The Effect of Diabetes-Specific Enteral Nutrition Formula on Cardiometabolic Parameters in Patients with Type 2 Diabetes: A Systematic Review and Meta–Analysis of Randomised Controlled Trials

Background: The prevalence of diabetes is on the increase in the UK and worldwide, partly due to unhealthy lifestyles, including poor dietary regimes. Patients with diabetes and other co-morbidities such as stroke, which may affect swallowing ability and lead to malnutrition, could benefit from enteral nutrition, including the standard formula (SF) and diabetes-specific formulas (DSF). However, enteral nutrition presents its challenges due to its effect on glycaemic control and lipid profile. Aim: The aim of this review was to evaluate the effectiveness of diabetes-specific enteral nutrition formula versus SF in managing cardiometabolic parameters in patients with type 2 diabetes. Method: This review was conducted in accordance with the preferred reporting items for systematic reviews and meta-analyses. Three databases (Pubmed, EMBASE, PSYCInfo) and Google scholar were searched for relevant articles from inception to 2 January 2019 based on Population, Intervention, Comparator, Outcomes and Study designs (PICOS) framework. Key words, Medical Subject Heading (MeSH) terms, and Boolean operators (AND/OR) formed part of the search strategy. Articles were evaluated for quality and risks of bias. Results: Fourteen articles were included in the systematic review and five articles were selected for the meta-analysis. Based on the findings of the review and meta-analysis, two distinct areas were evident: the effect of DSF on blood glucose parameters and the effect of DSF on lipid profile. All fourteen studies included in the systematic review showed that DSF was effective in lowering blood glucose parameters in patients with type 2 diabetes compared with SF. The results of the meta-analysis confirmed the findings of the systematic review with respect to the fasting blood glucose, which was significantly lower (p = 0.01) in the DSF group compared to SF, with a mean difference of −1.15 (95% CI −2.07, −0.23) and glycated haemoglobin, which was significantly lower (p = 0.005) in the DSF group compared to the SF group following meta-analysis and sensitivity analysis. However, in relation to the sensitivity analysis for the fasting blood glucose, differences were not significant between the two groups when some of the studies were removed. Based on the systematic review, the outcomes of the studies selected to evaluate the effect of DSF on lipid profile were variable. Following the meta-analysis, no significant differences (p > 0.05) were found between the DSF and SF groups with respect to total cholesterol, LDL cholesterol and triglyceride. The level of the HDL cholesterol was significantly higher (p = 0.04) in the DSF group compared to the SF group after the intervention, with a mean difference of 0.09 (95% CI, 0.00, 0.18), although this was not consistent based on the sensitivity analysis. The presence of low glycaemic index (GI) carbohydrate, the lower amount of carbohydrate and the higher protein, the presence of mono-unsaturated fatty acids and the different amounts and types of fibre in the DSF compared with SF may be responsible for the observed differences in cardiometabolic parameters in both groups. Conclusion: The results provide evidence to suggest that DSF is effective in controlling fasting blood glucose and glycated haemoglobin and in increasing HDL cholesterol, but has no significant effect on other lipid parameters. However, our confidence in these findings would be increased by additional data from further studies.

