ASSOCIATION OF MEDITERRANEAN AND BALTIC SEA DIETARY PATTERNS WITH MULTIMORBIDITY IN ELDERLY WOMEN:
OSPTRE-FPS STUDY
Huyen Nguyen Master's Degree in Public Health
University of Eastern Finland Faculty of Health Sciences School of Medicine huyen@student.uef.fi May 2020
Public health
NGUYEN, HUYEN: Association of Mediterranean and Baltic Sea dietary patterns with multimorbidity in elderly women: OSPTRE-FPS Study
Master's thesis, 61 pages.
Supervisors: Arja Erkkilä, Adjunct Professor, PhD; Fatemeh Ramezan Alaghehband, MPH.
May 2020
Key words: Mediterranean diet, Baltic Sea diet, multimorbidity.
Life expectancy at birth has increased worldwide, particularly since the 19th century. Among positive aspects, this trend raises one of the recent public health concerns, which is the growing number of certain non-communicable diseases (NCDs) as people get older. Notably, in the old population, multiple health conditions tend to occur more often in the same individual which referred to as multimorbidity.
Dietary factors can have direct or indirect effects on health. While an unhealthy diet is among the main risk factors for NCDs, a healthful diet improves health and protects people from many diseases including the major types of NCDs.
Previously, several studies have examined the relationship between multimorbidity and different factors, such as lifestyles, education, and socioeconomic factors. In terms of nutrition, the links between multimorbidity and certain food groups have been studied. Mediterranean diet (MED) and Baltic Sea diet (BSD) have been known as healthy dietary patterns; however, there have been no studies assessing their association with multimorbidity status. Therefore, this study primarily aimed to investigate the association of these two factors. Moreover, the relationships between several food groups and multimorbidity were also assessed as well as an additional analysis was done to discover the multimorbidity patterns among the participants.
The data for this cross-sectional study came from 554 elderly women who participated in the Osteoporosis Risk Factor and Fracture Prevention Study (OSTPRE-FPS), completed a 3-day food record and a self-administrated questionnaire at the baseline. Based on these data, the consumption of foods and nutrient intakes was calculated for each participant. The MED score was determined by 8 components, and 9 components were used to calculate the BSD score.
These two dietary scores were divided into quartiles which indicate different levels of conformance with the MED and the BSD. Multimorbidity was defined as having three or more health conditions among 20 chronic conditions. Multinomial logistic regression was used to examine the association of the BSD and the MED with multimorbidity. Three models of adjustment for covariates were used.
The results showed no associations between the MED, the BSD, or any food groups with multimorbidity status in the study population. The findings also indicated that the most common multimorbidity pattern was comprised of hypertension, CHD, and high cholesterol.
Although this study did not find any association of the MED, the BSD, or food groups with multimorbidity in the cohort of elderly women, it contributes a piece of information to this new topic. Further research is needed to provide a clear insight into the possible association of MED and BSD with multimorbidity.
First and foremost, I would like to express my sincere gratitude to my primary supervisor, Professor Arja Erkkilä, for the valuable advice, support, and thorough guidance from the very beginning to the end of my thesis process. Moreover, I am enormously grateful to my second supervisor, Fatemeh Ramezan Alaghehband, who was always available to help and support me at every stage of the thesis work.
My deep appreciation also goes to the University of Eastern Finland’s, Department of Public Health, for accepting me to this program, giving me the opportunity to broaden my knowledge and flexible time to finish my thesis.
Furthermore, I would like to give special thanks to my dear husband and son, I am ultimately grateful for our love and companionship. Also, many thanks to my wonderful parents, siblings, and in-laws, for their endless care and encouragement throughout my years of study and through the process of this thesis despite the geographical distances among us.
Finally, I wish to extend my thanks to other family members and friends for the unconditional support and encouragement all through the thesis process.
Oulu, May 2020 Huyen Nguyen.
2 LITERATURE REVIEW ... 8
2.1 Multimorbidity ... 8
2.1.1 Definition and measurement ... 8
2.1.2 Prevalence and patterns ... 9
2.1.3 Risk factors ... 11
2.1.4 Impacts ... 12
2.2 Dietary patterns ... 13
2.2.1 Definition ... 13
2.2.2 Measurements ... 14
2.3 Mediterranean dietary pattern ... 15
2.3.1 Definition ... 15
2.3.2 Studies on the association of MED and health outcomes ... 15
2.4 Baltic sea dietary pattern ... 22
2.4.1 Definition ... 22
2.4.2 Studies on the association of BSD and health outcomes ... 22
2.5 Dietary factors and multimorbidity ... 29
3 AIMS OF THE STUDY ... 33
4 METHODOLOGY ... 34
4.1 Study design and data collection ... 34
4.2 Measurements of multimorbidity and dietary intake ... 36
4.2.1 Dietary and alcohol intakes ... 36
4.2.2 Mediterranean dietary pattern ... 36
4.2.3 Baltic sea dietary pattern ... 36
4.2.4 Multimorbidity ... 38
4.3 Potential confounding factors ... 38
4.4 Statistical analysis ... 38
4.5 Ethical considerations ... 39
5 RESULTS ... 40
5.1 Basic Characteristics ... 40
5.3 Association of food groups with multimorbidity ... 46
5.4 Multimorbidity patterns ... 47
6 DISCUSSION ... 48
6.1 Association of the MED and BSD with the components of the common multimorbidity48 patterns ... 48
6.2 Association of food groups and multimorbidity ... 49
6.3 Multimorbidity patterns in the elderly population ... 50
6.4 Strengths and limitations ... 51
7 CONCLUSIONS ... 52
8 REFERENCES ... 53
BMI Body mass index BP Blood pressure BSD Baltic Sea diet
CHD Coronary heart disease
COPD Chronic obstructive pulmonary disease CVD Cardiovascular disease
DASH Dietary Approaches to Stop Hypertension HDL High-density lipoprotein
LDL Low-density lipoprotein
LMIC Low and middle-income country MCI Mild cognitive impairment MED Mediterranean diet
MI Myocardial infarction MUFA Monounsaturated fatty acid NCD Non-communicable disease PAL Physical activity level PUFA Polyunsaturated fatty acid T2D Type 2 diabetes
TG Triglyceride SFA Saturated fatty acid
RCT Randomized controlled trial WHO World Health Organization
1 INTRODUCTION
Globally, life span has increased from 52.5 years in 1960 to 72,4 years in 2017 (World Bank, 2019). Among positive aspects, this trend raises one of the recent public health concerns, which is the growing number of certain non-communicable diseases (NCDs). According to the World Health Organization (WHO), certain NCDs such as cardiovascular diseases (CVDs), cancer, type 2 diabetes (T2D), and chronic obstructive pulmonary disease (COPD) are among the most common challenges, and the prevalence of those diseases increase with age (WHO, 2019).
Multimorbidity is defined by WHO as “the coexistence of two or more chronic conditions in the same individual” and occurs more often in the disadvantaged groups which needs complex and long-term care (WHO 2016). The prevalence of multimorbidity has been increasing since several past decades, and the most reasonable explanation for this trend is highly likely because of the aging population worldwide (Lutz et al. 2008). The optimal nutritional status is one of the crucial factors, which determines people’s health and has huge effects on the aging process.
Several studies have indicated that healthy dietary patterns have positive effects on health in general and in preventing certain chronic diseases (Tapsell et al. 2016, Kelly et al. 2017, Satija
& Hu 2018).
