Pelvic organ prolapse after subtotal and total hysterectomy: a long-term follow-up of an open randomised controlled multicentre study

Authors

  • P Persson,

    Corresponding author
    1. Department of Women's and Children's Health, Uppsala University, Uppsala, Sweden
    • Correspondence: Dr P Persson, Department of Women's and Children's Health, Uppsala University, S-751 85, Uppsala, Sweden. Email par.persson@kbh.uu.se

    Search for more papers by this author
  • J Brynhildsen,

    1. Division of Obstetrics and Gynaecology, Department of Clinical and Experimental Medicine, Faculty of Health Sciences and Department of Obstetrics and Gynaecology, Linköping University, County Council of Östergötland, Linköping, Sweden
    Search for more papers by this author
  • P Kjølhede,

    1. Division of Obstetrics and Gynaecology, Department of Clinical and Experimental Medicine, Faculty of Health Sciences and Department of Obstetrics and Gynaecology, Linköping University, County Council of Östergötland, Linköping, Sweden
    Search for more papers by this author
  • and on behalf of the Hysterectomy Multicentre Study Group in South-East Sweden


Abstract

Objective

To analyse the development of pelvic organ prolapse (POP) after subtotal (SH) and total hysterectomy (TH) in the long-term, and to assess patient-reported symptoms regarding pelvic floor dysfunction (PFD).

Design

Long-term follow-up study of a randomised controlled multicentre study.

Setting

Seven hospitals and one private clinic in the south-east of Sweden.

Population

Of the 184 women who were eligible from the original trial, 151 (82%) responded to a postal questionnaire and 128 (70%) were clinically examined.

Methods

Postal questionnaire using the short-form version of the Pelvic Floor Distress Inventory (PFDI-20) and clinical examination using the POP-Q system. Multivariate analyses were used.

Main outcome measures

POP-Q measurements and symptoms of PFD.

Results

Follow-up time was a median of 11.3 years. Less than 3% had stage-3 prolapse. No significant difference was found in the presence of stage-2 or higher stage prolapse between the two hysterectomy groups (39% in SH versus 37% in TH; OR 1.28, 95% CI 0.59–2.80). Nor was there any significant difference in the quality-of-life measurement between the SH and TH groups [summary score PFDI-20: median 93 (range 60–201) versus 87 (range 60–186); Fisher's protected least significant difference post hoc test, = 0.78 ). None of the symptoms of PFD revealed statistically significant differences between the hysterectomy groups.

Conclusions

This long-term follow-up study of PFD showed basically no significant differences in subjective or objective measurements of POP, or in specific pelvic floor quality-of-life aspects after SH and TH. However, because of the low statistical power the results are inconclusive. Larger trials, and probably also a longer follow-up period, are necessary.

Introduction

Hysterectomy is the most frequently performed major gynaecological intervention.[1] Subtotal hysterectomy is an option for the removal of the uterus for benign gynaecological disorders. This procedure has gained popularity among women and doctors because of reports indicating potential advantages.[2-5] It also seems to be a mode of hysterectomy that is preferred and recommended by gynaecologists to women with certain gynaecological disorders.[6-8]

Relatively few randomised trials have compared subtotal abdominal hysterectomy (SH) and total abdominal hysterectomy (TH).[9-15] In a recently updated meta-analysis no major differences have been shown in physical, including urinary and bowel function, psychological, or sexual outcomes between the two modes.[1] Women are more likely to experience continued cyclical bleeding up to a year after surgery with SH, and because of the preserved cervix the women should be encouraged to continuously and regularly perform check-ups for cervical neoplasia.