In patients with diabetes who are on enteral nutrition, the enteral feeds provided can be in the form of either Standard Formulas (SF) or Diabetes Specific Formulas (DSF). Enteral feeding formulas have a tendency to promote hyperglycaemia and insulinemic responses in patients with diabetes and in healthy subjects [16,17]. In addition, the effect of enteral nutrition on blood glucose parameters may be due to the fact that continuous enteral feeding is a source of continuous supply of glucose, providing 10-20 g of carbohydrates per hour, which is not the same during normal eating [15]. The absence of the normal postprandial glucose peak in patients with diabetes on enteral nutrition makes the management of hyperglycaemia difficult [15]. On the other hand, the effect of different types and amounts of fibre and mono-unsaturated fatty acids in various enteral feeds may influence lipid profile and other cardiometabolic parameters such as fasting blood glucose and glycated haemoglobin in patients with type 2 diabetes [18]. The role of the different enteral feeding formulas such as SF and DSF and their impact on cardiometabolic parameters in patients with diabetes continues to generate interest and controversy, and there appears to be no consensus among researchers on the most effective management strategy for these patients.
DSFs usually contain carbohydrates with low GI such as fructose and large amounts of monounsaturated fatty acids in varying amounts, which have effect on glycaemic control [17][18][19][20]. On the other hand, SFs are often high in carbohydrate and contain only low to moderate levels of lipids and do not have dietetic fibre [17].
Previous reviews on the use of enteral nutrition in patients with diabetes [16,17,[21][22][23] either lacked consensus in the recommendations, were based only on glycaemic control or did not involve meta-analysis. In addition, concerns remain with the use of DSF in terms of the safety and tolerance of relatively high levels of fat and fructose with respect to lipid metabolism and lactic acidosis, despite its advantage in improving blood glucose compared with SFs [16,19]. Therefore, this review provides a quantitative assessment of the relative effectiveness of DSF compared with SF.
Aim: The aim was to evaluate the effectiveness of diabetes specific enteral nutrition formula versus SF in managing cardiometabolic parameters in patients with type 2 diabetes.

Methods
This study was conducted in accordance with the preferred reporting items for systematic reviews and meta-analyses (PRISMA) [24].

Types of Studies and Participants
Only randomised controlled studies were included in this review and participants were patients with type 2 diabetes.
Inclusion and Exclusion Criteria.
The criteria for considering studies for the review are outlined in Table 1.

Inclusion Criteria Exclusion Criteria Population
Patients with type 2 diabetes and on enteral nutrition irrespective of type of feeding tube.
Patients with type 1 diabetes. Pregnant women with gestational diabetes. Healthy individuals without diabetes on enteral nutrition.
Patients with diabetes on parenteral nutrition and parenteral plus enteral nutrition.

Intervention
Diabetes specific formulas Parenteral nutrition, parenteral plus enteral nutrition. (Oral nutrition supplement or enteral tube feeding)

Type of Intervention
The intervention for this review was based on diabetes-specific enteral formula, irrespective of the type of feeding tube, mode and rate of delivery of the enteral feed and clinical settings.

Types of Outcome Measures
The following were the outcome measures of interest; • Blood glucose parameters-Fasting blood glucose and glycated haemoglobin.

Search Strategy
Databases encompassing Pubmed, EMBASE, PSYCInfo and Google scholar were searched for relevant articles based on the Population (Patients with diabetes), Intervention (Diabetes Specific Formula), Comparator (Standard enteral formulas), Outcomes (outcome measures) and Study designs (Randomised controlled studies)-PICOS framework ( Table 2) [25]. The use of key words, truncation symbols, Medical Subject Heading (MeSH) terms and Boolean operators (AND/OR) formed part of the search strategy. Searches were conducted from the date of inception of databases until 2 January 2019.
The screening of studies and the evaluation of their eligibility and inclusion were in line with PRISMA [24] guidelines ( Figure 1). These procedures were conducted by five researchers (OO, SMW, TT, RC, X-HW) and differences were resolved through consensus.

Data Extraction
All the articles from different databases were exported to ENDNote (Analytics, Philadelphia, PA, USA) for de-duplication. Data extraction was carried out by one researcher (OO) and cross-checked by the other four researchers (SMW, TT, RC, X-HW).

Assessment of Risk of Bias and Evaluation of Quality
A critical appraisal skills programme (CASP) tool was used to appraise the quality of the articles [26]. In addition, the researchers carried out an assessment of the risk of bias using the domain-based tool (random sequence generation, allocation concealment, blinding of participants, personnel and outcome assessment, reporting bias and selective reporting) to evaluate the studies included [27].

