Healthy lifestyles and cardiovascular risk profiles in young Australian adults: the Childhood Determinants of Adult Health Study

Abstract

Introduction We examined whether a healthy lifestyle was associated with cardiovascular risk factors in a cohort of young adults.

Design Cross-sectional.

Method Data from the 2004–2006 Childhood Determinants of Adult Health Study (age range 25–36 years) were used. A lifestyle score [0 (unhealthy) to 8 (healthy)] was derived from eight behaviours (normal body mass index, nonsmoking, low alcohol, salt, meat and regular fish consumption, leisure time physical activity and skim milk use). Using linear regression to adjust sociodemographic characteristics, we examined relationships between the lifestyle score and blood pressure (BP), low-density lipoprotein (LDL) and high-density lipoprotein (HDL) cholesterol, triglycerides, insulin, glucose and an estimate of insulin resistance (IR).

Results In men (n = 863), after adjustment, higher lifestyle scores were linearly associated with lower diastolic BP [regression coefficient (β): – 0.83; 95% confidence interval (CI): –1.23 to –0.43], LDL cholesterol (β: –0.05; 95% CI: –0.09 to –0.007), HDL cholesterol (β: 0.01; 95% CI: 0.001–0.003), insulin (β: –0.50; 95% CI: –0.75 to –0.25) and IR (β: –0.10; 95% CI: –0.16 to –0.04). In women (n = 941), after adjustment, healthy lifestyles were linearly associated with HDL cholesterol (β: 0.02; 95% CI: 0.01 to 0.04) and glucose (β: –0.02; 95% CI: –0.04 to –0.003). There were significant nonlinear associations of the lifestyle score with triglycerides in males and with diastolic BP, LDL cholesterol, triglycerides, insulin and IR in females.

Conclusion Even in young adults, a healthy lifestyle is clearly associated with a better cardiovascular risk profile.

Keywords

adults cardiovascular risk factors risk reduction behaviour

Introduction

Lifestyle contributes heavily to mortality, predominantly by increasing the incidence of cardiovascular disease (CVD) [1]. Capturing the quality of a person's lifestyle is complex because of the breadth and combination of behaviours that influence health and the differing ways to measure those behaviours. A simple lifestyle score that comprises eight factors has been developed by Spencer et al. [2]. Using data from a large cohort of elderly Australian males, these investigators demonstrated that this lifestyle score predicted mortality over a 5-year period in men with [3] and without vascular disease [2]. The usefulness of this score in younger populations is potentially high as, in contrast to other risk scores, it does not require invasive tests, which clinicians may feel are unnecessary in young people. Furthermore, as modification of behaviour is likely to be the first-line intervention to reduce cardiovascular risk in younger people this lifestyle score may be useful as a health improvement tool.

Prior to any such application of the score, it is important to establish its appropriateness in a younger cohort. We therefore sought to investigate the pattern of lifestyle scores in a cohort of young adults, compare the prevalence of healthy lifestyle behaviours in males and females, and assess whether having a healthier lifestyle is associated with a more favourable cardiovascular risk profile.

Methods

This analysis used cross-sectional data from the Childhood Determinants of Adult Health study, a follow-up of participants in the 1985 Australian Schools Health and Fitness Survey [4]. Data were collected from 2004 to 2006 at 34 study clinics around Australia, when participants were aged 25–36 years. Of the original 8498 participants in the 1985 survey, 6840 (80.5%) were traced, 5170 were enrolled in the study, 2881 completed survey questionnaires and 2410 (28.4%) attended a study clinic.

