Study Type – Symptom prevalence (retrospective cohort)
Level of Evidence 2b
Study Type – Symptom prevalence (retrospective cohort)
To investigate temporal continuity in ejaculatory dysfunction by comparing self-reported experiences of premature ejaculation (PE) at first intercourse with self-reported PE and delayed ejaculation at present, and to clarify whether and to what extent genetic or environmental factors affect continuity in ejaculatory dysfunction, as previous studies indicate moderate heritability for PE at first intercourse.
SUBJECTS AND METHODS
The study comprised retrospective self-reported data on ejaculatory performance at first sexual intercourse and a concurrent self-report of the same at the time of data collection in a population-based sample of 2633 Finnish twins and their siblings aged 18–48 years (mean 26.63, sd 4.68). The continuity of ejaculatory function was assessed by correlation and multiple regression. Reasons for continuity were separated into genetic and environmental sources using twin-model fitting.
Ejaculatory function, particularly PE, was stable over time. Genetic effects accounted for ≈30% of the variance in PE both at first intercourse and when measured at data collection. Unshared environmental effects accounted for most of the variance (≈70%). Genetic effects were almost identical between the sample occasions, but there was a substantial discrepancy between unshared environmental effects affecting PE at first intercourse and unshared environmental effects affecting PE later in life. Age effects were generally negligible. Data were self-reported and retrospective, and thus vulnerable to response bias.
Ejaculatory dysfunction seems to be temporally stable both in the short and long term. Genes that contribute to the variance in PE at first intercourse are similar to those that contribute to the variance in PE later in life, whereas there are, in this regard, substantial differences in the unshared environmental factors that are a cause of PE.
PE at first intercourse
(intravaginal) ejaculation latency time
Complex Samples General Linear Model
degrees of freedom.
Although premature ejaculation (PE) is regarded as one of the most common sexual dysfunctions in men , little is known about its persistence over time. No long-term follow-up study has been conducted to investigate the temporal dimension of ejaculatory dysfunction (EjD). However, stability of ejaculatory problems is very much expected, and even taken for granted both in the academic and clinical setting. For example, the very definition of lifelong PE , as suggested and accepted by the International Society of Sexual Medicine, implies that the condition is stable and permanent. However, the definition itself only refers to the present: ‘Lifelong PE [. . .] is characterized by ejaculation that always or nearly always occurs [. . .] and inability to delay ejaculation on all or nearly all vaginal penetrations’. There is no suggested period for the duration of the condition.
While several studies have indeed measured ejaculatory function over short periods (especially clinical drug-trial studies), relatively few have reported the test-retest reliability statistics for the measures. Symonds et al. reported intraclass correlation coefficients of 0.73 for test-retest reliability on patient-reported outcomes during a 4-week study with data collected on three separate occasions during this time, while Patrick et al. found intraclass correlations of 0.63–0.87 for intravaginal ejaculation latency time (IELT) and patient-reported outcomes in subjects with PE and their partners over a 2-week period. A Belgian study found high within-subject repeatability for ejaculation latency times with intraclass correlations of 0.88–0.93 . Althof et al. found test-retest reliability correlations of 0.70–0.90 for a questionnaire measuring sexual satisfaction, control and distress associated with PE for a period of 7–10 days. These findings indicate a substantial temporal continuity in EjD, at least over short periods. For longer periods there are no such statistics, but the stability of PE over longer periods is indirectly suggested by Pryor et al., whose participants reported a mean duration of PE of ≈16 years in a clinical drug-trial study. The temporal stability of PE is also implied by Hawton et al., who found that 75% of men who initially responded to behavioural therapy for PE showed no lasting improvement after 3 years of follow-up.
