For about a century various psychological treatments have been proposed to treat lifelong premature ejaculation (PE). However, the psychological treatments and underlying theories mostly relied on the opinions of leading psychotherapists and sexologists, and were not based on controlled studies but on anecdotal experience, a typical example of authority‐based research. Despite the availability of clomipramine , an effective ejaculation‐delaying tricyclic antidepressant, in the early 1970s, possible secondary medical use with clomipramine to treat PE remained unrecognized. In contrast to authority‐based research, evidence‐based medicine (EBM)  has been accepted as the current standard for clinical research and medical practice. Particularly in last decade randomized clinical trials with clomipramine [3, 4] and some selective serotonin reuptake inhibitors (SSRIs) [5-9] have repeatedly shown the efficacy of serotonergic antidepressants in delaying ejaculation. However, even now, despite evidence‐based drug treatment research, many sexologists maintain their view that PE is a psychological disorder caused by learned behaviour or unconscious psychological conflicts. The aim of this article is to review the current scientific state of PE by applying the principles of EBM to both the psychological and neurobiological approaches to the disorder and its treatments.
HISTORICAL DEVELOPMENT OF THEORIES
Waldinger  distinguished four periods in the approach to and treatment of PE. Throughout these periods PE has been considered both from a medical and a psychological view, resulting in contrasting psychotherapeutic and drug treatment approaches.
THE FIRST PERIOD (1887–1917): RAPID EJACULATION
In 1887 Gross  described the first case of rapid ejaculation; a second report by von Krafft‐Ebing  followed in 1901. Although publications were rare, notably during the first 30 years of its existence in the medical literature, rapid ejaculation was viewed as an abnormal phenomenon and not particularly as a psychological disturbance.
THE SECOND PERIOD (1917–50): NEUROSIS AND PSYCHOSOMATIC DISORDER
In 1917 Abraham  described rapid ejaculation as ejaculatio praecox and stated that it was a symptom of a neurosis caused by unconscious conflicts, maintaining that treatment should consist of classic psychoanalysis. On the other hand, some physicians stated that PE was a result of anatomical urological abnormalities, e.g. too short a foreskin frenulum or changes in the posterior urethra, which had to be treated with incision of the foreskin or electrocautery of the verumontanum. In 1943, Schapiro  argued that PE was neither a pure psychological nor a pure somatic disorder, but a psychosomatic disturbance caused by a combination of a psychologically overanxious constitution and a weak ejaculatory system. Schapiro described two types of PE; type B, in which rapid ejaculation existed from the first intercourse, and type A, which led to erectile dysfunction. Many years later both types became distinguished as the primary (lifelong) and secondary (acquired) forms of PE .
THE THIRD PERIOD (1950–90): LEARNED BEHAVIOUR
The biological component of PE and therefore also drug treatment, advocated by Schapiro, was ignored by most sexologists, who advocated psychoanalytical treatment. This neglect became even more pronounced after Masters and Johnson  claimed high success rates for behavioural therapy in the form of the ‘squeeze’ technique, an adaptation of the stop‐start technique published by Semans  in 1956. Masters and Johnson stated that men with PE had learned to be rapid because their initial attempts at sexual intercourse were hurried.
THE FOURTH PERIOD (1990 TO PRESENT): NEUROBIOLOGY AND GENETICS
Although behavioural treatment for PE prevailed in sexology until the present, very few studies have been conducted to investigate its effectiveness. In contrast, since the 1990s there have been many publications on the efficacy of SSRIs, clomipramine and topical anaesthetic ointments in delaying ejaculation. At the same time a new neurobiological view has been developed, arguing that PE is related to disturbances of serotonin metabolism in specific areas of the CNS and a possible genetic vulnerability is postulated [10, 18].
EBM means that the formulation of a seemingly attractive hypothesis of the cause of a disease is not enough for scientific acceptance. Instead, empirical evidence must be produced, preferably replicated in various controlled studies.
The psychoanalytic idea of unconscious conflicts being the cause of PE has never been investigated in a manner that allowed generalization, as only case reports on psychoanalytic therapy have been published. However, this is also true for behavioural therapy. Masters and Johnson  deliberately refuted a definition of PE in terms of a man's duration of ejaculation. Instead, they insisted on defining PE in terms of the female partner's response, e.g. as a man's inability to inhibit ejaculation long enough for the partner to reach orgasm in half of the attempts at intercourse. It is obvious that their definition is inadequate because it implies that any partners of a women who has difficulty in reaching orgasm in half the attempts have PE.
