Intervention Review

You have free access to this content

Vitamin A and fish oils for retinitis pigmentosa

  1. Sobharani Rayapudi1,
  2. Stephen G Schwartz2,*,
  3. Xue Wang1,
  4. Pamela Chavis3

Editorial Group: Cochrane Eyes and Vision Group

Published Online: 19 DEC 2013

DOI: 10.1002/14651858.CD008428.pub2


How to Cite

Rayapudi S, Schwartz SG, Wang X, Chavis P. Vitamin A and fish oils for retinitis pigmentosa. Cochrane Database of Systematic Reviews 2013, Issue 12. Art. No.: CD008428. DOI: 10.1002/14651858.CD008428.pub2.

Author Information

  1. 1

    Johns Hopkins Bloomberg School of Public Health, Department of Epidemiology, Baltimore, Maryland, USA

  2. 2

    Bascom Palmer Eye Institute, Miami, Florida, USA

  3. 3

    Medical University of South Carolina, Department of Ophthalmology, Charleston, South Carolina, USA

*Stephen G Schwartz, Bascom Palmer Eye Institute, Miami, Florida, USA. sschwartz2@med.miami.edu.

Publication History

  1. Publication Status: New
  2. Published Online: 19 DEC 2013

SEARCH

 
Characteristics of included studies [ordered by study ID]

Methods
  • Study design: RCT, 2x2 factorial design
  • Number randomized: 601
  • Vitamin A group = 146; trace group = 149; vitamin A + vitamin E group = 151; vitamin E group = 155
  • Number analyzed: 572
  • Enrolment period: May 1984-June 1987
  • Length of follow-up: planned = 4 years; actual = 4 years for all participants and 5.2 years for a proportion of participants
  • Sample size estimation: calculated sample size based on expected rate of decline in ERG amplitude by 4% among participants taking supplements that would preserve a detectable ERG response for 5 additional years; 95% power


Participants
  • Country: USA and Canada
  • Age (mean ± SD): vitamin A = 32.5 ± 0.7 years; trace = 32.2 ± 0.7 years; vitamin A + E = 32.3 ± 0.6 years; vitamin E = 31.5 ± 0.6 years
  • Sex: both men and women included (62% men and 38% women)
  • Key inclusion criteria:
    • Common forms of retinitis pigmentosa; participants aged 18-49 years; 1 study participant per family
    • Snellen visual acuity ≥ 20/100
    • Visual field ≥ 8° diameter to V-4-e white test light on Goldman perimeter
    • 30 Hz ERG ≥ 0.12 μV or 0.5 Hz ERG ≥ 2.5 μV in at least 1 eye
    • People with Usher's syndrome type 2 (i.e. RP with mild congenital deafness) were included
  • Key exclusion criteria:
    • Atypical forms of RP (including paravenous RP,clumped pigmentary retinal degeneration, sector RP, or unilateral RP)
    • Posterior subcapsular cataracts > 11% of the total lens area in both eyes
    • X-linked carrier, Refsum disease, Bardet-Biedl syndrome, or Usher's syndrome type 1 (i.e. RP with profound congenital deafness)
    • RP associated with hereditary abetalipoproteinemia (i.e. Bassen-Kornzweig syndrome)
    • Weight < the 5th percentile of weight for a given age, sex, and height
    • Serum retinol level of ≥ 3.50 pmol/L (100 μg/dL), serum a-tocopherol levels > the normal range adjusted for serum triglyceride level, total cholesterol level, and sex
    • Total estimated intake of preformed vitamin A in diet plus pills > 11500 IU/d
    • Total estimated intake of vitamin E in diet plus pills > 40 IU/d
    • Pregnancy, lactation
    • Diseases affecting the absorption or metabolism of vitamins A or E
  • Genetic profile
    • Autosomal dominant - no (%) vitamin A = 45 (31%); trace group = 40 (27%); vitamin A + E = 40 (26%); vitamin E = 40 (26%)
    • Autosomal recessive- no (%) vitamin A = 22 (15%); trace group = 22 (15%); vitamin A + E = 26 (17%); vitamin E = 29 (19%)
    • X-linked no (%) vitamin A = 15 (10%); trace group = 17 (11%); vitamin A + E = 12 (8%); vitamin E = 12 (8%)
    • Isolate no (%) vitamin A = 59 (40%); trace group = 62 (42%); vitamin A + E = 62 (41%); vitamin E = 63 (41%)
    • Undetermined no (%) vitamin A = 5 (3%); trace group = 8 (5%); vitamin A + E = 11 (7%); vitamin E = 11 (7%)
  • Baseline clinical status: retinal arteriolar attenuation, elevated dark adaption thresholds, reduced ERGs with delayed b wave implicit times. 94% had intraretinal pigment in the mid-peripheral fundus
  • Comparability of baseline characteristics: comparable


