*Article* **High Cortico-Trabecular Transitional Zone Porosity and Reduced Trabecular Density in Men and Women with Stress Fractures**

**Afrodite Zendeli 1,\*, Minh Bui 2, Lukas Fischer 3,4, Ali Ghasem-Zadeh 5, Wolfgang Schima <sup>6</sup> and Ego Seeman <sup>5</sup>**


**Abstract:** To determine whether stress fractures are associated with bone microstructural deterioration we quantified distal radial and the unfractured distal tibia using high resolution peripheral quantitative computed tomography in 26 cases with lower limb stress fractures (15 males, 11 females; mean age 37.1 ± 3.1 years) and 62 age-matched healthy controls (24 males, 38 females; mean age 35.0 ± 1.6 years). Relative to controls, in men, at the distal radius, cases had smaller cortical cross sectional area (CSA) (*p* = 0.012), higher porosity of the outer transitional zone (OTZ) (*p* = 0.006), inner transitional zone (ITZ) (*p* = 0.043) and the compact-appearing cortex (CC) (*p* = 0.023) while trabecular vBMD was lower (*p* = 0.002). At the distal tibia, cases also had a smaller cortical CSA (*p* = 0.008). Cortical porosity was not higher, but trabecular vBMD was lower (*p* = 0.001). Relative to controls, in women, cases had higher distal radial porosity of the OTZ (*p* = 0.028), ITZ (*p* = 0.030) not CC (*p* = 0.054). Trabecular vBMD was lower (*p* = 0.041). Distal tibial porosity was higher in the OTZ (*p* = 0.035), ITZ (*p* = 0.009), not CC. Stress fractures are associated with compromised cortical and trabecular microstructure.

**Keywords:** bone microstructural deterioration; cortical porosity; high resolution peripheral quantitative computed tomography; stress fracture

#### **1. Introduction**

Stress fractures are commonly the result of repetitive loading and are seen in athletes, military recruits and professional dancers, but stress fractures also occur in recreational athletic individuals as vigorous exercise increases the risk for injuries [1–5]. These fractures exist across a spectrum from low-grade stress reactions, bone marrow edema to complete fractures visible radiologically [6–8]. Despite there being little or no deficit in areal bone mineral density (aBMD) reported in most studies, there is evidence suggesting that abnormalities in bone microarchitecture are associated with these fractures and so may contribute to bone fragility [6,9–17].

As stress fractures are commonly reported in young adults abnormalities in bone morphology associated with these fractures are likely to be due, in part, to abnormalities in the growth and development of peak bone macro- and microstructure, not necessarily bone loss. In the diaphysis, a region composed virtually exclusively of cortical bone, growth

**Citation:** Zendeli, A.; Bui, M.; Fischer, L.; Ghasem-Zadeh, A.; Schima, W.; Seeman, E. High Cortico-Trabecular Transitional Zone Porosity and Reduced Trabecular Density in Men and Women with Stress Fractures. *J. Clin. Med.* **2021**, *10*, 1123. https://doi.org/10.3390/ jcm10051123

Academic Editor: Lorenzo Drago

Received: 9 February 2021 Accepted: 1 March 2021 Published: 8 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

in size and mass occurs by periosteal apposition with concurrent endocortical resorptive modeling producing a radial 'modeling drift' so the cortex becomes thinner relative to its increasing total cross-sectional area (CSA) [18,19].

By contrast, the metaphyseal region contains both cortical and trabecular bone. The cortical bone is formed by condensation or 'corticalization' of trabeculae as they arise from the periphery of the growth plate. Adjacent trabeculae coalesce by bone formation upon their surfaces while more centrally placed trabeculae form the metaphyseal trabecular compartment. The cortico-trabecular transitional region so formed between the cortical bone and the more centrally placed trabecular bone is composed of mineralized bone matrix that transitions from the more compact cortical configuration to a more open spongy trabecular configuration with more void volume (porosity) [20].

The purpose of this study was to quantify any association between bone microstructure and the presence of a stress fracture in the lower limb. We hypothesized that young adult individuals sustaining stress fractures have thinner and more porous cortices and reduced trabecular density characterised by reduced numbers and thinner trabeculae.

#### **2. Materials and Methods**

This was a cross sectional case control study. Thirty-four patients presenting with acute focal lower limb pain were assessed for possible stress fractures using magnetic resonance imaging (MRI) [21]. Participants were recruited from the osteoporosis outpatients department of the St. Vincent Hospital and the Austrian Military Hospital in Vienna/Austria.

Patients were included if they were above 18 years of age and sustained a recent MRI-diagnosed stress fracture within 14 days prior to inclusion. Patients receiving any medication affecting bone metabolism (parathyroid hormone, intravenous or oral bisphosphonate, strontium ranelate, raloxifene, hormonal replacement therapy or anabolic steroids) were excluded. Other exclusion criteria included the presence of metabolic bone disease, a history of malignancy, hypo- or hyperparathyroidism, pregnancy and lactation.

Of the 34 subjects, 26 were included (11 females, 15 males). The study group included five military recruits, three long distance runners, one sprinter and one professional tennis player, the other 16 patients were recreational sportsmen and women. We compared these results with data obtained from 62 healthy age- and sex matched Caucasian controls (42 Austrian subjects, 20 Australian subjects). The study was approved and supervised by an independent local ethics committee in Vienna/Austria.