The role of hysterectomy as a risk factor for post-hysterectomy pelvic organ prolapse (POP) has been discussed for many years, but there is still no consensus about the role of mode of hysterectomy as an aetiologic factor.[16] The impact of mode of hysterectomy on the development of post-hysterectomy POP has not been well investigated. Only two of the randomised trials comparing SH and TH have focused on the development of POP in the long-term,[17, 18] and both found no significant differences in outcome of POP development. POP is a common gynaecological problem with a multifactorial aetiology.[19] The pelvic floor anatomy is complex, and it is important to recognise this complexity in order to understand the development of POP.[20]

We have previously published a randomised trial comparing SH and TH in order to analyse short-term recovery and psychological wellbeing 1 year after surgery.[15, 21] Based on the women participating in that randomised trial we aimed to investigate the development of POP subjectively and objectively in the long term. In addition to the information such a study will provide concerning the association between mode of hysterectomy and later development of prolapse, there are quality-of-life and health-economic aspects of this issue to be considered if one of the hysterectomy methods is found to increase the risk of subsequently developing a prolapse that requires surgery.

The aims of this study were to analyse the development of POP in SH and TH in the long term, and to assess patient-reported symptoms regarding pelvic floor dysfunction.

Methods

This is a long-term follow-up study of women who participated in a previously published randomised controlled open multicentre study comparing SH and TH for benign gynaecological diseases.[15, 21] The randomised study was conducted between 1998 and 2004, and the follow-up study was performed between May 2011 and April 2012.

The flow chart is shown in Figure 1. All 184 women who were operated on in the randomised trial and were alive in April 2011 were approached by mail. The letter contained a section explaining the purpose of the study, including an informed consent form, and a questionnaire with 33 questions dealing with symptoms of pelvic floor dysfunction (PFD). Each woman was asked to answer whether she wanted to participate in the study or not. If affirmative, she signed the informed consent form and returned it together with the questionnaire. A reminder was sent to the non-responders within 4 weeks. One hundred and sixty women answered the letter: eight of these did not want to participate at all, and one wished to take part in the clinical examination only. In all, 24 women did not respond.

Figure 1.

Flow chart at long-term follow-up.

Surgical technique of hysterectomy

The hysterectomy was performed or supervised by a consultant and conducted under general anaesthesia. The decision of mode of skin incision, transverse or low midline, was made by the surgeon based on the size of the uterus and/or previous lower abdominal surgery. The surgical technique of the hysterectomy was left to the surgeon's discretion. Basically, the surgical techniques of the TH and SH were carried out according to the principles described by Thompson,[22] with the exception that no pelvic peritonealisation was made in either SH or TH. The sacrouterine and cardinal ligaments were anchored to the vaginal corner in TH, whereas no plication or reinforcement of the ligaments was performed in SH.

Postal questionnaire concerning PFD and quality of life

To assess symptoms and impact on quality of life (QoL) of PFD, the women completed a postal questionnaire including a modified version of the condition-specific QoL questionnaire, the short-form version of the Pelvic Floor Distress Inventory (PFDI-20).[23] The PDFI-20 is a composite of three different scales: the Urinary Distress Inventory (UDI-6), the Pelvic Organ Prolapse Distress Inventory (POPDI-6), and the Colorectal–Anal Distress Inventory (CRADI-8). The PFDI is a reliable and commonly used instrument in pelvic floor disorders, and is useful in clinical practice.[23, 24] Furthermore, it has been validated for use in Sweden.[25] In order to keep the questionnaire to a reasonable size so as to obtain a high response rate, we excluded some questions from the PDFI-20. The excluded items were four questions from the CRADI-8 that reflected on topics other than prolapse-related perceived discomfort, such as functional bowel disorders, gastrointestinal diseases, dietary problems, and anorectal disorder. One question from the POPDI-6 was excluded, as a similar question concerning digital assistance in pushing the vaginal bulge was given in the CRADI-8. In summary, our modification of the PFDI-20 excluded one question from POPDI-6 (‘Do you ever have to push up on a bulge in the vaginal area with your finger to start or complete urination?’) and four questions from CRADI-8 (‘Do you feel you need to strain too hard to have a bowel movement’?; ‘Do you usually have pain when you pass your stool?’; ‘Do you experience a strong sense of urgency and have to rush to the bathroom to have a bowel movement?’, and ‘Does a part of your bowel even pass through the rectum and bulge outside during or after bowel movement?’). In addition, the questionnaire contained five generic questions concerning the female genitals, five questions concerning urinary symptoms and habits, three questions concerning the impact of urinary and faecal leakage on social and daily living, and finally five questions concerning the frequency of use of sanitary pads and leakage of urine, gas, and loose and solid stool. In total, the questionnaire thus contained 33 questions. The questions were constructed as simple sentences. Each question was answered by putting an X in the box that was most appropriate. The answers were given on a Likert-type scale with the options: ‘No’ (0); ‘Yes, but it does not bother me at all’ (1); ‘Yes, and it bothers me somewhat’ (2); ‘Yes, it bothers me moderately’ (3); and ‘Yes, it bothers me quite a bit’ (4). For the questions about frequency of use of sanitary pads and leakage, the options were: ‘Never’ (0); ‘Sometimes yearly’ (1); ‘Sometimes monthly’ (2); ‘Sometimes weekly’ (3); or ‘Daily’ (4). For each of the three scales in the PFDI the score was calculated as the mean score of the questions belonging to the scale, multiplied by 25. Thus each of the three scales scored from 0 (least distress) to 100 (greatest distress). The sum of the scores of these three scales serves as the overall summary score of the PFDI-20, and ranges from 0 to 300.

