Abstract
Background
Previous research has shown that dietary patterns are related to the risk of several adverse health outcomes, but the relation of these patterns to skeletal fragility is not well understood.
Objective
Our objective was to determine the relation between dietary patterns and incident fracture and possible mediation of this relation by body mass index, bone mineral density, or falls.
Design
We performed a retrospective cohort study based on the Canadian Multicentre Osteoporosis Study—a randomly selected population-based cohort. We assessed dietary patterns by using self-administered food-frequency questionnaires in year 2 of the study (1997–1999). Our primary outcome was low-trauma fracture occurring before the 10th annual follow-up (2005–2007).
Results
We identified 2 dietary patterns by using factor analysis. The first factor (nutrient dense) was strongly associated with intake of fruit, vegetables, and whole grains. The second factor (energy dense) was strongly associated with intake of soft drinks, potato chips, French fries, meats, and desserts. The nutrient-dense factor was associated with a reduced risk of fracture per 1 SD in men overall [hazard ratio (HR): 0.83; 95% CI: 0.64, 1.08] and in women overall (HR: 0.86; 95% CI: 0.76, 0.98). An age trend (P = 0.03) was observed, which yielded an HR of 0.97 in younger women (age <70 y) compared with an HR of 0.82 in older women (age ≥70 y). The associations were independent of body mass index, bone mineral density, falls, and demographic variables. The energy-dense pattern was not related to fracture.
Conclusion
A diet high in vegetables, fruit, and whole grains may reduce the risk of low-trauma fracture, particularly in older women.
INTRODUCTION
In 2005, 2 million fractures occurred in the United States, which were associated with >17 billion in direct costs, and these figures are expected to increase by ≥50% over 20 y (1). Fractures also led to long-term disability (2), decreased health-related quality of life (3), and increased mortality (4). Identification and treatment of individuals with low bone mineral density (BMD) can reduce the burden of fracture (5, 6). Alternative strategies for population health are necessary because many individuals who fracture do not have a low BMD (7). Applied across the population, even relatively small effects can affect the burden of disease. Adequate calcium and vitamin D has long been recommended, and many studies have been undertaken to test supplementation with these nutrients (8). Other studies suggest that dietary intakes other than calcium and vitamin D may also play an important role in bone health (9–12). In many studies, BMD has been used as a proxy for fracture because it is one of the strongest and most consistent risk factors for fracture (13). However, it is also important to consider fracture as an outcome, because diet may impinge on other risk factors, including the propensity to fall and body mass index. We previously assessed the relation of dietary patterns and BMD in a large population-based cohort of men and women and found that, although there was a clear association between diet and body mass index, there was only an indirect association with BMD (14). Similar dietary patterns have been noted in other studies, which showed fairly broad generalizability of the analysis. Moreover, these patterns have been associated with many health outcomes in both men and women (15–18). The objective of the present analysis was to determine whether dietary patterns in postmenopausal women and men aged ≥50 y are related to fracture, independently of other lifestyle variables, and whether the association found is mediated by body mass index, BMD, and/or falls.
SUBJECTS AND METHODS
Subjects
The Canadian Multicentre Osteoporosis Study (CaMos) is an ongoing cohort study with study enrollment from 1995 to 1997. Eligible participants were men aged ≥50 y and postmenopausal women (≥1 y after cessation of menses) enrolled in the study at year 2 (1997–1999). The study cohort consisted of eligible participants who completed the food-frequency questionnaire (FFQ) with ≤10 missing responses in the food and drink section. A total of 9423 participants were enrolled in CaMos: 7637 were eligible, and 5188 were in the study cohort. An overview of the study appears on the study website (www.camos.org), and the methodologic details of CaMos have been published (19). A more recent publication includes much of the groundwork methodology necessary for the present study (14). For the sake of completeness, a brief summary of the methods is also included. Recruitment for CaMos was based on a random selection of households across Canada according to an age-stratified design. Of those invited to participate, 42% had a baseline interview. Participants were ≥25 y of age at the beginning of the study and lived in 1 of 9 regions, with centers in St John’s, Halifax, Quebec City, Toronto, Hamilton, Kingston, Saskatoon, Calgary, and Vancouver. Ethics approval was granted through McGill University and the appropriate research ethics board for each participating center. Signed informed consent was obtained from all study participants in accordance with the Helsinki Declaration.
Data collection
All participants were given a standardized interviewer-administered questionnaire at baseline (1995–1997). The questionnaire covered demographics, health, nutrition, lifestyle, and medical history. The nutrition part of the baseline questionnaire included items on milk, milk products, and other sources of calcium and vitamin D (20). Food models were included to demonstrate portion sizes. Medication and supplement use was assessed by using a complete inventory of prescriptions and bottles brought to the interview. The baseline clinical assessment included height, weight, and BMD. An FFQ was mailed to all participants in the second year of the study (1997–1999). Follow-up visits were scheduled in the 3rd (ages 40–60 y), 5th, and 10th year after enrollment.
FFQ
The FFQ used in CaMos for the year 2 follow-up was derived from items on the short-form Block questionnaire (21), with modifications according to the Canadian diet (22). A specified portion size was used; hence, the response options were similar to those of the Willett questionnaire (23). More specifically, possible responses were 1 of 9 ordinal frequency categories ranging from never/less than once a month to ≥6 times/d. The main variables used in this study were derived from the food and beverage portion of the questionnaire; there were 51 food items and 18 beverage items. Total energy intake was calculated by using the frequency and specified portion size from the questionnaire together with caloric information from the Canadian Nutrient File (24).
Derivation of the factor scores
The methods used to create the factors scores in this article were described elsewhere (14). As in our previous dietary study, we excluded participants with >10 missing responses. The median of the study sample was used to impute missing values for the remaining study population. A sensitivity analysis showed that this imputation did not affect the factor analysis or subsequent regression. Note that the study population used to create the factor scores consisted of all men and women who had a sufficiently complete questionnaire and thus includes younger men and premenopausal women. This younger population was not included in the present study because they have a low risk of fragility fracture.