Healthy Lifestyle Score

The lifestyle score used in this study comprises eight factors: body mass index (BMI) less than 25 kg/m², never smoker or ex-smoker ≥ 12 months, ≥ 3 h of moderate to vigorous leisure time physical activity per week, ≤ 20g of alcohol per day, fish/seafood consumption ≥ three times per week, red meat consumption less than five times per week, regular use of skim milk and not adding salt to food. Weight was measured in light clothing using Heine portable scales (Heine, Dover, Delaware, USA) and height with a Leicester stadiometer (Invicta Plastics Ltd., Leicester, UK). We measured moderate and vigorous leisure time activity using the International Physical Activity Questionnaire-long form (IPAQ). The IPAQ has good reliability (r =0.81 across 12 countries) and criterion validity (r =0.33) [5]. Information on smoking was gathered by questionnaire. Dietary data were collected using a food frequency and habits questionnaire that referred to usual consumption over the last 12 months. Weekly consumption of red meat and fish comprised of 10 and nine items, respectively. Participants were classified as not adding salt to food if they said they never or rarely added salt to food during or after cooking. Daily alcohol consumption in grams was estimated from the frequency of consumption of 10 alcoholic beverages multiplied by the average alcohol concentration of each beverage [6]. The Healthy Lifestyle Score was calculated by assigning a point for each healthy behaviour as defined by the criteria above. These were then summed, giving a total score ranging from 0 (no healthy behaviours) to 8 (all healthy behaviours).

Measurement of cardiovascular risk factors

Blood pressure (BP, mmHg) was measured with an Omron HEM-907 Digital Automatic Blood Pressure Monitor (Multipoint Technologies Pty Ltd., Victoria, Australia) after the individual had been sitting quietly for 5 min. The mean of three readings was used. Blood samples (32 ml) were collected after an overnight fast. Assays were conducted in a central laboratory to measure fasting insulin (μIU/l); glucose (mmol/l); total, high-density lipoprotein (HDL, mmol/l) and low-density lipoprotein (LDL, mmol/l) cholesterol and triglycerides (mmol/l). The homoeostasis model assessment (HOMA-IR) index was used to estimate insulin resistance. HOMA-IR is calculated from insulin × glucose/22.5 [7].

Covariates

Age, marital status (married/living as married or single/divorced/separated), education (school only, vocational training or tertiary) and occupation (nonmanual, manual and not working) and, in women, parity, were gathered with standard questions. We collected data on current medications prescribed for hypertension, lipid lowering, diabetes and oral contraception. Personal history of hypertension, diabetes, angina, stroke or myocardial infarction was self-reported. A positive family history of CVD and diabetes reflected a participant's father or brother having a myocardial infarction before the age of 50 years or a first-degree relative having diabetes.

Analysis of data

We examined the distribution of the Healthy Lifestyle Score using sex-specific histograms and Student's t-test. Differences between the sexes in the prevalence of lifestyle items were examined using χ² test. We dichotomised the score at the sample median of 5 to examine the characteristics associated with healthy (scores 5–8) and unhealthy (scores 0–4) lifestyles. Previous investigators have also dichotomised the lifestyle score at the median [2].

We examined the levels of biomedical risk factors (mean and standard deviation) in subgroups defined by ‘unhealthy’ and ‘healthy’ lifestyle scores using unpaired Student's t tests. Sex specific linear regressions of the individual continuous risk factors on the continuous Healthy Lifestyle Score were conducted. We present models that are both unadjusted and adjusted for confounding factors. Confounders were covariates that were associated with the outcome, not intermediates between the exposure and outcome and that changed the coefficient by more than 10% when included in the model. We examined the effect modification between the lifestyle score and sociodemographic factors with product terms. To assess the importance of BMI in the association between the lifestyle score and biomedical risk factors, we ran regression models again, using the lifestyle score excluding BMI [8].

We identified that relationships between the lifestyle score and some biomedical risk factors were nonlinear [9]. The nonlinearity was of the same type irrespective of which CVD risk indicator was the outcome. We transformed the lifestyle score to remove the nonlinearity because it seemed to be an inherent characteristic of the lifestyle score in its relationship with CVD risk. As suggested by Royston et al. [9] we graphed the fitted functions for the relevant biomedical risk factors (95% confidence interval) for each value of the lifestyle score.

We examined the potential importance of loss to follow-up by comparing participants with nonparticipants using 1985 data and also with the Australian population aged 24–35 years [10, 11]. Participants gave their informed consent and the Southern Tasmania Health and Medical Human Research Ethics Committee approved the study.

