Age and Ageing Advance Access originally published online on January 11, 2006
Age and Ageing 2006 35(2):110-115; doi:10.1093/ageing/afj031
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Influence of environmental factors on facial ageing
The Danish Twin Registry, Epidemiology, Institute of Public Health, University of Southern Denmark, DK-5000 Odense, Denmark
Address correspondence to: K. Christensen. Tel: (+45) 6550 3049. Fax: (+45) 6590 6938. Email: kchristensen{at}health.sdu.dk
 |
Abstract |
|---|
 Top Abstract BackgroundMaterials and methodsResultsDiscussionKey pointsFundingConflicts of interestReferences |
|---|
Background: a recent twin study has shown that ‘looking old for one’s age’ is associated with increased mortality. Approximately 40% of the variation in perceived age is due to non-genetic factors.
Objective: to examine environmental factors influencing perceived age controlling for diseases.
Design: a twin study.
Setting: in the 2001 wave of the population-based survey—the Longitudinal Study of Aging Danish Twins—participants provided information on a wide range of exposures and health indicators. Additionally, they were asked to have a face photograph taken.
Subjects: a total of 1826 elderly (70+) twins who had a high-quality face photograph taken.
Methods: ten nurses assessed the visual age of each twin from the face photograph. The mean of the nurses’ age estimates for each twin was used as the twin’s perceived age. Multivariate linear regression and intrapair comparison (for intact twin pairs) were used for analyses.
Results: statistically significant determinants of facial ageing associated with high perceived age for men were smoking (P = 0.01), sun exposure (P = 0.02) and low body mass index (BMI) (P<0.005), while for women they were low BMI (P = 0.05) and low social class (P<0.005). The number of children (men) and marital status (P = 0.08) and depression symptomatology score (women) were borderline significantly associated with facial ageing.
Conclusion: our study confirms previous findings of a negative influence of sun exposure, smoking and a low BMI on facial ageing. Furthermore, our study indicates that high social status, low depression score and being married are associated with a younger look, but the strength of the associations varies between genders.
Keywords: perceived age, facial ageing, twin study, elderly
|
Background |
|---|
|
TopAbstractBackgroundMaterials and methodsResultsDiscussionKey pointsFundingConflicts of interestReferences |
|---|
It is generally agreed that certain individuals look ‘old for their age’ or ‘young for their age’. In elderly individuals, ‘looking old for one’s age’ has been considered an indicator of poor health, and a recent twin study showed that ‘looking young for one’s age’ is in fact associated with prolonged survival [1]. The same study found that 40% of the variation in perceived age is due to non-genetic factors. In the present study, we wish to examine environmental causes influencing variation in perceived age.
One of the visual cues underlying the perception of facial ageing is skin wrinkling. A study demonstrated that the impression that people are older than their stated ages correlates with clinical and histologic evidence of wrinkling and solar elastosis [2]. Previous studies have pointed towards smoking [3–6] and sun exposure [7, 8] as environmental factors of major importance for premature skin wrinkling or facial ageing. Frequent intake of alcoholic beverages has also been shown to correlate with appearing older than chronological age [9], whereas a negative correlation between a higher body mass index (BMI) and wrinkles and between a better health and wrinkles has been found [10, 11]. Marital status has been shown not to influence perceived age [9].
Studies of same-sexed twins provide an opportunity for case–control studies that control for age, gender and genes (all in monozygotic twins and half in dizygotic twins). Previously, only one twin study based on 35 pairs has examined environmental factors influencing facial ageing [12]. Though inconclusive, the study does point at smoking and sun exposure as having a negative effect.
Here we use a population-based nationwide study of elderly Danish monozygotic and dizygotic twins to examine environmental causes influencing perceived age, controlling for diseases. Among individual data collected are health status, self-reported diseases, current use of analgesics, smoking, depression score, BMI, family social class, number of children and marital status. Our data do not include explicit information on sun exposure but on the type of profession that indicates exposure to sun (i.e. during working hours).
