ASSOCIATION BETWEEN DIETARY INFLAMMATORY INDEX AND BONE MINERAL DENSITY AMONG ELDERLY WOMEN
Grace Marshall Master’s Thesis Public Health School of Medicine
Faculty of Health Sciences University of Eastern Finland June 2017
UNIVERSITY OF EASTERN FINLAND, Faculty of Health Sciences Public health
MARSHALL, GRACE A.: Association between dietary inflammatory index and bone mineral density among elderly women
Master’s thesis, 52 pages.
Instructors: Arja Erkkilä, PhD and Masoud Isanejad, MSc June 2017
Key words: Dietary pattern, inflammation, Dietary Inflammatory Index, bone mineral density ASSOCIATION BETWEEN DIETARY INFLAMMATORY INDEX AND BONE
MINERAL DENSITY AMONG ELDERLY WOMEN
Diet and chronic inflammation are both known to be associated with bone health outcomes, such as osteoporosis. However, knowledge regarding the role that inflammatory potential of diet plays in bone health is scarce. This master’s thesis aimed to investigate the association between the inflammatory potential of diet measured using dietary inflammatory index (DII) score and bone mineral density (BMD) and bone mineral content (BMC) in a population of elderly Finnish women.
Data for this master’s thesis came from 535 participants in the Kuopio Osteoporosis Risk Factor and Prevention – Fracture Prevention Study. The women were 65−71 years old and had an average BMI of 27.4 (SD 4.2). A 3-day food record was completed at baseline and BMD was measured using dual-energy x-ray absorptiometry at baseline and after 3 years.
Data from the 3-day food records was used to calculate DII score. The calculation was done by linking the daily intake of each food parameter to the regionally representative world data base by calculating a Z-score, which was converted to a centered percentile score then multiplied by the inflammatory effect score for each food parameter. The values of all the food parameters were then summed to get the total DII score. Linear regression analysis was used to analyze the association between DII score and baseline BMD and BMC, as well as change in BMD and BMC over 3 years.
Average intakes of the anti-inflammatory nutrients fiber, vitamin A, vitamin C, vitamin D, zinc, and magnesium varied significantly between DII score quartiles. The highest intakes were associated with the lowest scores, indicating less dietary inflammation. After adjusting for confounders, the findings of the regression analysis were that higher, more inflammatory DII scores are significantly, inversely associated with baseline measures of lumbar spine BMD (β = -0.008, P = 0.041), femoral neck BMC (β = -0.032, P = 0.014), and total body BMC (β = -16.490, P = 0.024). The prospective analysis yielded no significant results.
Consuming a more pro-inflammatory diet may be associated with lower BMD and BMC among elderly Finnish women. These findings are only generalizable to elderly, female populations. Further research is needed to fully understand the association between DII score and bone health.
ABBREVIATIONS
BMC bone mineral content BMD bone mineral density
BMI body mass index
CRP C-reactive protein
CVD cardiovascular disease
DASH Dietary Approaches to Stop Hypertension DII dietary inflammatory index
DXA dual-energy x-ray absorptiometry
EU European Union
FFQ food frequency questionnaire
HEI Healthy Eating Index
hs-CRP high-sensitivity C-reactive protein
HT hormone therapy
IFN interferon
IL interleukin
MUFA monounsaturated fatty acids
NHANES National Health and Nutrition Examination Survey OSTPRE Osteoporosis Risk Factor and Prevention Study
OSTPRE-FPS Osteoporosis Risk Factor and Prevention – Fracture Prevention Study PUFA polyunsaturated fatty acids
RANK receptor activator of nuclear factor-κB
SD standard deviation
TNF tumor necrosis factor WHO World Health Organization
ACKNOWLEDGEMENTS
I would like to first thank my supervisors, Arja Erkkilä and Masoud Isanejad. I learned so much from them throughout this process and their insight played a major role in shaping this thesis into what it is today. They were incredibly patient with me as I sent them draft after draft and always had thoughtful, helpful feedback. This project would not have been possible without their in-depth knowledge of the subject and constant guidance.
I am also very grateful for the academic opportunities that have been provided to me by the University of Eastern Finland and the Institute of Public Health and Clinical Nutrition. I would like to thank all the staff and faculty who have assisted me in the pursuit of this degree, especially Annika Männikkö, who is always willing to talk and answer my questions.
I would also like to thank my classmates for their support and friendship. I always enjoyed discussing the trials and successes we all encountered while working on our theses. The people I studied alongside have taught me so much and brought me so much joy over the past two years. Lastly, thank you to my family, both in Finland and the United States, for always being there and providing unwavering support.
Contents
1. INTRODUCTION ... 6
2. LITERATURE REVIEW ... 8
2.1 Bone health ...8
2.1.1 Definition ... 8
2.1.2 Prevalence ... 9
2.1.3 Lifestyle factors and bone health ... 10
2.2 Inflammation and bone health ... 11
2.2.1 Systemic inflammation ... 11
2.2.2 Biomarkers of inflammation and bone health ... 11
2.2.3 Menopause, inflammation, and osteoporosis ... 12
2.2.4 Obesity, inflammation, and osteoporosis ... 13
2.3 Dietary Patterns ... 13
2.3.1 Dietary patterns ... 14
2.3.2 The dietary inflammatory index (DII) ... 15
2.4 Diet, inflammation, and bone health ... 16
2.4.1 Diet and bone health ... 16
2.4.2 Diet and inflammation ... 18
2.4.3 Dietary inflammatory index and bone mineral density ... 20
4. METHODOLOGY ... 22
4.1 Subjects... 22
4.2 Data collection ... 24
4.2.1 Dietary intake ... 24
4.2.2 Lifestyle questionnaire ... 24
4.2.3 Bone density measurements ... 25
4.2.4 Anthropometric measurements ... 25
4.3 Statistical analysis ... 25
4.3.1 Calculating dietary inflammatory index scores ... 25
4.3.2 Analyzing baseline characteristics ... 28
4.3.4 Regression analysis ... 29
5. RESULTS ... 31
5.1 Baseline characteristics ... 31
5.2 Association of DII score and BMD at baseline and over 3-year follow up ... 35
6. DISCUSSION ... 38
6.1 Findings in context ... 38
6.2 Strengths and limitations ... 40
6.3 Recommendations ... 42
1. INTRODUCTION
Poor bone health is a public health issue that impacts the lives of millions of individuals around the world. In particular, postmenopausal women are at higher risk of bone loss and osteoporosis than the general population. Previous research in the field of nutritional
epidemiology has investigated many aspects of the role that diet plays in bone health. This has been done by examining the role of chronic inflammation in the development of diseases like osteoporosis and by looking into the effect that consuming specific pro- and anti-
inflammatory foods has on bone health. The findings of previous studies suggest that one way diet influences bone health outcomes may be through increased levels of systemic
inflammation. This pathway is illustrated in Figure 1. Less is known about the association between the inflammatory potential of whole dietary patterns and bone health, which this thesis sought to address.
The dietary inflammatory index (DII) is a tool that has been designed to measure the inflammatory potential of an individual’s diet. This thesis aimed to further explore the associations between DII and bone health, using bone mineral density (BMD) and bone mineral content (BMC), which are important indicators of bone health. Using data from the Kuopio Osteoporosis Risk Factor and Prevention – Fracture Prevention Study (OSTPRE- FPS), this study assessed the association between DII score and measurements of bone health among a population of postmenopausal women living in and around Kuopio, Finland.
