Double incontinence among women globally: A systematic review and meta-analysis

Table 1 Characteristics of included studies

Ref.	Country	Study design	Setting	Population	Age range/mean (years)	Sample size (n)	DI cases (n)	Prevalence (%)	95%CI
Yuaso et al[7], 2018	Brazil	Prospective longitudinal	Community	Elderly women, São Paulo	65-90/74.6 ± 9.5	865	42	4.9	3.7-6.5
Ribeiro et al[5], 2019	Brazil	Cross-sectional retrospective	Public tertiary hospital	UI/FI patients	30-86/60.07 ± 11.10	227	32	14.1	9.9-19.2
Gallas et al[1], 2018	Tunisia	Cross-sectional	Hospital-based	Female doctors and nurses	23-60/36.8 ± 8.32	402	24	6.0	4.1-8.7
Güngör Uğurlucan et al[3], 2014	Turkey	Retrospective cohort	Medical school clinic	UI/POP patients	21-85/53.8 ± 12.0	2518	311	12.3	11.0-13.6
Wu et al[8], 2015	United States	Population-based cross-sectional	NHANES 2005-2010	Community women ≥ 50	50 - ≥ 80	3497	227	6.0	5.0-7.1
Matthews et al[12], 2013	United States	Cross-sectional	Community	Female nurses	62-87/72.7	64396	4660	7.2	7.0-7.4
Johannessen et al[2], 2018	Norway	Prospective cohort	Hospitals	Primiparous women, 1 year postpartum	19-42/28.6-29.8	976	82	8.4	6.8-10.2
	Norway	Prospective cohort	Hospitals	Primiparous women, late pregnancy	19-42	1031	127	13.0	10.8-15.3
Luo et al[4], 2020	China	Cross-sectional multisite	Community-based	Rural elderly women ≥ 65	65-92/72.71	700	73	10.4	8.0-12.8
Slieker-ten Hove et al[13], 2010	Netherlands	Cross-sectional	General population	General female population	45-85/58.0 ± 9.2	1397	143	10.3	8.7-12.2

DI: Double incontinence; CI: Confidence interval; UI: Urinary incontinence; FI: Fecal incontinence; NHANES: National Health and Nutrition Examination Survey; POP: Pelvic organ prolapse.

Table 2 Random-effects meta-analysis results for pooled prevalence of double incontinence

Parameter	Estimate	Standard error	95%CI	Statistical significance
Pooled logit prevalence	-2.3500	0.111	-2.568 to -2.131	P < 0.001, Z = -21.1
Pooled prevalence (%)	8.66	-	7.14-10.63	P < 0.001

CI: Confidence interval.

Table 3 Comprehensive assessment of heterogeneity

Heterogeneity metric	Value	95%CI	P value	Interpretation
I² (%)	32.7	0.0-62.8	-	Moderate heterogeneity (30%-60% range)
Tau²	0.0374	0.000-0.152	-	Moderate between-study variance
Tau (SD)	0.193	0.000-0.390	-	SD on logit scale
95% prediction interval (%)	-	6.5-11.5	-	Expected range of true study effects
Q-statistic	13.396	-	0.145	Not significant (low power with n = 10)
H²	1.49	1.11-2.69	-	Observed variance 49% > expected
Tau²/total variance (%)	27.3	-	-	Proportion of variance due to heterogeneity

I² confidence interval calculated using non-central χ² method; Tau² confidence interval using Q-profile method. Prediction interval represents expected range of 95% of true study-specific effects. CI: Confidence interval.