Pelvic organ prolapse quantification

Prior to the hysterectomy and at the long-term follow-up the women were interviewed and clinically evaluated by the doctor. Gynaecological examination was performed with the woman in the dorsal lithotomy position, and with an emptied bladder. The Pelvic Organ Prolapse-Quantification (POP-Q) classification system, as described by the International Continence Society (ICS) was used.[26]

Briefly, the POP-Q considers six defined points within the vagina: two anterior (Aa and Ba), two posterior (Ap and Bp), and two apical (C and D) (Figure 2, reproduced from Digesu et al.[27]). The hymen is considered as a landmark for reference. Each point is expressed as distance in centimetres from the hymen with the woman performing maximum Valsalva. Each point is defined as zero if it is seen at the level of the hymenal ring, and as negative or positive measures if they are seen above or below the hymen, respectively. POP-Q also considers three other measurements: total vaginal length (tvl), genital hiatus (gh), and perineal body (pb). These are expressed in centimetres, and have no sign because they denote lengths and not positions relative to the hymen. tvl is the greatest depth of the vagina in centimetres when C and D are reduced to full normal position. gh is measured from the middle of the external urethral meatus to the posterior midline hymen. pb is measured from the posterior margin of the GH to the middle of the anal orifice. By definition, the ranges of position of point Aa and Ap, respectively, relative to the hymen, run from −3 to +3 cm, and for point Ba and Bp, respectively, run from −3 cm and +tvl. Stages of the prolapse (stages 0–4) are assigned according to the most severe portion when the full extent of the protrusion has been demonstrated (Figure 3).[26]

Figure 2.

The POP-Q system, as described by the ICS.[26] (Reproduced from Digesu et al.[27])

Figure 3.

Quantitative staging of prolapse according to the POP-Q classification.

At the preoperative examination the POP-Q measurement included just a qualitative determination of the stage of prolapse. In the follow-up examination the POP-Q measurements were quantitative as well as qualitative. At the interviews the women were asked standardised questions about the occurrence of stress urinary incontinence, urgency and urge incontinence, faecal incontinence, feeling of incomplete bladder emptying, and frequency of bowel movements. Constipation was defined as bowel movements less than three times per week. At the follow-up interview additional information was obtained regarding heredity for POP surgery, previous POP and urinary incontinence surgery, current smoking habits, current age, height, and weight, and the birthweight of the largest baby delivered. The patient records of those who had had surgery for POP or urinary incontinence in the follow-up period were scrutinised, and information about surgical procedure and POP-Q status prior to the POP surgery was registered. The POP-Q staging measurements obtained just prior to the POP surgery were used as a substitute for measurement in the follow-up outcome in these cases. The patient records of the non-participating women were also examined in order to disclose surgery on the cervix, POP, or urinary incontinence surgery.