There have been different studies investigating the relationship between multimorbidity and different factors, such as lifestyles, education, and socioeconomic factors (Fortin et al. 2014, Dhalwani et al. 2017, Lund Jensen et al. 2017, Read et al. 2017, Amaral et al. 2018, Mercer et al. 2018, Singh-Manoux et al. 2018). Mediterranean diet (MED) and Baltic Sea diet (BSD) have been known as healthy dietary patterns, because they have been associated with reduced CVD incidence and mortality (Berild et al. 2017, Grosso et al. 2017, Dinu et al. 2018, Galbete et al.
2018b, 2018a, Martinez-Lacoba et al. 2018), lower risk of T2D (Lacoppidan et al. 2015, Jannasch et al. 2017, Galbete et al. 2018a, Rees et al. 2019), and might contribute to decrease the risk of sarcopenia in elderly women (Isanejad et al. 2018). However, the links between these two dietary patterns and multimorbidity in the elderly population have not been examined.
This study assessed the association of MED and BSD with multimorbidity in the cohort of older women from a randomized trial of the OSTPRE: Kuopio Osteoporosis Risk Factor and Prevention Study.
2 LITERATURE REVIEW
2.1 Multimorbidity
2.1.1 Definition and measurement
Definition
Multimorbidity is simply considered as the co-occurrence of multiple health conditions in a person. The first publication on multimorbidity was in German in 1976, and this concept has been used more universally since 1990 (Brandlmeier 1976, Heuft 1990, Le Reste et al. 2015).
According to a systematic review by Johnston and colleagues (2019), there was no consensus for the minimum number of multiple conditions. Although the cut-off point for multiple conditions was different among studies, the most common cut-off point for multimorbidity used in literature is two or more chronic diseases in the same individual (Johnston et al. 2019), which also corresponds to the definition by WHO (WHO 2016). Another systematic review indicated that in addition to the cut-off of 2 chronic conditions, most of the authors also reported the cut- off of 3 or more, or 4 or more diseases (Fortin et al. 2012). Furthermore, Fortin et al. (2012) suggested that the cutoff point of 3 could be more beneficial when studying the higher risk groups because it provides lower prevalence of multimorbidity which in turn helps to better identify patients with higher needs for treatment. The range of conditions can vary between 4 to 147; all definitions of multimorbidity included diseases, the majority included risk factors, while symptoms and severe conditions were less commonly included (Willadsen et al. 2016).
Measurement
Systematic reviews showed that there were challenges and heterogeneity for both definition and measurement of multimorbidity which suggest readers to be more cautious when interpreting the results (Fortin et al. 2012, Willadsen et al. 2016, Xu et al. 2017, Johnston et al. 2019).
Findings from an expert panel and online survey indicated that there is no best single measure of multimorbidity that can apply for all studies, and measurement is determined by several factors, namely the study purpose, the expected outcomes or the preferences of those who are involved in the study (Griffith et al. 2018). According to Johnston and colleagues (2019), two main multimorbidity measures were “disease counts and weighted indices such as the Charlson Index, the Cumulative Illness Rating Scale, the Index of Coexistent Disease, the Adjusted Clinical Groups System and the Duke Severity of Illness.” Each method has its own advantages.
Hence, choosing an appropriate method is important to the study outcome's accuracy. More specifically, based on the count of conditions, a wide range of health conditions can be included, and this method is more suitable when the outcomes of interest encompass various aspects, or when weighted measures have not been validated (Johnston et al. 2019). In addition, Johnston et al. (2019) also pointed out that applying weighted measures can be more beneficial than the disease count method when the study purpose is to assess outcomes in participants as well as to provide an explanation for the occurrence of multiple health conditions. Data on health conditions can come from case-note review, administrative data, or patient self-report; and the outcomes will be influenced by the differences of data resources (Johnston et al. 2019).
2.1.2 Prevalence and patterns
Prevalence
As described previously, the variations in the cut-off points of multimorbidity definition, the list of included diseases, and the method of measuring health conditions strongly affect the prevalence of multimorbidity. Moreover, the study context (primary healthcare settings, general population, and specific age groups), the difference in geographic settings, and the sample size also affect the prevalence of multimorbidity (Fortin et al. 2012).
It was estimated in a systematic review that the prevalence of multimorbidity worldwide is more than 60%, and more than 80% in the population aged ≥85 years old (Salive 2013). In 2008, 67% of Medicare beneficiaries in the US had multimorbidity, and the prevalence increased as people get older, from 62% between 65 and 74 years old to 81.5% among those
≥85 years old. This study also showed that women had a higher prevalence of multimorbidity than men in every age group (Salive 2013).
A cross-sectional study in almost 20 000 participants in Germany indicated that 39.6% of them lived with at least two chronic diseases, and the prevalence of multimorbidity escalated substantially corresponding to age: about 50% of those who were between 50–59 years suffered from these health condition (Puth et al. 2017). Another study conducted by Britt et al. (2008) in Australia estimated the prevalence of multimorbidity was 25.5% for the whole country and it increased with age. More specially, 83.2% of the participants who aged more than 75 had multimorbidity, of which 58.2% suffered at least three or more and 33.4% suffered four or more chronic conditions (Britt et al. 2008).
In a multi-country study which included nine countries in Europe, Asia and Africa in adults aged more than 50 years old showed that Russia had the highest prevalence of multimorbidity (71.9%), while China and Ghana had the lowest number of multimorbidity of 45.1% and 48.3%
respectively (Garin et al. 2016). Findings from this study also indicated that while the prevalence of multimorbidity is greater in high-income countries, it is also gradually increasing in low and middle-income countries (LMICs).
Multimorbidity patterns
Multimorbidity in young-age groups often contains both mental and physical health issues.
(Lyness et al. 2006). On the other hand, the multiple physical health conditions such as hypertension, pain, diabetes, coronary heart disease (CHD) were more common for the group aged more than 55 years old (McLean et al. 2014).
According to Salive (2013), among Medicare beneficiaries in the US who had multimorbidity, hypertension and hyperlipidemia made up the most common pair. Moreover, hypertension together with hyperlipidemia, and ischemic heart disease created the most common combination of three chronic conditions in the same individual (Salive 2013).
The multi-country study by Garin et al. (2016) showed that hypertension, arthritis, and cataract were most frequently co-occurring with other health conditions. Their results also mentioned several common disease patterns such as “metabolic” pattern (diabetes, obesity, and hypertension), “cardio-respiratory” pattern (angina, asthma, and chronic obstructive pulmonary disease), “mental-articular” pattern (arthritis, depression) among others (Garin et al. 2016).
The overview of systematic reviews by Xu et al. (2017) provided three main disease patterns which were cardiovascular and metabolic disease, musculoskeletal disorders, and mental health-related problems. The most common disease components of multimorbidity according to this review were diabetes, heart disease, cancer, hypertension, depression, COPD, stroke, arthritis/osteoarthritis, osteoporosis, and asthma (Xu et al. 2017).
2.1.3 Risk factors
Both modifiable and non-modifiable risk factors for multimorbidity were identified. Non- modifiable risk factors include demographic factors (age, sex, race), and socioeconomic factors;
the modifiable risk factors are overweight/obesity, elevated blood pressure (BP), and abnormal lipids, diet, physical activity, and smoking (Staimez et al. 2017). Understanding these factors clearly is fundamental due to the fact that it would be beneficial to develop prevention strategies (Salive 2013).
