Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Relevance of methodological design for the interpretation of efficacy of drug treatment of premature ejaculation: a systematic review and meta-analysis


The aim of this systematic review and meta-analysis is to evaluate whether the design and methodology of drug-treatment studies of premature ejaculation affect the efficacy outcome differently. Therefore, methodological, design and efficacy data from 79 studies (3034 males), published between 1943 and 2003, are reviewed. A meta-analysis is performed on 43 selective serotonin reuptake inhibitors (SSRIs) and clomipramine studies (1514 males), published between 1973 and 2003; these studies were pooled to provide a summary variance-weighted effect size. The antidepressant-induced percentage increase of the intravaginal ejaculation latency time (IELT) was calculated and examined against various methodological items. A significant difference in efficacy between SSRIs was observed. Using daily treatment, paroxetine appeared more effective than the other SSRIs. Retrospective use of a questionnaire, subjective reports, single-blind and open study designs generate far greater variability of ejaculation time both at baseline and during active drug treatment than real time assessment by stopwatch. In conclusion, at daily treatment, the overall efficacy of paroxetine, clomipramine, sertraline and fluoxetine is comparable, but paroxetine exerts the strongest ejaculation delay. Only eight (18.5%) studies on antidepressant treatment fulfilled all criteria used in evidence-based medicine, for example, randomised, double-blind studies with prospective real time (stopwatch) assessment of the IELT at each intercourse. Single-blind studies, open designs, retrospective reporting, or the use of a questionnaire to assess ejaculation time should be avoided.


Some selective serotonin reuptake inhibitors (SSRIs) and clomipramine have increasingly become the agents of first choice for the treatment of lifelong premature ejaculation.1, 2, 3 Based on animal4 and human psychopharmacological studies, Waldinger et al have postulated that lifelong premature ejaculation is a neurobiological phenomenon related to decreased central serotonergic neurotransmission, 5-HT2C receptor hyposensitivity and/or 5-HT1A hypersensitivity.5, 6, 7 In addition, Waldinger postulates that lifelong early ejaculation is not an acquired disorder due to learned behavior, as has been suggested by Masters and Johnson, but, instead, belongs to the normal biological variability of the intravaginal ejaculation latency time (IELT) in men, with a possible familial genetic vulnerability.1, 5, 6, 7, 8, 9 In this sense, early ejaculation is considered a neurobiological phenomenon, that may become perceived as premature ejaculation and consequently may secondarily lead to psychological distress.1

Due to many practical difficulties sexual psychopharmacological research is subjected to, it is questioned whether the methodology of research is of relevance for clinical outcome.10, 11 The way to address the issue of methodological bias is to perform a systematic review of the complete literature and to perform a meta-analysis on all data. Such a meta-analysis on methodology and drug treatment has not been performed yet on this issue. Consequently, there is no evidence-based consensus on methodology in research and treatment.11

In the current study, a review was performed on all available drug-treatment studies, published between 1943 and 2003, and a meta-analysis was conducted on all studies using serotonergic antidepressants for the daily and as-needed treatment of premature ejaculation.


Literature search

All drug-treatment reports and studies, whether drugs, ointments, creams or solutions, were included in the systematic review. A search by MEDLINE (1966–2002), Web of Science, PICA, and EMBASE (1980–2002), was performed, checking all publications that used the words premature, rapid, ejaculation, praecox in the title, abstract or keywords. The computer search was combined with manual cross-referencing of all papers. Apart from German12, 13, 14 and French15 articles, the publications in Italian,16, 17, 18 Spanish,19, 20 Portugese,21 Czech,22, 23, 24 and Hungarian25 language were translated. Three authors were contacted by e-mail or telephone for further explanation. Of two papers published in a South Korean journal,26, 27 the English abstracts were used and one of the authors was contacted for further information. Abstracts in journals without the full text being published elsewhere or without being in press were not included in the review. Two articles23, 24 of drug treatment combined with behavioral treatment were not included in the meta-analysis.

Data analysis

Only studies reporting quantitative data on the change of the IELT or equivalents of the ejaculation time due to treatment were included in the meta-analysis. Since various scales were used to quantify efficacy (in some papers, a stopwatch was used; in others, a clock, questionnaire or subjective assessment), we decided to bring these different quantifications on the same scale by calculating the percentage change from baseline. Percentage change is defined as 100%* (follow-up IELT−baseline IELT)/baseline IELT. In most papers, no numbers on the percentage change were reported, and we had to calculate it from the average baseline IELT and the average follow-up IELT, which is not complicated. For the meta-analysis, however, we needed the variance of the percentage change. The delta method28 is a mathematic–statistical technique to calculate the variance of the percentage change from the means and variances of the baseline and follow-up IELT. Standard errors of these average change values were calculated in a similar fashion. The effects of research design characteristics on the change values were assessed in a similar fashion.

A random-effects meta-analysis model was used. Basically, we assumed that each study had its own true percentage change (say, θi), which was estimated by the observed/inferred percentage change , where ei is normally distributed with observed variance si2. We further assumed that θi followed a normal distribution with mean μ and variance τ2. The parameters τ2 and μ were estimated by maximizing the likelihood (using SAS proc mixed). P-values were assessed by comparing the loglikelihoods under the null hypothesis to the loglikelihoods under the alternative models.


General results of review

Since the first publication in 1943, there have been 79 publications on drug treatment until 2003. These reports comprised the use of anesthetic ointments,12, 26, 27, 29, 30, 31, 32, 33, 34, 35, 36, 37 neuroleptics,20, 22, 38, 39, 40, 41, 42 monoamine-oxidase inhibitors,16, 43 sympatholytic drugs,18, 19, 44, 45, 46 antibiotics,47, 48 and a group of miscellaneous agents13, 14, 21, 49, 50, 51 (Table 1). Between 1973 and 2003, there have been 35 studies15, 17, 23, 24, 25, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81 on daily treatment with clomipramine, a tricyclic antidepressant, and on SSRIs (Table 2). In the same period, on-demand antidepressant treatment results were reported in eight studies82, 83, 84, 85, 86, 87, 88, 89 (Table 3). The current review consists of 3034 males obtained from 79 studies.

