Hysterectomy and risk of stress-urinary-incontinence surgery: nationwide cohort study
Hysterectomy for benign indications has been associated with an increased risk for lower-urinary-tract sequela, but results have been inconclusive. We aimed to establish the risk for stress-urinary-incontinence surgery after hysterectomy for benign indications.
Hysterectomy is a preferred treatment option in women because of its low perioperative morbidity and effectiveness in bringing a definitive cure to disorders such as menometrorrhagia, leiomyoma, uterine prolapse, adenomyosis, and postmenopausal bleeding. By age 55 years, about one in five British women will have undergone a hysterectomy. In the USA, 600 000 procedures are undertaken every year, of which 90% are done for benign indications.
Large variations exist in rates of hysterectomy between and within countries and between populations with similar characteristics. and Although health-economic factors and differences in morbidity can account for some of the variation, rates of hysterectomy for benign indications are largely affected by differences in medical practice and attitudes and Considering the generally high hysterectomy rates in premenopausal and perimenopausal women, increased knowledge of the long-term outcomes associated with undertaking hysterectomy for benign indications is essential.
For decades, the effects of hysterectomy on lower-urinary-tract function have been controversial. One theory suggests that hysterectomy could initiate stress urinary incontinence by interruption of the local nerve supply to the urethra, and the procedure might cause changes in urethral pressure dynamics by distortion of pelvic-organ anatomy. Results from many previous studies relating hysterectomy to adverse effects of the lower urinary tract have been inconsistent. The conflicting evidence could be attributed to methodological shortcomings including inadequate statistical precision owing to small sample sizes, scarcity of controls who have not had hysterectomy, short durations of follow up, and systematic errors, such as ascertainment, selection, or recall bias.
Female stress urinary incontinence, defined as involuntary leakage of urine on effort or exertion, has a population-based prevalence of nearly 40% in most industrialised countries, usually with severe implications for daily function, social interactions, sexuality, and psychological wellbeing. Stress urinary incontinence also has a major effect on health economy and is increasingly recognised as a global health concern. Identification and possible reduction of risk factors for stress urinary incontinence is, therefore, of importance for women at risk and for society’s direct health-care costs. The primary aim of this nationwide, population-based, cohort study was to assess short-term and long-term risk for stress-urinary-incontinence surgery after hysterectomy for benign indications.
The Swedish Inpatient Registry contains data for individual hospital discharges. The register was established in 1964; in 1973, it covered more than 60% of the population, in 1983 coverage was 85%, and from 1987 complete national coverage was achieved. Every inpatient discharge record contains: (1) dates of hospital admission and discharge; (2) up to eight discharge diagnoses, coded according to the International Classification of Diseases (ICD-7 until 1968, ICD-8 from 1968 to 1986, ICD-9 from 1987 to 1996, and ICD-10 thereafter); and (3) up to 12 operation codes from the Swedish Classification of Operations and Major Procedures. The Swedish Inpatient Registry also includes information about the unique national registration number assigned to all Swedish residents, allowing unambiguous record linkage across all national registers in Sweden. Correct coding for surgical procedures is achieved in 98% of records in the Registry, with a less than 1% yearly loss to registration.
We identified all records in the Swedish Inpatient Registry indicating hysterectomy from Jan 1, 1973, until Dec 31, 2003 (exposed cohort). For every patient undergoing hysterectomy, we randomly selected three controls from the register of the total population, who did not have hysterectomy and who were individually matched by year of birth and county of residence at year of hysterectomy (unexposed cohort). Using national registration numbers, our exposed and unexposed cohorts (ie, women who had and had not undergone hysterectomy) were linked to other nationwide Swedish registers, including the census of 1990, medical birth register, cancer register, emigration register, and cause of death register.
Within the Swedish Inpatient Registry, we identified all women in the exposed and unexposed cohorts having an operation because of stress urinary incontinence. Those undergoing hysterectomy were excluded from further analyses if stress-urinary-incontinence surgery preceded the hysterectomy, if hysterectomy was done for malignant indications, or if the woman appeared in the national cancer register at the time of hysterectomy. To address potential confounding by pelvic organ prolapse, we excluded women undergoing hysterectomy because of uterine prolapse or who had concurrent prolapse surgery at the time of hysterectomy throughout the main analyses, and we present data for this cohort separately.