The nutrient-dense and energy-dense factor scores were constructed as follows. The 69 food and drink items were grouped into 34 categories of similar items; monthly frequency was determined by adding the frequencies of each item in the category. The resulting variables were shifted by one so that all results were positive, log transformed, and finally rescaled to have a mean of 0 and an SD of 1. A 2-factor solution was chosen because it was easy to interpret, similar to that reported in other studies, and robust to choice of subgroup. Indeed, the 2 dietary patterns found in the whole group were present in men and women separately. The nutrient-dense factor was most strongly correlated with intake of fruit, vegetables, and whole grains and explained 9.3% of the total variance. The energy-dense factor was most strongly correlated with the intake of soft drinks, potato chips, French fries, meats (hamburger, hot dog, lunch meat, bacon, and sausage), and desserts (doughnuts, chocolate, and ice cream) and explained 7.8% of the total variance. A detailed presentation of the factor loadings and association between factor scores and other variables is presented in a previous article (14). In a subsequent post hoc analysis, we showed that the use of individual items, rather than grouped items, produces factors with much the same interpretation and nearly the same factor scores. The correlation between original scores and those created by using individual items was r = 0.97 for the nutrient-dense factor and r = 0.95 for the energy-dense factor.
The association between diet and health outcomes may often be related to total energy intake (25). Both factor scores were positively correlated with energy intake, which is not surprising given population variation in energy intake. Furthermore, of those with a given energy intake, an inverse relation between the nutrient-dense factor score and the energy-dense factor score was observed, so that the increase in one factor resulted in a decrease of the same magnitude in the other factor score. To assess variation in diet independent of energy intake, we created a new index—the difference score—which equals the energy-dense factor score minus the nutrient-dense factor score. The reason for using the difference score is that it is simple to explain how one can alter it (ie, one eats more foods from the nutrient-dense list and less foods from the energy-dense list), it is nearly independent of energy intake, and it easily translates both to individual and population health interventions. It also appears that the major shift in dietary patterns between older and younger generations occurs along this primary axis, ie, younger people tend to eat a diet that simultaneously includes more energy-dense foods and less nutrient-dense foods (14).
BMD
BMD was measured at the lumbar spine (L1-L4), femoral neck, trochanter, Ward’s triangle, and total hip. Seven centers used Hologic (Waltham, MA) densitometers, and 2 centers used GE/Lunar (Madison, WI) densitometers. All GE/Lunar measurements were converted to equivalent Hologic values by using a standard reference formulas (26). Calibration was done by using daily scans of phantom local to each study center; between-center calibration was done by circulating a European Spine Phantom between the 9 centers.
Fracture assessment
The self-reported incident clinical fractures were identified at the scheduled interview (3rd, 5th, and 10th year after study enrollment) or by postal questionnaire (all other years). Confirmation and further information concerning the fracture were gathered with the use of a structured interview that included the date, fracture site, circumstances leading to fracture, and medical treatment of the fracture. Radiology reports were obtained, where possible (with participants’ consent). Fractures that occurred without trauma or from a fall of standing height or less were considered to be low trauma. For the current analysis, all low-trauma clinical fractures that occurred up to the 10th annual follow-up were included. Fractures of the skull, face, hands, and feet were not included.
Statistical analysis
We hypothesized that there is an association between diet and fracture risk and that this association has a component that is independent of BMD. The main model was a Cox proportional hazards model with incident low-trauma fracture as the outcome and age as the time axis. All participants without incident low-trauma fracture were censored at the time last observed in the study or at the 10th y follow-up. The regression analyses were a priori stratified by sex based on known differences in both diet and bone mineral metabolism. The a priori–specified potential confounders were as follows: age, education (<12 y schooling, high school diploma, or postsecondary education), cigarette smoking (nonsmoker, former smoker, or current smoker), alcohol [nondrinker, moderate intake (<1 drink/d in women, <2 drinks/d in men), high intake (≥1 drink/d in women, ≥2 drinks/d in men)], activity (kcal/wk spent on moderate activity, vigorous work, or strenuous sports calculated from a weekly inventory of activities in these 3 categories), sedentary time (time per day spent sitting or sleeping), daily milk consumption, daily use of supplements (vitamin D and calcium), diagnosis of osteoporosis, history of low-trauma fracture after age 40 y, medication use (hormone therapy and bisphosphonates), and comorbidities (heart disease, kidney disease, type 1 or type 2 diabetes or hypertension, in-flammatory bowel disease, or eating disorder). Sample size considerations limited the inclusion of covariates in the fully adjusted model for men, although other variables were tested for confounding. The choice of covariates in the final model for men was determined by the literature and previous BMD regression analysis (14) and included prior fracture, comorbidities, smoking, milk intake, and daily use of supplements. We assessed direct and indirect associations in both men and women by running models excluding or including previous falls, baseline body mass index, and baseline femoral neck BMD. We tested the proportional hazards assumptions and possible interactions between age and dietary variables. All analyses were performed with Stata 9.2 (StataCorp, College Station, TX).
RESULTS
The study sample consisted of 1649 men and 3539 women. Our analyses excluded 758 men and 1690 women present at year 2 because the year 2 FFQ was missing or incomplete. The differences between the study population and those with missing or incomplete data were similar to those reported in the previous study (14). The baseline characteristics of the men and women, stratified by tertile of each factor score, are shown in Table 1. Many univariate relations were observed between demographics and the factor scores, the most notable of which was the relation between sex and both dietary factor scores. The bottom tertile for the nutrient-dense factor score was 57% women, and the top tertile was 80% women. The reverse association held for the energy-dense factor score, for which the bottom tertile was 79% women, and the top tertile was 52% women. This very strong relation led to associations between other variables and the dietary factors scores (eg, use of antiresorptive agents, vitamin D, and calcium supplements). Our main analysis was stratified by sex to control for strong confounding.
TABLE 1.
