Abstract
Background: Bone fractures are common in alcoholics. Aims: To analyse which factors (ethanol consumption; liver function impairment; bone densitometry; hormone changes; nutritional status, and disrupted social links and altered eating habits) are related to bone fractures in 90 alcoholic men admitted to our hospitalization unit because of organic problems. Methods: Bone homoeostasis-related hormones were measured in patients and age- and sex-matched controls. Whole-body densitometry was performed by a Hologic QDR-2000 (Waltham, MA, USA) densitometer, recording bone mineral density (BMD) and fat and lean mass; nutritional status and liver function were assessed. The presence of prevalent fractures was assessed by anamnesis and chest X-ray film. Results: Forty-nine patients presented at least one fracture. We failed to find differences between patients with and without fractures regarding BMD parameters. Differences regarding fat mass were absent, but lean mass was lower among patients with bone fracture. The presence of fracture was significantly associated with impaired subjective nutritional evaluation (χ2 = 5.79, P = 0.016), lower vitamin D levels (Z = 2.98, P = 0.003) and irregular eating habits (χ2 = 5.32, P = 0.02). Reduced lean mass and fat mass, and altered eating habits were more prevalent among patients with only rib fractures (n = 36) than in patients with multiple fractures and/or fractures affecting other bones (n = 13). These last were more closely related to decompensated liver disease. Serum vitamin D levels showed a significant relationship with handgrip strength (ρ = 0.26, P = 0.023) and lean mass at different parts of the body, but not with fat mass. By logistic regression analysis, only vitamin D and subjective nutritional evaluation were significantly, independently related with fractures. Conclusion: Prevalent fractures are common among heavy alcoholics. Their presence is related more closely to nutritional status, lean mass and vitamin D levels than to BMD. Lean mass is more reduced, nutritional status is more impaired and there is a trend to more altered eating habits among patients with rib fractures, whereas multiple fractures depend more heavily on advanced liver disease.
INTRODUCTION
Bone loss and fractures are common events in alcoholics, due in part to both direct and indirect effects of ethanol on bone remodelling (Diamond et al., 1989; Leslie et al., 2003; Peris et al., 1995; Spencer et al., 1986; Wezeman et al., 2000; ) in a dose-dependent fashion (Turner, 2000), but also to the peculiar style of life of the alcoholic, prone to falls, traffic accidents and violence. The term ‘Battered alcoholic syndrome’ was coined >30 years ago by Oppenheim (1977) to define the situation of a chronic alcoholic who presented with at least three bone fractures, at different sites and in different healing stages.
Among alcoholics, probably rib fractures constitute the most frequently observed ones (González-Reimers et al., 2005), and in many cases, the patients are not aware of them, partly due to the anaesthetic effect of ethanol, but also because of inebriation. Although rib fractures have not been considered as classic osteoporotic fractures, in a large study including nearly 6000 elderly men, rib fractures were the most commonly observed ones, and they were related in most cases to falls from the standing position (Barrett-Connor et al., 2010). Importantly, hip bone density was an independent risk factor for rib fracture, and also, rib fracture was an independent risk factor for a new hip fracture. Therefore, at least in elderly men (Barrett-Connor et al., 2010), rib fracture should be considered as an osteoporotic fracture, in a similar way as wrist or hip fractures. A similar conclusion was reached by Ismail et al. (2006), especially among women. However, falls from standing height being the main cause of rib fracture, inebriation, muscle atrophy and/or neuropathy, and the peculiar style of life of the alcoholic patient surely play a role in the high prevalence of these fractures. For instance, Keso et al. (1988) found that thoracic fractures were more commonly observed among unmarried, divorced or widowed alcoholics. However, in addition to these factors, it is clear that heavy prolonged ethanol consumption also constitutes a risk for classic osteoporotic fractures, such as hip fracture (Felson et al., 1988). Indeed, ethanol exerts direct effects on bone synthesis and resorption, leading to decreased bone mass. Besides these direct effects, altered nutrition (Santolaria et al., 2000), altered liver function (Chappard et al., 1991; Jorge-Hernández et al., 1988) and ongoing drinking (Alvisa-Negrín et al., 2009; Peris et al., 1994) also play a role.
The main objective of this study was to analyse which factors (ethanol consumption; liver function impairment; bone densitometry; hormone alterations; nutritional status and environmental factors related with job, social status and eating habits) are related to bone fracture in a cohort of 90 heavy-alcoholic men admitted to our hospitalization unit because of organic problems. Since, as commented, subtle differences, largely depending on the style of life and social environment of alcoholics, may exist between the mechanisms leading to rib fractures and the classical osteoporotic fractures in these patients, we have compared all the data mentioned before among patients with only rib fractures and patients with other types and/or multiple fractures.
PATIENTS AND METHODS
Patients and controls
We included 90 alcoholic patients, aged 50.14 ± 10.99 years (median = 49; inter-quartile range (IR) = 42–58), all of whom were heavy drinkers of more than 100 g ethanol/day (212 ± 79 g/day) for prolonged time periods (29.4 ± 9.7 years) until the day of admission, and 30 age-matched controls, who were sanitary workers and drinkers of <10 g/day. The total amount of ethanol consumed by the patients until the inclusion in the study was 33.88 ± 16.36 kg ethanol/kg body weight (median = 33; IR = 19–50). In 50 cases, serology for hepatitis C virus (HCV) and hepatitis B virus (HBV) was performed, being positive in 10 cases for HCV and in 4 for HBV (anticore antibodies; surface antigen was negative in all cases). Patients were also categorized into cirrhotics and non-cirrhotics according to clinical grounds and ultrasound findings, and to liver biopsy when clinically indicated. A total of 40 cirrhotics were included; Child's classification (based on the presence and characteristics of ascites and encephalopathy, and on serum albumin, bilirubin and prothrombin activity) was used to assess severity. According to this score, 2 patients belong to Child A (less severe), 13 to Child B (moderately severe) and 25 to Child C (most severe) groups. Strictly speaking, the Child-Pugh score is only applicable to cirrhotics. However, alcoholic liver disease is an ongoing process, and decompensated liver cirrhosis is the last stage of it. It is difficult, sometimes, to differentiate between compensated cirrhosis (i.e. Child A patients) and non-cirrhotic alcoholic liver disease on clinical grounds only, as it happens in this study with two patients who did not undergo liver biopsy, whereas it is relatively easy to clearly identify decompensated (Child C) cirrhotics. Therefore, we have also classified our sample in two groups: decompensated (Child C) patients and compensated (non-cirrhotics + Child A and Child B) ones.
The presence of fracture was recorded by anamnesis and examination of clinical records and thoracic X-ray. If a rib fracture was discovered in the X-ray plain film without having been reported by the patient, the patient and their relatives were asked again. Forty-nine patients had suffered fractures at the inclusion in the study: rib fracture, affecting one or two adjacent ribs, in 36 cases, three or more rib fractures affecting non-consecutive ribs (more than one fracturing episode), and/or non-simultaneous fractures in other places, in 9 further cases, 2 cases of hip fracture, distal radius in one case and multiple simultaneous fractures (traffic accident) in one further case. We classified our patients in those with only rib fractures (n = 36) and those with multiple + osteoporotic fractures (n = 13).
