Differential efficacy of pelvic floor muscle training in primary versus acquired premature ejaculation: an 8-week comparative study using non-invasive biomechanical assessment

Abstract

Background

Premature ejaculation (PE), including lifelong (LPE) and acquired (APE) subtypes, negatively affects men’s sexual, psychological, and relational well-being.

Aim

This study aimed to investigate pelvic floor muscle function differences between healthy individuals and premature ejaculation patients and evaluate the efficacy of pelvic floor muscle training in improving premature ejaculation subtype symptoms using an innovative assessment approach.

Methods

In this 8-week prospective study, 199 men were enrolled (LPE, n = 66; APE, n = 83; healthy controls [HC], n = 50). PE participants received PFMT comprising PFM awareness, biofeedback-based strengthening, and integration into sexual activity. PFM function was measured with a perineometry/biofeedback device; anxiety and depression with 7-item Generalized Anxiety Disorder scale (GAD-7) and 9-item Patient Health Questionnaire (PHQ-9); and intravaginal ejaculatory latency time (IELT) by stopwatch.

Outcome

The primary outcomes were improvements in pelvic floor muscle function, ejaculatory latency time, and premature ejaculation symptom severity.

Results

After 8 weeks of PFMT, significant improvements were observed in both groups. In the APE group, IELT increased from a median of 120 s to 180 s, and premature ejaculation diagnostic tool (PEDT) scores decreased by 2 points (P < 0.001). In the LPE group, IELT increased from 30 s to 60 s, and PEDT scores decreased by 1.5 points (P < 0.001). Reductions in PHQ-9 (APE: –3.0; LPE: –2.0) and GAD-7 (APE: –3.0; LPE: –1.0) scores were also observed, with APE showing greater overall benefits. No significant baseline differences were found in PFM strength; however, both groups showed post-training increases in maximum contraction pressure and sustained contraction average pressure (all P < 0.001).

Clinical Implications

Pelvic floor muscle training appears effective in addressing psychological and physical symptoms of premature ejaculation, particularly for acquired premature ejaculation.

Strengths and Limitations

This study is the first to use a non-invasive biomechanical approach to compare pelvic floor muscle function between healthy individuals and PE patients. It also uniquely examines how PFM training affects different PE subtypes, offering insights for more targeted treatment strategies. However, the research is limited by the absence of sham controls, which are necessary to distinguish placebo effects, and it also lacks long-term follow-up to assess enduring outcomes.

Conclusion

Pelvic floor muscle training improves ejaculatory control and psychological outcomes in premature ejaculation patients, with greater efficacy observed in acquired premature ejaculation; further research is needed to explore long-term effects and underlying mechanisms.

Introduction

Premature ejaculation (PE) is among the most common male sexual dysfunctions, and it significantly impacts men’s physical and psychological well-being while also affecting their relationship satisfaction.^1,2 Notably, PE has been categorized into four subtypes: lifelong (LPE), acquired (APE), natural variable, and premature-like ejaculatory dysfunction, each presenting unique challenges for patients and clinicians alike. The 2014 updated criteria of the International Society for Sexual Medicine (ISSM) define LPE as ejaculation typically occurring within 1 min of vaginal penetration. They define APE as clinically significant reduction in intravaginal ejaculatory latency time (IELT) to usually ≤3 min, the inability to delay ejaculation, and the presence of negative personal consequences from PE, such as frustration, bother, stress, and/or avoidance of sexual intimacy.^3,4

Notably, PE has a multifactorial etiopathogenesis, with a complex interplay of psychological and biological factors.⁵ Recently, pelvic floor muscle (PFM) function has emerged as an influencer of ejaculatory control. The strength and endurance of these muscles are reportedly crucial in regulating ejaculatory reflexes, and their dysfunction may contribute to PE.^6–8 However, the precise role of PFMs in the etiology and treatment of PE remains under investigation.⁹

Interestingly, PFM training (PFMT) is increasingly being explored as a non-pharmacological approach to improving ejaculatory control and enhancing sexual health. This intervention, which aims to strengthen and improve the coordination of PFMs, has shown promise in treating various types of sexual dysfunction, including erectile dysfunction^7,10 and PE.^11,12 Despite growing interest in PFMT, there remains a lack of comprehensive data regarding its efficacy, specifically for PE subtypes and particularly in differentiating between its effects on LPE and APE.

La Pera et al. highlighted that the vast majority of men with PE are unaware of the role of PFMs in voluntarily controlling the ejaculatory reflex, with only 6.8% of men with PE correctly identifying and using PFMs for this purpose compared to 82% of men without PE.¹³ In their study, PFM awareness was first facilitated via digital rectal examination to help patients identify and voluntarily contract these muscles. Building on this, they developed a structured rehabilitation protocol combining PFM awareness training, instruction on timely contraction during the pre-orgasmic phase, and pelvic floor rehabilitation via biofeedback. In a cohort of 78 men with LPE, this approach yielded a 54% cure rate, with the mean IELT increasing from <2 min to >10 min in a subset of patients. These findings, together with subsequent large-scale follow-up studies like Pastore et al.’s (2018), which used anal probe–based stimulation to elicit PFM contractions and reported sustained benefits in 56% of participants at 36 months, support the clinical utility of PFMT in PE management.¹² However, none of these studies objectively quantified changes in PFM function, limiting mechanistic insight into how PFMT exerts its effects.

