Sunscreen and Prevention of Skin Aging A Randomized Trial

Abstract

Chinese translation
Sunscreen use and dietary antioxidants are advocated as preventives of skin aging, but supporting evidence is lacking.
To determine whether regular use of sunscreen compared with discretionary use or β-carotene supplements compared with placebo retard skin aging, measured by degree of photoaging.
Randomized, controlled, community-based intervention. (Australian New Zealand Clinical Trials Registry: ACTRN12610000086066).
Nambour, Australia (latitude 26° S).
903 adults younger than 55 years out of 1621 adults randomly selected from a community register.
Random assignment into 4 groups: daily use of broad-spectrum sunscreen and 30 mg of β-carotene, daily use of sunscreen and placebo, discretionary use of sunscreen and 30 mg of β-carotene, and discretionary use of sunscreen and placebo.
Change in microtopography between 1992 and 1996 in the sunscreen and β-carotene groups compared with controls, graded by assessors blinded to treatment allocation.
The daily sunscreen group showed no detectable increase in skin aging after 4.5 years. Skin aging from baseline to the end of the trial was 24% less in the daily sunscreen group than in the discretionary sunscreen group (relative odds, 0.76 [95% CI, 0.59 to 0.98]). β-Carotene supplementation had no overall effect on skin aging, although contrasting associations were seen in subgroups with different severity of aging at baseline.
Some outcome data were missing, and power to detect moderate treatment effects was modest.
Regular sunscreen use retards skin aging in healthy, middle-aged men and women. No overall effect of β-carotene on skin aging was identified, and further study is required to definitively exclude potential benefit or potential harm.
National Health and Medical Research Council of Australia.

Full text

Sunscreen and Prevention of Skin Aging
A Randomized Trial
Maria Celia B. Hughes, MMedSci; Gail M. Williams, PhD; Peter Baker, PhD; and Ade`le C. Green, MBBS, PhD
Background: Sunscreen use and dietary antioxidants are advocated
as preventives of skin aging, but supporting evidence is lacking.
Objective: To determine whether regular use of sunscreen com-
pared with discretionary use or
␤
-carotene supplements compared
with placebo retard skin aging, measured by degree of photoaging.
Design: Randomized, controlled, community-based interven-
tion. (Australian New Zealand Clinical Trials Registry:
ACTRN12610000086066).
Setting: Nambour, Australia (latitude 26° S).
Patients: 903 adults younger than 55 years out of 1621 adults
randomly selected from a community register.
Intervention: Random assignment into 4 groups: daily use of
broad-spectrum sunscreen and 30 mg of
␤
-carotene, daily use of
sunscreen and placebo, discretionary use of sunscreen and 30 mg
of
␤
-carotene, and discretionary use of sunscreen and placebo.
Measurements: Change in microtopography between 1992 and
1996 in the sunscreen and
␤
-carotene groups compared with con-
trols, graded by assessors blinded to treatment allocation.
Results: The daily sunscreen group showed no detectable increase
in skin aging after 4.5 years. Skin aging from baseline to the end of
the trial was 24% less in the daily sunscreen group than in the
discretionary sunscreen group (relative odds, 0.76 [95% CI, 0.59 to
0.98]).
␤
-Carotene supplementation had no overall effect on skin
aging, although contrasting associations were seen in subgroups
with different severity of aging at baseline.
Limitation: Some outcome data were missing, and power to detect
moderate treatment effects was modest.
Conclusion: Regular sunscreen use retards skin aging in healthy,
middle-aged men and women. No overall effect of
␤
-carotene on
skin aging was identified, and further study is required to defini-
tively exclude potential benefit or potential harm.
Primary Funding Source: National Health and Medical Research
Council of Australia.
Ann Intern Med. 2013;158:781-790. www.annals.org
For author affiliations, see end of text.
Preservation of a youthful complexion has been the goal
of aging humans for thousands of years (1). Today,
many billions of dollars are spent annually on creams and
lotions that purport to treat or protect against skin wrin-
kling (2). Most changes associated with skin aging are due
to photoaging after cumulative sun exposure, superim-
posed on chronologic aging (3).
Photoaging describes the clinical and histologic skin
changes induced by sun exposure. Affected skin loses elas-
ticity and appears dry, wrinkled, and patchily pigmented
and often has dilated superficial blood vessels and actinic
keratoses (4– 6). Histologic changes include epidermal
thickening, atypical keratinocytes, and reduced collagen in
the dermis with abundant abnormal elastin (“dermal elas-
tosis”) (7).
Ultraviolet (UV) A and B components of solar radia-
tion are implicated in photoaging of the skin (8). Apart
from unwanted cosmetic effects, photoaging is a strong risk
factor for skin cancer (9). Ultraviolet radiation damages
nucleic acids and proteins in epidermal cells directly and
through reactive oxygen species (10), resulting in impaired
collagen and elastin homeostasis, local immune suppres-
sion, altered differentiation of keratinocytes, and ultimately
tumor development (8, 10, 11).
Although dermal elastosis is considered definitive con-
firmation of photoaging, several noninvasive techniques
can also assess its presence and severity: visual analog scor-
ing (12), skin extensibility (13), pulsed ultrasonography of
skin (14), and silicone impressions of skin surface topog-
raphy (15, 16). We have previously shown that severity of
dermal elastosis (17, 18) is predicted by standard grading
of the microtopography of the skin surface (19, 20), which
provides a valid measure of skin photoaging up to age 70
years (21).
Among myriad creams, drugs, and “cosmeceuticals”
available over the counter or by prescription, several pre-
ventive and therapeutic agents for photoaged skin are be-
lieved to be efficacious, the most common being sunscreen
(22). However, experimental evidence showing that sun-
screen protects against aging (23) is not matched by hu-
man evidence. A trial in 35 patients with a history of skin
cancer randomly assigned to sunscreen or placebo for 2
years showed no significant difference in dermal elastosis
with sunscreen use (24).
No known randomized studies in humans have evalu-
ated the effect of sunscreen on surface changes associated
with skin aging. We performed a randomized, controlled
trial to examine whether daily sunscreen use could prevent
progression of skin aging in adults younger than 55 years
(25). In addition, in view of experimental evidence that
oral antioxidants can reduce signs of oxidative skin damage
See also:
Print
Summary for Patients.......................I-28
Annals of Internal Medicine Original Research
© 2013 American College of Physicians 781

