Current Nutrition Reports

, Volume 4, Issue 1, pp 88–101

Diet Quality and Weight Change in Adults Over Time: A Systematic Review of Cohort Studies

Authors

  • Haya Aljadani
    • Priority Research Centre in Physical Activity and Nutrition, ATC Level 3, University DriveUniversity of Newcastle
    • School of Health Sciences, Faculty of Health and Medicine, HA12 Hunter BuildingUniversity Drive, University of Newcastle
    • Faculty of Nutrition and Health ScienceKing Abdul-Aziz University
  • Amanda Patterson
    • Priority Research Centre in Physical Activity and Nutrition, ATC Level 3, University DriveUniversity of Newcastle
    • School of Health Sciences, Faculty of Health and Medicine, HA12 Hunter BuildingUniversity Drive, University of Newcastle
  • David Sibbritt
    • Faculty of HealthUniversity of Technology
    • Priority Research Centre in Physical Activity and Nutrition, ATC Level 3, University DriveUniversity of Newcastle
    • School of Health Sciences, Faculty of Health and Medicine, HA12 Hunter BuildingUniversity Drive, University of Newcastle
Dietary Patterns and Behavior (LM Steffen, Section Editor)

DOI: 10.1007/s13668-014-0115-1

Cite this article as:
Aljadani, H., Patterson, A., Sibbritt, D. et al. Curr Nutr Rep (2015) 4: 88. doi:10.1007/s13668-014-0115-1
Part of the following topical collections:
  1. Topical Collection on Dietary Patterns and Behavior

Abstract

This systematic review examines the relationship between diet quality and weight gain in adults over time and is an update of our previous review of the same topic. The goal was to synthesise the best available current evidence on diet quality and weight change within longitudinal analyses. The inclusion criteria were case-control or cohort studies, and adults aged ≥18 years. The dependent variable was diet quality indexes and the independent variable was any measurement of body weight. The current systematic review identified 16 studies published between 1970 and 2014. Of these, eight were published since our last review. The findings of these recent studies confirm the results of our previous review, that higher diet quality is associated with relatively lower prospective weight gain, as well as a lower risk of becoming overweight or obese, compared with poor diet quality. Across the 16 studies, it appears that the diet quality indexes based on foods alone, or food and nutrient components, are more predictive of weight change. However, further research is needed to confirm this. Additionally, high-quality analyses that assess change in diet quality over time are needed.

Keywords

Diet quality indexWeight gainObesityAdultsCohort studySystematic review

Introduction

There is great interest in measuring overall diet intake using tools that reflect usual eating patterns, rather than focusing on single nutrients, and to evaluate the relationship with various health outcomes in cohort studies [13]. This stems from the acknowledgement of the limitations of single nutrient methods. One of these weaknesses is that people do not consume isolated nutrients; they consume a variety of foods and the resultant mix of nutrients interacts within the body in ways which are not fullyelucidated. Additionally, it is preferable to recommend a dietary pattern to the public than to make recommendations on individual nutrients [14]. One method for measuring dietary patterns is using an “a priori” diet quality index or score that complies with a country’s national dietary guidelines—for example, the Healthy Eating Index (HEI), indexes associated with lower risk of chronic disease such as the Dietary Approaches to Stop Hypertension diet score (DASH score), or scoring methods that align with the Mediterranean dietary pattern (MDP) [3, 5, 6].

Obesity and weight gain are risk factors for adverse health outcomes. Obesity prevalence has been increasing rapidly around the world [7••]. The older epidemiological literature on the relationship between diet and weight change is heterogeneous, particularly in cross-sectional analyses [8]. In 2011, we reviewed the best available evidence on the association between dietary patterns and weight change in adults over time [9, 10]. At that time, limited analyses (n = 8) had been conducted that examined the diet quality indexes in relation to body weight change, with considerable inconsistencies across methodologies and results. Although the evidence was limited and significant heterogeneity was found, we concluded that higher diet quality scores, reflecting greater alignment with national dietary guidelines, were associated with lower weight gain. Hence, the focus of this systematic review is to update the results of our previous review. In this current update we will capture the most recent studies on diet quality and prospective weight changes in adults, tabulate the data, and discuss the results of the included studies. Hence, using any pre-defined diet quality indexes or scores, the aim is to summarise the best available evidence on the association between diet quality and body weight change in adults over time.

Methods

The criteria for including studies in this review were the same as our previous review [9, 10]. Study designs included cohort or case control studies. The participants had to be adults aged ≥18 years. Dietary intake could be assessed by any method, but had to include a pre-defined diet quality or variety index or score. Weight change could be measured by different methods, including weight (kg), body mass index (BMI; kg/m2), waist circumference (WC), or body fat percentage.

Medline, Cinahl, Embase, and Scopus were searched for additional papers published since the 2011 review. As such, our search included literature reviews and original research papers published between 2011 and February 2014. Keywords included “diet quality,” “diet index,” “diet score,” “weight gain,” “BMI,” “obesity,” “adults,” “cohort,” “case-control,” and “prospective.” The quality of the included studies was assessed by two independent reviewers using standardised critical appraisal instruments from the Joanna Briggs Institute Meta Analysis of Statistics Assessment and Review Instrument (JBI-MAStARI, Fig. 1). The assessment included the following criteria: representativeness of the sample; assessment of participants at the same follow-up time; identification of confounding factors; outcome objective; follow-up period; description of withdrawals; measurement of outcome; and statistical analysis used.
https://static-content.springer.com/image/art%3A10.1007%2Fs13668-014-0115-1/MediaObjects/13668_2014_115_Fig1_HTML.gif
Fig. 1

The Joanna Briggs Institute Meta-Analysis of Statistics Assessment and Review Instrument (JBI-MAStARI)

Results

Figure 2 summarises the combined number of studies identified and subsequently included in the reviews for 2011 and 2014. From 3,136 citations identified by the initial and updated searches, a total of 16 studies were included in this review.
https://static-content.springer.com/image/art%3A10.1007%2Fs13668-014-0115-1/MediaObjects/13668_2014_115_Fig2_HTML.gif
Fig. 2

Flowchart of studies included in a systematic review of the relationship between diet quality and weight change in adults over time

Table 1 summarises the characteristics of the included studies, such as demographic characteristics, study aim, the follow-up period, and participant retention. Table 2 describes the dietary intake methods and diet quality indexes used, the main weight change variables, and the results.
Table 1

Characteristics of the studies included

 

First author & publication year

Study design & cohort’s name

Subjects

Country

Study length

Retention

Purpose

A) Diet quality indexes#

 1

Aljadani et al. 2013 [11]

The Australian Longitudinal Study on Women’s Health (ALSWH)

F = 7,155,

Australia

6 years

No =,588

To measure the effect of weight changes during six years of follow-up with baseline ARFS.