Recent advances have introduced non-invasive biofeedback-based devices capable of quantifying PFM strength and endurance with high reproducibility. For instance, Kim et al. used such a device to demonstrate that reduced maximal pelvic floor contraction strength was independently associated with erectile dysfunction severity, highlighting the feasibility of integrating objective pelvic floor assessments into sexual medicine research.¹⁴ Building upon this approach, our study applies a similar technology to characterize pelvic floor function in men with PE and to track functional changes following PFMT.

The current study aims to fill some gaps by investigating the correlation between PFM function and PE and evaluating the efficacy of an 8-week PFMT intervention in improving symptoms of PE. We herein introduce an innovative assessment approach using a novel measurement device that allows for comprehensive evaluation of PFM function. This device simplifies the evaluation process, allowing patients to sit on the equipment and perform pelvic floor contractions as instructed, based on which reliable and accessible data on muscle strength and endurance are acquired. By assessing outcomes like the ejaculatory latency time, PFM strength, and psychological factors (eg, anxiety and depression), we seek to provide insights into the therapeutic potential of PFMT for men with PE and highlight its role in addressing both physical and psychological symptoms.

Material and methods

Study design

This prospective observational study was conducted over an 8-week period, with recruitment occurring from December 2024 to April 2025. Follow-up assessments were conducted at baseline and at the end of the intervention period. Ethical approval was obtained from Ethics Review Committee of The Second Affiliated Hospital of Zhejiang Chinese Medical University (approval no.: Ethics Review 2024 research No. 105-IH01), and all participants provided written informed consent before enrollment.

A priori power analysis was conducted using G^*Power version 3.1.9.7 (Heinrich Heine University, Düsseldorf, Germany) to determine the minimum required sample size. For comparison among three independent groups (LPE, APE, and HC) using one-way analysis of variance (ANOVA), a medium effect size (f = 0.25), an alpha level of 0.05, and a power of 0.80 were assumed. The analysis indicated that a minimum total sample size of 159 was required. For between-group comparisons of LPE and APE using independent-samples t-tests, a medium effect size (d = 0.5), an alpha level of 0.05, and a power of 0.80 yielded a required sample size of 102 (51 per group).

Participants

Male participants diagnosed with LPE and APE were recruited through voluntary enrollment and referrals from urology and andrology specialists, whereas healthy controls (HCs) were recruited through voluntary enrollment and routine health examinations. Inclusion criteria for the PE groups included a diagnosis based on the ISSM criteria, self-reported IELTs of less than 1 min for LPE and reduced IELT for APE, and the absence of significant comorbidities affecting sexual function. HCs were required to have no history of sexual dysfunction. All participants were heterosexual men in a stable sexual relationship with a female partner for at least six months before enrollment. We excluded patients having received other interventions or pharmacological treatments that could affect their sexual function within the last 6 weeks, patients with a diagnosis of other types of sexual dysfunction, and patients with a history of conditions affecting PFM function, endocrine disorders, or allergies or contraindications to any interventions used in the study. Finally, the study included 199 participants and divided them into three groups: LPE (n = 66), APE (n = 83), and HC (n = 50) (Figure 1).

Figure 1.

Flowchart of participant recruitment and study design.

Intervention

Participants in PE groups underwent an 8-week biofeedback-enhanced PFMT program designed to improve PFM strength, coordination, and conscious control during sexual activity.^15,16 This program is a multimodal approach that integrates three complementary components:

1. Awareness of PFMs: Participants were educated about the PFMs and their crucial role in ejaculatory control. This initial phase was aimed at helping participants identify and consciously activate their PFMs, thus understanding their importance in regulating ejaculation.

2. Strengthening exercises: Participants engaged in supervised PFMT sessions using a Pulsed Magnetic Therapy Device (Model MP900, EDAN Instruments, Shenzhen, China). This device integrates biofeedback functions for perineometry and pelvic floor training. The biofeedback mode of the device provided real-time visual cues, guiding participants to correctly engage their PFMs while minimizing the use of abdominal muscles. Sessions were conducted at baseline (week 0) and at weeks 1, 2, 4, 6, and 8. During these sessions, participants performed exercises focused on improving both quick contractions and sustained muscle contractions. In addition, participants were instructed to perform daily PFM exercises at home to further reinforce these skills.

3. Integration into sexual activity: Participants were instructed to engage in sexual activity 1–2 times per week with their regular partner and apply PFM control strategies to delay ejaculation. Participants were encouraged to initiate penetration with a brief, controlled PFM contraction for a limited number of thrusts (~3–10, adjusted to their individual comfort), followed by transition to their usual sexual rhythm. In particular, they were trained to recognize the sensations leading up to ejaculation, stop stimulation, and relax their PFMs until the urge subsided. This technique was repeated 2–4 times during each sexual encounter to enhance ejaculatory control and help prevent premature ejaculation.

HCs did not undergo any intervention.

Measurements

Herein, PFM function was assessed at baseline and after the intervention using the same MP900 device, applied in its perineometry and biofeedback mode. Participants were instructed to sit relaxed on an evaluation chair equipped with an inflatable air-filled cushion positioned under the perineal area. The cushion contained sensors to measure pressure changes resulting from PFM contractions. Electrodes were also placed on the rectus abdominis muscle and the anterior superior iliac spine to monitor abdominal muscle activity, ensuring that participants engaged only their PFMs without compensatory abdominal contractions.

The assessment procedure included the following steps:

1. The cushion was inflated, and pressure changes were recorded in millimeters of mercury (mmHg) while monitoring abdominal muscle activity using the electrodes to ensure proper PFM engagement.

