Hip Geometry in Hip Fracture Patients in Greenland


 Background: Hip geometry influences hip fracture risk. Hip fractures are common and they are associated with pain, disability, premature death and marked costs on society. Osteoporotic fractures are frequent in Arctic populations and increase with advancing age in this society that has a steep rise in life expectancy. Greenland Inuit is a distinct ethnic group and data on hip geometry is missing. This let us to describe hip geometry in populations with hip fracture in Greenland.Methods: We evaluated collodiaphysial angle, femoral neck length, the outer and inner diameter of the femur at 2 and 5 centimetres below the centre of the lesser trochanter and the cortical thickness from pelvic and hip radiographs in all patients operated in Greenland over a 7.7-year period. We included all 84 patients with one non-fractured hip visible for geometric analysis. Analyses were conducted in duplicate.Results: We found collodiaphysial angle of 134.8/132.6o in men/women (p=0.06) and femoral neck length of 38.0/33.9mm in men/women (p=0.001). Cortical thickness was influenced by sex in the adjusted analysis (p<0.001). Cortical thickness index at 5cm below the centre of the lesser trochanter decreased with age (p=0.026) and may be influenced by height (2cm below the centre of the lesser trochanter, p=0.053). Conclusion: Our findings differed from European reference data and suggest a delicate balance in hip geometry that may change with lifestyle transition in Arctic populations. Ethnic peculiarities influence the hip structure and may have an impact on fracture risk. A focus on hip geometry and risk factors for osteoporotic fractures in Arctic populations is warranted.


Background
Fracture risk differs with ethnicity, and ethnic differences in hip fractures risk between South African Blacks and Caucasians Whites were attributed to differences in hip geometry [1,2,3]. Similarly, ethnic differences exist between Inuit and Caucasians in bone metabolism [4] while hip geometry in populations in Greenland remains unknown.
Hip fractures are important because they associated with pain, disability, premature death and marked costs on society [5]. They are associated with the highest frequency of physical deterioration and death. The occurrence of hip fractures accelerates with age they are the predominant fracture in the 8th decade [6]. Thus, the steep increase in life span among Arctic inhabitants calls for attention on the occurrence of hip fractures in the Arctic ageing populations.
Vitamin D levels, that are likely to be in uenced by Arctic habitat [7], in uence fragility fracture risk. This may affect the risk of fractures in Greenland. Also, Inuit is a distinct ethnic group with shorter limbs and a larger torso compared to Caucasians as illustrated by differences in sitting-height/height ratio and BMI [8,9,10]. Whether these differences carry through to hip geometry is unknown.
The powerful endocrine mechanisms that control bone mineralisation differ between Inuit and Caucasians [4].
Ethnic differences occur in bone mineral density (BMD) [11] but BMD did not differ between Inuit and Caucasians in Greenland when adjusting for differences in body weight [12]. Still, fragility fractures were frequent among old women in Greenland [13,14]. Finally, complying with hip geometry when performing hip fracture surgery is important for the outcome [15], and knowledge of hip geometry is necessary.
This led us to study hip geometry and cortical thickness parameters among subjects in Greenland.

Setting
Greenland is the world largest island and it is sparsely populated along the vast coastline. The total population of around 55,000 is mainly Inuit. The country hosts one orthopaedic department at the national hospital in the capital Nuuk. All hip fracture patients are transferred to Nuuk for surgery.

Radiographs
We retrieved radiographs of patients admitted to the Orthopaedic Department at Queen Ingrid's Hospital in Nuuk, Greenland for a hip fracture over 7.7-years. The study period was from January 1st 2007 through September 1st 2015. Radiographs were transferred to a separate le in Nuuk for evaluation of hip geometry. Radiographs included in the analysis had to ful l criteria set up prior to analysis of the radiographs. These criteria were one non-fractured hip that was readily visible for geometric analysis to at least 5 cm below the prominent tip of the lesser trochanter.
Eighty-four patients ful lled the criteria for inclusion in the analysis. The reason for exclusion from the analysis was mainly lack of radiographs of the pelvis as the routine was to take radiographs of the pelvis and the fractured hip only.

Analysis of radiographs
Radiographs were evaluated on a client review workstation (IMPAX 6.5. Solution). Measurements were conducted on plain anteroposterior radiographs of the pelvis, hip, or femur as illustrated in Fig. 1. All measurements were performed in duplicate. They were performed by a single evaluator (IF). The two measurements were done separately with at least four weeks between measurements. The second measurement was blinded to the results of the rst measurement. Measurements were conducted as described previously [2] and detailed in Fig. 1.
We evaluated collodiaphysial angle. This was done by rst identifying the centre of the femoral head and drawing the line through this centre parallel to the femoral neck ( Fig. 1). Second, the diaphyseal line was drawn through the centre of the femoral diaphysis. Third, the angle between these two lines was determined to depict the collodiaphysial angle ( Fig. 1). Femoral neck length was determined by measuring the distance from the centre of the femoral neck to the line drawn between the most prominent tip of the lesser and the upper tip of the major trochanter ( Fig. 1). The outer and inner diameter of femur was measured at 2 and 5 centimetres below the centre of the lesser trochanter ( Fig. 1). We calculated the cortical thickness as the sub-periosteal width minus the endocortical width divided by two. The cortical thickness index was then calculated as the cortical thickness divided by the outer diameter. The cortical thickness index was calculated to account for differences in radiographic magni cation and varying femoral size. Measurements were performed on the non-fractured femur at 2 and 5 cm below the most prominent tip of the lesser trochanter. Calibration was conducted individually for each radiograph.
Data retrieval and analysis were conducted after approval by the Ethics Committee for Scienti c Research in Greenland and individual consent was not required (ethics committee reference no 2013-16).