Based on the previous literature, we would expect smoking and former/present diseases to be associated with looking older than chronological age. Contrarily, we would expect a higher BMI to be correlated with a younger appearance. In addition, our data include information on individual depression symptomatology score, use of analgesics, family social class and number of children, all factors that, to our knowledge, have never been investigated in relation to visual ageing before and that we hypothesise could be associated with perceived age.
|
Materials and methods |
|---|
|
TopAbstractBackgroundMaterials and methodsResultsDiscussionKey pointsFundingConflicts of interestReferences |
|---|
Survey population
Our study is based on the Longitudinal Study of Aging Danish Twins 2001 (LSADT 2001) [13]. LSADT is a longitudinal study that started in 1995, where Danish monozygotic and dizygotic twins [same sex for all studied cohorts but also opposite sex from the oldest cohort (<1920)] who are 70+ years of age are asked to give a health-related interview every second year. In LSADT 2001, a total of 2,448 twins participated, comprising 524 same-sexed pairs (229 monozygotic and 295 dizygotic) and 1,398 single twins. Of these, 1,948 (80%) were cognitively intact and consented to have their face photographed. A digital camera was used at a distance of 0.6 m, person en face, head held straight with a neutral background, if possible. In 1,826 twins (840 men and 986 women), we had a high-quality picture taken, providing a sample size sufficient for the proposed analysis. Not everyone was photographed with a neutral facial expression, but that has been shown not to have any impact on perceived age [9].
Perceived age
Ten female nurses, aged 25–46 years, were engaged to assess the visual age of each twin from the face photograph. The mean of the nurses’ age estimates for each twin was used as the twins’ perceived age. Using ANOVA, the reliability of the mean age rating was estimated to be 0.89, and 
= 0.93 using Cronbach’s coefficient alpha.
Social class
The classification of a participant’s social group employed the social group classification of the Danish National Institute of Social Research [14]: Group I having the highest social rank and Group V the lowest. The social group was taken as the highest of the participant and his or her spouse.
Sun exposure
The degree of sun exposure was based on information on type of job during the longest period of working life. The participants were assessed in two categories: whether they had worked indoor or (partly) outdoor.
Children
The number of biological children was categorised into three groups: 0 children, 1–3 children and 4+ children.
Marital status
Marital status at the time of interview was categorised into three groups: married, separated/widowed and never married.
Depression symptomatology score
The depression symptomatology score was developed by McGue and Christensen [13]. The scale used in this work was computed by summing 17 symptoms: 9 symptoms primarily dealing with well-being and 8 dealing with cognitive difficulties, slowing and loss of energy. The score range is 17–49: the higher the score the more depressed.
Other variables
Also, use of analgesic, smoking (converted to pack-years), alcohol consumption (ever had more than three drinks of alcohol per day for a longer period), various diseases [diabetes, asthma/chronic bronchitis, cancer (except skin cancer), stroke and cardiovascular diseases], BMI and change in BMI since the 1999 survey were included in the analyses.
Analyses of associations
All analyses were made separately for the two sexes. To detect possible associations between perceived age and the suspected phenotypes, bivariate linear regression analyses were carried out, with perceived age as the dependent variable and chronological age as well as the other variables as the independent variable. Stepwise forward inclusion and backward elimination were carried out on multiple linear regression using P = 0.20 for inclusion/exclusion of covariates. Also, within twin pairs, comparisons of intact twin pairs were made using sign and McNemar tests.
|
Results |
|---|
|
TopAbstractBackgroundMaterials and methodsResultsDiscussionKey pointsFundingConflicts of interestReferences |
|---|
Summary statistics stratified by sex are shown in Table 1.
|
It has been previously shown that the perceived age in this sample regressed towards a mean of 77 [1] (Figure 1). Therefore, chronological age was included in all regression models.
|
Regrouping of categorical variables
Regression on social class and number of children showed that we could combine some of the groups, since the estimates for the regression coefficients in bivariate regression with age were comparable for these groups. For men, social Groups II–V were combined, and for women, Groups I–III were combined. As for number of children, the groups with no children and many children (4+) were combined for men, while for women the groups consisting of no children and 1–3 children were combined.