We hypothesized that postmenopausal women with a diet containing larger amounts of anti- inflammatory factors, such as fiber and vitamin C, and smaller amounts of pro-inflammatory factors, such as saturated fat, may have better bone health indicated by higher BMD and BMC levels, as compared to their counterparts with a diet containing more pro-inflammatory factors and less anti-inflammatory factors. If following a dietary pattern that decreases circulating levels of inflammatory biomarkers and DII score could decrease an individual’s risk of osteoporosis and fracture, this could be an important area for future research into the role of diet in determining bone health.
Figure 1. Potential pathway of the influence of diet on bone health (DII: dietary inflammatory index, BMD: bone mineral density)
2. LITERATURE REVIEW
2.1 Bone health
2.1.1 Definition
Osteoporosis is a major bone related disease that affects millions of people around the world, particularly in Europe, Japan, and the United States. Osteoporosis is characterized by low bone mass and this condition greatly increases an individual’s risk of low-trauma,
osteoporotic fractures and the associated morbidity (Sanchez-Riera et al. 2010). Hip and vertebral fractures are also associated with increases in premature mortality (Holroyd et al.
2008). These reasons, among others, are why osteoporosis is considered an important public health issue. The consequences of osteoporosis include these negative health outcomes, as well as significant social and economic costs. For instance, it has been estimated that osteoporotic fractures in the European Union (EU) cost $30 billion annually (Holroyd et al.
2008). The treatment of fractures and the associated rehabilitation can place a substantial burden on the families and individuals responsible for providing care. These facts highlight the necessity of studying ways to promote bone health to prevent fractures from occurring in the first place.
Osteoporosis is diagnosed in an individual by assessing their BMD and it is defined as a BMD more than 2.5 standard deviations (SD) below the mean peak bone mass (World Health
Organization 2007). The average femoral neck BMD from the young, white, female participants in the United States National Health and Nutrition Examination Survey
(NHANES) III is considered the international reference standard used with this definition of osteoporosis (Sanchez-Riera et al. 2010). The mean femoral neck BMD of non-Hispanic, white women aged 20-29 is 0.858 and one SD is 0.120 (Looker et al. 1998). BMC and bone size are the measures of bone health that are used to calculate BMD (Deng et al. 2002). Used on its own, BMC can also provide valuable information about the association between potential determinants of bone health and bone health outcomes, such as fractures and osteoporosis.
There has been discussion within the scientific and medical communities about whether BMD 2.5 SD below mean peak bone mass is the best definition for osteoporosis. Some argue that using BMD only considers demineralization of the bone and not micro-architectural changes, which can also weaken bone and lead to fractures, but which cannot be assessed using BMD (Holroyd et al. 2008). Others have stated that femoral neck BMD measured by dual-energy x- ray absorptiometry (DXA) should be the reference standard due to its extensive validation and the high gradient of fracture risk it provides (Kanis et al. 2008). While both of these methods allow for the identification of at-risk patients, it is also important to consider an individual’s history of fracture, which is included in the World Health Organization (WHO) definition of severe osteoporosis.
In addition to osteoporosis, three other categories of bone health were identified by the WHO in 1994 and have been in use ever since. The WHO report on fracture risk assessment and screening for osteoporosis defines “normal” as a BMD within 1 SD of this mean, “low bone mass” (osteopenia) as a BMD between 1 and 2.5 SD below the mean, and “severe
osteoporosis” as one or more fractures in addition to a BMD more than 2.5 SD below that mean value (World Health Organization 1994). It has been found that every decrease of one SD in an individual’s hip BMD is associated with a 2.6-fold increase in their risk of hip fracture (Marshall et al. 1996). These diagnostic definitions are a valuable tool in efforts to identify existing cases of osteoporosis and patients who are at heightened risk for osteoporotic fractures.
2.1.2 Prevalence
In 2010, there were 188,000 global deaths attributable to low BMD, but there is great variation in mean BMD values between countries in different regions (Sanchez-Riera et al.
2014). While osteoporosis effects populations around the world, more developed countries face a greater proportion of the burden. This is due to a number of factors, including the longer average life expectancy in these countries. Osteoporosis effects around 22 million women in the EU (Hernlund et al. 2013). Rates of osteoporosis in Northern Europe are higher than they are in most other regions of the world, which is one reason why it is important to study this condition in a Finnish context.
It has been estimated that in 2000 there were 9.0 million osteoporotic fractures globally, with 34.8% of those occurring in Europe (Johnell & Kanis 2006). These fractures are primarily
hip, forearm, and vertebral fractures. Around the world, 5.8 million Disability Adjusted Live Years (DALYs) were lost due to osteoporotic fractures, with 2.0 million lost in Europe alone (Johnell & Kanis 2006). A study conducted in Central Finland found that from 2005 to 2006, the crude incidence of all osteoporotic fractures among those 50 or older was 1254 per 100,000 person years (Koski et al. 2014). The incidence of hip fracture in Finland has decreased over the last few decades, but is expected to increase as the population of those over 50 years old grows. In 1997, the age-adjusted incidence of hip fracture among Finnish women was 515.7 per 100,000 persons, but in 2010 it had decreased to 382.6 (Korhonen et al.
2013). Korhonen and colleagues’ findings predict that as the Finnish population over 50 years old grows, so will the number of hip fractures in Finland, with around 10,000 annually by 2020 and 13,500 annually by 2030 (2013).
2.1.3 Lifestyle factors and bone health
There are a wide variety of modifiable and non-modifiable risk factors for low BMD, osteoporosis, and fractures. Major non-modifiable risk factors for osteoporotic fractures include age and gender, which are part of the reason why there are such great differences between rates of osteoporosis in various regions of the world (Sanchez-Riera et al. 2010).
Genetics and ethnicity are also considered non-modifiable risk factors (Cashman 2007). These factors cannot be altered through medical treatment or lifestyle interventions, but there are a number of modifiable lifestyle determinants that can impact bone health outcomes.
An individual’s lifestyle and the choices they make play an important role in determining bone health. Diet and physical activity are considered to be the major independent risk factors for osteoporosis, but smoking, alcohol consumption, and hormonal status are also important modifiable risk factors (Cashman 2007). Exercise has been associated with higher BMD and participating in regular physical activity throughout life helps develop and maintain greater bone mass (Tatsuno et al. 2013; Zhu & Prince 2015). Cigarette smoking has been well established as a risk factor for osteoporosis and is associated with lower BMD, possibly due to decreases in estrogen caused by smoking (Zhu & Prince 2015). Excessive alcohol
consumption has also been found to have a negative impact on bone health, which may be because it affects calcium absorption and may decrease rates of bone formation and increase bone resorption (Zhu & Prince 2015). Associations have also been found between some socioeconomic factors and bone health outcomes. For instance, some research shows that poverty is associated with low BMD (Amiri et al. 2008).
