SEARCH

SEARCH BY CITATION

Keywords:

  • Epidemiology;
  • hysterectomy;
  • morbidity;
  • rates;
  • record linkage

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion and conclusion
  7. Acknowledgements
  8. References

Objective  To investigate incidence trends and demographic, social and health factors associated with the rate of hysterectomy and morbidity outcomes in Western Australia and compare these with international studies.

Design  Population-based retrospective cohort study.

Setting  All hospitals in Western Australia where hysterectomies were performed from 1981 to 2003.

Population  All women aged 20 years or older who underwent a hysterectomy.

Methods  Statistical analysis of record-linked administrative health data.

Main outcome measures  Rates, rate ratios and odds ratios for incidence measures and length of stay in hospital and odds ratios for morbidity measures.

Results  The age-standardised rate of hysterectomy adjusted for the underlying prevalence of hysterectomy decreased 23% from 6.6 per 1000 woman-years (95% CI 6.4–6.9) in 1981 to 4.8 per 1000 woman-years (95% CI 4.6–4.9) in 2003. Lifetime risk of hysterectomy was estimated as 35%. In 2003, 40% of hysterectomies were abdominal. The rate of hysterectomy to treat menstrual disorders fell from 4 per 1000 woman-years in 1981 to 1 per 1000 woman-years in 1993 and has since stabilised. Low socio-economic status, having only public health insurance, nonindigenous status and living in rural or remote areas were associated with increased risk of having a hysterectomy for menstrual disorders. Indigenous women had higher rates of hysterectomy to treat gynaecological cancers compared with nonindigenous women, particularly in rural areas. The odds of a serious complication were 20% lower for vaginal hysterectomies compared with abdominal procedures.

Conclusion  Western Australia has one of the highest hysterectomy rates in the world, although proportionally, significantly fewer abdominal hysterectomies are performed than in most countries.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion and conclusion
  7. Acknowledgements
  8. References

Hysterectomy is one of the most common surgical procedures performed on women in Western countries.1 Concern over the perceived excessive use of hysterectomy to treat benign conditions, particularly dysfunctional uterine bleeding, has led to the introduction of alternative nonsurgical treatments and less invasive surgical procedures.2,3 Evidence of shorter hospital stays and/or lower relative costs of vaginal hysterectomy and laparoscopic-assisted vaginal hysterectomy (LAVH) compared with abdominal hysterectomy has also influenced patterns of surgical care.4–6 Evidence supports the use of vaginal hysterectomy as the method of choice where possible; yet, abdominal hysterectomy remains the predominant procedure in many countries.1,7 Factors associated with increased risk of hysterectomy include increasing parity, poor health, high body mass index, smoking, lower socio-economic status, gynaecologist’s length of time away from training and use of estrogen replacement therapy.8–13

Rates of hysterectomy vary considerably between and within countries;14 however, directly comparing incidence rates between studies can be problematic. Methodologies differ by the age ranges included, whether crude or age-standardised, exclusion of hysterectomy for malignant conditions, and studies that do not account for the proportion of women who have already had a hysterectomy, will underestimate the true rate.

The aim of this population-based study was to use record-linked health data to identify and compare the trends in the incidence rate of hysterectomy in Western Australia from 1981 to 2003 with international reports. Population denominators were adjusted for the prevalence of hysterectomy. Social and demographic factors associated with undergoing hysterectomy were also identified. Western Australia is the largest state in Australia, with an area of 2.5 million km2, equivalent in size to Western Europe, and with a population of 2 million inhabitants, the majority of who live in Perth, the only major metropolitan city. The indigenous population comprised Aboriginal and/or Torres Strait Islanders who represent 3.5% of the state population and who disproportionately reside in rural and remote areas compared with the nonindigenous population (Australian Bureau of Statistics [ABS]).

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion and conclusion
  7. Acknowledgements
  8. References

Study data were obtained through the Western Australia Safety and Quality of Surgical Care Project15 from the Western Australian Data Linkage System. A deidentified extraction of all hospital morbidity records belonging to women who underwent a hysterectomy from 1981 to 2003 with associated detailed demographic, diagnostic and procedural data was performed. International Classification of Diseases (ICD) procedure and diagnosis codes were used to classify hysterectomies, to identify the diagnoses associated with each procedure and to identify serious complications recorded during the hysterectomy admission (haemorrhage, puncture or laceration during procedure, deep vein thrombosis, pulmonary embolism and injury to pelvic organs).

