Habitus and Pelvic Floor Symptoms and Support 1 Year Postpartum

Abstract

Objective

To estimate the association between habitus measures and pelvic floor support and symptoms in primiparous women 1 year following term vaginal delivery.

Methods

In this cross-sectional study including women enrolled at 7 academic and community sites, we assessed pelvic floor support, weight, height, waist circumference and percent fat using air displacement plethysmography and participants completed questionnaires, all at one year postpartum. We tested the association of quintiles of habitus measure, including body mass index (BMI), waist circumference, percent body fat, and waist/height ratio with the primary outcomes: anatomic support, dichotomized as maximal vaginal descent <0 cm (better support) versus ≥ 0 cm (worse support) per the Pelvic Organ Prolapse Quantification examination and symptom burden (positive with bothersome symptoms in ≥ 2 of 6 symptom domains), and on 5 secondary outcomes.. The sample size provides 90% power to detect OR ≥ 1.78 between women at mean vs mean + 1 SD of habitus measure.

Results

Of 592 participants, 55 (9.3%) demonstrated worse support and 321 (54.2%) symptom burden. In multivariable analyses, habitus measures were not significantly associated with anatomic support or, except for the highest waist/height quintile, with symptom burden. Compared to women in the first quintile of each habitus measure, those in most higher quintiles demonstrated elevated odds of moderate to severe UI while increased odds for SUI were mainly limited to the highest quintile. After adjusting for percent body fat, the increased odds for BMI on SUI (OR 2.47 (95% CI 1.43, 4.28)) were no longer significant (OR 1.38 (95% CI 0.54, 3.51)).

Conclusion

Habitus in primiparous patients at 1 year postpartum was not associated with anatomic support or symptom burden. Habitus was more associated with moderate to severe UI than mild UI. The association of higher BMI with SUI was attenuated by fitness, reflected by fat percentage.

PRECIS

Habitus in primiparous patients at one year postpartum was not associated with anatomic pelvic floor support or symptom burden.

INTRODUCTION

The number of surgeries for pelvic floor disorders (PFD) is expected to increase by nearly 50 percent by 2050.¹ This estimate, based on the aging of the U.S., does not take into account the parallel rise in obesity.² A large body of literature demonstrates that overweight and obesity increase the risk of urinary incontinence (UI).³ Data for pelvic organ prolapse (POP) are less consistent and more sparse, but suggest the risk increases moderately in largely middle-aged women.⁴ Studies addressing the effect of habitus on UI and POP in younger or nulliparous women show inconsistent directions of effect.^{3, 5–12} There is scant information about the effect of habitus on PFDs other than UI and POP.

Few studies have evaluated measures of habitus other than body mass index (BMI) on PFDs. Those that have suggest that measures of central adiposity, such as waist circumference in older women and waist to height ratio in postpartum women may be more robust predictors of UI than BMI alone.^13–15 A more nuanced understanding of the role that habitus plays on a range of pelvic floor symptoms following vaginal birth^17–19 may advance preventive efforts long before women present for surgical care.

Thus, the aims of this cross-sectional study are to examine the strength of association for each of 4 measures of habitus 1 year postpartum on the prevalence of pelvic floor support and symptoms. We hypothesized that all measures of habitus would increase the odds of worse pelvic floor support and pelvic floor symptoms.

METHODS

Participants include women enrolled in a prospective cohort study, Motherhood And Pelvic health (MAP), who completed follow-up 1-year postpartum. Methods for the overarching study are described elsewhere.²⁰ All participants completed informed written consent and local institutional review boards approved this study.

We enrolled nulliparous women in the third trimester from one of 7 sites in the Salt Lake Valley between 09/01/15 and 07/11/18 and followed those that subsequently delivered vaginally at ≥ 37 weeks gestation for 1 year postpartum. We excluded women who lived > 60 miles from our study site, had no telephone or email access, anticipated moving before one year postpartum, did not speak English or Spanish, were unable to walk without assistance, or had connective tissue disorders predisposing to POP. For the current analysis, we excluded women who did not complete both the study visit and study questionnaires at 1-year follow-up.