2. For the maximum contraction pressure (MCP) assessment, participants were asked to contract their PFMs as quickly and forcefully as possible, followed by immediate relaxation; this sequence was repeated three times with a 10-s rest interval between contractions.

3. For the sustained contraction average pressure (SCAP) assessment, participants were instructed to contract their PFMs and hold the contraction for 10 s before relaxing; this sequence was repeated three times with a 10-s rest interval between contractions.

These assessments provided reliable measurements of each participant’s MCP and SCAP, which were used to evaluate PFM strength and endurance.

Self-estimated IELT was measured at baseline and after the intervention for PE participants, with both assessments conducted following a period of sexual abstinence of 3–5 days. Psychological assessments were conducted using the 7-item Generalized Anxiety Disorder scale (GAD-7) to assess anxiety levels and the 9-item Patient Health Questionnaire (PHQ-9) to assess depression. The premature ejaculation diagnostic tool (PEDT) was used to evaluate PE symptom severity (Figure 2).

Figure 2.

This system allowed for quantitative, real-time assessment of pelvic floor dynamics and was crucial for both baseline evaluation and biofeedback-assisted training throughout the intervention. (A) the perineometry and biofeedback system used for non-invasive pelvic floor muscle (PFM) assessment and training. The participant sits on a specially designed evaluation chair equipped with an air-filled pressure sensor placed beneath the perineum. This setup enables real-time recording of PFM pressure changes (mmHg) during voluntary contractions. (B) Electrode placement for monitoring abdominal muscle compensation. Two surface electrodes are placed on the rectus abdominis (2 cm lateral and 2 cm inferior to the umbilicus), and one reference electrode is positioned on the anterior superior iliac spine (ASIS) to measure abdominal involvement during PFM contractions. (C) Representative pelvic floor muscle pressure curve during assessment, showing stable resting baseline, sharp contraction peaks (maximum contraction pressure, MCP), and sustained contraction segments (sustained contraction average pressure, SCAP). These parameters are used to evaluate PFM strength, endurance, and coordination. (D) Abdominal muscle activity curve recorded via surface electromyography (sEMG). Low involvement rate suggests minimal compensation by abdominal muscles, ensuring the accuracy of PFM measurement.

Objectives and hypotheses

The primary objective of this study was to explore the differential effects of PFMT on LPE and APE, using a non-invasive biomechanical approach. We also aimed to investigate the differences in PFM function between HCs and patients diagnosed with LPE or APE. We hypothesized the following:

Hypothesis 1: PFMT would significantly improve PFM strength and ejaculatory control in both LPE and APE patients, but the extent of improvement would significantly differ between the two subtypes.

Hypothesis 2: Significant differences in PFM function would be observed not only between HCs and PE patients but also between the two PE subtypes, with PE patients showing significantly worse PFM function.

Outcome measures

The primary outcomes were improvements in PFM function (MCP and SCAP), IELT, and PEDT scores. Secondary outcomes included changes in PHQ-9 and GAD-7 scores to assess improvements in psychological well-being.

Statistical analysis

Statistical analysis was performed using SPSS version 29.0 (IBM Corp., Armonk, NY, USA). Descriptive statistics were used to summarize baseline characteristics. Categorical variables were expressed as frequencies and percentages. For continuous variables, the Shapiro–Wilk test was first used to assess normality. Normally distributed continuous data were expressed as the mean ± standard deviation (SD), whereas non-normally distributed data were presented as the median and interquartile range (IQR). Participants with missing data were excluded from the analysis.

For between-group comparisons, one-way ANOVA was used for normally distributed data involving more than two groups, whereas the Kruskal–Wallis H test was used for non-parametric comparisons with more than two groups. If significant differences were identified, the Mann–Whitney U test was used for post-hoc between-group comparisons. Paired t-tests were used to compare pre- and post-intervention measures within groups. A P value of <0.05 was considered statistically significant. No subgroup or interaction analyses were performed in this study. No sensitivity analyses were conducted because the primary analyses were considered robust and no alternative analytic assumptions were planned.

Results

A total of 199 participants were included in the study, comprising 66 LPE patients, 83 APE patients, and 50 HCs. Table 1 shows the demographic characteristics of all participants. The groups did not significantly differ in terms of age, body mass index (BMI), alcohol consumption, protein intake, smoking status, fruit and vegetable intake, or physical exercise (all, P > 0.05). Therefore, no further adjustment for these variables was made in the primary analysis.

Table 1.

Baseline characteristics of all participants^a.

	LPE (n = 66)	APE (n = 83)	HC (n = 50)	P value
Age (year), median (IQR)b	30.5 (10.0)	30.0 (8.0)	30.5 (9.25)	0.885b
BMI (kg/m2), mean ± SDc	23.931 ± 3.187	24.050 ± 3.634	23.974 ± 3.023	0.976c
Smoke				0.683
Yes	30 (45.5)	32 (38.6)	20 (40)
No	36 (54.5)	51 (61.4)	30 (60)
Alcohol				0.893
Yes	27 (40.9)	35 (42.2)	19 (38)
No	39 (59.1)	48 (57.8)	31 (62)
Protein intaked				0.769
Adequate	58 (87.9)	74 (89.2)	46 (92)
Inadequate	8 (12.1)	9 (10.8)	4 (8)
Dietary fiber intaked				0.694
Adequate	45 (68.2)	51 (61.4)	32 (64)
Inadequate	21 (31.8)	32 (38.6)	18 (36)
Physical exercise				0.721
Sufficient	18 (27.3)	24 (28.9)	17 (34)
Insufficient	48 (72.7)	59 (71.1)	33 (66)

Abbreviations: APE, acquired premature ejaculation; BMI, body mass index; HC, healthy control; LPE, lifelong premature ejaculation.