It is generally agreed that PE is not explained by one cause ; affective and cognitive psychological experiences , genetics [12,13], neurobiology , physical hypersensitivity (of, e.g. the glans penis ), and frequency of sexual intercourse and early sexual experiences  have all been suggested as (partly) causing PE. However, some of these aetiological factors are disputed (e.g. physical hypersensitivity ). By contrast, delayed ejaculation (DE) is less studied, but it has been suggested that various surgical and medical conditions could cause DE, the foremost of which is the use of anti-adrenergic or neuroleptic drugs . Waldinger and Schweitzer  note that several neurotransmitters have been found to delay (or even inhibit) ejaculation, and suggest a neurobiological cause for DE.
There is some evidence to indicate that sexual dysfunctions might be more common at first intercourse than in the general population. In a study by Santtila et al., 29.5% of the participants retrospectively reported ejaculating within 1 min from penetration, with 3.7% reporting experiencing ejaculation before penetration. In comparison, the proportion of men with IELTs of <1 min is ≈0.5% in stopwatch-based studies . It is not certain what would cause PE to be more common during first intercourse, but recent studies suggested that age has a rather weak, if any, effect on ejaculatory function, and that this effect actually increases problems of PE with age [12,20]. No studies have been conducted on young (<18 years old) adolescents, which would leave age effects in very young men and adolescents undetected. Santtila et al. found some contextual factors, such as alcohol or drug intoxication, and having sex with a completely unknown sexual partner, to be negatively associated with PE at first intercourse. Affective reactions, such as anxiety, might also be causal. Some contextual factors (e.g. peer pressure) might be substantially more common and severe in adolescents, which could (at least partially) explain the discrepancy in prevalence rates for PE at first intercourse and PE measured in sexually more experienced adults. Santtila et al. also found a significant heritable component for PE at first intercourse, with 22% of its variance accounted for by genetic effects and 78% by unshared environmental effects.
In summary, PE both at first intercourse and measured later in life have a moderate heritable component, while most of the variance is accounted for by unshared environmental effects. The different aetiological factors that contribute to PE suggested in previous reports support this view, with some factors (such as partner choice, drug intoxication or peer pressure) being examples of environmental components that could contribute to PE; these components are also unlikely to be very stable over time. Other aetiological factors (e.g. neurological) could instead be assumed to be genetic. However, environmental factors might also contribute to the temporal stability of EjD. For example, continuous medication that alters ejaculatory function would be an environmental factor that could be (at least partly) causal of EjD over a long time, whereas contextual factors such as the partner being unknown could be a situational factor that has an effect on ejaculatory performance, but not with a temporally stable effect. Also, genetically determined aetiological factors might also be situational (e.g. endocrine effects related to puberty).
The aim of the present study was two-fold. First, we attempted to investigate the temporal continuity in EjD, by comparing self-reported experiences of PE at first intercourse with self-reported PE and DE at present. Second, we attempted to clarify whether and to what extent genetic or environmental factors are affecting continuity in EjD. As indicated by previous studies, moderate heritability is to be expected for PE, also at first intercourse. Hence, we expected the continuity of PE to be under partial genetic control, but be mostly affected by environmental effects unique to the individual. Furthermore, as studies have shown that the heritability estimates for PE at first intercourse and PE measured at present are quite similar in size, we expect the same genes to be associated with PE both at first intercourse and measured at present.
SUBJECTS AND METHODS
The present study involved 2633 men, of which 1812 were twins and 821 were siblings of the twins. Response rates might fluctuate slightly between instruments (the lowest number of subjects for any statistical analysis was 2570). Previous studies have shown that twins do not differ significantly from controls in most aspects . The mean (sd) age of the respondents was 26.63 (4.68) years. Subjects were a subset from the second data collection of the Genetics of Sex and Aggression sample. These data were collected in 2006 and targeted 18–33-year-old twins and their over 18-year-old siblings. In all, 10 524 participants responded to the second data collection, yielding an overall response rate of 45%. The response rate was lower for men (34%) respondents. The response rate represents an underestimate due to changing addresses; according to statistics, ≈15% of Finns move each year . Considering that the data collections lasted in each instance for >6 months, the real response rate was about half. The research plan was approved by the Ethics Committee of the Åbo Akademi University.