Age 48-56

 2

Alajadani et al. 2013 [12••]

The ALSWH

F = 8,239,

Australia

6 years

No = 881

Assessed the relationship between tertiles of diet quality measured by the ARFS, FAVI, and Aus-DQI and changes in weight over the period from 2003 to 2009.

Age 27. 6 ± 1.5

 3

Arabshahi et al. 2012 [13]

Nambour Skin Cancer Study

F, M = 2,399,

Australia

15 years

Not reported

To investigate the relationship between food based dietary index and change in BMI and WC.

Age 25-75

 4

Asghari et al. 2012 [14]

Tehran Lipid and Glucose Study

F, M = 708,

Iran

6.7 years

66 %

To evaluate the association of diet quality indices with BMI and WC after 6.7 years.

Age > 19 yrs

 5

Beunza, et al. 2010 [15]

The Seguimiento Universidad de Navarra (SUN) cohort.

F, M = 10,376

Spain

Mean 5.7 ± 2.2 years

>90 % (for first 24 mo)

The correlation between MDSs and weight gain over time.

Mean age 38

 6

Boggs et al. 2013 [16••]

Black Women’s Health Study

F = 19,885,

America

6 years

Not reported

Assess diet quality in relation to the incidence of obesity.

Age 21-39

 7

Forget et al. 2013 [17]

Cohort study (no name specified)

F, M = 196,

Canada

4 years

Unclear (74 - 90 %)

Assess ideal Healthy Diet Score and its association with weight status over four years.

Age 20.5 ± 3.0

 8

Kimokoti et al. 2010 [18]

Framingham Offspring/Spouse (FOS) study.

F, M = 1,515

USA

16 years

67 %

Evaluate how diet quality compares with demographic, anthropometric, biological, clinical, and other lifestyle factors in predicting weight change in our participants; and impact of these factors, including smoking status, on the association between diet quality and weight change.

Age ≥30

 9

Lassale et al. 2012 [19••]

SU.VI.MAX study

F, M = 3,151

France

13 years

Not reported

1) Assess association between dietary scores and 13-year weight change

Age 45-60 at baseline

2) Assess the 13-year risk of becoming obese in non-obese participants (at baseline) and the association with dietary scores.

 10

Mendez et al. 2006 [20]

The European Prospective Investigation into Cancer and Nutrition (EPIC)-Spain.

F = 27,827

Spain

Mean 3.3 years

95 %

To examine whether adherence with MDP is associated with obesity incidence for three years of follow up.

Age 29-65

 11

Quatromoni et al. 2006 [21]

FOS study.

M, F = 2,245

USA

8 years

58 %

The relationship between eight-year weight change among FOS participants and adherence to the fourth edition of the Dietary Guidelines measured using DQI.

Age 49-56

 12

Romaguera et al. 2010 [22]

The European Prospective Investigation into Cancer and Nutrition–Physical Activity, Nutrition, Alcohol Consumption, Cessation of Smoking, Eating Out of Home, and Obesity (EPIC- PANACEA) project.

F, M = 373,803

Denmark, France, Germany, Greece, Italy, Netherlands, Norway, Spain, Sweden, and UK.

Median 5 yrs

Unclear

The association between adherence to MDP and weight change and incidence of overweight or obesity.

Age 25-70

 13

Sanchez-Villegas, et al. 2006 [23]

The Seguimiento Universidad de Navarra (SUN) cohort.

F, M = 6,319

Spain.

28 mo

90 %

The association between adherence to the MDP and weight or BMI change.

Mean age 34-40 yrs

 14

Wolongevicz, et al. 2010. [24]

FOS Study.

F = 590

USA

16 yrs

100 %

The relationship between diet quality and the development of overweight or obesity in women over 16 years in healthy, normal-weight (BMI , 25 kg/m2) women.

Aged 25-71

 15

Yannakoulia, et al. 2008 [25]

The ATTICA study that is a health and nutrition survey that is being carried out in the province of Attica.

M, F = 3,031

Greek

5 yrs

Not reported

To investigate potential effect of several socio-demographic and lifestyle habits on the incidence of obesity in a sample of CVD-free with normal and overweight adults at baseline during five years of follow-up.

Aged >18 yrs.

 16

Zamora et al., 2010 [26••]

Coronary Artery Risk Development in Young Adults (CARDIA)

M, F = 4,913, 18-30 yrs

America

20 years

72 %

Examine association between diet quality and the 20 yr risk of weight gain

#The Diet Quality indexes data adapted and updated from: [10] Aljadani H, Patterson A, Sibbritt D, Collins CE. The association between diet quality and weight change in adults over time: A systematic review of prospective cohort studies. (In: Diet quality: an evidence-based approach. Editor Victor Preedy. King’s College London. In Press January 2012, with permission from Springer; and [9] Aljadani H, Patterson A, Sibbritt D, Collins CE. The association between dietary patterns and weight change in adults over time: a systematic review of studies with follow up. The Joanna Briggs Institute's Database of Systematic Reviews and Implementation Reports. 2013;11.

Table 2

Dietary intake methods and body weight outcomes

 

Author & year

Dietary intake method

Diet quality tool /How the score is derived using the diet quality tool

Weight outcome

Results

A)Diet quality indexes#

 1

Alajdani et al. (2013) [11]

The Dietary Questionnaire for Epidemiological Studies Version 2 (DQES v2) FFQ.

ARFS contains seven main groups with different sub-scales.

• Vegetables

• Fruit

• Protein

• Grain

• Dairy

• Fat

• Alcohol

CWt

The ARFS shows no relationship with weight changes in this group of women. P > 0.05

 2

Aljadani et al. (2013) [12••]

( DQESv2) FFQ.

The ARFS as described above.

The FAVI is contains two sub-scales with top points 333 down to zero:

• The fruit sub-scale, which contains 13 items,

• The vegetable sub-scale which contains 24 items

Consumption frequency of all fruit and vegetable items was scored using the full range of the FFQ Likert scale from zero to nine, with “never” scored as zero and “≥3 times per day” scored as nine points.

