Introduction

Globally, hysterectomy is the most frequent gynaecological surgery1. In Germany, the prevalence of hysterectomy between 2008 and 2011 in 18–79-year-olds was 18%2. In 2021 alone, an estimated 90,000 inpatient hysterectomies were performed3, making it the third most common inpatient gynaecological surgery in the country. Nearly 90% of indications for hysterectomy are for benign conditions4,5. The most frequent being uterine fibroids, representing 60% of all in-patient hysterectomies performed in 20124, followed by abnormal uterine bleeding and pelvic prolapse. Of the malignant indications, endometrial cancer ranked first.

Over the past decades, there has been a proliferation of non-surgical treatments for benign gynaecological diseases6,7,8,9,10. Coincidentally, the rates of hysterectomy are reportedly decreasing in several countries. In analysing trends from 2006 to 2012 in Germany, a similar declining pattern was found, indicating a shift towards more conservative procedures11. However, when examining subtypes, the subtotal hysterectomy rates had increased.

The age-period-cohort analysis is a tool for analysing trends of population-level data aiming at disentangling the effects of age, period and cohort on the changes in the trends12. Age effects are those due to the intrinsic biological process of aging; period effects reflect changes associated with time, affecting all age groups simultaneously. Cohort effects are limited to particular cohorts because of specific exposures at a point in time13. Although commonly used in cancer studies, there has been recent use of this approach for diagnostic procedures as well14,15.

The objective of this study was to examine the age, period and cohort effects on trends of overall hysterectomies and subtypes in Germany based on nationwide Diagnoses Related Group data from 2005 to 2019.

Materials and methods

Using an ecological study design, population-level data was used to examine the trends of hysterectomy incidence in Germany. The introduction of the German Diagnoses Related Group remuneration system for inpatient services in 2004, which demands reporting of all inpatient services for reimbursement, enables the analysis of national trends in inpatient hysterectomy from 2005 onwards. Nationwide Diagnoses Related Group hysterectomy data was obtained using the following German Operation and Procedure Classification (OPS) codes: Subtotal hysterectomy: 5–682, total hysterectomy: 5–683, obstetric hysterectomy: 5–757 and radical hysterectomy: 5–685. This data is publicly available from the German Statistical Office16. The yearly population at risk, i.e. Germany’s female population from 2005 to 2019 by 5-year age group, was also retrieved from the German Statistical Office17.

Crude incidence rates were calculated with the absolute number of hysterectomies as the numerator and the denominator as the annual female population obtained from the Federal Statistical Office (per 100,000). Trends of all hysterectomies and the subtypes except for obstetric hysterectomy were visualised with trend graphs. Except for calculating the rate of all hysterectomies, obstetric hysterectomy was not included in the analyses because it is less frequently performed and limited to the child-bearing age group. Analyses by age group as well as age-period-cohort were restricted to the ages 20 years and above because of the rarity of hysterectomy in younger women.

For the age, period, cohort analysis, age and period were grouped in five-year intervals. Fourteen age groups were obtained, the first being the 20–24-year group and the last group consisting of people aged 85 years and above, but for this study, it was taken as a 5-year age group, 85–89. Three period groups were obtained: 2005–2009, 2010–2014 and 2015–2019. The midpoint of age and period were used to represent each group. The packages used created sixteen cohorts from 1920–1924 to 1995–1999 by subtracting the age from the period.

Descriptive plots were first created with the help of the rateplot function of the Epi R package for the three hysterectomy subtypes. Four classical plots for each data set were then produced: Age-specific rates by period, age-specific rates by date of birth, period-specific rates by age and cohort-specific rates by age.

The age period cohort effects were estimated using the National Cancer Institute APC web tool (available at /https://analysistools.cancer.gov/apc/). For our results, we focused mainly on the estimated functions: age-specific rates, period- and cohort-specific rate ratios (RR), local drifts (annual percentage changes for each age group) and the net drifts (indicating the overall annual percentage change). The decision on whether to report the age-specific longitudinal or age-specific cross-sectional rates was made after careful analysis of the descriptive plots utilising the theory proposed by Clayton18. This theory presupposes the presence of a strong period effect when “parallelism”—the consistent variation in the single curves—is present in the age-specific rates by period plot. The absence of parallelism suggests predominant cohort effects. The reference period was the median year for 2010 to 2014; 2012 and the reference cohort 1956–1960. The significance of estimable functions was determined using Wald Chi-square tests. The statistical significance level was set at 0.05.