To measure ejaculatory function at the first intercourse, we used an instrument created specifically for this data collection. The subjects indicated the extent to which the following items were true about their first intercourse experience: ‘Ejaculation before penetration’, ‘Ejaculation almost immediately after penetration (one to three penetrations)’, ‘Ejaculation in <1 min after penetration’, ‘Fear of PE’. Individuals who had not responded to this part of the questionnaire were excluded from all statistical analyses. A composite variable measuring PE at first intercourse (termed ‘PE-1st’) was computed by summing these four items. This composite measure was computed to replicate the structure of a previously conducted and reported factor analysis on the same sample . Higher values on the variable indicate more problems related to PE.
To measure current ejaculatory function, 10 questions were used to ascertain objective and subjective aspects of ejaculatory function. The questions were designed to take different definitional aspects into account. To avoid heterosexist bias and exclusion of both female–male and male–male anal intercourse, we used a gender-neutral definition of ELT inclusive of anal intercourse. The questions were adapted from an unpublished questionnaire developed by Grenier and Byers . Exploratory and confirmatory factor analyses on these variables have previously been conducted on a larger sample that included all individuals of the present sample . To maintain consistency between the studies, no separate factor analysis was conducted for this study, and composite variables measuring PE and DE, respectively, were computed based on this factor solution. A previous study on a sample with no overlap with the present study also revealed a very similar factor structure . Higher values on either variable indicate more problems related to PE or DE.
We calculated descriptive statistics for all variables and then investigated the associations between PE-1st, PE, DE, age and age at first intercourse. A variable measuring the age difference between the first intercourse experience and the present age was computed for each participant (simply by subtracting the age of the first intercourse from the present age).
To investigate the effects of age on PE-1st, PE and DE, Pearson correlations were computed between the variables measuring age, age at the time for the first intercourse, age difference between first intercourse and present age, PE-1st, PE and DE. Next, regression analyses were conducted with PE-1st, PE and DE as dependent variables, respectively, in separate analyses, with age and age at first intercourse as covariates. The regression analyses were conducted with the Complex Samples General Linear Model (CSGLM) module for SPSS 15.0, a procedure that allows the data to be correlated and adjusts the estimates of sem s. Subsequently, the age difference variable was used to divide the sample into two groups based on median value to indicate whether a relatively long or short time had elapsed since the first intercourse. Correlations between PE-1st and PE were then computed separately for the two groups. The significance of the difference between these correlations was assessed by computing an interaction term in two steps: first, the variable measuring age difference and PE-1st were centralized by subtracting the variables’ means from all observations. Second, these two centred variables were multiplied to achieve a measure of the interaction. This interaction term was then included as a covariate in CSGLM regression analyses with PE and DE as dependent variables. PE and DE were, separately, inserted as dependent variables with PE-1st and the variable measuring age difference as covariates. Finally, the interaction term was included as a covariate in the regression analysis with PE as dependent variable to test the correlations for significance as described above. A significant interaction term here would indicate a significant difference of the PE-PE-1st correlations between the groups created by dividing the sample by the dichotomized age difference variable.
The reasons for continuity in EjD were investigated in terms of genetic and environmental sources. The present sample yielded 134 complete monozygotic (MZ) pairs (rg = 1.0), 128 complete dizygotic (DZ) pairs with an additional 218 families with at least two brothers forming twin-sibling or sibling-sibling pairs (all types of brother pairings except MZ twins have a genetic resemblance of ≈50%; thus rg = 0.5 for all pairings except MZ twins). Data from participants from families with only one respondent (i.e. individuals who had no participating brothers) contribute to the variance of the variables. However, for genetic modelling, data from these participants is not informative and thus these respondents were not included in genetic model fitting. Subsequent analyses were conducted between these groups. For the genetic modelling, a maximum family size of five members was selected.