Clinical risk factors for osteoporotic fractures include increased age and taller height at age 25, while increased weight gain since age 25 decreases fracture risk (Cummings & Melton 2002). A history of previous fractures and maternal hip fracture is also associated with a higher risk of hip fractures (Cummings & Melton 2002). Osteoporosis, as determined by BMD, is itself a risk factor for fractures (Kanis et al. 2008). Falls are another factor that is associated with an increased risk of fractures and low BMD is the cause of a substantial proportion of falls-related deaths (Sanchez-Riera et al. 2010). Increasing physical activity and level of strength can decrease one’s likelihood of falling and in turn, decrease their fracture risk (Holroyd et al. 2008).
2.2 Inflammation and bone health
2.2.1 Systemic inflammation
Inflammation is a part of the human body’s complex response to harmful stimuli and it is a feature of a healthy immune system. It is necessary for healing and fighting infection, but should be self-limiting. In cases of chronic low-grade inflammation, it is not a normal aspect of the immune system, but rather a state of persistent immune response and accumulating tissue destruction marked by high levels of pro-inflammatory cytokines (Pawelec et al. 2014).
Systemic inflammation can be the result of obesity and other environmental factors (Lee et al.
2013). This type of inflammation is important in the context of public health, because it may play a role in the development of cardiovascular disease (CVD), some cancers, diabetes, and other chronic diseases (Elinav et al. 2013; Lee et al. 2013; Pawelec et al. 2014).
Pharmaceuticals, such as nonsteroidal anti-inflammatory drugs, can be used to control systemic inflammation, but altering dietary habits could be a much safer, long-term strategy for reducing the risk of chronic diseases.
2.2.2 Biomarkers of inflammation and bone health
Levels of inflammation can be measured or monitored using a number of biomarkers, including pro- and anti-inflammatory cytokines. The main pro-inflammatory cytokines are tumor necrosis factor (TNF)-α, interleukin (IL)-1, IL-6, and interferon (IFN)-γ and the main anti-inflammatory cytokines are IL-4 and IL-10 (Lee et al. 2013). C-reactive protein (CRP), and the more recent high-sensitivity C-reactive protein (hs-CRP), are other markers of
inflammation that are widely used to study the associations between various exposures or conditions and levels of inflammation (Nanri et al. 2007).
Pro-inflammatory cytokines are thought to play a crucial role in the process of bone remodeling. One way cytokines influence the remodeling cycle is through the receptor activator of nuclear factor-κB (RANK)/RANK ligand/osteoprotogerin system, which
regulates the processes of bone resorption and bone formation by coordinating the interaction between osteoclasts and osteoblasts (Clowes et al. 2005). More specifically, IL-1 and TNF-α have been found to increase osteoclast survival and TNF-α has also been found to inhibit osteoblast survival (Clowes et al. 2005). Additionally, some studies suggest that IL-1 and IL-6 play a part in the regulation of osteoclastogenesis (Abrahamsen et al. 2000). Bone loss results from an imbalance in osteoclast and osteoblast activity, which can be the result of elevated levels of inflammation due to aging or chronic disease. These are some of the ways that systemic inflammation may lead to increased fracture risk and osteoporosis.
Studies have looked into the associations between a variety of inflammatory biomarkers and bone health. The pro-inflammatory cytokines for which there is existing evidence linking them to increased osteoclast activity are IL-1, TNF-α, IL-6, IL-11, IL-15, and IL-17 (Mundy 2007). In a recent study, Apalset and colleagues found an inverse association between BMD and IFN-γ-mediated inflammation and some kynurenines (Apalset et al. 2014). Another study suggests that the connection between inflammation and reduced BMD could be increased levels of the cytokine transforming growth factor (TGF)-β1 (Ehnert et al. 2010). Hs-CRP is another marker of inflammation that has been found to be inversely associated with BMD (Koh et al. 2005). It has also been observed that increases in circulating levels of IL-6 can predict increased bone loss and resorption (Ding et al. 2008). These associations provide evidence that addressing high levels of inflammation may be one way to promote bone health and prevent osteoporosis.
2.2.3 Menopause, inflammation, and osteoporosis
The hormonal changes that occur during menopause and the associated treatments can also impact levels of inflammation. Estrogen plays an important role in regulating bone
metabolism and some immune function (D'Amelio 2013). During menopause, estrogen withdrawal causes the production of pro-inflammatory cytokines, which may lead to decreases in bone density due to the promotion of osteoclast activity (Mundy 2007).
Specifically, estrogen deficiency has been found to increase levels of IL-1, IL-6, IL-7, TNF-α, and IFN-γ (Clowes et al. 2005). Hormone therapy (HT) is used to control severe menopausal symptoms and has been found to prevent osteoporosis among postmenopausal women, but is only recommended for use for women at high risk of osteoporosis because it can increase an individual’s risk of CVD and dementia (Marjoribanks et al. 2017). Also, the use of oral HT has been found to be associated with increases in levels of CRP among postmenopausal women, which signals heightened levels of systemic inflammation (Lakoski & Herrington 2005). While HT is recommended to prevent osteoporosis among some women, the fact remains that both estrogen withdrawal and HT have been found to be associated with increases in chronic inflammation. Further research is needed to fully understand the relationships between menopause, hormone therapy, inflammation, and bone health.
2.2.4 Obesity, inflammation, and osteoporosis
There is a link between obesity and chronic inflammation (Pradhan 2007). This inflammation may contribute to the development of diabetes and CVD, but it also influences bone health.
Obesity is related to bone health because adipose tissue generates pro- and anti-inflammatory cytokines, including IL-4, IL-6, IFN-γ, and TNF-α (Lee et al. 2013; Pradhan 2007). A number of mechanisms have been suggested to explain the relationship between obesity and
osteoporosis, with one being that the pro-inflammatory cytokines released by the adipose tissue affect osteoclast and osteoblast activity, which impacts the bone remodeling process and leads to lower BMD (Kawai et al. 2012). There is some controversy about the nature of this relationship and the exact pathways involved. Some studies state that obesity protects against osteoporosis (Reid 2010). Lower weight has been associated with osteoporosis and higher weight has been associated with less bone turnover, even while higher CRP levels have been found to be associated with more abdominal fat (Gunn et al. 2014). While more research is needed in this area, it may be important to consider body mass index (BMI) when studying the association between inflammatory diets and bone health, because of the association between obesity and systemic inflammation.
2.3 Dietary Patterns
2.3.1 Dietary patterns
Nutritional epidemiology research typically examines how individual nutrients and foods impact health. However, much can be learned from studying whole dietary patterns in relation to health outcomes, rather than separate foods and nutrients. According to the United States Department of Agriculture, dietary patterns are defined as “the quantities, proportions,
variety, or combination of different foods, drinks, and nutrients (when available) in diets, and the frequency with which they are habitually consumed” (United States Department of Agriculture 2014).
Studies of dietary patterns can be conducted using either a-priori or a-posteriori methods. Α- priori analysis utilizes pre-determined dietary patterns based on nutrition recommendations or established diets and related indexes (Panagiotakos 2008). Some examples of commonly studied, pre-established dietary patterns are the Mediterranean diet and the Baltic Sea Diet.