Data Extraction
All the articles from different databases were exported to ENDNote (Analytics, Philadelphia, PA, USA) for de-duplication. Data extraction was carried out by one researcher (OO) and cross-checked by the other four researchers (SMW, TT, RC, X-HW).

Assessment of Risk of Bias and Evaluation of Quality
A critical appraisal skills programme (CASP) tool was used to appraise the quality of the articles [26]. In addition, the researchers carried out an assessment of the risk of bias using the domain-based tool (random sequence generation, allocation concealment, blinding of participants, personnel and outcome assessment, reporting bias and selective reporting) to evaluate the studies included [27].

Statistical Analysis
Articles that met the inclusion criteria for meta-analysis were exported to RevMan (Review Manager, 5.3) [28] for data analysis. Therefore, cross-over studies and other studies which presented with difficulty in extracting suitable data were excluded from the meta-analysis. The data analysis included both meta-analysis and sensitivity analysis, the latter being conducted to test the consistency of the effect of DSFs on the different cardiometabolic paramters. The random effects model was used for the parameters of interest due to the high level of heterogeneity measured by the statistic I 2 with values ranging from 34% to 100%. A p value of 0.10 was used to determine the statistical significance of heterogeneity.

Effect Size
A forest plot was used to present the results of the meta-analysis and statistical significance for the overall effect of the intervention was determined by a p value of <0.05.

Data Inclusion Decisions
Fasting blood glucose in the studies included was measured after overnight fasting, using standard measuring instruments. This is the standard method of measuring fasting blood glucose: the blood glucose concentrations were expressed as Means. However, the studies by Pohl et al. [29,30] were expressed as median and interquartile ranges and these were converted to means and standard deviations [27]. Fourteen studies were included in the systematic review (Table 3) while only five studies [29][30][31][32][33] were selected for the meta-analysis (Table 4).      Table).

Study Reference
Interventions    Figure 2 shows the risk of bias summary of the various studies included in the meta-analysis. All the studies demonstrated low risk of bias in all the areas, except with respect to incomplete outcome data (attrition bias) where two studies [29,33] showed high risk of bias.  Based on the findings of the review and the meta-analysis, two distinct areas were evident: the effect of DSF on blood glucose parameters and the effect of DSF on lipid profile.

Assessment of Risk of Bias in Included Studies
The effect DSF on blood glucose parameters: All the fourteen studies included in the systematic review showed that DSF was effective in lowering blood glucose parameters in patients with type 2 diabetes compared with SF. In particular, DSF improved glycaemic control and lowered insulin requirements [18,29,31,[35][36][37][38][39]. It provided better clinical outcomes, including reducing the risk of acquired infections and pressure ulcer, reduced body weight and was safer compared to SF [29,31,32,]. In addition, the use of DSF was shown to be effective in lowering postprandial blood glucose levels compared to SF [20,32,34,40].
Pohl et al. [30] observed that long-term tube feeding with a DSF significantly lowered fasting blood glucose and improved glycaemic control. Similarly, Vaisman et al. [33] reported that DSF significantly improved longer-term glycaemic control in diabetic patients compared to SF. The results of the meta-analysis confirmed the findings of the systematic review. With respect to the fasting blood glucose, it was significantly lower (p = 0.01) in the DSF group compared to SF, with a mean difference of −1.15 (95% CI −2.07, −0.23) (Figure 3). However, in relation to the sensitivity analysis, there were no significant differences (p > 0.05) between the two groups with the removal of Pohl et al. [29,30] studies. Based on the findings of the review and the meta-analysis, two distinct areas were evident: the effect of DSF on blood glucose parameters and the effect of DSF on lipid profile.
The effect DSF on blood glucose parameters: All the fourteen studies included in the systematic review showed that DSF was effective in lowering blood glucose parameters in patients with type 2 diabetes compared with SF. In particular, DSF improved glycaemic control and lowered insulin requirements [18,29,31,[35][36][37][38][39]. It provided better clinical outcomes, including reducing the risk of acquired infections and pressure ulcer, reduced body weight and was safer compared to SF [29,31,32]. In addition, the use of DSF was shown to be effective in lowering postprandial blood glucose levels compared to SF [20,32,34,40].
Pohl et al. [30] observed that long-term tube feeding with a DSF significantly lowered fasting blood glucose and improved glycaemic control. Similarly, Vaisman et al. [33] reported that DSF significantly improved longer-term glycaemic control in diabetic patients compared to SF. The results of the meta-analysis confirmed the findings of the systematic review. With respect to the fasting blood glucose, it was significantly lower (p = 0.01) in the DSF group compared to SF, with a mean difference of −1.15 (95% CI −2.07, −0.23) (Figure 3). However, in relation to the sensitivity analysis, there were no significant differences (p > 0.05) between the two groups with the removal of Pohl et al. [29,30] studies. The glycated haemoglobin was significantly lower (p = 0.005) in the DSF group compared to the SF group following meta-analysis ( Figure 4) and sensitivity analysis.

