Symptomatic Improvement After Mesh Removal: A Prospective Longitudinal Study of Women with Urogynecologic Mesh Complications

Abstract

Background:

Limited data exists on symptom improvement following the excision of urogynecologic mesh for a complication.

Objective:

To compare clinical characteristics and outcomes in patients undergoing complete excision of polypropylene urogynecologic mesh for pain, mesh exposure, or both.

Design:

Prospective, longitudinal cohort

Setting:

Academic tertiary referral center

Population:

Women undergoing complete vaginal mesh excision for the complication of mesh exposure and/or pain.

Methods:

Clinical assessments and patient reported outcomes (PRO) assessing pain (VAS), bother (PFDI) and functional impact (PFIQ) were collected at baseline, 6, 12, and 24 months after complete mesh excision. Baseline and postoperative PROs were compared by mesh type [sling, prolapse (transvaginal or sacrocolpopexy mesh) or both] and complication (pain, exposure, both).

Measure:

Much better or very much better on PGI-I up to two years after removal.

Results:

Of 173 women, 48 underwent removal for pain, 27 exposure, and 98 exposure plus pain. 75% reported “moderate to severe” baseline symptoms; the most prevalent and severe symptom was dyspareunia. Patients with pain alone were most bothered (PFDI median 234.2 (IQR: 83), p=.02) and had the highest functional impact (PFIQ median 181 (IQR: 138), p<.001). After excision, only 33.3% of women with pain alone reported “improved” symptoms (PGI-I), vs 73.9% with exposure, 58.3% with exposure plus pain (p=.03) with no differences in PGI-I by mesh type. While VAS scores decreased in all groups, PFDI and PFIQ did not improve in pain patients.

Conclusions:

In women experiencing a pain complication after urogynecologic mesh insertion, mesh removal often does not improve symptoms.

Introduction:

In the United States, the cumulative risk of surgical correction of stress urinary incontinence (SUI) or pelvic organ prolapse (POP) by age 80 is 13.6% and 12.6% respectively.¹ Synthetic mesh is widely used in these surgeries; however, there is ongoing concern regarding the risks of significant mesh complications. Pain at the site of implantation and exposure of mesh through the vaginal epithelium are the most commonly reported complications and a source of litigation world-wide.^2,3

Medical management of mesh complications includes analgesics or vaginal estrogen. Surgical management involves mesh trimming, releasing tension bands, or full excision.⁴ To treat a complication, current literature suggests that ~50% of women undergo mesh excision as the primary intervention and 79% of patients eventually undergo at least one mesh excision surgery.^4–6 However, reports indicate wide variability in post-operative outcomes with only 50–71% of women reporting improved symptoms.^5,7,8 Little is known about which symptoms are optimized by surgery or which patients benefit from mesh removal.

The objective of this study was to examine patient characteristics and prospective post-operative outcomes in women undergoing complete vaginal mesh excision for the two most common mesh-associated complications: exposure through the vaginal epithelium and pain. Specifically, we assessed whether symptoms resolved post-operatively using patient reported outcomes up to two years after mesh removal. We hypothesized that women presenting with pain would report worse symptom burden pre-operatively and less symptom improvement compared to patients with exposure.

Materials and Methods:

Study Design and Patient Population

This was a single center prospective, longitudinal cohort study of women undergoing surgical mesh removal between February 2012 and September 2019 (IRB# PRO10090194). Women in whom mesh was removed for exposure and/or pain were included in the current study. Prior to surgery, consented patients completed questionnaires assessing baseline demographics (age, race, body mass index, gravidity, parity, number of vaginal births, menopausal status, medications, and social, medical and surgical histories). Physical exam and surgical indication were recorded by the surgeon. Baseline questionnaires included the Pelvic Floor Distress Inventory (PFDI-20), Pelvic Floor Impact Questionnaire (PFIQ-7), Patient Global Impression of Severity (PGI-S), and visual analogue scales (VAS) assessing pelvic pain, dysuria and dyspareunia.⁹ These measures, as well as the Patient Global Impression of Improvement (PGI-I) were repeated at 6, 12, and 24 months (± 2, 6, and 6 months, respectively) after surgical mesh removal.^10,11 All questionnaires were optional and were completed in person or online with a secure, personalized link (REDCap 9.7.1©). Baseline and follow-up physical examinations were performed by trained clinicians including the pelvic organ prolapse quantification (POP-Q) exam and incontinence evaluation.