Table 4 Sensitivity analysis results

Analysis type	Studies (n)	Pooled prevalence (%)	95%CI	I² (%)	Tau²
Random-effects (primary)	10	8.66	7.14-10.63	32.7	0.0374
Fixed-effects	10	7.23	7.04-7.43	32.7	N/A
Excluding Matthews et al[12], 2013	9	9.08	7.21-11.42	35.4	0.0428
Excluding Ribeiro et al[5], 2019	9	8.52	7.02-10.35	30.1	0.0351
Excluding Yuaso et al[7], 2018	9	8.92	7.32-10.87	33.8	0.0395
Excluding Gallas et al[1], 2018	9	8.84	7.26-10.75	33.5	0.0388
Excluding Güngör Uğurlucan et al[3], 2014	9	8.47	6.93-10.33	29.8	0.0345
Excluding Wu et al[8], 2015	9	8.95	7.34-10.90	34.2	0.0398
Excluding Johannessen et al[2], 2018	9	8.71	7.13-10.63	34.0	0.0385
	9	8.31	6.81-10.14	27.5	0.0318
Excluding Luo et al[4], 2020	9	8.54	7.00-10.40	31.8	0.0365
Excluding Slieker-ten Hove et al[13], 2010	9	8.48	6.93-10.35	30.5	0.0352
Range (leave-one-out)	-	8.31-9.25	-	27.5-35.4	0.0318-0.0428

All leave-one-out estimates overlap with the primary 95% confidence interval, confirming robustness. CI: Confidence interval; N/A: Not applicable.

Table 5 Meta-regression analysis of predictors of heterogeneity

Covariate	Coefficient	SE	95%CI	Z	P value	R² (%)
Study setting
Clinical/hospital vs community	0.334	0.169	-0.007 to 0.675	1.98	0.063	14.2
Geographic region
Other vs Europe/North America	0.108	0.168	-0.231 to 0.447	0.64	0.524	2.1
Mean participant age
Per 10-year increase	-0.052	0.092	-0.238 to 0.134	-0.56	0.576	1.5
Sample size
Log (n)	-0.079	0.096	-0.274 to 0.116	-0.82	0.418	3.2
Study quality score
Per 1-point increase (JBI scale)	0.028	0.047	-0.067 to 0.123	0.59	0.557	0.0
Full model (all covariates)	-	-	-	-	-	23.8
Residual heterogeneity
Residual Tau²	0.0285	-	-	-	-	-
Residual I²	28.5%	-	-	-	-	-

Coefficients on logit scale. R² represents proportion of Tau² explained by each covariate. Full model R² of 23.8% indicates that 76.2% of heterogeneity unexplained. CI: Confidence interval; JBI: Joanna Briggs Institute.

Table 6 Cumulative meta-analysis (chronological order, 2009-2020)

Ref.	Cumulative n studies	Cumulative sample size	Cumulative prevalence (%)	95%CI	I² (%)
Slieker-ten Hove et al[13], 2010	1	1397	10.3	8.7-12.2	N/A
Matthews et al[12], 2013	2	65793	7.3	7.1-7.5	48.2
Güngör Uğurlucan et al[3], 2014	3	68311	7.8	7.2-8.5	58.7
Wu et al[8], 2015	4	71808	7.5	7.0-8.1	51.3
Johannessen et al[2], 2018	5	72784	7.6	7.1-8.2	47.8
	6	73815	7.9	7.3-8.5	45.9
Yuaso et al[7], 2018	7	74680	7.7	7.1-8.3	43.2
Gallas et al[1], 2018	8	75082	7.6	7.0-8.2	40.5
Ribeiro et al[5], 2019	9	75309	7.8	7.2-8.5	38.4
Luo et al[4], 2020	10	6009	8.66	7.14-10.63	32.7

Stabilization observed from 2015 onward (studies 4-10: Range 75%-8.66%). No temporal trend in prevalence; confidence interval narrowing reflects increased precision. CI: Confidence interval; N/A: Not applicable.

Table 7 Comprehensive publication bias assessment

Assessment method	Test statistic	Value	95%CI	P value	Interpretation
Egger’s regression test	Intercept	0.357	-1.582 to 2.296	0.721	No asymmetry detected
	Standard error	0.982	-	-	-
Begg’s test	Kendall’s Tau	-0.022	-	1.000	No rank correlation
Fail-Safe N	Rosenthal N	2554	-	< 0.001	Highly robust (threshold = 60)
Trim-and-fill (L0)	Imputed studies	0	-	-	No missing studies
	Adjusted prevalence	8.66%	7.14-10.63	-	No change from observed
Trim-and-fill (R0)	Imputed studies	0	-	-	No missing studies
	Adjusted prevalence	8.66%	7.14-10.63	-	No change from observed
Forced imputation	Imputed studies	1 (forced)	-	-	Sensitivity analysis
	Adjusted prevalence	8.58%	7.08-10.42	-	Negligible change (-0.08%)

Convergence of multiple methods (Egger’s, Begg’s, Fail-Safe N, trim-and-fill) provides robust evidence against publication bias despite limited test power with n = 10 studies. CI: Confidence interval.