#### *2.1. Magnetic Resonance Imaging*

Examinations were performed on 1.5 or 3.0 T scanners (Signa HDx, GE Healthcare, USA; Intera, Philips, The Netherlands, Avanto, Siemens, Germany) using a phased-array extremity coil. All MRI examinations included an axial T1-weighted spin-echo (SE) sequence and an axial fat-suppressed T2-weighted turbo spin-echo (TSE) or a short-tau inversion recovery (STIR) sequence. The presence of bone marrow edema and stress fracture was seen as decreased signal intensity on T1 weighted SE images, and markedly increased signal intensity on STIR images or T2 weighted fat-suppressed TSE images [22]. In 16 patients High Resolution Multi-Detector CT (HR-MDCT) scanning was performed with a 128-row MDCT scanner (Somatom AS+, Siemens, Forchheim, Germany). Images were acquired in ultrahigh resolution (UHR) mode covering the area of interest as defined by MRI. Slice thickness was 0.6 mm, field of view of 160–300 mm (*x*-axis) × 160–300 mm (*y*-axis), with a matrix of 512 × 512, which equates to a resolution of 0.3–0.6 mm (*x*-axis) × 0.3–0.6 mm (*y*-axis) × 0.6 mm slice thickness (*z*-axis). A linear stress fracture was visible in 16 of 26 cases (61.5%) [23]. In 8 patients the fracture was identified in the metatarsals, in 8 patients in the calcaneus and ankle, in 6 patients in the proximal tibia adjacent to the knee, in 2 patients in the femur condyle, and in 2 patients in the hip (one femoral head and one acetabulum).

#### *2.2. Measurement of Bone Microarchitecture*

Scans were obtained from the non-dominant distal radius and unfractured side of the distal tibia using by HR-pQCT scanner (XtremeCT; Scanco Medical AG, Brüttisellen, Switzerland) using the standard in vivo protocol (60 kVp, 900 μA, 100-ms integration time) [24]. StrAx1.0 is a new algorithm that segments bone from background, and then bone into its compact appearing cortex (CC), outer transitional zone (OTZ) and inner transitional zone (ITZ), and trabecular compartments (see Figure 1); and in so doing correctly confines the trabecularized cortex (i.e., cortical fragments) to the transitional zone rather than incorrectly allocating cortical fragments (which look like trabeculae) to the medullary compartment; a segmentation error which overestimates 'trabecular' density [25]. The OTZ is the trabecularized cortex adjacent to the CC whereas the ITZ is the trabecularized cortex adjacent to the medullary cavity. The latter also contains true trabecular bone (of growth plate origin).

**Figure 1.** HR-pQCT image of the distal tibia of an 18-year-old male patient with a stress fracture. Compact appearing cortex (green); outer transitional zone (white); inner transitional zone (red); medullary area (yellow).

Of the 110 slices in the region of interest imaged by HR-pQCT the analysis is restricted to the 40 most proximal slices because the 70 distal slices often have very thin cortices, and hence are unsuitable for unambiguous quantification of cortical porosity. Porosity was quantified as previously reported. The coefficient of variation for segmentation and quantification of porosity ranges from 0.54 to 3.98% depending on the compartment [26,27].

#### *2.3. Statistical Analysis*

Testing for normality of distribution of a variable was conducted using Shapiro– Wilk test. Two sample *t*-test was used to test the difference in mean between cases and controls if the data was approximate normal, otherwise nonparametric Mann–Whitney test was used to test the difference in the medians. Post hoc power calculations were done. Initially, the statistical test was computed without adjustment for other covariates, and then adjustment for age and height was carried out. Spearman's correlations were used to assess relationships between porosity and medullary CSA/total CSA. Univariate logistic analysis was used to examine the association between porosity of CC, OTZ, ITZ and trabecular volumetric BMD and fracture risk. All variables were standardized to have mean of 0 and SD of 1 and were used to predict the odds ratio (OR). The analyses were performed on data at both tibia and radius site for males and females separately. We used statistical software STATA (StataCorp, 2009), version 11, to conduct all analyses. *p*-values were computed for two-sided tests and values less than 0.05 were considered as significant.

#### **3. Results**

Characteristics of the entire study population are described in Table 1.


**Table 1.** Characteristics of cases and controls.

Results are shown as mean ± SE (Standard error of mean); *p*-value comparing cases and controls (*p* < 0.05).

#### *3.1. Males*

As shown in Figure 1 and Table 2, relative to controls, at the distal radius, cases had smaller cortical CSA (*p* = 0.012) due to a smaller CC CSA (*p* = 0.005) and a smaller OTZ CSA (*p* = 0.003). Cases had higher porosity of the OTZ (*p* = 0.006), ITZ (*p* = 0.043) and CC (*p* = 0.023). Cortical vBMD was lower (*p* = 0.002). Trabecular vBMD was lower (*p* = 0.002). Trabecular thickness was reduced (*p* = 0.001), connectivity was reduced (*p* = 0.012) and separation increased (*p* = 0.007) (Figure 1, Table 2).

**Table 2.** Comparison of microarchitecture of the distal radius and distal tibia in cases and controls.


Results are shown as mean ± SE (standard error of mean); *p*-value comparing healthy controls and fracture patients, adjusted for age and height, using \* two-sample *t*-test and \*\* nonparametric Mann–Whitney test on the residuals of regression of a variable on age and height (*p* < 0.05). TCSA = total cross sectional area, CSA = cross-sectional area; CC = compact-appearing cortex; OTZ = outer transitional zone; ITZ = inner transitional zone; vBMD = volumetric bone mineral density; Tr = trabecular; HA = hydroxyapatite.

