Subtotal versus total abdominal hysterectomy: randomized clinical trial with 14-year questionnaire follow-up

Subtotal versus total abdominal hysterectomy: randomized clinical trial with 14-year questionnaire follow-upGYNECOLOGY
Subtotal versus total abdominal hysterectomy: randomized clinical trial with 14-year questionnaire follow-up Lea Laird Andersen, MD; Bent Ottesen, MD, DMSc; Lars Mikael Alling Møller, MD, PhD; Christian Gluud, MD, DMSc; Ann Tabor, MD, DMSc; Vibeke Zobbe, MD; Elise Hoffmann, MD; Helga Margrethe Gimbel, MD, DMSc; for the Danish Hysterectomy Trial Group
OBJECTIVE: The objective of the study was to compare long-term (Short Form-36 questionnaire), hospital contacts, and vaginal
results of subtotal vs total abdominal hysterectomy for benign uter- ine diseases 14 years after hysterectomy, with urinary incontinence as the primary outcome measure.
STUDY DESIGN: This was a long-term follow-up of a multicenter, randomized clinical trial without blinding. Eleven gynecological departments in Denmark contributed participants to the trial. Women referred for benign uterine diseases who did not have contraindications to subtotal abdominal hysterectomy were randomized to subtotal (n ¼ 161) vs total (n ¼ 158) abdominal hysterectomy. All women enrolled in the trial from 1996 to 2000 who were still alive and living in Denmark (n ¼ 304) were invited to answer the validated questionnaire used in prior 1 and 5 year follow-ups. Hospital contacts possibly related to hysterectomy from 5 to 14 years postoperatively were regis- tered from discharge summaries from all public hospitals in Denmark. The results were analyzed as intention to treat and per protocol. Possible bias caused by missing data was handled by multiple imputation. The primary outcome was urinary incontinence; the secondary outcomes were pelvic organ prolapse, constipation, pain, sexuality, quality of life
From the Department of Obstetrics and Gynecology, Nykøbing Falster Hospi association with the University of Southern Denmark (Drs Andersen and Gim Center (Dr Ottesen), Department of Obstetrics and Gynecology (Drs Zobbe Copenhagen Trial Unit, Center for Clinical Intervention Research (Dr Gluud), Copenhagen University Hospital, Copenhagen; and Department of Obstetric Roskilde Hospital, Roskilde (Drs Alling Møller and Hoffmann), Denmark.
Received Sept. 9, 2014; revised Nov. 14, 2014; accepted Dec. 17, 2014.
This long-term follow-up study was supported by the research foundation o University of Southern Denmark, and the Department of Gynecology, Nykøb Rigshospitalet (Copenhagen University Hospital), and Roskilde Hospital, Den
B.O. and H.M.G. are members of the board of the Danish Hysterectomy and Database. The other authors report no conflict of interest.
Presented in oral format at the 39th biannual meeting of the Nordic Federatio Obstetrics and Gynecology, Stockholm, Sweden, June 10-12, 2014, and as annual congress of Leading Lights in Urogynecology, European Urogynecolo Athens, Greece, Oct. 2-4, 2014.
Corresponding author: L. L. Andersen, MD. [email protected]
0002-9378/$36.00 ª 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1
758.e1 American Journal of Obstetrics & Gynecology JUNE 2015
bleeding.
RESULTS: The questionnaire was answered by 197 of 304 women (64.8%) (subtotal hysterectomy [n ¼ 97] [63.4%]; total hysterectomy [n¼ 100] [66.2%]). Mean follow-up time was 14 years and mean age at follow-up was 60.1 years. After subtotal abdominal hysterectomy, 32 of 97 women (33%) complained of urinary incontinence compared with 20 of 100 women (20%) after total abdominal hysterectomy 14 years after hysterectomy (relative risk, 1.67; 95% confidence interval, 1.02e2.70; P ¼ .035). After a multiple imputation analysis, this difference disappeared (relative risk, 1.36; 95% confidence interval, 0.86e2.13; P ¼ .19). No differences were seen in any of the sec- ondary outcomes.
CONCLUSION: Subtotal abdominal hysterectomy was not superior to total abdominal hysterectomy on any outcomes. More women seem to have subjective urinary incontinence 14 years after subtotal abdominal hysterectomy. This result was not confirmed by multiple imputation analysis and should be interpreted cautiously.