Characteristics of the study sample (postmenopausal women and men aged ≥50 y) at baseline (1995–1997), stratified by tertile of nutrient-dense and energy-dense factor scores1
| Nutrient-dense factor score
|
P value2 | Energy-dense factor score
|
P value2 | |||||
|---|---|---|---|---|---|---|---|---|
| Lowest tertile (n = 1730) | Middle tertile (n = 1729) | Highest tertile (n = 1729) | Lowest tertile (n = 1730) | Middle tertile (n = 1729) | Highest tertile (n = 1729) | |||
| Age (y)3 | 65.2 ± 9.54 | 66.6 ± 9.1 | 68.2 ± 8.6 | <0.001 | 68.5 ± 8.7 | 67.2 ± 9.1 | 64.3 ± 9.2 | <0.001 |
| Height (cm) | 165.5 ± 9.9 | 164.2 ± 9.2 | 162.8 ± 8.6 | <0.001 | 162.5 ± 8.5 | 163.2 ± 9.0 | 166.7 ± 9.8 | <0.001 |
| Weight (kg) | 75.0 ± 15.7 | 73.3 ± 14.7 | 71.6 ± 13.7 | <0.001 | 70.0 ± 13.8 | 72.1 ± 13.9 | 77.9 ± 15.4 | <0.001 |
| BMI (kg/m2) | 27.3 ± 4.8 | 27.1 ± 4.6 | 27.0 ± 4.5 | 0.14 | 26.5 ± 4.5 | 27.0 ± 4.6 | 27.9 ± 4.7 | <0.001 |
| Femoral neck BMD (0.001 g/cm2) | 742 ± 128 | 733 ± 130 | 720 ± 127 | <0.001 | 712 ± 126 | 723 ± 126 | 759 ± 129 | <0.001 |
| SF-36 physical5 | 47.9 ± 9.6 | 47.9 ± 9.7 | 47.7 ± 9.6 | 0.74 | 47.6 ± 9.9 | 47.6 ± 9.7 | 48.3 ± 9.3 | 0.04 |
| SF-36 mental5 | 53.5 ± 9.0 | 54.1 ± 8.1 | 54.5 ± 7.6 | <0.001 | 54.3 ± 8.4 | 54.1 ± 8.2 | 53.8 ± 8.2 | 0.21 |
| Sedentary time (h/d)6 | 14.1 ± 3.0 | 14.0 ± 2.9 | 13.8 ± 2.9 | <0.001 | 13.6 ± 2.8 | 14.0 ± 2.9 | 14.3 ± 3.0 | <0.001 |
| Activity (1000 kcal/wk)7 | 4.6 ± 4.2 | 4.8 ± 3.9 | 4.8 ± 3.4 | 0.16 | 4.4 ± 3.6 | 4.5 ± 3.4 | 5.3 ± 4.4 | <0.001 |
| Vitamin D supplements (μg) | 2.3 ± 5.2 | 3.3 ± 5.8 | 3.5 ± 6.1 | <0.001 | 3.5 ± 6.0 | 3.2 ± 6.1 | 2.4 ± 4.9 | <0.001 |
| Calcium supplements (mg) | 225 ± 420 | 320 ± 488 | 352 ± 487 | <0.001 | 343 ± 496 | 315 ± 470 | 239 ± 434 | <0.001 |
| Calcium from milk (mg) | 365 ± 375 | 405 ± 369 | 439 ± 384 | <0.001 | 395 ± 384 | 409 ± 369 | 404 ± 380 | <0.001 |
| Other calcium from diet (mg) | 329 ± 252 | 398 ± 260 | 453 ± 281 | <0.001 | 371 ± 287 | 389 ± 250 | 419 ± 269 | <0.001 |
| Female sex [n (%)] | 988 (57.1) | 1174 (67.9) | 1377 (79.6) | <0.001 | 1367 (79.0) | 1270 (73.5) | 902 (52.2) | <0.001 |
| Incident fracture, during follow-up [n (%)] | 147 (8.5) | 134 (7.8) | 161 (9.3) | 0.32 | 171 (9.9) | 146 (8.4) | 125 (7.2) | 0.04 |
| Falls in the month prior to baseline [n (%)] | 100 (5.8) | 98 (5.6) | 111 (6.4) | 0.60 | 105 (6.1) | 102 (5.9) | 102 (5.9) | 0.97 |
| White [n (%)] | 1636 (94.6) | 1663 (96.2) | 1684 (97.4) | <0.001 | 1581 (91.4) | 1696 (98.1) | 1706 (98.7) | <0.001 |
| Smoker, current [n (%)] | 332 (19.2) | 225 (13.0) | 135 (7.8) | <0.001 | 172 (9.9) | 199 (11.5) | 321 (18.6) | <0.001 |
| High alcohol intake [n (%)]8 | 122 (7.1) | 111 (6.4) | 110 (6.4) | 0.65 | 80 (4.6) | 124 (7.2) | 139 (8.0) | <0.001 |
| Antiresorptive use [n (%)]3 | 494 (28.6) | 616 (35.6) | 690 (39.9) | <0.001 | 657 (40.0) | 681 (39.4) | 462 (26.7) | <0.001 |
BMD, bone mineral density; SF-36, Short-Form 36 Health Survey.
P values for between-group differences by ANOVA (continuous variables) and chi-square test (categorical variables).
Based on year 2 (administration of food-frequency questionnaire) rather than study enrollment (baseline).
Mean ± SD (all such values).
Scores range from 0 to 100: 0 indicates the worst possible health and 100 indicates the best possible health.
Total time spent sitting or sleeping.
Moderate, strenuous, or vigorous activity.
More than 2 units/d in men and 1 unit/d in women: 1 unit is equivalent to 335 mL (12 oz) beer, 118–148 mL (4–5 oz) wine, or 30–44 mL (1–1.5 oz) hard liquor.
The mean follow-up time in the study sample was 6.6 y in men and 6.8 y in women. There were 70 men and 372 women in the study sample who had low-trauma fracture up to the 10th annual follow-up. Differences between fracture cases and noncases within the study cohort are shown in Table 2. As expected, strong univariate relations were observed between both age and BMD and incident fracture status.
TABLE 2.