Following a previously reported protocol (Santolaria et al., 2000), we also recorded the following: (a) the eating habits of the patients, asking them where they usually eat (if at home or in bars or taverns), how many times a day they eat and what they eat (sandwiches or snacks, or normal dishes), classifying them in three categories: normal: the patient eats three times a day (breakfast, lunch and dinner), usually at home, with the family; irregular habits (loss of some meals, substitution of some dishes by snacks); and poor eating habits (usually in bars or taverns, in the form of sandwiches or snacks, once or twice daily); (b) the social environment of the patient , also classifying it into three categories (the best one: with stable family and work; deranged: living alone, widowed, unmarried or divorced, eventually unemployed; and the worst one: skid-row patients, i.e. lonely, unemployed and/or homeless patients, usually heavy drinkers).
Nutritional evaluation
In addition, subjective nutritional evaluation was performed to all the patients, as follows: we examined the muscle masses of the upper and lower limbs and of the temporal muscle, defining two degrees of atrophy (severe, moderate) and absence of atrophy, and assigned 2, 1 and 0 points to each category, respectively. We also recorded, by physical examination, the fat loss on the cheek and abdomen, Bichat's fat and subcutaneous fat atrophy, and classified them in a similar way, and defined a score (SNS), based on the sum of the assigned points, for which the poorest value was 10 and 0 the best one was. We further classified our patients as well nourished (0–2 points), moderately undernourished (3–4 points) and severely undernourished (5–10 points), since this classification is related to the prognosis (Hernández-Plasencia et al., 1991).
We recorded handgrip strength with a dynamometer, body mass index, triceps skinfold with a Holtain lipocaliper and brachial perimeter with a tap, and determined serum bilirubin, prothrombin activity and serum albumin and other routine laboratory tests. We also collected blood samples after overnight fast. Samples were stored at −80°C until the following hormones and biochemical markers were determined.
Whole body composition
After informed consent, the 90 patients and 30 controls underwent assessment of bone mineral density (BMD), as well as lean mass and fat mass (only 26 controls) at different body parts, such as arms, ribs, legs, trunk and total body, with a HOLOGIC QDR-2000 (Waltham, MA, USA). In 85 cases and 28 controls, BMD was assessed at the lumbar spine and hip, recording BMD and Z- and T-scores (defined following standard criteria, Cummings et al., 2002) at the femoral neck, trochanter, Ward's triangle, intertrochantereal area, total hip and lumbar spine (L2–L4).
Biochemical parameters
We determined serum osteocalcin (to 79 patients), by immunometric chemiluminiscent assay (recovery = 97–121%; variation coefficients of assays ranging from 3.5 to 7.1%; DPC, Los Angeles, CA, USA), as a marker of bone synthesis, and C-terminal telopeptide of type I collagen (CrossLaps), by one-step enzyme-linked immunosorbent assay (ELISA), with a recovery ranging from 94 to 107% and an intra- and inter-assay variation coefficient ranging 4.7–4.9 and 5.4–8.1%, respectively (Osteometer Bio Tech A/S, Herlev, Denmark), as a marker of bone breakdown in non-cirrhotics. This parameter was determined only to 53 patients. We also determined serum IGF-1 (Chemiluminiscent assay, DPC, Los Angeles, CA, USA) to 87 patients; 1, 25 dihydroxyvitamin D3 to 73 patients (radioimmunoassay, Nichols, San Juan Capistrano, CA, USA), parathyroid hormone (PTH) to 87 patients; serum free testosterone to 49 patients; serum RANKL, to 50 patients, by ELISA, with a sensitivity of 0.08 pmol/l and a variation coefficient of 5% or less (intra-assay) and 9% or less (Immundiagnostik, Bensheim, Germany); and osteoprotegerin to 64, by sandwich ELISA, with a sensitivity of 0.14 U/l, and intra-assay and inter-assay variation coefficients <10% (Biovendor, Brno, Czech Republic).
The study protocol was approved by the local ethical committee of our Hospital (2009/23) and conforms to the ethical guidelines of the 1975 Declaration of Helsinki.
Statistics
The Kolmogorov–Smirnov test was used to test for normal distribution, a condition not fulfilled by PTH, vitamin D, IGF-1 and serum telopeptide. Therefore, non-parametric tests, such as Mann–Whitney's U test and Kruskall–Wallis, were used to analyse differences of these parameters between groups. Student's t-test, variance analysis and Pearson's correlation analysis were used with the remaining parameters (related to bone mass), whereas Spearman's rho (instead of Pearson's correlation) was utilized in the case of non-parametric variables.
To assess which parameters the presence of fracture depends on, we performed stepwise logistic regression analysis between fracture and several variables, which showed a relation with fractures in the univariate analysis, or which could be potentially involved in the pathogenesis of fractures, as commented in the results section, and also, among those with fracture, a logistic regression analysis to discern which factors were associated with rib fractures or with multiple fractures. All these analyses were performed with the SPSS program (Chicago, IL, USA).
RESULTS
Differences between patients and controls
Differences in clinical and biochemical parameters between patients and controls, as well as T-score values for lumbar spine and hip, and BMD at several parts of the skeleton are shown in Table 1. Decreased BMD was observed in the included patients, who also showed lower T-scores than controls. In Table 1, we also show data relative to lean and fat mass. Clearly, patients showed less lean mass, but a similar amount of fat (besides at the trunk) than controls.