Key words: hysterectomy, long-term follow-up, pelvic organ prolapse, quality of life, urinary incontinence
Cite this article as: Andersen LL, Ottesen B, Alling Møller LM, et al. Subtotal versus total abdominal hysterectomy: randomized clinical trial with 14-year questionnaire follow-up. Am J Obstet Gynecol 2015;212:758.e1-54.
tal, Nykøbing Falster in bel); Juliane Marie and Tabor), and Rigshospitalet, s and Gynecology,
f Region Sjælland, ing Falster Hospital, mark.
Hysteroscopy
n of Societies of a poster at the 7th gical Association,
016/j.ajog.2014.12.039
pproximately 4500 benign hyster-
A ectomies are performed yearly in Denmark.1 Although the use of the less invasive laparoscopicmodeofhysterectomy is rising, 32% of Danish hysterectomies in 2012 were abdominal,2 andmore than 50% of hysterectomies in the United States were abdominal in 2010.3
Of abdominal hysterectomies, 10% were subtotal, and of laparoscopic hys- terectomies, 20% were subtotal in Denmark in 2011.1 In some hospitals in Germany, subtotal laparoscopic hyster- ectomy is the standard and accounts for more than 80% of laparoscopic hyster- ectomies.4 The background for prefer- ring subtotal hysterectomy is that it is simpler and quicker and may result in
ajog.org Gynecology Research
fewer complications.5 However, mor- cellation is part of this mode of hyster- ectomy, and because leiomyosarcomas are sometimes mistaken for fibromas, morcellation is no longer recommended by the Food and Drug Administration6; consequently, one may assume, in the future, large uteri will more often be removed by abdominal hysterectomy.
Studies in the 1980s7-9 suggested that subtotal abdominal hysterectomy (SAH) was superior to total abdominal hyster- ectomy (TAH) regarding sexual func- tion. This finding was not reproduced in randomized clinical trials (RCTs).10-12
The risk of cervical cancer in the remaining cervix is another important issue. Nevertheless, if a Papanicolaou smear is normal prior to surgery and the woman continues to participate in cer- vical cancer screening, the risk of cervical cancer is only approximately 0.03%.14,15
Three RCTs comparable with our Danish trial15,16 comparing SAH with TAH16-18 have performed long-term follow-up and found no significant dif- ferences between SAH and TAH on clinical outcomes. Few data on long- term outcomes after subtotal vs total laparoscopic hysterectomy are avail- able.19 Although open abdominal and laparoscopic surgery differ in many ways, the most recent Cochrane sys- tematic review20 on the topic included both methods and stated that there was no evidence to support the shift toward subtotal hysterectomy seen in laparos- copy. The authors of the review conclude that more long-term follow-up is needed because urogenital problems may occur years after surgery, especially in post- menopausal women.20
We aimed to compare 14-year out- comes after SAH vs TAH in women included in a randomized clinical trial for benign uterine diseases.14,15 The primary outcome is urinary inconti- nence (UI) 14 years after hysterectomy.