Characteristics of the study sample (postmenopausal women and men aged ≥50 y) at baseline (1995–1997), stratified by sex and incident fracture status before the 10th annual follow-up1
| Women
|
Men
|
|||||||
|---|---|---|---|---|---|---|---|---|
| All (n = 3539) | Fracture (n = 372) | Nonfracture (n = 3167) | P value2 | All (n = 1649) | Fracture (n = 70) | Nonfracture (n = 1579) | P value2 | |
| Age (y)3 | 67.6 ± 8.64 | 70.7 ± 8.1 | 67.2 ± 8.6 | <0.001 | 64.6 ± 10.0 | 69.7 ± 10.1 | 64.4 ± 9.9 | <0.001 |
| Height (cm) | 159.6 ± 6.3 | 159.3 ± 6.5 | 159.7 ± 6.2 | 0.25 | 173.8 ± 7.1 | 175.1 ± 7.4 | 173.8 ± 7.1 | 0.13 |
| Weight (kg) | 69.0 ± 13.3 | 68.5 ± 12.6 | 69.1 ± 13.4 | 0.41 | 82.5 ± 13.7 | 80.1 ± 13.4 | 82.6 ± 13.7 | 0.15 |
| BMI (kg/m2) | 27.1 ± 4.9 | 27 ± 4.9 | 27.1 ± 4.9 | 0.84 | 27.2 ± 4.0 | 26.1 ± 3.6 | 27.3 ± 4.0 | 0.01 |
| Femoral neck BMD (0.001 g/cm2) | 698 ± 117 | 650 ± 110 | 703 ± 116 | <0.001 | 803 ± 123 | 730 ± 125 | 806 ± 122 | <0.001 |
| SF-36 physical5 | 47 ± 10.0 | 43.8 ± 11.5 | 47.3 ± 9.7 | <0.001 | 49.8 ± 8.5 | 45.7 ± 10.6 | 49.9 ± 8.4 | 0.001 |
| SF-36 mental5 | 53.7 ± 8.6 | 53.3 ± 9.2 | 53.7 ± 8.5 | 0.37 | 54.7 ± 7.6 | 55.7 ± 7.2 | 54.7 ± 7.6 | 0.23 |
| Sedentary time (h/d)6 | 13.7 ± 2.9 | 13.3 ± 2.7 | 13.7 ± 2.9 | 0.01 | 14.5 ± 3.0 | 14.3 ± 3.2 | 14.5 ± 3.0 | 0.62 |
| Activity (1000 kcal/wk)7 | 4.4 ± 3.2 | 4.2 ± 3.1 | 4.4 ± 3.2 | 0.29 | 5.5 ± 4.9 | 5.9 ± 5.5 | 5.5 ± 4.9 | 0.60 |
| Vitamin D supplements (μg) | 3.5 ± 6.0 | 4.2 ± 7.1 | 3.4 ± 5.9 | 0.04 | 2.0 ± 4.9 | 3.0 ± 5.7 | 2.0 ± 4.8 | 0.14 |
| Calcium supplements (mg) | 373 ± 510 | 436 ± 505 | 365 ± 505 | 0.02 | 142 ± 315 | 157 ± 249 | 141 ± 317 | 0.60 |
| Calcium from milk (mg) | 397 ± 363 | 395 ± 334 | 397 ± 366 | 0.90 | 415 ± 406 | 347 ± 277 | 419 ± 411 | 0.04 |
| Other calcium from diet (mg) | 396 ± 274 | 393 ± 228 | 396 ± 278 | 0.78 | 388 ± 260 | 406 ± 318 | 387 ± 257 | 0.64 |
| Falls in the month prior to baseline [n (%)] | 210 (5.9) | 36 (9.7) | 174 (5.5) | 0.001 | 99 (6.0) | 5 (7.1) | 94 (6.0) | 0.68 |
| White [n (%)] | 3421 (96.7) | 361 (97.0) | 3060 (94.6) | 0.67 | 1562 (94.7) | 68 (97.1) | 1494 (94.6) | 0.36 |
| Smoker, current [n (%)] | 456 (12.9) | 38 (10.2) | 418 (13.2) | 0.10 | 236 (14.3) | 17 (24.3) | 219 (13.9) | 0.02 |
| High alcohol intake [n (%)]8 | 214 (6.1) | 26 (7.0) | 188 (5.9) | 0.42 | 129 (7.8) | 5 (7.1) | 124 (7.9) | 0.85 |
| Antiresorptive use [n (%)]3 | 1709 (48.3) | 179 (48.1) | 1530 (48.3) | 0.94 | 91 (5.5) | 4 (5.7) | 87 (5.5) | 0.94 |
BMD, bone mineral density; SF-36, Short-Form 36 Health Survey.
P values for between-group differences by ANOVA (continuous variables) and chi-square test (categorical variables).
Based on year 2 (administration of food-frequency questionnaire) rather than study enrollment (baseline).
Mean ± SD (all such values).
Scores range from 0 to 100: 0 indicates the worst possible health and 100 indicates the best possible health.
Total time spent sitting or sleeping.
Moderate, strenuous, or vigorous activity.
More than 2 units/d in men and 1 unit/d in women: 1 unit is equivalent to 335 mL (12 oz) beer, 118–148 mL (4–5 oz) wine, or 30–44 mL (1–1.5 oz) hard liquor.
The crude and fully adjusted associations between dietary variables and low-trauma fracture are shown in Table 3. Very little difference was observed between the crude and adjusted estimates, and the following discussion refers to the fully adjusted estimates. The nutrient-dense dietary pattern was associated with a reduced risk of fracture in women (HR: 0.86; 95% CI: 0.76, 0.98) per 1 SD, and a similar trend was observed in men (HR: 0.83; 95% CI: 0.64, 1.08) per 1 SD. The remaining associations between dietary variables (energy-dense dietary pattern, difference score, and energy intake) were all closer to the null in both women and men, and all 95% CIs included the null value. The association between dietary variables and low-trauma fracture was not changed when body mass index, BMD, or falls was excluded from the list of adjustment variables, even though all of these variables were associated with the outcome in women in the multivariate model, as described below. A strong association was observed between BMD and fracture in women (HR: 0.65; 95% CI: 0.55, 0.77) per 1 SD and in men (HR: 0.62; 95% CI: 0.45, 0.85) per 1 SD. There was a modest association between body mass index and fracture in women (HR: 1.04; 95% CI: 1.01, 1.06) per 1-kg/m2 increase, but not in men (HR: 0.98; 95% CI: 0.90, 1.06) per 1-kg/m2 increase. There was an association between falls and fracture in women (HR: 1.63; 95% CI: 1.11, 2.11) for those with prior falls compared with those without falls; however, for men, these results were inconclusive (HR: 0.66; 95% CI: 0.20, 2.15). We tested for a possible interaction with age and a found a slight age (P = 0.03) trend using age as a continuous variable. Stratifying by age category, we found an association between the nutrient-dense factors in older women (age ≥70 y; HR: 0.82; 95% CI: 0.71, 0.96) per 1 SD, but the finding was inconclusive in younger women (age <70 y; HR: 0.97; 95% CI: 0.76, 1.24) per 1 SD.