and wrinkling due to sun exposure (26), we evaluated
whether
␤
-carotene supplements could protect against skin
aging.
METHODS
Design Overview
The Nambour Skin Cancer Prevention Trial was a
randomized, community-based trial in Nambour, Australia
(latitude 26 °S). Aims, methods, and results have been fully
documented elsewhere (25, 27–29). In brief, the study was
conducted from 1992 to 1996 in 1621 randomly selected
adults and evaluated whether daily application of broad-
spectrum sunscreen or dietary supplementation with
␤
-carotene could reduce skin cancer and retard actinic ker-
atosis and photoaging (25). The Queensland Institute of
Medical Research Ethics Committee (Queensland, Austra-
lia) approved the study, and participants provided written
informed consent.
Setting and Participants
Per the protocol, this study was restricted to partici-
pants younger than 55 years because their skin aging is
caused predominantly by photoaging rather than by pho-
toaging and growing old. Persons receiving vitamin supple-
ments containing
␤
-carotene or applying sunscreen on a
strict daily basis were ineligible. Height and weight were
measured, and personal information, including skin color,
skin reaction to sun exposure, outdoor behavior, sunburn
history, and smoking status, were obtained at baseline by
using standardized questionnaires.
Randomization and Interventions
Using a 2 ⫻2 factorial design, one of our study inves-
tigators, who had no knowledge of the participants, ran-
domly assigned them by using a computer-generated ran-
domized list to daily application of sunscreen labelled
“sun-protection factor 15⫹,” containing 8% (by weight)
2-ethylhexyl-p-methoxycinnamate and 2% (by weight)
4-tert-butyl-4' methoxy-4-dibenzoylmethane (Ross Cos-
metics, Melbourne, Victoria, Australia), or discretionary
sunscreen use (placebo sunscreen was considered unethical)
and to 30 mg of
␤
-carotene or placebo supplements daily.
Those allocated to daily sunscreen use were asked to apply
the intervention sunscreen to their head, neck, arms, and
hands every morning, with reapplication after heavy sweat-
ing, bathing, or spending more than a few hours outdoors.
Outcomes and Follow-up
The primary outcome was change in photoaging from
1992 to 1996 in those in the intervention groups com-
pared with their respective controls. To assess photoaging,
trained personnel obtained skin surface replicas from the
back of the left hand by using silicone-based impression
material (SilFlo, Flexico, Potters Bar, United Kingdom),
avoiding scarred areas. Participants were asked to not use
moisturizer or sunscreen the day that the replicas were
taken.
Experienced assessors who were unaware of treatment
allocation graded replicas by using the Beagley and Gibson
scale of microtopography grades (15, 16). Grades increase
from 1 (undamaged skin with fine lines evenly spaced in a
2-directional network) to 6 (increasing severity of changes
characterized by surface flattening, deepening of horizontal
lines, and loss of vertical lines). Intra- and intergrader re-
peatability of assessors was high, with weighted
␬
statistics
of 0.81 and 0.86, respectively (19).
Every 3 months, adverse effects were assessed and ad-
herence was evaluated by measured weights of returned
sunscreen bottles for the daily sunscreen group and re-
maining tablet counts. Biennially, application frequency in
all participants was assessed by questionnaire, dermal
␤
-carotene was assessed by photometric measurement (28),
and sun exposure and smoking habits were updated.
Statistical Analysis
The number of trial participants younger than 55
years who were eligible for study was determined by the
original random sample drawn from the Nambour
community (25). Thus, our sample size was determined
by practical constraints rather than a priori power
calculations.
Data were analyzed according to treatment as ran-
domly allocated. Change in photoaging was assessed by
comparing change in microtopography grades from base-
line to the end of the trial among intervention and control
groups by ordinal logistic regression using generalized esti-
mating equations (GEEs) (30). The GEE model is based
on generalized linear regression and allows dependence be-
Context
Whether sunscreen or
␤
-carotene protects against skin
aging has not been established.
Contribution
After 4 years, participants randomly assigned to daily ap-
plication of sunscreen showed less skin aging than those
instructed to use sunscreen on a discretionary basis. No
difference in skin aging was shown with daily
␤
-carotene
compared with placebo.
Caution
Power was limited; although results suggest no effect,
a beneficial or harmful effect of
␤
-carotene cannot be
confidently excluded.
Implication
Daily sunscreen use protects against skin aging. Although
no effect on aging was seen with
␤
-carotene use, these
findings need confirmation before firm conclusions can
be made.
—The Editors
Original Research Sunscreen and Skin Aging
782 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 www.annals.org

tween repeated outcome measurements (30). It adjusts for
within-person correlation of outcomes over time and ac-
counts for the magnitude of differences in categories of
aging (rather than simply the odds of more aging, yes or
no) and the changes in odds of higher grades of skin aging
over time.
This method gives effect estimates in terms of the odds
of having higher microtopography grades in 1996 relative
to 1992 for each category of sunscreen and
␤
-carotene
intervention and the relative odds, assumed constant from
1 grade to the next (proportional odds assumption), com-
paring these. The effects of treatments on microtopogra-
phy grades over time were estimated by specifying an in-
teraction between trial allocation and time in the model.
Three effect estimates (with 95% CIs using robust SEs)
were calculated for each intervention (that is, daily sun-
screen and
␤
-carotene): change in microtopography grade
over time in the intervention and control groups and rela-
tive change over time in microtopography between the 2
groups. All estimates for the sunscreen intervention were
adjusted for the
␤
-carotene intervention and its interaction
with time, and vice versa. Models were fitted using the
REPOLR procedure (31), specifying exchangeable correla-
tion between repeated measures of microtopography grades
to obtain an estimate of relative change over time in the
microtopography intervention and control groups, supple-
mented with a contrast function written by an investigator
to obtain an estimate of change in microtopography grade
over time in the 2 groups (R, version 2.13.2; R Founda-
tion for Statistical Computing, Vienna, Austria). The score
test was used to assess adherence to proportional odds
assumption.
The chi-square test was used to assess whether baseline
photoaging grade was associated with missing follow-up
grade. To test for possible differences in the final study
sample, characteristics of the participants with 2 photoag-
ing grades were compared with those with only 1 by using
multiple logistic regressions applying a GEE. A binary vari-
able for data completeness was created for both time points
and used as the outcome variable, with time and the char-
acteristic being analyzed as the explanatory variables. Anal-
yses of the effects of intervention were repeated, including
factors that were significantly associated with having 1
missing microtopography grade.
To assess consistency of effect according to baseline
characteristics (age, sex, education, body mass index, smok-
ing status, phenotype, sun exposure, and history of skin
cancer), we performed subgroup analyses using ordinal lo-
gistic regression applying a GEE and incorporating an in-
teraction among time, treatment allocation, and the previ-
ously cited factors to detect heterogeneity of effects. Effect
estimates for each subgroup were obtained using the
REPOLR procedure in R specifying an exchangeable cor-
relation structure, as described earlier. Because assessment
of the overall significance of the third-order interaction
(to test heterogeneity of effect) is not available using
REPOLR, Pvalues for the interactions were estimated
from the score test in the GEE model by using the PROC
GENMOD procedure (SAS, version 9.2; SAS Institute, Cary,
North Carolina), assuming independent correlation among
repeated measurements of microtopography grades. Be-
cause of the high positive correlation among repeated mea-
surements (0.63 to 0.81), these Pvalues were smaller than
if dependency among repeated measures were considered.
We conducted 2 sensitivity analyses: We reanalyzed
the data only on participants with complete photoaging
grades, then we imputed values for missing photoaging
grades and covariates by using multiple imputation by the
logistic regression method with 10 iterations (32). To as-
sess whether treatment effects differed by preexisting level
of photoaging, we conducted separate exploratory analyses
for participants with baseline grades 3 to 4 and 5 to 6. All
Pvalues were 2-sided; a Pvalue less than 0.05 was consid-
ered significant.
Role of the Funding Source
This study was supported by the National Health and
Medical Research Council of Australia, Canberra, Austra-
lian Capital Territory, Australia; Ross Cosmetics, Mel-
bourne, Victoria, Australia; and Roche Vitamins and Fine
Chemicals, Nutley, New Jersey. None of these sources had
any role in the design or conduct of the study; collection,
management, analysis, or interpretation of the data; or
preparation, review, or approval of the manuscript.
RESULTS
Of the 1621 Nambour residents enrolled in the trial,
903 were younger than 55 years and eligible for the study
(Figure 1 and Appendix Figure, available at www.annals
.org). Good-quality replicas of the back of the hand were
obtained from 817 participants at baseline and 673 in
1996, and 886 participants (98% of 903) (mean age, 39
years [SD, 7]; 58% women) had at least 1 good-quality
skin replica. Of these, 604 contributed replicas in 1992
and 1996, 213 in 1992 only, and 69 in 1996 only. Com-
pared with those with 2 replicas, participants who contrib-
uted only 1 (n⫽282) were more likely to have severe
photoaging of the neck and 2 to 5 total sunburns. Having
a missing microtopography grade in 1996 was not associ-
ated with baseline microtopography grade.
Most participants were fair-skinned, and more than
90% burned on acute sun exposure. One half worked
mainly indoors; around 43% ever smoked regularly. There
were no differences in phenotype, sun exposure, or pretrial
sunscreen use between the intervention and control groups
at baseline, although slightly more people were randomly
assigned to
␤
-carotene than to placebo (P⫽0.06) (Table
1). Reported sun exposure was similar between the daily
and discretionary sunscreen groups during the trial (78%
of the daily sunscreen group and 76% of the discretionary
sunscreen group spent ⬍50% of weekend time outdoors)
(P⫽0.25). Use of sun-protection measures other than
sunscreen was also similar (54% of the daily sunscreen
Original ResearchSunscreen and Skin Aging
www.annals.org 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 783