Statistics

The data were analysed according to the intention-to-treat principle. Univariate analyses were performed with non-parametric statistical tests: Mann–Whitney U-test, Yates' corrected χ2 test, and Fisher's exact test, as appropriate. Dependent outcome variables were evaluated by means of multiple logistic regression analyses for dichotomous nominal variables and analyses of covariance (ancova) for continuous variables. The results were presented as crude and adjusted odds ratios (ORs and aORs), with 95% confidence intervals (95% CIs), or P values. In the multivariate models adjustments were performed simultaneously for the confounding factors age, parity, and body mass index (BMI). Additional potential confounding factors may be added to the models, and if so the confounding factors are specifically stated in the text, figures, or tables. The statistical analyses were performed with statview® 5.0.1 for windows, (SAS Institute Inc., SAS Campus Drive, Cary, NC, USA).

Results

The response rate to the questionnaire was 82% (151/184), and a clinical examination was carried out in 70% (128/184). The demographic and clinical baseline data of participants and non-participants in the follow-up study are presented in Table 1. The non-participants were significantly younger than the participants; otherwise, no significant differences were observed between participants and non-participants.

Table 1. Baseline data of participants and non-participants
 Participants (n = 151)Non-participants (n = 33) P a
  1. EBV, estimated bleeding volume perioperatively.

  2. Data are means and SDs or numbers and percentages.

  3. a

    Mann–Whitney U-test, Yates' corrected χ2 test, and Fisher's exact test, as appropriate.

  4. b

    Frequency of once a month or more often.

Age (years)46.2 (5.2)43.4 (5.1)0.0021
Parity2.1 (1.0)2.1 (1.4)0.5031
BMI (kg/m2)25.5 (3.9)26.5 (4.2)0.1959
Smokers39 (26%)13 (39%)0.1827
Follow-up time (years)10.8 (1.7)10.4 (2.0)0.3055
Indication for hysterectomy
Bleeding disturbances107 (71%)24 (73%)0.9736
Mechanical symptoms30 (20%)6 (18%)
Dysmenorrhea/endometriosis/pelvic pain14 (9%)3 (9%)
Mode of hysterectomy
Subtotal hysterectomy80 (53%)18 (55%)0.9999
Total hysterectomy71 (47%)15 (45%)
Occurrence of prolapse before hysterectomy
Anterior wall (≥stage 2)1 (0.7%)0 (0%)0.9999
Posterior wall (≥stage 2)1 (0.7%)0 (0%)0.9999
Apical (≥stage 2)0 (0%)0 (0%)1.0000
Symptoms of pelvic floor dysfunctionb
Stress urinary incontinence35 (23%)5 (15%)0.4355
Urgency and urge incontinence15 (10%)4 (12%)0.7524
Incomplete bladder emptying10 (7%)1 (3%)0.6919
Incontinence of stools0 (0%)0 (0%)1.0000
Constipation (<3 bowel movements per week)

(n = 132)

8 (6%)

(n = 28)

2 (7%)

0.6876
Bowel movements per week (number)7.1 (2.9)7.6 (5.1)0.8072

Demographic data of the hysterectomy groups at baseline and follow-up are shown in Table 2. The median follow-up period was 11.3 years (range 7.0–13.7 years). No significant differences were seen between the SH and TH groups concerning any of the demographic or clinical parameters depicted in Table 1 at follow-up, except for operating time, which shows a significantly shorter time for the SH group (data not shown).

Table 2. Demographic and clinical data on different occasions for participants in the study
 Occasion of evaluation
Baseline at hysterectomyAt follow-up questionnaireAt follow-up clinical examination
SH group (n = 98)TH group (n = 86) P a SH group (n = 80)TH group (n = 71) P a SH group (n = 70)TH group (n = 58) P a
  1. NA, not applicable or no data available.

  2. Data are means and SDs or numbers and percentages.

  3. a

    Mann–Whitney U-test.

  4. b

    Yates' corrected χ2 test.