For all women included in the study, the vaginal portion of the mesh was removed as close to its entireity as possible.¹² Subjects were categorized by type of mesh removed [sling, prolapse (transvaginal or sacrocolpopexy mesh), both] and the indication for removal (pain, exposure, or both). The approach for insertion of the sling (retropubic or transobturator) and the prolapse mesh (transvaginal vs sacrocolpopexy) was also recorded. Patients were excluded if they cancelled surgery, had a mesh exposure into an adjacent organ, incomplete bladder emptying, or ineffective sling as removal indication, or did not complete at least one of the post-op follow-up assessments. While our practice is to perform a “complete vaginal excision”, some patients underwent more than one excision. For these women, data from the second removal was used (Figure S1).

Questionnaires

The PFDI-20 was used to assesses symptom bother with three subscales: Urinary Distress Inventory (UDI), Pelvic Organ Prolapse Distress Inventory (POPDI) and Colorectal-Anal Distress Inventory (CRADI).⁹ The PFIQ-7 evaluates functional impact with three subscales: Urinary Impact Questionnaire (UIQ), Pelvic Organ Prolapse Impact Questionnaire (POPIQ), and Colorectal-Anal Impact Questionnaire (CRAIQ). Both PFDI and PFIQ subscales are scored 0–100 with maximum global scores of 300. Higher scores indicate worse health status or health related quality of life (HRQOL).⁹ Individual VAS were used to measure the average severity of pelvic pain, dysuria, and dyspareunia within the past month (range 0 to 100). The PGI-S was used to quantify the patient’s global perception of the severity of her complication on a 4-point scale from 1 (normal) to 4 (severe). The PGI-I assessed the patient’s perception of her symptom improvement as compared to before surgery scored on a 7-point scale from 1 (very much better) to 7 (very much worse).¹¹

Outcomes

The primary outcome was PGI-I up to two years after surgical removal of mesh. Women were categorized as ‘improved’ if they answered much better or very much better on the PGI-I as has been reported in previous large studies^13,14 and not improved if they reported a little better, no change, a little worse, much worse or very much worse. The secondary outcome was the proportion of women with moderate to severe symptoms (PGI-S) at baseline and up to two years after mesh removal. Exploratory analysis was performed to assess change in symptom bother (PFDI-20), functional impact (PFIQ), pain domains (VAS) and new onset or recurrent POP (≥ stage 2) or SUI after mesh removal. For POP, we used the answer “yes” to the PFDI question, “do you see or feel a bulge beyond the hymen?” or the presence of ≥ stage II prolapse on exam. For SUI, we used “yes” to UDI-6 question, “do you leak with physical activity” or demonstrable SUI on exam. Exploratory analyses were also performed to examine differences in baseline and postoperative outcomes by prolapse mesh type (sacrocolpopexy versus transvaginal mesh) and sling type (retropubic versus transobturator).

Statistical Analysis and Sample Size

Data are presented as means (standard deviation) and medians (interquartile range) for parametric and non-parametric continuous variables respectively, and frequency (percentage) for categorical variables. Patients were grouped according to the type of mesh that was removed [sling, prolapse (transvaginal or sacrocolpexy mesh), or both) and their primary complication (pain, exposure, or both). Baseline characteristics were compared using one-way analysis of variance (ANOVA), Fisher’s exact and Kruskal-Wallis tests.

Baseline and post-operative patient-reported outcomes were compared using the minimal clinically important difference (MCID) when available.^15–17 For the PFDI and PFIQ global scores, MCIDs were 45 and 36, respectively.⁹ Subscale scores for each were analyzed using MCID 16 (POPDI), 11 (UDI), 29 (POPIQ), and 16 (UIQ).^17,18 Changes in VAS between baseline and follow-up were evaluated with the MCID 13 for pelvic pain, dysuria and dyspareunia.¹⁹ Post-operative endpoints were compared using Fisher’s exact and Kruskal-Wallis tests. As questionnaires were optional, outcomes were assessed based on last completed questionnaire.

This was a convenience sample of all women undergoing surgical mesh removal for pain and/or exposure complications between February 2012 and September 2019. In a previous study, 86% of women who had complications from tension-free vaginal tape procedures for stress urinary incontinence answered much better or very much better on the PGI-I up to 12 months post-operatively.²⁰ In the current study, we assumed a 20% difference between the 3 complication groups (86%, 66%, 46%). When the sample size in each group was 27, a 0.05 level Chi-square test had 80% power to detect a difference in proportions assuming a variance of 0.027 and an average proportion of 0.66. Multivariable logistic regression was used to evaluate whether baseline characteristics confounded the association between measures of postoperative improvement and indication for mesh removal or type of mesh removed. Statistical analysis was performed at the 0.05 significance level with STATA software (version 15.0; StataCorp, College Station, Texas).