group and 53% of the discretionary sunscreen group usu-
ally sought shade, and 63% of the daily sunscreen group
and 67% of the discretionary sunscreen group usually wore
a hat).
In 1992, 58% of participants had moderate photoag-
ing (grades 3 and 4); in 1996, the corresponding propor-
tion was 49% (Tables 2 and 3). When the odds of having
higher microtopography grades in 1996 compared with
Figure 1. Study flow diagram of the Nambour sunscreen use and photoaging study, 1992–1996.
With good-quality skin surface replica in 1992 and/or 1996
(n = 442)
Good-quality skin surface replica in 1992 and 1996: 310
Good-quality skin surface replica in 1992 only: 107
Good-quality skin surface replica in 1996 only: 25
With good-quality skin surface replica in 1992 and/or 1996
(n = 444)
Good-quality skin surface replica in 1992 and 1996: 294
Good-quality skin surface replica in 1992 only: 106
Good-quality skin surface replica in 1996 only: 44
Randomly assigned
(n = 1621)
Assigned to daily suncreen
(n = 812)
Eligible for photoaging study (n = 453)
Residents of Nambour invited to attend
(n = 1850)
Did not attend survey (n = 203)
Attended baseline survey and were eligible
(n = 1647)
Analysis Follow-up Allocation Enrollment
Excluded (n = 359)
Age ≥55 y: 357
Not white: 2
Excluded (n = 359)
Age ≥55 y
Excluded (n = 26)
Declined: 22
Not examined: 4
With skin surface replica in 1992 and/or 1996
(n = 447)
No skin surface replica in 1992
and 1996 (n = 6)
No replica in 1992 (n = 16)
Died (n = 2)
Active participants did not
attend 1996 skin examination
(n = 14)
No replica in 1996 (n = 9)
Became passive participants
during follow-up (n = 71)
No skin surface replica in 1992
and 1996 (n = 4)
No replica in 1992 (n = 22)
Died (n = 1)
Active participants did not
attend 1996 skin examination
(n = 12)
No replica in 1996 (n = 5)
Became passive participants
during follow-up (n = 74)
Assigned to discretionary
suncreen (n = 809)
Eligible for photoaging study (n = 450)
With skin surface replica in 1992 and/or 1996
(n = 446)
Poor-quality replica in 1992 and no
replica in 1996 (n = 3)
No replica in 1992 and poor-quality
replica in 1996 (n = 2)
Poor-quality replica in 1992 (n = 9)
Poor-quality replica in 1996 (n = 11)
No replica in 1992 and poor-quality
replica in 1996 (n = 2)
Poor-quality replica in 1992 (n = 22)
Poor-quality replica in 1996 (n = 14)
Original Research Sunscreen and Skin Aging
784 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 www.annals.org