^aData are presented as number (%) unless noted otherwise. P values are based on the chi-square test unless noted otherwise.

^bKruskal–Wallis test.

^cOne-way analysis of variance.

^dAccording to the Dietary Reference Intakes for China (2023 Edition), the Estimated Average Requirement (EAR) for protein in men aged 18–50 years is 60 g/day, and the Adequate Intake (AI) for dietary fiber is 25–30 g/day.¹

Clinical data of the participants are detailed in Table 2, which shows significant differences among the three groups in PHQ-9, GAD-7, and PEDT scores (all, P < 0.001). In all three measures, the HCs had significantly lower scores than LPE and APE patients (both, P < 0.001); however, no significant differences were noted between LPE and APE patients (P > 0.05).

Table 2.

Clinical characteristics of all participants^a.

	LPE (n = 66)	APE (n = 83)	HC (n = 50)	P value
PHQ-9 score	9.0 (8.00)	9.0 (11.00)	5.5 (7.0)b	<0.001
GAD-7 score	8.5 (10.00)	8.0 (10.00)	5.0 (8.0)b	0.007
PEDT score	12.0 (5.00)	11.00 (5.00)	6.0 (3.00)c	<0.001
Self-estimated IELT, seconds	30.00 (30.00)	120.00 (120.00)d	870.00 (360.00)c	<0.001
MCP, mmHg	13.150 (7.090)	12.820 (7.77)	13.990 (9.270)	0.899
SCAP, mmHg	8.860 (6.450)	7.300 (5.96)	7.530 (7.210)	0.669

Abbreviations: APE, acquired premature ejaculation; GAD-7, Generalized Anxiety Disorder 7-item; HC, healthy control; IELT, intravaginal ejaculation latency time; IIEF-5, International Index of Erectile Function 5-item; MCP, maximum contraction pressure; PEDT, premature ejaculation diagnostic tool; PHQ-9, Patient Health Questionnaire 9-item; LPE, lifelong premature ejaculation; SCAP, sustained contraction average pressure.

^aData are presented as median (IQR). P values are based on Kruskal–Wallis H test.

^b P < 0.05 vs. each type of PE.

^c P < 0.001 vs. each type of PE.

^d P < 0.001 vs. LPE.

The self-estimated IELT was also found to significantly differ among the three groups, with HCs showing longer IELT than LPE and APE patients (both, P < 0.001). Furthermore, APE patients had significantly longer IELT than LPE patients (P < 0.001).

Notably, MCP and SCAP showed no significant between-group differences, with P values of 0.899 and 0.669, respectively.

A total of 149 patients (LPE group, n = 66; APE group, n = 83) participated in the 8-week systematic PFM training. Participant dropout occurred primarily due to personal reasons (n = 8), time constraints (n = 7), or non-compliance with the training protocol (n = 10), accounting for ~15% (n = 25) of participants; consequently, 124 (LPE group, n = 56; APE group, n = 68) participants completed the 8-week training and were included in the study. Table 3 shows the baseline characteristics of these participants. The two groups did not significantly differ in terms of age, BMI, smoking status, alcohol consumption, protein intake, fruit and vegetable intake, or physical exercise (all, P > 0.05). Similarly, Table 4 shows clinical characteristics of these 124 patients. They showed no significant between-group differences in PHQ-9, GAD-7, and PEDT scores (all, P > 0.05). However, self-estimated IELT was significantly lower in the LPE group than in the APE group (P < 0.001). In terms of PFM function, although the MCP and SCAP values were slightly higher in the LPE group than in the APE group, the difference did not reach statistical significance (P = 0.974 for MCP, P = 0.429 for SCAP).

Table 3.

Baseline characteristics of participants who completed systematic pelvic floor muscle training^a.

	LPE (n = 56)	APE (n = 68)	P value
Age (year), median (IQR)b	31.0 (10.5)	30.0 (8.0)	0.727
BMI (kg/m2), mean ± SDc	24.059 ± 3.303	24.157 ± 3.596	0.876
Smoke			0.068
Yes	28 (50.0)	23 (33.8)
No	28 (50)	45 (66.2)
Alcohol			0.733
Yes	33 (58.9)	38 (55.9)
No	23 (41.1)	30 (44.1)
Protein intake			0.677
Adequate	48 (85.7)	60 (88.2)
Inadequate	8 (14.3)	8 (11.8)
Dietary fiber intake			0.383
Adequate	38 (67.9)	41 (60.3)
Inadequate	18 (32.1)	27 (39.7)
Physical exercise			>0.999
Sufficient	14 (25.0)	17 (25.0)
Insufficient	42 (75.0)	51 (75.0)

Abbreviations: APE, acquired premature ejaculation; BMI, body mass index; HC, healthy control; LPE, lifelong premature ejaculation.

^aData are presented as number (%) unless noted otherwise. P values are based on the chi-square test unless noted otherwise.

^bMann–Whitney U test.

^cIndependent-samples T test

Table 4.

Pre-training clinical characteristics of participants who completed systematic pelvic floor muscle training^a.

