Regional practice variation in hysterectomy and the implementation of less invasive surgical procedures: A register-based study in the Netherlands.

Maarten DH Vink, France RM Portrait, and Marlies Y Bongers designed the study. Maarten DH Vink and Eric JE van der Hijden were involved in the compliance procedure to retrieve the data from Vektis. Tim van Wezep analyzed the data, and Xander Koolman provided feedback to the statistical analyses performed. Maarten DH Vink, France RM Portrait, and Wouter JK Hehenkamp drafted the manuscript. All authors have reviewed and approved the final manuscript.

The study was approved by the compliance officers of Vektis and was exempt from the need for institutional review board approval because all data were gathered retrospectively and analyzed anonymously. Vektis merged the results of the COROP regions 31 ( Zeeuwsch‐Vlaanderen ) and 32 ( Overig Zeeland ) to reduce the risk of identifiability, given the limited number of inhabitants in those COROP regions.

We included 6 798 030 women aged ≥25 years from the Netherlands between 2016 and 2020, representing 31 038 375 person‐years. The number of inhabitants differed significantly among the 40 COROP regions. The smallest COROP region included 18 174 women and the largest region 576 371 women. The baseline characteristics of the study population are presented in Table  1 . Baseline characteristics (study period 2016–2020). Abbreviations: COROP, Coordination Commission Regional Research Program; DTC, Diagnosis Treatment Combination. For 45 hysterectomies, two surgical routes were registered. During the study period, 23 007 hysterectomies were performed for uterine fibroids or bleeding disorders, giving a hysterectomy rate of 74 per 100 000 person‐years for these indications in the Netherlands. The average age of women undergoing hysterectomy was approximately 47 years. Among the various surgical approaches, laparoscopic hysterectomy was performed most often (58.9%), followed by abdominal (24.4%) and vaginal (16.9%) hysterectomies. The overall hysterectomy rate was 43–124 per 100 000 person‐years among the different COROP regions. When adjusted for case‐mix, this range narrowed to 49–107 (CV 0.17). Upon further categorization by indication, the total hysterectomy rates were 25–78 for bleeding disorders (AR 33–70; CV 0.21) and 16–48 for uterine fibroids (AR 17–42; CV 0.17). Substantial variation in medical practice was observed across COROP regions, particularly in the surgical approach chosen. The hysterectomy rate per 100 000 person‐years was 23–84 for the laparoscopic approach (AR 26–79, CV 0.24), 9–32 for the abdominal approach (AR 10–31, CV 0.28) and 3–42 for the vaginal approach (AR 5–37, CV 0.65). Figures  1 and 2 , together with Table  2 , present the variations in total hysterectomy rates and the proportional distribution of each surgery by indication (ie bleeding disorders or uterine fibroids) among COROP regions. Medical practice variation in hysterectomy rates for bleeding disorders. (A) Case‐mix adjusted hysterectomy rate per 100 000 person‐years. (B–D) Contribution of each surgical approach to the hysterectomy rate. Medical practice variation in hysterectomy rates for uterine fibroids. (A) Case‐mix adjusted hysterectomy rate per 100 000 person‐years. (B–D) Contribution of each surgical approach to the hysterectomy rate. Variation in hysterectomy rate per COROP region, stratified by surgical approach. Note : The analysis included women with DTC codes recorded for bleeding abnormalities (G11) or fibroids (G15) within this period. Abbreviations: AR, adjusted rate; COROP, Coordination Commission Regional Research Program; CV, coefficient of variation; DTC, Diagnosis Treatment Combination. Between 2016 and 2020, we identified 40 416 women (average age 50 years) and 4895 women (average age 46 years) who underwent therapeutic hysteroscopy for bleeding disorders and uterine fibroids, respectively. At a population level, the therapeutic hysteroscopy rate for bleeding disorders was 130 per 100 000 person‐years and varied from 67 to 220 (AR 69–224; CV 0.20) among COROP regions. For uterine fibroids, the corresponding rate was 16 and varied from 4 to 32 (AR 9–29, CV 0.25) among regions. Regression analyses, conducted at the level of a COROP region to examine associations between ARs for hysterectomy and therapeutic hysteroscopy, showed a positive but nonsignificant relation for both bleeding disorders ( n  = 39, β = 0.14; P  = 0.31) and uterine fibroids ( n  = 39, β = 0.10; P  = 0.34). Sensitivity analysis for the association between hysterectomy and therapeutic hysteroscopy for women with