Ethics approval

This study used publicly published secondary data with no individual data; therefore no ethical approval was needed for this study.

Results

A total of 1,974,836 inpatient hysterectomies were performed in Germany from 2005 to 2019. Overall, the hysterectomy rate reduced from 371.53 procedures per 100,000 women per year in 2005 to 239.85 procedures per 100,000 women per year in 2019. Total hysterectomy remains the most common type of hysterectomy performed in Germany, although the total numbers and proportions have decreased over time. Total hysterectomy rates decreased from 335.97 per 100,000 women in 2005 to 168.99 per 100,000 women in 2019. The radical hysterectomy rates and numbers also decreased over the period, but the proportions remained unchanged. In contrast, the numbers, proportions and rates of subtotal hysterectomy increased over the period; from 6.2% of all hysterectomies and a crude rate of 22.67 per 100,000 women in the 2005–2009 period to almost 21.9% and a rate of 56.50 per 100,000 women in the 2015–2019 period (Table 1 and Fig. 1).

Table 1 Number and crude rates of hysterectomies by subtypes. Germany, 2005–2019.
Full size table
Figure 1
figure 1

Crude incidence rate of hysterectomies performed by type. Germany, 2005–2019. AlH All hysterectomies, TH Total hysterectomy, SH Subtotal hysterectomy, RH Radical hysterectomy.

Full size image

In the descriptive graphs for total and radical hysterectomy, the age-specific rates by period plot show a consistent trend across the three periods, namely, 2005–2009, 2010–2014 and 2015–2019 (Fig. 2). There is a larger decline between the 2010–2014 and 2015–2019 periods than between the 2005–2009 and 2010–2014 periods. In contrast, the subtotal hysterectomy graph reveals a reversal in this trend, with increasing rates across the periods. The increase in rates is also higher in the 2005–2009 and 2010–2014 periods compared to 2010–2014 and 2015–2019 periods. The age-specific rates by period graph and age-specific rates by birthdate demonstrate a consistent morphology between the plots and across the subtypes. Since all the age-specific rates by period graphs are similar in all the three subtypes, the age-specific cross-sectional rates are reported, where age effects are presented as the age-specific incidence rates for the reference period.

Figure 2
figure 2

Crude rates of hysterectomies subtypes by age and period, age and cohort, period and age and cohort and age. Germany, 2005–2019. CIR Crude incidence rate.

Full size image

Figure 3 shows the (annual percentage change in expected overall age-adjusted rates) and the local drift (expected age-specific rates over time) by hysterectomy subtype. The overall net drift per year was − 6.13 (95% CI − 6.49, − 5.77) for subtotal hysterectomy, for subtotal hysterectomy 9.86 (95% CI 9.13, 10.58) and − 5.99 (95% CI − 6.27, − 5.70) for radical hysterectomy. Except for the 20–24-year age group, the local drifts are significant for all other age groups for all the hysterectomy subtypes. The local drifts decreased significantly for all age groups for total hysterectomy, with those age 60 years and above having the greatest decrease. Local drifts decreased significantly for radical hysterectomy, also, with the greatest decrease being in the 65–70-year age group. The local drifts increased in all age groups for subtotal hysterectomy, with the greatest drift being in the 55–80-year age group.

Figure 3
figure 3

Net drift (annual percentage change in expected overall age adjusted rates) and local drift (expected age-specific rates over time) by hysterectomy type in Germany: 2005–2019. (a) total hysterectomy, (b) subtotal hysterectomy, (c) radical hysterectomy. The solid and dashed horizontal lines represents the net drift and the accompanying confidence intervals, the points and the shaded area represents the local drifts and the confidence intervals.