Clinical and biochemical parameters in patients and controls. Results are given as mean ± standard deviation and, in those non-parametric variables, also median (inter-quartile range). Data relative to lean and fat mass include only 26 controls
| Patients (90) | Controls (30) | T (Z); P-value | |
|---|---|---|---|
| Age (years) | 50.14 ± 10.49 | 50.11 ± 10.40 | T = 0.03; NS |
| Serum ASAT (U/l) | 114.0 ± 99.5 | – | |
| Serum ALAT (U/l) | 59.4 ± 41.8 | – | |
| Serum GGT (U/l) | 302.5 ± 455.5 | – | |
| Osteocalcin (ng/ml) | 3.13 ± 3.18 | 7.37 ± 2.69 | T = 4.73; P < 0.001 |
| Serum telopeptide (nmol/l) | 0.603 ± 0.388 0.53 (0.43–0.65) | 0.211 ± 0.095 0.17 (0.14–0.34) | Z = 4.9; P < 0.001 |
| Serum vitamin D (pg/ml) | 28.27 ± 14.68 24.00 (17.0–37.5) | 85.37 ± 27.10 88.39 (59.18–112.55) | Z = 5.2; P < 0.001 |
| Serum IGF-1 (ng/ml) | 106.73 ± 97.82 62.6 (32.6–153.0) | 192.00 ± 106.13 135.00 (121.38–259.9) | Z = 4.21; P < 0.001 |
| Serum PTH (pg/ml) | 64.95 ± 86.12 41.80 (25.90–75.50) | 87.95 ± 141.15 40.84 (19.37–65.31) | Z = 0.7; NS |
| Serum free testosterone (ng/dl) | 10.06 ± 10.58 7.82 (3.07–13.38) | 18.09 ± 3.84 18.41 (14.81–20.44) | Z = 3.79; P < 0.001 |
| Serum cortisol (µg/dl) | 14.32 ± 5.58 | 18.19 ± 4.55 | T = 2.66; P = 0.009 |
| Serum osteoprotegerin (pmol/l) | 12.86 ± 6.54 | 6.83 ± 1.79 | T = 3.54; P < 0.001 |
| Serum RANKL (pmol/l) | 0.14 ± 0.22 0.06 (0.01–0.17) | 0.08 ± 0.07 0.08 (0.02–0.15) | Z = 0.03; NS |
| Subtotal BMD (g/cm2) | 0.98 ± 0.10 | 1.08 ± 0.08 | T = 4.96; P < 0.001 |
| Pelvis BMD (g/cm2) | 1.04 ± 0.15 | 1.18 ± 0.13 | T = 4.85; P < 0.001 |
| Right leg BMD (g/cm2) | 1.22 ± 0.13 | 1.35 ± 0.12 | T = 5.22; P < 0.001 |
| Left leg BMD (g/cm2) | 1.22 ± 0.13 | 1.36 ± 0.12 | T = 5.38; P < 0.001 |
| Left arm BMD (g/cm2) | 0.80 ± 0.10 | 0.85 ± 0.08 | T = 2.75; P = 0.007 |
| Right arm BMD (g/cm2) | 0.81 ± 0.08 | 0.83 ± 0.07 | T = 1.32; NS |
| Left rib BMD (g/cm2) | 0.61 ± 0.07 | 0.70 ± 0.07 | T = 6.45; P < 0.001 |
| Right rib BMD (g/cm2) | 0.60 ± 0.07 | 0.70 ± 0.06 | T = 6.96; P < 0.001 |
| Thoracic spine BMD (g/cm2) | 0.90 ± 0.13 | 0.98 ± 0.11 | T = 2.94; P = 0.004 |
| Total hip T-score | −1.05 ± 1.16 (n = 85) | −0.07 ± 1.18 (n = 28) | T = 3.48; P < 0.001 |
| Lumbar spine T-score | −1.15 ± 1.18 (n = 85) | −0.56 ± 0.90 (n = 28) | T = 2.12; P = 0.036 |
| Left arm lean (g) | 2535 ± 608 | 3000 ± 448 | T = 3.62; P < 0.001 |
| Right arm lean (g) | 2781 ± 595 | 3099 ± 469 | T = 2.51; P = 0.013 |
| Trunk lean (g) | 26,301 ± 3655 | 26,094 ± 3008 | T = 0.26; NS |
| Left leg lean (g) | 7191 ± 1596 | 8187 ± 1135 | T = 2.97; P = 0.004 |
| Right leg lean (g) | 7467 ± 1563 | 8467 ± 1024 | T = 3.07; P = 0.003 |
| Total lean (g) | 49,951 ± 6984 | 53,032 ± 5927 | T = 1.98: P = 0.05 |
| Left arm fat (g) | 1322 ± 749 | 1429 ± 516 | T = 0.69; NS |
| Right arm fat (g) | 1438 ± 916 | 1536 ± 575 | T = 0.52; NS |
| Trunk fat (g) | 8812 ± 5057 | 11,770 ± 3860 | T = 2.76; P = 0.007 |
| Left leg fat (g) | 3071 ± 1528 | 3330 ± 1042 | T = 0.81; NS |
| Right leg fat (g) | 3044 ± 1516 | 3379 ± 1057 | T = 1.05; NS |
| Total fat (g) | 18,706 ± 9273 | 22,366 ± 6364 | T = 1.89; NS |
| Patients (90) | Controls (30) | T (Z); P-value | |
|---|---|---|---|
| Age (years) | 50.14 ± 10.49 | 50.11 ± 10.40 | T = 0.03; NS |
| Serum ASAT (U/l) | 114.0 ± 99.5 | – | |
| Serum ALAT (U/l) | 59.4 ± 41.8 | – | |
| Serum GGT (U/l) | 302.5 ± 455.5 | – | |
| Osteocalcin (ng/ml) | 3.13 ± 3.18 | 7.37 ± 2.69 | T = 4.73; P < 0.001 |
| Serum telopeptide (nmol/l) | 0.603 ± 0.388 0.53 (0.43–0.65) | 0.211 ± 0.095 0.17 (0.14–0.34) | Z = 4.9; P < 0.001 |
| Serum vitamin D (pg/ml) | 28.27 ± 14.68 24.00 (17.0–37.5) | 85.37 ± 27.10 88.39 (59.18–112.55) | Z = 5.2; P < 0.001 |
| Serum IGF-1 (ng/ml) | 106.73 ± 97.82 62.6 (32.6–153.0) | 192.00 ± 106.13 135.00 (121.38–259.9) | Z = 4.21; P < 0.001 |
| Serum PTH (pg/ml) | 64.95 ± 86.12 41.80 (25.90–75.50) | 87.95 ± 141.15 40.84 (19.37–65.31) | Z = 0.7; NS |
| Serum free testosterone (ng/dl) | 10.06 ± 10.58 7.82 (3.07–13.38) | 18.09 ± 3.84 18.41 (14.81–20.44) | Z = 3.79; P < 0.001 |
| Serum cortisol (µg/dl) | 14.32 ± 5.58 | 18.19 ± 4.55 | T = 2.66; P = 0.009 |
| Serum osteoprotegerin (pmol/l) | 12.86 ± 6.54 | 6.83 ± 1.79 | T = 3.54; P < 0.001 |
| Serum RANKL (pmol/l) | 0.14 ± 0.22 0.06 (0.01–0.17) | 0.08 ± 0.07 0.08 (0.02–0.15) | Z = 0.03; NS |
| Subtotal BMD (g/cm2) | 0.98 ± 0.10 | 1.08 ± 0.08 | T = 4.96; P < 0.001 |
| Pelvis BMD (g/cm2) | 1.04 ± 0.15 | 1.18 ± 0.13 | T = 4.85; P < 0.001 |
| Right leg BMD (g/cm2) | 1.22 ± 0.13 | 1.35 ± 0.12 | T = 5.22; P < 0.001 |
| Left leg BMD (g/cm2) | 1.22 ± 0.13 | 1.36 ± 0.12 | T = 5.38; P < 0.001 |
| Left arm BMD (g/cm2) | 0.80 ± 0.10 | 0.85 ± 0.08 | T = 2.75; P = 0.007 |
| Right arm BMD (g/cm2) | 0.81 ± 0.08 | 0.83 ± 0.07 | T = 1.32; NS |
| Left rib BMD (g/cm2) | 0.61 ± 0.07 | 0.70 ± 0.07 | T = 6.45; P < 0.001 |
| Right rib BMD (g/cm2) | 0.60 ± 0.07 | 0.70 ± 0.06 | T = 6.96; P < 0.001 |
| Thoracic spine BMD (g/cm2) | 0.90 ± 0.13 | 0.98 ± 0.11 | T = 2.94; P = 0.004 |
| Total hip T-score | −1.05 ± 1.16 (n = 85) | −0.07 ± 1.18 (n = 28) | T = 3.48; P < 0.001 |
| Lumbar spine T-score | −1.15 ± 1.18 (n = 85) | −0.56 ± 0.90 (n = 28) | T = 2.12; P = 0.036 |
| Left arm lean (g) | 2535 ± 608 | 3000 ± 448 | T = 3.62; P < 0.001 |
| Right arm lean (g) | 2781 ± 595 | 3099 ± 469 | T = 2.51; P = 0.013 |
| Trunk lean (g) | 26,301 ± 3655 | 26,094 ± 3008 | T = 0.26; NS |
| Left leg lean (g) | 7191 ± 1596 | 8187 ± 1135 | T = 2.97; P = 0.004 |
| Right leg lean (g) | 7467 ± 1563 | 8467 ± 1024 | T = 3.07; P = 0.003 |
| Total lean (g) | 49,951 ± 6984 | 53,032 ± 5927 | T = 1.98: P = 0.05 |
| Left arm fat (g) | 1322 ± 749 | 1429 ± 516 | T = 0.69; NS |
| Right arm fat (g) | 1438 ± 916 | 1536 ± 575 | T = 0.52; NS |
| Trunk fat (g) | 8812 ± 5057 | 11,770 ± 3860 | T = 2.76; P = 0.007 |
| Left leg fat (g) | 3071 ± 1528 | 3330 ± 1042 | T = 0.81; NS |
| Right leg fat (g) | 3044 ± 1516 | 3379 ± 1057 | T = 1.05; NS |
| Total fat (g) | 18,706 ± 9273 | 22,366 ± 6364 | T = 1.89; NS |
Differences between patients with fractures and without fractures
As shown in Table 2, 49 patients presented at least one bone fracture at the time of inclusion. We failed to find differences between patients with and without fractures regarding BMD parameters. On the contrary, there were significant differences regarding lean mass, especially between patients with rib fracture and patients with other types of fractures and non-fractured patients. Although there was also a trend regarding fat parameters, differences were not statistically significant (probably due to the higher values of standard deviations).
