Pelvic floor muscle training (PFMT) is an efficacious, low risk intervention and is being proposed as first line treatment for every type of incontinence. We aimed to identify clinical prognostic factors affecting the outcome of PFMT in women with primarily stress incontinence (SUI)

Data from 188 women with either SUI or mixed urinary incontinence (MUI) with a primarily stress component were retrospectively reviewed. All women were recruited from female Urology outpatient clinics of a public teaching hospital and underwent a 3-month PFMT programme by a specialized physiotherapist. They were evaluated at baseline and on a monthly basis until the end of the intervention. Initial evaluation comprised: 1) medical history including concomitant medication, gynecological operations, number and type of deliveries, menopausal status, age and somatometric values, 2) urogynecological history/examination for type and burden of incontinence, presence of frequency, nocturia and prolapse, 3) digital evaluation of pelvic floor (strength, endurance, number of repetitions and number of fast contractions)(n=87 women with available data) and 4) 3-day bladder diaries (n=83 women with available data). Additionally, data from baseline urodynamic studies were analyzed in 51/188 women who had originally been referred for preoperative evaluation but opted for PFMT. Patients diagnosed with primarily urgency incontinence, operated for prolapse, incontinence or urethral reconstruction, were excluded. 
Statistical analysis. Parametric and non-parametric tests were used for comparisons before and after intervention (secondary outcomes). Univariate and multivariate logistic regression was conducted to identify prognostic factors for 50% improvement of incontinence and complete cure of incontinence in the total group as measured by number of pads used, and then separately for the SUI and MUI subpopulations (primary endpoint). Several factors (age, BMI, type of incontinence, number of deliveries, menopause, prolapse, constipation, medication, surgical history, birth of overweight neonate, pelvic floor evaluation, severity of leak upon stress test) were tested for their predictive value.

Mean patient age was 52.2±10 years, with a median ΒΜΙ 27.6kg/m2. Approximately half the participants suffered from SUI (n=90, 47.9%), and the rest from MUI (n=98, 52.1%). Median number of deliveries was 2 (IQR=1), with 32.9% of women reporting delivery of overweight neonates. The majority (122/188, 64.9%) of patients were postmenopausal, and 37(19.7%) had a history of gynecological surgery. One out of 5 women (22.3%, n=42) required simultaneous prescription of medication for incontinence; 13.8% (n=26) were using anticholinergics, 4.3% (n=8) duloxetine, 2.1% (n=4) mirabegron, and 4 patients (2.1%) received combination treatment (anticholinergic plus α-blocker in 2 and duloxetine plus anticholinergic in 2). 
Bladder diaries recorded a median 9 urinations daily (IQR=2), 3 episodes of incontinence (IQR=3) and usage of 3 pads daily (IQR=3) at baseline. Median volume of fluid consumed daily was 1615ml (IQR=600) and voided volume varied between 480 (max) and 100 ml(min) (IQR=260-60). 
In women undergoing urodynamic investigation, mean cystometric capacity was 402ml (IQR=265), Qmax 21ml/s (IQR=13) and PVR 0ml (ΙQΡ=40). Detrusor overactivity was diagnosed in 17 patients (33,3%) while urodynamic SUI was diagnosed in the majority of patents (60,8%) and UUI in only 15,7%.

Effect of PFMT on incontinence. The median number of pads used daily at baseline was 3 (IQR=2), with 38.8% of women reporting only loss of drops, 50.5% moderate incontinence and 10.6% severe urine leakage. The number of pads was reduced to 1(IQR=1) after PFMT (p<0,001) for the total sample.
At baseline, stress test was negative in 43.1% of women. In the remaining patients, the test was mildly positive (small leak) in 26.1%, moderately positive in 20.2% and severely positive in 10.6% of patients. At 3 months, stress test was negative in 88.8%, while 10.1% of patients had a small leak and 1.1% more severe incontinence.  Concerning urgency symptoms, 51% of patients with MUI had daily urgency incontinence, 26.5% reported weekly symptoms and the rest (22.4%) monthly. Severity of urgency incontinence varied from drops in 52% of patients with MUI, to moderate leak in 39.8% and severe urine loss in 8.2%.
At 3 months, 10% of the SUI and 11.2% of the MUI subgroups were cured from SUI (overall 20 [10.6%] patients), while 35 (35.7%) of those with MUI reported being free of urgency incontinence. Bivariate analysis of patients completely cured from SUI for predictive factors revealed a strong correlation only with negative or mildly positive stress test (p=0,014, x2 test Linear-by-Linear). However, this significance was lost on binary logistic regression including all the aforementioned factors. 
Similarly, complete treatment of urgency incontinence in women with MUI was investigated. Again, only lower severity of stress test (p=0.009, x2-test) related to complete relief of symptoms. Furthermore, binary logistic regression including the previous factors confirmed the relation of initially negative stress test (p=0,003) or mildly positive stress test (p=0,007) with complete cure of SUI in the MUI subgroup. 
Moreover, 50% improvement in number of pads was calculated for all patients and explored in terms of recognizing predictive variables. Assessment after PFMT revealed >50% reduction in number of pads in 107 (56.9%) patients (60% in those exclusively with SUI and 54.1% in MUI). Bivariate analysis showed that patients with a history of gynecological surgery had 51.3% less probability for improvement compared to other groups without being statistically significant (p=0.128). Multivariate analysis (binary logistic regression) showed medication consumption as possible independent variable but the relation was not statistically significant. Subgroup analysis between patients with SUI and MUI showed that only improvement in endurance in patients with MUI related independently with >50% reduction in pad numbers (OR=3.794, p=0.019), signifying that increase in endurance for 1 sec results in 133%  increase in possible reduction of pad numbers (by 50%).

Other secondary outcomes. At baseline, 61.2% of women suffered from frequency compared to 31.4% in the last follow-up (p<0,001, McNemar test). Nocturia also improved significantly (67% vs. 43.6% of patients, p<0,001, McNemar test). Constipation also improved significantly, reported from 67 patients at baseline vs. 48 at 3 months (p<0,001 McNemar).
Digital evaluation of pelvic floor showed significant improvement in endurance (p<0.001, md1=4, IQR1=2, md2=6, IQR2=2), number of repetitions (p<0.001, md1=5, IQR1=1 - md2=6, IQR2=2) and fast contractions (p<0.001, md1=6, IQR1=2 - md2=8, IQR2=3) but not in muscle strength (p=0.157, Wilcoxon signed ranks test).

As in previous studies, PFMT significantly improved incontinence in both SUI and MUI patients. Interestingly, PFMT improved more significantly the urgency component rather than the SUI component in the MUI subpopulation, at least in terms of complete cure. The concomitant use of medication did not appear to significantly improve the outcome of PFMT in MUI patients. Not unexpectedly, patients with negative or mildly positive stress test were more likely to achieve complete cure or significant improvement with PFMT. The lack of prospective design is the main limitation of the study

Pelvic floor muscle training significantly improved incontinence in both SUI and MUI patient populations. Lower severity or negative stress test upon initial evaluation was the single predictive factor for successful outcome of PFMT in both patient populations, and in terms of both the SUI and UUI components. Improvement in endurance in patients with MUI related independently with >50% reduction in pad numbers. PFMT also improved frequency, nocturia and constipation