Non-modifiable risk factors
As mentioned earlier, individuals were more prone to suffer from multiple long-term health conditions as they get older (Britt et al. 2008, Salive 2013, Olivares et al. 2017, Puth et al. 2017, Staimez et al. 2017). Moreover, elderly women were more vulnerable to multimorbidity compared to elderly men (Salive 2013, Olivares et al. 2017, Hu et al. 2019, Yao et al. 2019).
Regarding ethnicity, a cross-sectional retrospective study in Spain conducted by Gimeno-Feliu and colleagues (2017) found that in comparison to native-borns, foreign-borns have lower risk of multimorbidity; however, the prevalence of multimorbidity significantly rose with respect to the time lived in the host country. More specifically, the lowest risk was for Asians, followed by Eastern Europeans compared to Latin Americans with the highest risk (Gimeno-Feliu et al.
2017). In addition, it was pointed out that the odds of multimorbidity were substantially greater in ethnic minority groups which came from South-Asian Surinamese, African Surinamese, Ghanaian, Turkish, and Moroccan compared to the host country (Netherland) (Verest et al.
2019).
When it comes to socioeconomic status, the chances of getting multiple chronic diseases later in life grew 8% by childhood financial hardship, while higher-than-average lifetime earnings throughout youth to middle age reduced the chances by 5% (Tucker-Seeley et al. 2011).
Furthermore, the lowest level of education increased the odds of multimorbidity compared to the highest one in both men and women.
Modifiable risk factors
According to WHO, four leading risk factors for death due to NCDs are tobacco use, harmful use of alcohol, unhealthy diet, and physical inactivity (WHO 2018). CVDs are the leading cause
of death among NCDs; cancers, respiratory and diabetes are also common, and these NCDs are long-term conditions which usually coexist in the same person rather than a single disease (Violan et al. 2014).
Even though most of the deaths due to NCDs in LMICs happened in the group of people between 30-69 years old, the elderly were among the vulnerable groups because the rapidly increased prevalence of multimorbidity in the old population (WHO 2018). Additionally, the accumulative effect of unhealthy lifestyle factors (smoking, excessive alcohol use, low fruit and vegetable intakes, low physical activity, and high body mass index (BMI)) increased the risk of multimorbidity (Fortin et al. 2014, Olivares et al. 2017).
Moreover, the prevalence of multiple chronic conditions gradually increased with BMI category with the highest risk being in obese category III whose BMI >40.0kg/m2 (Booth et al.
2014, Jovic et al. 2016). Multimorbidity also tended to appear more commonly in hypertensive patients ( Wong et al. 2014, Staimez et al. 2017), and in those with dyslipidemia, metabolic syndrome (Olivares et al. 2017).
2.1.4 Impacts
Multimorbidity has been a major public health issue worldwide, it affects not only individuals’
life but the entire society. The co-occurrence of T2D and CHD or stroke and coronary disorders led to more impaired health-related quality of life (Hunger et al. 2011). Particularly, elderly people with multimorbidity have increased risk of being depressed and having other mental health problems (Barnett et al. 2012). A systematic review and meta-analysis indicated that higher mortality and lower quality of life were among the common consequences of multimorbidity in the group of people aged more than 65 (Nunes et al. 2016). More specifically, the risk of death increased 73% in those having two or more diseases, 172% in those having three or more diseases (Nunes et al. 2016). In Southern China, multimorbidity was found to be an independent risk factor, which significantly reduced functional independence of older population (Wang et al. 2017).
Multimorbidity not only affected people’s physical and mental health, it also involved in different adverse drug events, polypharmacy, more frequent use of health care services and high treatment burden (Kernick et al. 2017). Findings from a systematic review indicated that there
was a positive correlation between multiple chronic health conditions and the health care usage and costs, in which the use and costs substantially rose with each additional disease (Lehnert et al. 2011). In the case of Singapore, the total cost used for healthcare for elderly people with multimorbidity was three times higher compared to those with single chronic disease and more than five times compared to those with no long-term conditions (Picco et al. 2016).
2.2 Dietary patterns
2.2.1 Definition
The latest definition of a dietary pattern is “the quantity, variety, or combination of different foods and beverages in a diet and the frequency with which they are habitually consumed”
(Sánchez-Villegas & Martínez-Lapiscina 2018).
In recent years, dietary pattern analysis in relation to the risk of chronic diseases has been paid more attention compared to studies of a single or a few nutrients. Human meals contain a variety of different foods, and nutrients from the food might interact with each other; therefore, the study of a single compound may be confounded by others (National Research Council (US) Committee on Diet and Health 1989, Willett 2013). The cumulative effects of several nutrients in a dietary pattern might be easier to be observed and analyzed as compared to the small effects of specific components (Sacks et al. 1995, Willett 2013). Furthermore, the public health implications of healthy diets in promoting general health and preventing chronic diseases might benefit more from the study of overall, regularly consumed patterns (Hu 2002). For example, a low-fat diet was found to be associated with higher intakes of vegetables, fruits, fiber, folate, whole grains and the joint effects of the components were associated with reduced risk of CHD (Ursin et al. 1993). Studying overall dietary patterns also directly link to dietary guidelines for nutrition intervention and education (Hu 2002).
Examples of dietary patterns
There are two main groups of dietary patterns: healthy and unhealthy. According to the dietary guidelines for Americans 2015-2020, healthy eating pattern is characterized by high consumption of a variety of vegetables, fruits, whole grains, fat-free or low-fat dairy products, a variety of protein foods, oil while low in saturated and trans-fat, added sugar and sodium (U.S.
Department of Health and Human Services and U.S. Department of Agriculture 2015). These
dietary guidelines mentioned several healthy dietary patterns as examples of healthy eating patterns, for instance, the MED, the vegetarian dietary pattern, and the Dietary Approaches to Stop Hypertension (DASH) (U.S. Department of Health and Human Services and U.S.
Department of Agriculture 2015). The DASH diet is described by high consumption of vegetables, fruits, low-fat dairy products, whole grains, poultry, fish, beans, and nuts while low in unhealthy food choices (sweet, processed products), and it has been proven to have associations with lowered BP and low-density lipoprotein (LDL) cholesterol concentrations, which in turn reduce CVD risk (Challa et al. 2019).
In contrast, Western diet is considered as an unhealthy dietary pattern which is high in animal protein, saturated fats, refined grains, sugar, alcohol, salt, and corn-derived fructose syrup, while low in intake of fruits and vegetables (Statovci et al. 2017). This unhealthy dietary pattern has been associated with a wide range of negative health consequences such as hypertension, diabetes, obesity, and chronic kidney disease (Hariharan et al. 2015), and increased risk of CHD, CVD, stroke and death (Rodríguez-Monforte et al. 2015).
