METHODS

The study population consisted of 3498 women, aged 35 years or more and residing in Rome, who in 1997 underwent a hysterectomy, excluding pelvic evisceration (ICD-9-CM 68.3–68.7, 68.9). The data analysed were those reported on hospital discharge records, retrieved through the hospital information system run by the health authority of the Lazio Region. This system collects and manages data from all discharges from both public and private hospitals in the region as well as discharges of Lazio residents hospitalised elsewhere in Italy. In 1997 the region had 223 acute hospitals (186 public or contracted hospitals and 37 private hospitals), and all but five (all private) provided data to the system (the five hospitals not providing data had a total of 298 beds and represented 13.5% of the private hospitals and 2.2% of total hospitals). Completeness and accuracy of data collected through the regional information system were sufficiently good in 1997, as shown by quality of coding indicators (average number of procedures by discharge abstract = 1.7; percentage of missing or wrong principal diagnosis = 2%).

We applied an area based index of socioeconomic status (SES), developed using 1991 census data on the following characteristics of the census block of residence (CBR, average inhabitants equal to 480): percentage of people by educational level, percentage of people employed by occupational category, percentage of unemployed men of working age, percentage of one person families, percentage of families with five or more persons, crowding index (persons/room), and percentage of dwelling rented or owned. The index is described in detail elsewhere.²⁸ The index showed a 0.63 linear correlation with the mean CBR per capita income among residents in Rome. Four levels of SES were defined on the basis of the 20th, 50th, and 80th centiles of the SES distribution (SES level I included women with the highest SES; level IV included those with the lowest SES). The SES was determined for 3141 (89.7%) of the 3498 women who underwent a hysterectomy in 1997; for the remaining 360 women, information on residence was not complete.

Episodes of hysterectomy were then subdivided according to the primary cause of hospitalisation (both malignant and not-malignant diseases), using ICD-9 codes of principal diagnosis (except for cancer, which was selected even if reported as a secondary diagnosis). Specifically, we considered the following mutually exclusive groups: uterine cancer (ICD9: 179, 180, 182, 233.1, 233.2, 236.0), other malignancies (ICD9: 140–208, 230–239, excluding uterine cancer), leiomyoma (ICD9: 218), other benign neoplasms of the uterus (ICD9: 219), disorders of the uterus not classified elsewhere (ICD9: 621), genital prolapse (ICD9: 618), benign neoplasm and disorders of ovary (ICD: 220, 620), disorders of menstruation (ICD9: 626), and the remaining non-malignant causes.

Direct age standardised hospitalisation rates and their 95% confidence intervals (CI) were calculated for each of the four SES levels, with the 1996 Rome population as reference population.

The ratios of standardised rates and their 95% CI were calculated to measure the excess risk of lower SES levels (levels II, III, and IV) compared with the highest level (level I). A weighted linear regression was performed on the logarithm of the rates to test for the existence of a trend across the SES levels.

RESULTS

The hysterectomy rate was 36.7 per 10 000 women aged 35 years or more (corresponding to 22.7 per 10 000 women of all ages). Eighteen per cent of the hysterectomies were performed for malignant causes, whereas the majority of hysterectomies (82%) were performed for non-malignant diseases of the uterus or annexes (table 1); almost half were performed for leiomyomas and other benign neoplasms of the uterus and 16.4% for genital prolapse. While cancer and prolapse were more frequent after 50 years of age, leiomyoma is responsible of 57% of hysterectomies in women aged 35–49 years and of 26% in those aged 50 years or more, resulting in hysterectomy rates for leiomyoma of 28.6 and of 7.7 per 10 000 in the two age groups, respectively (table 1).

The age standardised hospitalisation rates of hysterectomy for all causes showed a significant relation with SES (table 2). The risk of hysterectomy was 34% higher for the least affluent group (level IV) compared with the most affluent group (level I), and the linear regression showed a significant increase in rates with decreasing SES (p=0.013).

For malignancies, although the hysterectomy rate was significantly higher for SES levels III and IV compared with level I among women aged 35 years or more, the linear trend was not significant at 0.05 level (table 3). The risk of hysterectomy was particularly high among the most disadvantaged women in age group 35–49 (88% higher risk compared with level I).

As regards non-malignant conditions (table 4), we observed a significant inverse association between SES and hysterectomy in women aged 35–49 years (p=0.009) but not in those aged 50 or more. Among the women of the youngest age group, the risks of hysterectomy in lower SES levels were all significantly higher compared with level I (22%, 44% and 60% increase for level II to IV respectively).

Table 5 reports the results for the two most frequent non-malignant causes of hysterectomy. A significant inverse association was found for both leiomyoma (p=0.019) and genital prolapse (p=0.022). In the case of leiomyoma, the trend was significant for both women aged 35–49 (p=0.039) and those aged 50 years or more (p=0.009).