Masters and Johnson argued that PE was conditioned by experiencing first sexual contacts hurriedly (e.g. on the back seat of a car, or with an impatient prostitute). However, Masters and Johnson, and sexologists who followed their ideas, provided no evidence‐based data for this assumption. As to their proposed behavioural ‘squeeze’ technique, Masters and Johnson claimed 97% success for delaying ejaculation, but this high percentage of success has never been replicated by others.
Usually a lack of reproducible data leads to critical comments; this is one of the basic principles of EBM, as the effects of a treatment intervention should be reproducible by others. However, critical comments were not appreciated in traditional sexological thinking of the late 20th century. This unscientifically supported and uncritical belief in behavioural treatment remains today, despite clear EBM research in favour of the neurobiological view. The criticism is justified; the methodological insufficiencies of the report of Masters and Johnson are very serious. Their report on the efficacy of the squeeze method contains numerous biases.
First, there was a bias in selection and allocation of the subjects; the patients were not randomized to either the new squeeze technique, the older stop‐start technique or a ‘nonsense’ behavioural technique. Second, the treatment design was open and not double‐blinded. Furthermore, the diagnosis of PE was not quantified and therefore inaccurate, particularly as Masters and Johnson used an obscure definition of PE. Baseline data were not reported and inclusion and exclusion criteria were lacking. The assessment of success was subjectively reported with no quantification or scoring scales. In addition, Masters and Johnson provided no information on their data processing. Despite all these methodological flaws their behavioural technique has been accepted uncritically worldwide and promoted as the first choice of treatment. Even the very poor results of two studies [19, 20] on behavioural therapy (also poorly designed) could not prevent sexologists from continue to claim the squeeze technique as the first choice of treatment. Not only the squeeze technique, but also other types of psychotherapy, including ‘thought stopping’, Gestalt therapy, transactional analysis, group therapy and bibliotherapy, were all proposed as being effective [21-24]. The efficacy of these psychotherapies has also only been suggested in case reports and was never investigated in well designed controlled studies.
How can the uncritical acceptance of the squeeze technique as the first choice of treatment be explained? In the 1960s, Masters and Johnson were highly esteemed professionals who worked with patients with sexual disorders. Their formulated and focused treatment approach to sexual dysfunction was revolutionary at that time. They soon became experts in the field and considered as authorities. Clearly, the uncritical acceptance of their squeeze method as the ideal treatment is an example of opinion‐ or authority‐based medicine.
In contrast to the easy acceptance of behavioural treatment by sexologists, drug treatment had to prove itself far more explicitly against the rejection of professionals in the field. Only a few physicians have tried to develop drug strategies to treat PE. Currently, despite the remaining ambiguous attitudes of many sexologists, drug treatment with serotonergic antidepressants is accepted as effective therapy. Despite all the circumstantial evidence, a scientific approach to investigate empirical evidence remains obligatory . To investigate to what extent differences in methods may influence the clinical outcome of drug treatment studies, Waldinger et al. conducted a systematic review and meta‐analysis of all drug treatment studies published from 1943 to the present. In that study several methodological evidence‐based criteria were compared, e.g. study design (open vs double‐blind), tools for diagnostic testing (stopwatch vs subjective reporting or questionnaire) and ways of assessment (prospective vs retrospective). The results revealed that of 79 publications on drug treatment, 35 involved serotonergic antidepressants. It was clear that single‐blind or open studies and studies using a questionnaire or subjective report on the ejaculation time led to a higher variability in ejaculatory delay. Only eight studies [3, 6, 8, 27-31] fulfilled all criteria of EBM, e.g. double‐blind studies prospectively using a real‐time stopwatch assessment at each attempt at intercourse, both at baseline and during the drug trial. For daily treatment a similar efficacy for paroxetine, clomipramine, sertraline and fluoxetine was confirmed, whereas that for paroxetine was clearly better than all the aforementioned drugs. Based on this meta‐analysis paroxetine is proposed as the reference standard treatment for lifelong PE .