2.2.2 Measurements
Dietary patterns can only be measured indirectly through collected dietary information and statistical methods. As the literature indicates, three approaches to describe dietary patterns were factor analysis, cluster analysis, and dietary indices (Hu 2002). While the first two methods are based on multivariate statistical techniques using information from food frequency questionnaires (Nicklas et al. 1989, Hu et al. 1999, 1999, 2000,) or from dietary records (Barker et al. 1992), the dietary indices are built up by dietary recommendations. Therefore, the dietary index approach is considered as ‘a priori’ (Hu 2002); on the other hand, factor analysis and cluster analysis are referred as ‘a posteriori’ (Trichopoulos & Lagiou 2001). Each method has its own limitations. According to Hu (2002), the multivariate technique creates dietary patterns solely based on available data, therefore, it might not provide best patterns. On the other hand, dietary index method is based on the current available knowledge, there is heterogeneity regarding components and cut-off points of the dietary score which may lead to biased results (Hu 2002). Dietary pattern scores can be used to assess the level of adherence to definite diets or recommendations such as the Healthy Eating Index which initially assessed the compliance levels of people’s diet with the recommendations in the US (Schulze & Hoffmann 2006).
2.3 Mediterranean dietary pattern
2.3.1 Definition
The Mediterranean diet was first coined by Keys and Grande (1957) which characterized the pattern by high consumption of fruits and vegetables, nuts, legumes, olive oil while moderate alcohol intake, low to moderate intake of meat and dairy products and low in saturated fat (Keys
& Grande 1957). The MED was a healthy dietary pattern mostly in the Mediterranean area;
however, it has gained more interest from both scientists and general population in Europe as well as becoming increasingly popular globally in recent years (Boucher 2017, Lăcătușu et al.
2019). Therefore, the definition of MED has been developed and varied, yet the original components of this diet have remained.
2.3.2 Studies on the association of MED and health outcomes
Due to the substantial number of studies focusing on the association of MED and health outcomes available in the literature, in this part we only chose systematic reviews and meta- analyses conducted within 5 years (2015-2020) to describe the association. Results are described below and a summary of included studies is presented in Table 1.
General morbidity and mortality
An umbrella review of meta-analyses including 13 meta-analyses of observational studies and 16 meta-analyses of randomized controlled trials (RCTs) conducted by Dinu and colleagues (2018) aimed to assess the links between MED and 37 health outcomes. The results showed that the MED contributed to lower risk of overall mortality compared to the others (Dinu et al.
2018).
On the other hand, study conducted by Martinez-Lacoba et al. (2018) including nine systematic reviews and 24 meta-analyses to evaluate the association of the MED and different health outcomes. Their findings indicated that a MED was inversely associated with all-cause mortality, but the quality of evidence was low (Martinez-Lacoba et al. 2018). Likewise, Rees et al. (2019) conducted a systematic review consisting of 22 RCTs examined the effect of adherence to a MED and primary prevention of CVD. Out of these, only one trial which was the PREDIMED trial (one group was supplemented with extra‐virgin olive oil, and the second
one was supplemented with nuts) reported the effect on overall mortality. Their results found no effect of the PREDIMED intervention compared to a low‐fat diet on total mortality (Rees et al. 2019).
CVD
An inverse association between adherence to a MED and CVD incidence and mortality was identified (Grosso et al. 2017, Galbete et al. 2018b). The results were also analysed further for the components of MED, and suggested that olive oil, vegetable, fruit, and legumes seem to have bigger influence compared to the other components (Grosso et al. 2017). Moreover, an umbrella review of meta-analyses conducted by Dinu and colleagues (2018) provided vigorous evidence on the association of adherence to a MED and reduced CVD incidence including stroke and myocardial infarction (MI) (Dinu et al. 2018). Similarly, results from the study by Martinez-Lacoba et al. (2018) also supported this association.
On the other hand, based on the PREDIMED intervention, Rees et al. (2019) found little or no effect of the MED compared to a low‐fat diet on CVD mortality or myocardial infarction.
However, there was a moderate‐quality evidence for a reduction in the number of strokes with the intervention (Rees et al. 2019).
Components of metabolic syndrome
Components of metabolic syndrome, with an exception of blood glucose level, were found to have an inverse association with the level of adherence to a MED, and therefore, it was suggested as a diet to prevent metabolic syndrome (Godos et al. 2017). In addition, the MED might help to prevent central obesity (Bendall et al. 2018), had an association with reduced systolic and diastolic BP as well as body weight and BMI (Martinez-Lacoba et al. 2018).
However, based on findings from Rees et al. (2019), there was only moderate‐quality evidence supporting the reduction in systolic and diastolic BP among those who followed a MED.
T2D
Jannasch and colleagues (2017) reviewed and analyzed 48 longitudinal studies on the effects of several dietary patterns and prevention of T2D. Their results showed that the MED was among the healthy diets which can be used for preventing T2D (Jannasch et al. 2017).
Additionally, findings from the above studies also indicated that adherence to a MED
contributed to prevent and manage T2D (Dinu et al. 2018, Galbete et al. 2018b, Martinez- Lacoba et al. 2018, Rees et al. 2019).
Cancer
A study of 83 cohort and case-control studies observed that the highest MED score was associated with a lower risk of overall cancer mortality, colorectal cancer, breast cancer, gastric cancer, liver cancer, and this positive effect was mostly attributed to higher intakes of fruits, vegetables, and whole grains (Schwingshackl et al. 2017). Similar results were found by the studies of Galbete et al. (2018) and Martinez-Lacoba et al. (2018).
However, based on findings from Dinu et al. (2018), the MED may help to reduce overall cancer incidence, but not cancer mortality. Specifically, there was weak evidence on the links between MED and several cancers (colorectal, liver, and pancreatic cancer), no evidence for bladder, endometrial and ovarian cancer, and conflicting evidence if considering breast, gastric, prostate, esophageal, respiratory and head/neck cancer (Dinu et al. 2018).
Cognitive functions
Based on findings from Martinez-Lacoba et al. (2018), better adherence to a MED was associated with improved cognitive function and reduced risk of several cognitive conditions (cognitive impairment, dementia, Alzheimer’s disease, depression). However, results for mild cognitive impairment (MCI) were inconsistent (Martinez-Lacoba et al. 2018). Likewise, Wu &
Sun (2017) and Dinu et al. (2018) supported the association between highest compared to lowest category of adherence to a MED and lower risk of cognitive diseases including dementia and Alzheimer’s disease. Additionally, their findings provided further information which indicated that the median category of the MED score as compared to the lowest category was not significantly associated with the risk of cognitive disorders (Wu & Sun 2017). On the other hand, Galbete et al. (2018) showed that compared to the lowest MED adherence category, the highest category was associated with reduced risk of MCI and Alzheimer’s disease, but not for dementia.
Bone and muscle health
Findings from a study by Craig et al. (2017) indicated that there was not enough evidence to conclude the effects of the MED on musculoskeletal health including fracture and sarcopenia in all age groups (Craig et al. 2017). The study by Martinez-Lacoba et al. (2018) also indicated
an inconsistent association between the MED and fracture risk. Nevertheless, Silva et al. (2018) found that the MED helped to improve the functional disability and frailty in the old people, but no association was found for sarcopenia (Silva et al. 2018).
Other health outcomes
There has been a small number of studies accessing the links between the MED and osteoarthritis (Morales-Ivorra et al. 2018), rheumatoid arthritis (Forsyth et al. 2018) and asthma (Papamichael et al. 2017, Martinez-Lacoba et al. 2018). Therefore, it seems that there is limited evidence to conclude about the association between the MED and these health outcomes.
Conclusion
Based on the review of literature, there has been a strong relationship between higher adherence to a MED and reduced CVD incidence and mortality, and T2D incidence. Even though adoption to a MED has seemed to have beneficial effects on overall mortality rate, metabolic syndrome profile, cancer as well as cognitive functions, the quality of evidence has been considered as low to moderate. Finally, there has not been enough evidence to provide the links between the MED and other health outcomes such as muscle and bone health, osteoarthritis, rheumatoid arthritis, and asthma.