Age (years)45.5 (5.3)45.8 (5.2)0.768256.9 (5.6)57.1 (5.3)0.849256.4 (5.6)57.2 (5.6)0.5266
BMI (kg/m2)25.6 (3.7)25.8 (4.3)0.824426.6 (3.9)27.1 (4.6)0.403726.8 (4.0)27.2 (4.9)0.5853
Parity2.2 (1.1)2.1 (1.1)0.27252.2 (1.0)2.1 (1.0)0.30882.2 (1.0)2.2 (1.0)0.6962
Smokers28 (29%)24 (28%)0.9999NANA 13 (19%)8 (14%)0.6310b
ERT usersNANA NANA 17 (24%)9 (16%)0.3140b
Follow-up period (years) 10.8 (1.6)10.8 (1.7)0.952410.7 (1.7)10.8 (1.8)0.9633

Seven women (7%, 7/98) in the SH group had the preserved cervix removed (9–120 months postoperatively). One woman (1%, 1/98) in the SH group underwent prolapse surgery during the follow-up period, compared with three women (3.5%, 3/86) in the TH group (9–32 months postoperatively). Three women in each group (3.1%, 3/98; 3.5%, 3/86) had surgery for urinary incontinence (TVT) (8–120 months postoperatively) during the follow-up period.

The qualitative and quantitative POP-Q measurements are shown in Tables 3 and 4, respectively. No significant differences were found for stage-2 or higher stage prolapse in any of the vaginal compartments, or overall, between the two hysterectomy groups. The number of women with stage-3 prolapse was low [anterior compartment, 2/70 (2.9%) in the SH group versus 1/58 (1.7%) in the TH group; posterior compartment, 0/70 (0%) versus 2/58 (3.4%); and apical compartment, 0/70 (0%) versus 1/58 (1.7%), respectively]. Overall, nearly 40% of the women presented with stage-2 or higher prolapse. The prolapse was beyond the hymen in only four women (9%, 4/44) of those with stage-2 prolapse. Concerning the quantitative measurements (Table 4), point D as well as tvl were significantly longer in the SH group compared with the TH group (mean differences of 1 and 0.8 cm, respectively).

Table 3. POP-Q measurements and occurrence of pelvic organ prolapse preoperatively, and at a median of 11 years after abdominal subtotal and total hysterectomy
 Occasion of evaluation
Baseline at hysterectomyFollow-up examination
SH group proportion (%)TH group proportion (%)SH group proportion (%)TH group proportion (%)OR (95% CI)Adjusted OR (95% CI)a
  1. a

    Adjusted for age, parity, BMI, mode of delivery: Caesarean section only, and birth weight of heaviest baby delivered.

  2. b

    Adjusted to the pre-POP surgery POP-Q stage for those who had POP surgery.

  3. c

    Overall POP-Q ≥ stage 2 = at least one compartment with stage 2 or higher stage.

Point Ba ≥ −1 cmb0/98 (0%)1/86 (1.2%)16/70 (23%)8/58 (14%)1.85 (0.73–4.69)2.39 (0.89–6.41
Point Bp ≥ −1 cmb0/98 (0%)1/86 (1.2%)18/70 (26%)17/58 (29%)0.83 (0.38–1.82)0.97 (0.42–2.22)
Point C/D ≥ −1 cmb0/98 (0%)0/86 (0%)2/70 (3%)1/58 (2%)1.68 (0.15–18.87)1,61 (0.06–43.48)
Overall POP-Q ≥ stage 2b,c0/98 (0%)1/86 (1.2%)27/70 (39%)22/58 (37%)1.03 (0.50–2.10)1.28 (0.59–2.80)
Table 4. POP-Q measurements at follow-up at a median of 11 years after subtotal and total hysterectomy
 SH group (n = 70)TH group (n = 58)Analysis of covarianceaPost hoc test Fisher's PLSD
MedianRangeMedianRangeF1,126 value P P
  1. ERT, estrogen replacement therapy; NA, not applicable; PLSD, protected least significant difference.