OPERATIONAL DEFINITION OF PE
For evidence‐based research it is most important to have a definition of PE, but because there are conflicting ideas about the essentials of PE sexologists have never reached agreement on such a definition. DSM‐IV  defines PE as ‘persistent or recurrent ejaculation with minimal sexual stimulation before, upon, or shortly after penetration and before the person wishes it’. Until recently, any scientific basis for the DSM‐IV definition was lacking. For instance, the meaning of ‘persistent’, ‘recurrent’, ‘minimal’ and ‘shortly after’ is vague and certainly needs further qualification. To obtain an empirically operational definition Waldinger et al. investigated 110 consecutively enrolled men with lifelong PE. In that study, men and their female partners were instructed to use a stopwatch at home during each coitus, for a period of 4 weeks (Fig. 1). About 10% of the men ejaculated at 1–2 min but most (90%) ejaculated within 1 min of intromission, and 80% were actually ejaculating within 30 s, whereas 60% ejaculated within 15 s. The age of the men and duration of the relationship were not correlated with the ejaculation time. Based on this study, Waldinger et al. empirically defined lifelong PE as an ejaculation of <1 min in > 90% of episodes of sexual intercourse, independent of age and duration of relationship. This definition defines PE as being a rapid ejaculation that is independent of psychological or relationship distress. Thus, assessment by stopwatch revealed that PE is a matter of seconds and not minutes. In this respect the ICD‐10 definition (e.g. ejaculation in ≤ 15 s) seems more appropriate than the DSM‐IV definition, but both need to be adapted to these recent data.
A NEW THEORY OF PE
Waldinger et al.[10, 34] formulated a new theory of the cause and genesis of lifelong PE; they postulated that lifelong rapid ejaculation is not an acquired disorder caused by initial hurried intercourse, as suggested by Masters and Johnson. Instead, rapid ejaculation is postulated to be part of a normal biological variability of the intravaginal ejaculation latency time (IELT) in men, with a possible familial genetic vulnerability [10, 18, 34]. The IELT is defined as the time between vaginal penetration and intravaginal ejaculation [5, 6]. Thus rapid ejaculation is considered to be primarily a neurobiological phenomenon, that may or may not secondarily lead to psychological or psychosocial distress. Dependent on intra‐ and interpersonal and probably also cultural factors, rapid ejaculation may become perceived as PE. Both animal and large‐scale human epidemiological stopwatch studies are needed to confirm the existence of a biological continuum of the IELT.
Based on animal and human psychopharmacological studies, Waldinger et al.[10, 35, 36] further postulated that lifelong PE is related to decreased central serotonergic neurotransmission, 5‐hydroxytryptamine (HT)2C receptor hyposensitivity and/or 5‐HT1A receptor hypersensitivity. Treatment should therefore consist of 5‐HT2C receptor stimulation and/or 5‐HT1A receptor inhibition. Evidence for the role of the 5‐HT2C receptor was found in four stopwatch studies in men with PE [6, 27-29], where 5‐HT2C receptor‐stimulating and the 5‐HT2C‐blocking antidepressants exerted an ejaculation delay and absence of ejaculation delay, respectively.
EJACULATION THRESHOLD HYPOTHESIS
To understand the suggested biological variation in the IELT in relation to the serotonergic system, Waldinger et al.[10, 34] proposed the existence of a threshold IELT. If there is a low threshold men can sustain only a little sexual arousal before ejaculation. Whatever these men do or fantasise during intercourse, they ejaculate easily even when they are not fully aroused. The low threshold is assumed to be associated with low 5‐HT neurotransmission and probably hypofunction of the 5‐HT2C receptor and/or a hyperfunction of the 5‐HT1A receptor.
If the threshold is higher men will have more control over their ejaculation time; they can sustain more sexual arousal before ejaculating. In these men 5‐HT neurotransmission varies around a normal or average level and the 5‐HT2C receptor functions normally. These men have the neurobiological ability to voluntarily decide to ejaculate quickly or after a longer duration of intercourse.
If the threshold is high or very high men may experience difficulty in ejaculating or cannot ejaculate even when fully sexually aroused. At a high threshold 5‐HT neurotransmission is supposedly increased, 5‐HT2C receptor sensitivity is enhanced and/or the 5‐HT1A receptor sensitivity decreased.