At the distal tibia, relative to controls, cases had a smaller cortical CSA (*p* = 0.008) due to smaller CC CSA (*p* = 0.015) and a smaller OTZ CSA (*p* = 0.012). Porosity was not higher at this site. Cortical vBMD was lower (*p* = 0.016). Trabecular vBMD was lower (*p* = 0.001). Trabecular thickness was reduced (*p* = 0.0004), connectivity was reduced (*p* = 0.006) and separation increased (*p* = 0.007) (Table 2).

#### *3.2. Females*

The results were similar in females. Relative to controls, at the distal radius, cases had significant higher porosity of the OTZ (*p* = 0.028) and ITZ (*p* = 0.030), but porosity was not significantly higher at CC (*p* = 0.054). Cortical vBMD was not lower, but trabecular vBMD was lower (*p* = 0.041). At the distal tibia, porosity was higher in the OTZ (*p* = 0.035) and ITZ (*p* = 0.009) but not at the CC. Cortical and trabecular vBMD were not lower. No differences in trabecular morphology were detected at either site (Table 2).

In both sexes, at both sites, associations were detected between porosity of CC, OTZ and ITZ and medullary CSA/TCSA (*p* < 0.001, except for the distal radius ITZ in males with *p* = 0.004) (Figure 2). Cortical porosity was associated with an increased odds of stress fracture, with ORs ranging from 1.24 to 3.13 depending on the cortical compartment, though not all sites demonstrated a statistically significant increase in odds. Higher trabecular vBMD was protective, demonstrating a statistically significantly lower odds of fracture at the distal radius in males and females and the distal tibia in males (Figure 3).

**Figure 2.** Porosity of the compact appearing cortex (CC), outer and inner transitional zones (OTZ, ITZ) as a function of medullary cross-sectional area (CSA)/total CSA at the distal radius and distal tibia in male cases (filled dots), male controls (open dots), female cases (filled squares) and female controls (open squares). All *p* < 0.001 except for the distal radius ITZ in males (*p* = 0.004).

**Figure 3.** Odds ratio (OR) for fracture (mean and 95% Confidence Intervals, CI) for porosity of the compact appearing cortex (CC), outer and inner transitional zones (OTZ, ITZ) and trabecular volumetric bone mineral density (vBMD) at distal radius and distal tibia in males and females. \* = significant *p*-values.

#### **4. Discussion**

We report that men and women with stress fractures had increased porosity, observed in the inner and outer cortico-trabecular transitional zones of the distal radius in both sexes and distal tibia in female cases. Porosity of the compact appearing cortex was increased only in males at the distal radius. Odds ratio for a stress fracture was associated with increased porosity of the outer cortico-trabecular transitional zone in three of four locations. Cortical vBMD, which in part, reflects porosity, was also reduced. Furthermore, trabecular vBMD was lower due to reduced trabecular thickness, not numbers. Males with stress fractures had thinner trabeculae with greater separation.

Stress fractures in military cadets and athletes have been subject of most studies using dual x-ray absorptiometry (DXA), x-ray and single photon absorptiometry or nuclear bone scanning [11–14,28]. Fewer studies have used peripheral quantitative computed tomography (pQCT), or HRpQCT to describe bone characteristics in association with stress fractures [10,17,29].

Our findings confirm some, not all previous studies. For example, Beck et al., reported thinner cortices in female cadets and narrower subperiosteal diameters in male cases [12].

Reduced cortical area was also reported by Popp et al., in runners with a stress fracture [30]. While the finding of greater total bone cross sectional was not statistically significant in our study, Weidauer et al., also reported greater periosteal circumference in athletes with a stress fracture [31]. Among military recruits and male runners, Giladi et al., and Popp et al., reported reduced tibial cross sectional area, while HRpQCT-analyses of our cases and work by Schanda et al., reported larger tibial and radial cross sectional area [14,17,29]. These subjects also had reduced cortical vBMD, reduced cortical CSA with increased cortical porosity. In both studies, and work by Schnackenburg et al., trabecular vBMD was found to be reduced [10].

While bone loss cannot be excluded as a cause of these deficits, 20 of the 26 subjects were under 50 years of age. Therefore, higher porosity in the cortico-trabecular junctional zone, with less consistently elevated porosity of the compact cortex, and deficits in trabecular bone may have their origin in the growth-related assembly of bone in some subjects rather than its age-related deterioration [18]. We cannot distinguish these alternatives in a cross sectional study.

We suggest that the thinner and more porous metaphyseal cortex in cases is, in part, the result of a reduction in bone formation upon trabeculae emerging from the periphery of the growth plate. These may fail to coalesce leaving a thinner and more porous cortex (failed corticalization of trabeculae). Reduced bone formation upon trabeculae emerging from the centre of the growth plate may result in the lower metaphyseal trabecular density [32]. This is supported by the presence of a smaller and more porous OTZ suggesting impairment of coalescence as thinner and more separated trabeculae fail to coalesce leaving larger pores adjacent to CC. The more porous cortex will also be thinner due to failure of cortical thickening taking place by adsorption of trabeculae upon the endocortical surface. The thinner cortices are unlikely to be due to reduced periosteal apposition because total CSA was not reduced. Less corticalization of trabeculae may also be partly responsible for the relatively larger medullary canal area as adsorption of trabecular bone upon the endocortical surface is partly responsible for reducing medullary canal area from the 'inside'.