We extracted delivery information from the electronic medical record. Participants self-reported ethnicity as Hispanic or non-Hispanic. One year postpartum, participants completed an on-line questionnaire in English or Spanish using RedCap.²¹ Instruments included the Epidemiology of Prolapse and Incontinence Questionnaire (EPIQ), the Incontinence Severity Index (ISI) and the Defecation Distress Inventory.^22–24 The EPIQ contains 14 questions about symptoms in six domains. The ISI includes items about UI frequency and volume, producing a score that categorizes women into no (0), mild (1–2) or moderate to severe (≥ 3) UI. We defined constipation according to the Defecation Distress Inventory if women reported both < 3 bowel movements per week and straining > 25% of the time during bowel movements.

At 1-year in-person study visits, trained and certified coordinators completed the Pelvic Organ Prolapse Quantification (POP-Q) examination with participants in lithotomy at 45 degree back flexion and during straining.²⁵ We measured weight and height, on a calibrated scale and wall stadiometer and waist circumference by determining the mean of two measures at the natural waist. We assessed percent body fat, using air displacement plethysmography (BodPod, COSMED).^{26, 27} Research staff collecting measures of habitus were masked to results from the POP-Q examination and questionnaires.

Primary outcome measures were anatomic support and pelvic floor symptom burden. Defining maximal vaginal descent as the most positive value of Ba, Bp and C during strain, we dichotomized support as maximal vaginal descent above the hymen (< 0 cm; better support) versus at or below the hymen (≥ 0 cm; worse support). We chose this cut-point because it reflects the level at which more women become symptomatic.^{28, 29} We dichotomized pelvic floor symptom burden as endorsement of symptoms associated with bother (> 0 on a 0 to 100 visual analogue scale) in ≥ 2 versus 0 or 1 EPIQ domain. We also specified five secondary outcomes: SUI, OAB and anal incontinence (each positive if at least one symptom with bother was reported in each domain), constipation, and UI severity.

Primary exposures, measures of habitus, included BMI, waist circumference, percent body fat, and waist to height ratio. A priori sample size calculation was completed with a Bonferroni adjusted alpha for 2 primary outcomes of 0.05/2=0.025. The anticipated sample size of 585 would provide 90% power to detect a minimal OR of 1.78 for women with a component of habitus at the mean + 1 SD versus women at the mean, assuming R² of the habitus measure regressed on other variables in the model is 0.5.

For univariate analyses, we used parametric (t-tests, ANOVA and chi-squared or Fishers exact) and non-parametric tests (Wilcoxon rank sum test and Kruskal-Wallis test) as appropriate. For the 4 habitus measures, we determined the functional form of the relationship with each outcome variable using the graphical method of Hosmer and Lemeshow³⁰ and the Akaike information criterion (AIC). These were not linear in most cases. Additionally, for BMI and waist circumference, there were no clear patterns between outcomes and the established clinical cut-points generally used to assess risks for cardiovascular disease. For consistency, we therefore categorized each habitus measure into quintiles and in analyses used quintile 1 as the reference group. We conducted a planned sensitivity analysis categorizing anatomic support as Stage 0 + I vs Stage ≥ II, as this categorization is commonly described in research.

We included adjustment variables in the multivariable logistic regression models based on published evidence and results from previous analyses of this population.^{31, 32} In addition to each habitus measure, we included age (18–25, 26–32, ≥ 33 years), ethnicity (Hispanic vs non-Hispanic), presence of a variable described as high-risk for levator ani muscle injury (second stage of labor duration ≥ 150 minutes, birthweight > 4000 grams, forceps delivery or anal sphincter tear³³), current breastfeeding, current hormonal contraception, and pelvic floor loading (endorsement of heavy work, chronic cough or chronic constipation). Some pelvic floor and habitus measures are highly correlated over time and/or with each other.³⁴ As our aim was to investigate the total effect of habitus on outcomes, all at 1 year postpartum, we chose a priori not to adjust for pre-delivery anatomic support, symptoms, or habitus.