bleeding disorders <50 years suggested a positive but nonsignificant relationship ( n  = 39, β = 0.09; P  = 0.43). Overall, 31 050 women (average age 47 years) underwent endometrial ablation for bleeding disorders. At the population level, 100 procedures per 100 000 person‐years were performed with a variation of 30–238 (AR 25–234, CV 0.33) among COROP regions. Regional hysterectomy rates were positively but nonsignificantly related to endometrial ablation ( n  = 39, β = 0.11; P  = 0.21). Overall, 886 women (average age, 45 years) underwent fibroid embolization and 1222 women (average age 38 years) underwent myomectomy. This equated to three embolization procedures per 100 000 person‐years, with a variation of 1–12 (AR 1–12; CV 1.06) among COROP regions, and four myomectomies per 100 000 person‐years, with a variation of 1–12 (AR 2–8; CV 0.48) among COROP regions. Regional hysterectomy rates for women with fibroids were negatively but nonsignificantly associated with embolization ( n  = 31, β = −0.08; P  = 0.08) and myomectomy ( n  = 31, β = −0.03; P  = 0.82).

We assessed practice variation in the hysterectomy rate stratified by surgical approach and examined the association between this rate and the incidence of less invasive surgical techniques for women with bleeding disorders or fibroids. The literature demonstrates that minimally invasive techniques are effective for mitigating symptoms of vaginal bleeding or fibroids and that this could reduce the number of hysterectomies, thereby also reducing healthcare costs and patient exposure to unnecessary surgical risk. 14 , 23 However, there is less literature on whether the wider implementation of minimally invasive techniques has been associated with a reduction in the number of hysterectomies. Our results show an overall hysterectomy rate for both bleeding disorders and fibroids of 74 per 100 000 person‐years in the Netherlands, ranging from 43 to 124 among COROP regions (AR 49–107; CV 0.17). Vaginal hysterectomy showed higher regional practice variation than either bleeding disorders or fibroids. For women with bleeding disorders, a positive nonsignificant relation was observed between the regional incidence of hysterectomy and therapeutic hysteroscopy (β = 0.14; P  = 0.31) and the endometrial ablation (β = 0.11; P  = 0.21). For women with fibroids, a positive but nonsignificant relation was observed between the regional incidence of hysterectomy and that of therapeutic hysteroscopy (β = 0.10; P  = 0.34). Nonsignificant negative relations were observed with both myomectomy (β = −0.03; P  = 0.82) and embolization (β = −0.08; P  = 0.08). Although none of the coefficients was statistically different from zero, this could reflect our inclusion of data from a limited number of COROP regions. When compared directly with hysterectomy, embolization procedures have previously been established as being cost‐effective in the Dutch setting. A 2008 cost‐effectiveness analysis revealed that only 23% of patients with fibroids underwent hysterectomy within 2 years after embolization and showed that the mean total costs of embolization were significantly lower compared with hysterectomy. 14 However, our results suggest that, at a regional level, an increase of 100 embolization procedures per 100 000 person‐years is associated with just eight fewer hysterectomies performed, after adjusting for case‐mix factors. As such, it appears that embolization is not just used as an alternative to hysterectomy. It is mostly used to serve other patient groups. While our results cannot be used to conclude whether or not embolization is cost‐effective, they do pose the question of whether the strict hysterectomy vs embolization comparison was appropriate, as embolization is mostly used as an alternative to other types of care. Consistent with the relevant literature, the lack of a negative relation between hysterectomy and therapeutic hysteroscopy and endometrial ablation, does not support the existence of a clear substitution effect for therapeutic hysteroscopy or endometrial ablation at a population level. A study from Japan showed an increase in the incidence of hysterectomies between 2014 and 2017 despite a concurrent increase in alternative treatment options. 18 Similarly, no reduction in hysterectomy rates was observed in England after introducing endometrial ablation. 