Effects of L-carnitine supplementation for women with polycystic ovary syndrome: a systematic review and meta-analysis

Identification and eligibility of study

All randomized control trials (RCTs) comparing LC alone or in combination with other standard medications or other dietary supplements for the treatment and supplementation of PCOS women were considered in the review. The comparators were selected according to the availability of comparative studies versus LC. The participants included women who had been diagnosed with PCOS based on the revised ESHRE and the ASRM diagnosis of PCOS, according to the Rotterdam criteria of 2003. We excluded cross-over trials and studies other than RCTs. Werestricted the publications to the English language only.

We used the search strategy in Appendix S1 and searched through Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, Cumulative Index to Nursing and Allied Health Literature (CINAHL), Psychological Information Database (PsycINFO), and the World Health Organization International Clinical Trials Registry Platform for all available studies comparing LC alone or in combination with other standard treatments to treat PCOS. For additional datasets, we modified the search strategy. Using the Boolean operators AND as well as OR, we combined the terms “polycystic ovarian syndrome” and “L carnitine” (refer to Appendix S1). To locate unpublished trials or trials that could not be found using electronic searches, we looked through the reference lists of recognized RCTs and read the relevant articles. We also reached out to experts in the field and used the World Health Organization International Clinical Trials Registry Platform (http://www.who.int/ictrp/en/) and http://www.clinicaltrials.gov to find active trials.

Three authors (MFMS, SB, AAK) scanned the repository of articles for trial selection from the titles and abstracts derived from the searches. Therein, we obtained full-text articles when they appeared to meet the eligibility criteria or when there was insufficient information to assess the eligibility. We documented the reasons behind exclusion after the authors independently reviewed the eligibility of the studies. Any differences were settled by discussion among the authors. If more information is required, then we will contact the authors. We utilized the procedure recommended by the Cochrane Handbook for Systematic Reviews of Interventions for searching and selecting studies (Higgins et al., 2021).

We retrieved 56 records from the search of the electronic databases, 22 records from Cochrane, 30 from MEDLINE, and four records from other databases. We screened 33 records after removing duplicates. Furthermore, we reviewed the complete text of 28 records—nine studies met the inclusion criteria, whereas 19 studies did not fulfill the inclusion criteria and were, therefore, excluded (refer to Fig. 1). The number of records retrieved, screened, included, and excluded was presented in the PRISMA study flow diagram (Fig. 1).

Data collection and analysis

Three authors independently extracted data. We extracted data on the study setting, participant characteristics (age), methodology (inclusion and exclusion criteria, number of participants randomized and analyzed, and duration of follow-up), description of interventions used (dose, frequency, preparation, and duration used), and the measured outcomes. We also extracted data pertaining to the number of intrauterine gestational sacs and fetal heart rate visible by transvaginal ultrasound within 12 weeks of intervention (clinical pregnancy rate), the number of visible leading follicles of more than or equal to 18 mm by transvaginal ultrasound within 12 weeks of intervention (ovulation rate), BMI in kg/m2, serum LDL, serum HDL, TG, total cholesterol in mmol/l or mg/dl, and fasting blood glucose (FPG) in mg/dl, serum FSH and LH in IU/L, mental health status assessment using any questionnaires, and adverse side effects such as gastrointestinal disturbances (abdominal pain, nausea, and vomiting. Disagreements between the review authors (MFMS, SB, AAK) were resolved by discussion with the fourth author (NMN).

Assessment risk of bias

We used the Cochrane Collaboration’s risk-of-bias tools to assess the risk of bias in the included studies (Higgins et al., 2021). Three authors (MFMS, SB, AAK) assessed the selection bias (randomization and allocation concealment), performance bias (blinding of participant and health personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other biases (recall bias, transfer bias and etc.) independently. We classified the risk of bias as very low, low, moderate, or high. We also resolved disagreements by conducting discussions with the fourth author (NMN). In addition, we assessed the quality of evidence for primary and secondary outcomes, according to the GRADE methodology for risk of bias, inconsistency, indirectness, imprecision, and publication bias and classified it as very low, low, moderate, or high (Guyatt et al., 2008). Furthermore, we assessed the presence of the risk of bias, inconsistency or unexplained heterogeneity, indirectness of evidence, imprecision, and publication bias. We classified them as very low, low, moderate, and high.