Using SAS version 9.4, we checked models for model assumptions, multicollinearity, effects of sparseness, influential observations, and goodness of fit using standard regression diagnostics.

For associations between habitus measures and outcomes that were significant on multivariable analyses, we performed 2 additional analysis: 1) To estimate whether one habitus measure was most predictive for an outcome, we constructed receiver operating curves for each of the 4 habitus measures on the outcome and compared areas under the curve (AUC) using a contrast matrix to test differences of the areas with the lowest AUC as the reference group,³⁵ and 2) To evaluate whether the effect of BMI on the outcome changed when adjusting for other habitus measures, we planned to repeat the final multivariable model adjusting BMI also for other habitus variables. High (>5.0) Variance Inflation Factors and correlations between BMI and waist circumference (Variance Inflation Factor 7.8, r=0.93) and waist to height ratio (Variance Inflation Factor 7.8, r=0.93) indicated problematic multicollinearity³⁶ and therefore we analyzed only BMI adjusted for percent body fat (Variance Inflation Factor 3.6, r=0.85).

RESULTS

Of 825 women eligible after delivery, 645 completed 1-year foll0w-up (Fig. 1). Mean interval from delivery to follow-up was 384 (SD 35) days. Of these, 53 did not complete both the study visit and questionnaire, leaving 592 (71.8%) in this analytic sample. Mean age at delivery was 29.0 (SD 5.0) years (Table 1). Enrollees who completed the 1-year visit were older (28.4 (SD 5.0) vs 25.7 (SD 5.3) years) and less likely to report Hispanic ethnicity (17.4% vs 30.2%) than enrollees who did not.

Figure 1.

Participant flowchart.

Table 1.

Participant Characteristics

Characteristic	N=592
Age at enrollment, years; mean ± SD	29.0 ± 5.0
Ethnicity*
Hispanic	103 (17.4%)
Non-Hispanic	489 (82.6%)
Education*
High school or less	61 (10.3%)
Some college/completed college	352 (59.5%)
Graduate or professional degree	177 (29.9%)
Presence of high-risk delivery variable	157 (26.5%)
AT 1 YEAR POSTPARTUM
Heavy lifting (other than baby) or heavy work in past 7 days*	257 (43.4%)
Chronic cough*	12 (2.0%)
Constipation*	22 (3.7%)
Work status
Working full-time (≥30 hours per week)	303 (51.2%)
Working part-time (< 30 hours per week)	123 (29.8%)
Other	166 (28.0%)
Currently breastfeeding	269 (45.4%)
Hormonal contraception	327 (55.2%)
HABITUS VARIABLES
Weight (kg); mean ± SD	68.6 ±16.6
Height (cm); mean ± SD	165.7 (6.7)
Body mass index (kg/ m2); mean ± SD†	24.9 (5.7)
Normal/underweight (< 25)	351 (59.3%)
Overweight (25 to < 30)	147 (24.8%)
Obese (≥ 30)	94 (15.9%)
Waist circumference (cm); mean ± SD*	78.5±11.5
≤ 88 cm	488 (82.4%)
>88 cm	103 (17.4%)
Percent fat; mean ± SD*	29.8 ±10.2
Waist to height ratio; mean ± SD*	0.474 ± 0.070‡

^*Ethnicity was collected to ensure that we represented both predominant ethnic groups in our geographic area.

No data are missing unless indicated by “*”. Data were missing for education in 2, heaving lifting in 2, chronic cough in 1, constipation in 4, waist circumference in 1, percent fat in 4 and waist to height ratio in 1.

^†Quintile ranges for body mass index are provided in the legend for Table 3.

^‡Carried to 3 decimals according to standard practice for this variable

One year postpartum, 55 (9.3%) demonstrated worse support (maximal vaginal descent ≥ 0 cm) and 321 (54.20%) met criteria for symptom burden. Mean (SD) POP-Q points were: C (apex) −6.2 (1.1) cm, Ba (anterior vagina) −1.5 (0.8) cm and Bp (posterior vagina) −2.1 (0.7) cm. Two hundred seventy-seven (46.8%) demonstrated stage II POP, the sensitivity analysis outcome. No woman had POP greater than stage II.

