Dietary Patterns, Bone Mineral Density, and Risk of Fractures: A Systematic Review and Meta-Analysis

The aim of this systematic review was to assess the evidence on the relation between dietary patterns, bone mineral density (BMD), and risk of fracture in different age groups. Medline and Embase were searched for articles that identified dietary patterns and related these to BMD or risk of fracture through May 2018. Multivariable adjusted odds ratios (ORs) and 95% confidence intervals (95%CI) comparing the lowest and highest categories of dietary pattern were combined by using a random effects meta-analysis. In total, 31 studies were selected for review, including 18 cohorts, 1 case-control, and 12 cross-sectional studies, in the meta-analysis of Prudent/healthy and Western/unhealthy dietary pattern, BMD, and risk of fracture. There was evidence of a lower risk of fracture when intakes in the highest categories were compared with the lowest categories of Prudent/healthy dietary pattern (OR = 0.81; 95%CI: 0.69, 0.95; p = 0.01). In contrast, when intakes in the highest categories were compared with the lowest categories of Western/unhealthy dietary pattern, a greater risk of fracture (OR = 1.10; 95%CI: 1.02, 1.19; p = 0.01) was observed among men. The present systematic review and meta-analysis provides evidence of an inverse association between a Prudent/healthy dietary pattern and risk of low BMD and a positive relation between Western/unhealthy dietary pattern and risk of low BMD.


Introduction
Osteoporosis has been defined as a chronic disease characterized by low bone mass and bone tissue deterioration, with an increase in bone fragility and risk of osteoporotic fracture [1,2]. Due to its prevalence and contribution to morbidity, osteoporosis has been considered an important public health concern [3]. Osteoporotic fracture frequently results in disability, higher healthcare costs, and excess mortality [3,4]. Globally, osteoporosis-related fractures affect approximately 33% of women and 20% of men over the age of 50 years [5]. Low bone mineral density (BMD) has been considered to be a hallmark of osteoporosis, as well as a predictor of osteoporotic fracture [6][7][8]. Thus, a better understanding of the main prognostic factors of low BMD would have significant repercussions for public health. included in the evidence tables. We also reviewed the reference lists of all identified manuscripts for further identification of possible studies.

Selection of Studies
Two reviewers independently screened titles and abstracts for all of the potentially relevant studies, and coded them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. Fulltext study reports/publications were retrieved, and two reviewers screened them to identify studies for inclusion, or recording reasons for exclusion of those ineligible. Any disagreements were discussed and resolved and, if required, a third reviewer was consulted. Duplicates were excluded and multiple reports of the same study were collated so that each study (analysis) was unique in the review (Figure 1).

Data Extraction and Management
One reviewer extracted study characteristics from included studies. A second reviewer spotchecked characteristics for accuracy against the trial report. For each study, we describe: methods (study design, study duration, details of any 'run-in' period, number of study centers and location (country), study setting, and date of study); participants (sample size, mean age, age range, sex, ethnicity, calcium and vitamin D intake (if available), and inclusion and exclusion criteria); and interventions (types of dietary patterns).

Data Extraction and Management
One reviewer extracted study characteristics from included studies. A second reviewer spot-checked characteristics for accuracy against the trial report. For each study, we describe: methods (study design, study duration, details of any 'run-in' period, number of study centers and location (country), study setting, and date of study); participants (sample size, mean age, age range, sex, ethnicity, calcium and vitamin D intake (if available), and inclusion and exclusion criteria); and interventions (types of dietary patterns).

Assessment of Potential Bias in Included Studies
Two reviewers independently assessed potential bias in each study using the Cochrane handbook and the criteria outlined in the GRADEpro program [55,56]. Disagreements were discussed and resolved and, if necessary, a third reviewer was consulted.

Data Synthesis
We conducted a meta-analysis with available information from the identified studies with appropriate data.

