INTRODUCTION

Despite the availability of a wide therapeutic arsenal and bone densitometry, many patients at risk of fragility-related osteoporotic fractures remain undiagnosed and untreated. Back in 2019, it was estimated that 2,985,000 individuals in our country suffered from osteoporosis (79 % of them, women). Only 36 % of the 1,827,000 women diagnosed with osteoporosis receive treatment, underscoring a treatment gap of 64 % (¹). It is estimated that up to 80 % of individuals who have experienced, at one time or another at least, 1 fragility fracture lack proper diagnosis, have not been identified correctly, and consequently, do not receive the necessary diagnosis or subsequent proper management.

The concern surrounding the clinical management of osteoporosis has stimulated the development of procedures to assess the risk of fracture using key risk factors. A group of experts developed a tool called FRAX to identify individuals at higher risk of sustaining fractures within the next 10 years, by combining key factors of risk of fracture and adding bone mineral density (BMD) when available (²). Among fracture-related factors, decreased BMD has been identified as the primary risk factor because of its close relationship with bone strength. Other factors contributing significantly to the risk of fracture have also been identified including family and personal history of fragility fractures, low body weight, smoking, and age, which is known to be one of the most important independent predictors of fracture other than BMD.

However, we should admit that FRAX is not a perfect tool and has been criticized since its inception for not including certain risk factors widely discussed in the medical literature available, such as spine BMD, bone markers, use of benzodiazepines, or history of risk of fall (³-⁷). Therefore, the objective of this study was to assess if the variables not included in this algorithm, particularly certain gynecological variables, should be included based on the role they play on fragility fractures.

MATERIAL AND METHODS

A total of 316 women older than 50 years were randomized from the Oviedo municipal registry in Oviedo, Asturias, Spain. These women had previously participated in the EVOS study, which was initially designed to determine the prevalence of vertebral fractures across Europe. Also, they were asked to fill out a specifically designed survey for the EVOS study, which exhibited good reproducibility (⁸,⁹). This survey included questions on clinical variables such as weight and height to calculate the body mass index (BMI), osteoporosis-related risk factors such as previous fractures, parental history of hip fractures, smoking habits, and use of glucocorticoids for over 3 months. Additionally, other variables were analyzed, such as age at menarche, age at menopause, fertile years, and nulliparity. The femoral neck BMD of this cohort was also established.

This cohort was prospectively followed for 8 years through 4 postal surveys to establish the incidence of non-vertebral osteoporotic fractures during this period. All osteoporotic fractures (hip, Colles, humerus, rib, pelvis, and tibia) excluding skull and limb fractures were confirmed through x-rays, or medical reports. The total number of participants in the final follow-up was 223, with a participation rate of 81.3 % at the 8-year follow-up (excluding deaths). The participation rates for the 3 previous postal follow-ups were 87.1 %, 87.5 %, and 82.4 %, respectively.

RESULTS

Table I illustrates the baseline clinical characteristics of the variables used by the FRAX tool, as well as other variables not included in this algorithm, such as age at menarche and menopause, fertile years, and nulliparity. The mean age was 65 years and the BMI, 28.6 kg/m². The femoral neck BMD (0.683 g/cm²) fell within the range observed for the Spanish population aged 60-69 years (0.694 g/cm²). The parental history of hip fracture was reported by 7.7 % of participants, and the prevalence of previous non-traumatic fractures was 20 %. Less than 10 % (8.5 %) had used corticosteroids, and 4.7 % of women were smokers. Regarding the gynecological and obstetric variables, notable figures include a mean menarche, menopause, and fertile age of 14, 48.8, and 34.8 years, respectively. Approximately 17.5 % of women had never been pregnant (nulliparity).

Table I. Description of the baseline clinical characteristics of the study population. 

The univariate analysis showed that, among the variables used to build the FRAX algorithm, only age and parental history of hip fracture were actually associated significantly with the incidence of hip fracture. However, age at menopause and fertile years were also associated (Table II). We should mention that, among fractured women, no use of corticosteroids or smoking habits were reported. The multivariate analysis adjusted for age and BMI only showed significant associations between age [OR, 1.09; 95 % confidence interval (CI), 1.01-1.17] and age at menopause [OR, 0.90; 95 %CI, 0.82–0.99] with incidental hip fractures.

Table II. Association between clinical variables and the presence or absence of incidental hip fractures. 

Italicized values indicate significant differences between the two groups.

In the univariate analysis, BMI, femoral neck BMD, and nulliparity were associated with the occurrence of incidental Colles fractures (Table III). According to the multivariate analysis, only nulliparity was associated significantly with the occurrence of Colles fractures [OR, 4.59; 95 %CI, 1.59-13.26)].