Interventions
  • Group A: vitamin A 15000 IU/d, plus vitamin E 3 IU/d
  • Trace group: vitamin A 75 IU/d, plus vitamin E 3 IU/d
  • Group A + E: vitamin A 15000 IU/d, plus vitamin E 400 IU/d
  • Group E: vitamin A 75 IU/d, plus vitamin E 400 IU/d


  • Vitamin A administered as retinyl palmitate and vitamin E as di-a-tocopherol in soft gelatin capsules


Outcomes
  • Primary outcome of the study: cone ERG response amplitudes to 30 Hz flicker stimulus
  • Visual field assessed using Kinetic Goldmann perimeter
  • Visual acuity assessed using transilluminated ETDRS charts at a distance of 3.2 m
  • ERG: first, single flashes of white light (3.8 log foot-lamberts) presented every 2 seconds (i.e. 0.5 Hz flashes) were used to elicit mixed cone-rod responses, then flashes of the same white light presented at 30/second (30 Hz) were used to obtain cone-isolated responses. Responses to 0.5 Hz flashes were summed as 2 waveforms of 32 responses each by the computer; amplitudes as low as 1 μ could be detected. Responses to 30 Hz flashes of flickering light were recorded as 8 consecutive waveforms of 256 summations each. Responses to brief (10 μs in duration) full-field flashes of light were amplified, and summed on a computer
  • Frequency outcomes assessed: at baseline, and annually for 4-6 years depending on the length of follow-up
  • Losses to follow-up: 29/601 (5%); 4 participants died, 25 declined to continue participation
  • Adverse events: none reported


Notes
  • Funding sources:National Eye Institute (NEI), RP Foundation Fighting Blindness, vitamin capsules provided by Hoffman-LaRoche Inc, Nutley, NJ
  • Statistical analyses: appropriate; factorial analysis of variance was used to evaluate main and interaction effects of vitamin A and E
  • Subgroup analyses: post hoc analysis conducted in higher amplitude cohort (cohort of participants with cone ERG > 0.68 μV (354 participants)).


Notes: the trial was originally planned to allow for 4 years of follow-up for each participant. However, due to slow recruitment (required 3 years), follow-up was continued on all participants until the last randomized participants had completed their 4th year of follow-up. The DSMB recommended cessation of this protocol in September 1991 because by then all participants had completed their 4th year of follow-up and additional follow-up data would probably not lead to conclusions that would be substantially more precise. The smaller sample sizes at year 5 (n = 472) and year 6 (n = 261) reflect the fact that the study was stopped after the last 4th-year follow-up visit. The mean duration of follow-up was 5.2 years for all randomized participants.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk“A separate set of randomization assignments was maintained for each stratum based on a computer-generated set of random numbers to facilitate the above randomization” (p 764)

Allocation concealment (selection bias)Low risk“Each bottle contained 100 capsules labeled with a lot number and instructions for storage at room temperature but not labeled as to content.” (p 763)

Blinding of ParticipantsLow risk“Patients did not know the contents of the supplements under study or their group assignment and also agreed not to know the course of their retinal degeneration until the end of the study” (p 764)

Blinding of CaregiversLow risk“All members of the staff in contact with the patients, including the principal investigator (E.L.B.), were masked as to the treatment group assignment of each patient.” (p 764)