Table 1. Summary of included studies about the association of MED and health outcomes
Author Participant/
Sample size
Study aim Main findings
Craig et al. 2017 3 prospective studies Assess the association between MED and musculoskeletal health.
There is lack of evidence regarding the association between the MED and musculoskeletal health in all age groups
Godos et al. 2017 8 cross-sectional studies and 4 prospective studies
Review and assess the links between adherence to a MED and the risk of metabolic syndrome.
Most components of metabolic syndrome, with an exception of blood glucose level, were found to have an inverse association with level of adherence to the MED.
Grosso et al. 2017 17 prospective studies and RCTs
Examine the association between the MED and CVD incidence and mortality.
A MED is associated with lower risks of CVD incidence and mortality, including CHD and MI.
Schwingshackl et al.
2017
83 cohort and case-control studies
Evaluate the association of MED and cancer incidence and mortality.
The highest adherence score to the MED was associated with a lower risk of overall cancer mortality, and certain cancer types. The higher intakes of fruits, vegetables, and whole grains played the most important role.
Wu & Sun 2017 9 prospective cohort studies Access the association between MED and cognitive function.
Only the highest category of the MED score was inversely associated with the developing of cognitive disorders.
Bendall et al. 2018 18 intervention trials Evaluate the effectiveness of the MED on preventing central obesity.
The MED possibly helps to prevent central obesity.
Dinu et al. 2018 13 meta-analyses of observational studies and 16 meta-analyses of RCTs
Evaluate the association between the MED and 37 different health outcomes.
An inverse association between greater adherence to the MED and reduced risk of overall mortality, CVD, CHD, MI, overall cancer incidence, neurodegenerative diseases, and diabetes. No evidence reported for other cancers, as well as for LDL cholesterol levels.
Galbete et al. 2018 14 publications reporting 27 meta-analyses from 70 primary studies
Review and evaluate the evidence on the effects of the MED.
An inverse association between greater adherence to the MED and lower incidence of T2D, lower incidence/mortality of CVD and cancer.
Martinez-Lacoba et al. 2018
9 systematic reviews and 24 meta-analyses
Summarize and synthesize the effects of the MED pattern on different health outcomes.
Quality of evidence on the inverse association of the MED and all-cause mortality was low, but moderate for BMI, cancer, cognitive functions, bone health. Strong evidence supports the association of better adherence to a MED and reduced risk of CVD incidence/mortality and T2D.
Silva et al. 2018 8 cohort studies and 4 cross- sectional studies
Access the links between the MED and musculoskeletal health in the old population.
The MED helped to improve the functional disability and frailty in the old people, but no effects on sarcopenia.
Rees et al. 2019 22 RCTs Examine the effect of adherence
to a MED and primary prevention of CVD.
No effects on total mortality, no or little effects on CVD mortality or MI, risk of stroke, T2D and hypertension.
BMI: body mass index, CHD: coronary heart disease, CVD: cardiovascular disease, LDL: low-density lipoprotein, MED: Mediterranean diet, MI:
myocardial infarction, T2D: type 2 diabetes, RCT: randomized controlled trial
2.4 Baltic sea dietary pattern
2.4.1 Definition
The Baltic Sea is an area in Northern Europe, surrounded by Finland, Sweden, Estonia, Denmark among others (World Map 2015). Therefore, the BSD is a dietary pattern referring to those areas around the Baltic Sea, and Nordic diet is another term for this type of diet. It is also a healthy one which has been modified from the traditional MED due to the differences in climate, resources, and food culture in the Nordic countries. The BSD indicates a balanced food intake from different resources which meets the current recommended intakes and average requirements from the Nordic Nutrition Recommendations 2004.
The healthy food choices from BSD were first introduced in 2011 through a pyramid by the collaboration of different organizations in Finland such as the SYSDIET study (a Nordic Centre of Excellence on Food, Nutrition and Health) of the University of Eastern Finland, the Finnish Heart Association and the Finnish Diabetes Association (Kanerva et al. 2014a). The BSD pyramid includes ten groups of food, most of the healthy food choices are Nordic vegetables (legumes, cabbages, tomato, leafy vegetables, and cucumber), Nordic fruits (berries, apples, and pears). These food types should be consumed in high amount and are located at the bottom of the pyramid; wholegrain products (rye, oat, and barley) placed in the center of the pyramid, followed by fish, fat-free or low-fat milk products, and rapeseed oil. Moreover, other foods that are considered to be unhealthy and should be consumed in moderate or low amount locating at the pyramid’s top (processed meat, butter, chocolate, sweets, and sweet bakery products) (Kanerva et al. 2014a).
2.4.2 Studies on the association of BSD and health outcomes
The BSD is only popular among Nordic countries, thus the number of studies on the association of BSD and health outcomes in the literature is much less compared to the studies on MED.
Therefore, all related studies found in literature were included in this part. Although these studies considered a range of different confounding factors, several common covariates were included such as energy intake, BMI as well as basic demographic variables (age, sex,
education), and lifestyle elements (physical activity level, smoking status). More detailed information is presented below and a summary of included studies is presented in Table 2.
General morbidity and mortality
Healthy foods of the BSD promoted general health and contributed to lower mortality rate in the cohort of 57,053 Danish men and women (Olsen et al. 2011). Moreover, based on data from a large prospective cohort study in Swedish women aged 29-49 years old, adherence to the BSD was found to inversely associated with overall mortality (Roswall et al. 2015b).
CVD
A systematic review based on several dietary interventions in Nordic countries indicated that the BSD helped to reduce various risk factors for the developing of CVD among those at high risk (Berild et al. 2017). In addition, based on a Danish cohort study, the BSD was suggested to possibly prevent stroke incidence (Hansen et al. 2017). Another study came from the same Danish cohort study showed that the BDS contributed to reduce the risk of MI in both men and women (Gunge et al. 2017). Furthermore, findings from the EPIC-Potsdam longitudinal study showed that the risk of MI in the general population and stroke in men could be partly reduced by following the BSD (Galbete et al. 2018a). The mortality due to any disease, especially CVD, was higher among those with lower adherence to the BDS in a cohort of men from the region of Eastern Finland (Tertsunen et al. 2020).
Components of metabolic syndrome
A study conducted by Kanerva et al. (2013) found no association between the BSD score and BMI. However, the BSD might help to reduce the risk of abdominal obesity, especially when it is combined with a moderate alcohol intake (Kanerva et al. 2013). On the other hand, a systematic review and meta-analysis of RCTs indicated that the BDS improved BP, significantly reduced the total and LDL cholesterol, but did not affect high-density lipoprotein (HDL) cholesterol and triglyceride (TG) levels (Ramezani-Jolfaie et al. 2019).
T2D
Kanerva et al. (2014) conducted a prospective cohort study which followed participants for a period of ten years. Their results showed no association between adherence to the BSD and the incidence of T2D (Kanerva et al. 2014b). On the other hand, based on data from a longitudinal study including 57,053 Danish who were between 50 to 64 years old at the baseline, an inverse
association was found between better adherence to the BSD and the risk of T2D for both men and women (Lacoppidan et al. 2015).