Full size image

Total hysterectomy rates increase steeply with increasing age until it peaks at 45–49 years with 608.63/100,000 women (95% CI 565.70, 654.82), then decrease until around age 60 years, and then increase slightly again in older years (Fig. 4a). The period effects show an increasing total hysterectomy risk over the period. The cohorts born after 1960 had a lower risk of having a total hysterectomy performed compared to those born before this period. Like total hysterectomy incidence rates, subtotal hysterectomy is rarely performed in younger women. Rates then increase steeply, peaking at age 45–49 with rates of 151.30/100,000 women (95% CI 138.38, 165.44), before decreasing rapidly again in the older age groups (Fig. 4b). The risk of undergoing a subtotal hysterectomy increased over the study period. The cohort born after 1960 had a higher risk of undergoing a subtotal hysterectomy compared to the cohorts born before 1960. Radical hysterectomy rates are almost zero at 20 years, increasing sharply until 50 years, and then the slope of increase becomes gradual, peaking at 65–69 years with 40.63/100,000 (95% CI 38.84, 42.52) and decreases after 75 years (Fig. 4c). The risk of radical hysterectomy has decreased with increasing birth cohort, with cohorts born after 1960 having a decreased risk of having radical hysterectomy performed. Supplementary Table 2 shows all model estimates and confidence intervals. All estimated functions (cohort and period rate ratios and net and local drifts) were statistically significant (p < 0.01) in all hysterectomy subtypes (Supplementary Table 1).

Figure 4
figure 4

Age, period and cohort effects: Hysterectomy subtypes. Germany, 2005–2019. (a) total hysterectomy, (b) subtotal hysterectomy, (c) radical hysterectomy. Age effects are the expected age-specific rates in reference period adjusted for cohort effects, Ratio of age-specific rates in each period relative to reference period (2010–2014), Ratio of age-specific rates in each cohort relative to reference cohort (1956–1960).

Full size image

Discussion

This study aimed to analyse the trends of all hysterectomy cases and subtypes performed in Germany and to estimate the age, period, and cohort effects using an age-period-cohort analysis. The analysis showed decreasing hysterectomy rates overall and for all subtypes except for subtotal hysterectomy, with a peak age of 45–49 years for total and subtotal hysterectomy and a later peak at 65–69 years for radical hysterectomy. All hysterectomy subtypes showed predominant period effects with decreasing risk over the study period, whilst subtotal hysterectomy showed a reverse pattern.

The decreasing hysterectomy incidence rates observed in this study are similar to trends in other European countries such as Austria19, Portugal20 and Poland21. In recent years, there has been a plethora of options available to treat benign conditions, such as the levonorgestrel-releasing intrauterine system for abnormal uterine bleeding, myomectomy, uterine artery embolisation and Magnetic Resonance-guided focused ultrasound surgery for fibroids, which have made invasive procedures such as hysterectomy a second choice for benign indications9,10. Indeed, the German Society of Obstetricians and Gynaecologists S3 guidelines released in 2015 recommend these conservative therapies as the first line for managing most benign indications, especially when fertility preservation is desired8. In managing pelvic prolapse, for example, routine hysterectomy should no longer be performed except when indicated.

With the availability of these uterus-sparing treatment options, several criticisms have emerged since the early 1990s over the possible overuse of hysterectomy to treat benign diseases22. A probable response to this is an intentional attempt to reduce the overuse of hysterectomy for benign conditions across all age groups, an effect that is seen as the reduced risks of having a total hysterectomy (the most performed type of hysterectomy) over the study period. In a Danish analysis of hysterectomy trends, the observed decreasing hysterectomy incidence trends between 2000 and 2015 were due to the decrease in hysterectomies performed for benign diseases23.

Hysterectomies are performed using either the vaginal approach, open abdominal approach or the more recent laparoscopic approach, which is now more commonly performed24,25,26. In Germany, a study at a tertiary facility looking at surgical approaches between 2007 and 2016 revealed that the use of abdominal hysterectomy decreased from 61.4 to 13.4%, laparoscopic hysterectomy increased from 4.1 to 69.7% and vaginal hysterectomy decreased to 14.6%, despite an initial increase from 21 to 45.5% between 2007 and 201324. Under the laparoscopic approach, there are three subtypes, the total laparoscopic hysterectomy, laparoscopic subtotal hysterectomy and laparoscopic-assisted vaginal hysterectomy, which is a combination of the vaginal and laparoscopic approaches25,26.

The laparoscopic subtotal hysterectomy, compared to the total, was said to be relatively easier to perform, requiring shorter operating time because the difficult dissection around the cervix and bladder is not required with this procedure25,27. The procedure was, therefore, perceived to be associated with less bleeding, lower urinary tract complications and better pelvic floor integrity27,28. Initial case series and retrospective analyses also reported better sexual function with subtotal hysterectomy29,30. These reports and suggestions may have influenced the choice of surgical approach for hysterectomy, resulting in the increase in subtotal hysterectomy rates found in our analysis. However, in recent larger randomized trials and meta-analyses, the reported better sexual well-being, pelvic floor integrity, and urinary tract complications with subtotal hysterectomies could not be confirmed31,32. Although subtotal hysterectomy is still associated with less operating time and blood loss, this has not been found to be clinically significant31,32.