On univariate analysis, habitus measures were not associated with either pelvic floor support or symptom burden. All habitus measures were positively associated with SUI and UI severity (Table 2).

Table 2.

Univariate associations between habitus variables and outcomes±

	Body mass index (kg/m2) Mean ± standard deviation	Waist circumference (cm) Mean ± standard deviation	Percent fat Mean ± standard deviation	Waist to height ratio Mean ± standard deviation
Primary outcomes (n)
Support*
Worse (55)	24.1 ± 4.2	76.5 ±8.9	27.7 ± 9.1	0.465 ± 0.056†
Better (537)	25.0 ± 5.9	78.7 ±11.7	30.0 ± 10.3	0.475 ± 0.071
	P=0.12	P=0.10	P= 0.11	P=0.22
Symptom burden
Present (321)	25.1 ± 5.7	78.9 ± 11.2	30.1 ±10.1	0.477 ± 0.069
Absent (271)	24.8 ± 5.8	78.0 ± 11.8	29.5 ±10.3	0.470 ± 0.071
	P=0.52	P=0.34	P=0.49	P=0.21
Secondary outcomes (n)
Stress urinary incontinence
Present (319)	25.7 ± 6.1	80.1 ± 12.1	31.1 ±10.2	0.484 ± 0.073
Absent(273)	24.0 ± 5.1	76.6 ± 10.4	28.4 ±10.0	0.462 ± 0.063
	P=0.002	P=0.001	P=0.001	P<0.001
Overactive bladder
Present (280)	25.2 ± 5.9	79.0 ± 11.3	30.1 ±10.1	0.477 ± 0.069
Absent (312)	24.7 ± 5.6	78.0 ± 11.6	29.6 ±103	0.471 ± 0.070
	P=0.24	P=0.30	P=0.54	P=0.28
Constipation**
Present (22)	25.6 ± 6.2	80.2 ± 13.2	31.9 ±10.5	0.491 ± 0.080
Absent (566)	24.9 ± 5.7	78.4 ± 11.4	29.7 ±10.2	0.473 ± 0.069
	P=0.56	P=0.47	P=0.35	P=0.23
Anal incontinence
Present (142)	24.6 ± 5.3	78.2 ± 10.9	29.3 ±10.2	0.472 ± 0.067
Absent (450)	25.0 ± 5.8	78.6 ± 11.7	30.0 ±10.2	0.475 ± 0.071
	P=0.40	P=0.75	P=0.48	P=0.69
Urinary Incontinence severity‡
None (253)	24.2 ± 5.3	77.0 ± 10.8	28.9 ±10.1	0.464 ± 0.064
Mild (248)	24.7 ± 5.7	78.3 ± 11.7	29.5 ±10.4	0.473 ± 0.071
Moderate/severe (86)	27.6 ± 6.4	83.6 ± 11.6	33.6 ± 9.5	0.506 ± 0.074
	P<0.001	P<0.001	P<0.001	P<0.001

^*Worse support: Maximal vaginal descent ≥ 0 cm (at or below the hymen); Better support: maximal vaginal descent < 0 cm (above the hymen)

^†Carried to 3 decimals according to standard practice for this variable.

^‡Missing data: Overactive bladder, n=4; Urinary Incontinence Severity, n=5

The adjusted odds of each habitus measure on the primary outcomes, secondary outcomes, and sensitivity analyses are summarized in Table 3. Constipation was not further explored in multivariable models because of the small prevalence. There were no statistically significant associations between habitus measures and the primary anatomic support outcome (maximal vaginal descent ≥ 0 cm).

Table 3.