24 This discouraging result could have several causes. First, the presence of minimally invasive options for managing bleeding disorders or health complaints related to fibroids could lead to scenarios where women opt for a surgical intervention sooner than they might have otherwise, particularly if hysterectomy were the sole choice. Second, using minimally invasive therapies might extend to broader indications than those for which their effectiveness has been demonstrated in comparative effectiveness studies. A clear example has been reported for endometrial ablation, which is predominantly offered to women experiencing heavy menstrual bleeding. Research indicates a satisfaction rate exceeding 85% after 2 years for this indication, with a subsequent hysterectomy rate of 19%. 11 However, limited data indicate an effectiveness of only 50% when used for dysmenorrhea. 25 Given that endometrial ablation is used more often for dysmenorrhea, increased use would not result in a substitution effect for hysterectomy. Other examples also exist for therapeutic hysteroscopy. Women of reproductive age who experience bleeding disorders, possibly attributed to an intrauterine polyp, may undergo therapeutic hysteroscopy. However, these women are unlikely to opt for hysterectomy if they want to conceive in the future. Regarding the implementation of less invasive surgical techniques as a substitute for hysterectomy, the age of the patient should play a significant role in counseling. When a young patient with bleeding disorders or symptoms related to fibroids, undergoes a treatment other than hysterectomy (eg therapeutic hysteroscopy), the patient has a long period during which recurrent symptoms may develop. Literature indicates that, when performed for the appropriate indication, hysterectomy is a cost‐effective treatment for relatively young women who are not close to a natural menopause. 26 Another significant finding in our study concerns the vaginal approach to hysterectomy for women with idiopathic bleeding disorders, which only saw limited uptake and notable variability. There has been a decreasing trend in the number of vaginal hysterectomies in many countries, despite clear guidelines recommending this as the preferred option when technically feasible. 27 Vaginal hysterectomy is particularly suitable for patients with complaints of vaginal bleeding not accompanied by uterine enlargement. A plausible explanation for the decreasing use of vaginal hysterectomy among women with idiopathic bleeding disorders could be the greater use of alternatives such as endometrial ablation in everyday clinical practice. Literature also shows that the choice of hysterectomy approach depends on the provider, indicating that the decreasing trend in vaginal hysterectomies could reflect surgeon preference for the laparoscopic approach. 28 A recent survey among Irish residents uncovered a low sense of proficiency with abdominal and vaginal hysterectomy. This trend may also partially contribute to the decline in vaginal hysterectomies compared with the laparoscopic approach. 29 It is also possible that the number of vaginal hysterectomies has declined more in comparison with abdominal and laparoscopic hysterectomies, since the levonorgestrel intrauterine device is now commonly utilized to address bleeding disorders in women without fibroids. Specifically, this group of women with bleeding disorders and no anatomical causes may have undergone vaginal hysterectomies more frequently in the past. Another hypothesis is that the decline in abdominal and laparoscopic hysterectomies is lagging behind that of vaginal hysterectomy because these procedures are often performed in women with uterine fibroids. Our data show that embolization, the alternative treatment for these women, is rarely performed and has relatively large practice variation. The observed regional practice variation in the overall hysterectomy rate can also be explained partly by differences in surgical culture among regions and hospitals. Gynecologists and patients in a specific region might favor surgical intervention instead of more conservative medical treatments. Existing literature underlines that well‐established surgical cultures and practices play a pivotal role in the difference in adherence to surgical guidelines. 