Obesity is an epidemic which is worsening in most parts of the world and a major health concern in Malaysia that recorded an increment for the last 20 years [1] [2] . Obesity is defined as a condition of excess adiposity and physiological state occurs in chronic imbalance of whole-body energy metabolism that caloric intake exceeding the energy expenditure [3] . Global Health Observatory Data by WHO also recorded a doubling in the prevalence of obesity from 1980-2014 where in 2014, more than 1.9 billion adults aged 18 years and older were overweight, out of which over 600 million adults were obese [4] . Additionally, according to Snodgrass et al. (2016) adults are progressing from being overweight to being obese or morbidly obese at alarming rates [5] .

Evidences have shown the effect of increase in body weight or BMI are associated with the risk of chronic diseases such cardio vascular disease (CVD), type 2 diabetes mellitus (T2DM), hypertension, stroke, and some type of cancers [6] [7] [8] . Report from Malaysian National Health and Morbidity Survey (NHMS) shows an increasing trend in the percentage of overweight and obese individuals in the country from 1996 to 2002-2003 where the prevalence of obesity was more than doubled and that the number of adults overweight increased more than 60% [9] . Changes in the life style and dietary fat consumption may result in endocrine disorders [10] . Studies have shown a strong positive association between saturated fat intake and the development of obesity and CVD [7] [8] [11] .

Saturated fats such as myristic, palmitic and lauric have been reported to cause disadvantages to health compared with the shorter chain saturated fats and unsaturated fatty acid such as oleic acid [12] [13] . Two types of oils that are produced by oil palms are palm kernel oil and palm oil, which nowadays are important in world trade. Palm oil which is obtained from the mesocarp of the palm fruit is composed of 50% saturated fatty acids, 40% monounsaturated fatty acids and 10% polyunsaturated fatty acids. The saturated fatty acid to unsaturated fatty acid ratio of palm oil is close to unity and it contains a high amount of the antioxidants, β-carotene, and vitamin E [13] .

Palm oil is claimed to have high saturated fat and its consumption supposedly raises the levels of blood cholesterol, thereby increasing the risk of coronary heart disease as well as obesity [14] . Based on individual studies, it is difficult to clearly determine if palm oil consumption is directly associated to overweight and/or obesity, as obesity is likely to be multifactorial, with factors such as other dietary habits, level of activity as well as lifestyle playing important parts.

Food manufacturers prefer palm oil because of its nutritional benefits and versatility. The oil is highly structured to contain more than 95% neutral triacylglycerols (TAGs, or triglycerides) and less than 0.5% free fatty acids (FFAs). The quality and nutritional benefits have been assured for the variety of foods that can be manufactured from the oil directly or from blends with other oils while remaining trans-free [15] .

Their resistance to two major chemical reactions i.e. hydrolysis and oxidation will determine their quality. The hydrolysis reaction occurs slowly in the presence of water but is catalysed by lipase enzymes as well as acid whereas oxidation is mediated by initiating free radicals to produce more radicals from fat molecules, which then readily combine with oxygen, a diradical (O-O), of air to give hydroperoxy radicals and then hydroperoxides. The role of fats in health and nutrition should be evaluated in the context of the digestion, absorption, and metabolism of lipids.

Triacylglycerols with long-chain fatty acids are mainly hydrolyzed by the pancreatic lipase to 2-monoacylglycerols or 2-monoglycerides and free fatty acids from positions −1 and −3 of the triglycerides as shown in Figure 1. The ease or difficulty of absorption of these components will affect their subsequent metabolism and finally their role, if any, in cardiovascular diseases.

Fatty acids in the 2-monoacylglycerols are mainly unsaturated (87%), although overall 50% of the fatty acids in the three positions of triacylglycerols are saturated, a consequence of having the 1- and 3-positional fatty acids more highly saturated. There are two very important dietary consequences. First, the absorbed fatty acids are mainly mono- and di-unsaturated (87% for palm oil and up to 96% for palm olein), despite the relatively high saturation in the total glycerides.. Assuming that all the unsaturated fatty acids at the 1- and 3-positions in palm oil are preferentially absorbed while the saturated fatty acids from these same positions are excreted as salts, then only 8% of saturated fatty acids from the 2-position will be absorbed, an ideal situation to enhance the nutritional attribute of palm oil.

All oils and fats have the triacylglycerol or triglyceride structures shown below. Their differences were on the types of fatty acids (RCOOH) attached to the glycerol backbone in positions 1, 2, and 3 [16] .

There are many hypotheses linking palm oil and increase in weight or BMI. This review aims at synthesising the available evidence reporting the effect of palm oil consumption with weight changes or BMI including obesity specifically or any other anthropometric measurements.

A search strategy was developed to identify studies for this review. The search strategy contained population terms, exposure terms and outcome terms. The search terms included thesaurus terms or Medical Subject Headings (MeSH) as well as title and abstract text searches.

An electronic search for eligible papers were conducted in April 2018 from their earliest record up to 3^rd April 2018; Medline, EMBASE, PsychINFO, Web of Science [ISI Web of Knowledge], Cumulative Index to Nursing and Allied Health Library (CINAHL) [Ebscohost], SCOPUS, LILAC and The Cochrane Library. The search strategy was developed in Medline (Appendix 1) prior to adaptation for the other databases. A grey literature search for unpublished and continuing research was undertaken in March 2018. Search in the metaRegister of Controlled Trials, Clinical Trials.gov (www.clinicaltrials.gov) were also conducted up to 3^rd April 2018. The following search term was used “palm oil or related word variation” and “obesity or related word variation” (pre-obese, body mass index, body weight, waist circumference) or anthropometry wherever available. The references of included studies and systematic reviews were inspected for any additional studies.