Results of the Search and Study Selection
From the literature search through the Medline-PubMed and Ovid-Embase databases, we identified and screened 3346 titles. Of these, we excluded 3256 because they did not conform to the PICOS criteria. To assess discrepancies between the reviewers, a concordance analysis was conducted; in general, "very good" strength of agreement was found (PubMed 0.85, p = 0.0001; and Ovid-Embase 0.86, p = 0.0001). After the first screening, abstracts of 90 articles were reviewed and, of these, a total of 68 papers were selected for full-text revision. Of the remaining 68 articles, the following were excluded: 34 studies for duplicity, 6 articles because they did not use an "a posteriori" method to assess the dietary patterns, 2 studies that focused on athletes, and 1 study that focused on the same population. Thus, a total of 25 studies were eligible for analysis. Subsequently, after a detailed review of the 25 articles, 18 new studies (including two related systematic reviews) were identified, their full texts were retrieved for detailed evaluation, and an additional 6 studies were included in the analysis. Finally, 31 studies were included in our systematic review ( Figure 1). Of these, 12 studies, based on "a posteriori" methods to derive dietary patterns, were considered for the meta-analysis [27][28][29]34,36,39,42,44,49,50,52,53]. These 31 observational studies reported data on 175,060 participants. Six studies included information on children and adolescents; 9 included information on younger adults (<50 years); and 16 included information on older adults (≥50 years). Twenty-nine studies reported BMD or BMC measures using dual-energy X-ray absorptiometry (DXA). Only four studies reported fracture as an outcome and only in adults ≥50 years. For this clinical outcome, the population followed was 123,193 participants.