DISCUSSION

Our results show that hysterectomy rate in Rome is similar to that observed in England, much lower compared with the United States and Finland, and almost double those in Scandinavian countries.^4,26 Hysterectomy rate was much higher among women aged 35–49 years compared with women aged 50 years or more, and the difference is largely depending on interventions for leiomyoma. A similar pattern of hysterectomy rates by age and by cause has been reported from the United States by Wilkox et al.⁵

The higher risk of hysterectomy among women with a low SES, compared with more affluent and educated women, can for the most part be attributed to a higher frequency of surgery performed to treat benign disorders, particularly leiomyomas and genital prolapse. However, we were not able to establish whether this higher frequency reflects actual variations in disease prevalence or a different pattern of primary or secondary care, as no data on morbidity for benign uterine disorders by social class are available. Nevertheless, risk factors for leiomyoma, such as low parity¹⁷ and hormonal treatment with oestrogen for menopause symptoms,²⁹ have been reported to be more frequent among affluent socioeconomic groups.^26,30 It is thus unlikely that leiomyoma is more prevalent among less affluent women, and alternative explanations must therefore be found. It has been argued that women of low SES are less likely to undergo regular gynaecological examinations than women of high SES and they may present to consultation when leiomyoma and genital prolapse are too advanced to be treated with less radical means.²¹ Moreover, women with a low educational level have been reported to be less knowledgeable about treatment and thus less capable of expressing preferences, resulting in their being more likely to comply with the offer of hysterectomy.³¹ Another possible explanation for the inverse relation between hysterectomy for benign disorders and socioeconomic position can be found in the Tudor Hart's inverse care law.³² According to the law, the most affluent groups would have a more prompt uptake of innovative, less invasive techniques for removing uterine leiomyoma, with less affluent women being more vulnerable to inappropriate hysterectomy.

Although the significant association between hysterectomy and socioeconomic position observed only in age group 35–49 seems to be dependent on the higher frequency of leiomyoma before menopause, there is no effect modification of age on this relation, as a significant inverse association between hysterectomy rate and SES was found among women of both younger and older age groups. From a public health perspective however, it is worrying that reproductive age may not be a deterrent against potentially inappropriate interventions.

Although the results of our study confirm those of most other studies on socioeconomic patterns of hysterectomy,^16–24 they are not concordant with reports from Finland, which have found a positive association between disposable family income and surgical rates for hysterectomy, as well as rates for cataract removal, suggesting that there exist socioeconomic disparities in access to health care services.^26,27 In Rome, lower socioeconomic groups have equal access to facilities for low discretionary, effective treatment, such as surgery for hip fracture or cataract removal.³³ At the same time, they are more frequently hospitalised than higher socioeconomic groups for interventions considered to be specifically or generically inappropriate, such as appendicectomy or minor skin diseases³³ despite the fact that low socioeconomic position is not considered a risk factor for these conditions.

There is evidence that providing better information to the patient on the potential advantages and risks of treatment can reduce hospital use.³⁴ In Switzerland, the prevalence of hysterectomy was found to be significantly higher among privately insured patients, compared with those who were generally insured, and among the general population, compared with female doctors and lawyers' wives.³⁵ None the less, hysterectomy rates sharply fell after a mass media public education campaign,³⁶ emphasising the importance of information in modifying both the demand for hospital services and the offer of these services. Improving patient-doctor communication may reduce unnecessary surgery when effective, less radical treatment becomes available.

Our study shows that women of lower socioeconomic position aged 35–49 years have higher rates of hysterectomy because of malignancies, compared with more affluent women, albeit the linear trend was not statistically significant. This result is consistent with 1990–95 mortality data for uterine cancer in Rome, which show a significant inverse relation with socioeconomic position.²⁸

In interpreting our results, several potential biases must be taken into consideration. Specifically, we assigned socioeconomic status using a small area index, previously used in studies on differentials in mortality in Rome.²⁸ Thus misclassification may have occurred, though geographical indices are considered to be a sufficiently robust and valid option when individual information is lacking.^37–39 None the less, it is more likely that the effects of individual characteristics were underestimated, as compared with being overestimated.^40,41

As mentioned, five hospitals did not provide data in 1997; as they are all private hospitals, most patients probably have a fairly high socioeconomic status. However, these hospitals only represented a small percentage of the private hospitals and a negligible percentage of total hospitals.

Differences in access to treatment do not necessarily signify inequality, unless they are adjusted for need⁴²; thus, as morbidity data by socioeconomic position were not available for benign conditions, we discussed the prevalence of risk factors among socioeconomic groups. Finally, discharge abstracts provide no indications of the clinical appropriateness of interventions.

In conclusion, this study provides data on hysterectomy rates by age groups in Rome, which can also be used for comparisons with other countries. We show that hysterectomy for leiomyoma is widely practised in Rome regardless of age despite the availability of less radical interventions. Furthermore, this study shows that women of low SES undergo hysterectomy, especially for leiomyoma, more frequently than those with a high SES. The results of this study show the need to implement educational interventions for both women and health professionals regarding alternatives to hysterectomy for uterine disorders.