30 An important strength of our study is the data completeness. Everyone in the Netherlands is required to have basic health insurance that covers all of the healthcare examined in this study (except for a mandatory deductible). Vektis then receives the claims data from all healthcare insurers, resulting in a nationwide dataset for the Netherlands from 2016 to 2020. However, our study also has some limitations. For example, we did not correct for therapeutic hysteroscopy performed for postmenopausal bleeding due to an intracavitary polyp. Information about the specific indication for hysteroscopy could not be obtained from the claims data. However, we performed a sensitivity analysis for women <50 years of age as a proxy for menopause, which suggested a positive, nonsignificant relations between the hysterectomy rate and therapeutic hysteroscopy (β = 0.09; P  = 0.43). Additionally, we could not perform case‐mix correction at the individual patient level (except for age) and had to rely on an aggregated postal code level. Moreover, given the relatively limited number of COROP regions, there is a potential for a type 2 error in all the analyses that were performed. To mitigate the risk of Type 2 errors, especially in the regression analyses of interventions with low prevalence such as myomectomy and embolization, we opted to merge COROP regions that had fewer than five interventions during our study period and were in close proximity to each other. In the regression analyses comparing myomectomy and embolization vs hysterectomy, this resulted in larger regions with at least five interventions (either embolization or myomectomy). Lastly, our study lacks data on more conservative treatments for women with bleeding disorders or fibroids, such as medical management or the use of the levonorgestrel intrauterine device. While we acknowledge that these therapies may have had a significant impact on reducing the number of hysterectomies in previous decades, well before our study period, we chose not to include these therapies in our analyses. Our study focused specifically on the substitution effect of minimally invasive surgical techniques. Future studies could employ case investigations to examine why minimally invasive techniques are not employed only as substitutes for hysterectomy. Does this reflect a broader scope of indications, or are they being utilized as supplementary methods for patients who do not want hysterectomy? Research could also address the question of why there is a declining trend in the use of vaginal hysterectomy, despite it being recommended as the preferred surgical approach in guidelines.

Major regional variation existed in the hysterectomy rate and surgical approaches used in the Netherlands. The absence of a significant negative relation between the regional rates of minimal invasive techniques and hysterectomy indicates that, contrary to their purpose, minimal invasive techniques do not tend to replace hysterectomy in clinical practice. However, the embolization technique could be an exception. It showed a nonsignificant inverse relationship with hysterectomy, suggesting that greater implementation of embolization could ultimately reduce the number of hysterectomies performed for women with fibroids.

Hysterectomy is among the most common major gynecological surgeries performed worldwide for both malignancy and benign causes. 1 , 2 , 3 Relatively benign indications include uterine bleeding disorders, fibroids, endometriosis and prolapse. Although several approaches are available (eg vaginal, abdominal, laparoscopic, including robot‐assisted and vaginal natural orifice transluminal endoscopic surgery), guidelines usually recommend the vaginal approach for women with bleeding disorders and no anatomical cause. This is because it remains the most cost‐efficient technique with the lowest risk of complications for relatively simpler hysterectomies that are not complicated by fibroids, endometriosis or extensive intra‐abdominal adhesions. 4 , 5 , 6 Hysterectomy rates have declined over recent decades in several Western countries. 7 A Finish study that stratified hysterectomy by indication reported a decreasing incidence for benign diseases in particular. 8 This might reflect the introduction of new minimally invasive techniques, such as uterine artery embolization, endometrial ablation or therapeutic hysteroscopy. For example, studies into the effectiveness of uterine artery embolization, introduced in 1995, have demonstrated its ability to prevent hysterectomy in 75% of women with uterine fibroids during the initial 2 years after treatment, decreasing to 65% by 10 years. 9 , 10 Furthermore, in women with idiopathic heavy menstrual bleeding, endometrial ablation has a satisfaction rate exceeding 85% after 2 years and a subsequent hysterectomy rate of 19% from 2 to 5 years. 11 Practice variation in hysterectomy rates has been of interest for decades, with literature confirming that variation exists both between and within countries. 12 , 13 This is important because excessive reliance on hysterectomy leads to avoidable healthcare costs and surgical risks. 