Study Characteristics of Studies With Urinary Incontinence Outcomes

The characteristics of the 140 included studies (in 151 articles) that reported UI outcomes are summarized in Appendix C, Table C-3. They are listed in alphabetical order. We identified 51 studies in the update search and included 89 from the 2012 AHRQ report.

Across all trials, the mean or median age of enrollees ranged between 33 and 85 (median 55; interquartile range [IQR] 50 to 59) years. Analyzed sample sizes ranged between 18 and 2393; median 85 (IQR 50 to 218).

Study Characteristics of Studies With Quality of Life Outcomes

Quality of life outcomes were evaluated in 96 studies. Appendix E (Table E-1) gives the baseline data for these studies. Ninety-five studies were randomized controlled trials (RCTs), and one was an nonrandomized comparative study (NRCS); 60 were newly identified in the update. The mean or median ages ranged from 32 to 85 years. Analyzed sample sizes ranged between 14 and 2393 for the RCTs, with a median of 57 (interquartile range [IQR] 33 to 128); the non-randomized study had 6844 participants. Appendix C contains details of study design and baseline (Table C-1) and interventions (Table C-2) for the new studies; information for the studies in the 2012 AHRQ report are given in the appendixes of that report.

The studies evaluated several quality of life domains: bother, daily activities, distress, general health, mental health, pain, sexual health, and sleep/energy. Results are given by KQ below. Appendix E contains summary (Table E-2) and detailed (Table E-3) quality of life results. Quality of life was not evaluated by network meta-analysis, but instead by within-study statistical significance and directionality.

Study Characteristics of Studies Reporting Adverse Events

Adverse events outcomes were evaluated in 127 studies. Most included studies are RCTs or NRCS, but 16 are single group (noncomparative) studies. Results for adverse events are given for each KQ below. Details of design and intervention for the new studies are given in Appendix C; information for the studies in the 2012 AHRQ review are given in the appendixes of that report. Full adverse events data are given in Appendix F.

Comparisons Across Intervention Categories

In total, 51 RCTs (7049 women) were included in this analysis of cure; studies ranged in size from 29 to 2487 people. ^3,27–74 Table 5 describes the intervention categories compared, the number of women who received each intervention, and the numbers of studies (and women) analyzing each comparison between intervention categories. Forty-five RCTs (83%) were deemed to be at low or moderate risk of bias.

Table 5

Summary description of all studies reporting on cure.

Table 6 shows the ORs for cure comparing all 14 intervention categories that have been evaluated. Further details about the network meta-analyses, including the analysis of individual interventions in each intervention category, are in Appendix G. Only 20 of 91 possible comparisons are informed by direct (head-to-head) comparisons, of which 9 are comparisons with no treatment. In Table 6, the direct comparisons are in the unshaded cells. Shaded cells correspond to comparisons that were inferred from the network meta-analysis model but had not been examined in the included RCTs. For example, alpha agonists have been compared only with neuromodulation and placebo; periurethral bulking and intravesical pressure release devices have each only been compared with no or sham treatment. Comparisons with other active intervention categories are indirect through comparisons other interventions. Indirect comparisons are more uncertain than those for which head-to-head data exist. The added uncertainty of indirect comparisons is partly reflected in the width of their respective confidence intervals, which are broader (often much broader) than for interventions with direct comparisons. For all comparisons that are empirically observed with direct comparisons (all nonshaded cells in the table), results using only head-to-head data (i.e., standard pairwise meta-analysis) agree well with the results from the network meta-analysis (data not shown).

Full Network Summary

First, we describe results from the full network regardless of their primary use for urgency or stress UI or whether they are first, second, or third line therapies. In the subgroup analyses section below, we restrict summaries to those interventions used primarily for stress UI separately from those interventions used primarily for urgency UI.

All active treatments appear to result in higher rates of cure than sham, placebo, or no treatment. The differences versus no treatment are statistically significant for the following active interventions: BTX, anticholinergics, combined hormones and behavioral therapy, neuromodulation, combined neuromodulation and behavioral therapy, and behavioral therapy (alone). There was also a near-significant effect compared with placebo (based on statistically nonsignificant ORs of >2.0 for which the lower bound of the confidence interval is ≥0.80) for combined anticholinergics and behavioral therapy and for intravesical pressure release.

Regarding comparisons of active interventions, based on only statistically significant differences, bladder BTX results in higher cure rates than alpha agonists, anticholinergics, and periurethral bulking. Both neuromodulation (alone) and behavioral therapy (alone) result in higher cure rates than either alpha agonists or anticholinergics. Combined neuromodulation and behavioral therapy also results in higher cure rates than alpha agonists.

Other evidence of possible comparative benefits suggest that intravesical pressure release devices and combined anticholinergics and behavioral therapy each result in higher rates of cure than no treatment; combined hormones and behavioral therapy is favored over alpha agonists; combined neuromodulation and behavioral therapy is favored over anticholinergics; BTX is favored over behavioral therapy; and periurethral bulking agents result in lower rates of cure than neuromodulation, behavioral therapy, and the combination of the two.

Of note, the evidence regarding hormones, combined hormones and anticholinergics, combined anticholinergics and behavioral therapy, and combined hormones, neuromodulation, and behavioral therapy was generally too sparse to allow confident comparisons with other interventions (and in most instances no treatment); 95 percent confidence intervals for all comparisons with these interventions were very wide.