Multivariable analyses (Odds Ratio (95% Confidence Interval)) for each outcome according to each habitus measure^*

Outcome	Measure	Q1**	Q2	Q3	Q4	Q5
Primary outcomes
MVD† (≥ 0 cm vs <0 cm)	BMI**	Ref	1.60 (0.66, 3.90)	1.23 (0.49, 3.12)	1.38 (0.55, 3.49)	0.65 (0.22, 1.93)
	WC**	Ref	0.60 (0.24, 1.48)	0.85 (0.37, 1.98)	0.78 (0.33, 1.82)	0.55 (0.21, 1.42)
	% body fat	Ref	0.92 (0.39, 2.16)	0.89 (0.38, 2.10)	0.74 (0.30, 1.83)	0.63 (0.24, 1.64)
	Waist/height	Ref	0.90 (0.38, 2.14)	0.94 (0.39, 2.25)	0.73 (0.29, 1.82)	0.74 (0.29, 1.90)
Symptom Burden (Yes vs No)	BMI	Ref	0.86 (0.51, 1.46)	1.55 (0.91, 2.65)	0.90 (0.53, 1.53)	1.28 (0.75, 2.20)
	WC	Ref	1.11 (0.66, 1.88)	1.03 (0.61, 1.73)	1.18 (0.69, 2.00)	1.60 (0.93, 2.75)
	% body fat	Ref	0.65 (0.38, 1.10)	0.95 (0.56, 1.62)	1.22 (0.71, 2.09)	1.16 (0.67, 2.01)
	Waist/height	Ref	1.27 (0.75, 2.16)	0.96 (0.57, 1.63)	1.13 (0.66, 1.92)	1.84 (1.06, 3.19)
POP-Q† stage (II vs 0, I)‡	BMI	Ref	1.65 (0.97, 2.79)	1.30 (0.77, 2.21)	1.63 (0.95, 2.78)	1.15 (0.67, 1.97)
	WC	Ref	1.25 (0.74, 2.11)	1.40 (0.83, 2.37)	1.50 (0.88, 2.54)	1.17 (0.68, 2.00)
	% body fat	Ref	1.55 (0.91, 2.62)	1.60 (0.94, 2.73)	1.62 (0.95, 2.77)	1.09 (0.63, 1.89)
	Waist/height	Ref	1.82 (1.07, 3.09)	1.67 (0.99, 2.84)	1.86 (1.09, 3.18)	1.25 (0.72, 2.16)
Secondary outcomes
SUI (Yes vs No)	BMI	Ref	0.87 (0.52, 1.47)	1.99 (1.17, 3.38)	1.48 (0.87, 2.52)	2.47 (1.43, 4.28)
	WC	Ref	1.06 (0.63, 1.78)	1.40 (0.83, 2.35)	1.65 (0.97, 2.80)	3.00 (1.72, 5.23)
	% body fat	Ref	0.95 (0.56, 1.60)	1.53 (0.90, 2.59)	1.58 (0.93, 2.67)	2.50 (1.44,4.35)
	Waist/height	Ref	1.22 (0.72, 2.05)	1.40 (0.83, 2.35)	1.31 (0.77, 2.22)	3.80 (2.14, 6.76)
Overactive bladder (Yes vs No)	BMI	Ref	0.67 (0.39, 1.14)	1.91 (1.13, 3.25)	1.01 (0.59, 1.72)	1.18 (0.69, 2.02)
	WC	Ref	0.92 (0.54, 1.55)	1.26 (0.74, 2.12)	1.38 (0.82, 2.34)	1.26 (0.74, 2.15)
	% body fat	Ref	0.80 (0.47, 1.34)	1.10 (0.65, 1.86)	1.24 (0.73, 2.11)	1.05 (0.61, 1.79)
	Waist/height	Ref	1.26 (0.74, 2.13)	1.10 (0.65, 1.85)	1.43 (0.84, 2.43)	1.42 (0.83, 2.43)
Anal incontinence (Yes vs No)	BMI	Ref	1.20 (0.67, 2.17)	0.84 (0.45, 1.56)	0.77 (0.41, 1.46)	1.13 (0.61, 2.10)
	WC	Ref	1.76 (0.97, 3.18)	0.71 (0.37, 1.39)	0.93 (0.49, 1.77)	1.45 (0.78, 2.70)
	% body fat	Ref	0.89 (0.49, 1.64)	1.03 (0.57, 1.89)	0.90 (0.48, 1.67)	1.11 (0.60, 2.07)
	Waist/height	Ref	1.78 (0.98, 3.25)	0.80 (0.41, 1.55)	1.16 (0.61, 2.20)	1.54 (0.82, 2.91)
ISI 1,2 vs 0	BMI	Ref	0.97 (0.56, 1.67)	1.30 (0.73, 2.30)	1.32 (0.75, 2.33)	1.45 (0.80, 2.62)
	WC	Ref	0.98 (0.57, 1.69)	1.10 (0.63, 1.93)	1.17 (0.66, 2.06)	1.66 (0.91, 3.02)
	% body fat	Ref	0.90 (0.52, 1.57)	1.52 (0.86, 2.69)	1.21 (0.68, 2.14)	1.47 (0.82, 2.66)
	Waist/height	Ref	1.21 (0.70, 2.12)	1.18 (0.68, 2.06)	1.07 (0.61, 1.89)	2.18 (1.18, 4.03)
ISI ≥ 3 vs 0	BMI	Ref	1.24 (0.40, 3.80)	3.96 (1.44, 10.9)	2.62 (0.92, 7.46)	6.68 (2.47, 18.1)
	WC	Ref	1.96 (0.67, 5.72)	2.83 (1.01, 7.94)	3.48 (1.24, 9.73)	7.33 (2.70, 19.9)
	% body fat	Ref	1.83 (0.67, 4.97)	2.93 (1.08, 7.92)	3.19 (1.21, 8.40)	4.88 (1.84, 12.9)
	Waist/height	Ref	3.07 (1.08, 8.71)	2.93 (1.03, 8.33)	2.97 (1.06, 8.31)	10.1 (3.66, 28.0)