Blinding outcome assessors for primary outcome (visual field)Low risk“All members of the staff in contact with the patients, were masked as to the treatment group assignment of each patient. Each ocular examination and ERG was performed without review of previous records” (p 764)

Blinding outcome assessors for secondary outcome (visual acuity)Low risk“All members of the staff in contact with the patients were masked as to the treatment group assignment of each patient. Each ocular examination and ERG was performed without review of previous records” (p 764)

Blinding outcome assessors for secondary outcome (ERG)Low risk“Each ocular examination and ERG was performed without review of previous records.” (p 764)

Incomplete outcome data addressed for primary outcome (visual field)Low risk“Only 5% (29/601) of patients failed to complete this study; four of these patients died and 25 patients declined to continue participation, most after the fourth year.” (p 770)

Incomplete outcome data addressed for secondary outcome (visual acuity)Low risk“Only 5% (29/601) of patients failed to complete this study; four of these patients died and 25 patients declined to continue participation, most after the fourth year.” (p 770)

Incomplete outcome data addressed for secondary outcome (ERG)Low risk“Only 5% (29/601) of patients failed to complete this study; four of these patients died and 25 patients declined to continue participation, most after the fourth year.” (p 770)

Selective reporting (reporting bias)Unclear riskInsufficient information to permit judgment, as the study protocol is not available

Other biasLow riskWe did not detect other bias


Methods
  • Study design: RCT, parallel design
  • Number randomized: 221 (numbers of participants in each group were not specified)
  • Number analyzed: 208 (DHA + vitamin A group = 105; control + vitamin A group = 103)
  • Enrolment period: 1 year (13 May 1996-26 September 1997)
  • Length of follow-up: planned = 4 years; actual = 4 years
  • Sample size estimation: estimated that 220 participants were needed to provide sufficient power to observe a statistically significant difference (29 dB) between mean change in the DHA + vitamin A and control + vitamin A groups with respect to HFA 30-2 total point score over a 4-year interval and allowing for 5% attrition; 90% power


Participants
  • Country: USA
  • Age (mean ± SD): DHA + vitamin A = 37.8 ± 0.90 years; control + vitamin A = 36.0 ± 1.00 years
  • Sex: both men and women included (49% women)
  • Key inclusion criteria:
    • Common forms of RP; participants aged 18-55 years; 1 study participant per family
    • Snellen visual acuity ≥ 20/100
    • HFA 30-2 program total point score ≥ 250 dB
    • 30 Hz cone ERG amplitude of ≥ 0.68 uVin at least one eye
    • included participants with Usher's syndrome type 2 (i.e. RP with mild congenital deafness)
  • Key exclusion criteria:
    • Atypical forms of RP (including paravenous RP, clumped pigmentary retinal degeneration, sector RP, or unilateral RP)
    • Posterior subcapsular cataracts > 11% of the total lens area in both eyes
    • X-linked carrier, Refsum disease, Bardet-Biedl syndrome, or Usher's syndrome type 1 (i.e. RP with profound congenital deafness); RP associated with hereditary abetalipoproteinemia (i.e. Bassen-Kornzweig syndrome)
    • Weight < the 5th percentile of weight for a given age, sex, and height
    • Serum retinol level of ≥ 3.50 pmol/L (100 μg/dL), serum cholesterol levels > 300mg/dl total, total estimated intake of preformed vitamin A in diet plus pills > 5000 IU/d, average or vitamin E intake > 30 IU/d
    • Pregnancy, lactation
    • Diseases affecting the absorption or metabolism of vitamins A or DHA
  • Genetic profile:
    • Autosomal dominant - no (%) DHA + vitamin A = 20 (19); control + vitamin A = 19 (18)
    • Autosomal recessive- no (%) DHA + vitamin A = 16 (15); control + vitamin A = 13 (13)
    • X-linked no (%) DHA + vitamin A = 5 (5); control + vitamin A = 8 (8)
    • Isolate no (%) DHA + vitamin A = 50 (48); control + vitamin A = 50 (49)
    • Others no (%) DHA + vitamin A = 8 (8); control + vitamin A = 9 (9)
  • Baseline clinical status: retinal arteriolar attenuation, elevated dark adaption thresholds, reduced ERGs with delayed b wave implicit times. 97% had intraretinal pigment in the mid-peripheral fundus. 50% of participants in the DHA + vitamin A group and 64% of participants in the control + vitamin A group had cataracts in at least 1 eye at baseline. 11% of participants reported partial hearing loss
  • Comparability of baseline characteristics: comparable