Table III. Association between clinical variables and the presence or absence of incidental Colles fractures. 

Italicized values indicate significant differences between the two groups.

Regarding the incidence of major osteoporotic fractures, the univariate analysis showed significant associations with the parental history of hip fracture, nulliparity, and fertile years (Table IV). According to the multivariate analysis, the parental history of hip fracture [OR, 3.26, 95 %CI, 1.23-8.61)], nulliparity [OR, 3.07, 95 %CI, 1.48-6.37)], and fertile years [OR, 0.92, 95 %CI, 0.87-0.98)] were associated with the occurrence of incidental major osteoporotic fractures.

Table IV. Association between clinical variables and the presence or absence of incidental major osteoporotic fracture. 

Italicized values indicate significant differences between the two groups.

DISCUSSION

The results from this study illustrate the significance of some gynecological variables not included in the FRAX algorithm that could carry specific weight. We should mention the significant protective effect of delaying menopause age on the incidence of hip fractures, or the protective effect of fertile years or pregnancy on major osteoporotic fractures.

Undoubtedly, FRAX is a valuable tool to aid clinicians in the identification of individuals needing osteoporosis treatment and those who do not. Also, it is the only algorithm validated across various population cohorts, with or without BMD data, and available in multiple languages. It can equally serve as an inclusion criterion in clinical trial design where fracture is an endpoint. However, it does have well-recognized limitations (¹¹).

Within these limitations, particularly pertinent to this study, is that FRAX has not been validated in Spain (¹²). Also, FRAX does not include variables that could be challenging for general practitioners to obtain, such as physical activity measurements, vitamin D deficiency, bone turnover markers, or bone loss. Another variable not included is falls despite being a recognized risk factor for non-vertebral fractures, especially hip fractures. Falls are excluded mainly due to the lack of standardized assessment methods. Perhaps this is why gynecological variables were never added to the FRAX algorithm, despite having a specific weight on the occurrence of fragility-related osteoporotic fractures (¹³). Also, it is possible that the confinement of these variables to women alone may require 2 separate calculators for each gender, which could be operationally cumbersome.

A study of our geographical context showed that not all risk factors included in the FRAX algorithm were significant when fractured and non-fractured women were compared (¹⁴). Specifically, age, previous fracture, and baseline osteoporosis were significant. However, excessive alcohol consumption, use of glucocorticoids, rheumatoid arthritis, low BMI, and parental hip fracture history did not show any significance, despite the latter 2 being considered strong or very strong risk factors in FRAX (¹⁴). In our case, BMI was associated only with the occurrence of osteoporotic Colles fractures, an association that wasn't found in the multivariate analysis. Similarly, the history of parental hip fracture was associated with the occurrence of hip and major osteoporotic fractures in the univariate models, but only remained significant with major osteoporotic fractures at the multivariate level.

Some authors suggest that the FRAX tool should add gynecological variables, such as menopause, including its onset and duration (¹⁵). The findings from our study are not intended to challenge the utility of FRAX. We view FRAX as a highly useful tool to assess individuals at high risk of sustaining fragility fractures. However, clinicians should consider these gynecological variables during the clinical examination of women older than 50 years, in addition to using the FRAX tool, because these variables have shown an association with the incidence of osteoporotic fractures (¹³,¹⁶-²⁰). In our study, regarding hip fractures, delaying menopause by 1 year reduced its incidence by 10 %, which is similar to the 9 % increased risk associated with 1 additional year of aging. Similarly, for major osteoporotic fractures, the parental history of hip fracture was associated with a 3-fold higher risk, which is somewhat similar to the risk associated with nulliparity. Each fertile year cut the risk of major osteoporotic fracture by 8 %.

Our study has several limitations. The survey used to collect osteoporosis risk factor data was interviewer-administered rather than self-administered, which potentially introduces response biases. Some FRAX variables, such as rheumatoid arthritis, secondary osteoporosis, or alcohol consumption, could not be assessed due to the survey design. Another possible, yet inevitable, limitation is the time elapsed between gynecological changes and the study data collection. The study's limited number of incidental osteoporotic fractures and the relatively small cohort size (n = 316) could also be considered limitations.

Despite these limitations, we believe the study has significant strengths. On the one hand, the cohort analyzed participated in the EVOS-EPOS study, being our center one of the 5 centers that completed all the study guidelines. Additionally, the response rate across the 4 postal follow-ups over 8 years exceeded 80 %, strongly supporting the sample's representativeness (²¹).

In conclusion, some gynecological variables like age at menopause, fertile years, and nulliparity should be considered risk factors for the occurrence of fragility fractures in the patients' medical records. Future studies should clarify whether adding these gynecological variables alongside the FRAX algorithm could enhance the predictive capabilities of this tool widely used in the routine clinical practice.