Cancer
Higher adherence to the BSD was found to be associated with lower risk for colorectal cancer in Danish women, but not in men (Kyrø et al. 2013). Conversely, findings from another longitudinal study of 45,222 Swedish women (29–49 years) indicated no association between BSD and the incidence of colorectal cancer after a 20 year period of following up (Roswall et al. 2015a). Adherence to the BSD also did not show any effects on the risk of breast cancer in the cohort of young women in Sweden (Li et al. 2015). In addition, no association was found between adherence to the BSD and risk of cancer in the EPIC-Potsdam cohort study (Galbete et al. 2018a).
Cognitive functions
Results from a longitudinal study conducted by Männikkö et al. (2015) suggested that following a BSD may provide benefits for cognitive abilities of individuals with normal cognitive functions (Männikkö et al. 2015). No other studies have been identified regarding this association.
Bone and muscle health
In women, adherence to the healthy Nordic diet enhances overall physical performance (Perälä et al. 2016), and it can be a protective factor from weaker muscle strength but not from lower muscle mass (Perälä et al. 2017). In addition, higher adherence to the BSD may contribute to decreased risk of sarcopenia in women of old age (Isanejad et al. 2018).
Conclusion
Based on the literature, it seems that the BSD has been associated with lower mortality and morbidity rate and incidence of CVD. However, the links between adherence to the BSD and various risk profiles for metabolic syndrome, T2D, cancer, cognitive functions, and bone and muscle health have been still inconclusive due to little evidence and conflicting results from different studies.
Table 2. Summary of included studies about the association of BSD and health outcomes
Author Study design Participants/
Sample size
Study aim Main findings
Kanerva et al.
2013
Cross-sectional study
4720 Finns (25-74 years) Association between BSD score and the risk of obesity and abdominal obesity.
No association between the BSD and BMI, however the BSD might help to reduce the risk of abdominal obesity.
Kyrø et al. 2013 Longitudinal study
57 053 Danish men and women aged 50-64 years;
follow up:13 years
Association between adherence to the BSD and the risk of colorectal cancer.
BSD → colorectal cancer in women (not in men).
Kanerva et al.
2014b
Longitudinal study
6 745 Finnish men and women (56-70 years);
follow up: 10 years
Association between adherence to the BSD and incidence of T2D.
Adherence to the BSD was not associated with T2D incidence.
Lacoppidan et al. 2015
Longitudinal study
57 053 Danish men and women aged 50-64 years;
follow up: 15.3 years
Association between BDS and T2D.
BSD → T2D for both men and women.
Li et al. 2015 Longitudinal study
44,296 Swedish women (29–49 years); follow up:
20 years
Association between the BSD and the risk of breast cancer.
No association was found between adherence to the BSD and the risk of breast cancer.
Männikkö et al.
2015
Longitudinal study
1140 Finnish women and men (57-78 years);
follow up: 4 years
Association between the BSD and the cognitive function.
Suggested beneficial effects of the BSD to cognitive function among normal people.
Roswall et al.
2015a
Longitudinal study
49,259 Swedish women (29–49 years); follow up:
20 years
Association between adherence to the BSD and the risk of colorectal cancer.
No association was found between adherence to the BSD and the risk of colorectal cancer.
Roswall et al.
2015b
Longitudinal study
49,259 Swedish women (29–49 years); follow up:
20 years
Association between adherence to the BSD and all-cause mortality and mortality due to some specific diseases.
BSD → overall mortality and mortality due to cancer in general, but not for the risk of CVD.
Perälä et al.
2016
Longitudinal study
1072 Finnish men and women, mean age: 61 years; follow up: 10 years
Association between the BDS and the physical performance after the period of 10 years.
Adherence to the BSD enhanced overall physical wellness and may contribute to the prevention of disability in the elderly women. No association was detected in men.
Perälä et al.
2017
Longitudinal study
1072 Finnish men and women, mean age: 61 years; follow up: 10 years
Association between the BDS and muscle strength and mass after the period of 10 years.
Adherence to the BSD can be a protective factor from weaker muscle strength, but not from muscle mass in women. No association was found in men.
Berild et al.
2017
Systematic review of RCTs
15 studies based on 4 interventions in Nordic countries
Review the effects of the BDS in relation with CVD.
The BSD helped to reduce various risk factors for the developing of CVD among those at high risk.
Gunge et al.
2017
Longitudinal study
57 053 Danish men and women (50-64 years);
follow up: 13.6 years
Association between the BSD and the risk of MI.
The BDS contributed to reduce the risk of MI in both middle age Danish men and women.
Hansen et al.
2017
Longitudinal study
55 338 Danish men and women (50-64 years);
follow up: 13.5 years
Association between the BSD and the risk of stroke.
The BSD was suggested to possibly prevent stroke incidence in the Danish men and women cohort.
Galbete et al.
2018a
Longitudinal study
27 548 men and women aged 35-65 years, follow up:10.6 years
Association between the BSD and the risk of chronic diseases (MI, stroke, T2D, cancer).
The risk of MI in the general population and stroke in men could be partly reduced by following the BSD. No association was found for risk of cancer.
Isanejad et al.
2018
Longitudinal study
554 Finnish women (65- 72 years); follow up: 3 years
Association between adherence to the BSD, MED, and the incidence of sarcopenia in the elderly women.
Higher adherence to the BSD may contribute to decrease the risk of sarcopenia in the women of old age.
Ramezani- Jolfaie et al.
2019
Systematic review and meta-analysis
5 RCTs including 513 men and women aged 39 to 60 years
Investigate the effect of the BSD on circulating levels of cholesterol, TG and BP in adults.
The BSD improved BP, significantly reduced the total and LDL cholesterol, but did not affect HDL cholesterol and TG levels.
Tertsunen et al.
2020
Longitudinal study
1547 men aged 42- 60 years; follow up:
23.6 years
Association between the BSD and the risk of chronic diseases (CVD and T2D).
The mortality due to any disease, especially CVD, was higher among those with lower adherence to the BDS in a cohort of men from the region of Eastern Finland.
BMI: body mass index, BP: blood pressure, BSD: Baltic Sea diet, CVD: cardiovascular disease, HDL: high-density lipoprotein, LDL: low-density lipoprotein, MED: Mediterranean diet, MI: myocardial infarction, T2D: type 2 diabetes, TG: triglyceride, RCT: randomized controlled trial
2.5 Dietary factors and multimorbidity
There has been no study on the association of the MED and BSD with multimorbidity even though these patterns have been linked with single diseases and risk factors. However, many studies have been focusing on the association between different dietary factors and multimorbidity. Dietary factors, mainly the consumption of fruit and vegetable, have been studied as a single exposure or more often in combinations with other lifestyle factors such as smoking status, alcohol consumption, physical activity level, fruit and vegetable intake, and BMI. More detailed information is presented below and a summary of included studies is presented in Table 3.
Longitudinal studies
A longitudinal study conducted by Ruel and colleagues (2014) showed that the prevalence of multimorbidity increased 20% after five years of following up the participants, and higher intakes of fruit, vegetable, whole grain products seemed to reduce the likelihood of multimorbidity in the general population (Ruel et al. 2014). Prospective cohort studies conducted by Wikström et al. (2015) and Dhalwani et al. (2017) also studied the association of fruit and vegetable consumption among other clinical and lifestyle factors with multimorbidity.
Their results indicated that low fruit and vegetable consumption together with physical inactivity increased the risk of multimorbidity in patients with CVD (Wikström et al. 2015).