Table 1 Drug treatments for premature ejaculation: anesthesia, sympathicolytic drugs, and miscellaneous drugs (daily and as-needed strategies)
Table 2 Drug treatments for premature ejaculation: selective serotonin reuptake inhibitors, clomipramine and modern antidepressants (daily strategy)
Table 3 Drug treatments for premature ejaculation: selective serotonin reuptake inhibitors, and clomipramine (as-needed strategy)

Operational definition

A total of 46 (58.2%) studies mentioned an operational definition of premature ejaculation. In all, 41 (51.8%) studies used an exact cutoff point of the ejaculation time as criterion for inclusion: 1 min or less in 19 studies, 2 min or less in 11 studies, 3 min or less in eight studies, 30 s, 4 and 5 min in one study each. Six studies used the DSM classification. With the exception of six publications,29, 90, 44, 20, 21, 72 all studies that mentioned a definition of premature ejaculation used the ejaculation time as outcome measure.

Double-blind vs single-blind/open design

Of all 79 studies, 35 (44.3%) used a double-blind, four (5%) a single-blind, and 30 (37.9%) an open design. There were eight case reports and two reviews. From all studies, 34 (43%) were placebo-controlled.

Ejaculation time vs satisfaction

In total, 53 (67%) studies used the ejaculation time or equivalents as a clinical outcome, of which 10 mentioned the IELT.56, 62, 63, 65, 70, 77, 78, 79, 81, 87Altogether, 25 studies used qualitative outcome measures: feeling of satisfaction (10),44, 51, 46, 54, 55, 59, 17, 71, 72, 75 feelings of improvement (10)12, 13, 14, 16, 19, 20, 21, 42, 15, 64 and feelings of control (5).29, 30, 43, 31, 90

Stopwatch vs subjective report/questionnaire

A total of 47 (59.4%) studies used subjective reporting, eight (10.1%) a questionnaire, 19 (24%) a stopwatch and four (5%) a clock or watch to assess the clinical outcome. Of the 23 stopwatch and clock studies, 15 (65.2%) were double-blind,49, 27, 34, 35, 42, 57, 60, 63, 65, 77, 78, 79, 80, 81, 87 three (13%) used a single-blind26, 68, 86 and five (21.7%) used an open design.33, 69, 74, 85, 89

Prospective vs retrospective assessments

In all, 22 (27.8%) studies used a prospective design to measure the clinical outcome at each intercourse. In 56 (70.8%) studies, the outcome was assessed retrospectively.

Baseline period

Altogether, 23 (29.1%) studies mentioned a baseline assessment. However, in six of these studies, the duration of the baseline period or number of baseline assessments were not mentioned.33, 34, 35, 67, 76, 85 Baseline periods ranged from 168, 69 or 226, 27 to mostly 4 weeks.49, 57, 65, 77, 78, 79, 80, 89 In some studies, a certain number of intercourses41, 60, 74, 81 was taken as baseline measurement.

Duration of studies

The majority of studies used a fixed duration period. However, the duration of studies differed considerably, from 1 to 68 weeks. A total of 10 studies (13%) had a variable duration.

On average, however, the duration was between 4 and 8 weeks, particularly in the daily treatment studies with SSRIs and clomipramine.

Meta-analysis of 35 daily treatment studies with serotonergic antidepressants

The current meta-analysis was limited to those studies which used either a questionnaire/subjective report of the patient or stopwatches. Reviewing the available literature on this issue, it appeared that this was only feasible in studies reporting on the effects of daily taken serotonergic antidepressants, including 44 treatment groups: 10 with placebo, four with clomipramine, seven with fluoxetine, nine with paroxetine, nine with sertraline, two with citalopram and one each with fluvoxamine, mirtazapine and nefazodone.

Number of subjects in relation to treatment efficacy

With regard to the efficacy, expressed as the percentage of IELT increase of several SSRIs and clomipramine in relation to the number of included individuals per study, no correlation was found between treatment efficacy and patient numbers, indicating a lack of publication bias in our meta-analysis.

Efficacy differences among serotonergic antidepressants ( Figure 1 )

Figure 1

Plot showing the percentage increase of the IELT (95% CI) of serotonergic antidepressant drugs and placebo. The figure shows 35 daily treatment studies (I) (red bars) with heterogenous study designs (eg single blind/open design/double blind) and heterogenous assessments of the IELT (eg subjective report/questionnaire/stopwatch) vs eight double-blind, daily treatment studies (II) (green bars) with prospective use of real time stopwatch assessment of the IELT at each intercourse. The small number of studies involving as-needed strategies prevented them from being included in the meta-analysis.

Regardless of the major differences in design and drug doses of the several studies, it appeared that paroxetine, sertraline, fluoxetine and clomipramine were significantly efficacious in delaying ejaculation compared with placebo (P<0.001). Moreover, paroxetine induced a stronger ejaculatory delay than fluoxetine (P<0.001), clomipramine (P<0.03), and sertraline (P<0.05). Sertraline-induced delay was significantly stronger than that of fluoxetine (P<0.006), whereas clomipramine-induced delay was not significantly different from fluoxetine and sertraline. The rank order of efficacy was therefore: (1) paroxetine (1492% IELT increase, 95% CI: 918–2425%), (2) sertraline (790%, 95% CI: 532–1173%), (3) clomipramine (512%, 95% CI: 234–1122%), (4) fluoxetine (295%, 95% CI: 172–506), (5) placebo (45%, 95% CI: 27–87%), noting that the effect of clomipramine had about the same effect as fluoxetine and sertraline.

Methodological subgroups

In order to investigate the influence of methodology, design and the pharmacological drug properties of the various studies on efficacy, the following items were investigated: dose–response relationship, baseline IELT value vs percentage IELT increase, the definition of premature ejaculation (IELT within 1 or 2 min), stopwatch use vs questionnaire use or subjective report, prospective vs retrospective assessment and double-blind design vs single-blind and open design.