Results

The numbers of participants with each cardiovascular risk factor measured were as follows: BP, n =1913; HDL cholesterol, n = 1812; LDL cholesterol, n = 1808; triglycerides, n =1821; glucose, n =1819; insulin, n = 1815. These figures exclude those who were pregnant (n = 82), had missing data for risk factors (n =371) or covariates (n = 172). The total number of participants with complete data on lifestyle, sociodemographic factors, medications, fasting blood samples, BP and BMI was 1804. The denominator was higher than this for some analyses depending on which factors were included as confounders.

In terms of characteristics at baseline in 1985 (childhood), those included in analyses (n = 1804) were more likely to be female (52 vs. 49%) and residing in high socioeconomic postcodes (28 vs. 21%), were marginally older [mean age (standard deviation) 11.1 (2.5) vs. 10.9 (2.6) years], had lower BMI [18.1 (2.9) vs. 18.3 (2.7)] and were less often smokers (29 vs. 37%) than those not included in analyses (n = 6694). At follow-up, our sample compared favourably with the Australian population aged 24–35 years in terms of the proportion of never smokers (study sample, 58%; Australian population, 52%), married (study sample, 67%; Australian population, 62%), normal weight (study sample, 48%; Australian population, 49%) and low-risk alcohol consumption (study sample, 93%; Australian population, 87%). Our sample had more often completed postschool education than the Australian population of a similar age (75 vs. 52%).

Overall, women had more healthy lifestyles than men (P >0.01), as shown in Fig. 1, and were more likely to engage in most healthy behaviours except for consumption of fish, physical activity and being a never smoker or an ex-smoker for a year or more (Table 1). People with lower education and manual occupation were more likely to be in the unhealthy lifestyle group (scores >5), principally because of differences in smoking, BMI, consumption of meat and skim milk (data not shown).

Table 1

Prevalence (%) of healthy lifestyle factors

	Males		Females
Lifestyle factors	n	Percentage	n	Percentage	P value
≤ 20 g alcohol per day	1076	84	1487	94	>0.001
≥ 3 serves of fish per week	644	50	724	45	0.014
Never smoked or quit smoking > 1 year ago	709	55	869	54	0.670
>5 serves of meat per week	211	16	2436	27	<0.001
Never adds salt to food	318	25	538	34	>0.001
> 3 h of physical activity per week	415	35	397	26	>0.001
Uses reduced fat milk	556	44	904	57	>0.001
BMI ≤ 25 kg/m²	443	39	735	63	>0.001

BMI, body mass index. P values are from χ² tests.

Men with unhealthy scores (0–4) had higher levels of diastolic BP, LDL cholesterol, triglycerides, insulin and HOMA-IR (Table 2). Women showed similar inverse associations between the number of healthy behaviours and cardiovascular risk factors, with significantly different LDL cholesterol, triglycerides, glucose, insulin and HOMA-IR between healthy and unhealthy groups.

The multivariable linear regression analyses showed the associations in Table 2 were robust to adjustment for confounding factors (Table 3). In general, the associations were stronger in men than they were in women. Figure 2 shows the nonlinear associations that existed between the lifestyle score and triglycerides in men and the lifestyle score and diastolic blood pressure, LDL cholesterol, triglycerides, insulin and HOMA-IR in women. The graphs in Fig. 2 showed that the worst levels of the biomedical risk factors were in those with very few healthy behaviours, but inverse relationships were still evident across the range of lifestyle scores. We found no effect modification of the association between the lifestyle score and biomedical risk factors by sociodemographic factors.