Statistical analysis

We analyzed data using Review Manager 5.4 software (Manager, 2020) for the statistical analyses. Moreover, we used a random-effects model to pool data. We measured the treatment effect using risk ratios (RR) for dichotomous outcomes and mean differences (MD) with 95 percent confidence intervals (CI) for continuous outcomes. We assessed the presence of heterogeneity in two steps. First, the assessment was performed at face value by comparing populations, settings, interventions, and outcomes (Higgins et al., 2021). Second, the statistical heterogeneity was assessed by using the I² statistic (Higgins et al., 2021). We used the interpretation of heterogeneity as follows: 0% to 40% might not be important; 30% to 60% may represent moderate heterogeneity; 50% to 90% may represent substantial heterogeneity; and 75 percent to 100 percent would indicate considerable heterogeneity (Higgins et al., 2021). We checked the included trials for the unit of analysis errors. The unit of analysis errors can occur when trials randomize participants to intervention or control groups in clusters but analyze the results using the total number of individual participants. Based on the mean cluster size and intra-cluster correlation coefficient, we adjusted the results from trials with the unit of analysis errors (Higgins et al., 2021). Thereafter, we contacted the trial’s original authors to request data that had been missing or incorrectly reported. If missing data was not accessible, we conducted analyses using the available data. We performed a sensitivity analysis to investigate the impact of the high risk of bias on sequence generation and allocation concealment of included studies. If there were sufficient studies, then we used funnel plots to assess the possibility of reporting biases or small study biases, or both.

GRADEPro software was used to analyze the quality of evidence or certainty in the body of evidence for each outcome, and we classified the quality of evidence as high, moderate, low, and very low.

Trial selection

We retrieved 56 records from the electronic searches that were available from inception until June 2021. We screened a total of 33 records after duplicates were removed, and we excluded five studies that did not meet the eligibility criteria. Out of these 28 studies, another 19 studies were excluded. Five out of 19 studies were not RCT studies (Celik et al., 2017; Eyupoglu et al., 2019; Fenkci et al., 2008; Maleki et al., 2019; Salehpour et al., 2019), and 12 studies were excluded because they did not report outcomes of interest for this review (Chen et al., 2020; Chen et al., 2016; Cree-Green et al., 2019; Dong et al., 2015; Hamed, 2016; Jia et al., 2019; Karakas et al., 2016; Selen Alpergin et al., 2017; Sheida et al., 2021; Sun et al., 2019; Vonica et al., 2019; Zhao et al., 2015). Two other studies reported the effects of other supplementations other than LC and did not fulfil the eligibility criteria (Nct, 2019; Vigerust et al., 2012). We have summarized the results of the search strategy in Fig. 1.

Characteristics of included trials

We included nine trials with a total of 995 participants (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Ismail et al., 2014; Jamilian et al., 2017; Jamilian et al., 2019a; Jamilian et al., 2019b; Kortam, Abdelrahman & Fateen, 2020; Samimi et al., 2016; Talari et al., 2019). All nine trials recruited women who had been diagnosed with PCOS based on the Rotterdam criteria. Six trials involved the participants aged between 18 and 40 years (El Sharkwy & Abd El Aziz, 2019; Jamilian et al., 2017; Jamilian et al., 2019a; Jamilian et al., 2019b; Samimi et al., 2016; Talari et al., 2019). On the other hand, two trials include BMI >25 kg/m2 as one of the inclusion criteria (Jamilian et al., 2019b; Samimi et al., 2016), and three trials used clomiphene citrate resistant PCOS as the inclusion criteria (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Ismail et al., 2014). All nine trials reported hyperprolactinemia as the exclusion criteria. Eight trials excluded participants with endocrine disorder, and the duration of the study was 12 weeks, with the exception of one trial (Kortam, Abdelrahman & Fateen, 2020) that did not mention the study duration. Four out of nine included trials excluded women who were pregnant in the trial (Jamilian et al., 2017; Jamilian et al., 2019a; Jamilian et al., 2019b; Talari et al., 2019). Three studies excluded diabetic patients as participants in the trial (Jamilian et al., 2019a; Jamilian et al., 2019b; Samimi et al., 2016).