Bivariate regression
The results of the bivariate regressions including age and one other covariate, one at a time for each sex, are summarised in Table 2. For men, the results show that chronological age, smoking, sun exposure, having none or many children (4+) and suffering from strokes or asthma/chronic bronchitis are associated with a higher perceived age, while having a higher BMI is associated with a decrease in perceived age. Also, higher depression symptomatology score, lower social class and not being married have a tendency to increase the perceived age.
|
For women, chronological age, smoking, lower social class, higher depression symptomatology score and suffering from cardiovascular diseases increase perceived age, and there is a tendency for alcohol, sun exposure and suffering from asthma/chronic bronchitis to also do so. On the other hand, there is a slight tendency for a high BMI or being married to be associated with a younger look.
Overall, the results for men and women are very similar, with a few exceptions. For both sexes, the effect of chronological age on perceived age stayed consistently at the same levels: for men, at a level of 
0.3, meaning that in this age group (70+) it takes 3 years for men to visually age 1 year, while the effect for women in the same age group is slightly larger, namely 0.4, meaning that women visually age 1 year per 2.5 years, reflecting the regression towards the mean previously noticed [1].
Multivariate regression
Stepwise regression on all variables (forward substitution and backward elimination—using P = 0.20 for both inclusion and exclusion gave the same results) resulted in models as summarised in Table 2.
For men, the effect of chronological age (P<0.005), smoking (P = 0.01), sun exposure (P = 0.02), lower BMI (P<0.005) and suffering from stroke (P = 0.03) increase the perceived age significantly, while the effects of having one to three children compared with none or more than three (P = 0.08) have a tendency to decrease the perceived age. Lower social class (P = 0.08) and suffering from asthma/chronic bronchitis (P = 0.11) have a tendency to increase the perceived age.
For women, the effect of chronological age (P<0.005), lower social class (P<0.005) and suffering from cardiovascular diseases increase the perceived age significantly, while being married (P = 0.04) tends to decrease the perceived age. There is a tendency that also lower BMI (P = 0.09) and higher depression symptomatology score (P = 0.10) have an increasing effect on the perceived age.
For both sexes, there is no significant effect of alcohol consumption, which may be due to a lack of statistical power as the proportion reporting a substantial alcohol consumption for a longer period was small (men 10%, women 2%).
Table 2 summarises the results of a multivariate regression on variables selected for either men or women in the stepwise selection algorithm. All the regression coefficients are comparable to the results of the reduced models attained by stepwise regressions, but, more interestingly, it shows that practically all environmental factors influence the perceived ages in men and in women in the same direction.
Intrapair comparison
Sign and McNemar tests on intact twin pairs showed the same tendencies as regression, but due to the smaller number of intact pairs only significant effects from smoking (men: P = 0.05; women: P = 0.10) and marital status (men: P = 0.08; women: P = 0.07) were found.
|
Discussion |
|---|
|
TopAbstractBackgroundMaterials and methodsResultsDiscussionKey pointsFundingConflicts of interestReferences |
|---|
This study, aimed at investigating environmental causes influencing variation in perceived age, confirms our hypotheses to a large extent. Our main results show that sun exposure and smoking are associated with an increase in perceived age, whereas a high BMI, high social class, low depression score and being married are associated with a younger look, but the strength of the statistical association differs in men and women. Furthermore, having one to three children seems to have a favourable influence on perceived age in men.