First, phenotypic correlations (with adjacent CIs) for the rg = 1.0 and rg = 0.5 groups were computed for all three composite variables using the Mx statistical software package . Next, to explore the reasons for the continuity between PE-1st and current PE and DE (respectively), a bivariate model was fitted using the composite variables PE-1st and current PE, after which this process was repeated using the variables measuring PE-1st and current DE. Genetic and environmental influences can be separated in the twin design because genetic resemblance varies as a function of zygosity, whereas familial resemblance due to shared environmental influences does not. Specifically, MZ twins are genetically identical, whereas DZ twins (and all other biological siblings) on average share half of their segregating genes. By contrast, environmental influences that contribute to familial resemblance are assumed to affect MZ and DZ twins equally . Raw age-regressed variables age were used in all model-fitting scripts, using Mx with maximum likelihood estimation. The fit of the different models was compared by taking the fit function (−2 × log-likelihood of data) and the degrees of freedom (df) of the full model and subtracting it from the fit function and the d.f. of the nested restricted models. The subtraction gives a chi-squared value and associated df that can be tested for significance. In addition, the Akaike Information Criterion (AIC = χ2 − 2df) was considered, with lower values indicating better fit. Detailed descriptions of twin modelling analyses can be found in Posthuma et al..
Genetic (and environmental) correlations (i.e. to what extent the genetic effects on both phenotypes are caused by the same genes; or environmental factors) are calculated in the bivariate ACE model. Variance components can be tested for significance by being ‘dropped’; if, for example, model fit does not decrease significantly (compared to that of the full ACE model) by omitting the shared environmental component C, there is no reason to believe shared environmental effects to have a significant effect on the phenotype. The simpler, or nested, AE model should be preferred over the full ACE model. For theoretical reasons, the unshared environmental component E cannot be dropped, as it contains any error variance . In our model fitting analyses, AE, CE and E models were tested against the full model, and 95% CIs were computed for each variance component as well as the A, C, and E correlations.
The mean (sd) age at first sexual intercourse was 17.72 (2.77) years. The PE-1st variable comprised of four dichotomous variables (ranging from 0 to 1, with 1 indicating an affirmative response to the item): ‘ejaculation before penetration’ which occurred with a frequency of 4%; ‘ejaculation almost immediately after penetration’ (one to three penetrations) occurring with a frequency of 12%; ‘ejaculation in <1 min after penetration’ occurring with a frequency of 30%; and ‘fear of premature ejaculation’, which occurred with a frequency of 46%. The composite variable, which was also re-scaled to a range of 0–1, had a mean of 0.23 (0.24), indicating that problems related to PE at first intercourse were rather commonly reported. The mean (sd, range) age difference between the first intercourse and present was 8.93 (5.02, 0–32) years. A detailed description of the items measuring present ejaculatory function is given in Table 1.
|Variable/Description||Mean (sd, range)||Options and frequencies (valid %)|
|Number of thrusts*|
|How many thrusts have you typically been able to perform before ejaculation?||4.88 (0.55, 1–5)||1) No thrusts at all (0.2%)|
|2) 1–5 thrusts (1.2%)|
|3) 6–10 thrusts (3.5%)|
|4) more than 10 thrusts (93.8%)|
|5) Usually do not ejaculate (1.4%)|
|On average, during intercourse, how much time elapses between when you first enter your partner (vaginally or anally) with your penis and when you first ejaculate?||3.15 (0.83, 1–5)||1) <1 min (1.9%)|
|2) 1–5 min (21.2%)|
|3) 5–10 min (38.7%)|
|4) >10 min (36.8%)|
|5) I never ejaculate (1.4%)|
|Ejaculation before intercourse*|
|In what percentage of sexual intercourse experiences do you involuntarily ejaculate before intercourse has started?(85.3%)||1.20 (0.54, 1–5)||1) Never or very rarely|