The Effect of DSF on Lipid Profile
Based on the systematic review, the outcomes of the studies selected to evaluate the effect of DSF on lipid profile were variable. Craig et al. [31] did not find significant differences with respect to LDL cholesterol and triglyceride between the DSF and the SF groups, but differences were significantly higher (p < 0.05) in the DSF group in relation to HDL cholesterol. In two other studies [33,35], the level of HDL cholesterol was significantly higher (p < 0.05) in the DSF group compared with the SF group after intervention, but differences were not significant (p > 0.05) in relation to triglycerides, total cholesterol and LDL cholesterols. Differences between DSF and SF were also not significant (p > 0.05) in terms of triglyceride, total cholesterol, HDL cholesterol and LDL cholesterol in other studies [30,32,38]. In contrast, Pohl et al. [29] reported that there was a significant difference (p < 0.05) between the DSF group and the SF group with respect to triglycerides, but differences were not significant (p > 0.05) in relation to total cholesterol, HDL and LDL cholesterol. Other studies [36,40] have also shown that DSF is effective in controlling plasma triglyceride.
Following meta-analysis, no significant differences (p > 0.05) were found between the DSF and SF groups with respect to total cholesterol, LDL cholesterol and triglyceride (Figures 5-7). However, the DSF group had a significantly higher level (p = 0.04) of HDL cholesterol compared to the SF group after the intervention, with a mean difference of 0.09 (95% CI, 0.00, 0.18) (Figure 8). The results of the sensitivity test for HDL cholesterol demonstrated no significant differences (p > 0.05) between the two groups when the Craig et al. [31] and Vaisman et al. [33] studies were removed from the analysis. In addition, the sensitivity analysis showed no significant differences (p > 0.05) between the two groups with respect to total cholesterol and triglyceride, while significant differences (<0.05) were observed in relation to LDL cholesterol when the Craig et al. [31] study was removed. The glycated haemoglobin was significantly lower (p = 0.005) in the DSF group compared to the SF group following meta-analysis ( Figure 4) and sensitivity analysis. The glycated haemoglobin was significantly lower (p = 0.005) in the DSF group compared to the SF group following meta-analysis ( Figure 4) and sensitivity analysis.