The relatively larger medullary canal area, thinner cortices and higher porosity in the cases may also be due to greater growth-related endocortical resorptive modeling excavating a larger medullary canal during growth and concurrent intracortical resorptive remodeling forming osteons, each with their central Haversian canal. The cases were slightly taller and had wider bones. Taller persons assemble their wider bones by excavation of a disproportionately larger medullary canal such that the wider bone has a thinner cortex relative to its total CSA [19]. Wider bones are assembled with less material relative to their size because resistance to bending is a fourth power function of their radius–less material is needed to achieve a given bending strength than is needed in a narrower bone [33,34].

Wider bones may also be more porous. The greater amount of modeling required to assemble a larger total and medullary CSA is accompanied by more intracortical remodeling forming secondary cortical osteons, each with their Haversian canal (which forms most 'porosity' as seen in cross section) [35]. This is suggested by the positive correlation between medullary canal area and porosity reported here, and elsewhere [19]. This correlation is consistent with the notion of coordinated assembly of the external size, shape and internal architecture of bone by periosteal, intracortical, endocortical surface dependent modeling and remodeling [36].

The increased risk for fracture reported in taller persons is inconsistent with greater resistance to bending observed in wider bones [19]. However, the greater porosity and relatively thinner cortices of wider bones may offset the advantage of greater width. Resistance to bending is a 4th power function of the distance a unit volume of bone is placed from the neutral axis of a long bone [33]. However, resistance to bending is a 7th power function of cortical porosity and a 3rd power function of trabecular density, so the deleterious effect of microstructural abnormalities may offset any benefit achieved by greater bone width [37]. The ability of bone to deform without cracking decreases

as porosity increases [37,38]. The porosity may form stress concentrators predisposing a higher risk to fracture following repetitive loading [39]. Additionally, porosity may reduce compressive strength as the cross-sectional area of the cortex is less mineralized bone matrix and more void area [38].

Both deterioration in cortical and trabecular bone may result from impaired growth, particularly because cortical bone at metaphysis is partly formed by condensation of trabeculae in the periphery of the growth plate to form the cortex while trabeculae emerging from the center of the growth plate form metaphyseal trabecular bone.

Bone loss contributing to the deficits cannot be excluded but appears to be less likely given the age of the cases. Remodeling initiated upon Haversian canal surfaces excavate resorptive cavities enlarging the canal focally producing higher intracortical porosity and stress concentrators predisposing to micro cracks in cortical bone [39]. Even though remodeling may be still balanced in young adults, the refilling phase takes about 3 months, so resorption cavities upon trabeculae surfaces form stress concentrators which may predispose to microdamage in the face of repetitive strain [40]. The repetitive strains lead to accumulation of unrepaired microdamage ultimately producing a stress fracture [41–43].

This study has several limitations. First, it was a cross-sectional case control study. We cannot distinguish whether the porosity contributed to the stress fracture or whether the sequence of events was the reverse. However, measurements were performed within 14 days after injury on the contralateral side, which makes the latter less likely. Most of the cases were recreational sportsmen and women as were controls. We have no data concerning daily physical activity of the controls. Second, most of the male controls were from Australia. However, results were similar in females. Female cases and controls were Austrian. Third, the small number of cases, particularly in females, may have limited the power to achieve statistical significance of differences between cases and controls. Further investigation with a prospective study design and larger sample size are needed. The stress fractures in the cases were diagnosed in their lower limbs whereas differences in bone microstructure were measured at the distal tibia and the distal radius [26,44–46]. Despite this limitation Mikolajewicz et al., and others have reported fracture prediction using HRpQCT. [24,47,48]. Finally, the contribution of pore size and pore number to the reduction in total porosity was not evaluated.

#### **5. Conclusions**

In summary, stress fractures are common among young adults and are likely to be partly the result of deficits in cortical and trabecular bone microstructure. Whether these deficits have their origin established during growth, during advancing age or both requires further study.

**Author Contributions:** Study design: A.Z. Study conduct: A.Z. and E.S. Data collection: A.Z., L.F., A.G.-Z. and W.S. Data analysis and statistical calculations: A.Z., M.B. and E.S. Data interpretation: all authors; Drafting manuscript: A.Z. and E.S. All authors contributed to the development and critical revision of the manuscript and approved the final version for submission. A.Z. takes responsibility for integrity of the data analysis. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was not sponsored by any pharmaceutical company. Vinforce receives academic funding grants.

**Institutional Review Board Statement:** The study was approved by the local ethical committee (VINFORCE-013AZ) and conducted in accordance with the Declaration of Helsinki.

**Informed Consent Statement:** All patients agreed to participate to the study and signed an informed consent form.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors thank Colonel Christian Rizzi (Van Swieten Kaserne-Austrian Military Hospital Vienna/Austria) for the help in recruitment of several cases with stress fracture. We further thank Christina Marterer and Rahbarnia Arastoo for HR-pQCT scanning of the participants.

**Conflicts of Interest:** A.G.-Z. is one of the inventors of the StrAx1.0 algorithm. E.S. is an inventor of the StrAx1.0 algorithm and Directors of Straxcorp. All other authors state that they have no conflicts of interest.