Outcomes

The nine included trials had diverse groups, which addressed various comparisons and outcomes, resulting in several comparisons that contributed to each of predefined outcomes. All studies had methodological limitations, and there were too few studies to allow pooling of all primary and secondary outcomes.

Four included trials reported on the clinical pregnancy rate and the ovulation rate (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Ismail et al., 2014; Kortam, Abdelrahman & Fateen, 2020), whereas seven out of nine included trials reported BMI (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Jamilian et al., 2017; Jamilian et al., 2019b; Kortam, Abdelrahman & Fateen, 2020; Samimi et al., 2016; Talari et al., 2019). The lipid profile, including serum LDL, HDL, total cholesterol, and TG levels, were reported in four trials (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Jamilian et al., 2019b; Samimi et al., 2016), and FPG was reported in four trials (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Jamilian et al., 2019b; Samimi et al., 2016).

Five trials reported secondary outcomes, which are hormonal levels, including the serum FSH levels, and LH levels, and mental health status. The serum FSH and LH levels were reported in three trials (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Kortam, Abdelrahman & Fateen, 2020), and the mental health status was reported in two trials (Jamilian et al., 2017; Jamilian et al., 2019a).

Assessment risk of bias

The assessment of risk of bias has been presented in Figs. 2 and 3. The details of these trials are summarized in Table 1. All nine trials described the method of randomization used. Eight trials randomized the participants using computer-generated randomization (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Ismail et al., 2014; Jamilian et al., 2017; Jamilian et al., 2019a; Jamilian et al., 2019b; Samimi et al., 2016; Talari et al., 2019), with the exception of one trial (Jamilian et al., 2019b) in which the randomization sequence was manually executed at the clinic. Therefore, we judged a high risk of random sequence generation bias for this trial (Jamilian et al., 2019b), whereas a low risk of bias was assigned to the other eight trials. Allocation concealment was reported in all trials. All trials conducting the study using placebo capsules, which were designed to be identical to LC capsules. Three trials (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Ismail et al., 2014) distributed the capsules using opaque and sealed envelopes. Therefore, for allocation concealment, all trials had a low risk of bias. Eight trials mentioned blinding of participants and personnel with the exception of one trial (Kortam, Abdelrahman & Fateen, 2020), which resulted (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Ismail et al., 2014; Jamilian et al., 2017; Jamilian et al., 2019a; Jamilian et al., 2019b; Samimi et al., 2016; Talari et al., 2019), in an unclear risk of bias. Seven trials had a low risk of bias (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Ismail et al., 2014; Jamilian et al., 2017; Jamilian et al., 2019a; Jamilian et al., 2019b; Samimi et al., 2016), highlighting that the patients and physicians were blinded to the treatment allocation. Only one trial (Talari et al., 2019) mentioned that the researchers and participants were not blinded to the allocation concealment, thereby resulting in a high risk of bias.