Women undergoing hysterectomy contributed person-time to the study until first occurrence of any cancer, death, emigration, or end of the observation period (Dec 31, 2003). Controls contributed person-time by the same criteria but were excluded from the unexposed cohort at the time of hysterectomy, at which time they entered the exposed cohort unless the procedure was done for malignant indications.
We used specific codes from the Swedish Classification of Operations and Major Procedures for 1973–1996 and 1997–2003 to identify various surgical procedures in the Swedish Inpatient Registry. Hysterectomy was classified into six different categories: abdominal total hysterectomy (operation codes 7210 and 7220 for 1973–1996 and LCD00 for 1997–2003), abdominal subtotal hysterectomy (7211 for 1973–1996 and LCC10 for 1997–2003), vaginal hysterectomy (7261 and 7262 for 1973–1996 and LCD10 for 1997–2003), laparoscopy-assisted vaginal hysterectomy (LCD11 for 1997–2003), laparoscopic total hysterectomy (LCD01–04 for 1997–2003), and hysterectomy with concurrent pelvic-organ-prolapse repair (7467 for 1973–1996 and LEF13 for 1997–2003).
Our primary outcome measure was surgical procedures for stress urinary incontinence. These included: Kelly sutures, Kennedy sutures, abdominal and laparoscopic Burch colposuspensions, Stamey procedures, Marshall-Marchetti-Krantz procedures, Ingelman-Sundberg plasty, intravaginal slingplasty, suburethral slings, and tension-free vaginal tapes (operation codes 6355, 6356, 6358, 7470, 7471 for 1973–1996, and LEG00, LEG10, LEG20, and KDG10–40 for 1997–2003). As a potential source of confounding, pelvic-organ-prolapse surgery was categorised as: anterior repair, posterior repair, Manchester procedure, abdominal sacrocolpopexy, vaginal sacrospinous fixation, abdominal and vaginal enterocele obliteration, and colpocleisis (operation codes 7120, 7121, 7460–7464, 7466, 7469, and 7541 for 1973–1996 and LEF00, LEF03, LEF10–50, LEF53, LEF96 for 1997–2003).
We derived age at hysterectomy from the Swedish Inpatient Registry and information about parity and mode of childbirth from the medical birth registry, which has recorded data for all deliveries in Sweden since 1973. From the census of 1990 we obtained information on highest level of education, classified as compulsory school, high school, or college-university education.
We calculated crude and age-specific rates for stress-urinary-incontinence surgery as the number of surgical procedures per 100 000 person-years, with 95% CIs based on the Poisson distribution. The risk for incontinence surgery related to hysterectomy was estimated with Cox’s proportional-hazards model, conditioned on the matching variables age, county, and calendar time. Effect modifications of age at surgery (in population quartiles) and time since surgery (in 5-year bands) were measured by stratification. We compared the effect of various hysterectomy operative techniques on subsequent risk for incontinence surgery by Cox’s regression, stratified by time from surgery in 5-year intervals.
By multivariate analyses, we assessed the effects of number of vaginal deliveries (as time-dependent covariates for controls) and educational level in a subsample of women. Information for number of childbirths and mode of delivery was only available from 1973 onwards—ie, after introduction of the population-based Swedish medical birth registry. Thus, we restricted these analyses to women born in 1952 or later. We categorised people with no vaginal births reported in the birth register as nulliparous. In this restricted subpopulation, information about highest level of education was available from the census of 1990 in 92% of women.
To investigate whether women in the exposed cohort might be inclined to also seek elective surgery for other benign disorders, we analysed risks for hallux valgus surgery (osteotomy of the big toe) and stripping of varicose veins. These highly elective surgical procedures were chosen since a biological relation with the pathways leading to hysterectomy and stress-urinary-incontinence surgery is implausible.
We present the results of the different Cox’s regression analyses as hazard ratios with 95% CIs. Statistical analyses were undertaken with SAS software (version 9.1; Cary, NC, USA). The study was approved by the research ethics committee at Karolinska Institutet, Stockholm, Sweden.
Daniel Altman, Fredrik Granath, Sven Cnattingius och Christian Falconer
Hysterectomy and risk of stress urinary incontinence surgery: nationwide cohort study
The Lancet, 27 oktober 2007, ref 370: 1494-1499.
Altman D, Granath F, Cnattingius S, Falconer C.