A hysterectomy was classified as being abdominal, vaginal or LAVH. Prior to 1998, ICD codes did not distinguish between vaginal and LAVH, therefore, if a vaginal hysterectomy was accompanied by a separate code for laparoscopy in the same record, it was considered to be laparoscopic-assisted. Reasons for hysterectomy were identified by the primary or secondary ICD diagnosis codes associated with the hysterectomy admission and were divided into seven categories; menstrual disorders, genital prolapse, fibroids, endometriosis, cancer, pelvic inflammatory disease and pain and other conditions combined category, which included obstetric-related procedures. Unique ICD codes for endometrial ablation were not available prior to 1997, so the extent of information on this procedure was limited.

Residential location was defined as metropolitan Perth, rural or remote and hospital type as either private or public. Insurance status was classified as public, private and self-insured. Population census collector’s districts or postcodes were used to assign one of four socio-economic status levels using the Socio-economic Index of Disadvantage compiled by the ABS.

Population denominators were adjusted for the number of women who had already had a hysterectomy and were consequently no longer at risk of another. This was performed by back projection to estimate the cumulative probabilities of retaining an intact uterus by age group, birth cohort and calendar period as described.16 Reported rates were directly age-standardised using hysterectomy-adjusted population denominators and the Australian 2001 population as weights for women aged 20 years to 85+ years unless otherwise stated. For comparison of other published hysterectomy rates, Western Australia equivalent rates were calculated using the reported age ranges, calendar years, diagnostic inclusions, admission type and standard population weights unless otherwise stated.

Multivariate Poisson regression models using hysterectomy-adjusted population denominators were used to estimate the average yearly changes in hysterectomy rates and to identify the relative rates of hysterectomy by residential location and indigenous status, after adjusting for age and birth cohort effects. Multivariate logistic regression models were used to identify factors associated with having a hysterectomy to treat menstrual disorders compared with all other diagnoses after adjusting for age, year and birth cohort effect. Similarly, logistic regression was used to estimate the relative odds of a serious complication by type of hysterectomy after adjusting for age, calendar period, associated diagnosis, type and location of hospital, obstetric admission and co-morbidity. Statistical analyses were performed using STATA Version 8 (StataCorp, College Station, TX, USA).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion and conclusion
  7. Acknowledgements
  8. References

There were 83 068 hysterectomies performed on women aged 20 years and older. The average age at hysterectomy increased from 41.7 years in 1981 to 48.0 years in 2003 (P < 0.001). Women living in rural and remote areas had their hysterectomies at 0.83 and 4.1 years younger, respectively, than women living in metropolitan areas (P < 0.001). The rate of hysterectomy decreased 23% over the study period (Figure 1). By 2003, the crude hysterectomy rate was 5.4 per 1000 women, the age-standardised rate was 3.9 per 1000 women and the age-standardised rate adjusted for the underlying prevalence of hysterectomy in the community was 4.8 per 1000 women. This adjustment increased incidence rate estimates by 20–23%. A comparison of the equivalent hysterectomy rate in Western Australia with reported international and Australian rates is shown in Figure 2.

image

Figure 1. Age-standardised rate of hysterectomy in women aged 20 years and older from 1981 to 2003 overall and by type of procedure using hysterectomy-adjusted population as denominators. (–), all hysterectomies; (•), LAVH; (▴) abdominal hysterectomy and (□), vaginal hysterectomy.

Download figure to PowerPoint

image

Figure 2. Comparison of published hysterectomy rates with derived Western Australian equivalent calculated using the same calendar period, age ranges and age-standardised population weights as reported in each study. 1, excludes pelvic evisceration and radical hysterectomy; 2, European standard population was used as weights; 3, estimated from figure 3 in cited reference; 4, calculated from reported crude numbers to include hysterectomy for malignant conditions; 5, Canadian 1991 population was used as weights. NSW (Aust), New South Wales (Australia); WA, Western Australia.

Download figure to PowerPoint

The abdominal hysterectomy rate decreased by an average of 3.7% (95% CI 3.6–3.8) each year from 1981 to 2003, whereas the rate of vaginal hysterectomy increased by an average of 1.4% (95% CI 1.2–1.6) each year and LAVH increased by an average of 13.3% (95% CI 12.4–14.2) each year since its introduction in the early 1990s (Figure 1). By 2003, 45% of all hysterectomies performed were vaginal, 40% abdominal and 15% were LAVH. The relative proportion of abdominal hysterectomies performed compared with vaginal and LAVH in Western Australia is equivalently compared with international reports in Figure 3.

image

Figure 3. Proportion of hysterectomies performed by the abdominal route in Western Australia in comparison with reported international and national studies calculated using the same age ranges and calendar periods as reported in each comparison study. NSW (Aust), New South Wales (Australia); WA, Western Australia.

