Methods
This systematic review and meta-analysis was performed according to the
PRISMA guideline (Liberati et al., 2009). The study protocol was
registered in the PROSPERO international prospective register of
systematic review.
Search strategy
PubMed, Scopus, and ISI web of knowledge databases were searched for
relevant medical literature without language or publication date
restrictions up to September 2019. The keywords used in our search
strategy were (“Omega-3 Fatty Acid, γ-Linolenic Acid,Antioxidant
Supplementation”). Further details about the search strategy in
different databases are provided in Supplementary Table 1.
Reference lists of previous reviews that investigated this
immunomodulatory formula were also checked for any additional studies
not identified by the database searches. All titles and abstracts were
screened by two authors (MM and NP) to identify eligible studies.
Eligibility criteria
Titles, abstracts and full texts of retrieved articles were screened by
two reviewers (MM and NP) according to the following inclusion
criteria:1) the studies were a RCT with either a parallel or crossover
design; 2) Study population were adult patients admitted to a ICU; 3)
the mean and SDs, SEMs, or interquartile range or 95% CIs for baseline
and final values or change values reported for critically ill parameters
(length of hospital stays, duration of ICU stays, ICU- free days,
duration of mechanical ventilation, ventilator free-days, level of
oxygenation, SOFA (sequential organ failure assessment) and MOD
(multiple organ dysfunction) score and overall 28 days mortality); 4)
the only difference between the treatment group and control was the
immunomodulatory formula containing omega-3 fatty acid, γ-linolenic acid
and antioxidants. Studies were excluded as per the following criteria:
1) participants younger than 18 years; 2) pregnant and lactating women;
3) trials designed without concurrent controls; and 4) trials that were
not performed in ICU settings and subjects were not critically ill.
Data extraction
Data were extracted from the eligible papers by three independent
investigators (SF and NP and SR). Data listed from the eligible studies
included: first author’s name, country of the study, year of
publication, study design underlying diseases, number of participants in
the intervention and control group, mean age of the participants and
their gender, intervention duration, type of intervention in control and
intervention group, before and after measurement of the expected
outcomes in the intervention and control group. Outcome included:
duration of ICU stays, level of oxygenation (PaO2/FiO2), duration of
mechanical ventilation, length of hospital stays, ICU- free days,
ventilator- free days, SOFA (sequential organ filure assessment) and MOD
(multiple organ disfunction) score, 28 days mortality (for mortality
relative risk based on raw data was extracted).
Risk of bias assessment
Two investigators (NP and SR) independently assessed the risk of bias in
each included studies on the basis of Cochrane Risk of bias assessment
tool (Higgins and Altman, 2008). Disagreements were resolved in
consultation with a third investigators (SS). The following criteria
were evaluated: random sequence generation, allocation concealment,
incomplete data outcome, selective outcome reporting, the blinding of
participants and investigators. Each item was classified as yes (low
risk of bias), no (high risk of bias) or unclear. The overall quality of
included studies was considered as poor if they had less than four
points for low risk of bias. They were classified as fair if they had
four points and good if they had more than four points for low risk of
bias.
Assessment of the quality of
meta-evidence
The quality of meta-evidence for this review was evaluated by using the
NutriGrade (Grading of Recommendations Assessment, Development, and
Evaluation) scoring system (Schwingshackl et al., 2016). This system for
systematic review of RCTs has maximum 10 points and includes: 1) risk of
bias, study quality, and study limitations, 2) precision, 3)
heterogeneity, 4) directness, 5) publication bias, 6) funding bias, 7)
study design [22]. The overall quality of meta-evidence for each
outcome was classified: high (≥8 points), moderate (6–7.99 points), low
(4–5.99), and very low (0–3.99).
Data synthesis and
analysis
To determine the clinical effect of immunomodulatory supplements
containing γ-Linolenic Acid on potential outcome, changes and standard
deviations from baseline within intervention and comparator groups were
calculated for each trial. The results were expressed as mean
differences
(MDs) with 95% CIs. Regarding to malnutrition score (SOFA and MOD),
Hedges’g and with 95% CIs were calculated because of different
questionnaire applied. For 28-days mortality, relative risk (RR) with
95% CIs were considered as an effect size. Meta-analyses were performed
using the inverse-variance weighting method with the random-effects
model (DerSimonian and Laird, 1986). Heterogeneity of the included
trials was assessed using the chi-squared test and the
I2 test (Higgins and Thompson, 2002, Higgins et al.,
2011). The substantial heterogeneity was defined as an
I2 value of more than 50%. To detect source of
heterogeneity a series of predefined subgroup analyses were performed
according to 1) underlying diseases (sepsis, ALI and ARDS), 2) quality
of the study (good, poor and fair), 3) type of intervention in control
group (standard formula and High fat, low carbohydrate formula), 4)
duration of intervention (less than 14 days and equal or more than 14
days), 5) mean age of patients (less than 60 years and more than 60
years). The probability of publication bias was checked by funnel plot
if the number of included studies for each outcome were at least 10
studies. The sensitivity analysis was performed to evaluate the
robustness of finding using a random-effects model (Borenstein et al.,
2010). Analyses were conducted using STATA software, version 13 (Stata
Corp, College Station, TX). P values less than 0.05 were considered as
statistically significant.