  2. a

    Adjusted for age, parity, BMI, current smoking habits, use of ERT, surgery for prolapse, and cervix extirpation.

Point Aa (cm)−2.5−3.0, 2.0−2.0−3.0, 0.00.3690.54490.5313
Point Ba (cm)−2.3−3.0, 2.0−2.0−3.0, 0.00.9290.33720.3457
Point Ap (cm)−2.5−3.0, 1.0−2.0−3.0, 0.00.2610.61020.5793
Point Bp (cm)−2.5−3.0, 1.0−2.0−3.0, 0.50.0180.89360.9231
Point C (cm)−7.0−11.0, −0.5NANANA
Point D (cm)−8.0−12.0, −5.0−7.0−9.0, −5.019.385<0.0001<0.0001
tvl (cm)9.06.0, 13.09.06.0, 11.011.7160.00090.0008

The outcome of the PFD-specific QoL instrument PFDI-20 and the subscales are presented in Figure 4. No significant differences were found in any of the QoL measurements between the SH and TH groups, according to the analysis of covariance and the subsequent post hoc analysis.

Figure 4.

Summary score of PFDI and the subscales at a median of 11 years after subtotal (SH) and total hysterectomy (TH), and result of the analysis of covariance. Boxes represent median values and 25–75 per centiles. Bars represent 10–90 per centiles.

Symptoms of PDF recorded from the preoperative interviews and at the follow-up are reported in Table 5. None of the characteristics revealed statistically significant differences between the hysterectomy groups. The use of estrogen replacement therapy (ERT), locally or systemically, did not differ significantly between the SH and TH groups at the clinical follow-up [24.3% (17/70) versus 15.5% (9/58); aOR 1.76 (adjusted for age and BMI), 95% CI 0.71–4.35]. In all, 27% (21/78) of the women in the SH group declared that they had troublesome problems with their genitals, compared with 29% (20/69) in the TH group [aOR 0.96 (adjusted for age, BMI, ERT, prolapse, and incontinence surgery), 95% CI 0.43–2.12].

Table 5. Symptoms of pelvic floor dysfunction occurring with a frequency of once monthly or more often, reported by women at the preoperative interview and at the follow-up examination
 Occasion of evaluation
Baseline at hysterectomyFollow-up examination
SH group (n = 98)TH group (n = 86)SH group (n = 70)TH group (n = 58)OR (95% CI)Adjusted OR (95% CI)a
  1. a

    Adjusted for age, parity, BMI, mode of delivery: caesarean section only, birthweight of heaviest baby delivered, and preoperative symptom. The urinary irritation and incontinence symptoms were also adjusted for occurrence of incontinence surgery and use of estrogen replacement therapy. Incomplete bladder emptying and constipation were likewise adjusted for the occurrence of prolapse surgery.

Stress urinary incontinence20 (20.4%)20 (23.3%)27 (38.6%)17 (29.3%)1.51 (0.72–3.18)1.62 (0.71–3.67)
Urgency and urge incontinence8 (8.2%)11 (12.8%)10 (14.3%)7 (12.1%)1.21 (0.43–3.42)1.15 (0.37–3.62)
Incomplete bladder emptying7 (7.1%)4 (4.7%)8 (11.4%)7 (12.1%)0.94 (0.32–2.77)1.03 (0.31–3.39)
Incontinence of stools0 (0%)0 (0%)1 (1.4%)1 (1.7%)0.83 (0.05–13.51)NA
Constipation (<3 bowel movements/week)

(n = 83)

7 (8.4%)

(n = 77)

3 (3.9%)

2 (2.9%)2 (3.4%)0.82 (0.11–6.04)0.87 (0.09–7.99)

The prevalence and frequency of urinary and anal incontinence, and use of sanitary pads, according to the questionnaire at the follow-up, are shown in Table 6. The use of sanitary pads and loose stool incontinence revealed a significant difference between the groups, in favour of the TH group.