In general, SSRIs and clomipramine activate the 5‐HT2C receptor and therefore switch the threshold to a higher level, leading to a delay in ejaculation. Stopping treatment results in a uniform resetting of the threshold within 3–5 days to the lower individually determined reference level, that is assumed to be genetically determined.
COURSE OF RAPIDITY
It is generally believed that ageing delays ejaculation; this assumption might be true for men with a normal or average ejaculation time but has never been investigated in men with PE. However, in a stopwatch study  of 110 consecutively enrolled men (aged 18–65 years) with lifelong PE, 76% reported that throughout their lives their speed of ejaculation had remained as rapid as at the first sexual contacts in puberty and adolescence; 23% reported that it had become even faster with ageing and only 1% reported that it had become slower. From these data it is questionable whether the fixed rapidity and even paradoxical shortening of the IELT while ageing should be recognized as a part of the pathogenetic process of PE. Rapid ejaculation is part of a normal biological variation of the IELT in men, but its paradoxical or fixed course throughout life is considered as being pathological. Chronic PE appears to be the clinical syndrome of primary (lifelong) PE. As yet, there is no real cure for lifelong PE, although drugs may alleviate the symptoms, but only while they are being taken.
Together with the increasing number of clinical studies on the psychopharmacological treatment of PE there has been increasing evidence from animal neuroanatomical and neuropharmacological studies that ejaculation is regulated by various areas in the CNS, and that the rapidity of ejaculation is controlled by neurotransmitters, like serotonin and dopamine, in specific areas in the CNS. Most knowledge of the functional neuroanatomy of ejaculation is derived from studies of male rats. In the copulatory behaviour of male rats, there is a need to distinguish between brain, brainstem and spinal cord regions that become activated before and after ejaculation, when sensory information returns from the genitals (Fig. 2). The medial preoptic area (MPOA) in the rostral hypothalamus and the nucleus paragigantocellularis (nPGi) in the ventral medulla [37, 38] are suggested to be important in the process leading towards ejaculation. Electrical stimulation of the MPOA promotes ejaculation . It is hypothesized that ejaculation is tonically inhibited by serotonergic pathways descending from the nPGi to the lumbosacral motor nuclei. The present hypothesis is that the nPGi itself is inhibited by inhibitory stimuli from the MPOA. Dis‐inhibition of the nPGi is supposed to lead to ejaculation. The discovery of serotonergic neurones in the nPGi and the well‐known ejaculation delay induced by serotonergic antidepressants suggests an action of SSRIs on the nPGi. However, the precise location in the CNS on which SSRIs act to inhibit ejaculation has not yet been determined.
However, brain areas activated as a result of the occurrence of one or more ejaculations have been detected in several mammals . Using the expression of the immediate early gene, c‐fos, as a marker for neural activity in male rats, Coolen et al.[40-44] showed the presence of distinct ejaculation‐related neural activation in several brain regions after ejaculation; the posteromedial part of the bed nucleus of the stria terminalis, a lateral subarea in the posterodorsal part of the medial amygdala, the posterodorsal preoptic nucleus and the medial part of the parvicellular subparafascicular nucleus of the thalamus. These brain regions containing ejaculation‐induced activation are extensively interconnected and reciprocally connected with the MPOA , forming an ejaculation‐related subcircuit within the larger brain circuits underlying male sexual behaviour . The functional significance of this ejaculation‐subcircuit is still poorly understood but it might be that these areas have a role in ‘satiety’ and thus in mediating the postejaculatory interval.
Recently, an important study by Truitt and Coolen  highlighted the role of the lumbar spinal cord in the processing of ejaculation. In male rats they identified a group of lumbar spinothalamic (LSt) cells that are specifically activated after ejaculation and provide direct genital sensory inputs to the subparafascicular nucleus in the thalamus and the ejaculation‐related subcircuit in the brain. The LSt cells also project to sympathetic and parasympathetic neurones related to the genitals. It is suggested that the LSt cells contribute to triggering the ejaculatory reflex and to the sensation of ejaculation, i.e. orgasm.
