TABLE 2.

Should ketogenic therapies compared with control be used for people with Alzheimer's disease?

GRADE domain	Judgement	Concerns about certainty domains
Methodological limitation of the studies	All trials in this sample (44, 46–48, 53) were judged to be at low risk of methodological bias in the assessed domains. There were some concerns in 2 studies (44, 47) about baseline differences between groups, however, we did not warrant this as serious enough to downgrade the level of evidence in this category	Not serious
Indirectness	Overall, the subsample of studies in AD was fairly well characterized. All studies were performed in people with mild to moderate AD. All studies used exogenous ketogenic agents in the form of MCTs, however, there are differences in fatty acid composition across MCT products which may affect the efficacy of the intervention. All studies used the ADAS-Cog as the validated scale to measure cognition (primary outcome) (46–48, 53), except for the trial by Torosyan (44) that measured regional cerebral flood flow. We judged the evidence to have no serious indirectness	Not serious
Imprecision	The total sample size across studies was 634, or 618 when considering the studies that all used the same primary outcome measure (ADAS-Cog). Based on Optimal Information Size (OIS), the sample exceeds the number of patients generated by a conventional sample size calculation for a single adequately powered trial. All studies were randomized and placebo-controlled, so the comparator was “no intervention.” Overall, 4/5 studies (44, 46–48) reported nonsignificant results when subgroup analyses are not considered	Not serious
Inconsistency	The direction and magnitude of effect varied across the different trials. Overall, the results showed either a small reduction in symptoms (i.e. an improvement in cognition) or no statistically significant change. When participants were stratified for APOE-ε4 status, APOE-ε4(–) participants experienced statistically significant improvements from baseline. One study (53) reported statistically significant reductions in the ADAS-Cog after 30 d treatment (52 participants), whereas other studies with larger samples (46–48) showed no statistically significant effects overall. In 1/5 studies (48), there was no change, and in some outcomes, participants worsened (MMSE, APOE-ε4(+) participants in ADAS-Cog). No significant between-group differences were observed. In 3/5 studies (44, 46, 48), there was no statistically significant change after the treatment overall (notwithstanding subgroup analysis). 1/5 study (small sample, short study duration, overall low ROB) (53) showed statistically significant improvements in cognition in the treatment group (MCT). We judged the evidence to be at serious risk of inconsistency	Serious
Publication bias	We did not suspect publication bias, due to the presence of both positive and negative trials in the evidence base. Furthermore, the systematic search strategy was comprehensive and covered an extensive number of databases	Not suspected

AD, Alzheimer's disease; ADAS-Cog, Alzheimer's Disease Assessment Scale-Cognitive Subscale; APOE-ε4, apolipoprotein E allele ε4; GRADE, Grading of Recommendations, Assessment, Development and Evaluation; MCT, medium-chain triglycerides; MMSE, Mini-Mental State Examination; ROB, risk of bias.