	LPE (n = 56)	APE (n = 68)	P value
PHQ-9 score	9.0 (7.75)	10.5 (11.75)	0.224
GAD-7 score	8.0 (8.00)	8.0 (9.75)	0.700
PEDT score	12.0 (5.00)	11.0 (5.00)	0.876
Self-estimated IELT, seconds	30.00 (30.00)	120.00 (120.00)	<0.001
MCP, mmHg	13.150 (6.860)	12.540 (6.950)	0.974
SCAP, mmHg	8.470 (6.530)	7.240 (5.780)	0.429

Abbreviations: APE, acquired premature ejaculation; GAD-7, Generalized Anxiety Disorder 7-item; HC, healthy control; IELT, intravaginal ejaculation latency time; IIEF-5, International Index of Erectile Function 5-item; MCP, maximum contraction pressure; PEDT, premature ejaculation diagnostic tool; PHQ-9, Patient Health Questionnaire 9-item; LPE, lifelong premature ejaculation; SCAP, sustained contraction average pressure.

^aData are presented as median (IQR). P values are based on the Mann–Whitney U test.

Table 5 shows the post-training outcomes and comparisons of participants who completed the 8-week systematic PFM training. Both LPE and APE groups showed significant improvements in all assessed outcomes after the training, including reduced PHQ-9, GAD-7, and PEDT scores and increased self-estimated IELT, MCP, and SCAP values (all, P < 0.001).

Table 5.

Post-training outcomes and comparison for participants who completed systematic pelvic floor muscle training^a.

	LPE (n = 56)				APE (n = 68)				Z valued	P valued	r value c
	Post-training	Z valueb	P valueb	r value c	Post-training	Z valueb	P valueb	r value c
PHQ-9 score	7.0 (7.50)	-4.514e	<0.001g	0.603	7.0 (8.00)	-6.564e	<0.001g	0.796	-0.433	0.655	0.039
GAD-7 score	7.0 (8.00)	-4.554e	<0.001g	0.609	5.0 (7.00)	-6.464e	<0.001g	0.784	-1.420	0.156	0.128
PEDT score	10.5 (5.0)	-5.362e	<0.001g	0.717	9.0 (4.75)	-5.701e	<0.001g	0.691	-2.528	0.011	0.227
Self-estimated IELT, seconds	60.00 (0.00)	-6.332e	<0.001g	0.846	180.00 (150.00)	-7.040f	<0.001g	0.854	-9.183	<0.001	0.825
MCP, mmHg	16.740 (9.070)	-6.477f	<0.001g	0.866	16.315 (8.710)	-7.167f	<0.001g	0.869	-0.010	0.992	<0.001
SCAP, mmHg	10.220 (7.700)	-6.510f	<0.001g	0.870	8.445 (6.210)	-7.168f	<0.001g	0.869	-1.155	0.248	0.104

Abbreviations: APE, acquired premature ejaculation; GAD-7, Generalized Anxiety Disorder 7-item; HC, healthy control; IELT, intravaginal ejaculation latency time; IIEF-5, International Index of Erectile Function 5-item; MCP, maximum contraction pressure; PEDT, premature ejaculation diagnostic tool; PHQ-9, Patient Health Questionnaire 9-item; LPE, lifelong premature ejaculation; SCAP, sustained contraction average pressure.

^aData are presented as median (IQR).

^bThe Z values and P values are derived from the Wilcoxon signed-rank test, comparing pre- and post-intervention changes within each patient group.

^cEffect size r was calculated using r = , effect size interpretation: 0.1 = small, 0.3 = medium, and 0.5 = large.

^dThe Z values and P values are based on the Mann–Whitney U test, comparing the intervention effects between LPE and APE groups. ^eBased on positive ranks.

^eBased on positive ranks.

^fBased on negative ranks.

^g P < 0.001 vs. the respective pre-training clinical characteristics values as shown in Table 4.

Upon comparing post-training outcomes between the two PE groups, the APE group showed significantly lower PEDT scores than the LPE group (P = 0.011), indicating greater improvement in ejaculatory control. In addition, the self-estimated IELT was significantly longer in the APE group than in the LPE group (P < 0.001). The two groups did not significantly differ in terms of PHQ-9, GAD-7, MCP, or SCAP values after the training.

Table 6 shows the changes in outcome measures for LPE and APE patients. This change was calculated by subtracting pre-training values from post-training values. The analysis showed that compared to LPE patients, APE patients showed significantly greater reductions in PHQ-9 and GAD-7 scores and significant increases in self-estimated IELT (all, P < 0.001). These findings suggest that APE patients derive more substantial psychological and ejaculatory benefits from the PFM training intervention than LPE patients.

Table 6.

Changes in pre- and post-training outcome measures for PE patients and group comparisons.^a

	LPE (n = 56)	APE (n = 68)	Z valueb	P valueb	r valuec
PHQ-9 score reduction	1.0 (3.00)	3.5 (3.00)	4.911	<0.001	0.441
GAD-7 score reduction	1.0 (2.00)	3.0 (3.75)	4.786	<0.001	0.430
PEDT score reduction	2.0 (3.00)	3.0 (4.75)	1.689	0.091	0.152
Self-estimated IELT improvement, seconds	30.00 (30.00)	90.00 (90.00)	7.808	<0.001	0.701
MCP improvement, mmHg	3.315 (2.200)	3.405 (2.260)	-0.249	0.804	0.022
SCAP improvement, mmHg	2.015 (1.610)	1.73 (1.47)	-0.894	0.371	0.080

Abbreviations: APE, acquired premature ejaculation; GAD-7, Generalized Anxiety Disorder 7-item; HC, healthy control; IELT, intravaginal ejaculation latency time; IIEF-5, International Index of Erectile Function 5-item; MCP, maximum contraction pressure; PEDT, premature ejaculation diagnostic tool; PHQ-9, Patient Health Questionnaire 9-item; LPE, lifelong premature ejaculation; SCAP, sustained contraction average pressure.