Funding

This study was funded by the NIH (NICHD 5R01HD083383-03).

NIH/ORWH Building Interdisciplinary Research Careers in Women’s Health (BIRCWH) NIH K12HD043441

Results

Patient Demographics

Of the 191 patients enrolled during the study period, 173 were included in the analysis (Figure S1). Patients were excluded after consent due to canceled surgery (n=1), mesh excised for exposure into an organ other than the vagina (n=2); mesh excised for retention or ineffective sling (n=5); or no follow up (n=4). Six patients required multiple excisions and only data from last excision was included. Follow-up questionnaires were completed by 81 patients (46.8%) at 6-months, 67 (38.7%) at 12-months and 50 (28.9%) at 24 months. In total, 108 patients (62%) completed at least one follow-up survey. Patients were primarily white (97.1%), overweight (BMI 29.1 kg/m²), multiparous (parity 2), and postmenopausal (75.1%) with a mean age of 55.6 ± 6.9 years (Table 1). Patients with prolapse mesh and those with exposure were older than those with a sling (p=.001)). Over half smoked and used either vaginal or oral hormone therapy (HT). 62.8% of the patients had a history of hysterectomy, 14.5% had undergone a prior mesh removal, 12.7% had diabetes, and 2.4% used chronic steroids. 28.3% took opioid analgesics at baseline, while 10.4% took non-opioid analgesics. Median time between initial mesh insertion and symptom onset was 26 weeks; however, median time between mesh insertion and removal was 6.7 years.

Table 1.

Baseline characteristics of women undergoing mesh removal

Characteristic N=173	Total N=173	Slinga N=111	Prolapseb N=54	Sling & Prolapse N=8	P value	Pain N=48	Exposure N=27	Exposure + Pain N=98	P value
Age, mean (SD), years	55.6 (11.5)	52.5 (11.5)	61.5 (9.7)	59 (7.5)	.001	52 (11.8)	62.6 (12.2)	55.4 (10.5)	.001
BMI, median (IQR), kg/m2	29.1 (6.9)	30 (6.7)	27.2 (6.7)	29.5 (6.6)	.03	29.7 (5.9)	28.1 (8.3)	28.9 (6.8)	.80
Percent white (%)	97.1	97.3	96.2	100	.73	100	96.3	95.9	.81
Gravidity, median (IQR)	3 (2)	3 (2)	3 (2)	2.5 (1.5)	.34	3 (2)	3 (2)	3 (2)	.67
Parity, median (IQR)	2 (1)	2 (1)	3 (1)	2 (1.5)	.26	2 (1)	2 (1)	2 (1)	.88
Vaginal births, median (IQR)	2 (1)	2 (1)	3 (1)	2 (2)	.17	2 (1)	2 (2)	2 (1)	.67
Postmenopausal, N (%)	130 (75.1)	73 (65.7)	51 (94.4)	6 (75)	<.001	35 (72.9)	22 (81.5)	73 (74.5)	.73
Vaginal atrophy, N (%)	64 (38.1)	38 (35.2)	23 (43.4)	3 (42.9)	.60	15 (33.3)	11 (40.7)	38 (39.6)	.76
On hormone therapy, N (%)	88 (50.9)	54 (48.7)	36 (66.7)	6 (75)	.05	25 (52.1)	19 (70.4)	52 (53.1)	.26
Hormone therapy type, N (%)					.03
Oral	8 (4.6)	5 (4.5)	1 (1.9)	2 (25)		4 (8.3)	0	4 (4.1)	.14
Vaginal	69 (39.8)	40 (36)	27 (50)	2 (25)		17 (35.4)	15 (55.6)	37 (37.8)
Both	8 (4.62)	3 (2.7)	4 (7.4)	1 (12.5)		1 (2.1)	3 (11.1)	4 (4.1)
None	50 (28.9)	63 (56.8)	22 (40.7)	3 (37.5)		26 (54.2)	9 (33.3)	53 (54.1)
Prior hysterectomy, N (%)	118 (68.2)	66 (59.5)	45 (83.3)	7 (87.5)	.004	29 (60.4)	17 (63)	72 (73.5)	.20
Prior mesh removal, N (%)	25 (14.5)	16 (14.4)	6 (11.1)	3 (37.5)	.15	7 (14.6)	5 (18.5)	13 (13.3)	.72
Smoking history, N (%)	90 (52)	66 (59.5)	20 (37)	4 (50)	.02	23 (47.9)	12 (44.4)	55 (56.1)	.72
Sling type, N (%)					1.00				.58
TOT	29 (53.7)	28 (53.9)	0	1 (50)		9 (47.4)	4 (66.7)	16 (55.2)
TVT	21 (38.9)	20 (38.5)	0	1 (50)		9 (47.4)	2 (33.3)	10 (34.5)
TVT & TOT	1 (1.9)	1 (1.9)	0	0		1 (5.3)	0	0
Other	3 (5.6)	3 (5.8)	0	0		0	0	3 (10.3)
Diabetes, N (%)	22 (12.7)	16 (14.4)	6 (11.1)	0	.66	7 (14.6)	3 (11.1)	12 (12.2)	.95
Chronic Steroids, N (%)	4 (2.4)	3 (2.8)	1 (1.9)	0	1.00	1 (2.1)	1 (3.7)	2 (2.1)	.80
Opioid use, N (%)	49 (28.3)	31 (27.9)	16 (29.6)	2 (25)	.96	19 (39.6)	4 (14.8)	26 (26.5)	.07
Non-opioid analgesic, N (%)	18 (10.4)	13 (11.7)	5 (9.3)	0	.75	5 (10.4)	2 (7.4)	11 (11.2)	.94
Mesh Complication, N (%)					.03
Pain	48 (27.8)	39 (35.1)	7 (13)	2 (25)	.007	NA	NA	NA	NA
Exposure	27 (15.6)	14 (12.6)	12 (22.2)	1 (12.5)	.26	NA	NA	NA	NA
Exposure & Pain	98 (56.7)	58 (52.3)	35 (64.8)	5 (62.5)	.291	NA	NA	NA	NA
Symptom onset after mesh insertion, median (IQR), weeks	26 (154)	20 (190)	51 (123)	26 (51)	.63	39 (155)	104 (308)	18 (100)	.19
Time between onset and mesh removal, median (IQR), weeks	348.8 (431.8)	321.4 (3012)	391.1 (241)	448.4 (5566.6)	.22	347.1 (446)	432.1 (5284.7)	317.3 (362.9)	.55