The Mediterranean diet is characterized by a high intake of fruits, vegetables, whole grains, legumes, fish, and healthy fats such as olive oil (Widmer et al. 2015). The Baltic Sea Diet is characterized by high consumption of the fruits, vegetables, grains, and fish that are
commonly consumed in the Nordic region (Kanerva et al. 2014). Alternatively, a-posteriori methods identify unique dietary patterns by analyzing similarities in the food consumption patterns of the participants (Panagiotakos 2008). The Western diet, an example of a dietary pattern commonly identified using a-posteriori methods, is high in saturated fats, sodium, and refined carbohydrates, mostly from highly processed sources (Myles 2014).
Nutrients are not consumed in isolation, but rather as part of a whole diet. For this reason, research into dietary patterns can often represent how nutrients and foods are consumed and the impact they have on individuals’ health better than studies of singular nutrients (Fung et al. 2001). When entire dietary patterns are studied, combinations of foods and the interactions between them are taken into consideration (Wirfält et al. 2013). The proportions of different nutrients consumed in various dietary patterns can also affect health outcomes, such as osteoporosis (Cashman 2007). An additional benefit of studying dietary patterns rather than individual nutrients, is that the results can be easier to translate to practical public health recommendations. However, when developing public health programs and recommendations it is important to consider results from studies of individual nutrients alongside the findings from dietary patterns research (United States Department of Agriculture 2014). Studies of
healthy dietary patterns can help the public understand how to make better overall dietary choices.
2.3.2 The dietary inflammatory index (DII)
The DII was developed by researchers at the University of South Carolina’s Cancer Prevention and Control Program. It is a tool for measuring the inflammatory potential of one’s diet, whether anti-inflammatory or pro-inflammatory (Shivappa et al. 2014a; Shivappa et al. 2014b). There are 45 dietary components that make up the DII, and each can either increase or decrease an individual’s overall score. The dietary components include macronutrients, micronutrients, flavonoids, and some herbs, spices, and aromatics. 1,943 studies on the impact of certain foods and nutrients on biomarkers of inflammation were reviewed and that data was used to determine the effect of each dietary component on one’s DII score (Steck et al. 2014). The 6 biomarkers that were included in the search parameters were IL-1β, IL-4, IL-6, IL-10, TNF-α and CRP. If a food was found to significantly raise levels of IL-1β, IL-6, TNF-α and CRP or decrease levels of IL-4 or IL-10 it was assigned a
“+1.” If it lowered the levels of the pro-inflammatory biomarkers or increased the levels of the anti-inflammatory biomarkers it was given a “-1.” If a food did not have any impact on the biomarkers of inflammation it was assigned a “0” value (Steck et al. 2014). The studies were weighted on a scale of 1 to 10 based on quality. For example, the results of human studies were given more value than those from animal studies (Shivappa et al. 2014a). All of this information can be used to determine the DII scores of a population by applying the effect score, global mean, and SD for each dietary component to dietary data from food frequency questionnaires (FFQs), 24-hour recalls, or 3-day food records.
The DII has been validated in a number of settings. Construct validation of the DII using data from the 24-hour and 7-day dietary recalls that were a part of the Seasonal Variation of Blood Cholesterol Study found that the index could predict odds of elevated hs-CRP (Shivappa et al.
2014b). In this study, higher DII scores calculated using the 24-hour recall and were associated with increased odds of elevated hs-CRP (odds ratio = 1.08) and similar results were found when calculating the DII scores using dietary data from the 7-day recall (odds ratio = 1.10) (Shivappa et al. 2014b). Additionally, a construct validation of the DII was conducted using FFQ data from postmenopausal women enrolled in the Women’s Health Initiative Observational Study and found that scores were significantly associated with IL-6, TNF-α receptor 2, hs-CRP, and a combined inflammatory biomarker score (Tabung et al.
2015b). The DII has also been validated using data from NHANES 1999-2002, where higher DII scores were found to be associated with higher levels of CRP (Shivappa et al. 2017). A positive association between DII score and CRP was also observed in a population of police officers from Buffalo, NY (Wirth et al. 2014). On the other hand, there was no significant long-term, prospective association found between DII score and long-term CRP when studied over 12 years in the French Supplémentation en Vitamines et Minéraux Antioxydants cohort (Julia et al. 2017). Previous studies have shown significant associations between DII scores and chronic diseases, including some cancers, CVD, and asthma (Shivappa et al. 2016;
Tabung et al. 2015a; Wirth et al. 2016). However, very few studies have investigated the association between DII and bone health.
A study published in 2016 found that more pro-inflammatory DII scores were associated with less healthy scores on the Healthy Eating Index 2010, Alternative Healthy Eating Index, and Dietary Approaches to Stop Hypertension (DASH) Index (Wirth et al. 2016). Other studies have also identified that anti-inflammatory DII scores are often associated with dietary patterns that are considered “healthy,” like Mediterranean, macrobiotic, and vegetarian diets (Shivappa et al. 2014a; Steck et al. 2014). On the other hand, Western or fast food dietary patterns are usually found to be more inflammatory (Tabung et al. 2016).
2.4 Diet, inflammation, and bone health
2.4.1 Diet and bone health
Healthy bones are constantly being remodeled and proper nutrition is necessary to support and manage this process (Levis & Lagari 2012). Throughout an individual’s life, bone is being broken down by osteoclasts and rebuilt by osteoblasts, which synthesize an osteoid matrix that is then calcified (Lanham-New 2008). This process relies on certain levels of calcium and vitamin D, which are crucial for proper bone health according the WHO, the European Commission, and the United States Surgeon General (Cashman 2007). Along with calcium and vitamin D, the nutrients most commonly associated with skeletal health, there is a wide array of macronutrients and other micronutrients that also determine the health of an individual’s bones.
Vitamin D and calcium have been studied extensively in relation to bone health and have been found to play important roles in the prevention of fractures and osteoporosis (Levis &
Lagari 2012). Sufficient intake and absorption of calcium through-out the lifecycle is essential for proper bone development and maintenance, and calcium supplementation has been found to reduce bone loss and fracture risk in older populations (Cashman 2007; Lanham-New 2008). While calcium intake recommendations vary around the world, there is a consensus that insufficient calcium intake leads to more bone turnover and bone loss (Levis & Lagari 2012). The most recent Nordic Nutrition Recommendations, which were released in 2012 and developed for use in Finland and other Nordic countries, recommend a daily intake of 800 mg per day for adult women and men (Nordic Council of Ministers 2014). Vitamin D is necessary for bone matrix formation. It is necessary for the proper mineralization of children’s skeletons as they are growing, the prevention of osteomalacia among adults, and to ward off the
development of osteoporosis in elderly populations (Cashman 2007; Lanham-New 2008).
Additionally, the transcellular movement of calcium is dependent on vitamin D, and is an important pathway for the absorption of calcium when there is a limited amount of calcium available (Cashman 2007). The Nordic Nutrition Recommendations set the recommended intake for vitamin D at 10 µg per day for men and women under 75 and 20 µg per day for those aged 75 and older (Nordic Council of Ministers 2014). Calcium and vitamin D deficiency becomes more of a concern as individuals age, because the body is not able to process these nutrients and regulate their levels as efficiently as it does earlier in life (Clowes et al. 2005). The available evidence shows that any dietary recommendations for promoting bone health and preventing osteoporosis should include calcium and vitamin D.