Table 6

Odds ratios for cure between all intervention categories.

The league table (Table 7) offers complementary information from the same analysis. For each intervention category, it shows the mean and forecasted (from the network meta-analysis model) cure rates across the included RCTs. Bladder BTX (B), neuromodulation (N), behavioral therapy (T), and their evaluated combinations (N+T, H+T, H+N+T) had mean cure rates in the 30 to 45 percent range. Anticholinergics with or without behavioral therapy (C, C+T) hormones (H), alpha-agonists (A), periurethral bulking (U) or intravesical pressure release devices (V) had mean cure rates in the 14 to 28 percent range. Sham, placebo, or no treatment had a mean cure rate of 12 percent.

It should be noted that these summary results do not take into account characteristics of the women included in the studies that may be associated with resistance to treatment; thus, the summary findings may be confounded by study. In other words, the network meta-analyses assume that the women across all studies (and all other study characteristics) are generally similar. For example, they do not account for possible differences among women being considered for (and treated with) oral medications, injected or invasive interventions, or nonpharmacological interventions. Subgroup meta-analysis results are presented in the next section.

Descriptions of the comparisons across all individual interventions can be found in Appendix G. Briefly, the results of the analyses of intervention categories are congruent with the corresponding results of the analyses of individual interventions. However, many more of the specific comparisons have very broad confidence intervals because the comparisons across individual interventions are even more sparse than for comparisons of intervention categories.

Table 7

Mean and forecasted cure rates by intervention category (all).

Subgroup Analyses

Comparisons Across Intervention Categories

In total, 64 RCTs (13375 women) were included in this analysis; studies ranged in size from 29 to 5584 women. ^{17,27–29,32,33,36–38,41–43,48,52,54,55,57,58,60,63,65,67–70,73,75–111} Table 8 describes the intervention categories compared, the number of women who received each intervention, and the numbers of studies (and women) analyzing each comparison between intervention categories. Studies reported improvement outcomes for evaluations of 13 intervention categories. In contrast with the cure outcome, no studies reported improvement for the combination of hormones and behavioral therapy. Fifty-five RCTs (89%) were deemed to be at low or moderate risk of bias.

Table 8

Summary description of all studies reporting on improvement.

Table 9 shows the ORs for improvement comparing all 13 intervention categories that have been evaluated. Further details about the network meta-analyses, including the analysis of individual interventions in each intervention category, are in Appendix G. Only 19 of 78 possible comparisons are informed by direct (head-to-head) comparisons, of which 9 are comparisons with no treatment. In Table 9, the direct comparisons are in the unshaded cells. Shaded cells correspond to comparisons that were inferred from the network meta-analysis model but had not been examined in the included RCTs. For example, alpha agonists have been compared only with neuromodulation and placebo; periurethral bulking and intravesical pressure release devices have each only been compared with no or sham treatment. Comparisons with other active intervention categories are indirect through comparisons with other interventions. Indirect comparisons are more uncertain than those for which head-to-head data exist. The added uncertainty of indirect comparisons is partly reflected in the width of their respective confidence intervals, which are broader (often much broader) than for interventions with direct comparisons. For all comparisons that are empirically observed with direct comparisons (all nonshaded cells in the table), results using only head-to-head data (i.e., standard pairwise meta-analysis) agree well with the results from the network meta-analysis (data not shown).

Full Network Summary

First, we describe results from the full network regardless of their primary use for urgency or stress UI or whether they are first, second, or third line therapies. In the subgroup analyses section below, we restrict summaries to those interventions used primarily for stress UI separately from those interventions used primarily for urgency UI.

All active treatments appear to result in higher rates of improvement than sham, placebo, or no treatment. Hormones and combined hormones and anticholinergics had estimates with wide confidence intervals not suggesting a difference compared with placebo. The differences versus no treatment are statistically significant for the following active interventions: alpha agonists, BTX, anticholinergics, combined anticholinergics and behavioral therapy, neuromodulation, combined neuromodulation and behavioral therapy, the triple combination of hormones and neuromodulation and behavioral therapy, behavioral therapy (alone), and intravesical pressure release. More interventions were found to be more effective to achieve improvement, compared with placebo, than to achieve cure. Alpha agonists and the triple combination of hormones, neuromodulation, and behavioral therapy were not significantly different than no treatment to achieve cure. Intravesical pressure release and combined anticholinergics and behavioral therapy were only nonsignificantly more likely to achieve cure.

Regarding comparisons of active interventions, based on only statistically significant differences, hormones are less effective than all other interventions (excepting combined hormones and anticholinergics and periurethral bulking, for which there are nonsignificant differences). Alpha agonists were found to be statistically less effective to achieve improvement than neuromodulation, behavioral therapy, and combined neuromodulation and behavioral therapy. Anticholinergics were found to be significantly less effective than behavioral therapy.

Other evidence of possible comparative benefits, based on statistically nonsignificant ORs of >2.0 (for which the lower bound of the confidence interval is ≥0.80) suggest that alpha agonists are also less effective to achieve satisfaction than the triple combination of hormones, neuromodulation and behavioral therapy. Anticholinergics are also less effective than the same triple therapy, combined neuromodulation and behavioral therapy, and behavioral therapy alone.