Table 2

Mean (SD) level of cardiovascular risk factors by high and low healthy lifestyle scores

	Healthy Lifestyle Score
	0–4			5–8
	n	Mean	SD	n	Mean	SD	P value
Males
Systolic BP (mmHg)	734	125	11	229	125	11	0.858
Diastolic BP (mmHg)	734	75	9	229	74	9	0.023
LDL cholesterol (mmol/l)	696	3.17	0.88	222	2.90	0.76	>0.001
HDL cholesterol (mmol/l)	707	1.28	0.27	223	1.31	0.24	0.220
Triglycerides (mmol/l)	707	1.32	1.04	223	1.07	0.66	>0.001
Glucose (mmol/l)	705	5.18	0.47	223	5.14	0.40	0.284
Insulin (μIU/l)	705	7.72	5.42	222	6.39	3.89	>0.001
HOMA-IR	704	1.81	1.39	222	1.48	0.95	>0.01
Females
Systolic BP (mmHg)	682	112	11	389	111	10	0.445
Diastolic BP (mmHg)	682	71	9	389	70	8	0.069
LDL cholesterol (mmol/l)	634	2.87	0.84	368	2.75	0.69	0.023
HDL cholesterol (mmol/l)	637	1.52	0.35	368	1.58	0.33	0.015
Triglycerides (mmol/l)	637	0.98	0.61	368	0.86	0.43	0.001
Glucose (mmol/l)	637	4.87	0.40	368	4.81	0.40	0.032
Insulin (μIU/l)	634	7.36	4.74	368	6.04	3.10	>0.001
HOMA-IR	634	1.62	1.11	368	1.30	0.72	>0.001

BP, blood pressure; HDL, high-density lipoprotein; HOMA-IR, homeostasis model of assessment estimate of insulin resistance; LDL, low-density lipoprotein; SD, standard deviation. P-values are from Student's t tests.

Table 3

Associations between the Healthy Lifestyle Score and cardiovascular risk factors

	Unadjusted		Adjusted
	β coefficient	95% CI	β coefficient	95% CI
Males
Systolic BP	–0.42	–0.91 to 0.07	–0.25	–0.77 to 0.26
Diastolic BP	–0.90	–.30 to –0.50	–0.83	–1.23 to –0.43
LDL cholesterol	–0.07	–0.11 to –0.03	–0.05	–0.09 to –0.007
HDL cholesterol	0.01	0.001 to 0.03	0.01	0.001 to 0.003
Triglycerides^a	–0.02	–0.04 to –0.005	–0.02	–0.03 to –0.003
Glucose	–0.02	–0.04 to 0.003	–0.01	–0.03 to 0.01
Insulin	–0.47	–0.70 to –0.23	–0.50	–0.75 to –0.25
HOMA-IR	–0.12	–0.18 to –0.06	–0.10	–0.16 to –0.04
Females
Systolic BP	–0.29	–0.76 to 0.17	–0.33	–0.83 to 0.16
Diastolic BP^a	–0.21	–0.40 to –0.03	–0.29	–0.46 to –0.11
LDL cholesterol^a	–0.03	–0.04 to –0.01	–0.02	–0.04 to –0.003
HDL cholesterol	0.03	0.02 to 0.05	0.02	0.01 to 0.04
Triglycerides^a	–0.03	–0.04 to –0.02	–0.03	–0.05 to –0.02
Glucose	–0.02	–0.04 to –0.01	–0.02	–0.04 to –0.003
Insulin^a	–0.31	–0.39 to –0.23	–0.28	–0.37 to –0.20
HOMA-IR^a	–0.07	–0.09 to – 0.05	–0.07	–0.09 to –0.05

β coefficients are from linear regression analyses. Confounders in men: systolic BP, occupation; diastolic BP, age; LDL cholesterol, age, education, occupation; HDL cholesterol, hypertension medications; triglycerides, education; glucose, age, occupation, lipid-lowering medications; insulin, hypertension medications; HOMA-IR, diabetes medications. Confounders in women: systolic BP, age; diastolic BP, age; LDL cholesterol, age, education, occupation, parity; HDL cholesterol, education, oral contraceptives; triglycerides, marital status, oral contraceptives; glucose, age, lipid-lowering medications; insulin, education; HOMA-IR, education. BP, blood pressure; CI, confidence interval; HDL, high-density lipoprotein cholesterol; HOMA-IR, homeostasis model assessment estimate of insulin resistance; LDL, low-density lipoprotein cholesterol. ^aNon-linear association (see Fig. 2): the values represent the difference in the level of a risk factor between a lifestyle score one point above the mean and a lifestyle score at the mean.