Eight trials reported the number of participants who completed the study, including the number of patients who dropped out from the study with justified reasons (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Ismail et al., 2014; Jamilian et al., 2017; Jamilian et al., 2019a; Jamilian et al., 2019b; Samimi et al., 2016; Talari et al., 2019). The missing participants for these trials were less than 15 percent (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Ismail et al., 2014; Jamilian et al., 2017; Jamilian et al., 2019a; Jamilian et al., 2019b; Samimi et al., 2016; Talari et al., 2019), and one trial (Talari et al., 2019) did not have any missing participants from both the control and intervention groups. Only one trial (Kortam, Abdelrahman & Fateen, 2020) did not mention the number of participants who completed or withdrew from the study. Neither did it summarize the patients’ flow diagram, resulting in an unclear risk of bias.

All nine trials reported the outcomes as specified in their methods section (El Sharkwy & Sharaf El-Din, 2019; El Sharkwy & Abd El Aziz, 2019; Ismail et al., 2014; Jamilian et al., 2017; Jamilian et al., 2019a; Jamilian et al., 2019b; Kortam, Abdelrahman & Fateen, 2020; Samimi et al., 2016; Talari et al., 2019). Four trials registered their protocols, and three trials (Jamilian et al., 2017; Jamilian et al., 2019a; Samimi et al., 2016) were registered in the Iranian Registry of Clinical Trials. Only one trial (El Sharkwy & Abd El Aziz, 2019) was registered in the National Clinical Trials.

LC supplementation for women with PCOS

We performed meta-analysis for each of the five comparison groups in this review. For the first comparison, the combination of clomiphene citrate and LC was compared with the combination of clomiphene citrate and placebo (Ismail et al., 2014; Kortam, Abdelrahman & Fateen, 2020). In total, 250 milligrams (mg) of oral clomiphene citrate was administered along with 3 grams (g) of LC in one study in comparison with the same 250 mg clomiphene citrate combined with placebo (Ismail et al., 2014). In another study, the researchers used 100 mg clomiphene citrate daily in combination with 3 g of LC when compared with the use of 100 mg clomiphene citrate and the placebo (Kortam, Abdelrahman & Fateen, 2020). The second comparison comprised the study that used 150 mg clomiphene citrate, 850 mg metformin and 1 g LC versus 150 mg clomiphene citrate, 850 mg metformin, and placebo (El Sharkwy & Sharaf El-Din, 2019). The third comparison included the studies that used a combination of 150 g clomiphene citrate and 600 mg oral N-Acetylcysteine in comparison with 150 mg clomiphene citrate and 3 g LC (El Sharkwy & Abd El Aziz, 2019). The fourth comparison included the studies that used 250 mg of LC in comparison with placebo (Jamilian et al., 2017; Samimi et al., 2016; Talari et al., 2019). Finally, the fifth comparison comprised the studies that used 200 mg chromium picolinate and 1g LC daily in comparison with the placebo (Jamilian et al., 2019a; Jamilian et al., 2019b).

Comparison 1: Clomiphene citrate and LC versus clomiphene citrate and placebo

We performed meta-analysis in this comparison. No difference was observed in terms of the clinical pregnancy rate between the two groups (Risk ratio (RR) 7.12, 95% CI [0.14–350.06]; I² = 90%, P = 0.32; two trials, n = 264; low quality evidence) (Ismail et al., 2014; Kortam, Abdelrahman & Fateen, 2020). However, a difference was observed in terms of the primary outcome, ovulation rate between the two groups, which favored combination with placebo (RR 2.37, 95% CI [0.99–5.66]; I² = 88%, P = 0.05; two trials, n = 264; low quality evidence) (Ismail et al., 2014; Kortam, Abdelrahman & Fateen, 2020). Figure 4 showed the Forest plot, comparing the use of clomiphene citrate and LC in comparison with the use of clomiphene citrate and placebo for primary outcomes, clinical pregnancy rate, and ovulation rate. There is a difference in terms of the primary outcome, BMI within one group, which favored combination with placebo (MD 1.10, 95% CI [0.32–1.88]; P = 0.006; one trial, n = 94; moderate quality evidence) (Kortam, Abdelrahman & Fateen, 2020). No difference is observed for the secondary outcome, FSH within one group (MD −0.10, 95% CI, [−0.50–0.70]; P = 0.75; one trial, n = 94; moderate quality evidence) (Kortam, Abdelrahman & Fateen, 2020). There is no difference for the secondary outcome, LH within one group (MD (95% CI) −0.20 (−0.91–0.51); P = 0.58; one trial, n = 94; moderate quality evidence) (Kortam, Abdelrahman & Fateen, 2020). Therefore, in this comparison group, there was no significant difference in the pregnancy rate, FSH, and LH levels. However, there was a significant difference, favoring the placebo in terms of the ovulation rate and BMI. Table 2 showed the summary of findings and GRADE quality assessment for primary and secondary outcomes of Comparison 1.