Download figure to PowerPoint

The rate of hysterectomy to treat menstrual disorders fell four-fold from 4.1 per 1000 woman-years in 1981 to 0.9 per 1000 woman-years in 1992 where it has since remained stable. The drop in menstrual disorder-related hysterectomies was partly offset by an average yearly increase of 6.4% (95% CI 6.2–6.7) in the rate of hysterectomy to treat fibroids, while the rate of hysterectomy to treat genital prolapse decreased by an average of 1.8% (95% CI 1.5–2.1) each year (Figure 4). In 2003, most hysterectomies were performed for fibroids (27%), followed by genital prolapse (23%), menstrual disorders (21%), endometriosis (12%), cancer (8%), pelvic inflammatory disease (3%) and all other conditions (8%).

image

Figure 4. Age-standardised rate of hysterectomy in women aged 20 years and older from 1981 to 2003 associated with the three most common diagnostic codes calculated using hysterectomy-adjusted population as denominators. (•), menstrual disorders; (▴), uterine fibroids and (□), uterine prolapse.

Download figure to PowerPoint

The rate of hysterectomy to treat menstrual disorders was significantly higher for women who lived in rural areas (Rate ratios [RR] 1.26, 95% CI 1.22–1.30) and remote areas (RR 1.17, 95% CI 1.11–1.23) compared with those who lived in metropolitan Perth but only for nonindigenous women. After adjusting for age, birth cohort effects and calendar period, other socio-demographic factors associated with increased odds of hysterectomy to treat menstrual disorders included nonindigenous status (OR 1.38, 95% CI 1.19–1.59), lower socio-economic status compared with highest socio-economic status (OR 1.24, 95% CI 1.18–1.30) and having surgery in a private hospital (OR 1.22, 95% CI 1.16–1.28).

Indigenous women had half the overall rates of hysterectomy compared with nonindigenous women, independent of where they lived (Figure 5). Yet, the rate of hysterectomy to treat gynaecological cancer was higher in indigenous women compared with nonindigenous women, reaching statistical significance in rural areas. For indigenous women, the rates of hysterectomy to treat menstrual disorders, genital prolapse and endometriosis were considerably lower than nonindigenous women, whereas similar rates to treat pelvic inflammatory disease were observed.

image

Figure 5. Relative rate of hysterectomy in women aged 20 years and older by residential location and indigenous status compared with nonindigenous women living in metropolitan areas. RR and 95% CI (capped bars) from Poisson regression models that are simultaneously adjusted for age at hysterectomy, calendar year and birth cohort effects, with nonindigenous women living in metropolitan areas set as the reference group. Indigenous women (open symbols), nonindigenous women (closed symbols), metropolitan Perth (triangle symbols), rural areas (circle symbols) and remote areas (square symbols).

Download figure to PowerPoint

There were 1818 (2.2%) hysterectomy admissions associated with at least one serious procedural-related adverse event or complication as identified through hospital morbidity administrative records (Table 1), with more complications reported in more recent calendar periods as coding guidelines and standards improved. Haemorrhage and accidental puncture and/or laceration were the most common adverse events associated with hysterectomy in this study. The odds of a complication was reduced in women undergoing a vaginal hysterectomy compared with abdominal hysterectomy (OR 0.8, 95% CI 0.7–0.9) after adjustment for potential confounders. There was no significant difference in the odds of serious complication between abdominal and LAVH in this study (OR 1.0, 95% CI 0.8–1.2). The average length of stay in hospital for hysterectomy decreased from 11.5 (SD ± 13.3) days in 1981 to 5.0 (SD ± 3.7) days in 2003. LAVH resulted in significantly shorter hospital stays of 4.5 days, compared with 5.5 days for vaginal hysterectomy (P < 0.001) and 7.4 days for abdominal hysterectomies (P < 0.001) when analysed from 1990 onwards.

Table 1.  Number and percentage of women who had a serious complication or adverse event recorded during their hysterectomy admission by the type of procedure undergone
Complication identified by ICD codes at index admissionNo. of cases (%)
AbdominalVaginalLAVH
Haemorrhage complicating a procedure772 (1.6)391 (1.3)78 (1.9)
Accidental puncture or laceration during a procedure298 (0.6)135 (0.44)46 (1.1)
Deep vein thrombosis46 (0.1)11 (0.04)4 (0.1)
Pulmonary embolism36 (0.07)10 (0.03)4 (0.1)
Injury to pelvic organs16 (0.03)5 (0.02)3 (0.07)
Vascular complications following procedure11 (0.02)6 (0.02)0
Total1135 (2.4)553 (1.8)130 (3.1)

Discussion and conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion and conclusion
  7. Acknowledgements
  8. References

Western Australia has one of the highest reported hysterectomy rates in the world—similar to that reported for the USA.17 Australian hospital statistics for 2002–03 show that Western Australia also had the highest age-adjusted separation rate for hysterectomy of all Australian states.18 These high rates are likely due to a combination of patient perceptions and demands and clinician preferences within the context of a health system that includes both private- and public-funded insurance and hospitals.