Table 1. Participant Characteristics at Baseline in 1992, According to Sunscreen and
␤
-Carotene Allocation*
Characteristic Intervention,
n (%)
Daily Sunscreen
(
n
ⴝ442)
Discretionary Sunscreen
(
n
ⴝ444)
␤
-Carotene
(
n
ⴝ447)
Placebo
(
n
ⴝ439)
Sex
Men 189 (42.8) 185 (41.7) 185 (41.4) 189 (43.1)
Women 253 (57.2) 259 (58.3) 262 (58.6) 250 (57.0)
Age
25 to ⬍40 y 226 (51.1) 220 (49.6) 239 (53.5) 207 (47.2)
40 to ⬍55 y 216 (48.9) 224 (50.4) 208 (46.5) 232 (52.9)
Country of birth†
Australia/New Zealand 409 (92.5) 405 (91.4) 409 (91.5) 405 (92.5)
Other 33 (7.5) 38 (8.6) 38 (8.5) 33 (7.5)
Education†
High school 187 (52.4) 185 (49.7) 192 (50.5) 180 (51.4)
Higher education 170 (47.6) 187 (50.3) 188 (49.3) 170 (48.6)
Skin color†
Fair 257 (58.1) 248 (56.0) 260 (58.2) 245 (55.9)
Medium 161 (36.4) 172 (38.8) 160 (35.8) 173 (39.5)
Dark 24 (5.4) 23 (5.2) 27 (6.0) 20 (4.6)
Skin reaction to acute sun†
Burn, never tan 94 (21.3) 92 (20.8) 97 (21.7) 89 (20.3)
Burn, then tan 326 (73.8) 318 (71.8) 322 (72.0) 322 (73.5)
Only tan 22 (5.0) 33 (7.5) 28 (6.3) 27 (6.2)
Previous occupations†
Mainly outdoors 82 (18.6) 73 (16.5) 77 (17.2) 78 (17.8)
Indoors and outdoors 130 (29.4) 147 (33.2) 146 (32.7) 121 (29.9)
Mainly indoors 230 (52.0) 223 (50.3) 224 (50.1) 229 (52.3)
Sunburns†
0 25 (5.7) 18 (4.1) 22 (4.9) 21 (4.8)
1 53 (12.0) 48 (10.8) 56 (12.5) 45 (10.3)
2–5 218 (49.3) 230 (51.9) 227 (50.8) 221 (50.5)
⬎5 146 (33.0) 147 (33.2) 142 (31.8) 151 (34.5)
Nevi on back†
0 55 (12.7) 51 (11.7) 55 (12.6) 51 (11.9)
1–10 285 (65.8) 283 (65.1) 287 (65.5) 281 (65.4)
ⱖ11 93 (21.5) 101 (23.2) 96 (21.9) 98 (22.8)
History of skin cancer
No 379 (85.8) 379 (85.4) 385 (86.1) 373 (85.0)
Yes 63 (14.3) 65 (14.6) 62 (13.9) 66 (15.0)
Clinical photoaging of neck†
None 158 (35.8) 146 (33.0) 157 (35.2) 147 (33.6)
Low to moderate 220 (49.9) 218 (49.2) 226 (50.7) 212 (48.4)
Severe 63 (14.3) 79 (17.8) 63 (14.1) 79 (18.0)
Body mass index†
⬍25.0 kg/m
2
170 (50.5) 166 (46.4) 176 (48.1) 160 (48.6)
25.0–29.9 kg/m
2
112 (33.2) 140 (39.1) 130 (35.5) 122 (37.1)
ⱖ30.0 kg/m
2
55 (16.3) 52 (14.5) 60 (16.4) 47 (14.3)
Sunscreen use outdoors before randomization†
Never 71 (16.1) 77 (17.4) 72 (16.1) 76 (17.4)
⬎50% of the time 197 (44.6) 181 (40.9) 194 (43.4) 184 (42.0)
ⱖ50% of the time 150 (33.9) 157 (35.4) 155 (34.7) 152 (34.7)
Always 24 (5.4) 28 (6.3) 26 (5.8) 26 (5.9)
Continued on following page
Original ResearchSunscreen and Skin Aging
www.annals.org 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 785

1992 were examined after adjustment for sunburns and
photoaging of the neck, only the daily sunscreen group
showed no detectable increase in microtopography grade
(model 2 in Table 4). Compared with discretionary sun-
screen users, persons randomly assigned to daily sunscreen
were 24% less likely to show increased aging (relative odds,
0.76 [95% CI, 0.50 to 0.98]).
There was no difference in increases in microtopogra-
phy grades among persons allocated to
␤
-carotene and pla-
cebo (relative odds, 0.95 [CI, 0.74 to 1.22]) (model 2 in
Table 4), and odds were consistent across photoaging lev-
els (score test P⫽0.51). Results were not materially dif-
ferent from models without these covariates (model 1 in
Table 4) and were consistent with results using only those
participants with complete photoaging grades (n⫽604)
and with results incorporating multiple imputations of
missing data.
With regard to long-term self-reported treatment ad-
herence, by 1996 a total of 77% of daily sunscreen users
were applying sunscreen at least 3 to 4 days per week com-
pared with 33% of discretionary users. Supplement adher-
ence (defined as taking at least 80% of the prescribed tab-
lets) was 68% in the
␤
-carotene group and 67% in the
placebo group. The
␤
-carotene group had significantly
greater mean skin reflectance (measured in integers of 3 to
11 with an SE of 0.04 at baseline) on the palm at
follow-up than at baseline (6.2 vs. 6.0; P⬍0.001, paired
ttest), and their follow-up values were greater than those of
the placebo group (5.3; P⬍0.001).
Effect of sunscreen did not vary according to baseline
characteristics of participants (Figure 2). Exploratory anal-
ysis of effects of sunscreen and
␤
-carotene according to
baseline microtopography grades suggested stronger and
inverse associations with both treatments in those with less
severe skin aging at baseline. For participants with micro-
topography grades of 3 or 4 at baseline, skin aging was
reduced among those in the daily sunscreen (odds ratio
[OR], 0.77 [CI, 0.48 to 1.23]) and
␤
-carotene (OR, 0.52
[CI, 0.32 to 0.84]) groups than in their respective compar-
ison groups. For those with baseline microtopography
grades 5 or 6, daily sunscreen use was not associated with a
change in skin aging (OR, 0.90 [CI, 0.44 to 1.87]),
whereas the
␤
-carotene group tended to experience more
Table 1—Continued
Characteristic Intervention,
n (%)
Daily Sunscreen
(
n
ⴝ442)
Discretionary Sunscreen
(
n
ⴝ444)
␤
-Carotene
(
n
ⴝ447)
Placebo
(
n
ⴝ439)
Recreational activity†
None 127 (35.6) 122 (32.9) 131 (34.6) 118 (33.8)
Low 94 (26.3) 83 (22.4) 97 (25.6) 80 (22.9)
Moderate 50 (22.4) 93 (25.1) 84 (22.2) 89 (25.5)
High 56 (15.7) 73 (19.7) 67 (17.7) 62 (17.8)
Smoking status†
Never smoker 225 (58.0) 219 (54.9) 234 (57.5) 210 (55.3)
Former smoker 107 (27.6) 123 (30.8) 110 (27.0) 120 (31.6)
Current smoker 56 (14.4) 57 (14.3) 63 (15.5) 50 (13.2)
␤
-Carotene allocation
Placebo 233 (52.7) 206 (46.4) – –
␤
-Carotene 209 (47.3) 238 (53.6) – –
Sunscreen allocation
Daily sunscreen – – 238 (53.2) 206 (46.9)
Discretionary sunscreen – – 209 (46.8) 233 (53.1)
*Numbers and percentages show distribution of 886 respondents by sunscreen and
␤
-carotene allocation independently to demonstrate numbers used in analyses.
†Missing responses for country of birth, skin color, skin reaction to acute sun, previous occupations, sunburns, sunscreen use before randomization (n⫽1), education
(n⫽156), nevi on back (n⫽18), clinical photoaging of the neck (n⫽2), body mass index (n⫽191), recreational activity (n⫽158), and smoking status (n⫽99).
Table 2. Change in Skin Aging Grades From 1992 to 1996
Among Participants, by Sunscreen Allocation
Skin Aging
Grade in
1992
Participants, by Skin Aging Grade in 1996,
n (%)
*
No Grade 3 4 5 6
Daily sunscreen
No grade 0 (0.0) 1 (4.0) 9 (36.0) 9 (36.0) 6 (24.0)
3 8 (28.6) 8 (28.6) 10 (35.7) 2 (7.1) 0 (0.0)
4 52 (24.9) 11 (5.3) 113 (54.1) 28 (13.4) 5 (2.4)
5 22 (20.0) 1 (0.9) 16 (14.6) 62 (56.4) 9 (8.2)
6 25 (35.7) 0 (0.0) 2 (2.9) 6 (8.6) 37 (52.9)
Discretionary
sunscreen
No grade 0 (0.0) 0 (0.0) 10 (22.7) 18 (40.9) 16 (3.6)
3 13 (38.2) 6 (17.7) 14 (41.2) 1 (2.9) 0 (0.0)
4 45 (22.2) 7 (3.5) 108 (53.2) 38 (18.7) 5 (2.5)
5 18 (19.2) 2 (2.1) 9 (9.6) 46 (48.9) 19 (20.2)
6 30 (43.5) 0 (0.0) 3 (4.4) 8 (11.6) 28 (40.6)
*The percentage is the number of participants/number of participants who had a
skin aging grade in 1992 ⫻100.
Original Research Sunscreen and Skin Aging
786 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 www.annals.org