Of note, the evidence regarding periurethral bulking and intravesical pressure release was generally too sparse to allow confident comparisons with other interventions; 95 percent confidence intervals for almost all comparisons with these interventions were very wide.

Table 9

Odds ratios for improvement between all intervention categories.

The league table (Table 10) offers complementary information from the same analysis. For each intervention category, it shows the mean and forecasted (from the network meta-analysis model) improvement rates across the included RCTs. With use of most interventions between about 60 and 70 percent of women had improvement in their symptoms (variously defined across the studies). These more successful interventions included behavioral therapy, neuromodulation, both combinations of the two and with addition of hormones, BTX, combination anticholinergics and behavioral therapy, and intravesical pressure release. With use of anticholinergics (alone), periurethral bulking, and alpha agonists, about 40 to 50 percent of women had improvement. About 25 percent of women on no treatment or placebo had improvement.

It should be noted that these summary results do not take into account characteristics of the women included in the studies that may be associated with resistance to treatment; thus, the summary findings may be confounded by study. In other words, the network meta-analyses assume that the women across all studies (and all other study characteristics) are generally similar. For example, they do not account for possible differences among women being considered for (and treated with) oral medications, injected or invasive interventions, or nonpharmacological interventions. Subgroup meta-analysis results are presented in the next section.

Descriptions of the comparisons across all individual interventions can be found in Appendix G. Briefly, the results of the analyses of intervention categories are congruent with the corresponding results of the analyses of individual interventions. However, many more of the specific comparisons have very broad confidence intervals because the comparisons across individual interventions are even more sparse than for comparisons of intervention categories.

Table 10

Mean and forecasted improvement rates by intervention category (all).

Subgroup Analyses

Comparisons Across Intervention Categories

In total, 12 RCTs (3008 people) were included in this analysis; studies ranged in size from 24 to 975 women. ^{29,31,32,35,52,88,92,103,113,114} Table 11 describes the intervention categories compared, the number of women who received each intervention, and the numbers of studies (and women) analyzing each comparison between intervention categories. Five RCTs (63%) were deemed to be at low or moderate risk of bias.

Table 11

Summary description of all studies reporting on satisfaction.

Table 12 shows the ORs for achieving satisfaction with control of UI symptoms. The table includes all seven intervention categories evaluated. Further details about the network meta-analyses, including the analysis of individual interventions in each intervention category, are in Appendix G. Only 8 of 21 possible comparisons are informed by direct (head-to-head) comparisons, of which 3 are comparisons with no treatment. In Table 12, the direct comparisons are in the unshaded cells. Shaded cells correspond to comparisons that were inferred from the network meta-analysis model but had not been examined in the included RCTs. For example, BTX has been compared only with neuromodulation. Comparisons with other active intervention categories are indirect through comparisons with other interventions. Indirect comparisons are more uncertain than those for which head-to-head data exist. The added uncertainty in indirect comparisons is partly reflected in the width of their respective 95 percent confidence intervals, which are broader (often much broader) for comparisons without direct comparisons than for those with direct comparisons. For all comparisons that are empirically observed with direct comparisons (all nonshaded cells in the table), results using only head-to-head data (i.e., standard pairwise meta-analysis) agree well with the results from the network meta-analysis (data not shown).

Full Network Summary

First, we describe results from the full network regardless of their primary use for urgency or stress UI or whether they are first, second, or third line therapies. In the subgroup analyses section below, we restrict summaries to those interventions used primarily for stress UI separately from those interventions used primarily for urgency UI.

All active treatments appear to result in higher rates of satisfaction with control of UI symptoms than sham, placebo, or no treatment.

Regarding comparisons of active interventions, based on only statistically significant differences, treatment with either BTX or neuromodulation resulted in more women being satisfied than anticholinergics or combination anticholinergics and behavioral therapy. Behavioral therapy alone was found to be significantly more effective than anticholinergics alone.

Other evidence of possible comparative benefits, based on statistically nonsignificant ORs of >2.0 (for which the lower bound of the confidence interval is ≥0.80) suggests that anticholinergics are also less effective than combination neuromodulation and behavioral therapy.

Table 12

Odds ratios for satisfaction between all intervention categories.

The league table (Table 13) offers complementary information from the same analysis. For each intervention category, it shows the mean and forecasted (from the network meta-analysis model) satisfaction rates across the included RCTs. Most women were satisfied with any of the active treatments. BTX, neuromodulation, behavioral therapy, and combination neuromodulation and behavioral therapy achieved satisfaction rates of about 80 to 85 percent; 66 percent of women using combination anticholinergics and behavioral therapy, as did 55 percent of women using anticholinergics alone. Only 32 percent of women in no treatment study arms achieved satisfactory control of their UI symptoms.

It should be noted that these summary results do not take into account characteristics of the women included in the studies that may be associated with resistance to treatment; thus, the summary findings may be confounded by study. In other words, the network meta-analyses assume that the women across all studies (and all other study characteristics) are generally similar. For example, they do not account for possible differences among women being considered for (and treated with) oral medications, injected or invasive interventions, or nonpharmacological interventions. Subgroup meta-analysis results are presented in the next section.

