The analyses in this study are based on data from an RCT conducted between 2009 and 2011 (ID: NCT01032265). Women were enrolled in the project via an online screening survey that evaluated the type of incontinence and inclusion and exclusion criteria. The inclusion criteria were: female, age 18–70 years, SUI ≥1 time/week, ability to read and write Swedish, and access to a computer with the Internet. The exclusion criteria were: leakage associated with urgency, previous UI surgery, pregnancy, known malignancy in the lower abdomen, difficulties passing urine, macroscopic hematuria, intermenstrual bleeding, severe psychiatric disorders or Hospital Anxiety and Depression Scale (HADS) score >15 for depression or anxiety, and neurological disease affecting sensitivity in the legs or lower abdomen. Women considered eligible were sent self-assessment questionnaires for additional evaluation, including the validated questionnaires, the Lower Urinary Tract Symptoms Quality of Life (ICIQ-LUTSqol) and the International Consultation on Incontinence Modular Questionnaire Urinary Incontinence Short Form (ICIQ-UI SF), medical history, and a 2-day bladder diary. After their completion, a telephone interview was performed by a urotherapist to confirm the diagnosis of SUI. A total of 250 women aged 18–70 years with SUI ≥1 time/week were randomized by computer-generated block randomization to either an Internet-based training program or a program delivered by post to compare the effect of the two programs after a 3-month treatment period. The programs consisted of PFMT of increasing intensity three times per day for 3 months including contractions for strength and endurance, quick contractions, and the “knack maneuver” with information and lifestyle advice. No face-to-face education was provided before or during the PFMT program. Follow-up was performed at 4 months and 1 year using self-assessed questionnaires. The RCT was described in detail by Sjöström et al. .
All women who answered the 1-year follow-up questionnaires were included in the present study (n = 169). Participants in both groups achieved highly significant and clinically relevant improvements in the primary outcomes symptom severity and condition-specific quality of life, and the groups did not significantly differ in these measures . In the present study, the analyses include all participants regardless of treatment group. At 1 year, 3 % of the women (5 out of 169, 3 from the postal group and 2 from the Internet group) had undergone surgery for SUI and were also included in the study. None of the women received new medications for incontinence during the study period.
Definition of success
We used three outcomes to define success after 1 year:
Patient Global Impression of Improvement (PGI-I), a validated , patient-reported evaluation of the result of treatment consisting of one question: “How is your urinary leakage now compared with before treatment?” On a scale with seven alternatives ranging from very much worse to very much better, participants answering that they were much better or very much better were considered to have a successful outcome.
The ICIQ-UI SF is a validated and highly recommended symptom scoring instrument [3, 8, 16] that evaluates symptoms such as frequency, amount of leakage, and overall inconvenience. It consists of four questions: three adding up to a score (0–21 points) and the fourth used to determine the type of incontinence. The minimal important difference (MID) in improvement is the difference between the ICIQ-UI SF score at inclusion and at follow-up. An overall score reduction of 2.5 can be considered clinically relevant . We used ≥3 (rounded up to the closest integer) to define success.
“Sufficient treatment” is a question from the follow-up questionnaire: “Do you currently think that the treatment you underwent is sufficient?” answered with “No,” “Yes, I am completely cured from my urinary leakage,” or “Yes, I think the treatment is sufficient even though I am not completely cured.” We considered the two answer alternatives beginning with “yes” as a successful outcome.
The possible predictors of long-term success we chose to analyze included known risk factors for SUI [5–7] and suggested predictors from short-term results [11, 12, 18–20]. Because short-term success likely affects long-term success, we included data from the 4-month follow-up . We also included the amount of PFMT performed after 1 year. The data at baseline consisted of age, body mass index (kg/m2), education, menopausal status, parity, vaginal delivery of a child ≥4,000 g, local estrogen use, physical activity, time since onset of SUI, previously sought medical contact for SUI, ICIQ-UI SF, condition-specific quality of life as measured by the questionnaire ICIQ-LUTSqol, tea drinking, motivation to perform PFMT (Likert scale 1–10), and self-rated ability to perform PFMT (Likert scale 1–10). The data from the 4-month follow-up consisted of PGI-I, MID ICIQ-UI SF, sufficient treatment, and MID ICIQ-LUTSqol. The data from the 1-year follow-up concerned how often PFMT had been performed in the last 3 months, which could be answered with “Never,” “Sporadically, less than once a week,” “Regularly, 1–3 times/week,” “Regularly, more than 3 times/week,” or “Regularly, daily.”
The baseline variables ICIQ-UI SF, ICIQ-LUTSqol, and age were treated as continuous variables and the other possible predictors as categorical variables. At the 4-month follow-up, the possible predictors PGI-I, ICIQ-UISF, and sufficient treatment were analyzed in the same way as at the 1-year follow-up. The ICIQ-LUTSqol was divided into two categories based on the score reduction between inclusion and the 4-month follow-up; a score reduction exceeding the established MID (i.e., ≥4)  was considered a relevant improvement. The categorical variables were divided into categories on the original questionnaire. To have enough participants in each category to proceed with analyses, we had to create fewer categories by adding them together. The categories were reduced based on clinical relevance and the distribution of answers for the following predictors: body mass index (kg/m2), education, physical activity, time since onset of SUI, tea drinking, motivation to perform PFMT, self-rated ability to perform PFMT, and how often PFMT had been performed in the last 3 months at the 1-year follow-up. PFMT at 1 year was changed into two categories in which the answers beginning with “regularly” were in one category and “never” or “sporadically, less than one time/week” were another category, as shown in Table 2.
The outcome variables were recoded as 1 for success and 0 for failure. The relationships between each of the potential predictors and the outcomes were tested using the Chi-squared test for categorical variables, independent t test for continuous variables, and univariate logistic regression. Age and predictors with p < 0.25 were used in a final multivariate model in which non-significant (p > 0.05) predictors were manually removed one-by-one starting with the predictor with the highest p value, leaving only predictors with p < 0.05. The multivariate analysis was also performed without the five women who had undergone surgery.
All statistical analyses were performed using SPSS version 22.0 software.
The study was ethically approved by the Regional Ethical Review Board, Umeå University (number 08-124 M and 2015-79-32 M).