These and other animal studies clearly show the existence of neural circuitry for ejaculation in mammals. The role of the recently identified LSt cells for PE remains to be elucidated. Data from fundamental animal research remain basic for understanding the neurobiological foundations of rapid ejaculation.
Interestingly, a neurophysiological study  showed shorter latencies and greater amplitudes of somatosensory evoked potentials from the (glans) penis in men with lifelong PE than in matched normal controls. This study and studies with sacral‐evoked potentials [46-48] suggest a sensory hyperexcitability. Currently, a key issue for research is the question of whether rapid ejaculation is caused by a higher excitability in the genital part of the somatosensory cortex (sensory input side) or by an inability to delay ejaculation (motor output side) or both. Brain imaging studies (e.g. PET studies) in humans are needed to unravel the neural substrate of the ejaculatory process in men and may contribute to a better understanding of which part of the neural circuitry is disturbed in PE.
There are three drug treatment strategies: (i) daily treatment with serotonergic antidepressants; (ii) ‘as‐needed’ treatment with antidepressants; and (iii) anaesthetic topical ointments. Daily treatment can be with paroxetine (20–40 mg), clomipramine (10–50 mg), sertraline (50–100 mg) and fluoxetine (20–40 mg). A meta‐analysis of all drug treatment studies showed that paroxetine causes the greatest ejaculation delay . Paroxetine, sertraline and fluoxetine may give rise to side‐effects like fatigue, yawning, mild nausea, loose stools or perspiration. These side‐effects often start in the first week after intake and gradually disappear within 2–3 weeks. Ejaculation delay usually manifests at the end of the second week and sometimes even earlier with a daily oral dose. Except for fluoxetine it is advised not to stop the SSRIs acutely, but to do so gradually over 3–4 weeks, to avoid withdrawal symptoms. The side‐effects of clomipramine may consist of nausea, dry mouth and fatigue. Sometimes clomipramine and the SSRIs may give rise to reversible feelings of diminished libido or moderately decreased rigidity of the penis. The SSRIs may very rarely lead to a prolonged bleeding time and even more rarely also to priapism. It is advisable to inform patients of all these side‐effects when starting treatment.
Since 1993 only eight studies [4, 49-55] were published investigating an as‐needed (on‐demand) strategy in the treatment of PE. Because of the few studies and their inadequate designs, a meta‐analysis  was insufficiently powered to provide final conclusions about differences in efficacy and dose‐relationships. Despite these scientific limitations, clomipramine (10–50 mg) taken at least 4–6 h before intercourse may be effective, lasting for at least 15 h. Another strategy is the daily use of paroxetine, sertraline and fluoxetine in a low dose combined with as‐needed higher doses shortly before intercourse.
The use of topical anaesthetic ointments is probably the oldest treatment for delaying ejaculation, but only a few controlled studies have been reported. In one the results of lidocaine‐prilocaine cream 10 min before intercourse were described . In the Far East good results were reported with SS‐cream, a regionally manufactured cream consisting of various herbs [57-59], used 1–2 h before intercourse.
EBM supports the decisions that must be made in medicine by providing evidence from good scientific research. By critically appraising research evidence it appears that there is none to support the hypothesis of Masters and Johnson that PE is caused by learned behaviour. The worldwide use of the squeeze technique seems to be more related to authority‐based suggestions than EBM. In contrast, a meta‐analysis of 35 studies with serotonergic antidepressants showed the efficacy of paroxetine, clomipramine, sertraline and fluoxetine in delaying ejaculation, paroxetine being the most effective. The first choice of treatment for PE should thus be drug treatment. However, psychotherapy is only indicated for those men or couples who cannot cope with or cannot accept rapid ejaculation . In contrast to the classical psychological view, the purpose of psychotherapy, whether psychodynamic or cognitive, is not to learn how to delay ejaculation but how to cope with rapid ejaculation .
Animal and human studies suggest that PE is related to a decreased serotonergic neurotransmission in specific areas of the CNS and to hyposensitivity and/or hypersensitivity of the 5‐HT2C and 5‐HT1A receptor, respectively. Animal and large‐scale human epidemiological studies are needed to investigate the hypothesis that rapid ejaculation is part of a normally distributed biological continuum of the IELT in the general male population, neglecting the natural course of ageing‐induced ejaculation delay.