^aData are presented as median (IQR) and represent the changes (post-training minus pre-training) in PHQ-9 scores, GAD-7 scores, PEDT scores, self-estimated IELT, MCP, and SCAP.

^bThe Z values and P values are derived from the Mann–Whitney U test.

^cEffect size r was calculated using r = , effect size interpretation: 0.1 = small, 0.3 = medium, and 0.5 = large.

To compare the treatment response between APE and LPE subtypes while adjusting for potential covariates, ANCOVA analyses were conducted for both post-intervention IELT and PEDT scores, with age and corresponding baseline measures entered as covariates.

For IELT, the overall model was significant, F(3,120) = 190.27, P < 0.001, adjusted R² = 0.822. Baseline IELT (P < 0.001, partial η² = 0.672) and age (P = 0.006, partial η² = 0.062) were significant predictors. Importantly, PE subtype remained a significant factor, F(1,120) = 5.14, P = 0.025, partial η² = 0.041, with the APE group showing a greater increase in IELT than the LPE group [B = 31.13, 95% CI: 3.95–58.30, Figure 3A].

Figure 3.

Adjusted post-treatment outcomes (mean difference ± 95% CI) between APE and LPE groups. (A) Estimated marginal mean difference in IELT (seconds), with APE showing significantly greater increase compared to LPE. (B) Estimated marginal mean difference in PEDT score, with APE showing significantly greater symptom improvement than LPE.

For PEDT, the ANCOVA model was also significant, F(3,120) = 37.30, P < 0.001, adjusted R² = 0.470]. Baseline PEDT was the strongest predictor (P < 0.001, partial η² = 0.468), whereas age showed no significant effect (P = 0.528). The APE group showed a significantly greater reduction in PEDT scores than the LPE group, F(1,120) = 4.69, P = 0.032, partial η² = 0.038; B = –1.026, 95% CI: –1.964 to –0.088, Figure 3B. These findings suggest that even after adjusting for baseline severity and age, APE patients experienced more favorable improvements in both objective and subjective outcomes after PFMT.

Upon comparing post-training outcomes between the two PE groups, the APE group showed significantly lower PEDT scores than the LPE group (P = 0.011), indicating greater improvement in ejaculatory control. In addition, the self-estimated IELT was significantly longer in the APE group than in the LPE group (P < 0.001). The two groups did not significantly differ in terms of PHQ-9, GAD-7, MCP, or SCAP values after the training.

In this study, a linear regression model was constructed to predict post-training premature ejaculation diagnostic tool (post_PEDT) scores. To construct a model for predicting post_PEDT scores, we initially included all variables that we considered relevant based on our research hypotheses. To achieve an optimal regression model, we used a stepwise regression approach that incorporated both forward selection and backward elimination. During the stepwise regression process, we aimed to identify the best model by adding or removing variables. Variables were excluded from the model if their P values were large, which indicated their insignificant association with the dependent variable, or if they showed high multicollinearity (VIF > 10). The final model was selected based on maximizing the adjusted R² and improving the overall significance of the model. The final model included the following variables: the PE type, protein intake, physical exercise, pre-training PEDT (pre_PEDT), average muscle strength improvement, PHQ-9 score reduction, and self-estimated IELT improvement.

Besides physical exercise (P = 0.272), all other retained variables significantly influenced post_PEDT (P < 0.05). The rationale for retaining physical exercise was its contribution to enhancing the model’s explanatory power. The final model achieved an R value of 0.833, an R² of 0.695, and an adjusted R² of 0.676, indicating good model performance. Collinearity diagnostics revealed no severe multicollinearity among the variables (VIF < 10). Notably, pre_PEDT had the most significant effect on post_PEDT (P < 0.001).

Residual analysis was performed to ensure the validity of the linear regression assumptions. The normality test of the residuals indicated that the P values for both the Kolmogorov–Smirnov and Shapiro–Wilk tests were > 0.05, suggesting that the residuals followed an approximately normal distribution.

Discussion

To our knowledge, this study is the first to investigate the differences in pelvic floor muscle function between healthy individuals and PE patients using a non-invasive biomechanical approach. We believe it is also the first to describe the differential effects of PFM training on different PE subtypes. These findings underline the distinct therapeutic outcomes between the two PE subtypes, which enhance our understanding of their pathophysiological mechanisms and inform more effective clinical management strategies.

This study was aimed at investigating whether PFM function differs between HCs and PE patients and drawing comparisons between APE and LPE patients. Contrary to the primary hypothesis, our findings did not reveal any significant differences in PFM function between HCs and the PE group or between APE and LPE patients. These results suggest that PFM dysfunction is not as pronounced a factor in the pathophysiology of PE as previously hypothesized, at least not in the context of muscle strength as measured by MCP and SCAP.