Smoking [3, 4] and sun exposure [7, 8] have been shown to be associated with premature skin wrinkling or facial ageing. We found that sun exposure was highly significant for increasing perceived age in men, but not in women. This could most likely be explained by the fact that our data on sun exposure stem from profession data. Men exposed to sun during working hours have been highly exposed, e.g. fishermen and farmers, whereas only a very small number of women had been exposed to sun during working hours (8.3%). Our data on female sun exposure should thus be interpreted cautiously.
Similar tendency is seen in relation to smoking. Our data show that smoking is significantly associated with an increase in perceived age in men: smoking 20 cigarettes per day for 20 years has the same effect on perceived age as ageing 1 year. The effect on women’s perceived age is not quite as dramatic; here, the facial ageing caused by smoking 20 cigarettes per day for 40 years compares to ageing 1 year. However, in this study, fewer women were smokers and women smoked less than men, which may also explain why the effect of smoking is not statistically significant for women.
The effect of high BMI, high social class and low depression score, though not all significant, was, for both sexes, associated with looking younger than one’s age. A decrease of 2 BMI units for men and 7 BMI units for women has the same effect on perceived age as being 1 year older, shifting social class from highest to lowest compares to 3.5–4 years’ difference in chronological age, while the effect of a change of depression symptomatology score from 17 to 49 is the same as ageing 2.4 years for men and 3.9 years for women. Marital status does not have a statistically significant effect on the perceived age for either men or women, but the tendency is that marriage decreases the perceived age more for women than for men. For men, the effect of being married compared with never married compares to a 1-year decrease in perceived age; for women, 1.9 years. Apart from BMI, these factors have never been investigated in connection with perceived age; however, it makes sense that they would have an influence as they can all be seen as contributors to living the ‘good’ and affluent life. Less worrying, economic stability and social security are likely to be cue factors here.
Our analyses show that a high BMI makes men and women 70+ of age look younger. This corresponds well to the fact that BMI has been proven to be inversely related to degree of wrinkling [10] as well as to the fact that illness often is associated with loss of weight. Not surprisingly, asthma/chronic bronchitis, stroke (in men) and cardiovascular diseases (in women) are associated with looking older, most often with smoking as the underlying cause.
Interestingly, the number of children seems to influence men’s perceived age more than women’s. Having one to three children makes men look younger but has no influence on women’s perceived age. We had no specific expectations concerning this issue.
Except for stroke (in women) and cardiovascular diseases (in men), all the covariates have the expected effects on the perceived age. For stroke, this might be due to the small number (males, 11.4%; females, 4.7%). But it is worth noting that for the significant effects of these two diseases (stroke in men and cardiovascular diseases in women), the effects are in the expected directions: diseases are associated with higher perceived age.
Our study has a number of strengths. It is population based, with a high participation rate. Furthermore, the fact that the photographs were taken as part of a survey, where the interviewers were visiting the participants in their homes, ensured the inclusion of the more frail participants also, whereas this is usually not the case when participants have to attend a clinical survey located at a research institution.
The weakness of our study is primarily that the photographs could not be taken as standardised as in a clinical setting due to various conditions in the participants’ residences. However, the high Cronbach’s alpha value (0.93) shows that the mean age rating (perceived age) is very reliable. Furthermore, a possible underreporting of smoking and alcohol consumption leads to an underestimation of the effects of these factors. Also, our data on sun exposure are based on type of job only and hence do not express cumulative sun exposure, but it is reassuring that we were able to confirm the previous established association between facial ageing and our measure of sun exposure. Moreover, we have not considered the duration of certain diseases and conditions.
All in all, it could be argued that, though statistically significant, the influence of the present investigated environmental factors on facial ageing seems sparse. However, again it must be considered that perceived age in elderly persons (70+) tends to regress towards a mean of 77 [1], and therefore, any influence could be expected to be modest, e.g. it takes 2.5 and 3 years for women and men, respectively, to increase perceived year by 1 year. It is likely that the influence of the same environmental factors on perceived age is larger in younger generations.
|
Key points |
|---|
|
TopAbstractBackgroundMaterials and methodsResultsDiscussionKey pointsFundingConflicts of interestReferences |
|---|
- The present study of 1,826 twins aged 70+ confirms an influence of sun exposure, smoking and BMI on facial ageing.