Body composition analysis, including bone mineral density (BMD, in g/cm2), lean mass and fat mass and biochemical and epidemiological parameters in patients with and without fractures. Results are given as mean ± standard deviation and, in those non-parametric variables, also median (inter-quartile range)
| Only rib fracture (n = 36) | Hip + multiple fracture (n = 13) | No fracture (n = 41) | ||
|---|---|---|---|---|
| Left arm BMD | 0.79 ± 0.10 | 0.78 ± 0.09 | 0.81 ± 0.10 | F = 0.82; NS |
| Right arm BMD | 0.81 ± 0.08 | 0.79 ± 0.08 | 0.82 ± 0.09 | F = 0.62; NS |
| Left ribs BMD | 0.60 ± 0.08 | 0.62 ± 0.08 | 0.60 ± 0.07 | F = 0.43; NS |
| Right ribs BMD | 0.60 ± 0.07 | 0.62 ± 0. 09 | 0.60; ± 0.07 | F = 0.50; NS |
| Thoracic spine BMD | 0.89 ± 0.11 | 0.88 ± 0.15 | 0.92 ± 0.15 | F = 0.84; NS |
| Lumbar spine BMD | 0.95 ± 0.12 | 0.93 ± 0.18 | 0.98 ± 0.18 | F = 0.44; NS |
| Pelvis BMD | 1.03 ± 0.13 | 1.04 ± 0.19 | 1.05 ± 0.16 | F = 0.15; NS |
| Left leg BMD | 1.20 ± 0.12 | 1.20 ± 0.15 | 1.24 ± 0.14 | F = 1.03; NS |
| Right leg BMD | 1.21 ± 0.12 | 1.20 ± 0.16 | 1.23 ± 0.13 | F = 0.43; NS |
| Subtotal BMD | 0.98 ± 0.09 | 0.97 ± 0.12 | 0.99 ± 0.10 | F = 0.29; NS |
| Osteocalcin (ng/ml) (31/12/36) | 2.38 ± 2.01 1.70 (1.10–2.85) | 2.61 ± 1.48 2.40 (1.35–3.80) | 3.95 ± 4.14 2.05 (1.00–6.45) | KW = 0.65; NS |
| Serum telopeptide (nmol/l) (23/9/21) | 0.58 ± 0.27 0.54 (0.44–0.64) | 0.75 ± 0.81 0.46 (0.34–0.75) | 0.56 ± 0.20 0.54 (0.42–0.67) | KW = 0.37; NS |
| Serum vitamin D (pg/ml) (28/12/32) | 24.06 ± 10.97 24.0 (15.0–27.0) | 24.00 ± 17.45 19.0 (15.0–24.8) | 33.69 ± 15.15 35.0 (21.0–44.0) | KW = 9.18; P = 0.01 |
| Serum PTH (pg/ml) (35/13/39) | 56.63 ± 48.20 40.6 (25.4–72.4) | 125.71 ± 190.67 54.8 (36.6–91.6) | 52.17 ± 41.61 39.5 (24.2–67.5) | KW = 2.69; NS |
| Free testosterone (ng/dl) (15/6/28) | 7.40 ± 5.31 6.9 (3.2–12.9) | 8.10 ± 6.78 6.90 (2.05–12.93) | 11.91 ± 12.95 8.27 (2.17–17.70) | KW = 0.56; NS |
| Serum cortisol (µg/dl) (33/12/36) | 13.34 ± 4.35 | 14.24 ± 6.23 | 15.25 ± 6.31 | F = 1.00; NS |
| Osteoprotegerin (pmol/l) (27/8/29) | 13.60 ± 7.47 | 12.50 ± 5.24 | 12.27 ± 5.89 | F = 0.30; NS |
| Serum RANKL (pmol/l) (19/8/23) | 0.15 ± 0.25 0.06 (0.01–0.21) | 0.17 ± 0.30 0.05 (0.013–0.23) | 0.11 ± 0.16 0.04 (0.02–0.11) | KW = 0.07 NS |
| Serum IGF-1 (ng/ml) (33/13/33) | 122.9 ± 103.6 94.3 (35.7–195) | 84.8 ± 81.9 32.6 (25.9–148.9) | 99.2 ± 97.7 55.4 (41.7–127.0) | KW = 3.13; NS |
| Age (years) | 49.22 ± 8.47 | 52.31 ± 15.11 | 50.26 ± 11.63 | F = 0.38; NS |
| Total hip T-score (35/13/37) | −1.14 ± 1.14 | −1.14 ± 1.26 | −0.90 ± 1.11 | F = 0.48; NS |
| L2–L4 T-score (35/13/37) | −1.13 ± 0.82 | −1.54 ± 1.58 | −1.13 ± 1.35 | F = 0.65; NS |
| Left arm lean (g) | 2344 ± 539 | 2745 ± 547 | 2636 ± 600 | F = 3.27; P = 0.04, 1 vs. 2 |
| Right arm lean (g) | 2598 ± 529 | 2830 ± 611 | 2925 ± 615 | F = 3.09; P = 0.051 |
| Trunk lean (g) | 25,094 ± 2742 | 28,790 ± 3296 | 26,571 ± 4060 | F = 5.62; P = 0.005, 2 vs. 1,3 |
| Left leg lean (g) | 6794 ± 1358 | 7665 ± 1209 | 7388 ± 1834 | F = 2.05; NS |
| Right leg lean (g) | 7130 ± 1381 | 7778 ± 1292 | 7664 ± 1758 | F = 1.44; NS |
| Total lean (g) | 47,528 ± 5447 | 53,576 ± 6259 | 50,930 ± 7765 | F = 4.68: P = 0.012, 1 vs. 2,3 |
| Left arm fat (g) | 1101 ± 598 | 1499 ± 609 | 1458 ± 867 | F = 2.70; NS |
| Right arm fat (g) | 1172 ± 677 | 1606 ± 684 | 1618 ± 1105 | F = 2.61; NS |
| Trunk fat (g) | 7713 ± 5998 | 10,614 ± 3595 | 9206 ± 4386 | F = 1.83; NS |
| Left leg fat (g) | 2764 ± 1471 | 3483 ± 1303 | 3211 ± 1623 | F = 1.38; NS |
| Right leg fat (g) | 2695 ± 1470 | 3537 ± 1182 | 3194 ± 1611 | F = 1.88; NS |
| Total fat (g) | 16,466 ± 10333 | 21,725 ± 6653 | 19,715 ± 8722 | F = 2.03; NS |
| Handgrip strength (pounds) | 142.8 ± 69.3 | 108.3 ± 60.9 | 163.7 ± 80.6 | F = 2.64; NS |
| Triceps skinfold (mm) | 8.14 ± 5.34 | 8.69 ± 6.69 | 8.45 ± 6.07 | F = 0.05; NS |
| Brachial perimeter (cm) | 27.46 ± 4.68 | 26.85 ± 3.65 | 27.24 ± 4.06 | F = 0.10; NS |
| Daily ethanol consumption (g) | 208 ± 79 | 213 ± 80 | 215 ± 81 | F = 0.07; NS |
| Years of consumption | 30 ± 8 | 33 ± 14 | 27 ± 9 | F = 1.73 NS |
| Body mass index (kg/m2) | 24.02 ± 3.42 | 25.99 ± 3.11 | 25.54 ± 4.03 | F = 1.83; NS |
