This was a multicenter, prospective, parallel group, randomized, open-label, flexible dose study. It was a 12-week trial conducted simultaneously at 11 outpatient medical centers for alcohol abuse treatment across Israel from January 1, 2005 to June 31, 2006. All centers belong to the Israeli Ministry of Health, and are audited by the Department for the Treatment of Addictions. The Ministry of Health institutional review board approved the study protocol, and all patients provided written informed consent prior to participating in the study.

Similar to our preliminary report [37], male patients were eligible if they met the following inclusion criteria: 1) were between ages 18 and 50 years; 2) had a ICD-10 diagnosis of AD (F10.2); 3) sought treatment with the aim to stop alcohol consumption; 4) completed a detoxification program not later than the one month preceding the study recruitment; 5) had complaints of ED for at least 12 weeks preceding the study; and 6) had a regular female partner for the study duration.

AD was diagnosed according to International Classification of Diseases, Tenth Edition (ICD-10) criteria [38] adopted in Israel since 1994. The alcohol history was described by the following parameters: 1) age at first alcohol consumption, 2) age at first binge, 3) duration of harmful alcohol consumption, 4) number of prior inpatient or outpatient detoxifications, 5) average alcohol intake in last six months (gram alcohol/drinking day), and 6) number of drinking days during last month. The severity of AD was evaluated according to ICD-10 criteria for three categories according to the frequency of drinking (during the previous 6 months) and amount of alcohol intake. These were: 1) continuous drinkers = (almost) daily alcohol consumption without binges; 2) frequent heavy drinkers = frequent alcohol consumption (more than 3 days/week) with frequent intoxication (more than one/week); and 3) episodic drinkers = less frequent, irregular alcohol consumption with longer (>5 days) sober periods, and some binges (less than one/week).

Alcohol-associated sexual dysfunction (AASD) was defined by carefully elaborated DSM-IV criteria for alcohol-induced sexual dysfunction [39], which includes specific items for impaired desire, arousal (ED), orgasm, and sexual pain. Patients had to have substantial impaired sexual function that caused significant distress, defined by at least one of the following criteria: ED as defined by persistent or recurrent inability to attain an adequate erection until completion of sexual activity; inability to have an orgasm, or ejaculatory delay of at least 10 minutes for masturbation or intercourse.

Patients were not enrolled if they had anatomical penile deformities (e.g., Peyronie’s disease), primary or prior diagnosis of a sexual disorder other than AASD, co-morbid serious medical illnesses (hepatic, renal, neurological, cardiovascular, hematological, diabetes mellitus), suicide risk, acute psychosis, severe depression (with psychotic features), organic brain syndromes or current use of other than alcohol psychoactive substances, drugs or therapies, such as benzodiazipines, sedatives, antidepressants, barbiturates, and antipsychotics. The psychiatric diagnostic assessments were made according to ICD-10 criteria.

Patients were enrolled during 1.5 years and recruited from outpatient settings and referrals. All patients were evaluated for eligibility at screening (N = 127) (Figure 1) and all consenting patients (N = 108) completed the International Index of Erectile Function (IIEF) [40] to establish their ability for self-evaluation of sexual dysfunction. All patients received a physical examination, including blood pressure, electrocardiogram, and standard biochemistry tests (blood urea nitrogen, uric acid, glucose, total protein, albumin, total bilirubin, total cholesterol, triglycerides, fibrinogen, alkaline phosphatase, SGOT, SGPT, sodium, potassium, chloride, calcium, inorganic phosphorus, bicarbonate, creatinine, and creatine phosphokinase [CPK]) and hematological laboratory tests (hemoglobin, hematocrit, erythrosyte count, white blood cell count [WBC], total and differential WBC count, and platelet count).

Using a random number generator for each incoming participant for treatment decision-making [41], a computer-generated randomization schedule was developed. This resulted in 54 patients assigned to sildenafil citrate as add-on to standard ongoing outpatient program for treatment of AD and another 54 patients assigned to only the standard program without sildenafil. This program involved education and therapy, addressing problems contributing to or resulting from the alcoholism, and learning skills to manage the alcoholism over time. The only restriction to this randomization was that the groups be of equal size. The largest difference in numbers assigned to the two groups at endpoint of the study was 3. Additionally, there were no statistically significant differences between assigned groups at baseline in socio-demographic characteristics (Table 1). At baseline, eligible patients randomly assigned to receive sildenafil were instructed to take one tablet approximately one hour before anticipated sexual activity but not more than once daily. They were also instructed to make at least two attempts at sexual activity weekly. The dose of the drug could be adjusted from 1 to 2 tablets (VIAGRA^® was provided by Pfizer Pharmaceuticals Israel Ltd, Herzliya Pituach, Israel). Drug accountability and self-rated and physician-rated assessments were performed at baseline and 12 weeks later. Throughout the study, the investigators monitored and collected any spontaneous reports of adverse events and evaluated their severity and relationship to the study medication.

Efficacy was evaluated using a battery of validated measurements. These included the IIEF [40], the Global Life Functioning inventory (GLF) [42], Beck Depression Inventory-short form (BDI-13) [43], General Self-Esteem Scale (RGSES) [44], and the Multidimensional Scale of Perceived Social Support (MSPSS) [45]. Time frame for all measures was the 10 days preceding the assessment.