#### **References**


## *Article* **Fragility Fractures and Imminent Fracture Risk in the Spanish Population: A Retrospective Observational Cohort Study**

**Maria-José Montoya-García 1, Mercè Giner 2,\*, Rodrigo Marcos 3, David García-Romero 3, Francisco-Jesús Olmo-Montes 4, Mª José Miranda 4, Blanca Hernández-Cruz 5, Miguel-Angel Colmenero <sup>4</sup> and Mª Angeles Vázquez-Gámez <sup>1</sup>**


**Abstract:** Fragility fractures constitute a major public health problem worldwide, causing important high morbidity and mortality rates. The aim was to present the epidemiology of fragility fractures and to assess the imminent risk of a subsequent fracture and mortality. This is a retrospective population-based cohort study (n = 1369) with a fragility fracture. We estimated the incidence rate of index fragility fractures and obtained information on the subsequent fractures and death during a follow-up of up to three years. We assessed the effect of age, sex, and skeletal site of index fracture as independent risk factors of further fractures and mortality. Incidence rate of index fragility fractures was 86.9/10,000 person-years, with highest rates for hip fractures in women aged ≥80 years. The risk of fracture was higher in subjects with a recent fracture (Relative Risk(RR), 1.80; *p* < 0.01). Higher age was an independent risk factor for further fracture events. Significant excess mortality was found in subjects aged ≥80 years and with a previous hip fracture (hazard ratio, 3.43 and 2.48, respectively). It is the first study in Spain to evaluate the incidence of major osteoporotic fractures, not only of the hip, and the rate of imminent fracture. Our results provide further evidence highlighting the need for early treatment.

**Keywords:** osteoporosis; fragility fracture; fracture risk; imminent fracture risk

#### **1. Introduction**

Fragility fractures caused by osteoporosis constitute a major public health problem worldwide. The annual costs attributable to fragility fractures in the European Union (EU) currently equate to €37 billion, and these numbers are expected to rise due to population aging [1]. Fragility fractures are an important cause of disability, morbidity, and mortality in the population [2]. Such massive burden highlights the importance of risk assessment of fragility fractures and the need to adapt prevention strategies to individual risk patterns.

A large number of risk factors for fragility fracture have been identified [3]. Among them, a previous fragility fracture has been generally recognized as an assessment criteria of fracture risk in osteoporosis, regardless of bone mineral density (BMD) [4]. Although it is generally recognized that the risk of fracture increases throughout life with a previous fracture, a recent osteoporotic fracture increases even more the risk of an imminent fracture [5–7], and the magnitude of this risk and the contribution of other clinical risk

**Citation:** Montoya-García, M.-J.; Giner, M.; Marcos, R.; García-Romero, D.; Olmo-Montes, F.-J.; Miranda, M.J.; Hernández-Cruz, B.; Colmenero, M.-A.; Vázquez-Gámez, M.A. Fragility Fractures and Imminent Fracture Risk in the Spanish Population: A Retrospective Observational Cohort Study. *J. Clin. Med.* **2021**, *10*, 1082. https://doi.org/ 10.3390/jcm10051082

Academic Editor: Heinrich Resch

Received: 5 February 2021 Accepted: 2 March 2021 Published: 5 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

factors still demand further research. Previous evidence suggests that the risk is not constant but rather fluctuates over time, being greatest within the first few years of the initial fracture [5,8–11]. The predictive importance of imminent (12 to 24 months) fracture risk is widely accepted. However, the significance of the early years after an index fracture need to be further explored. After a first fragility fracture, the skeletal location of the index fracture may also influence the magnitude of imminent fracture risk. However, only a few studies have measured its effect on the level of further fracture risk, and conclusions do not hold for solid generalization. [12–14]. Older age is also a well-defined clinical risk factor for fractures [15]. However, there is controversy between studies over its contribution on the risk of further fracture events. Some authors have observed a marked increase by age in the risk of a second major fragility fracture [11,16], whereas the association between age and subsequent fracture was not confirmed by other reports [3,17]. As with age, the predictive value of sex in the risk of further fracture events is also controversial. Originally, women were considered at higher risk of both initial and subsequent fracture and, indeed, some authors confirmed this hypothesis [11,16]. However, other studies have not reported differences in the risk of subsequent fracture events between men and women [18,19].

The primary objective of the present study was to explore the epidemiology of fragility fractures in southern Spain by using data from a population-based cohort of women and men, aged 50 or older, admitted to the emergency unit at Virgen Macarena University Hospital. Our study also provides an estimation of the incidence rate of subsequent fractures over 1–3 years following an index fracture. As secondary objectives, the study evaluated mortality over three years following an index fracture, analyzing the role of subsequent fractures as an independent risk factor.

#### **2. Materials and Methods**

This study was designed as a retrospective observational population-based cohort study involving men and women aged ≥50 years with an index fragility fracture (caused by an injury that would be insufficient to fracture normal bone; the result of a/or resistance to bone torsion [20]) occurring between 1 January 2014 and 31 December 2014.

Eligible study participants were followed until 31 December 2016, to obtain information on the outcomes: subsequent fragility fractures and deaths events. Age, sex, time from index fracture, and skeletal site were assessed as potential independent risk factors. Patients were identified using the emergency unit's medical records at Virgen Macarena University Hospital in Spain. Virgen Macarena University Hospital is a public tertiary hospital and its Emergency Unit serves up to 481,879 inhabitants, of whom 157,428 subjects are aged ≥ 50 years, within the healthcare area of North Seville. Furthermore, this is the single reference hospital of the healthcare area including emergency care attention. General medical records (Diraya) were also used to collect any relevant information on the study outcomes including demographic information, index fracture type, and causality as well as any relevant radiological confirmatory findings. Eligible incidental fragility fracture locations included axial (hip, pelvis, dorsal, and lumbar vertebrae) as well as appendicular bones (proximal humerus and wrist), according to the International Classification of Diseases, ICD-9 codes (Supplementary Table S1). Diagnosis was based on symptoms but must have included a radiologic confirmation of the fracture. Non-clinical radiographic vertebral fractures as well as other pathological or traumatic fractures were excluded. The identification of subsequent fractures required a similar main diagnosis of fracture. To distinguish subsequent fractures from previous events recorded at follow-up visits and/or patient history, the following criteria were applied: (1) Fractures in the same skeletal site of index fracture were only captured if a minimum of four months had elapsed since the index fracture; (2) hip fractures were only captured if an inpatient hospital admission was required; (3) all medical visits identified as follow-up examination of a previous fracture were excluded as further fracture events; (4) patients who died following a fracture were captured as having both outcomes; and (5) if the index fracture involved more than one skeletal sites, to avoid double counting, the fracture was assigned to the site of highest

severity. Time at risk for subsequent fracture events began the day after the date of the index fracture and continued until outcome occurrence, either fracture or patient death.