The Effect of DSF on Lipid Profile
Based on the systematic review, the outcomes of the studies selected to evaluate the effect of DSF on lipid profile were variable. Craig et al. [31] did not find significant differences with respect to LDL cholesterol and triglyceride between the DSF and the SF groups, but differences were significantly higher (p < 0.05) in the DSF group in relation to HDL cholesterol. In two other studies [33,35], the level of HDL cholesterol was significantly higher (p < 0.05) in the DSF group compared with the SF group after intervention, but differences were not significant (p > 0.05) in relation to triglycerides, total cholesterol and LDL cholesterols. Differences between DSF and SF were also not significant (p > 0.05) in terms of triglyceride, total cholesterol, HDL cholesterol and LDL cholesterol in other studies [30,32,38]. In contrast, Pohl et al. [29] reported that there was a significant difference (p < 0.05) between the DSF group and the SF group with respect to triglycerides, but differences were not significant (p > 0.05) in relation to total cholesterol, HDL and LDL cholesterol. Other studies [36,40] have also shown that DSF is effective in controlling plasma triglyceride.
Following meta-analysis, no significant differences (p > 0.05) were found between the DSF and SF groups with respect to total cholesterol, LDL cholesterol and triglyceride (Figures 5-7). However, the DSF group had a significantly higher level (p = 0.04) of HDL cholesterol compared to the SF group after the intervention, with a mean difference of 0.09 (95% CI, 0.00, 0.18) (Figure 8). The results of the sensitivity test for HDL cholesterol demonstrated no significant differences (p > 0.05) between the two groups when the Craig et al. [31] and Vaisman et al. [33] studies were removed from the analysis. In addition, the sensitivity analysis showed no significant differences (p > 0.05) between the two groups with respect to total cholesterol and triglyceride, while significant differences (<0.05) were observed in relation to LDL cholesterol when the Craig et al. [31] study was removed.

The Effect of DSF on Lipid Profile
Based on the systematic review, the outcomes of the studies selected to evaluate the effect of DSF on lipid profile were variable. Craig et al. [31] did not find significant differences with respect to LDL cholesterol and triglyceride between the DSF and the SF groups, but differences were significantly higher (p < 0.05) in the DSF group in relation to HDL cholesterol. In two other studies [33,35], the level of HDL cholesterol was significantly higher (p < 0.05) in the DSF group compared with the SF group after intervention, but differences were not significant (p > 0.05) in relation to triglycerides, total cholesterol and LDL cholesterols. Differences between DSF and SF were also not significant (p > 0.05) in terms of triglyceride, total cholesterol, HDL cholesterol and LDL cholesterol in other studies [30,32,38]. In contrast, Pohl et al. [29] reported that there was a significant difference (p < 0.05) between the DSF group and the SF group with respect to triglycerides, but differences were not significant (p > 0.05) in relation to total cholesterol, HDL and LDL cholesterol. Other studies [36,40] have also shown that DSF is effective in controlling plasma triglyceride.
Following meta-analysis, no significant differences (p > 0.05) were found between the DSF and SF groups with respect to total cholesterol, LDL cholesterol and triglyceride (Figures 5-7). However, the DSF group had a significantly higher level (p = 0.04) of HDL cholesterol compared to the SF group after the intervention, with a mean difference of 0.09 (95% CI, 0.00, 0.18) (Figure 8). The results of the sensitivity test for HDL cholesterol demonstrated no significant differences (p > 0.05) between the two groups when the Craig et al. [31] and Vaisman et al. [33] studies were removed from the analysis. In addition, the sensitivity analysis showed no significant differences (p > 0.05) between the two groups with respect to total cholesterol and triglyceride, while significant differences (<0.05) were observed in relation to LDL cholesterol when the Craig et al. [31] study was removed.

Discussion
The findings of the systematic review and meta-analysis revealed that DSF was effective in lowering blood glucose (fasting blood glucose and glycated haemoglobin) compared with SF in patients with type 2 diabetes. However, the sensitivity analysis for the fasting blood glucose did not demonstrate a significant difference (p > 0.05) with the removal of the Pohl et al. [29,30] studies. In addition, there were no significant differences (p > 0.05) between the DSF and SF groups with respect to total cholesterol, LDL cholesterol and triglyceride (although a few studies reported significant differences with respect to triglyceride). Differences in the outcomes of studies in the systematic review were observed with respect to the effect of DSF on HDL cholesterol and the meta-analysis also showed significantly higher levels for the DSF group compared with the SF group. The high level of heterogeneity in the studies included in the meta-analyses may explain why the results of the metaanalysis and the sensitivity analysis were not consistent with respect to HDL cholesterol and fasting blood glucose.