Adjusting for the prevalence of intact uteri in Western Australia increased the estimated rate of hysterectomy by 20–23%. This adjustment was essential for obtaining a more accurate estimate of the true rate, particularly in countries like Australia that have a history of high rates. In a US study, prevalence adjustment increased the estimate of lifetime risk of hysterectomy from 35 to 43%.19 This adjustment has important implications for health services planning such as cervical cancer screening services.

The decrease in the rate of hysterectomies performed via the abdominal route in this study was marked and coincided with the introduction and increasing use of LAVH in the early 1990s together with a concurrent increase in the rate of vaginal hysterectomies. By 2003, only 40% of hysterectomies were performed abdominally in Western Australia, which is consistently lower than reported elsewhere. These changes in practice occurred around the time of publication of results from randomised clinical trials in the mid-1990s that demonstrated better outcomes in women undergoing LAVH or vaginal hysterectomy compared with abdominal hysterectomy.7 This suggests willingness on the part of gynaecological surgeons to use new techniques and keep abreast of evidence-based practice.

The decline in hysterectomies to treat menstrual disorders in the 1980s may have been in response to the concern over the high rates of discretionary surgery and the increasing hospital admission rates in the late 1970s in Australia, a topic given wide media coverage.20,21 Despite the introduction of endometrial ablation into Australia in the late 1980s, the rate of menstrual-related hysterectomy has remained stable. A recent UK study reported a sustained decline in the number of hysterectomies to treat menstrual disorders attributed to the Mirena® (Berlex, Montville, NJ, USA) levonorgestrel intrauterine system.22 The Mirena intrauterine device was approved for use in Australia in 2000 and it will be of interest to observe the extent of its impact in the Australian context with its large private health sector complementing the universal public health system.

We found that the increased risk of hysterectomy in rural and remote areas was the result of menstrual disorders and, to a lesser degree, pelvic inflammatory disease, particularly for nonindigenous women. Reduced ease of access to alternative therapies such as regular iron infusions in the case of heavy menstrual blood loss may influence the treatment decision for women in more rural or remote areas. Indigenous women were found to have a different hysterectomy profile and appear to have a higher ‘threshold’ for undergoing the procedure, reserving it for more serious reasons such as cancer and pelvic inflammatory disease and less for menstrual-related disorders. The rate of cervical cancer in indigenous women is up to three times that of nonindigenous women23 which may translate into an increased rate of hysterectomy. To account for the increased rate of cervical cancer, one might expect the hysterectomy rate in indigenous women to be at least three times higher than the nonindigenous population; however, there is evidence that indigenous women are more likely to present with advanced disease24 when perhaps hysterectomy is no longer a curative option. Together with the choices that indigenous women may make for cultural reasons, particularly if treatment requires travelling large distances, the differences in the rate of hysterectomy compared with nonindigenous women is indicative of inequitable access to health care.

A recent meta-analysis found better short-term outcomes in women who had vaginal hysterectomies compared with abdominal procedures,7 which supports the shorter length of hospital stay and reduced odds of complication found in this study. This is in contrast to a Finnish study that found in-hospital rates of complication were higher for vaginal hysterectomies compared with abdominal hysterectomy.25 However, the high rate of vaginal hysterectomy in Western Australia implies a level of general expertise in the procedure that may not be found in countries where the abdominal route is still favoured.

This study was limited by the use of administrative data that were not collected specifically for research. Nonspecific coding affected reliability of estimates such as the impact of endometrial ablation on the rates of hysterectomy, the type and extent of laparoscopy used during hysterectomy procedures and the temporal pattern of procedural complications. It is also unlikely that the reported diagnosis-related rate changes can be totally explained by real variations in disease burden. Continual efforts to improve data quality over the study period influenced coding patterns so that earlier nonspecific symptomatic coding such as abnormal bleeding would have been replaced by more specific diagnostic codes, specific procedural codes were introduced and procedural complications were recorded more readily.

In conclusion, Western Australia has one of the highest hysterectomy rates in the world but performs proportionally few abdominal hysterectomies compared with most countries. There were large variations in the rates of hysterectomy to treat underlying conditions over time, a likely response to concern regarding over servicing. Indigenous women and women living in rural and remote areas have a different hysterectomy profile compared with nonindigenous women living in metropolitan areas that may reflect issues relating to culture and access of care.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion and conclusion
  7. Acknowledgements
  8. References

We thank the Western Australia Data Linkage Unit for supplying the data analysed in this study and the National Health and Medical Research Council for financial support of the Western Australia Safety and Quality of Surgical Care Project.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion and conclusion
  7. Acknowledgements
  8. References