The Aus-DQI:A person gives one point for each of:

• total fat <35 % kJ

• saturated fat ≤7 % kJ

• carbohydrate ≥45 % kJ

Sodium <2300 mg/d

CWt

The ARFS and FAVI were significantly associated with lower weight gain in the sub-sample of the plausible TEI young age women in the fully adjusted model. Those who achievied the highest score in FAVI (the top tertile) gained the lowest weight compared with last tertiles ( = − 1 . 6 , CI: −2.4 to −0.3, P = 0 . 0 1 ). Also those who were in the top tertile of ARFS gained less weight than those who were in the last tertile ( = − 1 . 6 kg (95 % CI: −2.67 to −0.56), P = 0 . 0 0 3).

 3

Arabshahi et al. 2012 [13]

1992 & 1996 – 129 item FFQ. In 2007 – 151 item FFQ.

DGI, consists of 11 items and each scored from zero to ten points which included the following;

1. Vegetables and legumes

2. Fruit

3. Total cereals

4. Meat and alternatives

5. Total dairy

6. Diet variety

7. Saturated fat

8. Alcoholic beverages

9. Added sugars

10. Extra foods

11. Proportion of lean meat relative to total meat

CBMI, CWc

1)Men with a higher DGI score gained less BMI and Wc over time. Those in the highest quartile of DGI (86.4 to 122.9 scores) had the lowest gain in BMI compared with those in the lowest quartile with scores 22.2 to 62.3 point of DGI, (0.05v, 0.11 kg/m2/year, with 95 % CI: (0.00, 0.09 ) and (0.06, 0.16); p = 0.01) and in Wc (0.04v, 0.26 cm/year, p = 0.04).

2)In women, the DGI score was not associated with any change in anthropometric measures.

 4.

Asghari et al. 2012 [14]

2 x 24 hr recall, at approx. 10 day interval.

• HEI-2005 developed by Guenther (2008) [27] but omitting alcohol and simplified the scoring of sodium

• DQI-I, four aspects of healthy diet including variety, adequacy, moderation and balance

MDS proposed by Trichopoulou [28] but with alcohol item removed and divided the grain into whole grains and refined grain. Processed meat and red meat were into one group. Polyunsaturated fatty acids (PUFAS) was substituted for monounsaturated fatty acids.

CBMI, CWc

1)The change of HEI-2005 associated with lower BMI during the follow-up period. (-0.022 ± standard error 0.011) with p = 0.043.

2)The change of MDS, and DQI-I were significantly associated with changes in BMI and WC.

 5

Beunza et al. 2010 [15]

FFQ with 136 food item.

MDS. There are nine food items: The amount intake of these food are for M and F respectively: (1) Vegetables 550 and 500 (g/d) (2) Fruit and nuts 360 and 360(g/d) (3) Legumes 9 and 7(g/d) (4) Cereals, bread and potatoes 180 and 140(g/d) (5) Ratio of monounsaturated fatty acids to saturated fatty acids 1.7 and 1.7(g/d) (6) Moderate alcohol (10-15 g alcohol /d)for M and 50-25 g for F. (7) Fish 24 and 19 (g/d) (8) Meat and poultry120 and 90 (g/d) (9) Dairy products 200 and 190(g/m)

1) Annual CWt

2) Incidence of wt gain ( ≥3 or ≥5 kg during the first 2 or 4 years of follow up.

1) Participants in the highest tertile of MDS had the lowest average yearly wt gain relative to lowest tertile of MDS - .059 kg/y; 95 % CI: (-0.111, -0.008 kg/y).

2) There were significant inverse association between all MDS and wt change

OR for incident of wt gain for highest tertile relative to lowest tertile:

OR; CI 95 % ( 2 yr) for ≥ 3 kg: 0.8(07, 0.9)

OR:CI 95 %(2 yr)for : ≥5 kg. 0.76; CI:0.62-0.92

OR; CI 95 % ( 4 yr) for ≥ 3 kg: : 0.80; CI:0.71-0.91

OR:CI 95 %(4 yr)for : ≥5 kg. : 0.76; CI:0.64-0.9

3) Those in the highest tertile of MDS were less likely to have absolute wt gain relative to those in the lowest tertile.

A maximum of 9 points total from the last components.

 6

Boggs et al. 2013 [16••]

Self-administered modified version of the Block-National Cancer Institute FFQ

AHEI-2010, 11 food and nutrient items, each scored 0 – 10 where 10 shows compliance to dietary recommendations.

1. Vegetables

2. Fruits

3. Whole grains

4. Nuts and legumes

5. Long chain (omega 3) fatty acids (EPA + DHA) and PUFAS

6. Sugar-sweetened beverages and fruit juice

7. Red and processed meats

8. Trans fat

9. Sodium

10. Alcohol

DASH contains eight item with maximum 80 points:

Fruits and fruit juice

Vegetables

Nuts and legumes

Whole grain

Low fat dairy

Sodium

Red and processed meat

Sweet sugar beverages

CWt, Incidence of obesity.

1)Overall analysis, it found no relationship between AHEI-2010 and DASH during the follow-up period.

2)In analysis which measured the diet quality scores at two occasions and considered the initial BMI, it found consistently higher quality diet scores were inversely associated with obesity incidence among women with a normal BMI (18.5 – 24.9 kg/m2) at baseline. When comparing highest with lowest quintiles of AHEI-2010 in this group the HR = 0.76 (CI 95 %: 0.58, 0.98) vs 0.68 (CI: 0.53, 0.88).

3)No significant association for women who were overweight (BMI 25-29.9 kg/m2) at baseline. HR 0.85 (CI 95 %: 0.67, 1.07).

 7

Forget et al. 2013 [17]

3 day food dairy (2 weekdays, 1 weekend day)

Healthy Diet Score (HDS), 5 components

1. ≥4.5C F/V per week

2. ≥2 serves fish/week

3. ≥3 serves fibre rich whole grains/day

4. <2300 mg sodium/day

5. ≤36 oz sugar sweetened beverages/ week

CBMI

Participants with a poor HDS (0-1/5) at baseline were more likely to gain more weight by 0.70 kg/m2 over the 4 years compared with intermediate or ideal scores (p = 0.03).