MATERIALS AND METHODS
In 1996e2000, 319 women from 11 gy- necological departments in Denmark were randomized to SAH vs TAH.14
Details about eligibility criteria, con- sent, inclusion, randomization, and sur- gical procedures have been published.14
The sample size of the original trial was calculated based on an assumed prevalence of the primary outcome, UI, 1 year after TAH of approximately 23%.21,22 With a power of 0.80, a type I error of 5%, and a 15% abso- lute difference in UI between the surgical groups, 160 participants had to be included in each intervention group.14
Results from 1 year of follow-up14
showed that significantly more women in the SAH group were urinary inconti- nent compared with the TAH group. A decrease in UI after hysterectomy was seen in both surgical groups. The sec- ondary outcomes postoperative compli- cations, quality of life (Short Form-36 [SF-36]), constipation, pelvic organ prolapse, satisfaction with sexual life, and pelvic pain did not show any dif- ference between surgical groups. Neither did the further analyses of lower urinary tract symptoms23 and sexuality.13 At 1 year, 20% of the SAH group still expe- rienced vaginal bleeding. At 5 years,15
the significant difference between SAH and TAH regarding UI was reproduced. The number of incontinent women was higher than at 1 year. In the SAH group, 11% still experienced vaginal bleeding. All participants still alive and living in
Denmark in September 2012 were con- tacted by letter, and it contained the validated questionnaire24 (Appendix; Supplemental Material) used in prior follow-ups.14,15 The questionnaire assessed primary and secondary out- comes (presented in the following text). Reminders were sent 2 and 7 months later to nonresponders. Participants were encouraged to return the ques- tionnaire unanswered if they did not wish to participate, thus avoiding re- minders. Age at follow-up and follow-up time was calculated with January 2013 as the cutoff point. The primary outcome, UI, was
defined as a subjective complaint of involuntary loss of urine often or always (question 35 in the question- naire). Because this result could re- flect a difference in treatment-seeking behavior between surgical groups rather than in the occurrence of UI, we also analyzed the number of
JUNE 2015 Ameri
women who reported UI at any time since hysterectomy including prior follow-ups.14,15
Secondary outcomes were hospital contacts, pelvic organ prolapse (POP), pelvic pain, satisfaction with sex life, constipation, quality of life (QoL), and vaginal bleeding after SAH. All out- comes, except QoL, were dichotomized, and the SAH and TAH groups were compared using a c2 test. Analyses were conducted as intention to treat as well as per protocol excluding participants that did not receive the allocated intervention (Figure 1). As in prior follow-ups,14,15
the conclusions are based on the intention-to-treat analyses. Additionally, satisfaction with sex life was analyzed separately for those stating they had a partner and those who did not.
QoL was assessed by the validated SF- 36 questionnaire25 included in our questionnaire (Supplemental Material). SF-36 was scored according to the spec- ifications by Quality Metric using the official scoring software. For each participant a physical component score (PCS) and a mental component score (MCS) were calculated. These scores are validated and a norms based mean of 50 is interpreted as average QoL. Means were compared between surgical groups using the Wilcoxon rank sum test because the scores were not normally distributed.
Some women did not answer all questions resulting in different totals for each analysis. The number in each group for the particular analysis is stated in Table 1. To account for possible bias caused by missing data because of the loss to follow-up and incomplete ques- tionnaires, multiple imputation (MI) was carried out using the FCS method in SAS (version 9.3; SAS Institute, Cary, NC) using the PROC MI and MIANA- LYZE functions. The 14-year outcomes imputed were UI, pelvic pain, POP, satisfaction with sex life, QoL, and constipation.
The following variables were included in the imputation model because they were associated with (P < .1) one or more of the outcomes in themultivariate logistic regression: baseline variables in- cluded type of surgery, number of
can Journal of Obstetrics & Gynecology 758.e2
FIGURE 1 Flowchart of participants
The figure shows participants at each stage of the trial from randomization through all follow-ups and
reasons for dropouts.
SAH, subtotal abdominal hysterectomy; TAH, total abdominal hysterectomy.
Andersen. SAH vs TAH, RCTwith 14 year follow-up. Am J Obstet Gynecol 2015.
Research Gynecology ajog.org
deliveries, largest baby greater than 4000 g, smoking more than 5 cigarettes per day, alcohol consumption greater than 14 units per week (1 unit ¼ 12 g of alcohol), UI, pain, and constipation. Follow-up variables included were as follows: UI, pelvic pain, POP, con- stipation, satisfaction with sex life at 1 and 5 years, and physical (PCS) and mental (MCS) QoL scores at 1 year.
One hundred imputed datasets were created using a maximum of 100 itera- tions. This was chosen to obtain high
758.e3 American Journal of Obstetrics & Gynecol
precision in the analyses. The MI method assumes that missingness is missing at random, meaning that missing data are related to other observed variables but not to unob- served variables or to the missing items.26 The imputed datasets were analyzed by a c2 test, and the pooled analyses were carried out using the MIANALYZE function (SAS Institute). Relative risks were logarithmically transformed before pooling to comply with Rubin’s rules for pooling imputed
ogy JUNE 2015
results.27 The QoL scores were not nor- mally distributed, and despite using the transformations of inverse, log, squared, gamma, exponential, box cox, and beta, we could not approximate the normality. We entered the untransformed scores in the MI model and analyzed them using the Wilcoxon rank sum test. Because a normal distribution is assumed in MI, this could potentially skew other out- comes. However, we ran theMIwith and without QoL included, and it had no impact on other outcomes.