Descriptions of the comparisons across all individual interventions can be found in Appendix G. Briefly, the results of the analyses of intervention categories are congruent with the corresponding results of the analyses of individual interventions. However, many more of the specific comparisons have very broad confidence intervals because the comparisons across individual interventions are even more sparse than for comparisons of intervention categories.

Table 13

Mean and forecasted satisfaction rates by intervention category (all).

Subgroup Analyses

Cure

All nonpharmacological treatments were associated with better cure rates than sham or no treatment (Table 14). Across all studies, the point estimates for the cure rates are between 27 and 35 percent for the active treatments versus 12 percent for sham or no treatment. With one exception, cure rates were similar in the studies restricted to stress or urgency UI (among relevant interventions). Behavioral therapy (1^st line therapy) was somewhat higher in studies of women with stress UI (46%) than across all studies (30%). Except for intravesical pressure release, all nonpharmacological treatments resulted in statistically significantly higher cure rates than no or sham treatment, with ORs ranging from 3.1 to 4.0 across all studies (see Table 6). Intravesical pressure release (3^rd line therapy) attained near-significance against no treatment (OR 2.7, 95% CI 0.8 to 9.0) across all studies, although the estimate was highly imprecise in studies of women with stress UI only.

With respect to cure, there were no statistically significant differences between categories of nonpharmacological interventions. From Table 6, the point estimates of the ORs between any two nonpharmacological treatments are smaller than 1.45 (or greater than its inverse). However, the confidence intervals are generally broad and cannot exclude relatively large differences. Estimates of differences were similar, but less precise, from studies restricted to women with either stress UI or urgency UI (Appendix H, Tables H-1 to H-2).

Table 14

Summary league table displaying percent of women with UI outcomes for each nonpharmacological treatment.

Improvement

All nonpharmacological treatments were associated with better improvement rates than sham or no treatment (Table 14). Across all studies, the point estimates for the improvement rates are between about 60 and 70 percent for the active treatments versus 25 percent for sham or no treatment. Improvement rates were for all evaluated nonpharmacological treatments were higher in studies of women with urgency UI only than across all studies. In studies of women with stress UI combination neuromodulation and behavioral therapy resulted in a low average rate of improvement (27%). All nonpharmacological treatments resulted in statistically significantly higher improvement rates than no or sham treatment, with ORs ranging from 4.2 to 6.7 across all studies (see Table 9). Estimates of effect compared with no or sham treatment were similar in studies restricted to women with stress UI or urgency UI, except that combination neuromodulation and behavioral therapy was not significantly different than control (see Appendix H, Tables H-4 and H-5).

With respect to improvement, there were no statistically significant differences between categories of nonpharmacological interventions. From Table 9, the point estimates of the ORs between any two nonpharmacological treatments are nonsignificant and imprecise, with confidence intervals that cannot exclude relatively large differences. Estimates of differences were mostly similar from studies restricted to women with either stress UI or urgency UI, except that in studies of stress UI, combination neuromodulation and behavioral therapy (3^rd line therapy) was significantly less effective than either neuromodulation (3^rd line therapy) or behavioral therapy (1^st line therapy) alone (Appendix H, Tables H-4 and H-5).

Satisfaction

The three evaluated nonpharmacological treatments were associated with better satisfaction rates than sham or no treatment (Table 14). Across all studies, the point estimates for the satisfaction rates are about 80 percent for the active treatments versus 32 percent for sham or no treatment. Rates of satisfaction after behavioral therapy were similar in studies of women with only stress UI or urgency UI. Satisfaction rates after neuromodulation were also similar in studies of women with stress UI, but were lower (51%) in studies of women with urgency UI. All nonpharmacological treatments resulted in statistically significantly higher satisfaction rates than no or sham treatment, with ORs ranging from 8.0 to 10.7 across all studies (see Table 12). Estimates of effect compared with no or sham treatment were similar in studies restricted to women with stress UI or urgency UI (see Appendix H, Tables H-7 to H-8).

With respect to satisfaction, there were no statistically significant differences between categories of nonpharmacological interventions. From Table 12, the point estimates of the ORs between any two nonpharmacological treatments are nonsignificant and imprecise, with confidence intervals that cannot exclude relatively large differences. From studies restricted to women with either stress UI, the only available comparison found a nonsignificant difference between neuromodulation (3^rd line therapy) and behavioral therapy (1^st line therapy) among women with stress UI, based in part on direct head-to-head comparisons (Appendix H, Tables H-7 to H-8). Among studies of women with urgency UI, BTX (3^rd line therapy) was nonsignificantly favored to achieve satisfaction over neuromodulation (OR 1.40, 95% CI 0.93 to 2.12).

Quality of Life

Seventy-eight studies reported on quality of life outcomes for the comparison of nonpharmacological interventions versus placebo or other nonpharmacological interventions. ^{28,31,35,43,46,49,51–53,56,63,66,69,74,93,98,103,108,109,111,115–170} The summary results are presented in Tables 15 and 16. The table cells show the number of studies and the number of people (in parentheses), followed by the number of studies with the number of studies that found statistically significant differences and which intervention was favored, the number of studies that found discordant results (that is, within the same study, significant differences favoring one intervention were found on one scale or subscale, but nonsignificant differences were found on others), and the number of studies with nonsignificant differences. Study-level details, including citations, are presented in Appendix E.