 8

Kimokoti et al. 2010 [18]

3 day estimated food records

Framingham Nutritional Risk Score (FNRS19-nutrients: (1) Total energy(kJ) (2) Protein (% energy) (3) Total fat (% energy) (4) Monounsaturated fat (% energy), (5) Saturated fat (%energy) (6) Alcohol (% energy) (7) Cholesterol (mg/4184 kJ), (8) Sodium (mg/4184 kJ) (9) Carbohydrate (%energy), (10) Polyunsaturated fat (% energy) (11) Fiber (g/4184 kJ), (12) Calcium (mg/4184 kJ) (13)Selenium (ug/4184 kJ) (14) Vitamin C (g/4184 kJ) (15) Vitamin B-6 (g/4184 kJ) (16) Vitamin B-12 (ug/4184 kJ) (17) Vitamin E (g/4184 kJ) (18) Folate (ug/4184 kJ) (19) Beta-carotene (ug/4184 kJ).

CWt

FNRS was not associated with weight gain in both M and F; p = 0.16 and p = 0.61 respectively.

F who were former smokers and who were in the lowest tertile of FRNS gained an additional 5.2 kg compared with former smokers who were in the highest tertile of FNRS; P for trend = 0.03. FNRS was not a predictor for wt gain either in M or F.

 9

Lassale et al. 2012 [19••]

24 hr recall

15 – point PNNS-GS with top 15 points. Eight items are related to food servings including fruits, vegetables, starchy food, whole grain, dairy products, meat, seafood, added fat and vegetables fat. Four related to moderation in consumption, including sweet, salt, water and soda, alcohol. Additional one point to physical activity. Negative points deducted for excessive consumption of salt, sweets and when energy intake exceeds the needed energy level by more than 5 %.

DQI-I adapted for French populations from DQI-I developed by Kim (2003) [29] with 4 components (variety, adequacy, moderation and balance)

DGAI, range 0-20 and divided into 2 sets of components (11 food including dark green vegetables, orange vegetables, and all grains) and 9 healthy choices for example consuming >50 % of grains as whole grains)

MDS developed by Trichopoulou with range 0-9, based on 9 components (intake of grains, veg, fruit and nuts, milk and dairy, meat, legumes, alcohol, fish ratio of monounsaturated fatty acids to saturated fat.

The rMED, 0 – 18 and based on the same components as MDS except MUFA and saturated fat ratio is replaced with olive oil consumption.

MSDPS, 13 components, whole grain, cereals, fruit, veg, dairy, wine, fish, poultry, olives, legumes, nuts, potatoes, eggs, sweets, meats and olive oil.

CWt, BMI

1)In men (n = 1680), there was a statistically significant inverse association between all diet quality index scores, except for the MSDPS, and weight change and the risk of obesity. More specifically, a higher dietary index score of one standard deviation resulted in a lower weight gain during the follow up period, ranging from: 0.40 (95 % CI; -0.71 to -0.1) for PNNS-GS, to 0.87 (95%CI -1.15 to -0.58) for rMED.

2)The risk of becoming obese was strongly reduced for men in the top quartile compared to the bottom quartile with: OR = 0.20, CI 95 %; 0.10, 0.42 for the DGAI. OR = 0.45, CI 95 %; 0.26, 0.79 for the PNNS-GS and OR = 0.45, CI 95 %; 0.27, 0.76 the MDS.

3)In women, there was no association between dietary scores and obesity risk.

 10

Mendez et al. 2006 [20]

Diet history with >600 items

MDS. A maximum of 8 points derived from the following components: (1) Fish (g/MJ) (2) Vegetables(g/MJ) (3) Fruits and nuts (g/MJ) (4) Legumes(g/MJ) (5) Cereals(g/MJ) (6) The ratio of monounsaturated saturated fat(g/MJ) (7) Moderate ethanol intakes, defined as 5–25 g /d for women and 10–50 g/d for men. (8) Meat (g/MJ)

CWt in mean of follow-up 3.3 years

1)Participants in the highest MDS were less likely to be obese among overweight subjects. OR ; ( 95 % CI) of becoming obese for F and M respectively.

0.69, (0.54–0.89)

0.68, (0.53–0.89).

2) High MD adherence was not associated with overweight incidence in women or men.

OR; ( 95 % CI) of becoming overweight for F and M respectively. 0.99(0.78, 1.25).

1.11(0.81-1.52).

 11

Quatromoni et al. 2006 [21]

3 day estimated food records

Diet quality index (DQI) 5-point DQI to assess adherence to key US dietary recommendations.

One DQI point was contributed for each of five nutrients if

(1) Total fat ( <30%kcal) (2) Saturated fat ( <10 % kcal) (3) Carbohydrate ( >50 % kcal)]. (4) Cholesterol ( <300 mg/d) (5) Sodium ( <2400 mg/d)

DQI scores ranged from 0 to 5

CWt in 8 years

1)Participants in the highest DQI quintile gain less wt over 8 years; (P for trend <0.01). The mean ± SD of wt gain in F in the highest DQI quintile was 3.3 ± 17.4 Ib compared with 8.0 ± 13.0 Ib gained in those within the lowest DQI quintile. The mean ± SD of wt gain in M in the highest DQI quintile was 2.7 ± 10.1 Ib compared with 5.1 ± 13.3 Ib those in the lowest DQI quintile.

2)There is significant inverse association between the mean of DQI score for all participant and weight change in both adults; p = 0.025 and p = 0.008 respectively for M and F.

 12

Romaguera et al. 2010 [22]

FFQ

A revised MDS (rMDS). There is 11 component of this score as follow: ( 1) Vegetables, (2) Legumes (3) Fruit and nuts, (4) Cereals (5) Fish and seafood (6) Olive oil, (7) Moderate alcohol consumption (8) Meat and meat products (9) Dairy products

Each part measured by g/1000 kcal to express ED

CWt and Incidence of overweight

1) Those in the highest tertile of rMDS had less wt gain -0.16 kg(CI 95 %: -0.24, -0.07) and -0.04 kg(-0.07, -0.02) for the combined results.

2)Overall results showed that a 2-point increased in the rMDS was associated with becoming overweight; OR (CI 95 %) = 0.97 (0.95, 0.99)

 13

Sanchez-Villegas et al. 2006 [23]

FFQ with 136 food item

MDS. A maximum of 30 points total from the components: (1) Cereals (g/d) (2) Vegetables ( g/d) (3) Fruits (g/d) (4) Legumes (g/d) (5) Fish (g/d) (6) Nuts (g/d) (7) Olive oil (g/d) (8) Moderate red wine consumption (g/d) (9) Meat and meat products (g/d) (10) Whole-fat dairy products (g/d).