BMI, body mass index; IQR, interquartile range

Data reported as N (%) or median (interquartile range) except where specified.

^aSling mesh includes bladder sling, midurethral sling, retropubic sling, pubovaginal sling, tension free vaginal tape, transvaginal tape or transobtorator tape.

^bProlapse mesh includes sacrocolpopexy, sacrocervicopexy, hysteropexy, rectopexy with mesh, transvaginal mesh, or rectocele, cystocele or enterocele repair with mesh.

The most common indication for mesh removal was exposure plus pain (56.7%), followed by pain (27.8%), and exposure (15.6%). The most common type of mesh removed was a sling (64.2%), followed by POP mesh (31.2%) and sling plus POP mesh (4.6%). Patients with slings had a higher BMI (p=.03) and more smoked compared to patients with POP mesh removal (p=.02). More patients with POP mesh complications had prior hysterectomy (p=.004), were postmenopausal (p<.001), and used vaginal estrogen (p=.03). 46.7% of patients with slings had demonstrable stress incontinence prior to mesh removal. 78% of patients with prolapse mesh excision had baseline POPQs and and half of these patients had recurrent prolapse (≥ Stage 2).

Baseline Symptoms

The most common pain trigger prior to mesh removal was dyspareunia (50.3%), followed by spontaneous pain (31.2%), pain with activity (17.9%), and pain during pelvic exams (9.3%) (Table 2). Pelvic pain and dyspareunia scores were more severe in pain only patients (p<.001).

Table 2.