Hospital contacts were registered for all randomized women by looking up all discharge summaries from Danish pub- lic hospitals from 5 years postoperatively until July 2013 in the central registry of discharge summaries. Any hospital contact regarding abdominal, gyneco- logical, urological (including UI), plastic surgical, or dermatological complaints were scrutinized. If the contact might be related to the prior hysterectomy, it was registered. Hospital contacts from the time of surgery until 5 years post- operatively have been published else- where.14,15 Hospital contacts were divided into the following categories: recurrent urinary tract infection (including pyelonephritis), pain, UI, POP, cervical problems (bleeding or dysplasia), other urogenital, skin prob- lems/hernias, and others.
All data were handled and analyzed using SASjmp version 10 statistical software (SAS Institute) except for MI, which was carried out in SAS version 9.3 (SAS Institute).
The original trial as well as this follow- up was accepted by the regional ethics committee journal number, SJ-268, as well as the Danish Data Protection Agency journal number 2012-41-0286.
RESULTS
We contacted 304 women (95.3%) (SAH: 153; TAH: 151); 10 (3.1%) had died, from causes unrelated to hysterec- tomy, and 5 (1.6%) had left Denmark. Two hundred forty-nine women (82%) returned the questionnaire; however, 52 (17.1%) returned it blank stating that they did not wish to participate. A total of 197 (64.8%) answered the question- naire (Figure 1).
Outcome (n [ SAH/TAH) SAH TAH
Observed data Multiple imputation
RR 95% CI P value RR 95% CI P value
UI, % (n ¼ 96/100) 32 (33.3) 20 (20) 1.67 1.02e2.70 .035a 1.36 0.86e2.13 .19
Constipation, % (n ¼ 97/100)
14 (14.4) 7 (7) 2.06 0.87e4.89 .091 1.77 0.83e3.77 .14
Pelvic organ prolapse, % (n ¼ 93/97)
12 (12.9) 11 (11.3) 1.14 0.53e2.45 .74 0.97 0.50e1.86 .92
Satisfied with sexual life, % (n ¼ 75/78)
48 (64) 53 (67.9) 0.94 0.75e1.18 .61 1.09 0.76e1.58 .64
Pelvic pain, % (n ¼ 96/100)
14 (14.6) 10 (10) 1.46 0.68e3.12 .33 1.33 0.69e2.55 .40
Vaginal bleeding, % (SAH only, n ¼ 97)
0
Mean (95% CI) TAH
PCS mean (95% CI) 50.4 (48.5e52.4) 51.3 (49.4e53.2) .54 50.05 (48.5e51.6) 50.9 (49.1e52.8) .67
MCS mean (95% CI) 54.8 (52.9e56.7) 53.2 (51.4e55.1) .39 54.4 (52.5e56.1) 52.2 (50.7e54.2) .87
CI, confidence interval; MCS, mental component score; PCS, physical component score; QoL, quality of life; RR, relative risk; SAH, subtotal abdominal hysterectomy; TAH, total abdominal hys- terectomy; UI, urinary incontinence.
a Statistically significant; b Wilcoxon rank sum test.
Andersen. SAH vs TAH, RCTwith 14 year follow-up. Am J Obstet Gynecol 2015.
ajog.org Gynecology Research
The number of participants in the groups was similar: 97 of 153 (63.4%) in the SAH group and 100 of 151 (66.2%) in the TAH group. Characteristics of the participants and nonparticipants in this follow-up (Table 2) did not differ ac- cording to baseline variables from time of surgery except that fewer of the par- ticipants were smokers at the time of surgery and more participants had an alcohol consumption greater than 14 units per week at time of surgery than the nonparticipants. The 2 surgical groups of responders were comparable (Table 2). Mean age at follow-up was 60.1 years; mean follow-up timewas 14.1 years.
More women in the SAH group (32 of 97, 33.3%) than in the TAH group (20 of 100, 20%) reported UI often or always (P ¼ .035) (Table 1). The difference was also significant in the per-protocol analysis (P ¼ .024) (Table 3). Table 3 also shows the other analyses of UI as described in the Materials and Methods section.