Included Studies
The methodological characteristics of the 31 studies  incorporated into the present analysis are reported in Tables 1 and 2. These articles were published between 2002 and 2018 and included 18 cohort studies [24][25][26][30][31][32][33][39][40][41][42][43][44][45][46][47]52,53], 1 case-control study [54], and 12 cross-sectional studies [27][28][29][34][35][36][37][38][47][48][49][50][51]. Ten studies were conducted in Asia [27][28][29]36,38,43,44,49,50,54], 10 in Europe [24,25,31,37,[39][40][41]45,46,52], 7 in North America [26,[33][34][35]47,52,53], 3 in Australia [30,32,42], and 1 in South America [48]. Data from two studies [39,40] were from the Rotterdam Study. Table 1. The main characteristics of the epidemiological studies on the association between bone mineral density or bone mineral content and dietary patterns defined using the "a posteriori" approach. "Potatoes, rice, and vegetables", "Refined grains and confectionary", and "Dairy and whole grains" dietary patterns Sex, ethnicity, birth weight Z-score, adherence scores for the two-other dietary patterns, total energy intake, time interval between dietary assessment and visit, age at visit, height at visit, weight at visit, and maternal BMI 3 at enrolment Adherence to a "Dairy and whole grains" pattern was positively associated with BMD 4 . The other patterns were not associated with BMD Monjardino, et al., 2015 [25] Portugal 1007 17 years FFQ Cluster analysis "Healthier", "Dairy products", "Fast food and sweets", and "Lower intake" dietary patterns Height, weight, total energy intake, and age at menarche (in girls) Among girls, adherence to a "Lower intake" pattern was negatively associated with BMD, compared with subjects with a "Healthier" pattern Wosje, et al., 2010 [26] USA 5 325 6.8-7.8 years 3-day food records RRR 6 Pattern 1 and pattern 2 (not labeled) Race, sex, height, weight, energy intake, calcium intake, physical activity, and time spent viewing television and playing outdoors A pattern characterized by high intakes of dark green vegetables, deep-yellow vegetables, and low intakes of processed meats, fried chicken and fish, and fried potatoes was associated with higher bone mass Cross-sectional studies, children and adolescents Mu, et al., 2014 [27] China 1319 16-20 years FFQ Factor analysis (varimax rotation) Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern" Sex, physical activity, economic status, passive smoking, calcium supplements, body mass index The findings suggested that there was a positive association between a "Chinese traditional" dietary pattern and healthy BMD and that this same association was observed between "Calcium food pattern" and BMD. In contrast, "Western pattern" was negatively related with BMD; however, the relationship was not statistically significant These results indicate that the intake of milk and cereal is important for the bone health of Korean adolescents, whose diets are composed mainly of grains and vegetables "Healthy", "Traditional", "Meats and nuts" only for women, "Refined" only for men, and "Social" dietary patterns Age, BMI, smoking, physical activity, father's social class, and energy intake Women with higher scores of "Meats and nuts" pattern had significantly greater BMD. Men with higher scores of "Refined" pattern had significantly lower BMD. The other patterns were not associated significantly with BMD "Chicken", "Fish", "Processed foods", "Red meat", and "Low-fat milk" dietary patterns Age, sex, estrogen status, BMI, height, total energy intake, current smoking status, alcohol intake, calcium supplement use and vitamin D supplement use, and physical activity BMD was higher among subjects in the "Low-fat milk" pattern, compared with subjects in the "Processed foods" and "Red meat" dietary patterns "Meat, dairy, and bread", "Meat and sweet baked products", "Sweet baked products", "Alcohol", "Candy", and "Fruit, vegetables, and cereal" dietary patterns BMI, height, age, energy intake, physical activity score, smoking, vitamin D supplement use, calcium supplement use, season, and estrogen use for women Men and women in the "Candy" pattern had significantly lower BMD than in the "Fruit, vegetables, and cereal" pattern. Men in the "Fruit, vegetables, and cereal" pattern had the greatest average of BMD of all subjects "Healthy", "Red meat and refined cereals", "Low-fat dairy", "Sweet foods, coffee, and tea", and "Western" dietary patterns Energy intake, calcium intake, lean mass, height, and postmenopausal time The "sweet foods, coffee, and tea" dietary pattern was inversely and significantly associated with BMD. The other patterns were not associated significantly with BMD
Four studies used RRR to derive dietary patterns [26,30,39,41]; one was conducted in children and adolescents [26], another in young adults [30], and two in older adults [39,41]. In each of these, two dietary patterns were extracted; one was mainly characterized by high intakes of refined grains and cereals, red meats, processed meats, eggs, fats, sweetened beverages, and sweets. Only one study [26] reported a negative association between the first dietary pattern and BMC. The other three studies [30,39,41] did not find any significant association. Conversely, in all articles, a dietary pattern that was mainly represented by low-fat milk, fruit, dairy products, vegetables, fish, legumes, and whole grains was significantly related with greater BMD and/or BMC [26,30,39,41].
Four cohort studies derived a dietary pattern by means of a cluster analysis (CA) [25,35,46,47]. Of these, one was conducted in children and adolescents [25], one in young adults [35], and two in adults older than 50 years [46,47]. In general, these showed that individuals consuming a "Healthy" dietary pattern had higher BMD than those with an "Unhealthy" or a "Western" pattern. For example, in older adults, a diet high in fruits, vegetables, and breakfast cereal and limited in less nutrient-dense foods was associated with the highest BMD, particularly in men [47].
Only four studies, including three cohort studies [39,52,53] and one case-control study [54], reported associations between dietary patterns and risk of fracture. They were conducted in men and women over 50 years. One used RRR [39] and the others [52][53][54] employed factor analysis methods to construct dietary patterns. One study using a factor analysis [53] did not find any association between the "Prudent" or "Western" pattern and fracture risk among older men or women. The remaining three studies [39,53,54] identified significant associations between favorable dietary patterns (generally characterized by high intakes of vegetables, fruits, fish, lean poultry, legumes, nuts, whole grains, and water and labeled "Healthy" or "Prudent") were significantly associated with a lower risk of fracture [39,53,54]. In contrast, the unfavorable dietary patterns [39,53,54] (rich in red meat, processed meat, poultry with skin, animal organ meat, cooking oil, soft drinks, hamburgers, hotdog, ice cream, doughnuts, margarine, and butter, and labeled "Sweets, animal fat" [39], "Western" (53), or "Energy-dense"), were associated with a higher risk of fracture [39,53,54].

Dietary Patterns for the Meta-Analysis
In total, 12 studies (4 cohort, 1 case-control, and 7 cross-sectional) were included in the meta-analysis of "Prudent/Healthy" and "Western/Unhealthy" dietary patterns, BMD, and risk of fracture.
Of these, nine studies (three cross-sectional in children and adolescents, two cross-sectional in younger adults, and five (two cohorts and two cross-sectional studies) in older adults), analyzed the association between a "Prudent/Healthy" dietary pattern and BMD. Four studies (three cohort studies and one case-control) evaluated the relation between risk of fracture and "Prudent/Healthy" dietary pattern among older adults.
The association between "Prudent/Healthy" and "Western/Unhealthy" dietary patterns and BMD or fracture was estimated using a random-effects meta-analysis with 95% CI.