Interventions
  • DHA + A group: 1200 mg/d DHA + 15000 IU/d vitamin A
  • Control + A group: 15000 IU/d vitamin A


Each participant received either 6 capsules/day each containing 500 mg of fatty acids (200 mg of which was DHA, for a total of 1200 mg/d of DHA), or 6 placebo capsules/day containing 500 mg of fatty acids with no DHA, for 4 years. Vitamin A was administered as retinyl palmitate.


Outcomes
  • Primary outcome of the study: visual field - static perimetric sensitivity with the 30-2 program of the Humphrey field analyzer
  • Visual field assessed using Humphrey field analyzer size V target
  • Visual acuity assessed using transilluminated ETDRS charts
  • ERG: first, single flashes of white light (3.8 log foot-lamberts) presented every 2 seconds (i.e. 0.5 Hz flashes) were used to elicit mixed cone-rod responses, then flashes of the same white light presented at 30 per second (30 Hz) were used to obtain cone-isolated responses. Responses to 0.5 Hz flashes were summed as 2 waveforms of 32 responses each by the computer; amplitudes as low as 1 μ could be detected. Responses to 30 Hz flashes of flickering light were recorded as 8 consecutive waveforms of 256 summations each. Responses to brief (10 μs in duration) full-field flashes of light were amplified, and summed on a computer
  • Outcomes assessed at baseline, and annually for 4 years
  • Losses to follow-up: 13/221 (5%); 1 participant died of breast cancer
  • Adverse events: none reported


Notes
  • Funding sources: National Eye Institute (NEI), RP Foundation Fighting Blindness, vitamin capsules provided by Martek
  • Statistical analyses: appropriate; intention-to-treat analysis
  • Subgroup analyses: (post hoc) looked at subgroup of participants not taking vitamin A prior to enrollment and those taking vitamin A for 2 years prior to enrollment into the trial


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"A separate set of randomization assignments was maintained for each stratum based on a computer generated set of random numbers" (p 1299)

Allocation concealment (selection bias)Low risk" . . . computer generated set of random numbers that was available only to a programmer who provided assignment information to the data manager (C.W.D.) on a case-by-case basis. Group assignment was implemented by the data manager". (p 1299)

Blinding of ParticipantsLow risk"Patients did not know the contents of the supplement under study or their treatment group assignment and also agreed not to know the course of their retinal degeneration until the end of the study". (p 1299)

Blinding of CaregiversLow risk"All members of the staff in contact with the patients, including the principal investigator (E.L.B.), were masked with regard to each patient’s treatment group assignment" (p 1299)

Blinding outcome assessors for primary outcome (visual field)Low risk"Each ocular examination was performed without review of previous records" (p 1299)

Blinding outcome assessors for secondary outcome (visual acuity)Low risk"Treatment group assignments and plasma DHA and RBC PE DHA levels were placed in records separate from that used for ocular examinations as part of masking those in contact with the patients" (p 1299)

Blinding outcome assessors for secondary outcome (ERG)Low risk"Treatment group assignments and plasma DHA and RBC PE DHA levels were placed in records separate from that used for ocular examinations as part of masking those in contact with the patients" (p 1299)

Incomplete outcome data addressed for primary outcome (visual field)Low risk"Two hundred eight of these patients (221) completed all 4 annual follow-up visits. Analyses performed on patients with partial follow-up, but with missing values left as missing and after using multiple imputation methods to account for missing data among patients with incomplete followup". (p 1301)