| Only rib fracture (n = 36) | Hip + multiple fracture (n = 13) | No fracture (n = 41) | ||
|---|---|---|---|---|
| Left arm BMD | 0.79 ± 0.10 | 0.78 ± 0.09 | 0.81 ± 0.10 | F = 0.82; NS |
| Right arm BMD | 0.81 ± 0.08 | 0.79 ± 0.08 | 0.82 ± 0.09 | F = 0.62; NS |
| Left ribs BMD | 0.60 ± 0.08 | 0.62 ± 0.08 | 0.60 ± 0.07 | F = 0.43; NS |
| Right ribs BMD | 0.60 ± 0.07 | 0.62 ± 0. 09 | 0.60; ± 0.07 | F = 0.50; NS |
| Thoracic spine BMD | 0.89 ± 0.11 | 0.88 ± 0.15 | 0.92 ± 0.15 | F = 0.84; NS |
| Lumbar spine BMD | 0.95 ± 0.12 | 0.93 ± 0.18 | 0.98 ± 0.18 | F = 0.44; NS |
| Pelvis BMD | 1.03 ± 0.13 | 1.04 ± 0.19 | 1.05 ± 0.16 | F = 0.15; NS |
| Left leg BMD | 1.20 ± 0.12 | 1.20 ± 0.15 | 1.24 ± 0.14 | F = 1.03; NS |
| Right leg BMD | 1.21 ± 0.12 | 1.20 ± 0.16 | 1.23 ± 0.13 | F = 0.43; NS |
| Subtotal BMD | 0.98 ± 0.09 | 0.97 ± 0.12 | 0.99 ± 0.10 | F = 0.29; NS |
| Osteocalcin (ng/ml) (31/12/36) | 2.38 ± 2.01 1.70 (1.10–2.85) | 2.61 ± 1.48 2.40 (1.35–3.80) | 3.95 ± 4.14 2.05 (1.00–6.45) | KW = 0.65; NS |
| Serum telopeptide (nmol/l) (23/9/21) | 0.58 ± 0.27 0.54 (0.44–0.64) | 0.75 ± 0.81 0.46 (0.34–0.75) | 0.56 ± 0.20 0.54 (0.42–0.67) | KW = 0.37; NS |
| Serum vitamin D (pg/ml) (28/12/32) | 24.06 ± 10.97 24.0 (15.0–27.0) | 24.00 ± 17.45 19.0 (15.0–24.8) | 33.69 ± 15.15 35.0 (21.0–44.0) | KW = 9.18; P = 0.01 |
| Serum PTH (pg/ml) (35/13/39) | 56.63 ± 48.20 40.6 (25.4–72.4) | 125.71 ± 190.67 54.8 (36.6–91.6) | 52.17 ± 41.61 39.5 (24.2–67.5) | KW = 2.69; NS |
| Free testosterone (ng/dl) (15/6/28) | 7.40 ± 5.31 6.9 (3.2–12.9) | 8.10 ± 6.78 6.90 (2.05–12.93) | 11.91 ± 12.95 8.27 (2.17–17.70) | KW = 0.56; NS |
| Serum cortisol (µg/dl) (33/12/36) | 13.34 ± 4.35 | 14.24 ± 6.23 | 15.25 ± 6.31 | F = 1.00; NS |
| Osteoprotegerin (pmol/l) (27/8/29) | 13.60 ± 7.47 | 12.50 ± 5.24 | 12.27 ± 5.89 | F = 0.30; NS |
| Serum RANKL (pmol/l) (19/8/23) | 0.15 ± 0.25 0.06 (0.01–0.21) | 0.17 ± 0.30 0.05 (0.013–0.23) | 0.11 ± 0.16 0.04 (0.02–0.11) | KW = 0.07 NS |
| Serum IGF-1 (ng/ml) (33/13/33) | 122.9 ± 103.6 94.3 (35.7–195) | 84.8 ± 81.9 32.6 (25.9–148.9) | 99.2 ± 97.7 55.4 (41.7–127.0) | KW = 3.13; NS |
| Age (years) | 49.22 ± 8.47 | 52.31 ± 15.11 | 50.26 ± 11.63 | F = 0.38; NS |
| Total hip T-score (35/13/37) | −1.14 ± 1.14 | −1.14 ± 1.26 | −0.90 ± 1.11 | F = 0.48; NS |
| L2–L4 T-score (35/13/37) | −1.13 ± 0.82 | −1.54 ± 1.58 | −1.13 ± 1.35 | F = 0.65; NS |
| Left arm lean (g) | 2344 ± 539 | 2745 ± 547 | 2636 ± 600 | F = 3.27; P = 0.04, 1 vs. 2 |
| Right arm lean (g) | 2598 ± 529 | 2830 ± 611 | 2925 ± 615 | F = 3.09; P = 0.051 |
| Trunk lean (g) | 25,094 ± 2742 | 28,790 ± 3296 | 26,571 ± 4060 | F = 5.62; P = 0.005, 2 vs. 1,3 |
| Left leg lean (g) | 6794 ± 1358 | 7665 ± 1209 | 7388 ± 1834 | F = 2.05; NS |
| Right leg lean (g) | 7130 ± 1381 | 7778 ± 1292 | 7664 ± 1758 | F = 1.44; NS |
| Total lean (g) | 47,528 ± 5447 | 53,576 ± 6259 | 50,930 ± 7765 | F = 4.68: P = 0.012, 1 vs. 2,3 |
| Left arm fat (g) | 1101 ± 598 | 1499 ± 609 | 1458 ± 867 | F = 2.70; NS |
| Right arm fat (g) | 1172 ± 677 | 1606 ± 684 | 1618 ± 1105 | F = 2.61; NS |
| Trunk fat (g) | 7713 ± 5998 | 10,614 ± 3595 | 9206 ± 4386 | F = 1.83; NS |
| Left leg fat (g) | 2764 ± 1471 | 3483 ± 1303 | 3211 ± 1623 | F = 1.38; NS |
| Right leg fat (g) | 2695 ± 1470 | 3537 ± 1182 | 3194 ± 1611 | F = 1.88; NS |
| Total fat (g) | 16,466 ± 10333 | 21,725 ± 6653 | 19,715 ± 8722 | F = 2.03; NS |
| Handgrip strength (pounds) | 142.8 ± 69.3 | 108.3 ± 60.9 | 163.7 ± 80.6 | F = 2.64; NS |
| Triceps skinfold (mm) | 8.14 ± 5.34 | 8.69 ± 6.69 | 8.45 ± 6.07 | F = 0.05; NS |
| Brachial perimeter (cm) | 27.46 ± 4.68 | 26.85 ± 3.65 | 27.24 ± 4.06 | F = 0.10; NS |
| Daily ethanol consumption (g) | 208 ± 79 | 213 ± 80 | 215 ± 81 | F = 0.07; NS |
| Years of consumption | 30 ± 8 | 33 ± 14 | 27 ± 9 | F = 1.73 NS |
| Body mass index (kg/m2) | 24.02 ± 3.42 | 25.99 ± 3.11 | 25.54 ± 4.03 | F = 1.83; NS |