Likewise, a 65% increase in the risk of multimorbidity associated with inadequate fruit/vegetable intakes (less than five portions per day) in women, but a paradoxical reduction of multimorbidity risk by 40% in males (Dhalwani et al. 2017). Findings from a cohort of elderly Spanish showed a positive association between high intake of fruit and vegetable with the probability of surviving among participants with two chronic diseases, but no association was observed for those who had none, 1 or ≥3 chronic conditions (Olaya et al. 2019).
Cross-sectional studies
In cross-sectional studies, dietary factors including fruit, vegetable, and alcohol intake among other clinical and lifestyle factors were examined in relation with multimorbidity. Results have been varied among studies. According to findings from Fortin et al. (2014), not meeting recommendations on of fruit and vegetables intakes (at least five portions per day), physical activity level or alcohol consumption were not associated with the presence of multimorbidity in both men and women, but the accumulation of more unhealthy lifestyle factors increased the
risk of multimorbidity. In contrast, a study conducted by Olivares et al. (2017) indicated that low intake of fruit and vegetables among other unhealthy lifestyle factors were risk factors of multimorbidity. Even though excessive alcohol consumption is harmful and can cause many different chronic diseases, moderate amount daily or weekly was associated with lower likelihood of multimorbidity (Sakib et al. 2019).
Conclusion
There have been only longitudinal and cross-sectional studies assessing the association between dietary factors and the risk of multimorbidity. It seems that high consumption of vegetables and fruits has been a protective factor against multimorbidity. Other elements such as nutrients, foods, food groups, and dietary patterns have not been studied with multimorbidity.
Table 3. Summary of studies about the association of dietary factors and multimorbidity
Author Participant/
Sample size
Study aim Main findings
Longitudinal studies
Ruel et al. 2014 1020 Chinese men and women (36-62 years).
Follow up: 5 years
Examine the association of foods, macronutrients, and micronutrients with the development of multimorbidity (≥2 conditions among 11 health conditions).
The likelihood of multimorbidity seemed to lower in those who had higher intake of fruits, vegetable and whole grain products.
Wikström et al. 2015 32,972 Finnish men and women (25-64 years). Follow up: 10 years
Explore various clinical and lifestyle factors on the evolution of multimorbidity (≥2 conditions among 5 health conditions) among healthy people and those with diabetes or CVD.
Inadequate intake of fruit and vegetable increased the risk of multimorbidity in patients with CVD at baseline.
Dhalwani et al. 2017 5,476 English men and women (≥50 years)
Investigate the association of five lifestyle factors with incident multimorbidity (≥2 conditions among 18 health conditions).
Low consumption of fruit and vegetable increased 65% risk of multimorbidity in women, in contrast, it would reduce 40% the risk in men.
Olaya et al. 2019 1699 Spanish men and women (≥65 years).
Follow up: 6 years
Assess the association of fruit and vegetable intakes and time to dead as well as potential moderators including
High consumption of fruit and vegetable increased the probability of surviving among older adults with two chronic conditions, but no association for those who had none, 1 or ≥3 chronic conditions.
multimorbidity (≥2 conditions among 7 health conditions).
Cross-sectional studies
Fortin et al. 2014 1,196 Canadian men and women (≥45 years)
Determine association of lifestyle factors with multimorbidity (≥2 conditions among 14 health conditions).
The risk of multimorbidity was not associated with not meeting the recommendation on fruit and vegetable, physical activity, or alcohol.
Olivares et al. 2017 1044 Argentinian men and women (≥18 years)
Investigate risk factors for chronic diseases and multimorbidity (≥2 conditions among 10 health conditions) in a primary healthcare context.
Low intake of fruit and vegetables is a risk factor for multimorbidity.
Sakib et al. 2019 29,841 Canadian (45–
64 years)
Determine prevalence of multimorbidity (≥3 conditions among 27 health conditions) and its association with lifestyle factors.
Moderate consumption of daily or weekly alcohol lowered the risk of multimorbidity.
CVD: cardiovascular disease
3 AIMS OF THE STUDY
The primary aim of this study was to investigate the association of MED and BSD with multimorbidity in the elderly women from a randomized trial of the OSTPRE-FPS: Kuopio Osteoporosis Risk Factor and Prevention Study. The secondary aim was to examine the links between various food groups and the presence of multimorbidity in the study population.
Finally, we also did a separate analysis to explore the multimorbidity patterns among the participants.
4 METHODOLOGY
4.1 Study design and data collection
This study is based on the Osteoporosis Risk Factor and Fracture Prevention Study (OSTPRE- FPS) which was a 3-year randomized trial conducted by Kärkkäinen et al (2010) started in 2003 in Kuopio, Finland. Its initial aim was to examine the effect of vitamin D and calcium supplementation on preventing of falls and fractures among postmenopausal women (Kärkkäinen et al. 2010). OSTPRE-FPS was a part of the original study, the Osteoporosis Risk Factor and Prevention (OSTPRE) Study which started in 1989 in Kuopio region by the Kuopio Musculoskeletal Research Unit – KMRU (Kärkkäinen et al. 2010). OSTPRE-FPS included 3,432 participants; however, only 750 of these women were selected randomly as subsample for detailed measurements. After that, they were asked to fill in a 3-day food record, and in total there were 554 valid food records returned.
This present study used data at the baseline in a cross-sectional setting which includes 554 participants with valid food records as mentioned above. The participant flow diagram of the OSTPRE-FPS is presented in Figure 1.
Eligible subjects of OSTPRE cohort (n = 5407)
1975 Excluded
701 Did not return enquiry 962 Not willing to participate 312 Did not meet inclusion criteria
3432 Randomized
1718 to calcium + vitamin D group 1714 to control group
750 Randomly selected subsample 375 to both groups
237 Withdrew after randomization 132 From intervention group 132 Withdrew consent 105 From control group
83 Withdrew consent 15 Died before start 7 No contact information
313 Allocated to control group
290 Allocated to calcium + vitamin D group and received allocated intervention
554 women returned the valid food record
Figure 1. Osteoporosis Risk Factor and Fracture Prevention Study (OSTPRE-FPS) participant flow diagram (Adapted fromKärkkäinen et al. 2010).
4.2 Measurements of multimorbidity and dietary intake
4.2.1 Dietary and alcohol intakes
A food record, which included information of 3 consecutive days with one day during the weekend and 2 others during weekdays, was used to evaluate the dietary intake from participants at the baseline. Nutrica program from Finnish social insurance Institute was used to assess the consumption of foods and nutrient intakes (Järvinen et al. 2012). In addition, a self-administrated questionnaire and the instructions were sent to participants beforehand to assess the alcohol usage and other details. Participants were also asked to bring the filled questionnaire on the clinical visiting day (Kärkkäinen et al. 2010).
4.2.2 Mediterranean dietary pattern
Mediterranean Diet score was used to assess the conformance with the MED. The MED has eight components encompass five positive components, moderate alcohol intake (5-25g/day), and two negative components. For the five positive components, one point was given for those with the intake of at least equal the median; otherwise zero. These components represent healthy food choices, and are determined by high intake of (1) root and vegetable based foods, nuts and legumes, mushrooms (not potato); (2) fruit; (3) potatoes or cereal; (4) fish and (5) (PUFA + MUFA): SFA ratio (as surrogate of quality of dietary fat). Moreover, those had (6) moderate alcohol intake were also given one point; otherwise zero. The two negative components (7) total meat and (8) milk and dairy products were given points in a reverse way compared to the positive ones (Isanejad et al. 2018). The score of eight indicated the highest level of conformance with the MED. Construction of the MED score is presented in Table 4.