Incomplete outcome data addressed for secondary outcome (visual acuity)Low risk"Two hundred eight of these patients (221) completed all 4 annual follow-up visits. Analyses performed on patients with partial follow-up, but with missing values left as missing and after using multiple imputation methods to account for missing data among patients with incomplete followup". (p 1301)

Incomplete outcome data addressed for secondary outcome (ERG)Low risk"Two hundred eight of these patients (221) completed all 4 annual follow-up visits. Analyses performed on patients with partial follow-up, but with missing values left as missing and after using multiple imputation methods to account for missing data among patients with incomplete followup". (p 1301)

Selective reporting (reporting bias)Unclear riskStudy protocol is not available

Other biasLow riskWe did not detect other bias


Methods
  • Study design: RCT, parallel design
  • Number randomized: 44 (DHA group = 23; control (placebo) group = 21)
  • Number analyzed: 41 (numbers of participants in each group were not specified)
  • Enrolment period: 20 February 1995
  • Length of follow-up: planned = 4 years; actual = 4 years
  • Sample size estimation: derived from a predicted decrease of 0.085 log units/y in cone 31 Hz flicker amplitude for the placebo group, assuming a change of 0.34 log units over 4 years in the placebo group and that reducing the rate of progression by 40% is meaningful for a two-sided test with a .05 significance level and a power of 80%


Participants
  • Country: USA
  • Age (mean ± SD): 16 ± 9 years; age range 4-38 years
  • Sex: males only
  • Key inclusion criteria:
    • RP diagnosed by an ophthalmologist specializing in retinal disease
    • History that was not only consistent with X-linked inheritance, but that also ruled out dominant or recessive RP
    • HFA 30-2 program total point score ≥ 250 dB
    • 31 Hz cone ERG amplitude of ≥ 0.68 μV in at least 1 eye
    • No excessive dietary intake of fish or supplementation with fish oil
  • Key exclusion criteria:
    • Female
    • Genetic profile: X-linked RP gene mutations (RPGR, RP24)
  • Comparability of baseline characteristics: comparable


Interventions
  • Intervention # 1: DHA group = 400mg/d
  • Intervention # 2: placebo group = corn/soy oil triglyceride


Each participant received 2 capsules/day each containing 500 mg of fatty acids (200 mg of which was DHA, for a total of 400 mg/d of DHA), or 2 placebo capsules/day containing 500 mg of fatty acids with no DHA, administered for 4 years


Outcomes
  • Primary outcome of the study: cone ERG (31-Hz electroretinogram amplitude)
  • Visual field assessed using Humphrey field analyzer size V target
  • Visual acuity assessed using transilluminated ETDRS charts
  • ERG: 30 Hz electroretinogram amplitude


  • Outcomes assessed at baseline, and annually for 4 years
  • Losses to follow-up: 3/44 (6%) in total: 2 in the placebo group and 1 in the intervention group
  • Adverse events: none reported


Notes
  • Funding sources: National Eye Institute (NEI), orphan products development program of the US Food and Drug Administration and the RP Foundation Fighting Blindness Foundation Fighting Blindness, capsules provided by Martek
  • Statistical analyses: appropriate; intention-to-treat analysis
  • Subgroup analyses: rod ERG in children < 12 years and children ≥ 12 years


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear risk“Placebo and + DHA assignments were made following a block randomization schedule (10/block).” (p 705) Method used for random sequence generation is unclear

Allocation concealment (selection bias)Low risk“Relatives were randomized together to eliminate a potential for mixing of capsules; there were five sib-pairs in each cohort” (p 705)

Blinding of ParticipantsLow risk“All medications were labeled either A or B by the manufacturer. Both study oils were encapsulated with ethyl vanillin-flavored gelatin; thus, smell and taste of the capsules were identical.” (p 705)

Blinding of CaregiversLow risk“Martek retained the code and divulged group assignment to the Data and Safety Monitoring Committee when requested to or to a patient’s physician in case of a medical emergency.” (p 705)

Blinding outcome assessors for primary outcome (visual field)Low risk“The randomization code was not available to study personnel conducting visual function assessments until after completion of testing.” (p 705)