Baseline symptoms of women undergoing mesh removal

Measure	Slinga N=111	Prolapseb N=54	Sling & Prolapse N=8	P value	Pain N=48	Exposure N=27	Exposure + Pain N=98	P value
Pain triggers, N (%)
Pelvic exams	10 (9.3)	5 (9.3)	1 (12.5)	.81	5 (10.4)	0	11 (11.2)	.19
Intercourse	61 (55)	22 (40.7)	4 (50)	.21	30 (62.5)	0	57 (58.2)	<.001
Activity	23 (20.7)	5 (9.3)	3 (37.5)	.05	12 (25)	0	19 (19.4)	.009
Spontaneous	38 (34.2)	12 (22.2)	4 (50)	.14	17 (35.4)	0	37 (37.8)
VAS Score, median (IQR)
Dyspareunia	76 (74)	45 (74)	82 (50)	.20	96 (44)	1 (12)	73 (58)	<.001
Pelvic Pain	49 (52)	44 (44)	16 (50)	.32	63 (38)	7.5 (26.5)	47 (44)	<.001
Dysuria	20 (55)	5.5 (50)	29.5 (77)	.70	43 (80)	2.5 (5)	15 (47)	.002
PGI-Severity, N (%)				.09				.11
Normal-Mild	14 (18.7)	15 (36.6)	1 (16.7)		5 (15.2)	8 (42.1)	17 (24.3)
Moderate-Severe	61 (81.3)	26 (63.4)	5 (83.3)		28 (84.9)	11 (57.9)	53 (75.7)
PFDI-20, median (IQR)	213.3 (85.0)	212.5 (99.2)	200.0 (75.0)	.60	234.2 (82.9)	181.3 (125.0)	205.6 (78.3)	.02
PFIQ-7, median (IQR)	126.2 (145.2)	90.5 (123.8)	76.2 (95.2)	.11	181.0 (138.1)	48.8 (73.8)	109.5 (152.4)	<.001
Stress incontinence, N (%)	50 (107,46.7%)	10 (52, 19.2%)	2 (7,28.6%)	.002	17 (46, 37.0%)	10 (26, 38.5%)	35 (94,37.2%)	1.00
Prolapse stage ≥ 2, N (total per group, %)+	24 (46, 51.2%)	21 (42, 50.0%)	3 (4,75.0%)	.77	12 (24,50.0%)	8 (15, 53.3%)	28 (53,52.8%)	1.00

VAS, visual analog scale; PGI, Patient Global Impression; PFDI-20, Pelvic Floor Distress Inventory 20; PFIQ-7, Pelvic Floor Impact Questionnaire 7

130 patients completed baseline VAS scores and 153 patients completed baseline PFDI-20 and PFIQ-7 scores. Baseline prolapse data was available on 100 patients. Baseline SUI data was available on 166 patients

Data reported as N (%) or median (interquartile range) except where specified.

^aSling mesh includes bladder sling, midurethral sling, retropubic sling, pubovaginal sling, tension free vaginal tape, transvaginal tape or transobtorator tape.

^bProlapse mesh includes sacrocolpopexy, sacrocervicopexy, hysteropexy, rectopexy with mesh, transvaginal mesh, or rectocele, cystocele or enterocele repair with mesh.

Most patients reported moderate to severe baseline symptoms on the PGI-S. Baseline symptom bother (PFDI-20) was >200 in all mesh types and highest among patients with pain (median score: 234.2, (IQR: 82.9)) followed by exposure plus pain (205.6 (IQR: 78.3)) and exposure alone, (181.3 (125.0), p=.01). The overall impact of symptoms on daily function and quality of life were worst in the pain only group (median PFIQ-7: 181.0, (IQR: 138.1)) followed by exposure plus pain (109.5 (IQR: 152.4)) and exposure only, (48.8 (IQR: 73.8), p<.001). There were no differences in baseline symptoms by prolapse mesh type (sacrocolpopexy versus transvaginal, Table S1). More women with retropubic slings reported moderate to severe symptoms than those with transobturator slings (100% vs 57.1%, p=0.02) but all other baseline measures were similar (Table S2).

Outcomes Following Mesh Removal

Of the patients who had mesh removed for pain, just 33.3% reported “improved” symptoms (PGI-I) compared to 58.3% in the exposure plus pain and 73.9% in the exposure only group after 6–24 months (p=.03) (Table 3). A total of 26 women (25%) reported that their symptoms were overall worse (“a little worse,” “much worse,” or “very much worse,”) on PGI-I with a similar number reporting this outcome in the 3 groups (P = 0.66). Symptom severity on PGI-S improved across all groups; however, over 40% of patients with pain or exposure plus pain continued to report moderate to severe symptoms after complete vaginal mesh excision. Median VAS scores for dyspareunia and pelvic pain met MCID for improvement in 64.7% and 52.9% of patients with pain only, and 51.1% and 54.9% with exposure plus pain. Median VAS dysuria scores did not improve in any group. Mesh type (sling or POP) did not influence VAS improvement (Figure 1d–f and Table 3).

Table 3.