skin aging (OR, 1.38 [CI, 0.66 to 2.88]) than the placebo
group.
The main reported symptoms relating to use of sun-
screen were contact allergy or skin irritation (3%), greasi-
ness (1%), and interference with perspiration or stinging
eyes after facial perspiration (0.8%) (28).
DISCUSSION
In this community-based, randomized, controlled
trial, we have shown that regular application of sunscreen
by people younger than 55 years for 4.5 years significantly
retarded aging of the skin. This difference does not seem to
be due to changes in outdoor behavior or sun protection by
the intervention compared with the control group. Long-
term
␤
-carotene supplementation did not seem to influ-
ence progressive skin aging, although we could not rule out
a small decrease or increase in skin aging as a result of
supplementation.
Despite the widespread belief that by screening out
solar UV radiation implicated in skin aging (8), sunscreen
application can diminish its severity in young and middle-
aged adults as they grow older (22), to date there has been
evidence of this only in hairless mice (23, 33). A search of
relevant English-language papers in MEDLINE (1980 to
November 2012) using the terms “sunscreen” (and “beta-
carotene”) and “photoaging,” “skin aging,” or “skin wrin-
kling” identified a single trial involving 35 patients with
past skin cancer that evaluated the effect of sunscreen on
histologic skin aging. The study showed no difference in
dermal elastosis between sunscreen and placebo groups af-
ter analysis during which repeated measurements were ac-
counted for (24). To our knowledge, whether sunscreen
protects humans against visible rather than histologic pre-
mature skin aging has not previously been tested.
These results have important clinical implications. In
our data, a unit increase in microtopography grade is sig-
nificantly related to visible deterioration in skin texture
(coarser skin and increased wrinkling) and an increase in
visible small blood vessels and comedones on the face (as
assessed by dermatologists [4]). More important, a unit
increase in microtopography significantly correlates with
risk for actinic keratoses and skin cancer (16, 34). A reduc-
tion in the highly prevalent aging changes among middle-
aged adults by regular application of sunscreen will there-
fore be associated with cosmetic benefit (prevention of
visible aging changes and hence more youthful appearance)
and reduced risk for skin cancer.
The cost-effectiveness of promoting daily sunscreen
use based on skin cancer prevention alone (35) is probably
substantially higher after accounting for the additional pre-
vention of skin photoaging. Whether our results would
have differed with a sunscreen with a higher sun-protection
factor or one with greater absorption in the UVA spectrum
is debatable, because the overriding factor in achieving ad-
Table 3. Change in Skin Aging Grades From 1992 to 1996
Among Participants, by
␤
-Carotene Allocation
Skin Aging
Grade in
1992
Participants, by Skin Aging Grade in 1996,
n (%)*
No Grade 3 4 5 6
␤
-Carotene
No grade 0 (0.0) 1 (2.4) 15 (36.6) 14 (34.2) 11 (26.8)
3 12 (33.3) 8 (22.2) 16 (44.4) 0 (0.0) 0 (0.0)
4 43 (20.5) 8 (3.8) 130 (61.9) 27 (12.9) 2 (1.0)
5 22 (22.2) 1 (1.0) 8 (8.1) 53 (53.5) 15 (15.2)
6 19 (31.2) 0 (0.0) 4 (6.6) 4 (6.6) 34 (55.7)
Placebo
No grade 0 (0.0) 0 (0.0) 4 (14.3) 13 (46.4) 11 (39.3)
3 9 (34.6) 6 (23.1) 8 (30.8) 3 (11.5) 0 (0.0)
4 54 (26.7) 10 (5.0) 91 (45.1) 39 (19.3) 8 (4.0)
5 18 (17.1) 2 (1.9) 17 (16.2) 55 (52.4) 13 (12.4)
6 36 (46.2) 0 (0.0) 1 (1.3) 10 (12.8) 31 (39.7)
*The percentage is the number of participants/number of participants who had a
skin aging grade in 1992 ⫻100.
Table 4. Odds of Having Higher Microtopography Grades in 1996 Relative to 1992, by Sunscreen and
␤
-Carotene Intervention*
Intervention Model 1 Model 2
Odds of 1996 Compared With
1992 (95% CI)†
P
Value Odds of 1996 Compared With
1992 (95% CI)†
P
Value
Sunscreen‡
Daily sunscreen 1.19 (1.00–1.41) 0.046 1.18 (0.99–1.39) 0.060
Discretionary sunscreen 1.56 (1.29–1.88) ⬍0.001 1.54 (1.28–1.86) ⬍0.001
Relative odds, daily sunscreen/discretionary sunscreen 0.76 (0.59–0.98) 0.033 0.76 (0.59–0.98) 0.033
␤
-Carotene§
␤
-Carotene 1.32 (1.12–1.55) 0.001 1.31 (1.11–1.55) 0.001
Placebo 1.40 (1.16–1.70) ⬍0.001 1.38 (1.14–1.67) ⬍0.001
Relative odds,
␤
-carotene/placebo 0.94 (0.73–1.20) 0.61 0.95 (0.74–1.22) 0.69
*Represents 1490 records across 886 persons.
†Odds ratios derived from generalized estimating equation models.
‡In model 1, the analysis was adjusted for the
␤
-carotene intervention; in model 2, the analysis was also adjusted for factors associated with missing photoaging grade
(number of sunburns and clinical photoaging of the neck).
§In model 1, the analysis was adjusted for the sunscreen intervention; in model 2, the analysis was also adjusted for factors associated with missing photoaging grade (number
of sunburns and clinical photoaging of the neck).
Original ResearchSunscreen and Skin Aging
www.annals.org 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 787