Comparison 2: Clomiphene citrate, metformin, and LC versus clomiphene citrate, metformin, and placebo

We performed meta-analysis in this comparison. There is a significant difference in the primary outcome, clinical pregnancy rate in one group, which favored combination with placebo (RR 4.27, 95% CI [2.15–8.47]; P = 0.0001; one trial, n = 274; moderate quality evidence) (El Sharkwy & Sharaf El-Din, 2019). There is a significant difference in the ovulation rate in one group, which favored combination with placebo (RR 3.15 95% CI [1.86–5.35]; P = 0.0001; one trial, n = 274; moderate quality evidence) (El Sharkwy & Sharaf El-Din, 2019). There is a significant difference for BMI in one group, which favored combination with placebo (MD 1.10, 95% CI [0.32–1.88]; P = 0.006; one trial, n = 274; moderate quality evidence) (El Sharkwy & Sharaf El-Din, 2019). There is a significant difference for the primary outcome, FPG in one group, which favored combination with LC (MD −5.10, 95% CI [−6.25 to −3.95]; P = 0.00001; one trial, n = 274; moderate quality evidence) (El Sharkwy & Sharaf El-Din, 2019) (Table 3). In addition, there is a significant difference for the primary outcomes, LDL level in one group, which favored combination with LC (MD −25.00, 95% CI [−27.93 to −22.07]; P = 0.00001; one trial, n = 274; moderate quality evidence) (El Sharkwy & Sharaf El-Din, 2019), TC level in one group, which favored combination with LC (MD −21.00, 95% CI [−24.14 to −17.86]; P = 0.00001; one trial, n = 274; moderate quality evidence) (El Sharkwy & Sharaf El-Din, 2019) and TG level in one group, which favored combination with LC (MD −9.00, 95% CI [−11.46 to −6.54]; P = 0.00001; one trial, n = 274; moderate quality evidence) (El Sharkwy & Sharaf El-Din, 2019). There is a significant difference for the primary outcome, HDL level in one group, which favored combination with placebo (MD 15.50, 95% CI [12.42–18.58]; P = 0.00001; one trial, n = 274; moderate quality evidence) (El Sharkwy & Sharaf El-Din, 2019) (Table 3). There is a significant difference for the secondary outcomes, FSH level in one group, which favored combination with LC (MD −0.63, 95% CI [−0.92 to −0.34]; P = 0.00001; one trial, n = 274; moderate quality evidence) (El Sharkwy & Sharaf El-Din, 2019). There is a significant difference for LH level in one group, which favored combination with LC (MD−2.36, 95% CI [−3.04 to −1.68]; P = 0.00001; one trial, n = 274; moderate quality evidence) (El Sharkwy & Sharaf El-Din, 2019). In this comparison, there was a significant difference that favored combination with LC in terms of the FPG, LDL, TC, TG, HDL, FSH, and LH levels. There was a significant difference favoring the combination with placebo in pregnancy rate, ovulation rate, HDL level, and BMI. Table 3 showed the summary of finding and GRADE quality assessment for primary and secondary outcomes of Comparison 1.