We estimated the incidence rate over 12 months of (index) fragility fractures in the general population aged ≥ 50 years (based on a total estimated population of 157,428 individuals aged 50 or over served in Virgen Macarena Hospital catchment area on 1 January 2014). Then, we estimated the incidence rate of further fracture events during the study period among those who had a previous fragility fracture in 2014. Time at risk for subsequent fracture events began the day after the date of the index fracture and continued until outcome occurrence, either fracture or patient death. Fracture incidence rates per 10,000 person-years were calculated by age group, sex, and fracture type. The 95% confidence intervals (CIs) were calculated assuming a Poisson distribution. The excess risk of further fractures was compared to the general population using a Poisson regression model including age, sex, and location of previous fracture as covariates. Nelson–Aalen cumulative hazard estimates were plotted to analyze the time to a subsequent fracture event [21]. The risk of subsequent fracture was analyzed using a Cox proportional hazards regression model, as well as the model proposed by Fine and Gray [22], using death as a competing risk. All-cause mortality was also analyzed using the Cox hazard model. Furthermore, we included a time-dependent variable in this model to estimate the risk of all-cause death associated with the occurrence of subsequent fracture events during the study period. Estimates with *p* values < 0.05 were considered statistically significant. No imputation of missing data was necessary. All statistical analyses were performed using Stata software (STATA Corp., College Station, TX, USA).

#### **3. Results**

#### *3.1. Patient Baseline Characteristics*

Among a total population of 157,428 Caucasian individuals aged 50 or over served by Virgen Macarena University Hospital, 1068 women and 301 men (3.5 female/male ratio), with mean of age 75.1 and 72.1 years, respectively, registered eligible index fragility fractures in 2014 and were included in the analyses. Only 14 subjects were excluded due to miscoding of fractures (3/14) or traumatic (8/14) or pathological fractures (3/14), (Figure 1). The most frequent index fracture site in women was wrist (405 [37.9%]), whereas in men hip fractures were the most common (111 [36.9%]). Mean duration of follow-up was 2.3 years for all subjects, 2.2 years in males and 2.3 in females.

**Figure 1.** Flow diagram of study participants based on the Strobe statement. A total of 1383 subjects were assessed for eligibility, among a study population of 157,428 inhabitants. Fracture miscoding or traumatic or pathological causes of fractures were the causes for the exclusion of 14 subjects. Subjects were enrolled from January 2014 to December 2014 and followed up until December 2016. Mean duration of follow-up was 2.3 years for all subjects, 2.2 years in males and 2.3 in females. The most frequent types of index fracture were wrist (N = 506, 37.0%) and hip (N = 431, 31.5%), followed closely by proximal humerus (N = 264, 19.3%). Overall, 99 subjects registered a subsequent fracture, with 120 events of death occurring throughout follow-up. Fx: fracture. SsFx: subsequent fracture.

#### *3.2. Incidence Rate of Index Fracture in the General Population*

An overall incidence rate of 86.9 fractures/10,000 person-years was found in the general population aged ≥ 50 years (Table 1). The frequency of fragility fractures was significantly higher among women as compared to men: 123.9 (CI 95% 116.6–131.6) versus 42.3 (CI95% 37.7–47.4), respectively. Also, marked increase in the frequency was also observed with increasing age. The highest overall frequency was observed for wrist fractures. Despite this, hip fractures were the most frequent among women and men aged ≥ 80 years. On the other hand, the skeletal sites with the lowest incidence rate of fragility fractures were the pelvis and the spine.

**Table 1.** Incidence rate of index fracture in the general population aged ≥50 years, by index fracture site, sex, and age group. The total population values for males and females are marked in bold. The shading indicates that they are total values of the population, to differentiate them from the rest that are separated by gender and age.


Data are incidence rate per 10,000 person-years (95% confidence interval). \* Age is described as of index date.

#### *3.3. Incidence of Subsequent Fractures*

The frequency of clinical subsequent fracture events was 318.2/10,000 person-years. Incidence rate of subsequent fractures was higher in women than in men. However, the overall differences did not reach statistical significance (Table 2). By contrast, the frequency of clinical fracture events increased with age, with markedly higher incidences rates in men and women aged 70 years or older (Table 2). Overall, no significant differences in the frequency of subsequent fracture events were observed by index fracture type. However, a slight trend was observed toward increased incidence among subjects with pelvis and hip index fractures (Table 2). The rate of further fracture events was highest within the ≥80 years' age group, in women with a previous fracture in the pelvis and men with a previous wrist index fracture. The most frequent skeletal locations of further fractures were hip and wrist. The incidence rate of further fracture events during the first year after the index fragility fracture was not higher than the rates observed during the following second and third years of follow-up, regardless of sex and age (Figure 2). Similarly, no marked differences among follow-up periods were observed by site of index fracture. However, a slight trend was observed for wrist fractures (Supplementary Table S2).