Discussion
The findings of the systematic review and meta-analysis revealed that DSF was effective in lowering blood glucose (fasting blood glucose and glycated haemoglobin) compared with SF in patients with type 2 diabetes. However, the sensitivity analysis for the fasting blood glucose did not demonstrate a significant difference (p > 0.05) with the removal of the Pohl et al. [29,30] studies. In addition, there were no significant differences (p > 0.05) between the DSF and SF groups with respect to total cholesterol, LDL cholesterol and triglyceride (although a few studies reported significant differences with respect to triglyceride). Differences in the outcomes of studies in the systematic review were observed with respect to the effect of DSF on HDL cholesterol and the meta-analysis also showed significantly higher levels for the DSF group compared with the SF group. The high level of heterogeneity in the studies included in the meta-analyses may explain why the results of the metaanalysis and the sensitivity analysis were not consistent with respect to HDL cholesterol and fasting blood glucose.

Discussion
The findings of the systematic review and meta-analysis revealed that DSF was effective in lowering blood glucose (fasting blood glucose and glycated haemoglobin) compared with SF in patients with type 2 diabetes. However, the sensitivity analysis for the fasting blood glucose did not demonstrate a significant difference (p > 0.05) with the removal of the Pohl et al. [29,30] studies. In addition, there were no significant differences (p > 0.05) between the DSF and SF groups with respect to total cholesterol, LDL cholesterol and triglyceride (although a few studies reported significant differences with respect to triglyceride). Differences in the outcomes of studies in the systematic review were observed with respect to the effect of DSF on HDL cholesterol and the meta-analysis also showed significantly higher levels for the DSF group compared with the SF group. The high level of heterogeneity in the studies included in the meta-analyses may explain why the results of the metaanalysis and the sensitivity analysis were not consistent with respect to HDL cholesterol and fasting blood glucose.

Discussion
The findings of the systematic review and meta-analysis revealed that DSF was effective in lowering blood glucose (fasting blood glucose and glycated haemoglobin) compared with SF in patients with type 2 diabetes. However, the sensitivity analysis for the fasting blood glucose did not demonstrate a significant difference (p > 0.05) with the removal of the Pohl et al. [29,30] studies. In addition, there were no significant differences (p > 0.05) between the DSF and SF groups with respect to total cholesterol, LDL cholesterol and triglyceride (although a few studies reported significant differences with respect to triglyceride). Differences in the outcomes of studies in the systematic review were observed with respect to the effect of DSF on HDL cholesterol and the meta-analysis also showed significantly higher levels for the DSF group compared with the SF group. The high level of heterogeneity in the studies included in the meta-analyses may explain why the results of the metaanalysis and the sensitivity analysis were not consistent with respect to HDL cholesterol and fasting blood glucose.