The analyses conducted using a Healthy Lifestyle Score that excluded BMI showed that there were significant linear relationships between the score and LDL cholesterol (β: –0.09; 95% CI: –0.13 to –0.04) and triglycerides (β: –0.05; 95% CI: –0.10 to –0.004) in males. For females, the lifestyle score without BMI was linearly associated with triglycerides (β: –0.05; 95% CI: –0.08 to –0.02), insulin (β: –0.24; 95% CI: –0.44 to –0.04) and HOMA-IR (β: –0.06; 95% CI: –0.10 to –0.01). After adjustment for sociodemographic factors, only the associations between the score and LDL cholesterol (β: –0.07, –0.11, –0.03) in males and triglycerides (β –0.05, –0.08, –0.02) in females remained statistically significant.

Discussion

Our data demonstrate that even in young adults, having a healthy lifestyle is clearly associated with a better cardiovascular risk profile. BMI was largely responsible for these associations; however, even without BMI in the score, there were significant linear associations with lipid profiles in each sex. Clear differences existed between the sexes, with women having significantly healthier lifestyles than men.

Having a greater number of healthy behaviours was associated with better cardiovascular risk factor profiles in both the sexes. For several biomedical risk factors, the association with the lifestyle score was not linear and it was those with the least healthy behaviours that had the worst risk factor profiles. The score is potentially very useful because it is comprised of behaviours that have been shown, in one cohort of women, to be responsible for 72% of cardiovascular deaths [1]. Although comparable data are limited, our findings do support the notion that lifestyle is associated with cardiovascular risk factors in younger [12] and older populations [13].

BMI accounted for a large amount of the association between the lifestyle score and cardiovascular risk factors. We conducted additional analyses because BMI is strongly associated with the risk factors we examined [14]. Although we were concerned that BMI alone may have been solely responsible for the significant associations, with BMI removed, the lifestyle score was still significantly associated with LDL cholesterol in males and triglycerides in females in a linear manner. This shows that although BMI is influential, the other lifestyle factors in the score are also important for cardiovascular health in young adults.

We found that women had healthier lifestyles than men, which other investigators’ findings support [12]. The reasons for this difference are uncertain but could include that women tend to have higher health literacy than men [15] and are more likely to use primary health care services than men [16]. A previous validation study of the lifestyle score revealed no systematic differences between men and women in the extent to which self-reported behaviour was corroborated by the individual's spouse or partner, with high levels of agreement for all the items included in our analyses [16]. Having multiple healthy behaviours was common: 6% of males and 2% of females had none or one healthy behaviour, 18% of males and 10% of females had two healthy behaviours and 77% of males and 87% of females had three or more healthy behaviours. Our findings are supported by other data, albeit relating to unhealthy behaviours, from the Young Finns cohort, where 41% of 18, 21 and 24-year-olds reported two or more of smoking, low diet quality, frequent inebriation and physical inactivity [12].

This study had some limitations. The loss to follow-up of over 75% of the original cohort is a potential shortcoming. However, as these analyses were cross-sectional, it is more important to consider the comparability of the sample with the general population and to have heterogeneity in the distribution of exposures, outcomes and covariates. Overall, the sample was reasonably similar to the Australian population of the same age with the exception of level of education. Furthermore, because of the breadth of the covariates measured we were able to take account of these differences in the analyses. The cross-sectional nature of our analysis limits our ability to infer causality, though other evidence has clearly showed causal pathways from behaviours to CVD [1, 17].

There are also several strengths of the study. We had a large, national sample with comprehensive, valid and reliable measures of biomedical and behavioural risk factors, as well as of covariates. The lifestyle score itself is a strength because, unlike similar scores, it is simple to calculate, does not require clinical assessments, and can be translated into goals that are achievable.