Table 6. Results from the questionnaire concerning the frequency of use of sanitary pads, and urinary and faecal incontinence
 SH group n (%)TH group n (%)OR (95% CI)Adjusted OR(95% CI)a
  1. a

    Adjusted for age, parity, and BMI. Urinary incontinence also adjusted for incontinence and POP surgery; anal incontinence adjusted for prolapse surgery; and use of sanitary pads adjusted for incontinence and POP surgery, and extirpation of cervix.

Use of sanitary pads
Never or less than monthly53 (69%)58 (83%)2.19 (1.00–4.81)2.75 (1.15–6.54)
Monthly or more often24 (31%)12 (17%)
Urinary incontinence
Never or less than monthly44 (55%)43 (61%)1.26 (0.66–2.40)1.48 (0.73–2.98)
Monthly or more often36 (45%)28 (39%)
Gas incontinence
Never or less than monthly47 (59%)43 (61%)1.05 (0.54–2.01)1.13 (0.58–2.20)
Monthly or more often32 (41%)28 (39%)
Solid stool incontinence
Never or less than monthly66 (83%)59 (83%)1.04 (0.45–2.43)1.33 (0.53–3.34)
Monthly or more often14 (18%)12 (17%)

Discussion

Main findings

This long-term follow-up study of PFD after SH and TH showed basically no significant differences in subjective or objective measurements of POP, or in pelvic floor-specific QoL aspects, between the modes of hysterectomy.

Strengths and weaknesses

The strengths of this study are the prospective and randomised design and the long follow-up period. Originally the power calculation was carried out for the primary outcome of the study, and thus no sample size estimation has been performed for this secondary study. Assuming that the difference in occurrence of prolapse of the anterior compartment between the groups (23 versus 14%) is significant at a 5% level, for 80% power a sample size of nearly 600 individuals is needed. Thus, both our study, and the other randomised long-term study of SH and TH presenting objective data,[17] is underpowered. Accordingly, the non-significant findings in this study make the results inconclusive, and consequently should be interpreted with great caution. Another drawback of our study is the lack of pelvic floor-specific QoL assessment preoperatively; however, the preoperative prevalence of prolapse and urinary incontinence were similar in the two groups. Besides, we used two validated generic QoL forms (the Psychological General Well-Being and Women's Health Questionnaire) in the primary study, and found no significant difference between the modes of hysterectomy on any occasion of assessment, up to 1 year after surgery.[21] It therefore seems conceivable that even a pelvic floor-specific QoL would be similar preoperatively. At the follow-up we used a modified PFDI-20 form to measure the pelvic floor-specific QoL. We are aware that the validation of the form cannot therefore be guaranteed; however, we compared the scores obtained from the same questions between the two modes of hysterectomy and found no significant difference. Thus, by adding questions that, to a substantial degree, may cover problems other than pelvic floor organ dysfunction seems unlikely to improve the chance to detect a difference in pelvic floor-specific QoL.

Interpretation

In a meta-analysis of studies of SH and TH, Gimbel stated that prolapse formation and urinary incontinence were more often associated with SH than with TH.[28, 29] Most studies in the meta-analysis were observational, and the results of the two randomised trials that were included did not support the conclusion. The follow-up period was relatively short (1 year), or was not mentioned, and the prolapse identification was not based on clinical examination but on a questionnaire. Only two randomised trials of SH and TH have presented long-term follow-up data concerning PFD.[17, 18] Both of these studies conclude that there were no major long-term differences in outcome after SH and TH; however, these studies have substantial restrictions. The study by Greer et al. had a low number of participants, with only 37 women of the 135 randomised in the original study, and presented only subjective data obtained from a questionnaire.[18] The study by Thakar et al. suffered from a relatively low response rate (65%), and the clinical examination was conducted in less than 50% of the randomised women.[17] By contrast, our study had a response rate of 82%, and 70% of the randomised women were clinically examined.