In addition to those two main nutrients, other dietary factors that are beneficial for bone health include copper, zinc, silicon, fluoride, magnesium, phosphorus, potassium, vitamin A, vitamin C, vitamin E, vitamin K, B vitamins, n-3 fatty acids, and protein (Cashman 2007; Levis &
Lagari 2012; Sahni et al. 2015). A recent meta-analysis concluded that higher protein intake is associated with higher lumbar spine BMD (Shams-White et al. 2017). Vitamin K is
considered to be an important nutrient for bone health, because of its role in the carboxylation of osteocalcin and other bone-related proteins (Lanham-New 2008). Excess alcohol, caffeine, sodium and n-6 fatty acids are dietary factors that could have a negative impact on bone health (Cashman 2007). However, it is important to note that many studies suggest that moderate alcohol intake is beneficial for bone health (Sahni et al. 2015; Sommer et al. 2013).
Recently, there has also been research into and discussion of the impact of certain bioactive
compounds and dietary phytoestrogens on bone health, but more research is needed in this area before concrete recommendations can be made (Cashman 2007).
Studies have looked into the specific effects of the consumption of these nutrients on BMD and what types of foods these nutrient intake patterns may be related to. A study set in New Zealand found that intake of protein, vitamin B12, zinc, and potassium were positively
correlated with BMD (Gunn et al. 2014). A nutrient pattern high in folate, total fiber, vitamin B6, potassium, vitamin A, vitamin C, β-carotene, vitamin K, magnesium, copper, and
manganese, had a significant association with BMD. Women consuming a diet high in these nutrients, likely due to high consumption of fruits and vegetables, were found to be less likely to have low lumbar spine BMD (Karamati et al. 2014). These findings show that consuming a healthy diet that is rich in a variety of vitamins and minerals can promote bone health and help prevent osteoporosis.
Around the world, studies have been conducted that look into the associations between different dietary patterns and BMD. Among Australians aged 50 or older, those consuming a prudent dietary pattern high in fruits, vegetables, and dairy products were less likely to have low BMD than those consuming a Western diet (Melaku et al. 2016). A study of Scottish women in their 50s found that diets high in processed and snack foods were associated with low BMD (Hardcastle et al. 2011). In Iran it was observed that menopausal women who consume diets high in saturated fats or lacking in many vitamins and minerals are more likely to have low BMD (Karamati et al. 2012). A diet high in fish and olive oil intake, but low in red meat consumption was associated with higher BMD in a population of Greek women (Kontogianni et al. 2009). In a Spanish setting, the Mediterranean diet has been found to be associated with increased BMD (Rivas et al. 2013). Findings from these studies of dietary patterns are supported by a 2011 systematic review that identified positive associations between fruit and vegetable consumption and forearm, lumbar spine, and total hip BMD (Hamidi et al. 2011). Previous findings follow a pattern of healthy, or prudent, diets high in fruit and vegetable intake being associated with higher BMD and diets composed of high amounts of processed foods and saturated fats being associated with low BMD.
2.4.2 Diet and inflammation
Many dietary components have an impact on individuals’ levels of subclinical inflammation.
According to the DII’s overall inflammatory effect scores, total energy, carbohydrate, protein,
total fat, saturated fat, trans fat, cholesterol, iron, and vitamin B12 are pro-inflammatory dietary components (Shivappa et al. 2014a). The food parameters that are anti-inflammatory include alcohol, caffeine, fiber, zinc, magnesium, vitamins A, B6, C, D, E, n-3 fatty acid and n-6 fatty acid (Shivappa et al. 2014b). While most of these anti-inflammatory factors support bone health, excess alcohol, caffeine, and n-6 fatty acids could have a negative impact on bone health (Cashman 2007). All these food parameters are taken into consideration in the calculation of the DII score, which aims to represent how an individual’s diet impacts their levels of systemic inflammation.
The inflammatory effect scores in the DII are supported by findings from observational studies and review articles that have examined the associations between various nutrients and foods and levels of inflammation. A study using NHANES data from the United States found an inverse association between levels of fiber, polyunsaturated fatty acids (PUFA), vitamins A, B6, C, E, K, folate, magnesium, iron, copper, and potassium consumption and hs-CRP, but a positive association between sugar intake and hs-CRP levels (Mazidi et al. 2017). Giugliano and colleagues suggest that reducing consumption of saturated fat, trans fat, and refined grains, while ensuring one consumes enough n-3 fatty acids, fruits, vegetables, nuts, and whole grains, can decrease chronic inflammation (Giugliano et al. 2006). A study of older Scottish men and women observed significant inverse associations between CRP levels and apple intake, total fruit intake, and total combined fruit and vegetable intake (Corley et al.
2015). Additionally, moderate alcohol consumption, up to 40 grams per day, has been found to be associated with lower levels of inflammation when compared to non-drinkers and heavy drinkers in both observational and intervention studies (Giugliano et al. 2006). According to Lee and colleagues, consuming flavonoid-rich foods and choosing carbohydrates with a low glycemic index also helps reduce and maintain low levels of inflammation (Lee et al. 2013).
In general, it is understood that over-nutrition leads to immunoactivation and increases susceptibility to inflammatory diseases, which is why it is important to maintain a balanced diet (Lee et al. 2013). Emphasizing the consumption of these foods and nutrients that may lower levels of inflammation could possibly help reduce osteoporosis risk because of the associations between high levels of inflammation and reduced BMD.
Many healthy dietary patterns are composed of the foods and nutrients that are associated with lower levels of inflammation and some specific dietary patterns have been found to be associated with decreases in serum CRP and hs-CRP. Observational studies have found
adherence to prudent and Mediterranean dietary patterns to be associated with reductions in CRP levels, while following a Western dietary pattern is associated with increased levels of CRP (Barbaresko et al. 2013; Giugliano et al. 2006; Neale et al. 2016). In a meta-analysis of randomized trials, adhering to the DASH diet was found to decrease hs-CRP levels (Soltani et al. 2017). Another dietary pattern that has been found to be associated with lower circulating hs-CRP is the Baltic Sea diet (Kanerva et al. 2014). Following national nutritional
recommendations, such as the Dietary Guidelines for Americans or those developed by the French National Nutrition and Health Program, may also reduce systemic inflammation (Ahluwalia et al. 2013). Recommending these healthy dietary patterns to a population may help individuals consume less inflammatory diets, which, in turn, could promote bone health.
2.4.3 Dietary inflammatory index and bone mineral density
Since the introduction of the DII, a number of studies have looked into the association between DII score and bone health. One study among postmenopausal Iranian women found that higher DII scores were associated with lower lumbar spine BMD (Shivappa et al. 2016).
This cross-sectional study of 160 postmenopausal women used dietary data from an FFQ to calculate DII score and then analyze the association between DII score and BMD at the femoral neck and lumbar spine. The results were only significant between DII score and lumbar spine BMD, although the direction of the association with femoral neck BMD was similar (Shivappa et al. 2016). An analysis using data from the Women’s Health Initiative found that, among postmenopausal women, consuming a less inflammatory diet was
associated with less loss of BMD (Orchard et al. 2016). This was a prospective, observational study of over 160,000 participants with an average age of 63 years. In addition to the findings regarding BMD loss, higher baseline DII scores were found to be associated with greater risk of hip fracture among the younger, white women (Orchard et al. 2016). While it did not specifically look at BMD, another study examined the association between DII score and bone health. This case-control study of 1,050 pairs of Chinese men and women aged 52 to 83 found pro-inflammatory diets to be associated with increased risk of hip fracture (Zhang et al.