Post-operative patient reported outcomes

Measure	Slinga N=59	Prolapseb N=44	Sling & Prolapse N=5	P value	Pain N=21	Exposure N=23	Exposure + Pain N=64	P value
PGI-Improvement Post-op, N (%)				1.00				.03
Improved	33 (56.9)	23 (56.1)	3 (60)		7 (33.3)	17 (73.9)	35 (58.3)
Not Improved or worse	25 (43.1)	18 (43.9)	2 (40)		14 (66.7)	6 (26.1)	25 (41.7)
PGI-Severity Post-op, N (%)				.23				.08
Normal-Mild	31 (54.4)	29 (70.7)	4 (80)		11 (52.4)	18 (81.8)	35 (58.3)
Moderate-Severe	26 (45.6)	12 (29.3)	1 (20)		10 (47.6)	4 (18.2)	25 (41.7)
VAS Score
Dyspareunia post op, median (IQR)	23 (59)	12 (69)	16 (78)	.88	44 (94)	0 (2)	29 (72)	<.001
Dyspareunia MCID -13, N (%)	22 (55)	12 (40)	2 (50)	.44	11 (64.7)	2 (16.7)	23 (51.1)	.04
Pelvic Pain post op, median (IQR)	18(47)	23 (41)	0	.19	51 (38)	0 (7)	21 (45)	<.001
Pelvic Pain MCID -13, N (%)	25 (54.4)	17 (46)	1 (25)	.48	9 (52.9)	6 (31.6)	28 (54.9)	.23
Dysuria post op, median (IQR)	3 (10)	5 (19)	0 (38)	.69	48 (84)	3 (5)	15 (49)	.001
Dysuria MCID -13, N (%)	22 (41.5)	12 (30)	1 (25)	.49	4 (32)	0 (6)	5 (29)	.04
PFDI-20 post-op, median (IQR)	183.3 (94.2)	175 (143.3)	118.8 (143.8)	.58	222.9 (70.5)	118.8 (108.3)	185 (130.8)	.003
PFDI-20 MCID -45, N (%)	27 (47.4)	11 (29)	2 (50)	.15	6 (28.6)	9 (45)	25 (43.1)	.48
POPDI-6 post-op, median (IQR)	50 (37.5)	62.5 (50)	25 (62.5)	.50	75 (31.3)	50 (68.8)	55.6 (50)	.01
POPDI-6 MCID-16, N (%)	29 (50.9)	13 (34.2)	3 (75)	.14	9 (42.9)	8 (40)	28 (48.3)	.82
UDI-6 post-op, median (IQR)	66.7 (37.5)	71.9 (35.4)	50 (50)	.51	75 (25)	58.3 (25)	62.5 (37.5)	.07
UDI-6 MCID -11, N (%)	24 (42.1)	15 (39.5)	2 (50)	.94	7 (33.3)	9 (45)	25 (43.1)	.72
PFIQ-7 post-op, median (IQR)	58.6 (104.8)	57.1 (123.8)	28.6 (95.2)	.92	109 (119.1)	9.5 (52.4)	61.9 (114.3)	<.001
PFIQ-7 MCID -36, N (%)	29 (52.7)	10 (25.6)	3 (60)	.02	11 (52.4)	9 (40.9)	22 (39.3)	.60
POPIQ-7 post-op, median (IQR)	14.3 (33.3)	9.5 (42.9)	4.8 (38.1)	.95	28.6 (61.9)	0	19 (42.9)	<.001
POPIQ-7 MCID -29, N (%)	22 (38.6)	9 (22.5)	2 (40)	.20	10 (45.5)	2 (9.1)	21 (36.2)	.02
UIQ-7 post-op, median (IQR)	23.8 (50)	28.6 (52.4)	14.3 (47.6)	.61	55.6 (57.1)	0 (19.1)	28.6 (52.4)	<.001
UIQ-7 MCID -16, N (%)	27 (48.2)	7 (17.1)	2 (40)	.004	10 (47.6)	10 (43.5)	16 (27.6)	.16

PGI, Patient Global Impression; VAS, visual analog scale; IQR, interquartile range; PFDI-20, Pelvic Floor Distress Inventory 20; POPDI-6, Pelvic Organ Prolapse Distress Index 6; UDI-6, Urinary Distress Index 6; PFIQ-7, Pelvic Floor Impact Questionnaire 7; POPIQ-7, Pelvic Organ Prolapse Impact Questionnaire 7; UIQ-7, Urinary Impact Questionnaire 7; MCID, minimal clinically important difference

104 women completed PGI-I, 103 women completed PGI-S. 108 women completed at least one VAS follow up score. 99 women completed follow up PDFI, POPDI-6, UDI-6, PFIQ-7 and 102 women completed POPIQ-7 and UIQ-7.

Data reported as N (%) or median (interquartile range) except where specified.

^aSling mesh includes bladder sling, midurethral sling, retropubic sling, pubovaginal sling, tension free vaginal tape, transvaginal tape or transobtorator tape.

^bProlapse mesh includes sacrocolpopexy, sacrocervicopexy, hysteropexy, rectopexy with mesh, transvaginal mesh, or rectocele, cystocele or enterocele repair with mesh.