When patients with rib and multiple fractures were pooled together and compared with patients without fracture regarding lean and fat mass, significant differences were observed only regarding right arm lean mass, lower among those with fracture (2660 ± 555 g) than among those without fracture (2925 ± 615 g, t = 2.15; P = 0.034). The presence of fracture was significantly associated with subjective nutritional evaluation (χ2 = 5.79, P = 0.016). Indeed, 12 out of 49 individuals with fracture (24.49 %) were severely undernourished, in contrast with 3 out of 41 without fracture (7.32%). A non-significantly higher proportion of patients with multiple fractures (53.85%) were well nourished when compared with patients with only rib fractures (47.22%).
Among biochemical parameters, only vitamin D (KW = 9.18, P = 0.01; Table 2) was lower among those with fractures, both among those with only rib fractures and among those with multiple fractures. When all the patients with fracture were grouped and compared with those without fracture, serum levels of 1, 25 (OH)2 D3 were significantly lower in the former (median (IR) = 21 (15–27) pg/ml) than in the latter (35 (21–44) pg/ml; Z = 2.98; P = 0.003). Interestingly, serum vitamin D levels showed a significant relationship with handgrip strength (ρ = 0.24, P = 0.046), left arm lean mass (ρ = 0.26, P = 0.024), right arm lean mass (ρ = 0.31, P = 0.004), trunk lean mass (ρ = 0.26, P = 0.024) and total lean mass (ρ = 0.28, P = 0.017), but not with fat mass.
Irregular feeding habits were significantly associated with an increased risk of fractures (χ2 = 5.32, P = 0.021). Also, a nearly significant association was found between the type of fracture and eating habits: normal eating habits were recorded only in 18.2% of the alcoholics with multiple fractures, in 8.82% of those with only rib fractures, but in 40.62% of patients without fractures (P = 0.057).
A non-significant trend was observed between deranged social environment and fracture (P = 0.14). Social links were not disrupted in 69.7% of those without fractures, in 63.6% with multiple fractures and in 52.94% with only rib fractures.
Relationships between type of fracture, liver function, alcohol intake and tobacco
Thirteen patients were affected by multiple fractures. Of these, eight were cirrhotics and five non-cirrhotics, However, neither the association between cirrhosis and type of fracture (χ2 = 1.63, P = 0.2) nor the association between cirrhosis and any kind of fracture (χ2 = 0.87, P = 0.3; Table 3) were statistically significant. There were no relationships between both the type of fracture and the presence of fracture with any of the variables included in the Child-Pugh's score (ascites, encephalopathy, prothrombin, albumin and bilirubin). We also failed to find an association between Child's severity groups and the presence of fracture (χ2 = 0.04, P = 0.8) or the kind of fracture (χ2 = 0.79, P = 0.3). However, when patients of the Child C groups were compared with the remaining patients, the proportion of multiple fractures was significantly higher (32 vs. 7.7%) among Child C patients, and that of only rib fractures was slightly higher (43.1 vs. 32%) among compensated patients (χ2 = 5; P = 0.025; Table 3).
Fractures in Child C patients compared with the remaining ones, and among cirrhotics and non-cirrhotics
| Multiple fractures (%) | Rib fractures (%) | No fractures (%) | |
|---|---|---|---|
| Advanced liver disease (Child C patients) | 8 (32) | 8 (32) | 9 (36) |
| Non-advanced liver disease (Child A and B + non-cirrhotic alcoholics) | 5 (7.7) | 28 (43.1) | 32 (49.2) |
| χ2 | 5.00; P = 0.025 | ||
| Cirrhotics | 8 (20) | 16 (40) | 16 (40) |
| Non-cirrhotics | 5 (10.4) | 19 (39.6) | 24 (50) |
| χ2 | 1.63; NS | ||
| Multiple fractures (%) | Rib fractures (%) | No fractures (%) | |
|---|---|---|---|
| Advanced liver disease (Child C patients) | 8 (32) | 8 (32) | 9 (36) |
| Non-advanced liver disease (Child A and B + non-cirrhotic alcoholics) | 5 (7.7) | 28 (43.1) | 32 (49.2) |
| χ2 | 5.00; P = 0.025 | ||
| Cirrhotics | 8 (20) | 16 (40) | 16 (40) |
| Non-cirrhotics | 5 (10.4) | 19 (39.6) | 24 (50) |
| χ2 | 1.63; NS | ||
Cirrhotics showed in general lower T-score values at the lumbar spine (t = 3.15, P = 0.003), but not at the femoral neck, and also lower BMD values at different parts of the skeleton (Table 4). Cirrhotics showed lower values of vitamin D (Z = 2.54, P = 0.011) and IGF-1 (Z = 3.2, P = 0.001) than non-cirrhotics, but no differences in lean or fat mass.