- The study also indicates that high social status, low depression score and being married are associated with a younger look.
- The strength of the associations varies between genders.
|
Funding |
|---|
|
TopAbstractBackgroundMaterials and methodsResultsDiscussionKey pointsFundingConflicts of interestReferences |
|---|
The research was supported by the US National Institute on Aging research grant NIA-PO1-AG08761, Grete and Sigurd Petersen’s Foundation, the Velux Foundation and Unilever.
|
Conflicts of interest |
|---|
|
TopAbstractBackgroundMaterials and methodsResultsDiscussionKey pointsFundingConflicts of interestReferences |
|---|
There are no conflicts of interest.
|
References |
|---|
|
TopAbstractBackgroundMaterials and methodsResultsDiscussionKey pointsFundingConflicts of interestReferences |
|---|
- Christensen K, Iachine M, Rexbye H et al. "Looking old for your age": genetics and mortality. Epidemiology 2004; 15: 251–2.[Medline]
- Warren R, Gartstein V, Kligman AM, Montagna W, Allendorf RA, Ridder GM. Age, sunlight, and facial skin: a histologic and quantitative study. J Am Acad Dermatol 1991; 25: 751–60.[Web of Science][Medline]
- Daniell HW. Smoker’s wrinkles. A study in the epidemiology of "crow’s feet". Ann Intern Med 1971; 75: 873–80.
[Abstract/Free Full Text] - Model D.Smoker’s face: an underrated clinical sign? Br Med J 1985; 291: 1760–2.
[Abstract/Free Full Text] - Kadune DP, Burr R, Gress R, Kanner R, Lyon JL, Zone JJ. Cigarette smoking: risk factor for premature facial wrinkling. Ann Intern Med 1991; 114: 840–4.
[Abstract/Free Full Text] - Ernster VL, Grady D, Miike R, Black D, Selby J, Kerlikowske K. Facial wrinkling in men and women, by smoking status. Am J Public Health 1995; 85: 78–82.
[Abstract/Free Full Text] - Leyden JJ. Clinical features of ageing skin. Br J Dermatol 1990; 122 (Suppl 35): 1–3.
- Yaar M, Eller MS, Gilchrest BA. Fifty years of skin aging. J Investig Dermatol Symp Proc 2002; 7: 51–8.[CrossRef][Web of Science][Medline]
- Sherertz EF, Hess SP. Stated age. N Engl J Med 1993; 329: 281–2.
[Free Full Text] - Purba MB, Kouris-Blazos A, Wattanapenpaiboon N et al. Can skin wrinkling in a site that has received limited sun exposure be used as a marker of health status and biological age? Age Ageing 2001; 30: 227–34.
- Guinot C, Malvy DJ, Ambroisine L et al. Relative contribution of intrinsic vs extrinsic factors to skin aging as determined by a validated skin age score. Arch Dermatol 2002; 138: 1454–60.
[Abstract/Free Full Text] - Antell DE, Taczanowski EM. How environment and lifestyle choices influence the aging process. Ann Plast Surg 1999; 43: 585–8.[Medline]
- McGue M, Christensen K. Genetic and environmental contributions to depression symptomatology: evidence from Danish twins 75 years of age and older. J Abnorm Psychol 1997; 106: 439–48.[CrossRef][Web of Science][Medline]
- Hansen EJ. Socialgrupper i Danmark. København: Socialforskningsinstituttet. Studie 48.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. C Cosgrove, O. H Franco, S. P Granger, P. G Murray, and A. E Mayes Dietary nutrient intakes and skin-aging appearance among middle-aged American women Am. J. Clinical Nutrition, October 1, 2007; 86(4): 1225 - 1231. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Minerva BMJ, April 15, 2006; 332(7546): 922 - 922. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||