Nonpharmacological Interventions Versus Sham or No Treatment

Fifteen nonpharmacological interventions (including combinations of interventions) were compared with sham or no treatments in 36 studies (Table 15). Twenty-three of these studies evaluated behavioral interventions, which included bladder training, education, PFMT, biofeedback, electroacupuncture, weight loss, and yoga, and combinations of these interventions, ^{35,49,52,53,56,69,93,98,103,109,111,117,119–121,125,127,134,137,142,147,148,150,155,156,162–165,167–169} three combinations of behavioral and neuromodulation,^28,43,162 and one an intravesical pressure release device.⁶³ The studies mostly analyzed daily activity, bother, general health, and mental health.

Across interventions, the effect of the nonpharmacological interventions on various aspects of quality of life was mixed. Among interventions compared by more than one study, only TENS (alone) was found by all studies to be statistically significantly better than sham for any specific aspect of quality of life. TENS alone was found to result in better mental health quality of life (2 studies).^35,93 All interventions evaluated by more than one study were found to have a statistically significant improvement in at least one aspect of quality of life by at least one study, except for PFMT, in which only discordant or nonsignificant results were found in daily activities and general health. Only two interventions for which any aspect of quality of life was evaluated were not found to result in better quality of life than sham: combined bladder training and PFMT, and the intravesical pressure release device. However, the data for these interventions were sparse. The two studies evaluating these two interventions each reported on only a single aspect of quality of life.

Table 15

Quality of life outcomes for nonpharmacological interventions versus sham/no treatment.

Nonpharmacological Interventions Versus Other Nonpharmacological Interventions

Nineteen nonpharmacological interventions (including combinations of interventions) were compared with other active interventions (Table 16). Forty-two studies evaluated behavioral interventions versus other behavioral interventions (or combinations of behavioral interventions), including bladder training, education, PFMT, biofeedback, electroacupuncture, weight loss, and yoga. ^{31,46,51,66,74,108,115–118,122–124,126,128–133,135,136,138–141,143–146,149,151–153,157–162,166,170} One study evaluated neuromodulation (electric stimulation) versus behavioral therapy,³⁵ and ten evaluated combinations of behavioral therapy and neuromodulation versus behavioral therapy alone. ^{31,66,108,116,117,136,138,159,162} These studies mostly analyzed bother, daily activities, distress, general health, mental health, and sexual health.

Few interventions were compared by more than one study, and these contrasts generally were split between discordant or nonsignificant findings when analyzing bother, daily activities, or general health. Supervised PFMT was found to improve quality of life statistically significantly more than unsupervised PFMT or other specific types of exercise or physical therapy in one study for bother and in two studies each for the domains of daily activities and mental health. The other studies evaluating these comparisons reported discordant or nonsignificant differences between the interventions.^{46,51,74,117,118,123,124,126,128–130,133,139,141,146,149,152,153,157,170}

One study reported statistically significant improvements in the daily activities domain with PFMT and biofeedback compared with PFMT alone, and one study reported significant improvements in distress for bladder training combined with PFMT and biofeedback when compared to bladder training alone or PFMT with biofeedback^117,166 However, nine studies either reported discordant or nonsignificant differences across all other domains for this comparison.^{35,74,124,128,129,131,133,138,149}

Additionally, two studies found significant improvements among the control groups for two outcomes; bladder training alone was found to have significant improvements in bother ratings over combined TENS, PFMT, and biofeedback, and electroacupuncture alone was found to have significant improvements when compared to PFMT with weights in assessing impact on daily activities.^108,131

Table 16

Quality of life outcomes for nonpharmacological treatments versus other nonpharmacological treatments.

Adverse Events

Fifty-two studies reported on adverse events in studies of nonpharmacological interventions.^{14,29,32,46,54,60,69,92,95,98,109,115,119,120,123,126,128,129,132,133,136,138,140,144,147,149,150,154,156,157,160,161,165,168,169,171–187} The results are presented in Tables 17 and 18 (parts 1 and 2, respectively). For adverse events that were reported only by a single study (arm), the table cells show the percentage of people affected by the adverse event. Where more than one study arm reported on an adverse event, the cell gives the median and range of percentages (or just the range, if only two arms reported an adverse event). Detailed results are in Appendix F.

In general, the percentages of women with adverse events were low, but reporting was sparse with only one or two studies reporting adverse events for most of the interventions. No specific adverse event (e.g., urinary tract infection, diarrhea, dry mouth) was reported in more than one study for any specific intervention (e.g., acupuncture, bladder support). The most commonly reported adverse event was the group nonmajor or undefined adverse event. In two studies of bladder training, four studies of education, and two of three studies of magnetic stimulation (with 76 women), no adverse events were reported. The other study of magnetic stimulation reported that 3 of 60 women (5%) had an undefined adverse event. In two studies of InterStim™ (a form of sacral neuromodulation), 1 of 284 women total (0.4%) had a “serious” adverse event (an implant site erosion). In two studies of electroacupuncture (287 women) one reported no adverse events in 24 women, the other reported four adverse events in 247 women. A small study of bladder support (29 women) reported no adverse events.