2017). There is still a need for more investigation into the relationship between DII score and BMD in postmenopausal women from a variety of backgrounds, but the results of these studies suggest that a more pro-inflammatory diet may have a negative impact on bone health outcomes.
3. AIM OF THE STUDY
The aim of this master’s thesis was to evaluate the association between DII score and measures of bone health, specifically BMD and BMC, in a population of postmenopausal women from Kuopio, Finland, using data from OSTPRE-FPS.
4. METHODOLOGY
4.1 Subjects
The data used in this study were from OSTPRE-FPS, a fracture prevention trial based in Kuopio, Finland. OSTPRE-FPS was a 3-year intervention study, which began in 2002 and aimed to explore the effect of calcium (1000 mg/day) and vitamin D (800 IU/day)
supplementation on falls and fractures in this population of post-menopausal women
(Kärkkäinen et al. 2010). The OSTPRE-FPS participants were a group of 3,432 women who had been a part of the larger, population-based OSTPRE study that was launched in 1989.
OSTPRE-FPS participants were required to be 65-72 years old, be living in the Kuopio region, be willing to participate, and not have previously participated in the OSTPRE bone densitometry sample.
Participants randomized to the intervention study group were prescribed 800 IU of
cholecalciferol and 1,000 mg of calcium carbonate to take daily for 3 years (Kärkkäinen et al.
2010). Those randomized to the control group received no dietary supplements and all of the participants were asked not to change their dietary habits (Kärkkäinen et al. 2010). Out of the 3,432 women who volunteered to participate in OSTPRE-FPS, 750 were randomly selected to participate in a subsample that received detailed examinations including bone density
measurements, body composition measurements, and the completion of a 3-day food record.
554 of the subjects returned valid food records and had valid body composition measurements from baseline and after 3 years. However, data on alcohol consumption from the lifestyle questionnaire was missing for some participants so the final analysis included 535 women.
Figure 2 presents a description of participant selection and study group allocation.
All clinical measurements were conducted in the Kuopio Musculoskeletal Research Unit of the Clinical Research Center of the University of Kuopio.All participants provided written informed consent at baseline. The OSTPRE-FPS study was approved by the Kuopio University Hospital Committee on Research Ethics in 2001 and was registered at Clinictrials.gov on January 2, 2008 (NCT00592917).
Figure 2. Participant flow diagram of the OSTPRE cohort and OSTPRE-FPS 3-year trial (Kärkkäinen et al. 2010)
Eligible OSTPRE cohort subjects
(n = 5,407)
Randomized (n = 3,432)
Ca + vit D group (n = 1,718)
Control group (n = 1,714)
Randomly selected subsample (n = 750) In each group (n = 375)
Allocated to ca + vit D group (n = 290)
Allocated to control group (n = 313)
Valid food record and lifestyle questionnaire and included in the analysis (n = 535)
From ca + vit D group (n = 261) From control group (n = 274)
Excluded (n = 1,975)
Did not meet inclusion criteria (n = 312) Declined to participate (n = 962)
Did not return enquiry (n = 701)
Withdrew after randomization (n = 237)
From intervention group (n = 132) Withdrew consent (n = 132) From control group (n = 105) Withdrew consent (n = 83) Died before start (n = 15) No contact information (n = 7)
Lost to follow-up (n = 3) Died (n = 3)
Discontinued intervention (n = 40)
Adverse effects (n = 17) Other reasons (n = 22) No reason (n = 1)
Lost to follow-up (n = 7) Died (n = 1)
Not willing to continue (n = 1)
Not able to continue (n = 1)
Contact lost (n = 4)
4.2 Data collection
4.2.1 Dietary intake
The dietary data was from a 3-day food record, which was completed by OSTPRE-FPS participants at baseline in 2002. A physical questionnaire was mailed to the participants beforehand and they returned the completed forms on the visiting day. Participants were instructed to record their food consumption for 3 consecutive days, including exactly one weekend day. Dietary data was only collected at baseline, there was no additional
measurement over the 3 years of the study. If a participant’s submission was not clear, a nutritionist called the participant to clarify any issues (Erkkilä et al. 2012). Once the data was collected, nutritional intake was calculated using Nutrica version 2.5, a software program developed by the Kela, the Social Insurance Institution of Finland.
4.2.2 Lifestyle questionnaire
The lifestyle questionnaire was distributed to participants through the mail and was self- administered. It collected data from subjects on topics such as age, alcohol consumption, smoking status, supplement use, hormone therapy, menopause, medications, disease history, and physical activity. Alcohol consumption (g/day) was calculated by multiplying self-
reported portions per week, from the lifestyle questionnaire, by 12. This calculation was done because 12 g is the estimated amount of alcohol in a bottle of beer (330 ml), a glass of wine (120 ml), or a shot of hard liquor (40 ml) (Isanejad et al. 2017b). Duration of HT (years) was the reported length of time estrogen had been used to treat menopausal symptoms. Time since menopause (years) was defined as 12 months of amenorrhea and calculated based on the reported start of amenorrhea (Sandini et al. 2008). Self-reported calcium and vitamin D supplementation was defined as the current daily or almost daily use of calcium or vitamin D as a supplement. The list of diseases possibly affecting BMD included hyperthyroidism, disease of the parathyroid gland, chronic liver diseases, chronic intestinal disease, celiac disease, ventricle operation, chronic nephropathy arthritis, osteoporosis, and lactose
intolerance (Isanejad et al. 2017a). The medications that may have affected bone metabolism included corticosteroids, diuretics, cytotoxic drugs, and calcitonin (Kärkkäinen et al. 2010).
Physical activity in average hours per week was based on the amount and types of physical activity the participants engaged in in both winter and summer (Rikkonen et al. 2010).
4.2.3 Bone density measurements
BMC (g) of the lumbar spine (L2-L4), femoral neck, and total body was measured by trained nurses at baseline and over 3 years using DXA (Lunar Prodigy, Wisconsin, USA). BMD (g/cm2) was calculated by dividing BMC (g) by bone area (cm2) and absolute changes in BMD and BMC were calculated using the values from baseline and 3 years. DXA is a
standard technique and it is widely used to measure BMD (Miyabara et al. 2012). The quality of the measurements was double checked and if any measurement errors were found those readings were excluded from the statistical analysis. The long-term reproducibility of the DXA measurements for BMD during the study period, as determined by daily phantom measurements, was 0.3% (Kärkkäinen et al. 2010).
4.2.4 Anthropometric measurements
Both the height and weight of participants were measured in light indoor clothing without shoes. Measurements were taken at baseline using a calibrated digital scale and a calibrated wall meter, for weight and height respectively. Height (cm) and weight (kg) were measured to calculate the BMI (kg/m2) of the participants.
4.3 Statistical analysis
4.3.1 Calculating dietary inflammatory index scores
The data from the 3-day food record included 26 of the 45 total food parameters of the original DII (Shivappa et al. 2014b). These were alcohol, β-carotene, caffeine, carbohydrate, cholesterol, energy, total fat, fiber, folic acid, iron, magnesium, monounsaturated fatty acids (MUFA), niacin, n-3 fatty acids, n-6 fatty acids, protein, PUFA, riboflavin, saturated fat, thiamin, vitamin A, vitamin C, vitamin D, vitamin E, zinc, and tea. Table 1 lists the food parameters which were included, calculated then included, and not included in this analysis.