Discussion
The findings of the systematic review and meta-analysis revealed that DSF was effective in lowering blood glucose (fasting blood glucose and glycated haemoglobin) compared with SF in patients with type 2 diabetes. However, the sensitivity analysis for the fasting blood glucose did not demonstrate a significant difference (p > 0.05) with the removal of the Pohl et al. [29,30] studies. In addition, there were no significant differences (p > 0.05) between the DSF and SF groups with respect to total cholesterol, LDL cholesterol and triglyceride (although a few studies reported significant differences with respect to triglyceride). Differences in the outcomes of studies in the systematic review were observed with respect to the effect of DSF on HDL cholesterol and the meta-analysis also showed significantly higher levels for the DSF group compared with the SF group. The high level of heterogeneity in the studies included in the meta-analyses may explain why the results of the meta-analysis and the sensitivity analysis were not consistent with respect to HDL cholesterol and fasting blood glucose.
The presence of low glycaemic index (GI) carbohydrate in the form of isomaltulose, the lower amount of carbohydrate and the higher protein content in the DSF may have contributed to the findings of this review [37]. In addition, the presence of mono-unsaturated fatty acids (MUFA) and the different amounts and types of fibre in the DSF compared with SF may be responsible for the observed differences in the fasting blood glucose, glycated haemoglobin and lipid profiles in both groups [18,41]. DSFs are usually higher in fat (40-50% of energy with a significant portion of MUFA) and have a lower carbohydrate level (30-40% of energy) and about 15% of energy is derived from fructose and soluble fibre [20]. DSFs contain carbohydrates with low GI such as non-hydrolysed starches, disaccharides, fibre and fructose in varying amounts which are aimed at controlling postprandial glucose [17][18][19][20]. In contrast, SFs are high in carbohydrate (about 50%) and have low-moderate levels of lipids (about 30%) and do not contain dietetic fibre [17]. A study by Hofman et al. [42] demonstrated that, in 12 enteral formulas examined, the GI ranged from 12 for DSFs up to 61 for SFs. The GI of food is a measure of how quickly the food is digested and the glucose reaches the blood stream [22,43]. Foods with high GI rapidly increase blood glucose and insulin responses after consumption [43,44]. The results from meta-analysis showed that the intake of a low GI diet was associated with reductions in blood glucose parameters [35,43]. In addition, high soluble fibre-containing foods can improve glycaemic control partly due to delayed absorption [36].
Therefore, DSFs may improve glycaemic control through delay in gastric emptying, delayed intestinal absorption of carbohydrate and lower glycaemic response [20]. In the study by Alish et al. [34], the blend of DSF was made up of low glycaemic and slowly digestible carbohydrates, resistant maltodextrin, isomaltulose, sucromalt and prebiotic fibres, including fructo-oligosaccharides. These constituents collectively produce a slow and consistent release of glucose into the blood stream [34]. Isomaltulose is a naturally occurring low GI slowly digestible carbohydrate [18]. The slower hydrolysation of isomaltulose during digestion may be responsible for the slower rise in blood glucose in patients with diabetes on DSF [18]. In addition, the higher protein content of the DSF may have contributed to the lowering of blood glucose parameters by delaying gastric emptying [18].
The use of high fat content, including MUFA, in the DSF may slow the transit time in the gastrointestinal tract and slow the absorption of sugars which could help improve glycaemic control [35,41]. Diets that are high in MUFA have been shown to increase HDL cholesterol and reduce other components of the lipid profiles [35,41]. HDL cholesterol is useful for reducing the risk of cardiovascular disease [35]. The result of the meta-analysis of the current review confirmed the positive role of DSF in increasing HDL cholesterol. However, the sensitivity tests did not demonstrate consistency in terms of the effect of DSF on HDL cholesterol, which could explain why researchers may be reluctant to recommend the use of high fat content in DSF due to the risk of alterations in lipid profiles [41]. This may also be due to the fact that there have been differences in the outcomes of studies on the effect of DSF on lipid profile [41].

Limitation
The limitation of this review was that only five studies were included in the meta-analysis. In particular, there were fewer studies included for lipid outcomes (three for several parameters) and there was substantial variability in the studies. Therefore, the differences between the meta-analysis and the sensitivity analysis in some of the parameters suggest that those results were not quite consistent, which may be due to the high level of heterogeneity of the studies. Therefore, more studies are needed to address this problem.

Conclusions
The results provide evidence to suggest that DSF is effective in controlling fasting blood glucose and glycated haemoglobin. In addition, DSF was effective in increasing HDL cholesterol but had no significant effect on other lipid parameters. However, our confidence in these findings would be increased by additional data from further studies. Additional research would also provide the opportunity to refine our understanding of the effect of DSF on cardiometabolic parameters.