|2) <50% of times (11.1%)|
|3) ≈50% of times (2.6%)|
|4) >50% of times (0.7%)|
|5) almost always or always (0.3%)|
|Subjective experience of PE*|
|In what percentage of sexual intercourse experiences do you ejaculate sooner than you want to?||0.41 (0.55, 0–2)||0) Have not ejaculated sooner than desired (61.5%)|
|1) Usually too early (35.5%)|
|2) Always or nearly always too early (3.0%)|
|Pretending to ejaculate†|
|How often have you pretended to ejaculate?||1.05 (0.26, 1–5)||1) Never or very rarely (96.0%)|
|2) <50% of the time (3.4%)|
|3) ≈50% of the time (0.5%)|
|4) >50% of the time (0.1%)|
|5) almost always or always (0.004%)|
|Later ejaculation than desired†|
|In what percentage of sexual intercourse experiences do you ejaculate later than you want to?||0.05 (0.25, 0–2)||0) Have not ejaculated later than desired (95.5%)|
|1) Usually too late (3.8%)|
|2) Always or almost always too late (0.7%)|
|Trying to speed up intercourse†|
|How often do you try to quicken the intercourse?||0.05 (0.25, 0–2)||0) I do not try to quicken intercourse (95.4%)|
|1) Sometimes (4.0%)|
|2) Always or almost always (0.7%)|
|Trying to delay intercourse*|
|How often do you try to delay the intercourse?||0.74 (0.63, 0–2)||0) I do not try to delay intercourse (36.3%)|
|1) Sometimes (53.4%)|
|2) Always or almost always (10.3%)|
|Worrying about ejaculation†|
|How often have you been worried that you would ejaculate sooner than you would like to?||2.34 (1.14, 1–5)||1) Never or rarely (28.9%)|
|2) <50% of the time (29.2%)|
|3) ≈50% of the time (25.0%)|
|4) >50% of the time (13.0%)|
|5) Almost always or always (3.9%)|
|Feeling of control*|
|How often have you felt that you could decide when to ejaculate?||2.87 (1.20, 1–5)||1) Never or rarely (15.4%)|
|2) <50% of the time (25.1%)|
|3) ≈50% of the time (25.2%)|
|4) >50% of the time (26.3%)|
|5) Almost always or always (8.0%)|
The age at first intercourse had significant but weak positive associations with both current PE (r = 0.050, P < 0.05; R2 = 0.003; F = 7.488, t = 2.736, P < 0.006) and current DE (r = 0.063, P < 0.01; R2 = 0.006; F = 11.05, t = 3.324, P < 0.001), indicating that the subjects who had been older at the time of their first intercourse were more likely to report current problems related to EjD. Regression analyses were conducted with three predictors inserted simultaneously (current age, age at first intercourse and the interaction term). Age at first intercourse also had a significant positive association with current age (r = 0.172, P < 0.001), indicating that the older participants were comparatively older when having their first sexual intercourse, probably reflecting a cohort effect on age of sexual debut. However, current age had no significant association with PE. The correlation between DE and current age was not statistically significant, but there was a significant (but weak) association between them in the multiple regression analysis (r = 0.022, P < 0.252; R2 = 0.006; F = 5.393, t = 2.322, P < 0.020). Current age also had a significant positive correlation with PE-1st (r = 0.072, P < 0.01). However, this effect was not significant in the multiple regression analysis (in which neither age at the first intercourse nor current age had any significant association with PE-1st).
Subjects who reported experiencing PE at their first intercourse were significantly more likely to report PE symptoms at present. As can be seen in Table 2, the participants with a relatively small age difference between their first sexual intercourse and the present reported higher stability in PE than those with a relatively large age difference between the first intercourse and present. Likewise for DE, a comparatively stronger (albeit negative) correlation was detected for those with a relatively small age difference. The mean age of the low age-difference group was 23.88 (3.30) years, and the mean age for the high age-difference group was 30.22 (3.68) years. However, the difference between the groups was not significant for PE, although DE had a significant interaction term (F = 5.214, t = 2.283, P < 0.05), indicating that the difference between age-difference groups was significant for DE.