equate skin protection is application of a liberal quantity of
sunscreen; the sun-protection factor or precise shape of the
sunscreen-absorption spectrum is far less important (36,
37). The effect of sunscreen may vary depending on other
risk factors associated with skin aging, namely increasing
age (38), UV-susceptible phenotypes (fair skin and an
inability to tan), male sex, smoking, and body mass index
(4, 18, 39– 44); however, our data did not support this
theory.
Our null result for
␤
-carotene contrasts with the only
relevant clinical study identified, which involved 29 Ko-
rean women in whom photoaging measures (skin elasticity,
depth of skin wrinkling assessed by digitized images of
replicas of “crow’s feet” skin near the eyes, and immuno-
histochemical assessment of buttock skin samples) were
taken before and after a 3-month period of daily
␤
-carotene supplementation (15 women received 30-mg
capsules and 14 received 90-mg capsules) (45). After 3
Figure 2. Effect of sunscreen intervention on photoaging, according to baseline characteristics.
Variable
Sex
Male
Female
Age
25 to <40 y
40 to <55 y
Education
High school
Higher education
Skin color
Fair
Medium
Dark
Previous occupations
Mainly outdoors
Indoors and outdoors
Mainly indoors
Nevi on back
0
1–10
≥11
History of skin cancer
No
Yes
Body mass index
<25.0 kg/m2
25.0–29.9 kg/m2
≥30.0 kg/m2
Smoking status
Never smoker
Former smoker
Current smoker
Overall
Odds Ratio (95% CI) Relative Odds (95% CI) Approximate P Value
for Interaction*
Favors Daily Sunscreen Favors Discretionary
Sunscreen
Daily Sunscreen
1.35 (1.05–1.75)
1.10 (0.86–1.39)
1.24 (0.96–1.60)
1.07 (0.83–1.37)
1.01 (0.79–1.28)
1.49 (1.17–1.91)
1.04 (0.85–1.29)
1.58 (1.16–2.15)
0.95 (0.58–1.55)
1.16 (0.87–1.53)
1.21 (0.88–1.66)
1.21 (0.96–1.54)
0.82 (0.56–1.21)
1.27 (1.04–1.57)
1.22 (0.87–1.72)
1.20 (1.00–1.44)
1.14 (0.69–1.87)
1.17 (0.91–1.50)
1.21 (0.90–1.64)
1.02 (0.64–1.63)
1.17 (0.93–1.48)
1.21 (0.91–1.61)
1.35 (0.88–2.08)
1.18 (0.99–1.39)
Discretionary
Sunscreen
1.39 (1.05–1.86)
1.67 (1.30–2.15)
1.55 (1.17–2.05)
1.43 (1.10–1.87)
1.58 (1.19–2.11)
1.71 (1.32–2.20)
1.73 (1.35–2.22)
1.33 (0.99–1.78)
1.42 (0.67–2.99)
1.32 (0.84–2.06)
1.58 (1.15–2.17)
1.75 (1.36–2.27)
1.33 (0.84–2.12)
1.58 (1.25–1.98)
1.58 (1.09–2.30)
1.54 (1.26–1.90)
1.63 (1.05–2.50)
1.68 (1.29–2.19)
1.44 (1.02–2.02)
1.79 (1.01–3.17)
1.36 (1.05–1.77)
1.59 (1.14–2.23)
2.74 (1.85–4.06)
1.54 (1.28–1.86)
0.97 (0.66–1.42)
0.66 (0.46–0.93)
0.80 (0.55–1.17)
0.74 (0.52–1.07)
0.64 (0.44–0.93)
0.88 (0.62–1.24)
0.60 (0.44–0.83)
1.19 (0.78–1.83)
0.67 (0.30–1.51)
0.88 (0.52–1.49)
0.77 (0.49–1.19)
0.69 (0.49–0.98)
0.62 (0.34–1.13)
0.81 (0.59–1.10)
0.77 (0.47–1.27)
0.78 (0.59–1.02)
0.70 (0.36–1.34)
0.69 (0.48–1.00)
0.84 (0.54–1.32)
0.57 (0.27–1.19)
0.86 (0.61–1.22)
0.76 (0.49–1.18)
0.49 (0.28–0.88)
0.76 (0.59–0.98)
0.30
0.79
0.22
0.106
0.49
0.95
0.89
0.66
0.197
10.01.00.1
*Pvalues for heterogeneity of effects were derived using score tests from generalized estimating equations, assuming independent correlations between
repeated measures of skin microtopography grades.
Original Research Sunscreen and Skin Aging
788 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 www.annals.org