Dose–response relationships

Clomipramine was evaluated in doses of 25 mg (one study), 30 mg (one study) or 50 mg (two studies). The doses for the SSRIs were: fluoxetine 10 (one study), 20 (three studies), 40 (two studies), or 60 (one study) mg; paroxetine 20 (seven studies), or 40 (two studies) mg; sertraline 25 (one study), 50 (five studies), 100 (two studies) or 200 (one study) mg daily. There was no significant correlation between dose and response among the various drugs (P>0.13).

Relationship between baseline IELT and percentage increase on treatment

The average baseline IELT was (mean±s.d.) 41±23 s (range 13–81 s). No relationship was found between baseline IELT and percentage increase on treatment, regardless of which drug was used and of which definition of premature ejaculation was included (1 vs 2 min).

Stopwatch vs questionnaire or subjective reporting

The IELT or ejaculation time equivalents were measured with a stopwatch in 27 treatment groups, and with a questionnaire or subjective report in 15 groups. On average, the mean percentage IELT change was not significantly different between studies with a stopwatch and with a questionnaire. However, the variability of the outcome measure was far greater in those studies using a questionnaire or subjective report (P=0.001).

Prospective vs retrospective assessment

Of 24 prospective and 20 retrospective groups, no significant differences were found between the mean percentage of IELT increase. However, retrospective studies showed a significantly higher variability in outcome measure (P=0.001) than prospective studies.

Double-blind vs single-blind and open label studies

There were 35 double-blind treated groups and nine single-blind or open-treated groups. On average, single-blind and open treatment groups had significantly higher percentage IELT changes (P=0.004). The geometric mean IELT change was 864 (95% CI: 447–1667) s in single-blind/open studies and 252 (95% CI: 155–412) s in double-blind design studies. This significant difference was independent of treatment (P=0.85).

Prospective, double-blind, real-time-assessed stopwatch studies ( Figure 1 )

Excluding those studies subjected by greater variability of outcome measure (percentage IELT increase), for example, the use of a questionnaire and subjective report, retrospective and single-blind/open studies, eight prospective, double-blind, real time stopwatch studies57, 60, 65, 77, 78, 79, 80, 81 were left for final analysis. Four of these studies were performed by the same investigators.65, 77, 78, 79 In total, these eight studies included four placebo, two clomipramine groups (25/50 mg), three fluoxetine (20/40 mg), four paroxetine groups (20 mg), two sertraline groups (50 mg), two citalopram (20 mg) groups and three groups in which either fluvoxamine (100 mg), mirtazapine (30 mg) or nefazodone (400 mg) were used.

Of all these antidepressants, paroxetine, fluoxetine, sertraline and clomipramine exerted a significant delay of the IELT compared with placebo (P<0.001).

The percentage IELT increase on paroxetine was significantly greater than on sertraline (P<0.04) and fluoxetine (P<0.08), but did not differ from clomipramine (<0.06). The effects of clomipramine were not significantly different from those of fluoxetine (P=0.58) and sertraline (P=0.76). On the basis of data provided by this meta-analysis, the rank order of efficacy to increase the IELT was: (1) paroxetine (783% IELT increase, 95% CI: 499–1228%), (2) clomipramine (360%, 95% CI: 201–644%), (3) sertraline (313%, 95% CI: 161–608%), (4) fluoxetine (295%, 95% CI: 200–435), (5) placebo (47%, 95% CI: 29–76%), noting that clomipramine, sertraline and fluoxetine had about the same effect (Table 4).

Table 4 Efficacy of SSRIs and clomipramine in delaying ejaculation; percentage increase with 95% confidence intervals in all 35 daily treatment groups (meta-analysis I) and in eight selected daily treatment groups (double-blind, prospective, stopwatch studies) (meta-analysis II)

Meta-analysis of on-demand treatments

Parrallel to daily treatment, a rather similar rank order of efficacy of paroxetine, sertraline and clomipramine was demonstrated. Although clomipramine only showed a mild increase of the percentage IELT (mean (95% CI); 263 (60–1152)%, P<0.05), clear significance was achieved with paroxetine and sertraline. The delay after paroxetine had a mean (95% CI) 929 (398–2166)%, P<0.001 and after sertraline 553 (210–1457)%, P<0.01. However, caution is needed in interpreting the on-demand treatment data. One should keep in mind that the studies which were included in the meta-analysis greatly differed in methodology (Table 2). The studies were unbalanced for the antidepressants used, baseline IELT values, design (double-blind vs open) and assessment techniques (questionnaire vs stopwatch). Apart from these arguments, due to the very limited numbers of publications, no final conclusions could be drawn with regard to dose relationships and the influence of baseline IELT on predicting the drug response. Moreover, apart from one double-blind study with a watch,87 there are as yet no double-blind studies using a stopwatch for on-demand treatment. Similarly, all double-blind on-demand placebo treatments82, 83, 84 were evaluated without a stopwatch.


In the current study, a systematic review was performed on all drug-treatment studies of premature ejaculation, published between 1943 and 2003, with regard to methodology and design. The results of the systematic review have shown that a meta-analysis could not be performed on treatments with anesthetics, miscellaneuous drugs and on-demand treatment of serotonergic antidepressants, due to insufficiently provided data. In contrast, a meta-analysis was feasible on the daily treatment studies with clomipramine and the various SSRIs. The aim of the current study was in the first place to investigate the level and development of methodology in drug-treatment studies, and secondly to investigate whether methodology affected the treatment outcome, as would be expected according to evidence-based medical principles. We have consequently chosen for ‘the constant percentage change’ of the IELT and have based this choice on the results of the SSRI treatment studies with a stopwatch by our group.65 This choice was reasoned by the results of a double-blind placebo-controlled study with paroxetine 20 mg/day in men who complained of premature ejaculation.65 In this study, we compared men with an IELT of less than 1 min and men with an IELT between 1 and 3 min. After baseline measurements of the IELT at home with a stopwatch during 1 month, men with an IELT less than 1 min and those between 1 and 3 min were randomized either to placebo or active treatment with a follow-up of 6 weeks. It appeared that the absolute IELT values in both groups were different, but that the percentage increase of IELT in those men below 1 min (mean IELT 18 s) was identical with those who started with 1 and 3 min (mean IELT 82 s). This finding proves the identical ability of paroxetine to prolong the IELT, regardless of the initial absolute IELT values. The identical percentage increase of paroxetine (and we assume that this is also the case with other SSRIs) forms the basis of the approach to compare men with low and high IELTs at baseline.