|PE-1st||Mean (sd) age difference, years||Current PE||Current DE|
Next, two CSGLM regression analyses were conducted with PE and DE, respectively, inserted as dependent variables, and PE-1st together with the variable measuring age difference between first intercourse and present, as well as the age interaction term as covariates. PE-1st and age difference between first intercourse and present day had fairly strong associations with current PE (R2 = 0.145 for the full model with three covariates). PE-1st had the strongest association (F = 298.8, t = 17.29, P < 0.001), with age difference also being significantly associated (F = 11.45, t = 3.384, P < 0.001). This result suggests that PE-1st and current PE have a quite substantial amount of common causal factors. For DE (R2 = 0.041 for the full model with three covariates), PE-1st had, again, the strongest association (F = 108.5, t = 10.42, P < 0.001) with age difference having a weaker, but nonetheless significant, association (F = 7.523, t = −2.743, P < 0.01) with DE. This association was negative, indicating that those who report PE at the first intercourse are less likely to report DE later in life.
Before the model fitting analyses, all composite variables (PE-1st, PE and DE) were regressed to remove any effect of age. The phenotypic correlations for current PE and DE are presented in Table 3. In line with previous studies , the PE correlations indicated the presence of genetic effects for this phenotype. On the contrary, no such effects were suggested by the DE correlations. Much like PE, PE-1st had correlations that indicated the presence of genetic effects, which was expected, as the preceding analyses had indicated common causal factors.
|Pairing||Current PE||Current DE||PE-1st|
|rg = 1.00||0.305 (0.169–0.423)||−0.001 (−0.171–0.170)||0.300 (0.145–0.431)|
|rg = 0.50||0.130 (0.031–0.227)||0.088 (−0.004–0.180)||0.101 (0.005–0.197)|
Next, we fitted the model to the variables measuring PE-1st and current PE. The results of the bivariate model fitting analyses were much in line with previous findings [12,24], in that there was a significant genetic effect for PE of ≈30%. The genetic component was significant for both PE-1st and current PE, with the model fit significantly reduced when the genetic component was omitted for either variable (PE-1st, −2 × log-likelihood of data = 38444.2, Δχ2(1) = 8.616, P = 0.003, ΔAIC = 4.616; current PE, −2 × log-likelihood of data = 38 444.2, Δχ2(2) = 8.855, P = 0.012, ΔAIC = 4.855). The shared environmental components could be omitted without reducing the model fit for both variables, thus indicating that an AE model had optimal fit to the data. In the AE model, there were genetic influences of 28.0% for PE-1st, whereas current PE had a genetic effect accounting for 31.5% of the variance (Fig. 1). A substantial genetic correlation of rg = 0.989 indicated that mostly the same genes accounted for the genetic effects on PE both at the first intercourse and later in life. However, interestingly, very little of the variance (re = 0.111) was shared between the unshared environmental components, suggesting that unstable or situational effects might be active. The proportion of the phenotypic correlation between the variables measuring PE-1st and current PE that was explained by shared genetic effects was 0.293.
Next, we fitted the same model to the variables measuring PE-1st and current DE. Most of the variance in DE was accounted for by unshared environmental effects (94.8%). A weak genetic effect of 5.2% was detected for current DE, but model fit was not significantly reduced when the additive genetic component or shared environmental components were individually dropped for DE. However, when both were dropped simultaneously, model fit decreased significantly (−2 × log-likelihood of data = 36 815.8, Ä÷2(6) = 20.153, P = 0.003, ÄAIC = 8.153), suggesting that a significant familial effect might be present for DE. An AE model where the C components were dropped for both phenotypes had a very similar model fit to a modified AE model where the additive genetic component was dropped only for DE (but not for PE-1st), but the more complex AE model had a lower AIC value (ÄAIC = −1.510 between them), and was thus judged to be the best fitting model for the data. The correlation between the unshared environmental components in this model was rather modest (re = −0.092, 95% CI, −0.218 to 0.014), and negative, suggesting that those who experience PE at the first intercourse are less likely to develop DE later in life than those who do not, although this effect is rather small.