TABLE 3.
Hazard ratios (HRs) showing the association between dietary factor scores, energy intake, and differences between scores [energy-dense factor (EDF) – nutrient-dense factor (NDF)] and incident low-trauma fracture in the study sample of postmenopausal women and men aged ≥50 y1
| Variable | Women
|
Men
|
||||||
|---|---|---|---|---|---|---|---|---|
| Crude
|
Adjusted
|
Crude
|
Adjusted
|
|||||
| HR | 95% CI | HR | 95% CI | HR | 95% CI | HR | 95% CI | |
| NDF | 0.882 | 0.79, 0.992 | 0.862 | 0.76, 0.982 | 0.792 | 0.62, 1.002 | 0.83 | 0.64, 1.08 |
| EDF | 1.01 | 0.90, 1.13 | 1.01 | 0.89, 1.15 | 1.08 | 0.85, 1.36 | 1.06 | 0.82, 1.37 |
| Energy intake | 0.97 | 0.87, 1.08 | 0.96 | 0.85, 1.09 | 0.99 | 0.78, 1.26 | 1.02 | 0.80, 1.30 |
| Score difference (EDF – NDF) | 1.07 | 0.98, 1.16 | 1.08 | 0.98, 1.19 | 1.17 | 0.97, 1.40 | 1.13 | 0.92, 1.38 |
The parameter estimates are for each 1-SD increase in the NDF score, the EDF score, the log-transformed energy intake (1 SD is ≈36% change in energy intake), and the difference between the EDF and NDF scores. Analyses were run for the 2 factor scores and for the difference between the factor score and energy intake separately because of multicollinearity between intake and factor scores. All models had age as the time axis. Covariates in the adjusted models included BMI, bone mineral density, falls, prior fracture, comorbidities, smoking, milk consumption, and supplements (vitamin D and calcium). For women only, diagnosis of osteoporosis, antiresorptive use, education, alcohol use, physical activity, and sedentary hours were adjusted for. A high NDF score indicates a greater consumption of fruit, vegetables, and whole grains relative to other foods; a high EDF score indicates a greater consumption of chips or fries, processed meat, soft drinks, and certain desserts than of other foods. A high difference indicates more energy-dense foods than nutrient-dense foods.
Significant, P = 0.05.
DISCUSSION
In this study we found that an increased intake of nutrient-dense foods was associated with a lower risk of fracture, whereas the intake of energy-dense foods and total energy intake was not associated with fracture. These associations were similar in men and women and were independent of other important risk factors. The magnitude of the association between a diet high in nutrient-dense foods and fracture is comparable with the association between alcohol and fracture (per drink) (27), but less than that between smoking and fracture (smoker compared with non-smoker) (28) or between antiresorptive use and fracture (29). Smoking and alcohol are 2 risk factors included in the World Health Organization risk factor assessment tool (30), which includes implications concerning fracture prevention. The first is that dietary patterns might be considered a risk factor for osteoporotic fracture, and further work could be done to explore the development and validation of a short screening tool for inclusion in a risk factor assessment. The second is that population measures encouraging consumption of fruit, vegetables, and whole grains may reduce the population burden of fracture in addition to other putative health benefits (15, 17, 18, 31–33).
We also reported the association for the difference between factor scores. A higher difference score was associated with a higher (but not statistically significant) risk of fracture. This means that, at least for fracture, the most important direction of variation lies in the direction of the nutrient-dense factor, but it is possible that an increasing intake of energy-dense foods at the expense of nutrient-dense foods could increase the risk of fragility fracture.
Adjustment of the analysis for body mass index, BMD, or past falls did not alter the relation between a nutrient-dense diet and decreased fracture; hence, the association between diet and fracture risk was independent of these potentially mediating factors. The assessment of a role for falls was limited to the subjects’ recollection of falls during a specific time period, without an objective measurement of propensity to fall.
Few studies have assessed the association between overall dietary patterns and fracture. The Women’s Health Initiative Dietary Modification Trial showed that it was possible to increase the intake of fruit and vegetables and whole grains while reducing fat intake with dietary intervention and that those in the intervention group had a lower risk of falls (34). However, those who were in the intervention group did not have a lower risk of hip fracture or of other osteoporotic fracture. In this case the lower risk of falling in the intervention group might have been offset by lower serum estrogen and/or greater weight loss and hence bone loss (35, 36), which resulted in an overall null effect on fracture. Increased intake of fruit, vegetables, and low-fat dairy products is emphasized in the Dietary Approaches to Stop Hypertension, and a randomized trial based on this intervention was shown to reduce bone turnover (12).
Some studies on dietary patterns and fracture have shown that the consumption of diets with high ratios of animal to vegetable protein is associated with more bone loss, which may result in a higher risk of fracture (37, 38). It is likely that an energy-dense diet has a higher animal protein and fat content than does a nutrient-dense diet. It is hypothesized that animal protein provides a higher dietary acid load than does vegetable protein, and bone is lost, perhaps as a result of increased calcium excretion and negative calcium balance. However, other studies have not supported an association of negative calcium balance with animal compared with vegetable protein sources (39, 40). In general terms, protein intake has been associated with a bone benefit, as noted in a recent meta-analysis, which has called the dietary acid–ash hypothesis of bone loss into question (41).
One finding of our study was that diet was more related to fracture in older individuals. This might be related to a threshold effect, whereby food choices matter more in older individuals. It has been shown that the lack of dietary variety may lead to the inadequate intake of many crucial nutrients (42). The Canadian Community Health Survey shows that fewer than half of Canadians older than 65 y consume the recommended 5 servings of fruit and vegetables per day (43). The nutrient-dense factor score was strongly associated with both fruit and vegetable intake; hence, those with high scores were more likely to consume the recommended 5 servings of fruit and vegetables per day.
We previously found a strong relation between dietary patterns and body mass index, but not with BMD (14). We did, however, note an effect of the energy-dense pattern on BMD in subgroups of our population after adjustment for body mass index, an indirect association showing that higher body mass index does not necessarily mean higher BMD (14). Other researchers have found positive associations between a “prudent” or nutrient-dense diet and BMD in some populations (9) and others have noted beneficial aspects of some components of a nutrient-dense diet (10, 11, 44, 45). In contrast, a small intervention study showed no effect of increasing fruit and vegetable consumption on BMD, nor did it confirm posited mediation by an acid-balancing mechanism (46). It is likely that the dietary effects on BMD are incremental at best, and the differences noted might reflect long-term dietary patterns. It should be noted that we have not directly assessed lean mass, fat mass, or other measures of bone strength, all of which are possible mediating factors between diet and fracture (47).