Differences in BMD between cirrhotics and non-cirrhotics
| Cirrhotics (n = 40) | Non-cirrhotics (n = 48) | ||
|---|---|---|---|
| Left arm BMD (g/cm2) | 0.78 ± 0.08 | 0.82 ± 0.09 | T = 2.10; P = 0.039 |
| Right arm BMD (g/cm2) | 0.80 ± 0.08 | 0.83 ± 0.08 | T = 1.85; NS |
| Left ribs BMD (g/cm2) | 0.59 ± 0.07 | 0.62 ± 0.07 | T = 2.12; P = 0.037 |
| Right ribs BMD (g/cm2) | 0.59 ± 0.07 | 0.62 ± 0.06 | T = 2.57; P = 0.012 |
| Thoracic spine BMD (g/cm2) | 0.87 ± 0.12 | 0.93 ± 0.14 | T = 2.15; P = 0.034 |
| Lumbar spine BMD (g/cm2) | 0.93 ± 0.14 | 0.99 ± 0.17 | T = 1.82; NS |
| Pelvis BMD (g/cm2) | 1.01 ± 0.14 | 1.07 ± 0.15 | T = 2.08; P = 0.04 |
| Left leg BMD (g/cm2) | 1.21 ± 0.14 | 1.23 ± 0.12 | T = 0.93; NS |
| Right leg BMD (g/cm2) | 1.20 ± 0.11 | 1.24 ± 0.13 | T = 1.36; NS |
| Total BMD (g/cm2) | 1.06 ± 0.09 | 1.11 ± 0.10 | T = 2.33; P = 0.022 |
| Total hip T-score | −1.25 ± 0.96 | −0.80 ± 1.14 | T = 1.92; NS |
| Lumbar spine T-score | −1.57 ± 1.01 | −0.81 ± 1.22 | T = 3.15; P = 0.003 |
| Cirrhotics (n = 40) | Non-cirrhotics (n = 48) | ||
|---|---|---|---|
| Left arm BMD (g/cm2) | 0.78 ± 0.08 | 0.82 ± 0.09 | T = 2.10; P = 0.039 |
| Right arm BMD (g/cm2) | 0.80 ± 0.08 | 0.83 ± 0.08 | T = 1.85; NS |
| Left ribs BMD (g/cm2) | 0.59 ± 0.07 | 0.62 ± 0.07 | T = 2.12; P = 0.037 |
| Right ribs BMD (g/cm2) | 0.59 ± 0.07 | 0.62 ± 0.06 | T = 2.57; P = 0.012 |
| Thoracic spine BMD (g/cm2) | 0.87 ± 0.12 | 0.93 ± 0.14 | T = 2.15; P = 0.034 |
| Lumbar spine BMD (g/cm2) | 0.93 ± 0.14 | 0.99 ± 0.17 | T = 1.82; NS |
| Pelvis BMD (g/cm2) | 1.01 ± 0.14 | 1.07 ± 0.15 | T = 2.08; P = 0.04 |
| Left leg BMD (g/cm2) | 1.21 ± 0.14 | 1.23 ± 0.12 | T = 0.93; NS |
| Right leg BMD (g/cm2) | 1.20 ± 0.11 | 1.24 ± 0.13 | T = 1.36; NS |
| Total BMD (g/cm2) | 1.06 ± 0.09 | 1.11 ± 0.10 | T = 2.33; P = 0.022 |
| Total hip T-score | −1.25 ± 0.96 | −0.80 ± 1.14 | T = 1.92; NS |
| Lumbar spine T-score | −1.57 ± 1.01 | −0.81 ± 1.22 | T = 3.15; P = 0.003 |
Significant differences in total BMD (t = 2.15, P = 0.034), total hip T-score (t = 2.13, P = 0.036) and lumbar spine T-score (t = 2.42, P = 0.018) were observed between patients with decompensated liver disease (Child C patients, who showed lower values of the aforementioned parameters) and those without decompensated liver disease. In addition, decompensated patients showed lower IGF-1 (Z = 3.61, P < 0.001) and testosterone (Z = 2.19, P = 0.028) and also nearly significantly lower vitamin D levels (Z = 1.91, P = 0.056) than the remaining patients. No differences were observed in lean and fat mass, except trunk lean mass, which was significantly higher (t = 2.42, P = 0.018) in decompensated patients (a result which may be in relation with the presence of ascites).
Among cirrhotics, those with multiple fractures showed, in general, a tendency to lower BMD and T-score values, although the only significant difference was observed regarding lumbar spine T-score values, which were lower among those with multiple fractures (−2.06 ± 1.50) than those without fractures (−1.82 ± 0.70) and those with only rib fractures (−1.10 ± 0.80, F = 3.4, P = 0.04). On the contrary, no differences were observed, among non-cirrhotics, between patients with only rib fractures, multiple fractures or no fractures.
In 87 patients, tobacco consumption was recorded. No association was found between smoking and fracture (71.8% among smokers, 60.4% among non-smokers).
Logistic regression
By stepwise logistic regression analysis, introducing the parameters of age, cirrhosis, lumbar spine (L2–L4) T-score, total hip T-score, total BMD, total lean mass, total fat mass, subjective nutritional evaluation, duration of ethanol consumption and the hormones IGF-1, cortisol and vitamin D (classified in two groups, below or above the median), only vitamin D (P = 0.034) and subjective nutritional evaluation (P = 0.021) showed a significant relation with the presence or not of fracture in the univariate analysis, although subjective nutritional evaluation displaced vitamin D in the final formula (P = 0.019). On the contrary, vitamin D remained the only parameter related to fracture (P = 0.017), displacing subjective nutritional evaluation, when variables such as cortisol or IGF-1 (which did not show any relations with fracture in the present study) were removed. Also, vitamin D remained the only parameter related to fracture (P = 0.019) even if the variable ‘decompensated liver disease’ (Child C patients) was introduced.
Considering only patients with fractures and comparing those with rib fractures and those with multiple fractures, introducing the variables age, cirrhosis, lumbar spine (L2–L4) T-score, total hip T-score, total BMD, total lean mass, total fat mass, subjective nutritional evaluation, duration of ethanol consumption and the hormones IGF-1, cortisol and vitamin D (classified in two groups, below or above the median), the variables which entered the final formula were total lean mass (P = 0.024), total fat mass (P = 0.029) and liver cirrhosis (P = 0.046), in this order.
However, the variable ‘decompensated liver disease’ (Child C patients) displaces the variable ‘cirrhosis’ from the final formula. Again, total lean mass was the first parameter selected (P = 0.031), but ‘decompensated liver disease’ was the second (P = 0.023) and total fat mass, the third one (P = 0.035). These three parameters remain selected if we remove IGF-1 and cortisol.