Statistical analysis
Frequencies are given in both mean and standard deviation (SD) as well as median with 25 and 75 percentiles for descriptives of the participants. These are compared using the t-test as all variables follow the normal distribution.
The data are further tested in linear regression analysis with collum length, collodiaphysial angle, and cortical thickness and cortical thickness index at 2 and 5 cm below the tip of the lesser trochanter entered as dependent variables. Explanatory variables entered are age, sex, height and weight. Regression analysis was performed rst as univariate analysis and subsequently as multivariate analysis except for collodiaphysial angle as none of the variables in the univariate analysis in uenced the dependent variables.

Results
The study included 56 (66.7%) women and 28 (33.3%) men. Characteristics of the patients included in the study are given in table 1. The sex-difference in age was limited while a marked difference was seen in height and weight (table 1). BMI was relatively low for Inuit and did not differ with sex. Table 2 gives the measured and calculated geometric data for the proximal femur. The gender difference in collodiaphysial angle (mean, 134.8/132.6 degrees in men/women) was limited in the direct comparison (table 2).
Femur neck length differed with sex (p = 0.001) as shown in table 2. Outer diameter differed with sex at two and ve cm below the prominent tip of the lesser trochanter (both, p < 0.001) while the inner diameter differed with sex at only two cm (p = 0.028). The cortical thickness showed distinct gender differences at two and at ve cm below the tip of the lesser trochanter (both, p < 0.001), while the cortical thickness index differed most markedly at ve cm below the lesser trochanter (p = 0.003). Table 3 lists the univariate and multivariate analysis of factors important to collodiaphysial angle, collum length, cortical thickness and cortical thickness index. Collodiaphysial angle differed between men and women in the adjusted analysis (Table 3). Collum length was in uenced by height, weight, sex and age in the adjusted analyses.
Cortical thickness was in uenced by age, sex, height and weight at both 2 cm and 5 cm below the lesser trochanter in the unadjusted analysis while sex was the dominant factor in the adjusted analysis. Cortical thickness index was in uenced by age, sex and height at both sites in the unadjusted analysis while only by the height at 2 cm in the adjusted analysis and only by age at 5 cm below the lesser trochanter. This decrease is illustrated in Fig. 2 (adjusted comparison, p = 0.026).
The results of the repeated measurements of the radiographs did not show any systematic error in outer diameter or inner diameter, collum length, collodiaphysial angle, CT or CI (all, ns).