Among 13 studies of TENS, nine reported no adverse events in 571 women, but two reported any or any moderate adverse event in a total of 13 of 67 women (19%). Overall, adverse events were reported in 2.8 percent of 463 women receiving TENS. In three studies of TENS, 11 percent of women (n = 199) reported urinary tract infections; however, within studies, these rates were similar to or lower than urinary tract infection rates on an anticholinergic (6%) or with BTX (35%).

Table 17

Adverse events in nonpharmacological interventions, part 1 (acupuncture to MBSR).

Table 18

Adverse events in nonpharmacological interventions, part 2 (PFMT to yoga).

Cure

Among pharmacological treatments, BTX had an average cure rate of 44 percent, hormones 28 percent, and anticholinergics 21 percent, compared with other pharmacological treatments and placebo, which all had cure rates less that 16 percent (Table 19). However, only BTX (OR 5.7) and anticholinergics (OR 2.0) had statistically significant greater likelihood of cure than placebo; the estimate for hormones was less precise and nonsignificant (see Table 6). Estimates from studies of women with stress UI or urgency UI were generally similar, except for an atypically low cure rate for hormones in studies of stress UI. Most comparisons across were based on indirect comparisons and resulted in imprecise OR estimates. However, based in part on a direct head-to-head comparison, BTX (3^rd line therapy) was significantly more likely to result in cure than anticholinergics (2^nd line therapy, OR 2.91). Among studies of women with stress UI, hormones and alpha agonists had similar cure rates (no other pharmacological treatments were compared directly or indirectly). Among studies of women with stress UI, cure rates among pharmacological treatments were mostly nonsignificantly different than each (Appendix H, Tables H-1 to H-2), except for the comparison of BTX and anticholinergics described above.

Table 19

Summary league table displaying percent of women with UI outcomes for each pharmacological treatment.

Improvement

Among pharmacological treatments, BTX, anticholinergics, periurethral bulking agents, and alpha agonists had improvement rates, across studies of 42 to 67 percent. All other pharmacological treatments and placebo had improvement rates less than 25 percent (Table 19). Improvement rates in studies of women with stress UI or with urgency UI were mostly similar for the evaluated pharmacological treatments, except it was lower with anticholinergics among women with stress UI (25%). Alpha agonists (OR 2.2), BTX (OR 6.0), and anticholinergics (OR 3.0) each had significantly higher improvement rates than placebo (see Table 9). Other pharmacological interventions had imprecise OR estimates compared with placebo. Studies restrict to women with stress UI or with urgency UI had similar findings for evaluated pharmacological treatments. (see Appendix H, Tables H-4 and H-5).

With respect to improvement, hormones were significantly less effective than other pharmacological interventions, except in comparison with the combination hormones and anticholinergics for which the estimate was imprecise. BTX (3^rd line therapy) was likely more effective than alpha agonists and combination hormones and anticholinergics (both 2^nd line therapy), although the differences were nonsignificant. Findings were similar across evaluated treatments for women with urgency UI. In studies of women with stress UI alpha agonists were significantly more effective than hormones (OR 4.1, see Appendix H, Tables H-4 and H-5).

Satisfaction

Among studies that reported satisfaction, the only pharmacological interventions evaluated were BTX (3^rd line therapy) and anticholinergics (2^nd line therapy). With BTX, 86 percent of women had satisfaction with the control of their UI (Table 19). With anticholinergics, 55 percent of women had satisfaction. These compare to placebo treatment, with which 32 percent of women had satisfaction. Both pharmacological treatments were significantly more effective than placebo (BTX OR 12.7; anticholinergics OR 2.6; see Table 12). The two pharmacological treatments were not compared directly.

Quality of Life

Sixteen RCTs reported on quality of life outcomes in pharmacological interventions versus placebo or with other pharmacological interventions.^{38,45,62,70,84,96,114,117,188–195} Results are given in Table 20 and Appendix E. The table cells show the number of studies and the number of people (in parentheses), followed by the number of studies with the number of studies that found statistically significant differences and which intervention was favored, the number of studies that found discordant results (that is, within the same study, significant differences favoring one intervention were found on one scale or subscale, but nonsignificant differences were found on others), and the number of studies with nonsignificant differences. Study-level details are given in Appendix E. The studies evaluated several quality of life domains, as categorized across the columns of the tables: bother, daily activities, distress, general health, mental health, and sleep/energy.

Eight pharmacological interventions were compared with placebo interventions (Table 20). Six studies evaluated anticholinergic medications,^{114,117,190,191,193–195} two bladder BTX,^45,190 and one each an alpha agonist,⁸⁴, an antiepileptic (pregabalin),¹⁹² and a periurethral bulking agent.³⁸ Across interventions, the effects of the pharmacological interventions on various aspects of quality of life were mixed. Only two specific interventions (oxybutynin and tolterodine) were compared in more than one study, and none in more than two.

When anticholinergic medications were evaluated as a single group, four studies evaluated bother,^{114,117,191,194} and four studies evaluated daily activities.^{114,117,193,195} When these studies were significant they favored the anticholinergic medication over placebo,^114,193 though most were not significant and one gave discordant results.¹⁹⁵

Table 20

Quality of life outcomes for pharmacological interventions versus placebo/no treatment.