Table 8 Subgroup analysis with formal tests for differences

Subgroup	Studies (n)	Pooled prevalence (%)	95%CI	I² (%)	Q-between	df	P value
By study setting					3.18	1	0.075
Community-based	4	7.8	6.2-9.7	28.5
Clinical/hospital	4	10.9	8.1-14.5	31.2
Healthcare workers	1	6.0	4.1-8.7	N/A
By age group					0.16	1	0.689
Reproductive age (< 50)	3	9.1	6.8-12.1	42.3
Mixed/older (≥ 50)	7	8.4	6.7-10.5	29.8
By geographic region					0.35	1	0.554
Europe/North America	4	8.2	6.9-9.7	18.7
Other (South America, Asia, North Africa)	6	9.1	6.8-12.0	38.9

Q-between tests the null hypothesis that subgroup prevalence estimates are equal. Healthcare worker subgroup was excluded from formal testing due to it being a single study. None of the subgroup differences reached statistical significance (all P > 0.05). CI: Confidence interval; N/A: Not applicable.

Table 9 Study quality assessment summary

Ref.	Sample size	Study design score	Definition clarity	Overall quality
Yuaso et al[7], 2018	865	High	Good	High
Ribeiro et al[5], 2019	227	Moderate	Good	Moderate
Gallas et al[1], 2018	402	Moderate	Good	Moderate
Güngör Uğurlucan et al[3], 2014	2518	High	Good	High
Wu et al[8], 2015	3497	High	Good	High
Matthews et al[12], 2013	64396	High	Good	High
Johannessen et al[2], 2018	976	High	Good	High
	1031	High	Good	High
Luo et al[4], 2020	700	Moderate	Good	Moderate
Slieker-ten Hove et al[13], 2010	1397	Moderate	Good	Moderate

Table 10 Geographic and economic representativeness assessment

WHO region	Countries included	Studies (n)	Sample size (n)	% of total	Coverage status
Americas	Brazil, United States	4	66985	88.1%	Represented
Europe	Norway, Netherlands, Turkey	4	5922	7.8%	Represented
Western Pacific	China	1	700	0.9%	Limited
Eastern Mediterranean	Tunisia	1	402	0.5%	Limited
African Region (Sub-Saharan)	None	0	0	0%	Not represented
South-East Asia	None	0	0	0%	Not represented
Total coverage	7 countries	10	76009	100%	4 of 6 regions (67%)

WHO: World Health Organization.

Table 11 World bank income classification

Income level	Countries	Studies (n)	Sample size (n)	% of total	Coverage status
High-income	United States, Norway, Netherlands	3	66269	87.2%	Represented
Upper-middle-income	Brazil, China, Turkey, Tunisia	7	9740	12.8%	Represented
Lower-middle-income	None	0	0	0%	Not represented
Low-income	None	0	0	0%	Not represented

Table 12 Global standards compliance assessment

Criterion	Recommended standard	Study achievement	Met?
WHO regional coverage	≥ 4 of 6 regions	4 of 6 regions (67%)	Partial
Income diversity	All income levels represented	2 of 4 levels (high and upper-middle only)	Limited
Minimum sample size	≥ 10000 participants	76009 participants	Yes
Geographic diversity	Multiple continents	4 continents (Americas, Europe, Asia, North Africa)	Yes
Sample distribution	No single country >50%	United States = 84.7% of total sample	No
Sub-Saharan Africa inclusion	At least 1 study recommended	0 studies	No

WHO: World Health Organization.