Figure 1. Comparison of dyspareunia, pelvic pain and dysuria visual analog scale (VAS) scores before and after mesh removal.

(A-C) VAS scores before and after mesh removal stratified by primary complication symptom (pain, exposure, exposure + pain) A. Dyspareunia; B. Pelvic Pain; C. Dysuria. (D-F) VAS scores before and after mesh removal stratified by the type of mesh removed. D. Dyspareunia; E. Pelvic Pain; F. Dysuria. Comparisons were made between groups pre- and post-operatively with Kruskal-Wallis. The ends of the boxes represent the 25^th and 75^th percentiles, the horizontal line indicate the median and the whiskers represent upper and lower adjacent values. Outlier data are shown as circles plotted beyond the whiskers. The proportion of groups meeting MCID score cutoffs post-operatively were compared with Fisher’s Exact and significant differences are noted with asterisks (*).

VAS: visual analog scale

Most patients with pain only did not experience a substantial difference in quality of life after mesh removal. Median PFDI-20 score after mesh removal was significantly higher among patients with pain (222.9, IQR: 70.5) compared to those with exposure (118.8, IQR: 108.3) and exposure plus pain (185 (IQR: 130.8), (p=.003) (Table 3). Just 28.6% of these patients had PFDI scores that surpassed the MCID (versus 45% exposure and 43.1% pain plus exposure). Pelvic floor-related quality of life and physical function were also worse in the pain only group compared to exposure or pain plus exposure. The median post-operative PFIQ-7 score in the pain group was 109 (IQR: 119.1) compared to 61.9 pain plus exposure (IQR: 114.3)] and 9.5 exposure (IQR: 52.4) (p<.001) (Table 3). This difference was also evident in the POP and UI impact subscales (p<.001) (Table 3). These differences cannot be explained by new onset SUI or prolapse as less than 11% of patients across symptom groups developed new SUI after mesh removal (p=1.00) and less than 34% had new prolapse (p=.47). There were no differences in outcomes after mesh removal by prolapse mesh type or sling type (Tables S1 and S2).

Multivariable logistic regression was performed to evaluate whether selected variables showing significant differences between groups at baseline were driving the associations observed between the subjective measures of improvement and type of mesh and/or indication for mesh removal. The results of these analyses confirmed those of the univariate analyses; that is, the indication for mesh removal was the primary independent predictor for measures of postoperative improvement.

Discussion:

Main Findings

The main findings of our study were that patients with the complication of pain had more severe symptoms, lower quality of life, higher functional impact, and less improvement in these domains after mesh removal compared to patients with exposure. In addition, 25% of women reported overall worse symptoms.

Interpretation

Poor symptom resolution in women with pain corroborate the findings of previous smaller retrospective studies.^7,21,22,22In addition, the timing of symptom onset among patients undergoing mesh removal varied widely. Patients with pain symptoms reported median symptom onset within the first year after mesh placement; however, most patients with exposure did not develop symptoms until two years after implantation consistent with previous literature.²³ For reasons not appreciable in this study, the median time to mesh removal at our institution was 6.5 years for all patients in the cohort. Our analysis showed that patients with pain had mesh removed a median of 6.7 years after symptom onset while patients with exposure had mesh removed 8 years after symptom onset. This supports our finding that exposure may be associated with less symptom bother than pain. In addition, since most patients underwent their index surgery at an outside institution, a delay in care could represent initial conservative management by the primary institution.²⁴ 14% of patients also reported a history of a prior mesh excision but the details of these excisions that occurred prior to presentation into our system are not available. This may explain the contrast between our findings and prior retrospective studies in which the interval time from index surgery to mesh excision was 6 months to 2.5 years. ^4,8,25 A delay between symptom onset and removal may contribute to poor resolution of symptoms; however, data from our institution indicate that duration of symptoms or duration of implantation do not appear to be predictive of response to removal.²⁶

A significant number of patients in this study had at least one opioid prescription at the time of mesh excision, including 40% of patients with pain. Even though the distribution of opioid prescriptions in our cohort was higher than the general population (17.4% of U.S. adults received at least one prescription in 2017),²⁷ our data cannot determine the associated indication, prescribing physician, frequency or duration of opioid use. It is noteworthy that in the exposure only group, the percentage more closely mimicked that of the general population at 14.8%. Preoperative narcotic use is a known risk factor for ongoing postoperative pain in patients undergoing synthetic midurethral sling revision or removal (OR 6.9).²² The association between preoperative narcotics and poor postoperative outcomes has also been reported in other surgical subspecialties.^{28,29, 30}