Eight pharmacological interventions were compared with other pharmacological interventions (Table 21). Three studies evaluated either different anticholinergics ^96,189 or different doses of the same medication.¹⁹⁰ Only one found a significant difference in one aspect of quality of life, favoring 3.9 mg of oxybutynin over 2.6 mg of the same anticholinergic for improvements in daily activities. A single study evaluated an anticholinergic compared with BTX and reported a nonsignificant difference.⁷⁰ One study reported no significant difference in a comparison of an anticholinergic alone compared with an anticholinergic plus vaginal estrogen.¹⁸⁸ One study reported nonsignificant difference of the antiepileptic pregabalin over tolterodine in the bother domain, and the same study found no significant difference in bother between pregabalin versus pregabalin plus tolterodine.¹⁹²

Table 21

Quality of life outcomes for pharmacological versus pharmacological interventions.

Adverse Events

Ninety-five studies reported on adverse events in drugs.^{29,32,34,38,39,45,46,54,57–61,68–70,76,78,82–84,88,89,91,94,96,98,99,101,105,106,109,112,120,147,154,160,165,168,171–173,188–192,194,196–242} Results for anticholinergics are given in Table 22, and results for all other drugs, as well as for placebo arms (56 studies^{14,32,38,39,45,54,57–59,61,68,69,76,78,83,84,88,89,91,92,94,98,99,101,105,106,109,112,120,147,154,160,165,168,172,173,190–192,194,202,204,207–211,214,218,219,222,230,233,237,238,242}), are given in Table 23. The table cells show the percentage of people affected by the adverse event if only a single study arm reported the event. Where more than one study arm reported on an adverse event, the cell gives the median and range of percentages (or just the range, if only two arms reported an adverse event). Detailed results are in Appendix F. Fifty-one studies evaluated adverse events in anticholinergics,^{32,34,39,46,58,60,61,68,70,88,94,96,105,112,188–192,194,196–199,201–203,205,206,209–211,213,215–218,220–222,229,231–234,236–238,240–242} but with the exception of oxybutynin and tolterodine, each specific adverse event was evaluated in only one to three studies.

For most pharmacological interventions, serious adverse events (as described by authors) were rare or did not occur; although few studies defined serious adverse events. However, with periurethral bulking agents, 4.7 percent of 362 women in three studies had serious adverse events, including erosion and need for surgical excision of the bulking agents. The one study of a periurethral bulking agent currently available in the United States (macroplastique) reported 1.6% rate of erosion. In eight studies of anticholinergics, overall 2.4 percent of 2,583 women had serious adverse events, although the adverse events were mostly undefined. In two studies, 0.6 percent of 1,390 women taking the alpha agonist duloxetine had (undefined) serious adverse events. By comparison, in 10 studies, 0.2 percent of 2,852 women taking placebo (or no treatment) had (mostly undefined) serious adverse events.

The most commonly reported adverse event across interventions was dry mouth. The median percentage of women reporting dry mouth in studies of anticholinergic medications was 24.2 percent; oxybutynin was evaluated in the most studies (21) with a median of 36.1 percent (range 0 to 100).^{32,34,39,58,68,105,189,190,196,198,199,201,206,215,229,232,234,238,240,242–244} A single study each evaluated dry mouth in beta agonist (0.7%),²⁰⁰ BTX (30.8%),⁷⁰ estrogen (21.4%),²⁴³ and antiepileptic medications (10.5%).¹⁹² Fifteen studies evaluated dry mouth for duloxetine, an alpha agonist, and found a median of 12.9 percent (range 1.5 to 21.8). Placebo arms had a median of 4 percent across 29 studies (range 0 to 86.2).^{32,39,57–59,61,68,76,83,84,89,91,99,101,105,106,190,192,204,208–210,214,219,222,233,238,242} By comparison, in non-pharmaceutical treatments including PFMT and/or bladder training, dry mouth was reported in three studies with median rates that ranged from 0 to 34.9 percent.^32,179,245

Among the other active drugs (see Table 23), duloxetine, an alpha agonist, was evaluated in the largest number of studies (16).^{57,59,76,83,84,89,91,99,101,106,204,208,214,219,224,230} In addition to dry mouth, discussed above, dizziness (10.6% 14 studies), gastrointestinal upset, specifically constipation (11%; 14 studies) and nausea (23.2%; 15 studies), fatigue (8.6%, 12 studies), headache (8.3%; 11 studies), and insomnia (12.6%; 13 studies) were all reported in more than ten studies.

Only six studies reported on adverse events in bladder BTX.^{29,45,70,78,207,227} Five reported on urinary tract infections (UTI). The percentage of women reporting UTI ranged from 33.3 to 42.9 percent in the four studies with at least 20 participants (6/11 or 55% in the fifth study), with a median of 36.4 percent.^{29,45,70,78,207,227} This compares to a reported percentage of 1.5 percent for mirabegron, between 1.4 and 23 percent for various anticholinergics, and 4.3 percent in placebo arms. The sixth study reported hematuria in 2.3 percent of women using bladder BTX.^{29,45,70,78,207,227} Two studies reported on urine retention/voiding dysfunction in10.5 and 20.5 percent of women.^{29,45,70,78,207,227}

The most commonly reported adverse event for the periurethral bulking agents was UTI in five studies (median 6.6%; range 1.3% to 23.8%)^{38,223,225,226,235} and urinary retention/voiding dysfunction in seven studies (median 3.8%; range 0.9% to 9.5%).^{38,212,223,225,228,235,239}. However, only one of these studies (122 women) evaluated a periurethral bulking agent currently available in the United States (macroplastique). This study found high rates of urinary tract infection (23.8%), headache (18%), and urinary retention/dysuria (15.6%). Serious adverse events (erosion) were low (1.6%), as were pain (5%) and yeast infection (2.5%).³⁸ All other single and combination medications were evaluated in only one or two studies each.

