et al.
        A new perspective on the nitrate-phosphodiesterase type 5 inhibitor interaction.
        J Cardiovasc Pharmacol Ther. 2018; 23: 375-386https://doi.org/10.1177/1074248418771896
        • Webb DJ
        • Freestone S
        • Allen MJ
        • et al.
        Sildenafil citrate and blood-pressure-lowering drugs: Results of drug interaction studies with an organic nitrate and a calcium antagonist.
        Am J Cardiol. 1999; 83: 21C-28Chttps://doi.org/10.1016/s0002-9149(99)00044-2
        • Webb DJ
        • Muirhead GJ
        • Wulff M
        • et al.
        Sildenafil citrate potentiates the hypotensive effects of nitric oxide donor drugs in male patients with stable angina.
        J Am Coll Cardiol. 2000; 36: 25-31https://doi.org/10.1016/s0735-1097(00)00705-1
        • Kloner RA
        • Hutter AM
        • Emmick JT
        • et al.
        Time course of the interaction between tadalafil and nitrates.
        J Am Coll Cardiol. 2003; 42: 1855-1860https://doi.org/10.1016/j.jacc.2003.09.023
        • Oliver JJ
        • Kerr DM
        • Webb DJ.
        Time-dependent interactions of the hypotensive effects of sildenafil citrate and sublingual glyceryl trinitrate.
        Br J Clin Pharmacol. 2009; 67: 403-412https://doi.org/10.1111/j.1365-2125.2009.03375.x
        • Swearingen D
        • Nehra A
        • Morelos S
        • et al.
        Hemodynamic effect of avanafil and glyceryl trinitrate coadministration.
        Drugs Context. 2013; 2013212248https://doi.org/10.7573/dic.212248
        • Chang LL
        • Ma M
        • von Allmen H
        • et al.
        Co-possession of phosphodiesterase type-5 inhibitors (PDE5-I) with nitrates.
        Curr Med Res Opin. 2010; 26: 1451-1459https://doi.org/10.1185/03007991003771874
        • Huri HZ
        • Ling CF
        • Razack AH.
        Drug-related problems in patients with erectile dysfunctions and multiple comorbidities.
        Ther Clin Risk Manag. 2017; 13: 407-419https://doi.org/10.2147/TCRM.S118010
        • Williams D
        • Feely J
        Initial uptake in use of sildenafil in general practice.
        Eur J Clin Pharmacol. 2001; 56: 911-914https://doi.org/10.1007/s002280000255
        • Kukreja RC
        • Salloum FM
        • Das A.
        Role of cGMP signaling and phosphodiesterase-5 inhibitors in cardioprotection.
        J Am Coll Cardiol. 2012; 59: 1921-1927https://doi.org/10.1016/j.jacc.2011.09.086
        • Hutchings DC
        • Anderson SG
        • Caldwell JL
        • et al.
        Phosphodiesterase-5 inhibitors and the heart: compound cardioprotection?.
        Heart. 2018; 104: 1244-1250https://doi.org/10.1136/heartjnl-2017-312865
        • Guo W
        • Liao C
        • Zou Y
        • et al.
        Erectile dysfunction and risk of clinical cardiovascular events: A meta-analysis of seven cohort studies.
        J Sex Med. 2010; 7: 2805-2816https://doi.org/10.1111/j.1743-6109.2010.01792.x
        • Miner M
        • Parish SJ
        • Billups KL
        • et al.
        Erectile dysfunction and subclinical cardiovascular disease.
        Sex Med Rev. 2019; 7: 455-463https://doi.org/10.1016/j.sxmr.2018.01.001
        • Zhao B
        • Hong Z
        • Wei Y
        • et al.
        Erectile dysfunction predicts cardiovascular events as an independent risk factor: a systematic review and meta-analysis.
        J Sex Med. 2019; 16: 1005-1017https://doi.org/10.1016/j.jsxm.2019.04.004
        • Andersson DP
        • Trolle Lagerros Y
        • Grotta A
        • et al.
        Association between treatment for erectile dysfunction and death or cardiovascular outcomes after myocardial infarction.
        Heart. 2017; 103: 1264-1270https://doi.org/10.1136/heartjnl-2016-310746
        • Tzoumas N
        • Farrah TE
        • Dhaun N
        • et al.
        Established and emerging therapeutic uses of PDE type 5 inhibitors in cardiovascular disease.
        Br J Pharmacol. 2020; 177: 5467-5488https://doi.org/10.1111/bph.14920
        • Andersson DP
        • Landucci L
        • Lagerros YT
        • et al.
        Association of phosphodiesterase-5 inhibitors versus alprostadil with survival in men with coronary artery disease.
        J Am Coll Cardiol. 2021; 77: 1535-1550https://doi.org/10.1016/j.jacc.2021.01.045
        • Maas R
        • Rodionov RN.
        Phosphodiesterase-5 inhibitors and survival in men with coronary artery disease.
        J Am Coll Cardiol. 2021; 77: 1551-1553
        • Vestergaard N
        • Søgaard P
        • Torp-Pedersen AK
        Relationship between treatment of erectile dysfunction and future risk of cardiovascular disease: A nationwide cohort study.
        Eur J Prev Cardiol. 2017; 24: 1498-1505https://doi.org/10.1177/2047487317718082
        • Levine GN
        • Steinke EE
        • Bakaeen FG
        • et al.
        Sexual activity and cardiovascular disease: A scientific statement from the American Heart Association.
        Circulation. 2012; 125: 1058-1072https://doi.org/10.1161/CIR.0b013e3182447787