Conclusion

We have showed that a healthy lifestyle is clearly associated with a better cardiovascular risk profile in young adults. The score used in this study focuses on common healthy behaviours that adults can assess themselves, thus increasing the role of the individual in a lifelong process of the prevention of CVD.

Footnotes

Acknowledgements

S.L.G. is supported by a National Health and Medical Research Council (NHMRC) Public Health (Australia) fellowship. Childhood Determinants of Adult Health Study was supported by grants from the NHMRC, National Heart Foundation, Tasmanian Community Fund and Veolia Environmental Services. The authors acknowledge the contributions our sponsors (Sanitarium, ASICS and Target), the study's project manager, Ms Marita Dalton, the project staff and volunteers, and the study participants. Dr Obioha Ukoumunne provided valuable statistical advice.

Conflicts of interest: none declared.

References

Van Dam

Spiegelman

Franco

FB.

Combined impact of lifestyle factors on mortality: prospective cohort study in US women. BMJ 2008; 337:a1440.

Spencer

Jamrozik

Norman

Lawrence-Brown

A simple lifestyle score predicts survival in healthy elderly men. Prev Med 2005; 40:712–717.

Spencer

Jamrozik

Lawrence-Brown

Norman

PE.

Lifestyle still predicts mortality in older men with established vascular disease. Prev Med 2005; 41:583–588.

4 Venn

Thomson

Schmidt

Cleland

Curry

Gennat

et al. Overweight and obesity from childhood to adulthood: a follow-up of participants in the 1985 Australian Schools Health and Fitness Survey. Med J Aust 2007; 186:458–460.

5 Craig

Marshall

Sjostrom

Bauman

Booth

Ainsworth

et al. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc 2003; 35:1381–1395.

6 National Health and Medical Research Council. Australian alcohol guidelines: health benefits and risks Canberra: National Health and Medical Research Council; 2001.

Matthews

Hosker

Rudenski

Naylor

Treacher

Turner

RC.

Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 1985; 28:412–419.

8 Yeap

Almeida

Hyde

Norman

Chubb

Jamrozik

et al. Healthier lifestyle predicts higher circulating testosterone in older men. The Health In Men Study. Clin Endocrinol 2009; 70:455–463.

Royston

Ambler

Sauerbrei

The use of fractional polynomials to model continuous risk variables in epidemiology. Int J Epidemiol 1999; 28:964–974.

10.

10 Australian Bureau of Statistics. National Health Survey 2004–2005: Summary of Results. Cat. No. 4364.0 Canberra: ABS; 2006.

11.

11 Australian Bureau of Statistics. Australia (States) expanded community profile. Date Accessed: 1st March 2006; 2001.

12.

12 Raitakari

Leino

Rakkonen

Porkka

Taimela

Rasanen

et al. Clustering of risk habits in young adults. The Cardiovascular Risk in Young Finns Study. Am J Epidemiol 1995; 142:36–44.

13.

Villegas

Kearney

Perry

IJ.

The cumulative effect of core lifestyle behaviours on the prevalence of hypertension and dyslipidemia. BMC Public Health 2008; 8:210.

14.

Sowers

JR.

Obesity as a cardiovascular risk factor

Am J Med 2003; 115 (Suppl 8A):37S–41S.

15.

Von Wagner

Knight

Steptoe

Wardle

Functional health literacy and health-promoting behaviour in a national sample of British adults. J Epidemiol Community Health 2007; 61:1086–1090.

16.

16 Britt

Miller

Charles

Pan

Valenti

Henderson

et al. General practice activity in Australia 2005–2006. General practice series no. 19. AIHW cat. no. GEP 19Canberra: Australian Institute of Health and Welfare; 2007.

17.

Nocon

Hiemann

Muller-Riemenschneider

Thalau

Roll

Willich

SN.

Association of physical activity with all-cause and cardiovascular mortality: a systematic review and meta-analysis. Eur J Cardiovasc Prev Rehabil 2008; 15:239–246.