Table 22

Adverse events reported for anticholinergics.

Table 23

Adverse events reported for drugs other than anticholinergics.

Stress UI

Urgency UI

Improvement

The same pairs of interventions evaluated for cure have evidence regarding rates of improvement. Similarly, there was direct (head-to-head) comparisons and evidence from urgency UI studies only for anticholinergics versus behavioral therapy and BTX versus neuromodulation.

There is evidence that behavioral therapy is more effective than any of the three pharmacological treatments. Compared with anticholinergics, the OR was 1.82 (95% CI 1.03 to 3.23) across all studies and 4.17 (95% CI 1.61 to 11.1) in urgency UI studies. Compared with hormones, the OR was 10.0 (95% CI 3.57 and 33.3). Compared with combination hormones and anticholinergics, behavioral therapy was likely more effective to achieve improvement (OR 5.26, 95% CI 0.86 to 33.3).

The direct comparisons between BTX and neuromodulation found no significant difference between the two interventions (Table 9 and Appendix H, Table H-5).

Satisfaction

Hormones have not been studied among studies reporting satisfaction. Therefore, there are only two evaluable comparisons.

Behavioral therapy was found to be significantly more effective than anticholinergics (OR 3.12, 95% CI 1.85 to 5.26) across all studies; however, the difference was similar but not statistically significant in the smaller number of urgency UI studies (OR 3.12, 95% CI 0.71 to 14.3). The direct comparisons between BTX and neuromodulation found no significant difference between the two interventions (Table 12 and Appendix H, Table H-8A).

Quality of Life

Four RCTs reported on quality of life outcomes in nonpharmacological interventions versus pharmacological interventions, one for each comparison against oxybutynin, trospium, and BTX; the remaining two RCTs reported comparisons against tolterodine.^29,46,60,117 Results are given in Table 24 and Appendix E. The table cells show the number of studies and the number of people (in parentheses), followed by the number of studies with the number of studies that found statistically significant differences and which intervention was favored, the number of studies that found discordant results (that is, within the same study, significant differences favoring one intervention were found on one scale or subscale, but nonsignificant differences were found on others), and the number of studies with nonsignificant differences. No significant differences were seen, and one study found discordant results on daily activities when comparing neuromodulation to BTX.²⁹

Table 24

Quality of life outcomes for nonpharmacological versus pharmacological interventions.

Adverse Events

These comparisons have been described under the adverse events sections for KQ 1 and 2.

All UI Outcomes

A summary of the direct comparisons between combinations intervention categories, and sham or no treatment, pharmacological and nonpharmacological intervention categories is in Table 25. Refer to Tables 6, 9, and 12 for combined direct and indirect (network meta-analysis) comparisons with all other interventions and to Tables 7, 10, and 13 for the mean event rates for the outcomes of cure, improvement, and satisfaction, respectively.

Single (separate) studies reported that combined neuromodulation, behavioral therapy, and hormones (vaginal estrogen) resulted in significantly more women cured or improved than either hormones alone⁸² or no treatment.⁴²

Combined anticholinergics and behavioral therapy had similar cure and improvement rates compared with anticholinergics alone, but significantly higher rates of satisfaction.^17,79 In separate studies, combined behavioral therapy and hormones had similar cure rates as vaginal estrogen alone²⁴⁶ and behavioral therapy alone.⁴⁴

Table 25

Comparisons of cure, improvement, and satisfaction rates between intervention categories: combination versus no treatment, pharmacological, and nonpharmacological interventions.

Quality of Life

A single study of 69 people reported on quality of life in combined pharmacological and nonpharmacological versus nonpharmacological only.¹⁷⁷ Both arms received PFMT, electrostimulation, and biofeedback, but one arm also received vaginal estrogen. The arm that received estrogen did significantly better on the Incontinence Impact Questionnaire (IIQ-7), with a net difference of −7.8 (95% CI −9.6 to −6) on the 100-point scale.

Adverse Events

Rates of adverse events from combined interventions are presented in Table 26. Each adverse event was reported in only a single study. Detailed results are in Appendix F. One study reported on a combination of PFMT and the anticholinergic medication trospium, but reported only on three adverse events in 31 people.¹⁷⁹ In this study low percentages of women reported visual adverse events (3.2%) and discontinuation due to adverse events (3.2%), but more reported dry mouth (23%).¹⁷⁹ Four studies reported on adverse events from estrogen combined with PFMT, pessaries, and/or transcutaneous electrical nerve stimulation (TENS).^{82,171,177,224} Each adverse event was evaluated by only a single study, but the percentage of women reporting these adverse events were below 4 percent.

Table 26

Adverse events in combination interventions.