DISCUSSION
In contrast with some population-based studies, which report a reduced relative risk for hip fracture associated to mild or moderate alcohol consumption (Berg et al., 2008; Wosje and Kalkwarf, 2007), but in accordance with other authors who studied alcoholic patients (Clark et al., 2003; González-Calvín et al., 1993; Lindholm et al., 1991; Malik et al., 2009; Peris et al., 1992), we found marked differences between alcoholics and controls regarding BMD, and total hip and lumbar spine T-scores. Osteoporosis (T-score < − 2.5) was found in nine cases at the lumbar spine and in five cases at the hip. These are findings similar to others previously reported by our group and by many others, irrespective of the method used to assess osteoporosis (Alvisa–Negrín et al., 2009; Crilly et al., 1988; Diamond et al., 1989; Farley et al., 1985; Feitelberg et al., 1987; García-Valdecasas-Campelo et al., 2006; González-Calvín et al., 1993; Jorge-Hernández et al., 1988; Lalor et al., 1986; Peris et al., 1992; Santolaria et al., 2000, among others). In our study, decreased bone mass heavily depends on decreased bone synthesis, estimated on the basis of osteocalcin values. Biochemical variables related with bone remodelling, such as PTH and RANKL, were not significantly different in patients and controls, but serum telopeptide, which may be influenced by liver collagen metabolism (Ricard-Blum et al., 1996), was also higher among non-cirrhotics (0.53 ± 0.27 ng/ml) when compared with controls (Z = 4.00; P < 0.001).
More than 50% of the patients analysed showed bone fractures. This is quite a high figure compared with those reported by other authors, such as Keso et al. (1988), Wilkinson et al. (1985) or Peris et al. (1995), who found prevalence figures around 35%, although they focused mainly in detecting rib fractures or radiologically assessed vertebral fractures.
Low vitamin D levels and deranged nutritional status were the main factors associated with fracture. Interestingly, although, in general, BMD values were lower among those with fractures, differences were not statistically significant. This result is in agreement with those reported by Peris et al. (1995) and Santori et al. (2008) regarding vertebral fractures assessed by radiological criteria. BMD is the most widely used technique to define osteoporosis (Cummings et al., 2002), but it may not accurately estimate the fracture risk, due to the fact that factors related with bone quality are not assessed by this technique. Moreover, the accuracy of other recently introduced parameters seem to be higher than the classic BMD T-score value < − 2.5, at least for prediction of vertebral fractures (Diacinti et al., 2010). In contrast to BMD, serum vitamin D (1, 25 (OH)2 D3) levels were significantly lower among those with fractures. Although in some studies, there were no differences in serum vitamin D between alcoholic patients with and without fractures (Wilkinson et al., 1985), a similar result to that reported in this study was found by other authors, such as Santori et al. (2008), who reported no differences in BMD in a cohort of alcoholics with a high prevalence of vertebral and non-vertebral fractures, but indeed lower vitamin D values in the former; or Diamond et al. (1989) and Malik et al. (2009), who found lower 25 OH vitamin D levels in alcoholics than in controls. It is in this sense worthy of note that low vitamin D levels have been put in relation with increased risk of falls, and thus, with bone fractures; in fact, treatment with vitamin D reduced the risk of falling among older individuals (Bischoff-Ferrari et al., 2009). In the last decade, it was shown that a receptor for 1, 25 OH vitamin D3 is present in human skeletal muscle (Bischoff et al., 2001), activation of which leads to muscle cell proliferation and differentiation (Ceglia, 2008), and to an increase in the synthesis of calmodulin (Dittranti et al., 1990), leading to improved muscle function. Although it was not an aim of our study, it is important to remark that there was a significant relationship between handgrip strength and serum vitamin D levels, as well as between vitamin D and lean mass, fully in accordance with the aforementioned statements, and with the results of a recently published experimental study in which we also described a significant relationship between type II fibre atrophy and serum 1, 25 (OH)2 vitamin D levels in rats treated following the Lieber-deCarli model (González-Reimers et al., 2010).
In addition to impaired nutritional status, statistically significant differences were observed regarding several parameters related to lean mass and the presence of fractures. Also a trend was observed to lower values of handgrip strength among patients with fracture. Both findings are also in accordance with the relation observed between fractures and nutritional status, and in accordance with the current knowledge about this item (Huang et al., 1996). Nutritional status was evaluated in our patients with a subjective scale and it was related with fracture. Also, those with the worst nutritional status showed the lowest vitamin D levels (χ2 = 4.25; P = 0.039). Therefore, the presence of fracture in our population was related with vitamin D, subjective nutritional evaluation and lean mass, but vitamin D is related to nutritional status.
Some differences did exist between patients with only rib fractures and patients with other types of fractures. In general, patients with rib fractures showed worse nutritional parameters, especially lean mass parameters, than patients with other types of fractures, although there were no differences in BMD and in biochemical parameters. Rib fractures are among the most commonly suffered by alcoholics, so that they have been considered as markers of alcohol consumption in patients affected or not by chronic liver disease (Israel et al., 1980; Lindsell et al., 1982). As stated before, rib fractures are frequently observed in elderly, frail individuals, prone to falls (Barrett-Connor et al., 2010). In this sense, the alcoholics with rib fractures share some of these characteristics: they are those with less lean and fat mass, and also with the more deranged eating habits, also showing a trend to a more intense social margination. Nutritional status is in part related to bizarre eating behaviour in alcoholics (Santolaria et al., 2000). Usually, the heavy drinker disrupts his social links, become divorced or separated, and his eating habits evolve to more irregular ones. This may explain why we found an association between irregular feeding habits and bone fractures, and also the trend to an association of irregular eating habits and rib fracture. A similar finding was also reported by Keso et al. (1988), who showed that socio-economic status did not influence the prevalence of fractures, but thoracic fractures were more commonly observed among unmarried, divorced or widowed alcoholics.
It may seem counterintuitive that patients with multiple fractures showed, in general, better nutrition, reduced lean mass and a trend to more fat mass, and a even a trend to better social and familial environment, than those with only rib fractures. However, these findings may be interpreted in several ways. One explanation may consist in the fact that rib fractures may appear even after minor trauma, such as falling from a standing position, something which is very common in the advanced alcoholic patient, with muscle atrophy and reduced lean mass, severely impaired nutritional status, and, perhaps, accompanying polineuropathy and/or cerebellar atrophy, whereas multiple fractures may occur after major trauma, fighting and violence among less fragile alcoholics, with preserved lean and fat mass, and, in general, better environmental conditions.
However, an association was found between multiple fractures and liver cirrhosis and/or decompensated liver disease. These patients showed lower values of hormones directly involved in bone homoeostasis, such as vitamin D, IGF-1 and testosterone, decrease in which may play a role in the frequency of fractures. A third factor which must be considered is that the variables trunk and total lean mass, which were higher among cirrhotics and decompensated patients, are in fact measuring also water retention and not just muscle mass. Indeed, ascites is one of the criteria of decompensation of liver disease, and, as expected, it is by far more frequent among decompensated patients than among compensated ones (χ2 = 13.7; P < 0.0001).
Thus, our study shows that prevalent fractures are common among heavy alcoholics, but their presence is related more closely to impaired nutritional status and reduced lean mass (especially among alcoholics with only rib fractures) and low serum vitamin D levels, than to BMD. Interestingly, a relation was found between vitamin D levels, handgrip strength and lean mass, in accordance with the described effects of vitamin D on muscle structure and function. Multiple fractures depend more heavily on the presence of decompensated liver disease, with decreased levels of IGF-1, testosterone and vitamin D. In our study, these patients showed a trend to a more preserved nutritional status than those with only rib fractures.