**Table 2.** Incidence rate of subsequent fractures (any site) among patients with an index fracture, by index fracture site, sex, and age group. The total population values for males and females are marked in bold. The shading indicates that they are total values of the population, to differentiate them from the rest that are separated by gender and age.

Data are incidence rate per 10,000 person-years (95% confidence interval). \* Age is described as of index date.

**Figure 2.** Nelson–Aalen cumulative hazard of subsequent fracture events after an index fracture in men and women aged ≥50 years over a period of three years of follow-up. Cumulative risk of subsequent fracture increased over the years following initial fracture. However, no significant differences were found in the incidence rate of subsequent fractures over 1, 2, and 3 years, regardless of sex and age. Dotted line represents the cumulative incidence of subsequent fractures in all subjects. Solid line represents the cumulative incidence of subsequent fracture events in females. Dashed line represents the cumulative incidence of subsequent fracture events in males. SsFx: subsequent fracture. The Y-axis represents the cumulative hazard function. The X-axis represents follow-up years after the first fracture.

#### *3.4. Risk Factors of Subsequent Clinical Fracture*

Overall, the incidence rate of fractures was higher for subjects with a previous index fragility fracture at any site compared with the general population (relative risk [RR] 1.80, *p* < 0.01) (Figure 3 and Supplementary Table S3). Independent risk factors for subsequent fracture, as identified by multivariate analysis using Cox as well as Fine and Gray regression models, Higher age (≥70 years) was an independent risk factor for further fracture events, with a ≥1.5 increase in hazard risk (HR) observed for each decade from 60 years of age (Table 3). Multivariate analysis using Fine and Gray model revealed an increased risk of subsequent fractures in women. No effect of index fracture site on the level of risk of further fractures was observed (Table 3).

**Figure 3.** Relative risk of subsequent fractures among subjects with a previous fracture compared with the general population by sex (green box and square), age group (red box and circle), and index fracture site (blue box and triangle). Adjusted rate ratio estimated using Poisson regression models that included age, sex, and index fracture site as covariates. Corresponding data are presented in Supplementary Table S3. \* *p* < 0.05, \*\* *p* < 0.01.

**Table 3.** Hazard ratio (Cox) and subhazard ratio (Fine and Gray) of subsequent fracture events associated to sex, age, and index fracture site.



**Table 3.** *Cont.*

\* Baseline category; † hazard ratio (HR) and 95% confidence interval (CI) estimates adjusted for all variables in the table using a Cox proportional hazards model. ‡ Subhazard ratio (SHR) and 95% confidence interval (CI) estimates adjusted for all variables in the table using a Fine and Gray competing risks model using death as competing risk. <sup>+</sup> Appendicular fractures: wrist and proximal humerus. Central fractures: vertebral, pelvis, and hip fractures.

#### *3.5. Risk of Death Following Initial Fracture*

A total of 120 deaths occurred during the study follow-up period in patients aged ≥50 years with an index fragility fracture, with overall mortality rates reaching 37.20/1000 person-years (29.09 and 67.42 per 1000 person-years in women and men, respectively). Mortality rates were higher among men as compared to women (HR, 0.41; *p* < 0.01) (Table 4). Age was the strongest determinant of mortality with significant excess risk for subjects aged 80 or older (HR, 3.43 *p* < 0.01). Mortality risk was also significantly higher among patients presenting an index fragility fracture in the hip (HR, 2.48; *p* < 0.01). Lower mortality rates were observed in subjects with index fractures located in peripheral bone positions (wrist and proximal humerus). Mortality risk also increased after a subsequent fracture occurred, although this association did not reach statistical significance (HR, 2.14; *p* = 0.06) (Table 4).

**Table 4.** Mortality hazard ratio associated to sex, age, type of index fracture, and presence of subsequent fracture events.


\* Baseline category. † Mortality hazard ratio (HR) and 95% confidence interval (CI) estimates adjusted for all variables in the table using a Cox proportional hazards model. ‡ Mortality rate (MR) per 1000 person-years (PY). SsFx: subsequent fracture.

#### **4. Discussion**

This study presents the first report of the incidence rate of index of major fragility fractures and the risk of imminent fractures in a Spanish cohort of 1369 subjects (1068 women and 301 men) aged ≥50 years, by age group and fracture site. Our study confirmed

markedly higher rates in women, as well as an age-related increase in the risk, with highest frequency rates found in women aged ≥80 years. The prevalence of osteoporosis and osteoporosis fracture rates is higher in women compared to men. This is due to differences in BMD, bone size, bone geometry, and bone strength [23,24]. Estrogen deprivation after menopause is a major contributing factor, which could be the reason for the observed gender-related differences [25]. Age, on the other hand, is a well-studied risk factor of index fragility fractures, contributing to risk independently of bone mass density [26]. Previous information on the incidence rate of fragility fractures in Spain is scarce. As a first approach, using the Q-FRACTURE tool, González López-Valcárcel et al. [27] estimated a level of risk ranging from 1.8–21.5% in women and 0.7–10.8% in men. According to our findings, the frequency of osteoporotic fractures in Spain may be sensitively higher than reported. Conversely, our numbers underestimate the crude rates published by the International Osteoporosis Foundation for Spain as well as for other EU countries [28]. This discrepancy may be partly explained by the exclusion of non-clinical vertebral fractures and other fracture sites less commonly associated with osteoporosis. Overall, the most frequent index fracture type was wrist. However, the rate of hip fractures exceeds that of wrist in older aged groups. Similar age-related trends in the frequency of fragility fractures were reported previously [29].