The strengths of this study include the time-to-event analysis, measured height, measured weight, and a detailed questionnaire from which to assess potential confounding. This study also had some limitations. We did not include an assessment of weight or BMD change, which might be intermediate factors. The limited scope and specified portion size of the FFQ may yield biased estimates of absolute energy intake. Underrepresentation of ethnic minorities (<5% nonwhite) in the study may limit the generalizability of the findings. Some members of the cohort did not complete the FFQ. Those with a poor diet might be less likely to complete the FFQ and more likely to have died before having a fracture, with bias most likely toward the null. Finally, we cannot rule out the possibility of residual confounding because dietary patterns may be related to other unmeasured health behaviors.
In conclusion, a diet high in nutrient-dense foods (vegetables, fruits, whole grains) may reduce the risk of low-trauma fracture, especially among older women. Because older women are also at the highest risk of fracture, population measures to encourage increased intake of fruit, vegetables, and whole grains have the potential to lower the population burden of fracture, including hip fracture. Few recent studies have assessed dietary patterns related to fracture outcomes. The results here are complementary to those of studies that assessed the relation of specific foods and nutrients, because synergistic effects of food combinations might exist.
Acknowledgments
We thank all participants in CaMos, whose careful responses and attendance made this analysis possible.
The authors’ responsibilities were as follows—NK, SP, DAH, JCP, TA, and TT: data acquisition; LL, SP, DAH, JCP, TA, TT, and NK: study design; LL, SIB, and NK: data analysis; and LL, JCP, SIB, DG, SM, and NK: interpretation of the results. All authors were involved in drafting and revising the manuscript and read and approved the final manuscript. The following authors declared potential conflicts of interest as follows: DAH (advisory board, honoraria, or grants: Amgen, Eli Lilly, Novartis, and Warner-Chilcott), SIB (consulting: International Dairy Foods Association), TA (honoraria: Merck, Proctor & Gamble, Schering Plough, and Servier), TT (honoraria and grants: Abbott Laboratories, Bristol-Myers Squib, Novartis, and Sanofi-Aventis), DG (consulting: Eli Lilly, Novartis, Merck-Frosst, Proctor & Gamble, Sanofi-Aventis, and Servier), and SM (consulting: Proctor & Gamble, Sanofi-Aventis, Amgen, and Novartis). LL, SP, JCP, and NK did not have any conflicts of interest. The funding sources had no role in the conception of this analysis, statistical methods, or interpretation of the data.
The CaMos Research Group consists of the following participants: David Goltzman (co-principal investigator, McGill University); Nancy Kreiger (co-principal investigator, Toronto); Alan Tenenhouse (principal investigator emeritus, Toronto); the CaMos Coordinating Centre, McGill University, Mon-treal, Quebec: Suzette Poliquin (national coordinator emeritus), Suzanne Godmaire (research assistant), and Claudie Berger (study statistician); Memorial University, St John’s Newfoundland: Carol Joyce (director), Christopher Kovacs (co-director), and Emma Sheppard (coordinator); Dalhousie University, Halifax, Nova Scotia: Susan Kirkland and Stephanie Kaiser (co-directors) and Barbara Stanfield (coordinator); Laval University, Quebec City, Quebec: Jacques P Brown (director), Louis Bessette (co-director), and Marc Gendreau (coordinator); Queen’s University, Kingston, Ontario: Tassos Anastassiades (director), Tanveer Towheed (co-director), and Barbara Matthews (coordinator); University of Toronto, Toronto, Ontario: Bob Josse (director), Sophie Jamal (co-director), Tim Murray (past director), and Barbara Gardner-Bray (coordinator); McMaster University, Hamilton, Ontario: Jonathan D Adachi (director), Alexandra Papaioannou (co-director), and Laura Pickard (coordinator); University of Saskatchewan, Saskatoon, Saskatchewan: Wojciech P Olszynski (director), K Shawn Davison (co-director), and Jola Thingvold (coordinator); University of Calgary, Calgary, Alberta: David A Hanley (director) and Jane Allan (coordinator); and University of British Columbia, Vancouver, British Columbia: Jerilynn C Prior (director), Millan Patel (co-director), Yvette Vigna (coordinator), and Brian C Lentle (radiologist).
Footnotes
The Canadian Multicentre Osteoporosis Study was funded by the Canadian Institutes of Health Research, Merck-Frosst Canada Ltd, Eli Lilly Canada Inc, Novartis Pharmaceuticals Inc, The Alliance for Better Bone Health, Sanofi-Aventis, Procter & Gamble Pharmaceuticals Canada Inc, The Dairy Farmers of Canada, and The Arthritis Society.