Methodological considerations

Review of all studies

The review of all drug-treatment studies illustrates an improvement in methodology from initially published case reports and open label studies towards double-blind, placebo-controlled clinical trials. However, unfortunately, for unknown reasons, in recent years, a gradual increase in single-blind and open label studies has appeared to be again acceptable for peer review, particularly for on-demand treatment,85, 86, 88, 89 in contrast to the criteria of evidence-based medicine.

Apart from worries regarding the design issues of clinical trials, one of the major issues of premature ejaculation drug-treatment research remains the outcome measure. The review showed that, throughout the years, subjective feelings of ‘control’, ‘satisfaction’ and ‘improvement’ have been used, without clear operational definitions of what is meant by these vague terms. However, particularly since the mid-80s, the majority of studies used the ejaculation time as an outcome measure. A different terminology has been used for ejaculation time, like, for example, the duration of coitus, and an exact definition of ejaculation time is lacking in the majority of studies. An important progress in this methodological issue was made in 1994 by the introduction of the operationally defined IELT.56 There is a slow increase in the use of this concept.

Another key methodological problem is the instrument of assessment. It is clear that the only instrument to measure time objectively is a stopwatch. An important progress in methodology was made in 1983 by the introduction of a stopwatch49 and the use of a baseline period. However, in the majority (70%) of studies, a subjective estimate/report or questionnaire is still used. Only in 29% of the studies, a stopwatch, watch or clock was used. Unfortunately, of these 29% studies, 35% were single-blind or had an open design.

Related to the instrument is the moment of assessment. Drug-treatment studies usually have a prospective design. On the other hand, estimation of the ejaculation time either by spontaneous report or by questionnaire always implies retrospective assessment. In contrast, the use of a stopwatch always implies prospective real time assessments of the IELT at each intercourse. A combination of a drug-treatment study with one of the three instruments may erroneously lead to a mix-up of terminology. Caution is therefore warranted, if analysing prospective drug-treatment studies when retrospective IELT estimates have been used.

The use of the IELT and a stopwatch enabled an operational definition or premature ejaculation, which was empirically found in a study10 of a cohort of Caucasian men with lifelong premature ejaculation, being an ejaculation that takes place within 1 min after vaginal penetration. The current review showed that, in studies using a definition of premature ejaculation, the ejaculation time appeared to range from 30 s to 5 min, with a preference for the 1 min definition. It is of note that the DSM-III-R and DSM-IV definitions do not mention a clear quantifiable ejaculation time, and encompass vague terms.10 Therefore, these DSM definitions should be avoided as operational criteria in pharmacological studies. The ICD-10 definition (eg, ejaculation before or within 15 s) has not been mentioned in any of the drug-treatment studies.

The meta-analysis

The meta-analysis was performed on all available studies with serotonergic antidepressants, regardless of methodology and design (mixed-methodology studies). Despite the many confounders in the methodological approach in all of these studies, it appeared that there was a rank order of efficacy for the various antidepressants. Paroxetine exerted the strongest ejaculation-delaying effect compared to the effects of sertraline, clomipramine, fluoxetine and placebo. Clomipramine did not differ from sertraline and fluoxetine.

It has to be emphasised that the meta-analysis of the mixed-methodology studies demonstrated that single-blind and open studies have led to a significantly higher variability of the percentage IELT increase than was found in double-blind studies. This again demonstrates the methodological insufficiency of single-blind and open-design studies. In addition, the meta-analysis also showed that retrospective assessment by subjective report or the use of a questionnaire during prospective drug-treatment studies have led to a significantly higher variability of the percentage increase of the ejaculation time, compared with the prospective use of a stopwatch during each intercourse.

In order to avoid bias of insufficient methodology and design, a final meta-analysis included only those eight (18.6%) studies of proper methodology, for example, randomized, prospective, double-blind clinical trials with a stopwatch. Again, a similar rank order in efficacy was identified, but with far lower values of the percentage IELT increase. The similar outcome of rank order in efficacy supports the superiority of paroxetine as the most effective drug in delaying ejaculation. However, compared to the mixed-methodology meta-analysis, it was now clomipramine instead of sertraline being secondary efficacious. Similarly, clomipramine appeared to be equally efficacious as sertraline and fluoxetine.

Pharmacological considerations

Review of all studies

The review of all drug-treatment studies has shown that, throughout the years, two approaches seem to have become popular to treat premature ejaculation. The initial European approach to use anesthetic ointments is still being used, and research on these methods seems particularly undertaken in the far East with the herbal SS cream. Paroxetine, fluoxetine, clomipramine and sertraline are far more used and investigated in Western countries. This development seems to coincide with a specific interest to perform clinical and basic research on the mechanisms driving premature ejaculation, for example, a tendency to focus on the sensory input of tactile stimuli (eg sensitivity of the glans penis and evoked potentials of peripheral neurons) in the far East,91, 92 in contrast, a tendency to focus on the motoric output (eg serotonergic brain and spinal cord areas) in Western countries.4, 5, 93 It is of relevance to note that, in studies on ointments in the Far East, a 3 min criterion seems in use, in contrast to the 1–2 min inclusion criterion in drug treatments in Western countries.

The review illustrated a lack of well-designed studies on α-sympatholytic drugs, dopamine-antagonizing drugs and antibiotics, hampering the application of a meta-analysis. Head-to-head comparisons between anesthetic creams and serotonergic or sympatholytic drugs are currently not available, but need to be initiated.

The meta-analysis

In the mixed-methodology analysis, it was demonstrated that the percentage IELT increase was not related to the baseline IELT, irrespectively of the definition of either 1 or 2 min of baseline IELT and the drug dose that was used. The properly executed eight daily treatment studies did not allow to distinguish differences between 1 and 2 min definitions as well as proper dose–response relationships, due to a lack of data for a statistical valid meta-analysis.