The IIEF is a self-rated 15-item instrument to assess sexual function in five functional domains: erection, orgasm, desire, intercourse satisfaction and overall satisfaction. Questions are anchored on a 5-point scale with 1 corresponding to “almost never/never” and 5 corresponding to “almost always/always”. A score of 0 means the absence of sexual activity, stimulation or intercourse. The minimum possible total score is 5, and the maximum total is 75. The changes in the IIEF scores quantified the magnitude of the response.

The GLF inventory was developed to tape distress, well-being, functioning and life satisfaction and has been shown to be sensitive to changes and to discriminate well between efficacious treatments [46]. It is based on seven items of the Dupuy’s psychological general wellbeing index to measure wellbeing and distress [47], supplied with six items tapping general functioning and life satisfaction. Each item is rated on a 6-point Likert scale, with higher scores indicating better outcomes. In the present study the GLF was used as a self-evaluation measure of patient general wellbeing and functioning.

The severity of depressive symptoms was assessed using the standard abridged form of the BDI. Each of its 13 categories of symptoms and attitudes scores from 0 (absence of the symptom) to 3 (extreme severity of the symptom). The ranges of total scores are: 0–4, none or minimal, 5–7, mild, 8–15, moderate; and 16 and over, severe depressive symptoms.

The RGSES is a 10-item Likert scale with items answered on a 4-point scale - from “strongly agree” to “strongly disagree”, with higher scores indicating higher self-esteem.

The MSPSS was used as a self-report tool for assessing emotional help and the level of satisfaction with the social support obtained from family, friends and significant others. The scale includes 12 items, each of which refer to the people to whom the respondent would turn if he/she had problems of a personal, health or family nature, as well as financial and employment problems. Responses are scored on a 7-point scale from 1 (“completely disagree”) to 7 (“completely agree”). The MSPSS total score and three subscales scores are computed, with a higher score indicating a greater satisfaction with social support.

For the entire sample (N = 101), internal consistency reliability, as measured by Cronbach’s α coefficient, was satisfactory, specifically: 0.77 for the GLF; 0.80 for the BDI; 0.79 for the RGSES, and 0.91 for the MSPSS.

All analyses were performed with SAS version 9.1 (SAS Institute Inc, Cary, NC, USA). Baseline demographic characteristics were compared using descriptive statistics by χ² and Fisher exact tests (where cell sizes were <5). Mean scores and standard deviations (SDs) were computed and reported. The statistical significance of change from baseline to week 12 (Δ) was evaluated with two-tailed paired t-tests. Effect size is the primary outcome measure, and power for the study was calculated using effect size in each group. Expecting to find differences between the groups of effect size = 0.4 would clarify whether potential observed differences also merit clinical significance. Using a two-group analysis of effect sizes between the sildenafil and control groups, setting alpha at 0.05, power = 0.80 and a two-tailed t-test, a minimum of 52 subjects per group is required. Thus, 52 patients per group met power requirements for these specific aims.

Effect size (ES) for treatment efficacy of erectile dysfunction was calculated as the improvement in IIEF mean score for the sildenafil group minus the improvement in IIEF mean score of the control group over 12 weeks, divided by the standard deviation of the entire sample at baseline. ES for other variables studied was computed using the same equation. Following Cohen’s classic demarcation [48], Middel and associates [49] showed that ES reflects clinical relevance. An ES <0.20 indicates “no change,” an ES ≥0.20 but <0.50 indicates “a small change,” an ES ≥0.50 but <0.80 indicates “a moderate change,” and an ES ≥0.80 indicates “a considerable change”. We also use ES equal 0.5 SD as a universal measure of clinical significance [50].

A statistically significant increase in the IIEF mean scores of each sexual function domain was noted among all sildenafil-treated patients (Table 2). In contrast, the mean scores of all the sexual function domains among the control patients decreased over the trial, and in orgasmic function and sexual desire domains this decline was found to be statistically significant (P < 0.05). ES ranged from 0.91 for sexual desire to 1.23 for overall satisfaction domain, and 1.25 for IIEF total score, indicating clinically relevant improvement according Cohen classical demarcation [48]. However, only ESs for sexual desire and overall satisfaction demonstrated clinical importance of the changes using a conservative measure of clinical improvement as 0.5SD [50].

Baseline levels of depression were generally unrelated to efficacy or treatment satisfaction. Both groups demonstrated a statistically significant reduction in BDI symptom severity scores, with the ES = 0.47 (Table 3).

We observed a significant improvement in the GLF total scores as well as in the well-being and functioning subscale mean scores only in sildenafil-treated patients (all P < 0.001), but not in the control group. Correspondingly, ES values ranged from .44 for wellbeing to 0.57 for functioning domains, and 0.63 for GLF total score, reflecting clinically important changes (all > 0.5DS).

In parallel to a meaningful depression reduction, the sense of self-esteem improved substantially among individuals of the sildenafil group (P < 0.001), while this remained at pretreatment levels among the controls (ES = 0.61).

Similar to self-esteem, from baseline to endpoint a significant improvement was observed in the perception of overall social support (P = 0.02) as well as support from friends (P = 0.01) and significant others (P = 0.05) in the sildenafil-treated patients, while it remained unchanged in the control group. However, ESs for these measures were low, 0.31 for overall support, 0.24 for friends’ and 0.39 for significant others’ support, did not reflecting clinical relevance.

Sildenafil was well tolerated. The most common side-effect was headache, reported by 32% of sildenafil-treated patients (n = 16). Only one patient reported dyspepsia (2%). All of these adverse effects were transient and mild in nature. No serious adverse events related to the study drug were reported.