References
- 1.Burge R, Dawson-Hughes B, Solomon DH, Wong JB, King A, Tosteson A. Incidence and economic burden of osteoporosis-related fractures in the United States, 2005–2025. J Bone Miner Res. 2007;22:465–75. doi: 10.1359/jbmr.061113. [DOI] [PubMed] [Google Scholar]
- 2.Boyd CM, Xue QL, Guralnik JM, Fried LP. Hospitalization and development of dependence in activities of daily living in a cohort of disabled older women: the Women’s Health and Aging Study I. J Gerontol A Biol Sci Med Sci. 2005;60:888–93. doi: 10.1093/gerona/60.7.888. [DOI] [PubMed] [Google Scholar]
- 3.Adachi JD, Ioannidis G, Berger C, et al. The influence of osteoporotic fractures on health-related quality of life in community-dwelling men and women across Canada. Osteoporos Int. 2001;12:903–8. doi: 10.1007/s001980170017. [DOI] [PubMed] [Google Scholar]
- 4.Ioannidis G, Papaioannou A, Hopman WM, et al. Relation between fractures and mortality: results from the Canadian Multicentre Osteoporosis Study. CMAJ. 2009;181:265–71. doi: 10.1503/cmaj.081720. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wells GA, Cranney A, Peterson J, et al. Alendronate for the primary and secondary prevention of osteoporotic fractures in postmenopausal women. Cochrane Database Syst Rev. 2008:CD001155. doi: 10.1002/14651858.CD001155.pub2. [DOI] [PubMed] [Google Scholar]
- 6.Wells G, Cranney A, Peterson J, et al. Risedronate for the primary and secondary prevention of osteoporotic fractures in postmenopausal women. Cochrane Database Syst Rev. 2008:CD004523. doi: 10.1002/14651858.CD004523.pub3. [DOI] [PubMed] [Google Scholar]
- 7.Siris ES, Chen YT, Abbott TA, et al. Bone mineral density thresholds for pharmacological intervention to prevent fractures. Arch Intern Med. 2004;164:1108–12. doi: 10.1001/archinte.164.10.1108. [DOI] [PubMed] [Google Scholar]
- 8.Tang BM, Eslick GD, Nowson C, Smith C, Bensoussan A. Use of calcium or calcium in combination with vitamin D supplementation to prevent fractures and bone loss in people aged 50 years and older: a meta-analysis. Lancet. 2007;370:657–66. doi: 10.1016/S0140-6736(07)61342-7. [DOI] [PubMed] [Google Scholar]
- 9.Okubo H, Sasaki S, Horiguchi H, et al. Dietary patterns associated with bone mineral density in premenopausal Japanese farmwomen. Am J Clin Nutr. 2006;83:1185–92. doi: 10.1093/ajcn/83.5.1185. [DOI] [PubMed] [Google Scholar]
- 10.Macdonald HM, New SA, Golden MH, Campbell MK, Reid DM. Nutritional associations with bone loss during the menopausal transition: evidence of a beneficial effect of calcium, alcohol, and fruit and vegetable nutrients and of a detrimental effect of fatty acids. Am J Clin Nutr. 2004;79:155–65. doi: 10.1093/ajcn/79.1.155. [DOI] [PubMed] [Google Scholar]
- 11.Tucker KL, Chen H, Hannan MT, et al. Bone mineral density and dietary patterns in older adults: the Framingham Osteoporosis Study. Am J Clin Nutr. 2002;76:245–52. doi: 10.1093/ajcn/76.1.245. [DOI] [PubMed] [Google Scholar]
- 12.Lin PH, Ginty F, Appel LJ, et al. The DASH diet and sodium reduction improve markers of bone turnover and calcium metabolism in adults. J Nutr. 2003;133:3130–6. doi: 10.1093/jn/133.10.3130. [DOI] [PubMed] [Google Scholar]
- 13.Johnell O, Kanis JA, Oden A, et al. Predictive value of BMD for hip and other fractures. J Bone Miner Res. 2005;20:1185–94. doi: 10.1359/JBMR.050304. [DOI] [PubMed] [Google Scholar]
- 14.Langsetmo L, Poliquin S, Hanley DA, et al. Dietary patterns in Canadian men and women ages 25 and older: relationship to demographics, body mass index, and bone mineral density. BMC Musculoskelet Disord. 2010;11:20. doi: 10.1186/1471-2474-11-20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fung TT, Willett WC, Stampfer MJ, Manson JE, Hu FB. Dietary patterns and the risk of coronary heart disease in women. Arch Intern Med. 2001;161:1857–62. doi: 10.1001/archinte.161.15.1857. [DOI] [PubMed] [Google Scholar]
- 16.Fung TT, Schulze M, Manson JE, Willett WC, Hu FB. Dietary patterns, meat intake, and the risk of type 2 diabetes in women. Arch Intern Med. 2004;164:2235–40. doi: 10.1001/archinte.164.20.2235. [DOI] [PubMed] [Google Scholar]
- 17.Hu FB, Rimm EB, Stampfer MJ, Ascherio A, Spiegelman D, Willett WC. Prospective study of major dietary patterns and risk of coronary heart disease in men. Am J Clin Nutr. 2000;72:912–21. doi: 10.1093/ajcn/72.4.912. [DOI] [PubMed] [Google Scholar]
- 18.Newby PK, Muller D, Hallfrisch J, Andres R, Tucker KL. Food patterns measured by factor analysis and anthropometric changes in adults. Am J Clin Nutr. 2004;80:504–13. doi: 10.1093/ajcn/80.2.504. [DOI] [PubMed] [Google Scholar]
- 19.Kreiger N, Tenenhouse A, Joseph L, et al. Research notes: the Canadian Multicentre Osteoporosis Study (CaMos)—background, rationale, methods. Can J Aging. 1999;18:376–87. [Google Scholar]
- 20.Poliquin S, Joseph L, Gray-Donald K. Calcium and vitamin D intakes in an adult Canadian population. Can J Diet Pract Res. 2009;70:21–7. doi: 10.3148/70.1.2009.21. [DOI] [PubMed] [Google Scholar]
- 21.Block G, Hartman AM, Naughton D. A reduced dietary questionnaire: development and validation. Epidemiology. 1990;1:58–64. doi: 10.1097/00001648-199001000-00013. [DOI] [PubMed] [Google Scholar]
- 22.Villeneuve PJ, Johnson KC, Kreiger N, Mao Y. Risk factors for prostate cancer: results from the Canadian National Enhanced Cancer Surveillance System. The Canadian Cancer Registries Epidemiology Research Group. Cancer Causes Control. 1999;10:355–67. doi: 10.1023/a:1008958103865. [DOI] [PubMed] [Google Scholar]
- 23.Willett WC, Sampson L, Stampfer MJ, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol. 1985;122:51–65. doi: 10.1093/oxfordjournals.aje.a114086. [DOI] [PubMed] [Google Scholar]