The current meta-analysis showed that paroxetine, clomipramine, sertraline and fluoxetine are effective in delaying ejaculation. Remarkably, in spite of the above-mentioned methodological considerations, the meta-analysis showed that the rank order of efficacy was rather similar in the mixed-methodology analysis and in the final eight daily treatment studies. We assume that the almost similar rank orders could only be the result of the robust pharmacological efficacy of paroxetine, regardless of the many confounding variables related to inappropriate trials. It would be a misinterpretation, however, to conclude that methodology and design are of less relevance for drug-treatment trials in premature ejaculation.


The meta-analysis has shown that the rank order of efficacy of SSRIs, clomipramine and placebo is not extremely distorted by methodology and/or design. We explain this phenomenon to be the result of the strong delaying actions of paroxetine, in particular on ejaculation. However, the meta-analysis also demonstrated that open and single-blind studies will lead to an exaggerated response and that retrospective assessment of ejaculation time by a questionnaire or subjective report will lead to far more variability in clinical outcomes. Of all 43 serotonergic antidepressant studies, only eight (18.6%)57, 60, 65, 77, 78, 79, 80, 81 have been conducted according to the complete criteria of evidence-based medical research. Of all 79 studies, only 12 (15.1%)49, 27, 35, 42, 57, 60, 65, 77, 78, 79, 80, 81 have been performed in this way. Based on this systematic review and meta-analysis, it is recommended that genuine evidence-based research on drug treatment of lifelong premature ejaculation should be performed by randomised, double-blind studies, with the prospective real time use of a stopwatch at each intercourse both during a baseline period and during the active drug-treatment trial. The studies would further gain methodological quality when authors would apply the IELT, and mention an operational definition of premature ejaculation with a clear IELT cutoff point, the duration of the baseline period, and both the absolute IELT values as the percentage increase of the IELTs. The current meta-analysis provided clear empirical evidence that, in order to avoid unacceptable variabilities of outcome measure, open and single-blind studies with a questionnaire or subjective assessment of the IELT should be avoided. Editors and peer reviewers should be aware of the clinical relevance of above-mentioned methodological criteria for the outcome of drug-treatment studies of premature ejaculation.


  1. 1

    Waldinger MD . The neurobiological approach to premature ejaculation (review article). J Urol 2002; 168: 2359–2367.

    PubMed  Google Scholar 

  2. 2

    Waldinger MD, Olivier B . Selective serotonin reuptake inhibitors (SSRIs) and sexual side effects: differences in delaying ejaculation. In: Sacchetti E, Spano P (eds) Advances in Preclinical and Clinical Psychiatry, Vol I: Fluvoxamine: Established and Emerging Roles in Psychiatric Disorders. Excerpta Medica: Milan, Italy, 2000, pp 117–130.

    Google Scholar 

  3. 3

    Waldinger MD . Chapter 15: Rapid ejaculation. In: Levine SB, Risen CB, Althof SE (eds) Handbook of Clinical Sexuality for Mental Health Professionals. Brunner-Routledge: New York, 2003, pp 257–274.

    Google Scholar 

  4. 4

    Waldinger MD et al. The SSRIs fluvoxamine and paroxetine differ in sexual inhibitory effects after chronic treatment. Psychopharmacology 2001; 160: 283–289.

    Google Scholar 

  5. 5

    Waldinger MD et al. Premature ejaculation and SSRI-induced delayed ejaculation: the involvement of the serotonergic system. Behav Brain Res 1998; 92: 111–118.

    CAS  PubMed  Google Scholar 

  6. 6

    Waldinger MD, Olivier B . Selective serotonin reuptake inhibitor-induced sexual dysfunction: clinical and research considerations. Int Clin Psychopharmacol 1998; 13(Suppl 6): S27–S33.

    PubMed  Google Scholar 

  7. 7

    Olivier B, van Oorschot R, Waldinger MD . Serotonin, serotonergic receptors, selective serotonin reuptake inhibitors and sexual behaviour. Int Clin Psychopharmacol 1998; 13(Suppl 6): S9–S14.

    PubMed  Google Scholar 

  8. 8

    Waldinger MD et al. Familial occurrence of primary premature ejaculation. Psychiatr Genet 1998; 8: 37–40.

    CAS  PubMed  Google Scholar 

  9. 9

    Waldinger MD . Lifelong premature ejaculation: from authority based to evidence based medicine. Br J Urol Int, 2004 (in press).

  10. 10

    Waldinger MD, Hengeveld MW, Zwinderman AH, Olivier B . An empirical operationalization study of DSM-IV diagnostic criteria for premature ejaculation. Int J Psychiatry Clin Pract 1998; 2: 287–293.

    CAS  PubMed  Google Scholar 

  11. 11

    Waldinger MD . Towards evidence-based drug treatment research on premature ejaculation: a critical evaluation of methodology. J Impot Res 2003; 15: 309–313.

    CAS  Google Scholar 

  12. 12

    Moser A . Die Verwendung von Nupercainal in Geburtshilfe, Gynakologie und Eheberatung (German). Wien Med Wochnschr 1953; 103: 814–816.

    CAS  Google Scholar 

  13. 13

    Schöning FW . Behandlungsergebnisse bei Ejakulatio praecox mit opipramol (Insidon) (German). Z Haut-Geschl Kr 1972; 47: 135–140.

    Google Scholar 

  14. 14

    Wiederholt I . Behandlung der Ejaculatio Praecox mit einem Antiandrogen. (German). Psychiat Prax 1977; 4: 176–180.

    CAS  Google Scholar 

  15. 15

    Porto R . Essai en double aveugle de la clomipramine dans léjaculation premature (French). Med Hygiene 1981; 39: 1249–1253.

    Google Scholar 

  16. 16

    Nielsen NP . L' ejaculazione precoce e il suo trattamento: nota preliminare sull'attivita' del caroxazone (Italian). Rass Studi Psichiatr 1981; 70: 1065–1080.