Estimated incidence rate of subsequent fractures was 318.2/10,000 person-years in all subjects during the three years that followed index fracture (i.e., 3.2% of patients with a previous fracture experienced a new fracture every year). To date, no other studies have been published that measure the risk of imminent fractures after a sentinel fracture in Spain. Only Azagra et al. have published 10-year fracture data in a population cohort in Catalonia that presented clinical risk factors for osteoporotic fractures, with the aim of validating the Frax tool in the Spanish population [30].

Overall, the incidence of fracture was higher for subjects with a previous index fragility fracture at any site, compared with the frequency in the general population aged 50 or older (RR, 1.80). According to our findings, Kanis et al. [17] observed that, for any type of previous fracture, the RR of any further fracture ranged from 1.83 to 2.03 depending upon age. The effect of gender as a predictor of the risk was only significant when the analysis considered death as a competing risk. Previous reports observed similar risks in men and women, except among subjects over 85 years of age [31]. Our data also, however, proved the well-known independent effect of aging on the risk of further fracture events, [11,14], with significant differences in the HR among older age (≥70 years). They also found a marked age-related increase in the risk but did not observe any differences in the risk among women and men.

Noticeable differences in the frequency of subsequent fractures were found depending upon site, with highest rates found in subjects with a previous pelvis or hip fracture. Our findings, however, could not confirm the effect of the skeletal site on the risk of further fractures. Previous reports on the associations between prior and subsequent fractures are not consistent [7]. The time that follows initial fracture is key with regard to the risk and prevention of subsequent fracture events. Several previous studies report that the highest risk of further fracture events occurs within the first year after the index fracture [8–11] and that the incidence decreases thereafter. In the current study, however, the incidence of subsequent fracture events during the first year after the index fracture was not higher than the rates observed during the following second and third years of follow-up, regardless of sex and age. Despite this, a slight trend was observed for wrist index fractures. Authors reporting higher levels of risk during the first year after the index fracture have assessed longer timescales than our study [9]. Like in our study, Banefelt et al. [11] focused only in the early years following the index fracture and found the highest incidence during the second year (12%), rather than the first (7.1%).

Overall mortality rate reached 37.20/1000 person-years, which showed an up-to-2 fold excess mortality due to osteoporotic fractures among the younger-aged groups [32]. The risk of death was significantly higher in patients with a previous hip index fracture

(HR, 2.48; *p* < 0.01), with an estimated rate of 73.50/1000 person-years. No significant excess mortality was found for index fractures in other skeletal locations or for the event of subsequent fractures (*p* = 0.06). As expected, our data also confirmed a higher age-specific death risk (*p* < 0.01, in subjects aged ≥80 years) as well as lower death risk in women (*p* < 0.01). In fact, 58.3% of all deaths occurred in patients with a previous hip index fracture and a mean age of 83.3 years. The observed death rate after a hip fracture was sensitively lower than previously reported [33,34], which could be explained by the longer observation period of this study, as mortality is highest during the six months that follow the event [35].

The main limitation of our study is the lack of data from other clinical factors that could have contributed to in-depth understanding of the risk of subsequent fractures (bone mass density, previous record of falls, history of prior fractures, use of drugs affecting bone metabolism). One of the strengths of our study is that, as opposed to database studies, we manually reviewed clinical records of all 1369 cases to confirm eligibility as well as outcomes' information.

#### **5. Conclusions**

In summary, this report provides information on the magnitude and consequences of fragility fractures in the as-yet unexplored Spanish population aged ≥50 years, involving both genders, as well as major skeletal sites associated with osteoporosis, not limited to hip. This is the first study to report on the incidence rate of imminent fractures after an osteoporotic sentinel fracture in Spain, with age, sex, and skeletal location of the index fractures as possible risk factors. Our results support the increased risk of imminent fracture after a recent fracture and provide key elements for early identification of risk and the application of targeted strategies aimed at preventing future fractures and mortality, such as the interventions recommended by the Fracture Liaison Services [36].

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2077-038 3/10/5/1082/s1. Table S1: Fracture codes according to ICD-9, Table S2: Incidence rate of subsequent fracture events by type of index fracture and follow-up period (years), and Table S3: Relative Risk of subsequent fracture events by age, sex and index fracture site.

**Author Contributions:** Conceptualization, M.G. and M.-J.M.-G.; methodology, M.-J.M.-G. and M.A.V.-G.; validation, F.-J.O.-M., M.-A.C., M.J.M. and B.H.-C.; formal analysis, M.G. and M.-J.M.-G.; investigation, all authors; resources, R.M. and D.G.-R.; writing—original draft preparation, M.G. and M.-J.M.-G.; writing—review and editing, M.-J.M.-G. and M.A.V.-G.; supervision, M.A.V.-G.; project administration, M.-J.M.-G.; funding acquisition, M.-J.M.-G. and M.-A.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by Consejería de Salud de la Junta de Andalucía, Proyecto de Innovación PIN-0092-2016, which had no role in the design or conduct of the study; in the collection, analyses, and interpretation of the data; or in the preparation, review, or approval of the manuscript.

**Institutional Review Board Statement:** The study was approved by the Ethical Review Board of Seville (protocol code: CI#2147, 09/02/2013) and was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments. For this type of study, individual formal consent was not required.

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy.

**Acknowledgments:** We thank the participants in this study for their valuable contribution. Writing assistance was provided by Juliana Martinez.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