If we could find studies that assessed the effect of palm oil consumption to weight changes or BMI, particularly in large national or regional-level cohort studies, we would include these studies in the review, as well as any study that evaluated the effects of palm oil consumption to weight changes in human, such as randomized controlled trials, cluster-randomized controlled trials and quasirandomized controlled trials and all types of observational studies without any languages restriction. All studies that did not use palm oil as an exposure or any animal studies were excluded in this review.

We planned to include studies that use incidence or prevalence of obesity as outcome according to the definition of World Health Organization (WHO) as primary outcome measure. Overweight and obesity is also defined as abnormal or excessive fat accumulation that may impair health (WHO, 2017) [20] . BMI is a simple index of weight-for-height that is commonly used to classify overweight and obesity in adults. Additionally, as obesity can be measured indirectly as increase in body weight, changes in body mass index and changes in body fat measurement were also included. No limitations concerning publication date, disease course of participants, or language were made. We planned to include all adults and children of both sexes in this review. We screened all exposures by looking at the effect of dietary palm oil versus any other oils on obesity.

We used the Cochrane Risk of Bias tool to assess study quality. The risk of bias assessment was performed by two review authors. Any disagreements were resolved by another author. The risk of bias was made base on these domains:

・ Random sequence generation.

・ Allocation concealment.

・ Blinding of participants and personnel.

・ Blinding of outcome assessment.

・ Incomplete outcome data.

・ Selective outcome reporting.

・ Other bias.

Each potential source of bias was graded as high, low or unclear. We used the specific evidence grading system developed by the GRADE group to assess the study quality [21] .

We performed meta-analyses using Review Manager 5.3 (RevMan 2013) [22] . We followed the recommendations as stated in the Cochrane Handbook for Systematic Review of Interventions [23] [24] . If we had found any numerical outcome data that were missing, such as standard deviations or correlation coefficients, and they could not be obtained from the authors, we would have calculated them from other available statistics such as P values according to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions [23] [24] [25] . We created a summary of findings (SOF) table according to the Cochrane Handbook of Systematic Reviews of Interventions [26] . We chose three major outcomes to be included in our SOF table, which was generated using the GRADEpro software [27] . We expressed our results as MD or SMD.

A total of 177 studies were excluded from the analysis. We excluded 70 studies that did not measure directly or indirectly the outcome of interest, 33 studies that did not use palm oil or its derivatives as the intervention, 27 animal or in-vitro studies, 46 conference and commentary papers. We also excluded 1 studies that only measured palm oil without any comparator.

Out of the five included studies, four were RCTs, and one was quasi-randomised, cross-over study. Studies were conducted between 2005 to 2015 in Sweden, Denmark, Colombia and two studies were done in USA. All studies involved healthy adults of both sexes except study performed by Tholstrup et al. (2011) which included only adult men as their participants [28] . We defined the criterion as having been met if the study used palm oil in the intervention with a different list of comparisons. For example, study done by Kien 2005, the intervention is on high palmitic acid compared to high oleic acid from sunflower oil [29] . Whereas another four studies i.e. Tholstrup et al. 2011 between palm olein compared to olive oil and lard [28] ; Iggman 2014 between palm oil rich in saturated fatty acid compared to sunflower oil rich in polyunsaturated fatty acid [30] ; Kien 2014 between high palmitic acid (source from palm oil 89%, peanut oil 6.75%, olive oil 4.25%) compared low palmitic acid with high oleic acid (source from sunflower oil 19.3%, hazelnut oil 43.8%, palm oil 36.9%) [31] and lastly study done by Lucci et al. 2015 that compared between rich extra-virgin olive oil (EVOO) diet and hybrid palm oil-rich (HPO) diet.

All studies either assessed weight and/or BMI as the primary outcomes, and none of the studies assessed incidence or prevalence of obesity. Detailed description of study characteristics is depicted in Table 1.

Only study done by Iggman 2014 was considered to have low risks of bias across all domains [30] .

Risk of bias profiles of the remaining studies were unclear in most domains due to insufficient information provided by the authors. A detailed risk of bias assessment, displayed at the domain level and the study level, and are available in Figure 3 and Figure 4 respectively.

We included five studies that evaluated body weight and BMI of the participants using palm oil. However, all studies were only represented by a small number of

participants. Overall, the effect of palm oil on BMI or body weight has not been shown to achieve statistically significant association.

There is insufficient evidence to relate the association between dietary palm oil intake and obesity. There are no clear differences between palm oil and its comparators across all outcomes.

Through a comprehensive search strategy, five studies that matched our selection criteria in terms of population, intervention, comparison and outcomes. A total of 292 participants were assessed. We believe the studies gathered in this review, although small in number, represented the best available evidence to answer the question that we posed in conducting this review. Two studies were conducted in the US and another three studies in Sweden, Denmark and Colombia. These studies could limit the applicability of the findings to other parts of the world since we didn’t have any studies that represented Asia. We were unable to undertake any subgroup analyses to further determine the applicability of the findings due to insufficient data.

There was overall low-quality evidence provided by a small number of studies. The five included studies had low to high risk of bias in most domains except blinding, but the quality of evidence was downgraded by wide 95% CI in the estimates (see Appendix 2).

We performed a comprehensive search from multiple databases with independent screening, selection and assessment of eligible studies. The main downgrading factor was imprecision, as most outcomes were contributed by single study.

Our findings are consistent with a published article by Mancini et al. that covered palm oil as the most widely used vegetable oil in the world [15] . The study stated that no clear evidence has been provided to unequivocally demonstrate the association of palm oil consumption and increase in cardiovascular risk. However, to date, there was no study done directly for effect of dietary palm oil on obesity or adipose tissue cellularity in human.