    Google Scholar 

  17. 17

    Giammusso B, Morgia G, Spampinato A, Motta M . La paroxetina nel trattamento dell'eiaculazione precoce. Arch Ital Urol Androl 1997; 69: 11–13.

    CAS  PubMed  Google Scholar 

  18. 18

    Beretta G et al. Trattamento con un alfa-bloccante (fenossibenzamina) della eiaculazione precoce (Italian). Minerva Urol Nefrol 1988; 40: 29–32.

    CAS  PubMed  Google Scholar 

  19. 19

    Martinez Hurtado M, Gonzales Menendez R, Idelfonso Alavarez E, Goicolea Maiza S . La eyaculacion precoz: caracteristicas clinicas y respuesta al tratamiento con guanetidina (Spanish). Rev Hosp Psiquiatr Habana 1981; 22: 105–116.

    Google Scholar 

  20. 20

    Pico Costero J, Abreu de la Torre C, Roque Banos YL . Eyaculacion precoz. Tratamiento con tioridacina (Spanish). Rev Hosp Psiquiatr Habana 1986; 27: 277–280.

    Google Scholar 

  21. 21

    De Almeida Claro J et al. Experiencia brasileira com a trazodona no tratamento da ejaculacao precoce (Portugese). Rev Brasil Med 2000; 57: 931–934.

    Google Scholar 

  22. 22

    Hronek J, Kolomaznik M . Farmakoterapie prekotneho prubehu pohlavniho styku. (Czech). Cesk Psychiatr 1986; 82: 303–306.

    CAS  PubMed  Google Scholar 

  23. 23

    Kolomaznik M, Hronek J, Kolomaznik J, Kufner P . Sertralin v leche ejaculatio praecox: Intermitentne nebo kontinualne? (Czech). Psychiatr C'slo 1998; 2: 244–246.

    Google Scholar 

  24. 24

    Kolomaznik M et al. Fluoxetin v lecbe ejaculatio praecox (Czech). Psychiatrie 2002; 6: 6–9.

    Google Scholar 

  25. 25

    Kery S, Kozma A . Ejaculatio praecox citalopram kezelese (Hungarian). Magyar Urol 1995; 7: 359–364.

    Google Scholar 

  26. 26

    Choi HK, Xin ZC, Cho IR . The effect of herb cream (SS-cream) on premature ejaculation. J Korean Androl Soc 1993; 11: 99–105.

    Google Scholar 

  27. 27

    Xin ZC, Seong DH, Minn YG, Choi HK . A double blind clinical trial of SS-cream on premature ejaculation. Korean J Urol 1994; 35: 533–537.

    Google Scholar 

  28. 28

    Bishop YMM . The δ-method for calculating asymptomatic distributions. In: Bishop YMM, Fienberg SE, Holland PW (eds). Discrete Multivariate Analysis: Theory and Practice. The MIT Press: Cambridge, MA, and London, England, 1975, pp 486–502.

    Google Scholar 

  29. 29

    Schapiro B . Premature ejaculation, a review of 1130 cases. J Urol 1943; 50: 374–379.

    Google Scholar 

  30. 30

    Aycock L . The medical management of premature ejaculation. J Urol 1949; 62: 361–362.

    CAS  PubMed  Google Scholar 

  31. 31

    Damrau F . Premature ejaculation: use of ethyl aminobenzoate to prolong coitus. J Urol 1963; 89: 936–939.

    CAS  PubMed  Google Scholar 

  32. 32

    Berkovitch M, Keresteci AG, Koren G . Efficacy of prilocaine–lidocaine cream in the treatment of premature ejaculation. J Urol 1995; 154: 1360–1361.

    CAS  PubMed  Google Scholar 

  33. 33

    Xin ZC, Choi YD, Lee SH, Choi HK . Efficacy of a topical agent SS-cream in the treatment of premature ejaculation: preliminary clinical studies. Yonsei Med J 1997; 38: 91–95.

    CAS  PubMed  Google Scholar 

  34. 34

    Choi HK et al. Safety and efficacy study with various doses of SS-cream in patients with premature ejaculation in a double-blind, randomised, placebo controlled clinical study. Int J Impot Res 1999; 11: 261–264.

    CAS  PubMed  Google Scholar 

  35. 35

    Choi HK et al. Clinical study of SS-cream in patients with lifelong premature ejaculation. Urology 2000; 55: 257–261.

    CAS  PubMed  Google Scholar 

  36. 36

    Slob AK, van Berkel A, van der Werff ten Bosch JJ . Premature ejaculation treated by local penile anaesthesia in an uncontrolled clinical replication study. J Sex Res 2000; 37: 244–247.

    Google Scholar 

  37. 37

    Atikeler MK, Gecit I, Senol FA . Optimum usage of prilocaine–lidocaine cream in premature ejaculation. Andrologia 2002; 34: 356–359.

    CAS  PubMed  Google Scholar 

  38. 38

    Bartova D, Bouchal M . Thioridazine treatment of ejaculatio praecox. Act Nerv Super (Praha) 1965; 7: 244–245.

    CAS  Google Scholar 

  39. 39

    Mellgren A . Treatment of ejaculatio praecox with thioridazine. Psychother Psychosom 1967; 15: 454–460.

    Google Scholar 

  40. 40

    Falaschi P et al. Brain dopamine and premature ejaculation: results of treatment with dopamine antagonists. In: Gessa GL, Corsini GU (eds) Apomorphine and Other Dopaminomimetics. Vol. 1: Basic Pharmacology. Raven Press: New York, 1981, pp 117–121.

    Google Scholar 

  41. 41

    Wabrek AJ . Effect of metoclopramine on ejaculatory threshold in a group of premature ejaculators. Neurourol Urodyn 1984; 3: 155–161.

    Google Scholar 

  42. 42

    Greco E, Polonio-Balbi P, Speranza JC . Levosulpiride: a new solution for premature ejaculation. Int J Impot Res 2002; 14: 308–309.

    CAS  PubMed  Google Scholar 

  43. 43

    Bennett D . Treatment of ejaculatio praecox with monoamine-oxidase inhibitors (letter). Lancet 1961; 9: 1309.