We identified higher rates of past or current smoking among patients undergoing sling removal, however, there were no differences when patients were stratified by complication type. These findings corroborate previous studies which identified smoking as an independent predictor of mesh exposure and urogynecologic mesh revision or removal. ^31–33

Clinical Implications

Two thirds of patients with pain only, reported no improvement or worsening symptoms and half had moderate to severe persistent symptoms after their mesh had been removed. Rates of improvement were slightly better in those with exposure plus pain compared to pain only, suggesting that women with exposure had symptoms more amenable to correction and therefore, felt their overall condition was improved. Our finding that most women with exposure only improved after mesh removal strongly suggests that pain following mesh implantation is a serious complication associated with a high symptom burden that has a high likelihood of not resolving even with complete excision.

Research Implications

Our findings strongly suggest that patient factors that predispose to pain prior to mesh insertion is a critical next step in the management of urogynecologic surgical patients. In addition, further work is needed to identify intraoperative characteristics that predict response to removal. Karmarker et al reported success rates of 95% when MUS mesh was partially excised in patients with mesh exposure.³³ It is unclear whether partial excision as opposed to complete excision as performed in this study would be as effective for patients with pain. We also found that opioid use was high in this population and future studies will aim to describe opioid use in this study population in more detail. There was no difference in postoperative outcomes by type of prolapse mesh (transvaginal or sacrocolpopexy) or type of sling mesh, although this could be due to smaller numbers in this exploratory subanalysis. Recent literature suggests that women who undergo combined SUI and prolapse mesh revision surgeries have increased risk of complications compared to women with prolapse or SUI mesh alone (p=0.045).³⁴ This needs to be considered by patients and sugeons during operative planning because the incidence of serious complications six months was higher (7%) for combined MUS and prolapse mesh surgery compared with MUS alone (3.5%).³⁵ Finally, six of these patients required a second mesh removal and, although too small for meaningful subanalysis, this is also an area of interesting future work to examine risk factors for second surgery and the impact of this on improvement.

Strengths and Limitations

There are several strengths to this study, including its large prospective design. Patients were enrolled at a tertiary referral center with a large catchment area from both rural and urban locations. 89% of patients in our study had their index mesh surgery at an outside hospital and were referred to our institution for management of a complication. Previous literature suggests that ~50% of women presenting to tertiary referral centers for management of mesh complications had their index procedure at a different medical center.⁴ This may introduce referral bias, however, the geographic diversity of our patient population contributes to the generalizability of our findings. Post operative outcomes were assessed with validated patient centered questionnaires including condition-specific pelvic floor questionnaires, multiple pain domains and global assessments. Finally, we followed patients for up to two years with questionnaires and clinical exams in which new onset SUI and POP were documented.

This research should be considered in the context of several limitations. The primary limitation is that the follow-up response rate for the primary outcome, PGI-I, at 6, 12 and 24 months, was 68%. This rate was lowest among patients with pain (44%), followed by exposure plus pain (62%) and exposure (85%). Missing data may be non-random and informative because patients in the exposure group with better outcomes had a higher response rate. Sensitivity analysis revealed that women with follow up were slightly older and more often postmenopausal, post-hysterectomy, nonsmokers, and had exposure complications. However, our data are consistent with rates of improvement in prior studies and these data represent one of the largest cohorts available. Similarly, the lost to follow-up rate in this study was 71% at 24 months. We relied on in-person visits for study follow-up and many patients did not return to the clinic after 6 months. In addition, although we used validated questionnaires designed to assess global improvement and severity of symptoms, which we expect to capture our outcomes of interest, the PFDI and PFIQ are not validated in patients with mesh complications, therefore they may be less sensitive to detect clinical outcomes after mesh excision. Recently, a Pelvic Floor Complication Scale was created to assess peri- and postoperative outcomes from the patient and surgeon perspective.^36,37 Future research should incorporate this scale to assess patient outcomes. Finally, this study was designed to assess improvemet in quality of life after mesh excision compared to prior to excision in women with mesh complications. It does not provide any information on their quality of life before treatment of their pelvic floor disorders.

Conclusions

To our knowledge, this is the first large prospective study of women undergoing surgical mesh removal for the management of pelvic mesh-related complications. Our findings suggest that surgical management of exposure has better outcomes in the absence of pain. Patients with pain may continue to have symptoms and reduced quality of life. Our study adds to the literature on managing mesh complications and provides critical information to inform preoperative discussion with patients seeking mesh removal for pain.

Supplementary Material

sm1

Figure S1. Study Flowchart. For patients with >1 mesh removal during study period, data from the first removal was excluded and data from most recent removal was include in the analysis.