Table 1. Dietary inflammatory index food parameters included in this study Food parameters included
(n=23)
Food parameters calculated, then included (n=3)
Food parameters not available to be included (n=19)
Alcohol (g) Caffeine (g) Vitamin B12 (µg)
β-carotene (µg) n-3 fatty acids (g) Vitamin B6 (mg)
Carbohydrate (g) n-6 fatty acids (g) Eugenol (mg)
Cholesterol (mg) Garlic (g)
Energy (kcal) Ginger (g)
Total fat (g) Onion (g)
Fiber (g) Saffron (g)
Folic acid (µg) Selenium (µg)
Iron (mg) Trans fat (g)
Magnesium (mg) Turmeric (mg)
MUFA (g) Flavan-3-ol (mg)
Niacin (g) Flavones (mg)
Protein (g) Flavanols (mg)
PUFA (g) Flavonones (mg)
Riboflavin (mg) Anthocyanidins (mg)
Saturated fat (g) Isoflavones (mg)
Thiamin (mg) Pepper (g)
Vitamin A (RE) Thyme/oregano (mg)
Vitamin C (mg) Rosemary (mg)
Vitamin D (µg) Vitamin E (mg) Zinc (mg)
Green/black tea (g)
MUFA: monounsaturated fatty acids, PUFA: polyunsaturated fatty acids, RE: retinol equivalents
The data from the food records was provided in grams per day, except for caffeine, which was converted based on reported intake of coffee, tea, chocolate drinks, and chocolate in grams per day. This conversion was done by multiplying the amount of food or drink consumed (g/day) by the caffeine (g) in 100 g of the given food. The caffeine content (g) was calculated using data from a report by the European Food Safety Authority on the safety of caffeine, which reported the amounts of caffeine in these foods and drinks (mg/100g) (EFSA Panel on Dietetic Products, Nutrition & Allergies 2015). An example of this calculation is
demonstrated in Figure 3. N-3 and n-6 fatty acid consumption (g/day) was also calculated using data on the individual fatty acids. Intake (g/day) of linolenic acid, eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA) were summed to calculate n-3 fatty acid intake (g/day) and consumption levels of linoleic acid and arachidonic acid were summed to calculate n-6 fatty acid intake (g/day).
Calculating the DII required 4 steps of calculations for each food parameter being used and a fifth calculation to sum the DII scores of all 26 food parameters for each participant. To link the dietary data from the OSTPRE-FPS sample to the regionally representative world database, a Z-score was calculated using the global mean and global SD for each food parameter in the DII. These means and SDs are provided, along with the inflammatory effect scores, in a table in the paper by Shivappa and colleagues that details the design and use of the DII (2014a). The Z-score was then converted to a percentile score to minimize the effect of right skewing. Next, the percentile score was transformed into a centered percentile score that is centered on 0 and ranges from -1 to +1 by multiplying each percentile score by 2 and then subtracting 1. When this calculation is completed a score of -1 is the most anti-
inflammatory and a score of +1 is the most pro-inflammatory. Next, the centered percentile scores were multiplied by the inflammatory effect score for the given food parameter. Once each food parameter-specific DII score was obtained, all those values were summed to get the overall DII score for each participant. These steps are illustrated in Figure 4.
1. Convert caffeine amounts from mg/100g to g/100g
2. Calculate caffeine (g/day) for each food item using consumption (g/day) and caffeine content (g/100g)
3. Sum the specific caffeine consumption amounts to get total caffeine intake (g/day) Figure 3. Methods for calculating total caffeine intake
4.3.2 Analyzing baseline characteristics
After calculating the DII scores, the continuous variable was categorized into quartiles. The first quartile included DII scores less than -1.804, the second quartile included values greater than or equal to -1.803 but less than -0.332, the third quartile included values greater than or equal to -0.331 but less than 1.150, and the fourth quartile included values greater than or equal to 1.151. For analyzing the baseline characteristics, dietary intakes were reported in daily consumption level and either percent of total energy or amount per 1000 kcals. Intake per 1000 kcals was calculated by multiplying the daily consumption of the given food parameter by 1000 divided by total energy intake (kcal).
The baseline characteristics age (years), BMI (kg/m2), weight (kg), height (cm), alcohol consumption (g/day), and physical activity (hrs/week) were examined across the quartiles of DII score using one-way ANOVA for the continuous variables explained with means and SDs. Kruskal-Wallis, the corresponding non-parametric test, was used in instances when the variable was not normally distributed. Chi-squared analysis was used to analyze the three categorical variables: current smoking status (yes/no), use of hormone therapy (yes/no), and disease or medication that may have impacted bone health (yes/no). ANOVA was also
1. Calculate Z-score to link the dietary data to the “regionally representative world database”
2. Convert to a percentile score to minimize the effect of right skewing Rank Cases
3. Get a centered percentile value
4. Multiply the centered percentile score by the food parameter’s effect score
Figure 4. Methods for calculating the DII score variable
conducted for 12 food parameters to test for differences in mean intake among the DII quartiles.
ANOVA was then used to analyze baseline BMD and BMC values, as well as change in BMD and BMC over 3 years, across DII score quartiles. In this case, first an unadjusted ANOVA or Kruskal-Wallis test was used depending on the distribution of the variable, then an adjusted ANOVA was used with every variable to account for the effect of confounders.
Both the unadjusted and the adjusted P values were reported. The ANOVA was adjusted for age (years), BMI (kg/m2), intervention group, alcohol consumption (g/day), and duration of HT (years). Initially total energy intake (kcal/day), current smoking status (yes/no), self- reported calcium supplementation (yes/no), self-reported vitamin D supplementation (yes/no), time since menopause (years), disease or medication that may have affected BMD (yes/no), and physical activity (hours/week) were also considered for adjustment, but the final 5 variables that were selected were included in both the ANOVA and regression analyses because they resulted in the most complete and efficient model based on the adjusted r2 and F values. The corresponding baseline values of BMD and BMC were also included as
covariates in the analyses to account for changes in these measurements over 3 years.
4.3.4 Regression analysis
The association between DII score and BMD was evaluated using multiple linear regression analysis. The potential confounders that were considered in this analysis were age (years), BMI (kg/m2), total energy intake (kcal/day), intervention group, alcohol consumption (g/day), current smoking status (yes/no), length of HT (years), self-reported calcium supplementation (yes/no), self-reported vitamin D supplementation (yes/no), time since menopause (years), disease or medication that may have affected BMD (yes/no), and physical activity
(hours/week). Baseline BMD (g/cm2) and BMC (g) values were also considered as potential confounders for the prospective analysis. The factors that were tested as covariates were selected beforehand because they are known to possibly affect BMD and were in line with previously published analyses of OSTPRE-FPS data (Isanejad et al. 2017b; Järvinen et al.
2012).