    Google Scholar 

  44. 44

    Shilon M, Paz GF, Hommonai ZT . The use of phenoxybenzamine treatment in premature ejaculation. Fertil Steril 1984; 42: 659–661.

    CAS  PubMed  Google Scholar 

  45. 45

    Beretta G, Chelo E, Fanciullacci F, Zanollo A . Effect of an alpha-blocking agent (phenoxybenzamine) in the management of premature ejaculation. Acta Eur Fertil 1986; 17: 43–45.

    CAS  PubMed  Google Scholar 

  46. 46

    Cavallini G . Alpha-1 blockade pharmacotherapy in primitive psychogenic premature ejaculation resistant to psychotherapy. Eur Urol 1995; 28: 126–130.

    CAS  Google Scholar 

  47. 47

    Boneff AN . Topical treatment of chronic prostatitis and premature ejaculation. Int Urol Nephrol 1971; 4: 183–186.

    Google Scholar 

  48. 48

    Brown AJ . Ciprofloxacin as cure of premature ejaculation (case report). J Sex Marital Ther 2000; 26: 351–352.

    CAS  PubMed  Google Scholar 

  49. 49

    Cooper AJ, Magnus RV . A clinical trial of the beta blocker propranolol in premature ejaculation. J Psychosom Res 1984; 28: 331–336.

    CAS  PubMed  Google Scholar 

  50. 50

    Segraves RT . Treatment of premature ejaculation with lorazepam (letter). Am J Psychiatry 1987; 144: 1240.

    CAS  PubMed  Google Scholar 

  51. 51

    Fein RL . Intracavernous medication for treatment of premature ejaculation. Urology 1990; 35: 301–303.

    CAS  PubMed  Google Scholar 

  52. 52

    Eaton H . Clomipramine in the treatment of premature ejaculation. J Int Med Res 1973; 1: 432–434.

    Google Scholar 

  53. 53

    Goodman RE . An assessment of clomipramine (Anafranil) in the treatment of premature ejaculation. J Int Med Res Suppl 1980; 3: 53–59.

    Google Scholar 

  54. 54

    Girgis SM, El-Haggen S, El-Hermouzy S . A double-blind trial of clomipramine in premature ejaculation. Andrologia 1982; 14: 364–368.

    CAS  PubMed  Google Scholar 

  55. 55

    Assalian P . Clomipramine in the treatment of premature ejaculation. J Sex Res 1988; 24: 213–215.

    CAS  PubMed  Google Scholar 

  56. 56

    Waldinger MD, Hengeveld MW, Zwinderman AH . Paroxetine treatment of premature ejaculation: a double-blind, randomised, placebo-controlled study. Am J Psychiatry 1994; 151: 1377–1379.

    CAS  PubMed  Google Scholar 

  57. 57

    Althof SE et al. A double-blind crossover trial of clomipramine for rapid ejaculation in 15 couples. J Clin Psychiatry 1995; 56: 402–407.

    CAS  PubMed  Google Scholar 

  58. 58

    Mendels J, Camera A, Sikes C . Sertraline treatment for premature ejaculation. J Clin Psychopharmacol 1995; 15: 341–346.

    CAS  PubMed  Google Scholar 

  59. 59

    Montorsi F et al. Clomipramine for premature ejaculation: a randomized, double blind, placebo controlled study. Acta Urol Ital 1995; 1: 5–6.

    Google Scholar 

  60. 60

    Kara H et al. The efficacy of fluoxetine in the treatment of premature ejaculation: a double-blind, placebo controlled study. J Urol 1996; 156: 1631–1632.

    CAS  PubMed  PubMed Central  Google Scholar 

  61. 61

    Ludovico GM et al. Paroxetine in the treatment of premature ejaculation. Br J Urol 1996; 78: 881–882.

    Google Scholar 

  62. 62

    Lee HS, Song DH, Kim CH, Choi HK . An open clinical trial of fluoxetine in the treatment of premature ejaculation. J Clin Psychopharmacol 1996; 16: 379–382.

    CAS  PubMed  Google Scholar 

  63. 63

    Waldinger MD, Hengeveld MW, Zwinderman AH . Ejaculation retarding properties of paroxetine in patients with primary premature ejaculation: a double-blind, randomised, dose–response study. Br J Urol 1997; 79: 592–595.

    CAS  PubMed  Google Scholar 

  64. 64

    Raju GAR et al. Evaluation of fluoxetine in premature ejaculation. Indian J Pharmacol 1997; 29: 204–205.

    CAS  Google Scholar 

  65. 65

    Waldinger MD, Hengeveld MW, Zwinderman AH, Olivier B . Effect of SSRI antidepressants on ejaculation: a double-blind, randomized, placebo-controlled study with fluoxetine, fluvoxamine, paroxetine and sertraline. J Clin Psychopharmacol 1998; 18: 274–281.

    CAS  PubMed  PubMed Central  Google Scholar 

  66. 66

    Haensel SM, Klem TMAL, Hop WCJ, Slob AK . Fluoxetine and premature ejaculation: a double-blind, crossover, placebo-controlled study. J Clin Psychopharmacol 1998; 18: 72–77.

    CAS  PubMed  Google Scholar 

  67. 67

    Biri H et al. Sertraline in the treatment of premature ejaculation: a double-blind placebo controlled study. Int Urol Nephrol 1998; 30: 611–615.

    CAS  PubMed  Google Scholar 

  68. 68

    McMahon CG . Treatment of premature ejaculation with sertraline hydrochloride: a single-blind placebo controlled crossover study. J Urol 1998; 159: 1935–1938.

    CAS  PubMed  Google Scholar 

  69. 69

    McMahon CG . Treatment of premature ejaculation with sertraline hydrochloride. Int J Impot Res 1998; 10: 181–184.

    CAS  PubMed  Google Scholar 

  70. 70

    Kim SC, Seo KK . Efficacy and safety of fluoxetine, sertraline and clomipramine in patients with premature ejaculation: a double-blind, placebo controlled study. J Urol 1998; 159: 425–427.