Analysis of multiple imputed data showed no significant differences be- tween surgical groups regarding UI (P ¼ .19) (Table 1). Neither the physical (PCS) nor the
mental (MCS) score of the SF-36 QoL questionnaire differed between the sur- gical groups, and the means were consistent with the expected mean of 50 (Table 1). None of the participants experienced vaginal bleeding at 14 years. Twenty-one women (11.5%) stated
they did not have a partner. Of these, 9 (42.9%) stated that they did not know whether they were satisfied with their sex life. Six (28.6%) stated they were satis- fied and 6 (28.6%) stated they were not. Among those with a partner (n ¼ 162, 88.5%), 22 (13.6%) did not know whether they were satisfied with their sex life, 94 (58%) were satisfied, and 46 (28.4%) were not. Those who stated they did not knowwhether they were satisfied were excluded from the analysis of satisfaction with sex life in the 2 surgical groups (Table 1). There was no
JUNE 2015 Ameri
difference in satisfaction with sex life between SAH and TAH overall (Table 1) or when subdivided according to partner status (data not shown). The other sec- ondary outcomes are shown in Table 1. None of them showed significant dif- ferences between surgical groups in the analysis of observed data or in multiple imputation.
Hospital contacts from 5 to 14 years after hysterectomy are shown in Figure 2. There was no significant difference in the total number of hospital contacts (SAH, 29 [17.7%] vs TAH, 18 [11.3%]; relative risk, 1.57; 95% confidence interval, 0.91e2.71; P ¼ .10).
COMMENT
On observed data, we found that more women had UI after SAH than after TAH 14 years after surgery. This is consistent with prior results from our trial.14,15 At 14 years, the percentage of UI in the TAH group has reached approximately the prehysterectomy level, whereas the percentage in the SAH
can Journal of Obstetrics & Gynecology 758.e4
Characteristic Participants (n [ 197)
SAH participants (n [ 97)
TAH participants (n [ 100)
Age, y (SD) 60.1 (5.8) 60.5 (6.6) .58 60.7 (5.9) 59.6 (5.6)
Follow-up time, y (range) 14.1 (12e16) 14.04 (12e16) .77 14.2 (12e16) 14.03 (12e16)
Parity (range) 1.8 (0e5) 1.74 (0e5) .58 1.85 (0e5) 1.76 (0e4)
BMI, kg/m2 (SD)a 26.1 (6.7) 25.5 (4.6) .38 26.45 (7.1) 25.71 (6.3)
Indication for hysterectomy, %
Fibroids 115 (58.4) 70 (57.4) .86 58 (59.8) 57 (57.0)
Abnormal uterine bleeding 63 (31.9) 42 (34.7) .59 29 (29.9) 34 (33.6)
Dysmenorrhea 8 (4.1) 4 (3.31) .74 3 (3.1) 5 (4.9)
Pelvic pain 9 (4.6) 4 (3.31) .58 6 (6.2) 3 (2.9)
Endometriosis 0 1 (0.83) .16 0 0
Other 2 (1.02) 1 (0.83) .87 1 (1.03) 1 (0.99)
Type of surgery, %
46 (23.4) 57 (46.7) < .0001 18 (18.6) 28 (28.0)
Alcohol >14 units per week , %b,c
22 (11.2) 6 (4.9) .047 13 (13.4) 9 (9.0)
Chronic disease, %d 97 (49.2) 25 (29.4) .0018 49 (50.5) 48 (48.5)
Preoperative UI, % 48 (25.0%) (n ¼ 192) 20 (17.4%) (n ¼ 115) .110 26 (27.7%) (n ¼ 94) 22 (22.2%) (n ¼ 99)
BMI, body mass index; SAH, subtotal abdominal hysterectomy; TAH, total abdominal hysterectomy; UI, urinary incontinence.
a BMI at follow-up for participants but baseline for nonparticipants; b At time of surgery; c A unit of alcohol, in Denmark, is defined as 12 g of alcohol, which is the approximate content of a normal beer or a glass of wine; d At 14 year follow-up for participants and at 1 year for nonparticipants.
Andersen. SAH vs TAH, RCTwith 14 year follow-up. Am J Obstet Gynecol 2015.
Research Gynecology ajog.org
group is now higher.14 However, the MI analysis did not show a significant dif- ference. No significant differences were found between SAH and TAH on the secondary outcomes.
The strengths of the present results are that they represent the largest random- ized clinical trial on the topic and have the longest follow-up time. The Danish social security number enabled us to locate all participants and look up hos- pital…