Comparison 3: Clomiphene citrate plus LC versus clomiphene citrate plus n-acetylcysteine

We performed meta-analysis in this comparison. There is no difference for the primary outcome, clinical pregnancy rate in one group (RR (95% CI) 1.16 (0.72, 1.89); P = 0.54; one trials, n = 162; moderate quality evidence) (El Sharkwy & Abd El Aziz, 2019). There is no difference for the primary outcome, ovulation rate in one group (RR (95% CI) 1.11 (0.79, 1.56); P = 0.54; one trials, n = 162; moderate quality evidence) (El Sharkwy & Abd El Aziz, 2019). There is no difference for the primary outcome, BMI in one group (MD 0.10, 95% CI [−0.78–0.98]; P = 0.82; one trial, n = 162; moderate quality evidence) (El Sharkwy & Abd El Aziz, 2019). There is a significant difference for the primary outcome, FPG in one group, which favored combination with NAC (MD 2.30, 95% CI [1.02–3.58]; P = 0.0004; one trial, n = 162; moderate quality evidence) (El Sharkwy & Abd El Aziz, 2019). There is a significant difference for the primary outcome, LDL level in one group, which favored combination with LC (MD −12.00, 95% CI [−15.80 to −8.20]; P = 0.00001; one trial, n = 162; moderate quality evidence) (El Sharkwy & Abd El Aziz, 2019). There is a significant difference for the primary outcome, TC level in one group, which favored combination with LC (MD −24.00, 95% CI [−27.61 to −20.39]; P = 0.00001; one trial, n = 162; moderate quality evidence) (El Sharkwy & Abd El Aziz, 2019). There is a significant difference for the primary outcome, HDL level in one group, which favored combination with NAC (MD 9.60, 95% CI [5.30–13.90]; P = 0.0001; one trial, n = 162; moderate quality evidence) (El Sharkwy & Abd El Aziz, 2019). There is a significant difference for the primary outcome, TG level in one group, which favored combination with LC (MD −19.00, 95% CI [−22.79 to −15.21]; P = 0.00001; one trial, n = 162; moderate quality evidence) (El Sharkwy & Abd El Aziz, 2019). The summary of all findings and GRADE quality assessment for primary outcomes of Comparison 3 is shown in Table 4.

There is a significant difference for the secondary outcome, FSH level in one group, which favored combination with LC (MD−0.50, 95% CI [−0.84 to −0.16]; P = 0.004; one trial, n = 162; moderate quality evidence) (El Sharkwy & Abd El Aziz, 2019). There is no difference for the secondary outcome, LH level in one group (MD −0.40, 95% CI [−1.51–0.71]; P = 0.48; one trial, n = 162; moderate quality evidence) (El Sharkwy & Abd El Aziz, 2019). In this comparison, there was no significant difference in the pregnancy rate, ovulation rate, BMI, and LH level. There was a significant difference that favored the combination of LC in LDL, TC, TG, and FSH levels, and there was a significant difference that favored the combination with NAC in terms of the FPG and HDL levels. The summary of all findings and GRADE quality assessment for secondary outcomes of Comparison 3 is shown in Table 4.

Comparison 4: comparing LC with the placebo

We performed meta-analysis in this comparison. There was no difference for FPG in one group (MD −1.26, 95% CI [−7.50–4.98]); P = 0.69; one trial, n = 60; moderate quality evidence) (Samimi et al., 2016), LDL level in one group (MD 0.33, 95% CI [−0.05–0.71]; P = 0.09; one trial, n = 60; moderate quality evidence) (Samimi et al., 2016), total cholesterol level in one group (MD 6.84, 95% CI [−0.45–14.13]; P = 0.07; one trial, n = 60; moderate quality evidence) (Samimi et al., 2016), HDL level in one group (MD 0.00, 95% CI [−3.60–3.60]; P = 1.00; one trial, n = 60; moderate quality evidence) (Samimi et al., 2016), and TG level in one group (MD 0.15, 95% CI [−0.14–0.44]; P = 1.00; one trial, n = 60; moderate quality evidence) (Samimi et al., 2016). There was significant difference for BMI level in three groups, which favored LC group (MD −1.33, 95% CI [−1.52 to −1.44]; I²⁼ 0%, P = 0.00001; three trials, n = 180; moderate quality evidence) (Jamilian et al., 2017; Samimi et al., 2016; Talari et al., 2019). Figure 5 showed the Forest plot of Comparison 4, comparing LC with placebo for the primary outcome, BMI. The summary of findings of primary outcomes and GRADE quality assessment for Comparison 4 is shown in Table 5.