1) Annual CWt and CBMI

1)The highest quartile of MDS gained less weight (0.65 kg)(0.0.59, 0.8) compared with those in the lowest quartile

P for trend 0.291

2) Participants at the highest quartile of MDS gained smaller in BMI 0.23 (0.12, 0.33) compared with those in lowest quartile of MDS. P for trend 0.279.

 14

Wolongevicz et al. 2010 [24]

3 day dietary record

Framingham Nutritional Risk Score (FNRS) 19-nutrients: (1) Total energy (kj) (2) Protein (% energy), (3) Total fat (% energy) (4) Monounsaturated fat (% energy), (5) Saturated fat (% energy), (6) Alcohol (% energy) (7) Cholesterol (mg/4184), (8) Sodium ( mg/4184kj), (9) Carbohydrate (%energy), (10) Polyunsaturated fat (% energy), (11) Fiber (g4184kj), (12) Calcium (mg/4184kj) (13) Selenium (ug/4184kj) (14) Vitamin C (g/4184kj) (15) Vitamin B-6 (g/4184kj). (16) Vitamin B-12 (ug/4184kj) (17) Vitamin E (g/4184kj). (18) Folate (ug/4184kj). (19) Beta-carotene (ug/4184kj).

Incidence of overweight or obesity; BMI ≥25 km/m2.

FNRS was associated with incidence of overweight or obesity (p for trend =0.009). Women with lower diet quality were significantly more likely to become overweight or obese; OR: 1.76; CI: 95 %( 1.16 – 2.69) times compared with those with highest diet quality.

Higher; FNRS was associated with a reduce the risk of being overweight or obese.

 15

Yannakoulia et al. 2009 [25]

EPIC-Greece questionnaire, a validated semi-quantitiave FFQ

MDS ranged from 0 to 55 based on M patterns pyramid. The patterns are consistence with:

a) Daily consumption of non-refined cereals and products (like whole grain bread, pasta, rice, etc., 8 servings/day), vegetables (2–3 servings/day), fruits (4–6 servings/day), olive oil (in daily cooking as the main added lipid) and non fat or low-fat dairy products (like cheese, yoghurt, and milk, 1–2 servings/day),

(b) Weekly consumption of potatoes (4–5 servings/week), fish (4–5 servings/week), olives, pulses and nuts (>4 servings/week) and more rare poultry (1–3 servings/week), eggs and sweets (1–3 servings/week) and

c) Monthly consumption of red meat and meat products (4–5 servings/month).

d) Moderate consumption of wine (1–2 wineglasses/day), which usually accompanies meals.

e) Milk consumption is moderate, while of cheese and yogurt consumption is high [30].

Developing obesity

MDS was no related to the risk of developing obesity during the following-up period for those who were normal weight or over weight at baseline.

OR: 0.98, (95 %:CI: 0.95, 1.03), p = 0.51

 16

Zamora et al. 2010 [26••]

Diet history

2005 DQI, 10 components with zero to 100 points:

- 3 addressed total fat, saturated fat and cholesterol

- 4 quantified adequate intake of reduced fat dairy, fruit, veg and grain.

- 3 addressed variety of intake, reducing consumption of energy dense, nutrient poor foods and sodium

CWt, BMI

1)For the Caucasian participants with BMI <25 at the start of the study, there was a significant inverse association between diet quality score and the risk of gaining ≥10 kg over the follow-up period. Each 10 point increased in DQI, associated with 10 % lower risk of gaining ≥10 kg of weight during 20 years.

2) While amongst African American obese participants, it found that those with healthiest diet were gaining significant greater weight. Increased 10 points of DQI associated with a 15 % higher risk of gaining 10 kg or more during the same follow-up period.

Wt: weight, CWt changes in weight during follow-up, CWc: changes in waist circumferences during follow-up, EI energy intake, BMI body mass index, FFQ food frequency questionnaire, ED energy density, PA physical activity

#The Diet Quality indexes data adapted and updated from: [10] Aljadani H, Patterson A, Sibbritt D, Collins CE. The association between diet quality and weight change in adults over time: A systematic review of prospective cohort studies. (In: Diet quality: an evidence-based approach. Editor Victor Preedy. King’s College London. In Press January 2012, with permission from Springer; and [9] Aljadani H, Patterson A, Sibbritt D, Collins CE. The association between dietary patterns and weight change in adults over time: a systematic review of studies with follow up. The Joanna Briggs Institute's Database of Systematic Reviews and Implementation Reports. 2013;11.

Methodological Quality

Table 3 summarises the methodological quality of the included studies as assessed by the JBI-MAStARI tool. The studies were generally of good quality, and 12 of the 16 had low risk of bias, meaning that they met at least six of the eight criteria. Studies that were included used appropriate statistical tests, identified the main confounders in the analysis, and followed the participants over a significant period of time (one year or more).
Table 3

Quality assessment of studies included:

 

Authors

Is the sample representative of patients in the population as a whole?

Was everyone assessed at the same follow-up time?

Are the confounding factors identified and strategies to deal with them stated?

Are outcomes assessed using objective criteria?

Was follow up carried out over a sufficient time period?

Were the people who withdrew described and included in the analysis?

Were the outcomes measured in a reliable way?

Was appropriate statistical analysis used?

Ranking

A) Diet quality indexes#

 1

Aljadani et al. 2013 [11]

Y

Y

Y

N

Y

Y

Y

Y

7Y 1 N

 2

Aljadani et al. 2013 [12••]

Y

Y

Y

N

Y

Y

Y

Y

7Y I N

 3

Arabshahi et al. 2012 [13]

Y

Y

Y

Y

Y

N

Y

Y

7Y 1 N

 4

Asghari et al. 2012 [14]

Y

Y

Y

Y

Y

N

Y

Y

7Y 1 N

 5

Beunza et al. 2010 [15]

N

N

Y

N

Y

U

Y

Y

4Y 2N IU

 6

Boggs et al. 2013 [16••]

Y

Y

Y

N

Y

N

N

Y

5Y 3N

 7

Forget et al. 2013 [17]

Y

Y

U

Y

Y

U

Y

Y

6Y 2U

 8

Kimokoti et al. 2010 [18]

U

Y

Y

Y

Y

U

Y

Y

6Y 2U

 9

Lassale et al. 2012 [19••]

Y

Y

Y

Y

Y

N

U

Y

6Y I N 1 U

 10

Mendez et al. 2006 [20]

Y

Y

Y

N

Y

U

Y

Y

6Y 1 N 1U

 11

Quatromoni et al. 2006 [21]

U

Y

Y

Y

Y

U

Y

Y

6Y 2U

 12

Romaguera et al. 2010 [22]

N

N

Y

N

Y

U

N

Y

3Y 4N 1U

 13

Sanchez-Villegas et al. 2006 [23]

N

Y

Y

N

Y

U

Y

Y

5Y 2N 1U

 14

Wolongevicz et al. 2010 [24]

U

Y

Y

Y

Y

U

Y

Y

6Y 2U

 15

Yannakoulia et al. 2009 [25]

Y

Y

Y

Y

Y

Y

Y

Y

8Y

 16

Zamora et al. 2010 [26••]

Y

Y

Y

Y

Y

U

Y

Y

7Y 1U

N: no, Y: yes, U: unclear.