months, the authors reported a decrease in crow’s feet
wrinkles in the 15 women randomly assigned to 30-mg
capsules. However, because of this study’s (45) method-
ological limitations, including very small sample size, short
duration, lack of controls, and possible confounding by
sunscreen use, its findings are difficult to interpret and
cannot be compared with those from our long-term con-
trolled trial. Our results show a lack of effect of
␤
-carotene
and are unlikely to be explained by nonadherence to tablet
consumption, because photometric measurements of skin
color confirmed that the group receiving supplements
maintained significantly higher amounts of dermal
␤
-carotene than the placebo group.
Our study has limitations. One third of the partici-
pants had only 1 microtopography grade (mostly baseline).
A standard repeated-measures analysis would remove these
participants and reduce power, whereas a GEE opti-
mizes power by using all available data. Although pho-
toaging on the neck and sunburns were associated with
having only 1 microtopography grade, these factors are
unlikely to have affected trial findings because treatment
allocation was not associated with missing grades; more-
over, these factors were controlled for in the statistical
model. Baseline grade was unrelated to missing follow-up
grade and a complete case analysis, and estimates from
multiple imputation replicated the results presented in
Table 4. Measurement error occurred in study variables,
including sun exposure by questionnaire and assessment
of microtopography grades; however, it seemed to be
nondifferential with respect to treatment groups, partic-
ularly for microtopography, because assessors were blinded
to allocations.
Our sample size was determined by practical con-
straints. Although our estimate of the effect of
␤
-carotene
relative to placebo was 0.95 and was bounded by reason-
ably tight and symmetrical confidence limits implying no
effect, the lack of precision around this estimate leaves
open the possibility of
␤
-carotene supplementation having
either a protective effect (in the less severely aged sub-
group) or a small but harmful effect (in the severely aged
subgroup) on skin aging. Future research is needed to ver-
ify the effect of
␤
-carotene in persons with varying levels of
skin aging at baseline.
We conclude that regular sunscreen use by young and
middle-aged adults younger than 55 years can retard skin
aging. Although our study did not identify an effect of
␤
-carotene supplementation on skin aging, a small slowing
or accelerating effect cannot be ruled out.
From Queensland Institute of Medical Research and University of
Queensland, School of Population Health, Queensland, Australia, and
University of Manchester, Manchester Academic Health Sciences Cen-
tre, Manchester, United Kingdom.
Acknowledgment: The authors thank Mr. Toan Luong, who graded the
silicone replicas.
Financial Support: By the National Health and Medical Research Coun-
cil of Australia (NHMRC #922608), Ross Cosmetics, and Roche Vita-
mins and Fine Chemicals.
Potential Conflicts of Interest: Disclosures can be viewed at www
.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum
⫽M12-2280.
Reproducible Research Statement: Study protocol: See reference 25; also
available from Dr. Green (address below). Data set and statistical code:
Available from Dr. Green (address below).
Corresponding Author: Ade`le C. Green, MBBS, PhD, Queensland In-
stitute of Medical Research, Locked Bag 2000, Royal Brisbane Hospital,
QLD 4029, Australia.
Current author addresses and author contributions are available at www
.annals.org.
References
1. Barrett J. New secrets for youthful skin. Newsweek. 2006;147:74, 76. [PMID:
16669542]
2. Brandt FS, Cazzaniga A, Hann M. Cosmeceuticals: current trends and market
analysis. Semin Cutan Med Surg. 2011;30:141-3. [PMID: 21925366]
3. Yaar M, Gilchrest BA. Aging of skin. In: Freedberg IM, Eisen AZ, Wolff K,
Austen KF, Goldsmith LA, Katz SI, et al, eds. Fitzpatrick’s Dermatology in
General Medicine. 5th ed. New York: McGraw Hill; 1999:1697-706.
4. Green AC, Hughes MC, McBride P, Fourtanier A. Factors associated with
premature skin aging (photoaging) before the age of 55: a population-based
study. Dermatology. 2011;222:74-80. [PMID: 21196710]
5. Griffiths CE. The clinical identification and quantification of photodamage.
Br J Dermatol. 1992;127 Suppl 41:37-42. [PMID: 1390185]
6. Montagna W, Kirchner S, Carlisle K. Histology of sun-damaged human skin.
J Am Acad Dermatol. 1989;21:907-18. [PMID: 2808826]
7. Kligman AM, Kligman LH. Geriatric dermatology. In: Freedberg IM, Eisen
AZ, Wolff K, Austen KF, Goldsmith LA, Katz SI, et al, eds. Fitzpatrick’s Der-
matology in General Medicine. 5th ed. New York: McGraw Hill; 1999:1717-23.
8. Rabe JH, Mamelak AJ, McElgunn PJ, Morison WL, Sauder DN. Photoag-
ing: mechanisms and repair. J Am Acad Dermatol. 2006;55:1-19. [PMID:
16781287]
9. Foote JA, Harris RB, Giuliano AR, Roe DJ, Moon TE, Cartmel B, et al.
Predictors for cutaneous basal- and squamous-cell carcinoma among actinically
damaged adults. Int J Cancer. 2001;95:7-11. [PMID: 11241303]
10. Trautinger F. Mechanisms of photodamage of the skin and its functional
consequences for skin ageing. Clin Exp Dermatol. 2001;26:573-7. [PMID:
11696060]
11. Ichihashi M, Ueda M, Budiyanto A, Bito T, Oka M, Fukunaga M, et al.
UV-induced skin damage. Toxicology. 2003;189:21-39. [PMID: 12821280]
12. Marks R, Edwards C. The measurement of photodamage. Br J Dermatol.
1992;127 Suppl 41:7-13. [PMID: 1390188]
13. Adhoute H, de Rigal J, Marchand JP, Privat Y, Leveque JL. Influence of age
and sun exposure on the biophysical properties of the human skin: an in vivo
study. Photodermatol Photoimmunol Photomed. 1992;9:99-103. [PMID:
1300143]
14. de Rigal J, Escoffier C, Querleux B, Faivre B, Agache P, Le´veˆque JL.
Assessment of aging of the human skin by in vivo ultrasonic imaging. J Invest
Dermatol. 1989;93:621-5. [PMID: 2677155]
15. Beagley J, Gibson IM. Changes in Skin Condition in Relation to Degree of
Exposure to Ultraviolet Light. Perth, Western Australia: Western Australia Insti-
tute of Technology, School of Biology; 1980.
16. Holman CD, Armstrong BK, Evans PR, Lumsden GJ, Dallimore KJ,
Meehan CJ, et al. Relationship of solar keratosis and history of skin cancer to
objective measures of actinic skin damage. Br J Dermatol. 1984;110:129-38.
[PMID: 6696833]
Original ResearchSunscreen and Skin Aging
www.annals.org 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 789