14 Several patient and physician factors could explain the variation in hysterectomy rates, at least partly. Evidence suggests that the hysterectomy rate is positively correlated with age and socioeconomic factors (eg insurance status), whereas results about the influence of physician factors are contradictory. 15 Although some studies show a positive association with the density of gynecologists, a recent Swiss study could not confirm this finding. 16 , 17 Indeed, both the total hysterectomy rate and the chosen surgical approach (vaginal, abdominal or laparoscopic) appear prone to practice variation. 17 Given the effectiveness of minimally invasive techniques, greater adoption should reduce the hysterectomy rate. However, a cohort study has indicated that implementing minimally invasive techniques may not have this effect. 18 In this study of women with bleeding disorders and fibroids in the Netherlands, we aim to evaluate whether there has been a substitution effect from hysterectomy to minimal invasive techniques. Specifically, we study the regional variation in hysterectomy rate and approach, as well as the relation between rates of hysterectomy and minimally invasive surgical procedures.

The authors have stated explicitly that there are no conflicts of interest in connection with this article.

All Dutch residents have a compulsory and comprehensive healthcare package. Most importantly, this package includes specialist care, primary care, pharmaceuticals and, as such, all care for women diagnosed with bleeding disorders or fibroids. All individuals older than 18 are subject to a front end deductible of 385 € per year. General Practitioners (GPs) can refer patients with these disorders to one of the 69 hospitals in the Netherlands or to an independent healthcare clinic. In 2021, independent clinics were only responsible for about 5.5% of the total healthcare expenditure in hospitals and clinics combined. 19 Care provided by these clinics is also covered by the compulsory health insurance system. Our study incorporates invoice data from all Dutch hospitals and independent clinics. Not all hospitals provide every treatment; for example, embolization is only offered in approximately 30% of hospitals, whereas all hospitals offer laparoscopic hysterectomies. By contrast, independent clinics generally do not offer any invasive procedures, including hysterectomy or embolization. These patients may be referred to a hospital nearby providing these services. However, there is no centralized coordination or agreements regarding referral pathways in the Netherlands. In each region, one or several hospitals or clinics offer (surgical) care for bleeding disorders or fibroids and there is no central hospital with overall responsibility. This means that, apart from variations in the accessibility of the embolization procedure among hospitals, there are no differences across regions in technical capability to execute the three hysterectomy approaches. Healthcare reimbursement is facilitated through the Diagnosis Treatment Combination (DTC) system, a variant of the globally employed Diagnosis Related Groups (DRG) system. DTC codes are predicated on the specific health concern for which a patient is referred, and each code encompasses details regarding any medical procedures undertaken during hospitalization (eg diagnostic assessments and surgical interventions). Vektis functions as the operational agency for “ Zorgverzekeraars Nederland ”, an overarching consortium of 11 healthcare insurers in the Netherlands, and it has access to an extensive repository of claims data. 20 For the present comprehensive nationwide investigation, we procured data from Vektis for the period from 2016 to 2020. In 2013, the Dutch national guideline on “Heavy Menstrual Bleeding” was published, and it has since undergone several updates. Currently, the guideline provides recommendations for the care of women with heavy menstrual bleeding, both with and without fibroids, as included in our study. 21 A stepped‐care approach is recommended for women experiencing heavy menstrual bleeding without an identifiable anatomical cause, starting with conservative management. This should start with medical interventions or the use of a levonorgestrel intrauterine device; where these prove ineffective, the guideline suggests considering endometrial ablation before hysterectomy. The same applies to women with fibroids, although the guideline recommends considering an embolization procedure as an alternative to a hysterectomy. Following the publication of this guideline, the annual number of hysterectomies for bleeding disorders and fibroids in the Netherlands decreased from 6121 in 2013 to 3939 in 2020. 