In common with the findings of Thakar et al.,[17] we found no significant differences in POP-Q staging between the groups, but the prevalence of significant prolapse (stage 2 or higher) in our study was substantially higher than that reported by Thakar. The same quantification system was used to assess the prolapse. It may be difficult to explain the difference. Whether it depends on simple selection bias, differences in surgical technique causing different outcome, or truly a difference in English and Swedish populations, remains unresolved. We found a considerable difference in the location of the prolapse component, with the highest prevalence of prolapse in the posterior compartment equally distributed between women in the SH and TH groups. In contrast with this finding, prolapse in the anterior compartment occurred with an odds ratio in the SH group that was more than twice that in the TH group. Based on our findings, it seems reasonable to present data for each compartment, as the impact of hysterectomy theoretically may influence each compartment differently.

Women that are operated on for genital prolapse are on average in the sixth decade of life. The mean age at follow-up of the women in the Thakar study was about 53 years,[17] and in our study 57 years. Thus the higher prevalence in our study could reflect an effect of age, and the possibility of having developed a prolapse because of tissue factors and aging. The follow-up period may indeed be too short, even in our study, to show a significant difference in the occurrence of prolapse between the two groups. In women without POP at the time of hysterectomy, Dällenback et al. found a cumulative risk for pelvic repair surgery after hysterectomy in about 1% after 15 years.[16] Even though 38% of the women in our study had a prolapse of stage 2 or higher, it must be emphasised that the prevalence of women with a prolapse beyond the level of the hymen was less than 5%, indicating that the vast majority had a very small prolapse.

Women with SH used significantly more sanitary pads at a median of 11 years after the hysterectomy than women with TH. The clinical impact of this is uncertain, but might reflect a drawback of SH, giving more periodic bleeding and vaginal discharge, and therefore requiring the woman to use sanitary pads even after menopause.

Analogous with the findings by Greer and Thakar we found no significant difference in the occurrence of urinary or faecal incontinence, the feeling of incomplete bladder emptying, or constipation between women with SH and women with TH. The pelvic floor-specific QoL assessments in the present study revealed no difference between the groups. Although the American and English long-term follow-up studies did not use pelvic floor-specific QoL forms, they found no difference between groups in health-related quality of life (HRQoL) using the SF-36, a widely used generic HRQoL measure.

The occurrence of symptoms of urinary dysfunction seems to be in agreement between our study and the English study, whereas the prevalence of bowel symptoms differs substantially. Whether this phenomenon reflects a true difference in the two populations or is a matter of cultural, dietary, or functional differences is speculative.

In summary, the results are inconclusive because of a lack of statistical power; however, although we found no significant differences in symptoms of PFD or POP, interestingly almost all odds ratios in the comparisons between the two groups were unanimously in favour of TH. This might indicate that SH actually affects pelvic floor function more adversely that TH. Sufficiently powered randomised trials or meta-analysis of randomised long-term follow-up studies are needed to clarify the impact of SH and TH on pelvic floor function.

Disclosure of interests

None of the authors or study group reports any conflicts of interest concerning this article.

Contribution to authorship

PP and PK planned the study and wrote the article. PP, JB, and PK collaborated in the clinical follow-up and analysed the data. All the authors revised the article and have seen and approved the final version.

Details of ethics approval

The study was registered in ClinicalTrial.gov Protocol Registration System (NCT00876057), with initial release 04/03/2009. The study was approved by the Research Ethical Board in Linköping (Dnr 2011/50-31).

Funding

The study was financially supported by grants from the Medical Research Council of South East Sweden, the County Council of Östergötland, and Linköping University.

Acknowledgements

In addition to the main authors of the article, the Hysterectomy Multicentre Study Group in South East Sweden included the following gynaecologists: Laila Falknäs MD, Ryhov Central Hospital, Jönköping; Johan Holmberg, MD, the County hospital, Eksjö; Christina Gunnervik MD, the County Hospital Värnamo; Margareta Lindvall MD, the Central Hospital, Kalmar; Tanja Konstantinova, MD, Vrinnevi Central Hospital, Norrköping; Anna Saland MD, the County Hospital, Motala and Fatma Bäckman MD, the University Hospital, Örebro.

Ancillary