Blinding outcome assessors for secondary outcome (visual acuity)Low risk“The randomization code was not available to study personnel conducting visual function assessments until after completion of testing.” (p 705)

Blinding outcome assessors for secondary outcome (ERG)Low risk“The randomization code was not available to study personnel conducting visual function assessments until after completion of testing.” (p 705)

Incomplete outcome data addressed for primary outcome (visual field)Low risk“Of the 44 patients enrolled, all completed 3 years; 41 patients completed 4 years . . . test results from the previous year were used in place of an occasional missing value” (p 706, 709)

Incomplete outcome data addressed for secondary outcome (visual acuity)Low risk“Of the 44 patients enrolled, all completed 3 years; 41 patients completed 4 years . . . test results from the previous year were used in place of an occasional missing value” (p 706, 709)

Incomplete outcome data addressed for secondary outcome (ERG)Low risk“Of the 44 patients enrolled, all completed 3 years; 41 patients completed 4 years . . . test results from the previous year were used in place of an occasional missing value” (p 706, 709)

Selective reporting (reporting bias)Unclear riskInsufficient information to permit judgment as the study protocol is not available

Other biasLow riskWe did not detect other bias

 
Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion

Bergsma 1977Not randomized. A cohort of RP patients followed prospectively before and after administration of vitamin A without a control group

Berson 2010Vitamin A was given to all participants. Not able to justify the effectiveness of vitamin A

Dagnelie 2000Not randomized. A cohort of RP patients and patients with related retinal degenerations followed prospectively before and after administration of lutein. No control group

Fex 1996Uncontrolled trial, did not examine the outcome of interest to this review

Sibulesky 1999Secondary analysis assessing the safety of long term vitamin A supplementation in an RCT for RP (Berson 1993)

Tcherkes 1950Not randomized. A cohort of RP patients followed prospectively before and after administration of vitamin A. No control group

Wheaton 2003Secondary analysis assessing the safety of long term DHA supplementation in an RCT for X-linked RP

 
Table 1. Summary of included trials

Study IDBerson 1993Berson 2004Hoffman 2004

Design2x2 factorial designParallelParallel

Genetic profile of
participants
Autosomal dominantAutosomal dominantNot included

Autosomal recessiveAutosomal recessiveNot included

X-linkedX-linkedX-linked

Dominant with mutationDominant with mutationNot included

IsolateIsolateNot included

UndeterminedUndeterminedNot included

Age range18-49 years18-55 years4-38 years

Gender (% female)Men and women (38%)Men and women (49%)Only males (0%)

Number randomized60122144

Intervention(s)Vitamin A + vitamin E
trace = 146

Vitamin A + vitamin E =
151

Vitamin A trace + vitamin
E trace =149

Vitamin A trace + vitamin
E = 155
DHA + vitamin A = 105 (number analyzed)

DHA placebo + vitamin A = 103 (number analyzed)
DHA = 23

DHA placebo = 21

DoseVitamin A = 15000 IU/d

Vitamin A trace = 75 IU/d

Vitamin E = 400 IU/d

Vitamin E trace = 3 IU/d
DHA, 1200 mg/d

Vitamin A, 1500 IU/d
DHA, 400mg/d

Primary outcomeCone ERG amplitudeVisual field (total point score for 30-2 HFA)Cone ERG amplitude

Other outcomesRod ERG, visual acuity,
visual field
Cone ERG, visual acuity, visual field (total point score for 30-2 and 30/60-1
programs combined)
Rod ERG, visual
acuity, visual field, dark
adaptation

Length of follow-up4-6 years4 years4 years

 DHA: docosahexaenoic acid
ERG: electroretinogram
 
Table 2. Summary of analysis of visual outcomes (visual field and visual acuity) in included trials

OutcomeBerson 1993Berson 2004Hoffman 2004

Visual
field
Instrument
used
Goldmann perimeter
(V-4-e white test light)
Humphrey field
analyzer 30-2 program
640 Humphrey field
analyzer, Program 30-2

Effect
measure
Percent decline per
year of remaining
visual field area
Mean annual rate loss
of field sensitivity
Mean change in defect in
Humphrey spot size III field from baseline at 4
years