In our research, although baseline biomechanical parameters showed no intergroup differences, PFMT showed positive effects in both LPE and APE patients. After 8 weeks of PFMT, significant improvements were observed in PFM strength (both MCP and SCAP) and self-estimated IELTs. Pelvic floor exercises reportedly enhance body awareness, which is crucial for improving self-confidence and the ability to control the ejaculatory reflex.¹⁷ In particular, the bulbocavernous and ischiocavernous muscles, which are involved in the expulsion phase of ejaculation, reportedly show increased electromyographic activity during sexual arousal,¹⁸ and their voluntary relaxation during sexual activity has been shown to inhibit the ejaculation reflex.¹⁷

The PFMT intervention requires individuals to intensely focus on rapid contractions and sustained muscle contractions, fostering their ability to recognize and respond to ejaculatory impulses during sexual activity. PFMT may offer significant benefits in managing PE by enhancing attention regulation, which is a crucial component of interoceptive awareness.¹⁸ As individuals repeatedly engage in these exercises, their ability to direct attention toward both internal sensations and external actions is likely to improve. The underlying mechanism may involve the modulation of neural circuits within the sympathetic and parasympathetic nervous systems.^17,19 By enhancing the coordination between these systems, PFMT can potentially stabilize the reflex pathways that govern ejaculation, thus improving ejaculatory control.

A subset of PE patients initially had strong PFMs and showed limited further improvement in MCP and SCAP after the intervention; interestingly, even these patients reported markedly improved ejaculatory control and sexual satisfaction. This phenomenon supports previous research findings that most PE patients lack the awareness to use PFM contractions to control ejaculation.¹³ By improving their awareness through PFMT, patients can consciously apply this knowledge during sexual activities, effectively prolonging their ejaculation time. Such awareness is pivotal for recognizing the cues of impending ejaculation and for modulating PFM activity accordingly.

The intervention used in this study was intentionally designed as a biofeedback-enhanced PFMT program rather than a purely strength-based PFMT regimen. By combining neuromuscular training with real-time biofeedback, body awareness exercises, and behavioral strategies, such as the stop–start technique, the intervention targeted both the physiological capacity to modulate ejaculatory reflexes and the cognitive–behavioral skills needed to apply these controls in vivo. During intercourse, participants were encouraged to begin with a brief, controlled PFM contraction for a limited number of thrusts before returning to their usual rhythm, as a preparatory awareness exercise; upon recognizing pre-ejaculatory sensations, they were instructed to cease penile stimulation and consciously relax the PFMs until the urge subsided, before resuming. This individualized approach aligns with sex-therapy principles in previous study—a combination supported by PFMT studies showing enhanced awareness and ejaculatory control.^17,19 The multimodal nature of biofeedback-enhanced PFMT may partly explain the improvements observed even in patients with relatively strong baseline PFM function.

The association of PE with elevated anxiety and depression levels has been well documented.²⁰ Studies have shown that PE can increase anxiety levels in patients, which can consequently act as a risk factor for worsening PE symptoms.^21,22 In our study, PFMT significantly reduced both PHQ-9 and GAD-7 scores in both LPE and APE patients. These findings suggest that PFMT exerts psychotherapeutic benefits by addressing the psychological aspects of PE, which are often exacerbated by performance anxiety and lack of sexual success. The “vicious cycle” of anxiety and sexual failure is a common psychological burden for PE patients, wherein the fear of poor sexual performance can lead to avoidance behaviors and increased anxiety.^23,24

Beyond general distress, emerging evidence indicates that psychosexual context is particularly relevant for PE, particularly APE. Clinical reviews have linked APE to sexual performance anxiety, relationship stress, and comorbid erectile dysfunction, suggesting psychosocial drivers distinct from those implicated in LPE.²⁵ In addition, studies have shown that men with PE often show higher attachment anxiety and poorer social–cognitive performance compared to controls,²⁶ implying increased vulnerability to hyperarousal and heightened vigilance in intimate contexts.

This relationship between psychological distress and PE emphasizes the significance of addressing not only the physical but also the psychological components of the condition. By improving ejaculatory control, PFMT helps create a positive memory of sexual success, which disrupts this cycle of anxiety and sexual failure, fostering psychological resilience in patients. Our findings are further supported by the linear regression model, which showed that reductions in PHQ-9 scores were independently associated with improvements in PEDT scores (β = -0.24, P = 0.003). This highlights the interconnected nature of the physical and psychological aspects of PE, underlining the importance of integrating behavioral therapies, such as cognitive behavioral therapy, alongside PFMT in the management of PE, even in the absence of significant PFM dysfunction.

Notably, APE patients showed significantly greater improvements across these outcomes compared to LPE patients (all, P < 0.001), emphasizing the biological distinctions between these subtypes. The limited response of LPE patients to PFMT aligns with a neurobiological model that implicates serotonergic dysfunction, particularly heightened serotonin receptor sensitivity and impaired 5-HT neurotransmission, as a key contributor to the pathophysiology of LPE.

Several single-nucleotide polymorphisms in the serotonin-transporter-linked promoter region (5-HTTLPR), such as rs25531 and rs4795541, reportedly differ significantly between individuals with and without LPE,^27,28 suggesting downregulation of 5-HT reuptake and reduced synaptic serotonin availability in affected individuals. Furthermore, polymorphisms in serotonergic receptor genes, particularly HTR1A (5-HT1A receptor) and HTR2C (5-HT2C receptor), have also been implicated. For instance, the rs6295 variant in HTR1A may enhance autoreceptor expression, thus inhibiting 5-HT release,^29,30 whereas the rs518147 and rs3813929 variants in HTR2C may lead to receptor overactivation.^31,32 Together, these alterations may disrupt the excitatory–inhibitory balance of the ejaculatory reflex arc.