17. Fisher GJ, Kang S, Varani J, Bata-Csorgo Z, Wan Y, Datta S, et al. Mech-
anisms of photoaging and chronological skin aging. Arch Dermatol. 2002;138:
1462-70. [PMID: 12437452]
18. Uitto J. The role of elastin and collagen in cutaneous aging: intrinsic aging
versus photoexposure. J Drugs Dermatol. 2008;7:s12-6. [PMID: 18404866]
19. Battistutta D, Pandeya N, Strutton GM, Fourtanier A, Tison S, Green AC.
Skin surface topography grading is a valid measure of skin photoaging. Photoder-
matol Photoimmunol Photomed. 2006;22:39-45. [PMID: 16436180]
20. Hughes MC, Bredoux C, Salas F, Lombard D, Strutton GM, Fourtanier A,
et al. Comparison of histological measures of skin photoaging. Dermatology.
2011;223:140-51. [PMID: 21997520]
21. Hughes MC, Strutton GM, Fourtanier A, Green AC. Validation of skin
surface microtopography as a measure of skin photoaging in a subtropical popu-
lation aged 40 and over. Photodermatol Photoimmunol Photomed. 2012;28:
153-8. [PMID: 22548398]
22. Antoniou C, Kosmadaki MG, Stratigos AJ, Katsambas AD. Photoaging:
prevention and topical treatments. Am J Clin Dermatol. 2010;11:95-102.
[PMID: 20141230]
23. Sambandan DR, Ratner D. Sunscreens: an overview and update. J Am Acad
Dermatol. 2011;64:748-58. [PMID: 21292345]
24. Boyd AS, Naylor M, Cameron GS, Pearse AD, Gaskell SA, Neldner KH.
The effects of chronic sunscreen use on the histologic changes of dermatoheliosis.
J Am Acad Dermatol. 1995;33:941-6. [PMID: 7490363]
25. Green A, Battistutta D, Hart V, Leslie D, Marks G, Williams G, et al. The
Nambour Skin Cancer and Actinic Eye Disease Prevention Trial: design and
baseline characteristics of participants. Control Clin Trials. 1994;15:512-22.
[PMID: 7851112]
26. Baumann L. How to prevent photoaging? J Invest Dermatol. 2005;125:xii-
xiii. [PMID: 16185258]
27. Darlington S, Williams G, Neale R, Frost C, Green A. A randomized
controlled trial to assess sunscreen application and beta carotene supplementation
in the prevention of solar keratoses. Arch Dermatol. 2003;139:451-5. [PMID:
12707092]
28. Green A, Williams G, Neale R, Hart V, Leslie D, Parsons P, et al. Daily
sunscreen application and betacarotene supplementation in prevention of basal-
cell and squamous-cell carcinomas of the skin: a randomised controlled trial.
Lancet. 1999;354:723-9. [PMID: 10475183]
29. Green AC, Williams GM, Logan V, Strutton GM. Reduced melanoma after
regular sunscreen use: randomized trial follow-up. J Clin Oncol. 2011;29:257-63.
[PMID: 21135266]
30. Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized
estimating equation approach. Biometrics. 1988;44:1049-60. [PMID: 3233245]
31. Parsons NR. Repeated measures proportional odds logistic regression analysis
of ordinal score data in the statistical software package R. Computational Statis-
tics & Data Analysis. 2009;53:632-41.
32. Enders CK. Preserving interaction effects. In: Applied Missing Data Analysis.
New York: Guilford Pr; 2010:265-8.
33. Stern RS. Clinical practice. Treatment of photoaging. N Engl J Med. 2004;
350:1526-34. [PMID: 15071127]
34. Kricker A, Armstrong BK, English DR, Heenan PJ. Pigmentary and cuta-
neous risk factors for non-melanocytic skin cancer—a case-control study.
Int J Cancer. 1991;48:650-62. [PMID: 2071226]
35. Hirst NG, Gordon LG, Scuffham PA, Green AC. Lifetime cost-effectiveness
of skin cancer prevention through promotion of daily sunscreen use. Value
Health. 2012;15:261-8. [PMID: 22433757]
36. Diffey BL. Sunscreens and UVA protection: a major issue of minor impor-
tance. Photochem Photobiol. 2001;74:61-3. [PMID: 11460538]
37. Liu W, Wang X, Lai W, Yan T, Wu Y, Wan M, et al. Sunburn protection
as a function of sunscreen application thickness differs between high and low
SPFs. Photodermatol Photoimmunol Photomed. 2012;28:120-6. [PMID:
22548392]
38. Yaar M, Gilchrest BA. Photoageing: mechanism, prevention and therapy. Br
J Dermatol. 2007;157:874-87. [PMID: 17711532]
39. Martires KJ, Fu P, Polster AM, Cooper KD, Baron ED. Factors that affect
skin aging: a cohort-based survey on twins. Arch Dermatol. 2009;145:1375-9.
[PMID: 20026845]
40. Purba MB, Kouris-Blazos A, Wattanapenpaiboon N, Lukito W, Rothen-
berg E, Steen B, 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. [PMID: 11443024]
41. 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. [PMID: 7832266]
42. Kadunce 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. [PMID: 2014944]
43. Lucas RM, Ponsonby AL, Dear K, Taylor BV, Dwyer T, McMichael AJ,
et al. Associations between silicone skin cast score, cumulative sun exposure, and
other factors in the ausimmune study: a multicenter Australian study. Cancer
Epidemiol Biomarkers Prev. 2009;18:2887-94. [PMID: 19843682]
44. Suppa M, Elliott F, Mikeljevic JS, Mukasa Y, Chan M, Leake S, et al. The
determinants of periorbital skin ageing in participants of a melanoma case-control
study in the U.K. Br J Dermatol. 2011;165:1011-21. [PMID: 21787368]
45. Cho S, Lee DH, Won CH, Kim SM, Lee S, Lee MJ, et al. Differential
effects of low-dose and high-dose beta-carotene supplementation on the signs of
photoaging and type I procollagen gene expression in human skin in vivo. Der-
matology. 2010;221:160-71. [PMID: 20516658]
Original Research Sunscreen and Skin Aging
790 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 www.annals.org

Current Author Addresses: Ms. Hughes and Dr. Green: Queensland
Institute of Medical Research, Locked Bag 2000, Royal Brisbane Hospi-
tal, QLD 4029, Australia.
Drs. Williams and Baker: Level 2, Public Health Building, School of
Population Health, University of Queensland, Herston Road, Herston,
QLD 4006, Australia.
Author Contributions: Conception and design: G.M. Williams, A.C.
Green.
Analysis and interpretation of the data: M.C.B. Hughes, G.M. Williams,
P. Baker, A.C. Green.
Drafting of the article: M.C.B. Hughes, P. Baker, A.C. Green.
Critical revision of the article for important intellectual content: M.C.B.
Hughes, G.M. Williams, A.C. Green.
Final approval of the article: M.C.B. Hughes, G.M. Williams, A.C.
Green.
Provision of study materials or patients: A.C. Green.
Statistical expertise: G.M. Williams, P. Baker.
Obtaining of funding: A.C. Green.
Administrative, technical, or logistic support: P. Baker, A.C. Green.
Collection and assembly of data: M.C.B. Hughes, G.M. Williams, A.C.
Green.
www.annals.org 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 W-321

Appendix Figure. Study flow diagram of the Nambour
␤
-carotene and photoaging study, 1992–1996.
With good-quality skin surface replica in 1992 and/or 1996
(n = 447)
Good-quality skin surface replica in 1992 and 1996: 310
Good-quality skin surface replica in 1992 only: 96
Good-quality skin surface replica in 1996 only: 41
With good-quality skin surface replica in 1992 and/or 1996
(n = 439)
Good-quality skin surface replica in 1992 and 1996: 294
Good-quality skin surface replica in 1992 only: 117
Good-quality skin surface replica in 1996 only: 28
Randomly assigned
(n = 1621)
Assigned to -carotene
(n = 820)
Eligible for photoaging study (n = 457)
Residents of Nambour invited to attend
(n = 1850)
Did not attend survey (n = 203)
Attended baseline survey and were eligible
(n = 1647)
Analysis Follow-up Allocation Enrollment
Excluded (n = 363)
Age ≥55 y: 362
Not white: 1
Excluded (n = 355)
Age ≥55 y: 354
Not white: 1
Excluded (n = 26)
Declined: 22
Not examined: 4
With skin surface replica in 1992 and/or 1996
(n = 451)
No skin surface replica in 1992
and 1996 (n = 6)
No replica in 1992 (n = 20)
Died (n = 1)
Active participants did not
attend 1996 skin examination
(n = 17)
No replica in 1996 (n = 6)
Became passive participants
during follow-up (n = 61)
No skin surface replica in 1992
and 1996 (n = 4)
No replica in 1992 (n = 18)
Died (n = 2)
Active participants did not
attend 1996 skin examination
(n = 9)
No replica in 1996 (n = 8)
Became passive participants
during follow-up (n = 84)
Assigned to placebo
(n = 801)
Eligible for photoaging study (n = 446)
With skin surface replica in 1992 and/or 1996
(n = 442)
Poor-quality replica in 1992 and no
replica in 1996 (n = 3)
No replica in 1992 and poor-quality
replica in 1996 (n = 1)
Poor-quality replica in 1992 (n = 21)
Poor-quality replica in 1996 (n = 11)
No replica in 1992 and poor-quality
replica in 1996 (n = 3)
Poor-quality replica in 1992 (n = 10)
Poor-quality replica in 1996 (n = 14)
W-322 4 June 2013 Annals of Internal Medicine Volume 158 • Number 11 www.annals.org