Discussion
This is the rst systematic report of hip geometry in Inuit. We found the major in uence on hip geometry to be related to height, sex and age while an in uence of weight was limited. The impact of age on the cortical index was marked and the in uence of sex and height on cortical thickness was marked. Height had a major impact on collum length and, interestingly, age also in uenced collum length.
Measurements of hip geometry can be performed on plain radiographs [16]. Such data may provide important information to support fracture risk assessment [17,18,19,20] and osteoporosis evaluated from bone mineral density measurements [16]. Sah and colleagues reported cortical index measured on plain radiographs in patients with and without osteoporosis diagnosed by DXA in the same patients [16]. They found a marked difference in cortical index between non-osteoporotic and osteoporotic patients (0.55 vs 0.46) at 3 cm below the lesser trochanter [16]. This supports the impact of the cortical index on the risk of osteoporosis, and other measures of the cortical structure may be relevant to include to have enough geometry for a reliable strength estimate.
Beck reported data from NHANES III including cortical index and subperiostal cortical thickness [21]. The cortical index was 0.480 in men and 0.377 in women at 2 cm below the lesser trochanter [21]. We found a markedly lower cortical index of 0.160 and 0.150 respectively. Furthermore, Beck and colleagues reported subperiostal diameter of 3.62 and 3.20 cm in men and women respectively similar to the 3.15 in Caucasian women reported by Nelson [22].
We found lower values of 3.26 and 2.90 cm in men and women respectively. The values in our population from Greenland were low. Yet, the values reported from NHANES III and by Nelson were based on the general population [21,22] while our data are from patients with a fragility fracture at the hip. This lower femoral neck cortical thickness in hip fracture patients compared to individuals without fracture is in keeping with previous reports on the in uence of ethnicity on hip geometry and fracture risk [23]. Thus, our data support the notion that cortical thickness has the potential to be used for hip fracture risk assessment.
Koeppen and colleagues reported cortical index values among patients with femoral subtrochanteric or shaft fractures [24]. Their population is thus parallel to our patients except for the Swedish origin. Their fracture patients had a cortical index of 0.37 at 5 cm below the lesser trochanter. It was 0.14 higher than our patient's cortical index of 0.23. The cortical index decreased from the general Caucasian population to osteoporotic Caucasians [21,24] and decreased further to Inuit with an osteoporotic hip fracture. This lower cortical index among hip fracture patients in Greenland may suggest that other parameters than cortical index are at play and should be identi ed and included when aiming to provide a reliable estimate of bone strength and fracture risk.
Nelson reported the outer/inner diameter of the femoral shaft at 2 cm below the lesser trochanter to be 3.15/2.05 cm in White and 3.14/1.99 cm in Black women [22]. We found values of 2.90/2.04 cm. Thus, the cortical thickness was 0.43 cm in the patients from Greenland compared to 0.55 cm in White and 0.57 cm in Black Americans in the study by Nelson [22]. The markedly lower cortical thickness in our ndings from Greenland further supports the notion that other parameters may in uence hip fracture risk among patients in Greenland.
The collodiaphysial angle was reported previously to be between 124 and 129 degrees [1]. This angle was 133 degrees in our study. Such a higher angle would suggest a lower risk of hip fracture in the population in Greenland compared to Whites and Blacks. This may thus be one factor to add for obtaining a more reliable estimate of femoral fracture risk.
Finally, femoral neck length has been reported to be between 4.3 and 4.7 cm in non-Inuit women [1]. Ethnic differences in femoral neck length contributed to differences in hip fracture risk [25]. We found a femoral neck length in women to be 3.39 cm in Greenland hip fracture patients. This 20% shorter femoral neck is more than can be explained by the 6% difference in height between Caucasian and Inuit populations [9,10]. Our nding among hip fracture patients suggest shorter femoral neck among the general population in Greenland, and this could be an important hip structure parameter in the description of hip fracture risk.
The higher BMI in Inuit compared to Caucasians [9,10] could contribute to a lower risk of fragility fractures at the hip as weight in uenced both geometric strength and hip fracture risk [26]. This is thus an additional factor to include in the estimate of the in uence of ethnicity on fracture risk.
Data were collected in Greenland, which is Arctic environment. This area provides limited sun exposure and heavy clothing, and low vitamin D levels may be expected. However, the traditional Greenlandic diet is rich in vitamin D and populations with some intake of this diet and populations as far as 400 kilometres north of the Arctic Circle are not vitamin D de cient as dermal vitamin D production adds to the levels supplied by the diet [27,28,29]. Still, the powerful hormonal mechanisms that regulate calcium metabolism differ between Native and European populations [4]. Hence, ethnic differences are likely to apply to other aspects of skeletal health and may apply to bone structure and strength in populations in the Arctic.
Greenland is in a transition from a hunter to modern society. Life expectancy is only around 70 years, but it is rising steeply [30]. This in uences the risk of osteoporosis as vertebral fractures are present and frequent [13,14,31]. Hip fractures occur one decade later in life than vertebral fractures and a steep rise in hip fracture frequency may be expected based on the ageing population in the Arctic. Our ndings support careful monitoring of hip fracture occurrence and development of a fracture risk assessment method applicable to Arctic populations. The vast geography and logistical constraints in the Arctic necessitate the use of local technology, and analysis of femoral geometry from hip radiographs may provide an opportunity to raise health care service in rural areas.
The number of patients included in our study was limited. However, we included all hip fracture patients operated in Greenland over 7.7 years. This is in keeping with a low frequency of hip fractures seen in a population with a median life expectancy of approximately 70 years. There is only one hospital with expertise in orthopaedic surgery. Thus, hip fracture patients from all of Greenland are transported to the main hospital in the capital Nuuk for surgery. This logistic challenge and economic burden on health care will increase with the predicted rise in hip fracture frequency.
A limitation to our study is the lack of a control group. However, our ndings among hip fracture patients are comparable to ndings by others in similar patients and our results thus provide some insight into the risk of hip fracture among Arctic populations and should encourage further data on this topic.
In conclusion, we found a higher collodiaphysial angle and shorter femoral neck compared to non-Inuit suggesting a lower hip fracture risk. Conversely, we also found smaller outer diameter with a similar inner diameter and thus a lower cortical index compared to non-Inuit suggesting a higher hip fracture risk. Our ndings have opposite effects on fracture risk, and they may constitute a delicate balance. This may change with the ongoing lifestyle changes and thus add to the risk seen with the rise in life expectancy. In addition, ethnic peculiarities are likely and may in uence fracture risk prediction. Hence, a focus on hip geometry, bone metabolism and risk factors for osteoporotic fractures in Arctic populations is warranted.

Declarations Ethics approval
Data retrieval and analysis were conducted after approval by the Ethics Committee for Scienti c Research in Greenland and individual consent was not required (ethics committee reference no 2013-16).

Consent for publication
All authors have read the nal version and agreed to publication of the manuscript.

Availability of data
Data are made available upon request to the corresponding author.

Competing interests
This study poses no competing interests.