The purpose of the present study was, in the absence of robust longitudinal data, to elucidate the continuity of EjD and its stability over time. Our first aim was to investigate the temporal stability of PE; there were significant positive associations between reporting PE at the first intercourse and later in life, supporting the findings of previous research, e.g. . This finding was quite robust, with an effect size of 14.5% (combined with variables measuring age difference between first intercourse and age at the time of study participation, and an age interaction term). Participants who reported PE problems at the first intercourse were also significantly less likely to report DE later in life. In accordance with much recent research , age had quite weak effects on EjD. A probable age cohort effect on the age of first sexual intercourse was also detected, in that the older participants reported commencing their sexual debut at higher ages.
There were significant additive genetic effects of ≈30%, again suggesting that a firm heritable component is a cause of PE. A genetic effect on PE was previously established, e.g. , and recently a molecular genetic study  found an association between two alleles of the serotonin transporter promoter region 5-HTTLPR and IELT within a sample of patients with PE, suggesting that the genetic effect on PE might act through serotonergic neurotransmission. However, Janssen et al. found no differences in allele frequency between patients with PE and controls. It would thus be of great interest to repeat this analysis, perhaps on a larger sample.
The present results showed that the genetic effects that are affecting ejaculatory performance (both at the first intercourse and at present) are mostly caused by the same genes. In other words, the heritable component of PE seems to be stable over time. Theoretically, this would not necessarily need to be the case. For example, if there were genes that affect hormone regulation in puberty, and these endocrine effects (e.g. testosterone has been found to have an association with EjD ) in turn contribute to causing PE, there would be genetic effects that are only detectable in puberty. However, most of the variance in PE in the present study was caused by unshared environmental effects. Interestingly, there was very little covariation between the unshared environmental factors effective at the first intercourse and those that are effective later in life. In other words, this finding suggests that there are unique factors that affect ejaculatory performance only at the first intercourse which might be contextual, as suggested by Santtila et al.. Also, while any contextual factors (such as having sex with an unknown partner) can be repeated later in life, it might be that the effects of such factors are significantly more powerful when filtered through the thrill of the anticipation of the sexual debut. Furthermore, a persistent EjD might become a vicious circle over time, with the patient experiencing performance anxiety ahead of sexual intercourse, which in turn makes the PE problem worse. This then would be an environmental factor that is significantly less effective in the young and sexually inexperienced.
There are some limitations to the present study. First, the response rate (34%) might seem low. However, this is comparable with previously conducted sexuality-related mail survey studies both nationally  and internationally . Furthermore, data are self-reported, which might have implications for reliability. The data on EjD at the first intercourse are retrospective and are thus vulnerable to response bias from the participants.
Although a substantial amount of data indicate both short- and long-term temporal stability in EjD, a proper longitudinal study of ejaculatory functioning would be needed to more reliably and clearly chart the behaviour of PE over longer periods. Furthermore, to increase knowledge and validity of different PE measures, it would be interesting to conduct a (short- to medium-term) study within one (or few) participant(s) over several time points, the so-called P-technique . This approach could, in addition to yielding interesting data of within-subject fluctuations in IELTs, shed some light on contextual factors that affect ejaculatory performance.
In conclusion, EjD seems to be stable over both the short and long term. Unshared environmental factors are mostly accountable for this continuity, but a heritable component accounting for ≈30% of the variance is active in PE. The genetic effects that contribute to phenotypic variance in PE at first intercourse are the same that do so later in life. Unshared environmental factors that contribute to the variance in PE differ tremendously between first intercourse and later in life. Further research is needed to increase knowledge of the temporal dimension of PE, within- and between-subject variation of PE over time, and the individually unique factors that contribute to EjD.
CONFLICT OF INTEREST
None declared. Source of Funding: this research was financed by grant no. 210298 from the Academy of Finland and a Centre of Excellence Grant from the Stiftelsen för Åbo Akademi Foundation.