4.2.3 Baltic sea dietary pattern
Based on the Baltic sea diet score which was first created by Kanerva et al. (2014b) and slight modifications from previous study (Isanejad et al. 2018), the final BSD score included a total of nine components (foods or food groups, nutrients and alcohol consumption (g/d)). These components were divided into two groups: positive and negative. The positive components were (1) total fruits, and berries; (2) vegetables (root vegetables, legumes, nuts, mushrooms and vegetable products—potatoes excluded); (3) fiber from total cereal products; (4) total fish
intake; (5) low-fat milk (skim milk and milk with fat content less than 2%) and (6) ratio of PUFA to saturated fatty acid (SFA) indicating quality of fat intake (Isanejad et al. 2018). The negative components of the BSD score were (7) processed meat products (sausage) and (8) total fat intake expressed as a percentage of total energy intake (Isanejad et al. 2018). Intakes of all components were divided into quartiles. The positive components received zero point for the lowest quartile and three for the highest quartile. In contrast, points were given in a reverse direction for the negative components. According to the BSD, it is healthy to consume zero or a small amount of alcohol (ninth component); therefore, an intake of ≤12 g/d (1 portion) was given one point; otherwise zero. The range of BSD score was from 0 to 25; the higher score, the better the conformance with the BSD. Construction of the BSD score is presented in Table 4.
Table 4 Construction of the Baltic Sea diet and Mediterranean diet scores (adapted from (Isanejad et al. 2018).
Components Scoring
Mediterranean diet score components Vegetables: root vegetables, legumes and nuts, mushrooms, and vegetable products (g/d)
≥Median intake =1, <median intake=0
Total fruits (g/d) ≥Median intake =1, <median intake=0
Total cereals and potatoes (g/d) ≥Median intake =1, <median intake=0
Fish (g/d) ≥Median intake =1, <median intake=0
Ratio of PUFA+ MUFA: SFA ≥Median intake =1, <median intake=0 Total meat including sausage, and eggs (g/d) ≥Median intake =0, <median intake=1 Total milk and dairy products (g/d) ≥Median intake =0, <median intake=1
Alcohol (g/d) 5–25 g/d =1, <5 or >25 g/d=0
Baltic Sea diet score components
Total fruits and berries (g/d) Q1= 0, Q2=1, Q3=2, Q4=3 Vegetables: root vegetables, legumes and nuts,
mushrooms, and vegetable products (potato excluded) (g/d)
Q1= 0, Q2=1, Q3=2, Q4=3
Fiber from total cereal products (g/d) Q1= 0, Q2=1, Q3=2, Q4=3
Fish (g/d) Q1= 0, Q2=1, Q3=2, Q4=3
Milk, low fat<2% Q1= 0, Q2=1, Q3=2, Q4=3
Processed meat products, sausage (g/d) Q1= 3, Q2=2, Q3=1, Q4=0
Ratio of PUFA:SFA Q1= 0, Q2=1, Q3=2, Q4=3
Total fat intake energy % Q1= 3, Q2=2, Q3=1, Q4=0
Alcohol (g/d)a ≤12 g/d=1 and otherwise=0
Q: quartile, PUFA: polyunsaturated fatty acid, MUFA: monounsaturated fatty acids, SFA: saturated fatty acids.
a One portion of alcohol was calculated as 12 g.
4.2.4 Multimorbidity
Based on the review of literature and previous study which used the OSTPRE cohort (Afrin et al. 2016), we categorized chronic conditions which appeared in the same participant as: 0-1 condition; 2 conditions; ≥3 conditions. Multimorbidity in our study was defined as having at least 3 chronic health conditions in the same person. The conditions of long-term diseases were self-reported through questionnaire at baseline (2003). There were 20 included chronic medical conditions (hypertension, hypercholesterolemia, coronary heart disease, other heart diseases, cerebrovascular disease, venous thrombosis in leg, pulmonary embolism, diabetes, chronic kidney disease, chronic liver disease, rheumatoid arthritis, epilepsy, asthma or chronic obstructive pulmonary disease, lactose intolerance, hyperthyroidism, hyperparathyroidism, alcoholism, chronic mental disease, osteoporosis and cancer) and any additional conditions reported by participants (Kärkkäinen et al. 2010, Afrin et al. 2016).
4.3 Potential confounding factors
Information and variables were derived from self-administered questionnaires at the baseline including age, smoking status (no current and current smoking). In addition, BMI was calculated from measured weight divided by height squared, with weight was measured with a digital calibrated scale (Phillips, type HF 351/00) and height with a calibrated wall meter (Kärkkäinen et al. 2010).
According to WHO, old people should do exercise at least 2.5 hours per week (WHO 2020).
As a result, physical activity level (PAL) was categorized into two groups: ≥ 2.5h/week and
<2.5h/week. The PAL was obtained through questionnaire by asking what type of activity they did and duration per week in hours (Kärkkäinen et al. 2010).
4.4 Statistical analysis
The statistical analyses were done by using SPSS Statistics for Windows (Version 25.0; IBM Corp, Armonk, NY) and R (Version 3.6.2; R Foundation for Statistical Computing, Vienna, Austria).
MED and BSD scores were reported in quartiles and they were analysed as continuous variables. Morbidity (chronic condition) was categorized into three groups: 0-1 condition, 2 conditions, and ≥ 3 conditions (multimorbidity). Morbidity patterns were identified using R software. After loading the dataset to R software, the disease variables were replaced by the disease name if the subject had that disease; otherwise empty. This allowed us to get for each row (person's answers) with disease names, then n-grams functionality was used to find which diseases appear in a row together. Finally, different parameters were given to n-grams of how many diseases should appear together (2,3,4...) to be counted.
The Kolmogorov-Smirnov test was used to test the normality of distribution, and the skewed variables were transformed by log 10 transformation. Participant characteristics were compared according to BSD score quartiles, MED score quartiles, and morbidity status using One-way ANOVA test and non-parametric test (Kruskal-Wallis) for continuous variables, and Chi-square test for categorical variables. Post-hoc Tukey HSD test was used to check which groups differ from each other.
The association of MED score quartiles and BSD score quartiles with multimorbidity were assessed by multinomial logistic regression in three models: no adjustment in the model 1, model 2 was adjusted for age (years) and energy intake (kcal/d), and model 3 was adjusted for age (years), BMI (kg/m2), energy intake (kcal/d), smoking (no current and current smoking), and physical activity (<2.5h/week and ≥2.5h/week). Additionally, univariate ANOVA test and non-parametric test (Kruskal-Wallis) were used to examine the association of food groups and multimorbidity adjusted for age, BMI, energy intake, smoking, and physical activity. The level of significance was set to p-value less than 0.05 for all the tests, and two-tailed p-values were reported.
4.5 Ethical considerations
All the participants voluntarily took part in the study. They were informed about the purpose and other relevant information of this study. The Kuopio Musculoskeletal Research Unit of the Clinical research center of the University of Kuopio was in charge of different clinical tests and measurements. The study was registered in Clinical trials.gov by the identification NCT00592917 and was approved by the ethical committee of Kuopio University Hospital in October 2001.