Model 1 was left unadjusted, while model 2 was adjusted for age (years), BMI (kg/m2), intervention group, alcohol intake (g/day), and length of HT use (years) and baseline BMD (g/cm2) and BMC (g) in the prospective analysis. The variables listed were selected by
examining the adjusted r2 and F values. At first all potential confounders were included in model 2, but in the end total energy intake (kcal/day), current smoking status (yes/no), self- reported calcium supplementation (yes/no), self-reported vitamin D supplementation (yes/no), time since menopause (years), disease or medication that may have affected BMD (yes/no), and physical activity (hours/week) were excluded because they did not improve the
explanatory value while maintaining or increasing the efficiency of the model when added in one at a time. Additionally, to assess the effect of the intervention on the results of the prospective analysis, the data was stratified by study group and then analyzed using model 2 without adjustment for intervention group.
All statistical analyses were conducted using SPSS version 23. Results were significant if the P value was <0.05.
5. RESULTS
5.1 Baseline characteristics
DII score ranged from -5.5 to 4.8 and the mean was -0.3 (SD 2.1). The results of the ANOVA of baseline characteristics are reported in Table 2. There were not significant differences in mean values of BMI (kg/m2), weight (kg), alcohol consumption (g/day), current smoking status (yes/no), current HT use (yes/no), physical activity (hrs/week), and disease or medication affecting BMD (yes/no) in the DII score quartiles. The mean age of the
participants at baseline was 67.9 (SD 1.9) years old and age was slightly higher in the higher DII score quartiles (P = 0.024). Mean height was 160.5 (SD 5.0) cm in the first quartile and 157.4 (SD 5.8) cm in the fourth quartile (P <0.001).
Table 2. Selected demographic baseline characteristics of the participants by dietary inflammatory index score quartiles
Dietary inflammatory index score quartiles Q1
(≤-1.804)
Q2 (-1.803-
-0.332)
Q3 (-0.331-
1.150)
Q4 (≥1.151)
n 133 n 134 n 134 n 134
Mean SD Mean SD Mean SD Mean SD P
Age (years) 67.6 1.9 67.7 1.9 67.9 1.8 68.2 1.9 0.024
BMI (kg/m2) 27.0 4.4 27.2 3.7 27.6 4.5 28.1 4.3 0.092 Weight (kg) 72.2 13.0 71.2 10.5 72.2 11.7 73.3 12.5 0.569 Height (cm) 160.5 5.0 158.3 5.0 158.6 5.0 157.4 5.8 <0.001 Alcohol (g/d) 9.4 14.6 12.4 19.8 9.8 16.1 8.4 15.8 0.230 Current smoker (%) 3 (2.5) 7 (5.6) 5 (4.4) 11 (9.1) 0.232 Current HT use (%) 35 (26.3) 27 (20.1) 31 (23.1) 29 (21.6) 0.664 Physical activity
(hrs/wk)
12.1 15.5 13.4 19.9 12.9 14.3 8.9 11.1 0.449 Disease or medication
affecting bone (%)
42 (31.8) 51 (38.1) 53 (39.6) 49 (36.6) 0.588 Q: quartile, BMI: body mass index, HT: hormone therapy. Categorical data is presented as n (%).
ANOVA, Kruskal-Wallis, and chi-squared tests were used to evaluate the differences in mean values of baseline characteristics between the DII score quartiles.
Table 3 presents how mean dietary intakes varied across DII score quartiles. Mean total energy intake was 1569 (SD 373) kcal and there were significant differences found in the mean total energy intakes in the DII score quartiles, with the highest intake in the first quartile
and the lowest intake in the fourth quartile (P <0.001). There also were significant differences across the quartiles of daily intake for the rest of the food parameters reported. Significant differences were found for all the nutrients when they were reported proportionally, as percent of total energy or as intake per 1000 kcal, except for carbohydrate and fat. Both protein and PUFA intake as a percent of total energy were smaller in the higher quartiles and saturated fat intake as a percent of total energy was greater in the higher quartiles, with the highest mean intake in the fourth quartile (13.2%). Mean intakes of fiber (g/1000 kcal), vitamin A
(RE/1000 kcal), vitamin C (mg/1000 kcal), vitamin D (µg/1000 kcal), zinc (mg/1000 kcal), and magnesium (mg/1000 kcal) were highest in the first quartile and lowest in the fourth quartile.
Table 3. Selected baseline dietary intakes by dietary inflammatory index score quartiles Dietary inflammatory index score quartiles
Q1 (≤-1.804)
Q2 (-1.803-
-0.332)
Q3 (-0.331-
1.150)
Q4 (≥1.151)
n 133 n 134 n 134 n 134
Mean SD Mean SD Mean SD Mean SD P
Total energy (kcal/day)
1853 328 1670 298 1471 282 1286 317 <0.001 Protein
g/day 84.2 16.4 73.8 13.4 62.1 12.6 53.5 13.0 <0.001
% of total energy 18.4 2.4 18.1 3.4 17.3 3.3 17.0 3.2 <0.001 Carbohydrate
g/day 228.4 42.9 203.3 43.4 184.5 36.0 157.7 41.6 <0.001
% of total energy 49.1 5.3 48.5 5.7 50.0 6.0 48.8 6.1 0.179 Fat
g/day 62.9 18.6 57.8 16.8 50.4 16.3 45.3 16.2 <0.001
% of total energy 30.6 5.4 31.0 5.7 30.7 5.7 31.8 5.5 0.273 Saturated Fat
g/day 23.2 9.1 22.3 8.0 20.3 7.9 19.2 8.1 <0.001
% of total energy 11.2 3.0 11.8 2.9 12.3 2.9 13.2 3.2 <0.001 PUFA
g/day 11.4 3.4 9.6 3.4 7.9 2.5 6.4 2.2 <0.001
% of total energy 5.6 1.3 5.2 1.6 4.9 1.2 4.6 1.3 <0.001 Fiber
g/day 29.4 4.9 23.4 4.7 20.6 4.5 15.8 4.0 <0.001
g/1000 kcal 16.2 3.4 14.3 3.3 14.4 3.9 12.8 3.8 <0.001 Vitamin A
RE/day 1508.1 2039.1 1049.8 1184.7 883.5 855.6 601.8 626.2 <0.001 RE/1000 kcal 833.0 1116.5 644.4 730.0 628.8 641.7 476.8 442.9 <0.001 Vitamin C
mg/day 135.9 73.6 105.1 48.2 92.2 49.5 63.9 45.0 <0.001 mg/1000 kcal 75.4 41.4 65.2 33.2 65.3 37.8 50.9 32.7 <0.001 Vitamin D
µg/day 10.3 5.7 8.4 5.2 6.7 3.4 5.4 3.6 <0.001
µg/1000 kcal 5.5 2.8 5.2 3.4 4.6 2.3 4.3 2.9 <0.001
Zinc
mg/day 13.4 2.5 11.6 2.2 9.9 1.8 8.4 1.9 <0.001
mg/1000 kcal 7.3 1.0 7.0 1.2 6.8 1.2 6.7 1.4 <0.001
Magnesium
mg/day 417.3 60.0 349.2 48.9 311.1 46.3 256.2 51.7 <0.001 mg/1000 kcal 228.6 32.3 213.0 33.8 216.7 40.4 204.9 37.4 <0.001 Q: quartile, PUFA: polyunsaturated fatty acid. ANOVA and Kruskal-Wallis tests were used to evaluate the differences in mean dietary intakes between the DII score quartiles.