22 Women aged 25 years or older at study inclusion were enrolled and followed until December 31, 2020. This study assessed medical practice variation across different regions according to the COROP system. The analysis included women with DTC codes recorded for bleeding disorders (G11) or fibroids (G15) within this period. Claims data were also used to obtain information about the age and geographic region of each woman, as well as any surgical procedures they underwent. We identified specific surgical procedures within DTC codes G11 and G15 based on corresponding healthcare activity codes: abdominal hysterectomy (037111), laparoscopic hysterectomy (037113), vaginal hysterectomy (037131) and therapeutic hysteroscopy (037170 and 037175). However, endometrial ablation (037177) was only identified for cases falling under DTC G11, and myomectomy (037161) and uterine artery embolization (080828) were only identified within DTC G15. All women with DTCs recorded for gynecologic malignancies, endometriosis or prolapse during the study period were excluded from the dataset. First, hysterectomy rates were calculated as the number of hysterectomies per 100 000 person‐years at the national level and for each COROP region. Hysterectomy cases were also categorized into vaginal, laparoscopic or abdominal surgical approaches. To understand the contribution of each surgical approach to the overall number of hysterectomies, we divided the number of hysterectomies performed using a specific surgical method by the total number of hysterectomies at the national level and for each COROP region. The analysis was repeated for women with bleeding disorders (DTC G11) and for women with uterine fibroids (DTC G15). Second, for women with bleeding disorders (DTC G11), the crude rates per 100 000 person‐years were calculated separately for endometrial ablation and therapeutic hysteroscopy. For the women with uterine fibroids (DTC G15), the rates were calculated separately for embolization, myomectomy and therapeutic hysteroscopy. All crude rates were reported at the national level and for each COROP region. If a woman underwent hysterectomy or therapeutic hysteroscopy during the study period, and both DTC G11 and DTC G15 were documented, we classified the procedure as being performed for fibroids. We believe that if at least one healthcare professional has diagnosed a woman as having fibroids, this most accurately aligns with the actual clinical scenario. Third, an adjusted rate (AR) was derived from the crude rate calculated for each COROP region across all medical interventions. This adjustment was achieved by using the case‐mix variables age, socioeconomic status and ethnicity for the year 2018. Data were sourced from Statistics Netherlands at the level of postal code. The variable “household income” was used to approximate socioeconomic status, and the variable “% non‐Western immigrants” was used to approximate ethnicity. To estimate the extent of regional practice variation for all surgical procedures and both indications, the coefficient of variation (CV) between COROP regions was calculated as a descriptive measure of heterogeneity. For this, we divided the standard deviation by the case‐mix adjusted mean outcome per COROP region. A higher CV indicated greater relative variability (more practice variation) and a lower CV suggested less relative variability. Finally, the associations between the rates for total hysterectomy and other surgical procedures were assessed by regression analysis. The COROP case‐mix ARs for less invasive procedures were regressed on the COROP case‐mix ARs for hysterectomy. For bleeding disorders, we assessed the associations between regional hysterectomy rates and both endometrial ablation and therapeutic hysteroscopy rates. For uterine fibroids, we assessed the associations between regional hysterectomy rates and the embolization, myomectomy and therapeutic hysteroscopy rates. We could not exclude patients for whom a therapeutic hysteroscopy was performed for post‐menopausal bleeding in DTC G11. Therefore, a sensitivity analysis was performed for the relationship between hysterectomy and therapeutic hysteroscopy for women <50 years of age. We merged regions with fewer than five interventions during our observation period if they were in close proximity to each other, creating new, larger regions with at least five interventions. This merging process was exclusively conducted for individual regression analyses involving interventions with fewer than five interventions per region during our study period. SAS (version 9.4; SAS Institute Inc., Cary, NC, USA) was used for data storage and analysis.