^*Adjusted for age (18–25, 26–32, ≥ 33 years), ethnicity (Hispanic vs non-Hispanic), high-risk delivery variable, current breastfeeding, current hormonal contraception, pelvic floor loading variable. Significant results are bolded.

^†Q= Quintile; Ref = Reference group; MVD= maximal vaginal descent; POP-Q= Pelvic Organ Prolapse Quantification examination; WC=Waist circumference; BMI= Body mass index; SUI = Stress urinary incontinence; ISI= Incontinence Severity Index (0= none; 1–2 = mild; ≥ 3 = moderate to severe)

^‡Sensitivity analysis

Ranges for quintiles:

Body mass index (kg/m2): Q1 15.9–20.3; Q2 20.3–22.3; Q3 22.3–25.1; Q4 25.1–28.6; Q5 28.6–58.0

Waist circumference (cm): Q1 57.8–68.8; Q2 68.8–73.6; Q3 73.6–79.2; Q4 79.2–86.5; Q5 86.5–127.0

Percent fat : Q1 3.5–20.5; Q2 20.5–26.9; Q3 26.9–32.8; Q4 32.8–38.4; Q5 38.4–59.2

Waist to height ratio: Q1 0.35–0.41; Q2 0.41–0.44; Q3 0.44–0.48; Q4 0.48–0.52; Q5 0.52–0.78

Only waist to height ratio for women in the highest quintile increased the odds of the symptom burden outcome (OR 1.84 (95% CI 1.06,3.19)). BMI increased the odds of OAB only for women in the third BMI quintile (OR 1.91 (95% CI 1.13, 3.25)). No habitus measure was associated with anal incontinence.

Associations between habitus and UI varied according to UI severity (Figure 2). The odds of mild UI were increased relative to no UI only in women in the highest quintile of waist to height ratio. Habitus had a much greater association with moderate to severe UI with increased ORs for most quintiles compared to the first and with particularly higher odds for women in the highest quintile of each habitus measure. Women in the highest quintile for all 4 habitus measures had higher odds of SUI.

Figure 2.