There is a significant difference for the secondary outcome, mental health status, by using assessment score, BDI score in one group, which favored placebo (MD 2.50, 95% CI [2.35–2.65]; P = 0.00001; one trial, n = 60 ; moderate quality evidence) (Jamilian et al., 2017), general health questionnaire (GHQ) score in one group, which favored LC (MD −5.80, 95% CI [−6.10 to −5.50]; P = 0.00001; one trial, n = 60; moderate quality evidence) (Jamilian et al., 2017), and depression anxiety stress score (DASS) in one group, which favored LC (MD −6.80, 95% CI [−7.20 to −6.40]; P = 0.00001; one trials, n = 60; moderate quality evidence) (Jamilian et al., 2017). Therefore, in this comparison, there was no significant difference in terms of the FPG, LDL, TC, HDL, and TG levels, whereas there were significant differences that favored LC with respect to BMI, GHQ, and DASS scores, and significant differences that favored placebo in the BDI score. The summary of findings of secondary outcomes and GRADE quality assessment for Comparison 4 is shown in Table 5.

Comparisons 5: LC plus chromium and placebo

We performed meta-analysis in this comparison. There is no difference for the primary outcome, FPG in one group (MD −3.40, 95% CI [−7.60–0.80]; P = 0.11; one trial, n = 54; moderate quality evidence) (Jamilian et al., 2019b). There is no difference for the primary outcome, LDL level in one group (MD −0.60, 95% CI [−19.95–18.75]; P = 0.95; one trial, n = 54; moderate quality evidence) (Jamilian et al., 2019b). There is no difference for the primary outcome, TC in one group (MD −9.70, 95% CI [−28.53–9.13]; P = 0.31; one trial, n = 54; moderate quality evidence) (Jamilian et al., 2019b). There is no difference for the primary outcome, HDL level in one group (MD −3.40, 95% CI [−8.20–1.40]; P = 0.17; one trial, n = 54 moderate quality evidence) (Jamilian et al., 2019b). There is significance difference for the primary outcome, TG level in one group, which favored combination with LC (MD −28.10, 95% CI [−47.25 to −8.95]; P = 0.004; one trial, n = 54; moderate quality evidence) (Jamilian et al., 2019b). The summary of primary outcomes’ findings and GRADE quality assessment is shown in Table 6.

There is no difference for the secondary outcome, mental health status, by using BDI scoring in one group (MD −1.50, 95% CI [−4.17–1.17]; P = 0.27; one trial, n = 53; moderate quality evidence) (Jamilian et al., 2019a), GHQ scoring in one group (MD −1.80, 95% CI [−7.10–3.50]; P = 0.51; one trial, n = 53; moderate quality evidence) (Jamilian et al., 2019a), and DASS scoring in one group (MD −3.50, 95% CI [−11.42–4.42]; P = 0.39; one trial, n = 53; moderate quality evidence) (Jamilian et al., 2019a). Therefore, in this comparison, there was no difference in FPG, LDL, TC, HDL, BDI score, GHQ score, and DASS score. On the other hand, there was a significant difference that favored combination with LC in terms of the TG level. The summary of secondary outcomes’ findings and GRADE quality assessment is shown in Table 6.

Registration and Protocol

Our systematic review and meta-analysis protocol has been registered and published in PROSPERO (Registration number: CRD42021232433).