#The Diet Quality indexes data adapted and updated from [9] Aljadani H, Patterson A, Sibbritt D, Collins CE. The association between dietary patterns and weight change in adults over time: a systematic review of studies with follow up. The Joanna Briggs Institute's Database of Systematic Reviews and Implementation Reports. 2013;11.

Results of the Review

Diet Quality Indexes

The 16 studies included in the review used various methods to measure dietary intake, diet quality, and body weight outcomes. Among these, five studies used different versions of the scoring methods that align with the MDP [15, 20, 22, 23, 25]; two studies [18, 24] used the Framingham Nutritional Risk Score (FNRS); two used the Australian Recommended Food Score (ARFS) [11, 12••]; one used the USA five point scale of diet quality index (DQI) [21]; one used the Aus-DQI [12••]; one used the Australian Dietary Guideline Index (DGI) [13]; one used the Healthy Diet Score (HDS) [17]; one used the 100-point Diet Quality Index (2005 DQI) [26••]; and one used the Fruit and Vegetables Index (FAVI) [12••].

Regarding the variety of indexes used to measure the quality of diet, four studies [14, 15, 16••, 19••] used at least two indexes to measure the quality of dietary intake. Asghari et al. [14], for example, used three indexes in an Iranian population, including the following: the Mediterranean Diet Scale (MDS), the Healthy Eating Index-2005 (HEI-2005), and the Diet Quality Index-International (DQI-I). Another example of using more than one index is the study conducted by Lassale et al. [19••] that used six different tools to assess the quality of dietary intake, including the 15-point French Programme National Nutrition Santé-Guideline score (15-point PNNS-GS), the American DGI, DQI-I, MDS, the relative MDS (rMED) and the Mediterranean Style Dietary Pattern Score (MSDPS).

The Main Outcome of the Studies

In the previous review, we found the relationship between diet quality and weight status was reported over varying time periods. For example, Quatromoni et al. [21] examined the relationship over 8 years, Kimokoti et al. [18] reported over 16 years, and Mendez et al. [20] reported over 3.3 years of follow-up. In addition, the method by which weight was reported varied, with some studies reporting the development of overweight or obesity during the follow-up period [22, 24, 25], while others reported the change in body weight annually [15, 23].

Two studies [15, 21] found that those who reported higher dietary quality scores, as measured by DQI, had lower weight gain over time, compared with those who had lower diet quality scores (in the lowest quintile). Wolongevicz et al. [24] found that those with the lowest diet quality scores were more likely to become overweight or develop obesity during follow-up compared to those with higher or ‘healthy’ diet quality index scores. In contrast, two studies [18, 25] found that there was no relationship between diet quality and any change in weight status during follow-up.

Sanchez-Villegas et al. [23] used six different versions of the MDS index. They found that higher diet quality, as assessed using four versions of the MDS, was associated with lower weight gain over time. However, this was not the case for the other two versions. Romaguera et al. [22] found that for participants who were overweight at baseline, the risk of becoming obese at follow-up was greater if their diet quality was also poor. However, they found no relationship with becoming overweight amongst those who were in the healthy weight range at baseline [22].

The eight additional studies included in the updated review had varying conclusions. Arabshahi et al. [13] examined diet quality as per the DGI and measured changes in BMI and waist circumference during 15 years of follow-up in Australian men and women. Findings differed by gender, with no association between diet quality score and change in weight or BMI in women. However, in men there was a significant association between diet quality and change in BMI. In the fully adjusted model, those who achieved the highest DGI scores (fourth quartile) had a significantly lower annual BMI increase (mean = 0.05 kg/m2 with 95 % CI: (0.00, 0.09) compared to those with poor dietary intake (first quartile) (mean = 0.11 kg/m2, with 95 % CI: (0.06, 0.16). This model was adjusted for baseline age, education, smoking status, alcohol consumption, physical activity, and baseline WC.

Asghari et al. [14] used three diet quality indexes—the MDS, HEI-2005, and DQI-I—to evaluate the association between change in diet quality and changes in BMI during 6.7 years of follow-up. They found that higher HEI-2005 scores were associated with statistically significant lower BMI (-0.022 ± SE 0.011) over the follow-up period (p value = 0.043). However, neither the MDS nor DQI-I were associated with changes in BMI in this groups of adults (p > 0.05).

Boggs et al. [16••] estimated the hazard ratio (HR) for developing obesity in relation to baseline diet quality over a period from 1995 to 2011, as evaluated using two diet quality indexes, the AHEI-2010 and the DASH. The study recruited 19,885 African American women aged 21-39 years. Overall, there was no relationship between baseline diet quality and weight changes during the follow-up. However, in a sub-analysis considering the initial BMI and changes in diet quality scores at two points (1995 and 2001), different results appeared. The authors found that those who were of normal weight at baseline (BMI: 18.5 - 24.9 kg/m2) and maintained a higher diet quality score during follow-up, had significantly lower risk of developing obesity compared with those who had a lower diet quality score over time. The HRs, when comparing the highest to the lowest quintiles of AHEI-2010 and DASH, respectively, were 0.76 (95 % CI: 0.58, 0.98) and 0.68 (95 % CI: 0.53, 0.88). However, among women who were overweight at baseline, there was no relationship between diet quality and obesity risk.

Forget et al. [17] reported on a study of 196 young adults that evaluated the association between the HDS and changes in BMI over four years. The study found that those with a poor diet quality score (HDS = 0-1, 5 point-scale) gained a significantly greater (0.70 kg/m2) amount of weight during the follow-up period, compared to those with the healthiest or intermediate diet quality scores (HDS ≥ 2).