In 2014, the Food and Drug Administration warned against the risk of inadvertent spread of occult malignancy with the use of morcellators with the laparoscopic subtotal hysterectomy procedure33, which subsequently led to a 50% reduction in the procedure in the US34. Decreased conduct of the procedure has been reported in countries like Finland35 and Brazil36. In Germany, guidelines by the German Society of Obstetricians and Gynaecologists did not limit the use of this approach but recommended an investigative workup to exclude malignancy before the procedure8. However, one can still notice the effect of this warning by the relatively smaller increase in rates compared to the 2010–2014 period.

The high incidence of cervical cancer that led to the increase in the number of total hysterectomies in the mid-twentieth century37 has declined where population-based screening has been introduced. Additionally, the fear of the risk of cancer of the cervical stump that perpetuated this preference has also been reported not to be higher than the risk in the general population in areas where good screening measures are available and utilised post hysterectomy38,39. However, in a Danish study, having a subtotal hysterectomy after the age of 50 years was found to be associated with up to a five-fold increase in cervical cancer risk39. This could likely explain the rarity of the subtotal procedure in older women. The statutory cervical cancer screening program, which offers regular testing to women aged 20 years and older, was instituted in 1971 in West Germany and expanded to the East in 199440. This has resulted in a higher rate of diagnosis of precancerous lesions and early-stage cervical cancer41. Early diagnosis enables the use of less invasive surgeries such as conisation and radical trachelectomy in selected cases where fertility is desired42, contributing to the falling probability of hysterectomy overall and radical hysterectomy over the study period.

The earlier peak for total and subtotal hysterectomy at 45–49 years found in our study, is consistent with that of estimates in the United States43, Denmark23 and Brazil36. In a cross-sectional analysis of women who had undergone a hysterectomy in the city of Mainz and the Mainz-Bingen region in Germany, the probability of having a hysterectomy was highest in the 45–49-year age group44. The aetiology of benign gynaecological conditions such as fibroids, endometriosis, adenomyosis and endometrial hyperplasia that affect women in the reproductive age group is linked to dysregulation in reproductive hormones; estrogen and progesterone45. Ultrasound studies showed an increasing prevalence of uterine fibroids regardless of symptoms with increasing premenopausal age46. In a cohort study conducted in California, the 45–49-year age group had the highest incidence rate of uterine fibroids47. Even for those with earlier symptoms, the desire to preserve fertility might lead to the initial use of conservative options to fulfill patients' wishes, thereby pushing the need for hysterectomy to later ages. In a retrospective study of surgical treatments for fibroids, patients over 40 years preferred hysterectomy48, possibly because most women have no further intentions of preserving their fertility by then.

In Germany, the median age for endometrial cancer diagnosis is 6841. Since radical hysterectomy is generally indicated in invasive cervical42 and advanced endometrial cancers49 and endometrial cancer is the most common gynaecological cancer in Germany, it is not surprising that the highest rate for radical hysterectomy was in the 65–69 years age group. The higher incidence of gynaecological malignancies in older ages may also contribute to higher total hysterectomy rates after 60 years. With about 70% of endometrial cancers being diagnosed at an early stage41, these are, however, more likely to be treated surgically with total hysterectomy and bilateral adnexectomy, as strongly recommended by the German S3 endometrial cancer guidelines for endometrial cancer stage49.

Although this paper presents an age-period-cohort analysis of the national hysterectomy incidence trends in Germany, the relatively short study period may make it difficult to observe cohort effects and introduce data artifacts. Since this is an ecological study, analysis and interpretations are based on a population level and might not hold for all regions and their variations in Germany. In 2014, Stang and colleagues reported higher hysterectomy rates in West Germany compared to the East from six population cohorts from 1997 to 200650. A small proportion of hysterectomies are performed as outpatient cases, which are not captured in the Diagnoses Related Group data51.

The dynamics of the trends of hysterectomy subtypes in Germany are changing and has important implications for clinical practice. With increasing subtotal hysterectomy rates, the number of women who have undergone hysterectomy but have a cervix and, therefore, are still at risk of cervical cancer also increases. There is a need to educate this population on the need to continue cervical cancer screening. Continuous monitoring of trends is needed to detect these changing dynamics and guide practice accordingly.