Method used
for estimation
Longitudinal regression
analysis
Longitudinal regression
analysis
Mean change from
baseline

EstimateVitamin A + vitamin E
trace = 5.6%

Vitamin A + vitamin E =
6.2%

Vitamin A trace +
vitamin E trace = 5.9%

Vitamin A trace +
vitamin E = 6.3%
DHA + vitamin A = 36.95 ± 3.36
dB/year

Control + vitamin A = 37.68 ±
3.36 dB/year
DHA = 2.4 ± 3.66 dB
(0.24 logMAR)
Placebo = 1.4 ± 1.32 dB
(0.14 logMAR)

Data interpretationNo significant vitamin A

or vitamin E main effects

or interaction effects

were observed
No significant differenceNo significant difference

Visual
acuity
Instrument
used
EDTRS chartEDTRS chartEDTRS chart

Effect
measure
Number of EDTRS letters

lost per year
Annual rate of decline
of EDTRS visual acuity
over 4 years
Mean group difference in
log MAR visual acuity
between years 0 and 4
for the average of both
eyes

Method used
for estimation
Longitudinal regression
analysis
Longitudinal regression
analysis
Mean change from log
MAR baseline visual
acuity

EstimateVitamin A + vitamin E
trace = 1.1 letters/year

Vitamin A + vitamin E =
0.7 letters/year

Vitamin A trace +
vitamin E trace = 0.9
letters/year

Vitamin A trace +
vitamin E = 0.9
letters/year
DHA + vitamin A = 0.71 + 0.12
letters/year

Control + vitamin A =
0.68 + 0.12 letters/year
DHA = 0.05 ± 0.23 log
units (logMAR)

Placebo = 0.06 ± 0.2 log
units (logMAR)

Data interpretationNo significant differenceNo significant differenceNo significant difference

 DHA: docosahexaenoic acid
EDTRS: Early Treatment Diabetic Retinopathy Study
RCT: randomized controlled trial
 
Table 3. Summary of analysis of electroretinogram (ERG) in included studies

Berson 1993Berson 2004Hoffman 2004




Effect
measure


and estimate
Rate of decline of remaining 30 Hz ERG amplitude per yearVitamin A + vitamin E trace = 6.1%

Vitamin A + vitamin E = 6.3%

Vitamin A trace + vitamin E trace = 7.1%

Vitamin A trace + Vitamin E = 7.9%
Annual rate of decline of 30 Hz ERG amplitude, loge % declineDHA + vitamin A = 0.10 ± 0.01

Control + vitamin A = 0.11 ± 0.01
The mean (±1 SD) change in log cone ERG amplitude by 4th yearDHA = -0.199 ± 0.172 log μV

Placebo = -0.266 ± 0.173 log μV

Percentage of participants with less than 50% decline in 30 Hz ERG amplitude relative to baseline at year 6 (high amplitude cohort);Vitamin A + vitamin E trace = 62%

Vitamin A + vitamin E=50%

Vitamin A trace + Vitamin E trace = 48%

Vitamin A trace + Vitamin E = 27%
Mean annual rate of decline of remaining 30 Hz ERG functionDHA + vitamin A = 9.92%

Control + vitamin A = 10.49%
Mean change from baseline after 4 years in
log cone ERG amplitude (text) and for all years of follow-up
"A repeated-measures ANOVA showed a significant main effect of year(<.0001), with the population as a whole showing significant progression. The main effect of group was not significant (P=.16), and the interaction between group and year was not significant (P=.61)."

Mean change from
baseline for each year
of follow-up (for high-amplitude
cohort)
Data in figure only

Method
used for
estimation
Longitudinal regression analysis

Survival analysis

Mean change analysis
Longitudinal regression analysisSubtracting the mean baseline log amplitude from the mean follow-up log amplitude




Data interpretationThe vitamin A group had, on average, a slower rate of decline of retinal function than the 2 groups not receiving this dosageNo significant differenceNo significant difference

 DHA: docosahexaenoic acid
ERG: electroretinogram
SD: standard deviation