Lassale et al. [19••] conducted a study of 3,151 adults who were followed for 13 years in order to evaluate the association between six different diet quality indexes—the 15-point PNNS-GS, the DQI-I, DGI, MDS, rMED, and the MSDPS—and weight change status. When examining results by gender, it was found that there was no association between any of the diet quality indexes and weight change in women (n = 1,471). By contrast, in men (n = 1,680), there was a statistically significant inverse association between diet quality scores and weight change and obesity risk for all but the MSDPS index. More specifically, a higher dietary index score of one standard deviation resulted in lower weight gain during the follow up period, ranging from β = 0.40 (95 % CI; -0.71 to -0.1) for PNNS-GS, to β = 0.87 (95 % CI -1.15 to -0.58) for rMED. The risk of becoming obese was strongly reduced for men in the top quartile compared to the bottom quartile: OR = 0.20, 95 % CI 0.10, 0.42 for the DGI; OR = 0.45, 95 % CI 0.26, 0.79 for the PNNS-GS; and OR = 0.45, 95 % CI 0.27, 0.76 for the MDS.

Zamora et al. [26••] enrolled 4,913 adults (Caucasians and African Americans) aged 18-30 years in order to investigate the association between the 2005 DQI and the risk of gaining ≥10 kg during 20 years of follow-up. There was variation in the results by race but not by gender. For the Caucasian participants with a BMI <25 at the start of the study, there was a significant inverse association between diet quality score and the risk of gaining ≥10 kg over the follow-up period. Each ten-point increase in DQI (maximum of 100 points) was associated with a 10 % lower risk of gaining ≥10 kg of weight during 20 years of follow-up. While amongst African American obese participants, it was found that those with healthiest diet were gaining significantly greater weight. Each 10-point increase in the DQI score was associated with a 15 % higher risk of gaining 10 kg or more during the same follow-up period.

Aljadani et al. [12••] assessed the relationship between diet quality and weight gain in young Australian women over six years, as measured by the ARFS, the Aus-DQI, and the FAVI. In a sub-analysis of those who reported a plausible total energy intake (TEI), it was found that the top quartiles of the ARFS and FAVI indexes were only significantly associated with lower weight gain in the women. Results for the ARFS and FAVI were, respectively, β = −1.6 kg (95 % CI: -2.67 to -0.56), P = 0.003 and β = -1.6 kg (95 % CI: -2.4 to -0.3) P = 0.01.

In a second study by Aljadani et al. [11], it was demonstrated in 7,155 women aged 48 to 56 years that the ARFS score had no relationship with weight change at 6 years of follow-up. In multivariate linear regression analysis, there was still no relationship between diet quality measured by the ARFS and weight gain over six years (β = 0.016 and p = 0.08).

Discussion

The primary aim of this systematic review was to synthesise the best available evidence on the relationship between dietary patterns and weight change in adults in longitudinal analyses. Overall, the findings of the recent eight studies confirm our previous conclusions that optimal diet quality, characterised by higher index scores reflecting greater alignment with national dietary guidelines, is associated with lower weight gain in adults over follow-up periods of at least one year. Across these studies, it was found that those who achieved higher diet quality scores had relatively lower weight gain (BMI: 0.06 to 0.022 kg/m2) compared to those with the poorest diet quality scores. Furthermore, the findings of the recent studies highlight the fact that the majority of adults were gaining weight over time, even when they had eating patterns that met the national dietary guidelines. Across the included studies, it was reported that adults had an average increase in BMI of between 0.3 and 1.73 kg/m2 over 4 to 15 years of follow-up. The studies also found that the majority of participants generally had poor diet quality. In addition, the results varied depending on the index components, and it may be that indexes based on food alone or food combined with nutrient intake are better predictors of weight change than indexes based on nutrient intakes alone. There were also differences in the ability of diet quality indexes to predict weight change depending on gender. It appears that the relationship is stronger for men than for women.

It is apparent from this updated review that there has been increased research interest in capturing overall diet quality by proposing and adapting various diet quality indexes and investigating associations with secondary health outcomes. The eight new studies demonstrate the use of a variety of diet quality indexes, including the Aus-DGI, DGI, ARFS, DQI-I,PNNS-GS, various versions of MDP scores, HDS, AHEI, and others. In contrast, our previous review up to 2011 found that only three indexes had been used in the eight studies, and included FRNS, the five-point DQI, and various versions of indexes measuring adherence to the MDP. This emphasizes the growing interest in using and developing diet quality indexes in epidemiology and public health research to predict weight gain.

Once again it was found that the studies were heterogeneous, not only in the diet quality instruments that were used to quantify dietary patterns, but also in the way that body weight change or weight status was reported. In addition, there were variations in both follow-up period and statistical analysis approach. For example, some studies found different results by gender [13, 19••], with higher diet quality associated with lower weight gain in men but not in women [13, 19••]. It is still uncertain whether diet quality plays a more important role in weight change in men compared with women. In the study by Lassale et al. [19••], the percentage of men who were overweight was higher than that of women (47 % vs. 20 %). This could impact the status of weight at the end of the study, as well as the fact that they did not take into account the changes in eating behaviour over the 13 years of follow-up. There may also be unknown confounders, which could affect both diet and weight in women. Most studies examined measured diet quality only at baseline, and thus changes in eating behaviour over the follow-up period and their impact on weight often were not considered.

Among other studies, results varied based on initial participant BMI [16••] or race [26••]. For example, an analysis [16••] of women only found those who were of normal weight at the start of the study, and who maintained higher diet quality scores from the start and throughout the follow-up period, had a significantly lower risk of becoming obese (BMI ≥30 kg/m2). However, the same study found no relationship with diet quality for those who were overweight at the beginning of the study [16••]. Another study [26••] found higher diet quality scores were associated with lower risk of weight gain in Caucasian American adults, but not amongst African American adults.

Furthermore, when interpreting findings, the instrument used to quantify dietary patterns must be taken into account. Some of the indexes are designed to capture very specific dietary patterns, such as MDP scores, while others such as the FNRS or DQI may have been designed to look at relationships with chronic disease morbidity and/or mortality. Weight gain is a precursor to these, so you would not expect it to necessarily follow that the indexes predict rate of weight gain prospectively. For instance, in a longitudinal study of 467 adults, of the relationship of diet quality measured by three indexes, including the MDS, the HEI-2005, and the DQI-I, the only index that was significantly associated with changes in body weight during 6.7 years of follow-up was the HEI-2005. This relationship was not consistent for the MDS or DQI-I [14]. The study carried out in France also found inconsistent results across the indexes used; however, weight change in men was predicted by five out of six indexes [19••]. The ideal way to draw a straightforward conclusion across the studies is by comparing the studies that used the same diet quality indexes, or at least those containing similar components or subscales.