Risk of Bias in the Studies Included in the Meta-Analysis
Two reviewers (L.M.-S. and E.D.-G.) independently assessed the risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Review of Interventions [55]. Disagreements were solved by discussion or by involving a third author (P.C). We assessed the risk of bias according to the following domains of observational studies: 1. Risk of bias; 2. Inconsistency; 3. Indirectness; 4. Imprecision; and 5. Other considerations (Publication bias, large effect, plausible confounding, and dose-response gradient). We present the risk of bias analysis according to the GRADEpro Guideline Development Tool (56) in Table 3. Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

Prudent Dietary Pattern or Western Dietary Pattern for Bone Mineral Density Certainty Assessment Summary of Findings
Nutrients 2018, 10, 0 12 of 19  The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝ VERY LOW The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns LOW The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern.
"Chinese and western", "Westernization", and "Meat" dietary patterns  Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

Prudent Dietary Pattern or Western Dietary Pattern for Bone Mineral Density
Certainty Assessment Summary of Findings № of The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern.

Inconsistency Indirectness Imprecision Publication
"Chinese and western", "Westernization", and "Meat" dietary patterns Relation between "Prudent" DP and BMD in young adults ≥20 years to <50years (assessed with: PCA-factor analysis (varimax rotation))    Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

Prudent Dietary Pattern or Western Dietary Pattern for Bone Mineral Density
Certainty Assessment Summary of Findings № of

Summary of Findings
Relation between "Prudent" DP and BMD in Children and adolescents (assessed with: PCA-factor analysis (varimax rotation), Cluster analysis, and RRR)       The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝ VERY LOW The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns LOW The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

(3 observational studies) not serious not serious not serious serious a none
Nutrients 2018, 10, x FOR PEER REVIEW 27 of 36 Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

Prudent Dietary Pattern or Western Dietary Pattern for Bone Mineral Density
Certainty Assessment Summary of Findings № of

Inconsistency Indirectness Imprecision Publication Bias
Overall

3105
(3 observational studies) not serious not serious not serious not serious none ⊕⊕⃝ ⃝

LOW
The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝

VERY LOW
The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns VERY LOW The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns Relation between "Prudent" DP and BMD in young adults ≥20 years to <50years (assessed with: PCA-factor analysis (varimax rotation))  Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

3105
(3 observational studies) not serious not serious not serious not serious none ⊕⊕⃝ ⃝

LOW
The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝

VERY LOW
The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns  Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

3105
(3 observational studies) not serious not serious not serious not serious none ⊕⊕⃝ ⃝

LOW
The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝

VERY LOW
The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns  Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

3105
(3 observational studies) not serious not serious not serious not serious none ⊕⊕⃝ ⃝

LOW
The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝

VERY LOW
The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns  Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

3105
(3 observational studies) not serious not serious not serious not serious none ⊕⊕⃝ ⃝

LOW
The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝

VERY LOW
The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns   Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

3105
(3 observational studies) not serious not serious not serious not serious none ⊕⊕⃝ ⃝

LOW
The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝

VERY LOW
The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the  Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

Summary of Findings
Relation between "Prudent" DP and BMD in Children and adolescents (assessed with: PCA-factor analysis (varimax rotation), Cluster analysis, and RRR)

3105
(3 observational studies) not serious not serious not serious not serious none ⊕⊕⃝ ⃝

LOW
The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns Relation between "Western" DP and BMD in Children and adolescents (assessed with: Factor analysis (varimax rotation) and PCA-factor analysis (varimax rotation)) The dietary patterns: Four dietary patterns were LOW The dietary pattern: "Meat, alcohol, and sugar", "Vegetables and soya sauce", "White rice, kimchi, and seaweed" and "Dairy and fruit" dietary patterns. Dietary pattern 1, Dietary pattern 2, and Dietary pattern 3 (not labeled)   The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝ VERY LOW The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns LOW The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern.

(3 observational studies) not serious not serious not serious serious a none
Nutrients 2018, 10, x FOR PEER REVIEW 27 of 36 Table 3. The risk of bias analysis according to the Cochrane guideline to report the risk of bias analysis (using GRADEpro).

3105
(3 observational studies) not serious not serious not serious not serious none ⊕⊕⃝ ⃝

LOW
The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns Relation between "Western" DP and BMD in Children and adolescents (assessed with: Factor analysis (varimax rotation) and PCA-factor analysis (varimax rotation))

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝

VERY LOW
The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

VERY LOW
The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern.