    CAS  PubMed  PubMed Central  Google Scholar 

  71. 71

    Balbay MD et al. Treatment of premature ejaculation with sertralin. Int Urol Nephrol 1998; 30: 81–83.

    CAS  PubMed  Google Scholar 

  72. 72

    Basar MM et al. Comparison of sertraline to fluoxetine with regard to their efficacy and side effects in the treatment of premature ejaculation. Arch Esp Urol 1999; 52: 1.008–1.011.

    Google Scholar 

  73. 73

    Yilmaz U et al. The effects of fluoxetine on several neurophysiological variables in patients with premature ejaculation. J Urol 1999; 161: 107–111.

    CAS  PubMed  Google Scholar 

  74. 74

    McMahon CG, Touma K . Treatment of premature ejaculation with paroxetine hydrochloride. Int J Impot Res 1999; 11: 241–245.

    CAS  PubMed  Google Scholar 

  75. 75

    Atan A, Basar MM, Aydoganli L . Comparison of the efficacy of fluoxetine alone vs fluoxetine plus local lidocaine ointment in the treatment of premature ejaculation. Arch Esp Urol 2000; 53: 856–858.

    CAS  PubMed  Google Scholar 

  76. 76

    Rowland DL, De Gouveia Brazao CA, Slob AK . Effective daily treatment with clomipramine in men with premature ejaculation when 25 mg (as required) is ineffective. BJU Int 2001; 87: 357–360.

    CAS  PubMed  Google Scholar 

  77. 77

    Waldinger MD, Zwinderman AH, Olivier B . Antidepressants and ejaculation: a double-blind, randomized, placebo-controlled, fixed-dose study with paroxetine, sertraline, and nefazodone. J Clin Psychopharmacol 2001; 21: 293–297.

    CAS  PubMed  Google Scholar 

  78. 78

    Waldinger MD, Zwinderman AH, Olivier B . SSRIs and ejaculation: a double-blind, randomised, fixed-dose study with paroxetine and citalopram. J Clin Psychopharmacol 2001; 21: 556–560.

    CAS  PubMed  Google Scholar 

  79. 79

    Waldinger MD, Zwinderman AH, Olivier B . Antidepressants and ejaculation: a double-blind, randomised, fixed-dose study with mirtazapine and paroxetine. J Clin Psychopharmacol 2003; 23: 467–470.

    CAS  PubMed  Google Scholar 

  80. 80

    Novaretti JPT, Pompeo ACL, Arap S . Selective serotonin uptake inhibitor in the treatment of premature ejaculation. Brazil J Urol 2002; 28: 116–122.

    Google Scholar 

  81. 81

    Atmaca M, Kuloglu M, Tezcan E, Semercioz A . The efficacy of citalopram in the treatment of premature ejaculation: a placebo-controlled study. Int J Impot Res 2002; 14: 502–505.

    CAS  PubMed  Google Scholar 

  82. 82

    Segraves RT, Saran A, Segraves K, Maguire E . Clomipramine vs placebo in the treatment of premature ejaculation: a pilot study. J Sex Marital Ther 1993; 19: 198–200.

    CAS  PubMed  Google Scholar 

  83. 83

    Haensel SM, Rowland DL, Kallan KTHK, Slob AK . Clomipramine and sexual function in men with premature ejaculation and controls. J Urol 1996; 156: 1310–1315.

    CAS  PubMed  Google Scholar 

  84. 84

    Strassberg DS et al. Clomipramine in the treatment of rapid (premature) ejaculation. J Sex Marital Ther 1999; 25: 89–101.

    CAS  PubMed  Google Scholar 

  85. 85

    Kim SW, Paick J-S . Short-term analysis of the effects of as needed use of sertraline at 5 PM for the treatment of premature ejaculation. Urology 1999; 54: 544–547.

    CAS  PubMed  Google Scholar 

  86. 86

    McMahon C, Touma K . Treatment of premature ejaculation with paroxetine hydrochloride as needed: 2 single-blind, placebo-controlled, crossover studies. J Urol 1999; 161: 1826–1830.

    CAS  PubMed  Google Scholar 

  87. 87

    Abdel-Hamid IA, El Naggar EA, El Gilany AH . Assessment of as needed use of pharmacotherapy and the pause-squeeze technique in premature ejaculation. Int J Impot Res 2001; 13: 41–45.

    CAS  PubMed  PubMed Central  Google Scholar 

  88. 88

    Chia SJ . Management of premature ejaculation – a comparison of treatment outcome in patients with and without erectile dysfunction. Int J Androl 2002; 25: 301–305.

    PubMed  Google Scholar 

  89. 89

    Salonia A et al. A prospective study comparing paroxetine alone versus paroxetine plus sildenafil in patients with premature ejaculation. J Urol 2002; 168: 2486–2489.

    CAS  PubMed  PubMed Central  Google Scholar 

  90. 90

    Riley AJ, Riley EJ . Amitriptyline-perphenazine and the squeeze technique in premature ejaculation. J Pharmacother 1979; 2: 136–140.

    Google Scholar 

  91. 91

    Xin ZC et al. Penile sensitivity in patients with primary premature ejaculation. J Urol 1996; 156: 979–981.

    CAS  PubMed  Google Scholar 

  92. 92

    Xin ZC, Choi YD, Rha KH, Choi HK . Somatosensory evoked potentials in patients with premature ejaculation. J Urol 1997; 158: 451–455.

    CAS  PubMed  Google Scholar 

  93. 93

    Truitt WA, Coolen LM . Identification of a potential ejaculation generator in the spinal cord. Science 2002; 297: 1566–1569.

    CAS  PubMed  PubMed Central  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to M D Waldinger.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Waldinger, M., Zwinderman, A., Schweitzer, D. et al. Relevance of methodological design for the interpretation of efficacy of drug treatment of premature ejaculation: a systematic review and meta-analysis. Int J Impot Res 16, 369–381 (2004).

Download citation


  • premature ejaculation
  • meta-analysis
  • paroxetine
  • SSRI
  • serotonin

Further reading


Quick links