In addition, a recent study in the Chinese Han population identified several polymorphisms in the TPH2 gene, which encodes the rate-limiting enzyme for serotonin biosynthesis, as significantly associated with LPE susceptibility, among which the rs11178996 variant appeared to confer a protective effect. Interestingly, the study also found that different TPH2 genotypes correlated with levels of key neurochemical and metabolic markers, such as 5-HT, leptin, and folic acid, suggesting that TPH2 polymorphisms influence LPE risk through both serotonergic and metabolic pathways.³²

These converging findings collectively support the hypothesis that LPE is strongly linked to a genetically mediated serotonergic dysfunction and may thus require targeted pharmacologic strategies, such as the use of potent SSRIs like paroxetine or clomipramine, to enhance synaptic 5-HT availability.

In contrast, APE patients showed greater reversibility of symptoms, supporting the hypothesis of a functional etiology linked to maladaptive psychophysiological patterns, such as hyperarousal and poor awareness of sexual function. These factors are amenable to intervention through biofeedback-enhanced PFMT, which promotes neuromuscular re-education. The stronger response in APE patients suggests that PFMT can help restore normal ejaculatory function by correcting dysfunctional reflex patterns.

The perineometry and biofeedback device used in this study provides a reliable, non-invasive method for assessing pelvic floor dynamics. Perineometry is a well-established method for evaluating PFM strength and endurance,³³ offering greater convenience and accuracy over traditional techniques while also being more acceptable to patients. By simultaneously monitoring abdominal compensation through surface electromyography, this device ensures precise measurement of PFM activity, which is crucial for accurately tracking changes in muscle function. Furthermore, real-time biofeedback during PFMT sessions likely enhanced the neuromuscular re-education of patients, particularly of APE patients, who showed greater integration of contraction techniques into sexual activity.

However, strength-based metrics alone may not fully capture the spectrum of pelvic floor dysfunction relevant to PE. Clinical reports indicate that an overactive pelvic floor (OPF)—characterized by elevated resting tone, poor relaxation, or involuntary guarding—can negatively affect male ejaculatory function,^34,35 and a recent review identified premature ejaculation among the more frequently reported sexual dysfunctions in men with pelvic floor overactivity.³⁶ Overactivity may exacerbate ejaculatory problems by reinforcing maladaptive reflexes or limiting voluntary relaxation near the point of inevitability. These features are unlikely to be detected by contraction pressures alone and typically require targeted assessment of resting tone and coordination (eg, manual/digital examination).

Future work should therefore incorporate multimodal pelvic assessment to enable phenotype-driven stratification—combining clinical palpation of resting tone and trigger points with objective indices (eg, baseline perineal pressure traces, involuntary sEMG activity, and where feasible, ultrasound for dynamic relaxation/coordination). Such classification may distinguish hypertonic versus hypotonic/weakness-dominant presentations and guide tailored interventions and responder prediction.

Despite its strengths, this study has several limitations. The 8-week intervention period did not allow for the evaluation of long-term effects or relapse rates. In addition, the reliance on self-reported IELT introduces potential recall bias, even though the possibility of such a bias was mitigated by including objective assessments acquired using the PEDT. The absence of a sham-controlled group also limits the ability to definitively attribute the observed benefits to PFMT and distinguish them from placebo effects. Future longitudinal studies with 6–12-month follow-up are warranted to characterize sustainability and relapse trajectories and to test whether booster sessions are needed to maintain gains. Incorporating phenotype-based stratification and advanced neurobiological readouts (eg, neuroimaging) may further clarify central and peripheral mechanisms and optimize individualized care pathways for PE.

Clinical messages

1. Diagnostic refinement: Biomechanical pelvic floor assessment should be integrated into routine PE evaluations to identify candidates who would benefit the most from PFMT.

2. Therapeutic personalization: PFMT may represent a promising non-pharmacological option for PE management, particularly for APE patients with maladaptive psychophysiological patterns.

3. Multidisciplinary care: A holistic approach combining PFMT with cognitive behavioral therapy should be considered for APE patients with comorbid anxiety.

Conclusion

Taken together, given the lack of significant differences in PFM function between HCs and PE patients and between APE and LPE patients, PFM dysfunction alone may not be the primary mechanism underlying PE. Although muscle strength and function are important components of sexual health, the pathophysiology of PE likely involves complex neurobiological and psychophysiological factors. Our findings support the notion that improving PFM awareness and neuromuscular control through PFMT may offer a valuable non-pharmacological approach for managing ejaculatory control, particularly in APE patients. Future research should continue to explore the interplay among PFM function, neurobiological factors, and psychological components to better understand the mechanisms underlying PE and develop more targeted treatment strategies.

Author contributions

Yinan Lyu: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing—original draft, Visualization. Jiawei Gong: Data curation, Resources, Software, Formal analysis, Writing—review & editing. Houdong He: Investigation, Validation, Resources. Hongzhe Xu: Data curation, Investigation. Huiping Dong: Methodology, Supervision. Yue Duan: Funding acquisition, Supervision, Project administration, Resources, Writing—review & editing.

Funding

This work was supported by the Natural Science Foundation of Zhejiang Province (grant No.LY20H270003) and Zhejiang Traditional Chinese Medicine Administration (grant No. 2022ZZ017).

Conflicts of interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethical approval and informed consent statements

Ethical approval was obtained from Ethics Review Committee of The Second Affiliated Hospital of Zhejiang Chinese Medical University (approval no.: Ethics Review 2024 research No. 105-IH01), and all participants provided written informed consent before enrollment.

Consent for publication

Not applicable.