Multivariable analyses of associations between habitus measures and urinary incontinence severity 1 year postpartum. The y-axis demonstrates the OR with 95% CIs for each quintile of each habitus measure listed on the x-axis. The first quintile is the reference group. Body mass index measured in kg/m², waist circumference measured in cm. Incontinence Severity Index scores: 0 (reference category)=none; 1‒2=mild; ≥3=moderate to severe.

In additional multivariable models including BMI and percent body fat, as well as the other adjustment variables, we observed a similar direction and size of effect for associations between BMI and both mild and moderate to severe UI. However, after adjusting for percent body fat, BMI was no longer associated with SUI, with OR of 0.72 (95% CI 0.41, 1.27), 1.53 (95% CI 0.82, 2.85), 1.00 (95% CI 0.48, 2.09), 1.38 (95% CI 0.54, 3.51) for quintiles 2, 3, 4 and 5, respectively.

We assessed predictive values of each habitus measure on SUI and moderate to severe UI. For SUI, AUCs ranged from 0.62 (95% CI 0.57, 0.66) for percent fat to 0.64 (95% CI 0.60, 0.69) for waist to height ratio and for moderate to severe UI, from 0.71 (95% CI 0.64, 0.77) for percent fat to 0.75 (95% CI 0.69, 0.81) for waist to height ratio. Differences between AUCs were not statistically significant.

The finding that women in the highest BMI quintile demonstrated lower, albeit non-significant, adjusted odds of worse support was surprising. Therefore, we conducted post-hoc analyses to determine whether women in the highest BMI quintile differed in potential risk factors for worse support. There were no differences in rates of forceps delivery, severe anal sphincter laceration or high-risk delivery factor according to BMI quintiles (data not shown).

DISCUSSION

In this population of primiparous patients 1 year following vaginal birth, habitus measures demonstrated no or little association with anatomic support, symptom burden, anal incontinence or OAB.

Habitus was most associated with moderate to severe UI.. In contrast, only the highest quintile for waist to height ratio was associated with mild UI. Women in the highest quintile of all habitus measures had increased odds of SUI but not OAB. After adjusting for adiposity, BMI no longer increased the odds of SUI. None of the four habitus measures tested proved more predictive for either moderate to severe UI or SUI.

There is a large body of literature that, like our results, consistently shows a positive association of BMI with UI, broadly defined.³ Our study joins a small number that differentiates mild from more severe UI^{37, 38} and an important distinction emerges. Habitus has little to no association with mild UI but a stronger and nearly linear association with moderate to severe UI. Yet, despite having 3-fold greater odds of moderate to severe UI, women in BMI quintile 3 (ranging from 22.3–25.1 kg/m2) are categorized as normal weight (≤24.9) according to WHO categories. As studies typically use normal weight as the reference group, the trend we observed is usually masked.

Our finding that habitus plays little role on OAB fits with the inconsistent body of literature in this area, in which the effect of habitus is relatively small except at extremes. ^39,40 Others also found a greater effect of BMI on SUI than UUI.^{41, 42} However, in a population of older women, OAB increased linearly with waist circumference and BMI.⁴³ This may differ from our results because older obese women likely have longer exposure to elevated BMI, and symptom severity appears to worsen with duration of elevated BMI.⁴⁴ In a recent study, visceral fat surface area assessed with spiral computed tomography, but not BMI or waist circumference, increased odds of OAB.⁴⁵ Precise assessment of visceral adiposity was beyond our scope, though may be important in predicting OAB.

In some studies of older women, waist circumference was a better predictor of UI than BMI.^{14, 46} In our population, we found no differences in the predictive capabilities of UI models containing each of the 4 habitus measures but instead found that adiposity confounded the association between BMI and SUI. Obesity predisposes women to lower urinary tract symptoms through different mechanisms.⁴⁷ A common line of thinking is that higher BMI results in chronically elevated intra-abdominal pressure, which predisposes to SUI by increasing the intravesical pressure, and also possibly placing excessive strain on the pudendal nerve and its branches.⁴⁸ In addition, visceral adipose tissue may induce symptoms through metabolic factors related to systemic inflammation and oxidative stress. Our finding that adiposity attenuated the association between BMI and SUI questions the paradigm that elevated intra-abdominal pressure is the principal driver of SUI.