LOW
The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns Relation between "Western" DP and BMD in Children and adolescents (assessed with: Factor analysis (varimax rotation) and PCA-factor analysis (varimax rotation))

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝

VERY LOW
The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

LOW
The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns Relation between "Western" DP and BMD in Children and adolescents (assessed with: Factor analysis (varimax rotation) and PCA-factor analysis (varimax rotation)) The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the      The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝ VERY LOW The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

VERY LOW
The dietary patterns: Dietary pattern 1 ("Fruit, vegetables, and dairy", "Sweets, animal fat, and low meat"); Dietary pattern 2 ("Prudent or western"); and Dietary pattern 3 ("Nutrient-dense" and "Energy-dense"). Relation between "Western" DP and risk of fracture in WOMEN (assessed with: RRR: Reduced rank regression; PCA-factor analysis (varimax rotation))  The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝ VERY LOW The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns LOW The dietary patterns: Dietary pattern 1 ("Fruit, vegetables, and dairy", "Sweets, animal fat, and low meat"); Dietary pattern 2 ("Prudent or western"); and Dietary pattern 3 ("Nutrient-dense" and "Energy-dense"). Relation between "Prudent" DP and risk of fracture in MEN (assessed with: RRR: Reduced rank regression; PCA-factor analysis (varimax rotation))  The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns Relation between "Western" DP and BMD in Children and adolescents (assessed with: Factor analysis (varimax rotation) and PCA-factor analysis (varimax rotation))

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝ VERY LOW The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns LOW The dietary patterns: Dietary pattern 1 ("Fruit, vegetables, and dairy", "Sweets, animal fat, and low meat"); Dietary pattern 2 ("Prudent or western"); and Dietary pattern 3 ("Nutrient-dense" and "Energy-dense"). Relation between "Western" DP and risk of fracture in MEN (assessed with: RRR: Reduced rank regression; PCA-factor analysis (varimax rotation))  The dietary patterns: Four dietary patterns were identified: "Western food pattern", "Animal protein pattern", "Calcium food pattern", and "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, "the Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns Relation between "Western" DP and BMD in Children and adolescents (assessed with: Factor analysis (varimax rotation) and PCA-factor analysis (varimax rotation))

3105
(3 observational studies) not serious not serious not serious serious a none ⊕⃝ ⃝ ⃝ VERY LOW The dietary patterns: Four dietary patterns were identified: the "Western food pattern", the "Animal protein pattern", the "Calcium food pattern", and the "Chinese traditional pattern". Four different dietary patterns were identified: the "Traditional Korean" dietary pattern, the "Fast food" dietary pattern, the "Milk and cereal" dietary pattern, and the "Snacks" dietary pattern. "Chinese and western", "Westernization", and "Meat" dietary patterns LOW The dietary patterns: Dietary pattern 1 ("Fruit, vegetables, and dairy", "Sweets, animal fat, and low meat"); Dietary pattern 2 ("Prudent or western"); and Dietary pattern 3 ("Nutrient-dense" and "Energy-dense"

