TABLE 3.

Randomized, double-blind^*, vehicle-controlled, whole- or split-face design trials of retinols in facial skin aging

STUDY (JADAD SCORE)	INTERVENTION WITH NUMBER OF SUBJECTS	TREATMENT DURATION	OUTCOME MEASURES	RESULTS	STRENGTHS	WEAKNESSES/LIMITATIONS
Creidi et al.27 (Jadad score: 3)	Whole-face design Retinaldehyde 0.05%: 40 subjects Vehicle: 45 subjects	44 weeks	Optical profilometry of crow’s feet area, assessing for deepness and roughness of fine wrinkles	No statistically significant differences in retinaldehyde versus vehicle for both outcome measures	Active comparator with retinoic acid 0.05%: 40 subjects (internal validation) Objective outcome measure Demographic data and baseline values provided	No primary endpoint designation No power calculation Completer analysis
Green et al.28 (Jadad score: 3)	Whole-face design 80 subjects enrolled Retinyl propionate 0.15% Vehicle	48 weeks	Clinical assessment, subject self-assessment, photography, and optical profilometry of right crow’s feet area	No statistically significant differences in retinyl propionate versus vehicle for all outcome measures	Subjective and objective outcome measures Detailed reporting of outcome results with baseline values	No primary endpoint designation No power calculation No group-size numbers Completer analysis No correction for multiple comparisons but alpha set low at 0.005
Tucker-Samaras et al.21 (Jadad score: 3)	Split-face design (multiple products tested) Retinol 0.1%: 37 subjects Vehicle: 29 subjects	8 weeks	Clinical assessment with photography of 12 skin parameters, using a scoring system of 0 to 9	Statistically significant improvements (p<0.05) on retinol versus vehicle in 10 of the 12 clinically-assessed skin parameters	Absolute data as well as absolute and percentage changes from baseline provided	No primary endpoint designation No power calculation Completer analysis No baseline information No standard deviations or confidence intervals No correction for multiple comparisons
Kikuchi et al.24 (Jadad score: 3)	Split-face design Retinol 0.075%/vehicle: 54 subjects	26 weeks	Clinical assessment with photography of fine wrinkles, deep wrinkles, and pigmentation Rating scale: 0, no change; 1, slightly better; 2, better; and 3, far better	Statistically significant improvements (p<0.05) in retinol versus vehicle in fine wrinkles as of Week 4 and in deep wrinkles as of Week 26	Results reported in detail, including efficacy and tolerability (i.e., erythema, swelling, scaling)	No primary endpoint No power calculation No baseline information Completer analysis Power artificially increased by analysing observations rather than subjects No correction for multiple comparisons
Bellemère et al.20 (Jadad score: 3)	Whole-face design Retinol 0.1%: 24 subjects Vehicle: 24 subjects	36 weeks	Clinical assessment with high-resolution digital imaging of eye wrinkles, fine lines, brown spots, and skin-tone evenness (12-cm visual analog scale); three-dimensional profilometry of crow’s feet area	Statistically significant improvements (p<0.05) in retinol versus vehicle in eye wrinkles as of Week 12 and in fine lines as of Week 24	Subjective and objective outcome measures but profilometry results not provided Absolute data presented, including baseline values	No primary endpoint designation No power calculation Unclear whether analysis is based on intention-to-treat or completers No adjustment for multiple comparisons No brown-spot or profilometry results
Kim et al.13 (Jadad score: 4)	Split-face design Retinyl retinoate 0.06%/ Lotion P: 24 subjects	12 weeks	Clinical assessment of periorbital wrinkles using a 0-7 scoring system, subject self-assessment, and visiometry of crow’s feet area	Statistically significant improvemaent (p=0.031) retinyl retinoate versus placebo in periorbital wrinkles as of Week 12	Subjective and objective outcome measures provided but visiometry results not provided Absolute data presented, including baseline values, with (?standard) deviations and p-values but no numbers of observations	No primary endpoint designation No power calculation Unclear whether analysis is based on intention-to-treat or completers No adjustment for multiple comparisons No visiometry results
Lee et al.14 (Jadad score: N/A)	Split-face design Retinyl N-formyl aspartamate/vehicle: 24 subjects	24 weeks	Clinical assessment and subject self-assessment of crow’s feet area Rating scale: 0, no improvement; 1, mild improvement; 2, moderate improvement; 3, remarkable improvement Visiometry	Statistically significant improvement retinyl N-formyl aspartamate versus vehicle in crow’s feet area as of Week 24; visiometry showed statistically significant improvement (p=0.0014) versus vehicle for skin roughness as of Week 24	Subjective and objective outcome measures Bonferroni adjustment for multiple comparisons Absolute data presented for visiometry, including baseline values, standard deviations, and p-values	No primary endpoint designation No power calculation No baseline information for clinical assessment Rater not blinded as to treatment (single-blind) Completer analysis
Randhawa et al.25 (Jadad score: 3)	Whole-face design Retinol 0.1%: 35 subjects Vehicle: 32 subjects	52 weeks	Clinical assessment with photography of eight skin parameters, using a scoring system of 0 to 9, and subject self-assessment	Statistically significant improvements (p≤0.05) retinol versus vehicle in all eight skin parameters as of Week 24	N/A	No primary endpoint designation No power calculation Randomization unclear No baseline information Unclear whether analysis is based on intention-to-treat or completers No adjustment for multiple comparisons No results presented for vehicle Retinol results presented without standard deviations or confidence intervals
Gold et al.22 (Jadad score: 3)	Whole-face design Retinol 0.5%: 22 subjects Vehicle: 8 subjects	8 weeks	Clinical assessment of photodamage, wrinkles, ultraviolet spots, and photoaging and subject self-assessment of facial skin quality	No statistically significant differences in retinol versus vehicle for all outcome measures	Bonferroni adjustment for multiple comparisons	No primary endpoint designation No power calculation No baseline information

*The trial by Lee et al.14 is was a single-blind study, i.e., the subjects but not the evaluator(s) were blinded.