In a recent systematic review, compared to normal weight women, those overweight and obese had 36–47% higher meta-analysis risk ratios for POP.⁴ The single study in this review that included primiparous patients in the first postpartum year found no statistically significant association between overweight/obesity and ≥ Stage II POP ¹¹; in contrast, another study of a similar population reported a 41% increase in ≥ Stage II POP for each 1-unit increase in BMI.⁴⁹ The direction of effect in our study mirrored that of these studies when we used Stage II but not the hymen as a cut-point. This suggests that the relationships between habitus and POP depend on the POP phenotype under study.

We identified one other study of primiparous patients at 1 year postpartum that assessed the association between habitus and anal incontinence. In contrast to our results, increasing BMI was associated with slightly increased odds of anal incontinence. ⁵⁰

We were surprised to find that even small increases in habitus considered still within the range of normal were associated with increased odds of moderate to severe UI in this population. In contrast, the odds of SUI were increased primarily in women in the highest quintile of BMI, and furthermore, this deleterious association was attenuated after adjusting for adiposity. In addition to weight loss, shown to benefit women with SUI in randomized trials,⁵¹ our results suggest that decreasing adiposity without weight loss through muscle building interventions may provide an alternative approach to treating SUI. Further research is needed to understand why women in the third BMI quintile, who still meet criteria for normal weight have 3-fold higher odds of moderate to severe UI compared to normal- or under-weight women in the first BMI quintile, and, in particular, whether this might be mediated by the volume and distribution of adipose tissue.

Strengths of our study include a 1-year follow-up rate that was similar to or higher than most other published studies that followed women between late pregnancy and 1 year postpartum^{11, 17, 33, 52–54} We assessed several objective measures of habitus, rather than just BMI, and categorized each measure using quintiles based on preliminary inspection of relationships between each measure and each outcome. We used validated outcome measures. However, we used symptoms from the EPIQ as outcome measures to investigate experiences of young, postpartum women rather than the thresholds validated in middle-aged and older women for predicting women at high risk of diagnostic urogynecologic conditions. Additionally, we used a non-validated measure of symptom burden, as we were not aware of an existing measure appropriate for our population. We used a common anatomic outcome which represented a composite of POP-Q points. However, for most women, the anterior vaginal wall was the most distal POP-Q point (and thus defined maximal vaginal descent); the number of women with worse apical or posterior descent was too small to study whether habitus measures impact specific vaginal compartments differently.

Our finding that women in the highest quintile for each habitus measure had lower, though not statistically significant, odds of worse support was unexpected. Study coordinators underwent rigorous training and certification procedures in the POP-Q examination before and during the study period. However, considering that this examination is difficult to execute well in women with Class 3 obesity (BMI > 40 kg/m²), we repeated the functional form analysis between habitus and support removing participants with Class 3 obesity and found similar results.

Despite our large study population, we were not able to explore constipation in multivariable models, as the prevalence proved too small. The estimates of associations between habitus and moderate to severe UI had a consistent direction, though wide confidence intervals, given the smaller sample size for this outcome. Given the large number of associations tested, and the smaller number of cases in some subgroups, some associations detected may be spurious.

Prevalence rates vary according to how symptoms or support are defined, the population under study, and potentially the duration of obesity. Our results cannot, therefore, be extrapolated to women delivered by cesarean, multiparas or middle-aged or older women. Further, women that completed 1 year follow-up and make up our study sample are older and less likely to be Hispanic than enrolled women that did not complete follow-up, raising the possibility of participation bias.

That increased habitus is associated with moderate to severe UI is important because this is more likely to persist than mild UI.^{55, 56} However, while mild UI has a higher remission rate and a low rate of progression to moderate to severe UI, obesity has been shown to double the risk of progressing from mild to severe UI over time. Thus, weight loss interventions in young, overweight and obese women may have a substantial effect on bothersome UI throughout life.