- 24.Health Canada. Canadian nutrient file. 2008 Available from: http://2051939351/cnfonline/ (cited 18 July 2008)
- 25.Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol. 1986;124:17–27. doi: 10.1093/oxfordjournals.aje.a114366. [DOI] [PubMed] [Google Scholar]
- 26.Genant HK. Universal standardization for dual X-ray absorptiometry: patient and phantom cross-calibration results. J Bone Miner Res. 1995;10:997–8. doi: 10.1002/jbmr.5650100624. [DOI] [PubMed] [Google Scholar]
- 27.Kanis JA, Johansson H, Johnell O, et al. Alcohol intake as a risk factor for fracture. Osteoporos Int. 2005;16:737–42. doi: 10.1007/s00198-004-1734-y. [DOI] [PubMed] [Google Scholar]
- 28.Kanis JA, Johnell O, Oden A, et al. Smoking and fracture risk: a meta-analysis. Osteoporos Int. 2005;16:155–62. doi: 10.1007/s00198-004-1640-3. [DOI] [PubMed] [Google Scholar]
- 29.Langsetmo LA, Morin S, Richards JB, et al. Effectiveness of anti-resorptives for the prevention of nonvertebral low-trauma fractures in a population-based cohort of women. Osteoporos Int. 2009;20:283–90. doi: 10.1007/s00198-008-0656-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Kanis JA, Oden A, Johansson H, Borgstrom F, Strom O, McCloskey E. FRAX and its applications to clinical practice. Bone. 2009;44:734–43. doi: 10.1016/j.bone.2009.01.373. [DOI] [PubMed] [Google Scholar]
- 31.Fung TT, Stampfer MJ, Manson JE, Rexrode KM, Willett WC, Hu FB. Prospective study of major dietary patterns and stroke risk in women. Stroke. 2004;35:2014–9. doi: 10.1161/01.STR.0000135762.89154.92. [DOI] [PubMed] [Google Scholar]
- 32.Lutsey PL, Steffen LM, Stevens J. Dietary intake and the development of the metabolic syndrome: the Atherosclerosis Risk in Communities Study. Circulation. 2008;117:754–61. doi: 10.1161/CIRCULATIONAHA.107.716159. [DOI] [PubMed] [Google Scholar]
- 33.Kant AK, Graubard BI, Schatzkin A. Dietary patterns predict mortality in a national cohort: the National Health Interview Surveys, 1987 and 1992. J Nutr. 2004;134:1793–9. doi: 10.1093/jn/134.7.1793. [DOI] [PubMed] [Google Scholar]
- 34.McTiernan A, Wactawski-Wende J, Wu L, et al. Low-fat, increased fruit, vegetable, and grain dietary pattern, fractures, and bone mineral density: the Women’s Health Initiative Dietary Modification Trial. Am J Clin Nutr. 2009;89:1864–76. doi: 10.3945/ajcn.2008.26956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Ensrud KE, Fullman RL, Barrett-Connor E, et al. Voluntary weight reduction in older men increases hip bone loss: the osteoporotic fractures in men study. J Clin Endocrinol Metab. 2005;90:1998–2004. doi: 10.1210/jc.2004-1805. [DOI] [PubMed] [Google Scholar]
- 36.Ensrud KE, Ewing SK, Stone KL, Cauley JA, Bowman PJ, Cummings SR. Intentional and unintentional weight loss increase bone loss and hip fracture risk in older women. J Am Geriatr Soc. 2003;51:1740–7. doi: 10.1046/j.1532-5415.2003.51558.x. [DOI] [PubMed] [Google Scholar]
- 37.Sellmeyer DE, Stone KL, Sebastian A, Cummings SR. A high ratio of dietary animal to vegetable protein increases the rate of bone loss and the risk of fracture in postmenopausal women. Study of Osteoporotic Fractures Research Group. Am J Clin Nutr. 2001;73:118–22. doi: 10.1093/ajcn/73.1.118. [DOI] [PubMed] [Google Scholar]
- 38.Weikert C, Walter D, Hoffmann K, Kroke A, Bergmann MM, Boeing H. The relation between dietary protein, calcium and bone health in women: results from the EPIC-Potsdam cohort. Ann Nutr Metab. 2005;49:312–8. doi: 10.1159/000087335. [DOI] [PubMed] [Google Scholar]
- 39.Kerstetter JE, Wall DE, O’Brien KO, Caseria DM, Insogna KL. Meat and soy protein affect calcium homeostasis in healthy women. J Nutr. 2006;136:1890–5. doi: 10.1093/jn/136.7.1890. [DOI] [PubMed] [Google Scholar]
- 40.Roughead ZK, Hunt JR, Johnson LK, Badger TM, Lykken GI. Controlled substitution of soy protein for meat protein: effects on calcium retention, bone, and cardiovascular health indices in postmenopausal women. J Clin Endocrinol Metab. 2005;90:181–9. doi: 10.1210/jc.2004-0393. [DOI] [PubMed] [Google Scholar]
- 41.Fenton TR, Lyon AW, Eliasziw M, Tough SC, Hanley DA. Meta-analysis of the effect of the acid-ash hypothesis of osteoporosis on calcium balance. J Bone Miner Res. 2009;24:1835–40. doi: 10.1359/jbmr.090515. [DOI] [PubMed] [Google Scholar]
- 42.Marshall TA, Stumbo PJ, Warren JJ, Xie XJ. Inadequate nutrient intakes are common and are associated with low diet variety in rural, community-dwelling elderly. J Nutr. 2001;131:2192–6. doi: 10.1093/jn/131.8.2192. [DOI] [PubMed] [Google Scholar]
- 43.Riediger ND, Moghadasian MH. Patterns of fruit and vegetable consumption and the influence of sex, age and socio-demographic factors among Canadian elderly. J Am Coll Nutr. 2008;27:306–13. doi: 10.1080/07315724.2008.10719704. [DOI] [PubMed] [Google Scholar]
- 44.Prynne CJ, Mishra GD, O’Connell MA, et al. Fruit and vegetable intakes and bone mineral status: a cross sectional study in 5 age and sex cohorts. Am J Clin Nutr. 2006;83:1420–8. doi: 10.1093/ajcn/83.6.1420. [DOI] [PubMed] [Google Scholar]
- 45.Zalloua PA, Hsu YH, Terwedow H, et al. Impact of seafood and fruit consumption on bone mineral density. Maturitas. 2007;56:1–11. doi: 10.1016/j.maturitas.2006.05.001. [DOI] [PubMed] [Google Scholar]
- 46.Macdonald HM, Black AJ, Aucott L, et al. Effect of potassium citrate supplementation or increased fruit and vegetable intake on bone metabolism in healthy postmenopausal women: a randomized controlled trial. Am J Clin Nutr. 2008;88:465–74. doi: 10.1093/ajcn/88.2.465. [DOI] [PubMed] [Google Scholar]
- 47.Reid IR. Relationships among body mass, its components, and bone. Bone. 2002;31:547–55. doi: 10.1016/s8756-3282(02)00864-5. [DOI] [PubMed] [Google Scholar]