Discussion
To our knowledge, this is the first systematic review and meta-analysis of dietary patterns, BMD, and risk of fracture based on "a posteriori" derived dietary patterns. Our meta-analysis suggests that a "Prudent/Healthy" dietary pattern may decrease the risk of low BMD among children and adolescents, young adults, and older adults. Further, our results indicate that a "Western/Unhealthy" dietary pattern may increase the risk of low BMD in adults older than 50 years. Among older men, the pooled results of the cohort studies showed a significant association between "Prudent/Healthy" dietary patterns and a reduced risk of fracture, while "Western/Unhealthy" patterns were related to a higher risk of fracture. Among women, the associations between "Prudent/Healthy" or "Western/Unhealthy" were in the expected direction, but did not achieve statistical significance.
Our analyses identified two consistent "a posteriori" types of dietary patterns: "Prudent/Healthy" and "Western/Unhealthy". These total dietary patterns may contribute to diverse biological mechanisms that affect BMD and risk of fracture. In general, the "Prudent/Healthy" dietary pattern was characterized by high intakes of fruits, vegetables, whole grains, legumes, nuts, fish, low-fat dairy products, and low-fat milk, and low intakes of soft drinks, sugars, refined grains or cereals, red meat, and processed meat. Previous systematic reviews [62][63][64], have reported that the consumption of fruit and vegetables may reduce the risk of low BMD and fracture. Intake of fruits and vegetables, as well as whole grains, increases intake of several important vitamins, minerals, and phytonutrients, which may contribute to bone health through effects on acid-based balance [65], calcium metabolism [66], antioxidant capacity, which suppresses osteoclast action [67], and bone matrix formation [68], and by decreasing homocysteine concentration [69], among others.
Additional components of the "Prudent/Healthy" dietary pattern include fish and nuts. These contain polyunsaturated fatty acids (particularly n-3 fatty acids), which have been associated with anti-inflammatory properties that promote bone health [70]. Finally, dairy products and milk are an important source of calcium, magnesium, vitamin D, and protein, which are needed for bone matrix formation and preservation [71,72].
On the other hand, "Western/Unhealthy" dietary patterns tend to be characterized by red meat, processed meat, soft drinks, refined grains or cereals, fast food, and sweets. These dietary components contribute saturated fats, sodium, added sugars, and phosphorus [73,74], which have been linked to a higher risk of low BMD and fracture incidence. A higher intake of these nutrients contributes to imbalances that may decrease osteoblast differentiation and bone development [75], alter the equilibrium of calcium [76], or contribute to acid load [77].
There are some methodological limitations to the interpretation of our results. First, most of the studies included in our systematic review and meta-analysis were cross-sectional and do not allow for possible changes in diet over time. Second, most of the "a posteriori" dietary patterns included in our analysis were derived through factor analysis methods, which depend on subjective decisions during analysis (definition and number of food groups entered, number of factors to retain) and may lead to variation in pattern definition across studies [19][20][21][78][79][80]. In the studies we reviewed, food groups and loading factors in the dietary patterns derived were not identical between studies. Although this may result in a misclassification bias, the results across studies in the overall dietary patterns identified were quite consistent. To reduce potential bias in the meta-analysis, we selected only those dietary patterns that showed reasonably similar factor loadings for the most frequently consumed food groups. Such a methodology has been employed by other systematic reviews and meta-analyses evaluating dietary patterns and different health outcomes [81][82][83]. Most studies included in the present review and meta-analysis used food frequency questionnaires (FFQs), which often differ across studies. Nonetheless, previous studies comparing dietary patterns derived from FFQs as well as from 24-h dietary recalls or diet histories have found reasonable reproducibility [84][85][86]. Additionally, cultural, racial, and ethnic consumption patterns and food choices differ across populations. Despite this, we found similarities in the foods or food groups included in the different studies, as well as in the dietary pattern definitions. Finally, other dietary patterns could be related to BMD or BMC and fracture risk. In the present study, we only included the most commonly identified dietary patterns (similar foods or food groups); thus, it is possible that other dietary patterns, and, also, dietary patterns derived with different methods, such as an "a priori" score-based approach, could be associated with BMD or BMC and fracture risk.
The studies also varied in the confounding variables used in the analysis. However, most studies adjusted for the main documented risk factors for low BMD and risk of fracture, i.e., sex, age, physical activity, multivitamin use, height, weight, passive or active smoking, and, in women, age at menarche or menopausal status. Still, due to the observational nature of the included studies, residual confounding cannot be ruled out.
The quality of assessed evidence (GRADEpro GDT) [56] ranged from "moderate" to "very low", based on the observational design of the studies included in the meta-analysis. Further issues included identified limitations in study execution (explained withdrawals). In contrast, our study has several strengths, including the lack of identification of serious risk of bias, low heterogeneity across studies (<50%), and clear variability along the scores.

Conclusions
In summary, the present systematic review and meta-analysis provides evidence of an inverse association between a "Prudent/Healthy" dietary pattern and risk of low BMD across all age groups included. Conversely, a positive relation between "Western/Unhealthy" dietary pattern and low BMD was found only in older adults. Further, the meta-analyses contribute evidence that a "Prudent/Healthy" dietary pattern is protective against fracture risk among men, while a "Western/Unhealthy" dietary pattern is associated with greater fracture incidence. These results should encourage health professionals to emphasize the importance of consuming healthy diets that include fruits, vegetables, whole grains, fish, legumes, nuts, low-fat dairy products, milk, and water, while avoiding refined foods high in saturated fat and added sugars, for the prevention of low BMD and fracture risk.