Two studies, one in both men and women [18] and the other in women only [24], used the FNRS with the objective of evaluating its association with weight change status in subjects who were of healthy weight at the start of the study [24], and who were then followed over 16 years. The study conducted by Kimokoti et al. (2010) [18] found no relationship between the FRNS score and weight change in adults, while the study by Wolongevics et al. [24] in women found a statistically significant inverse association between FRNS and the risk of becoming overweight or obese. A consideration of important confounders, such as weight at baseline—which was adjusted for in the study conducted by Kimokoti et al. [18] but not in the other study—may help to explain the significant association in the second study by Wolongevics. The impact of initial BMI also needs to be considered. While one study enrolled participants of any BMI [18], the other enrolled only women with a healthy BMI at baseline [24].

Two analyses [12••, 21] used the five-point DQI. These found inconsistent results for weight gain in both men and women [21] or women only [12••] over at least six years. These studies did not measure changes in dietary patterns. This review identified only four analyses that used diet quality indexes based on nutrients only, the DQI and FRNS. From the negative results of these studies [12••, 18, 21, 24] we conclude that these indexes may have less predictive ability compared to other indexes in terms of prospective weight change.

Seven studies [14, 15, 19••, 20, 22, 23, 25] used various versions of the MDP score, but all had the same aim of reflecting greater adherence to Mediterranean eating patterns and had similar components or subscales. Eight versions of the MDP index found a significant association between greater MDP scores and lower weight gain in adults, while three others—the MSDPS [15, 19••], MDS [14], and MDS-Panagiotakos [23] —found no relationship with weight gain over time. Thus, we conclude that higher diet quality as reflected by greater adherence to a Mediterranean eating pattern, characterised by a diet rich in vegetables, fruit, nuts, legumes, fish, cereal, and olive oil, predicts lower prospective weight gain in adults. However, it is questionable whether it is appropriate to use these tools outside the relevant European populations. All of the studies that identified a relationship between MDP score and weight change were located in Europe, with the exception of one recent study carried out in Iran, which found no relationship between adherence to the MDP and weight change in Iranian populations [14].

There were two analyses that used the DQI-I [14, 19••], and these had inconsistent results. The study carried out in Iran [14] found no relationship between diet quality and body weight status change, while the study [19••] in France found that the DQI-I was significantly inversely associated with weight change in men but not in women.

In two other analyses, the American DGI was used in France and the Aus DGI in Australia [13, 19••]. These two studies found consistent conclusions that the highest score of the DGI was associated with significantly lower weight gain in men, but not in women.

Other studies found that AHEI-2010 [16••], HEI [14], HDS [17], DASH [16••], and the 15-point PNNS-GS [19••], were associated with lower body weight gain in adults. It is important to mention that these indexes are based on both nutrient and food components. In addition, some have additional components related to non-dietary factors, such as physical activity in the 15-point PNNS-GS used by Lassale [19••], that may affect the results.

The only two indexes based solely on food components or subscales were the ARFS [11, 12••] and the FAVI [12••]. In one of these studies, an association between higher quality diet and lower weight gain was found in young women at six years follow up. However, another study [11] conducted on middle-aged women, also followed for six years, found that the ARFS was not associated with lower weight gain. In both of these studies, no adjustment was made for changes in dietary patterns throughout the follow-up period.

Unfortunately, there is a gap in the current research regarding the relationship between changes in dietary patterns over time and how these changes relate to change in weight for both men and women. It may be changes in diet that are most important in determining weight gain over time, rather than a baseline diet quality score. From this review, at least five studies [13, 14, 16••, 21, 23] measured changes in diet quality over time, yet they used varied methods and it is difficult to compare results. As such, further research is needed in this area. For example the studies done by Arabshahi et al., Asghari et al., and Boggs et al. had significant differences in the method used to estimate the association between diet quality and weight changes over time [13, 14, 16••], and reported inconsistent results. Boggs et al. found a significant association between those who were of normal weight at the start of the cohort and maintained their higher diet quality scores and a lower risk of developing obesity, but not for those who were already overweight at baseline. However, Asghari et al. [14] found no relationship between diet quality score and weight change over time, except for the HEI-2005. Arabshahi et al. found no association between diet quality change and weight change over time in women, but there was an association in men [31]. Also, the impact on weight change for those who increased their diet quality scores over time is not certain, given insufficient numbers in this group. The study by Boggs et al. [16••] found in their overall analysis, without measuring changes in diet quality, that there was no relationship between diet quality and weight change. However, when they considered the changes in diet quality over time, they found in those who maintained their diet quality score and who were of a healthy weight at the start of the study that there was an inverse association between diet quality and the risk of obesity. Thus, further studies that include the change in diet quality over time into the analyses are needed.

Conclusion

In conclusion, from these 16 studies, diet quality indexes or scores appear to be good measures of overall dietary patterns and can be useful tools for predicting weight changes over time. In adults, an optimal diet rich in vegetables, fruit, whole grains and cereals, low fat dairy, and low total and saturated fat is associated with a reduced risk of gaining weight over time. It also suggests that baseline diet quality can be a strong predictor of prospective weight changes in men, but is less reliable in women. However, additional studies examining the impact of initial body weight at baseline and which consider the other important confounders, such as change in diet quality, are needed. This will help to develop more robust conclusions in the future.

In addition, the findings of this review suggest that diet quality indexes based on either both food and nutrient intake or on food items alone are likely to be better predictors of weight change over time than indexes based solely on nutrient intake. This is an important point to consider in terms of public health recommendations, as guidelines should be focused on food patterns rather than nutrient intake.

Acknowledgements

Haya Aljadani received funding from the King Abdul-Aziz University and Ministry of Higher Education, Kingdom of Saudi Arabia, to allow her to study at the University of Newcastle.

Compliance with Ethics Guidelines

Conflict of Interest

Haya Aljadani, Amanda Patterson, David Sibbritt, and Clare E. Collins declare that they have no conflict of interest.

Human and Animal Rights and Informed Consent

This article does not contain any studies with human or animal subjects performed by any of the authors.

Copyright information

© Springer Science+Business Media New York 2015