A summary of the evidence and evaluation of the effectiveness of nonpharmacological interventions for mild cognitive impairment

Table 2 presents the characteristics of the 14 SRs16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29 of exercise interventions and included data from 455 to 117,410 participants. All SRs were published in English. One SR was published in 2019, three in 2018, four in 2017, two in 2016, two in 2015 and two in 2014. Most SRs included randomized controlled trials (RCTs) and only two SRs23, 24 included prospective cohort studies. The studies examined various, diverse intervention types. In the case of Tai Chi, aerobic exercise, and walking were the most commonly used interventions. The average duration of the intervention was 30–90 min/session in 1–5 sessions/week for 3–12 months.

The results of the assessment of the methodological quality of the SRs of exercise interventions are presented in Table 3. Eleven of the remaining SRs16, 17, 18, 19, 20, 21, 24, 25, 27, 28, 29 were of moderate quality, two SRs22, 23 were considered low quality, and one SR26 was considered very low quality.

Commonly unreported items were the presence of an a priori design (n=11, 78.6%), an explanation of the method used to select the study designs for inclusion in the review (n=14, 100%), the lack of a comprehensive literature search strategy and the inclusion of grey literature (n=6, 42.9%), and the provision of a list of excluded studies and justification for the exclusions (n=11, 78.6%). Notably, 21.4% reviews did not report any potential sources of conflicts of interest. In addition, six SRs used narrative reviews to describe their findings, whereas eight reviews used meta-analysis methods.

(1) Global cognitive function

Five SRs16, 18, 22, 27, 28 analyzed the effect of physical exercise on cognitive function in MCI patients, as measured using the ADAS-Cog, MMSE, and MoCA. According to three SRs,18, 27, 28 aerobic exercise can improve global cognitive function significantly than controls (GRADE: moderate, moderate, and high). In another SR22 showed that aerobic exercise can significantly increase MoCA scores and MMSE scores, but it did not significantly influence ADAS-Cog scores (GRADE: low, low, and very low). One SR16 revealed significant beneficial effects of aerobic exercises (n=5), mind–body exercises (n=7), and resistance exercises (n=4) on improving global cognition in MCI patients compared to controls (GRADE: moderate, very low, and low).

(2) Executive ability

Three SRs18, 22, 28 analyzed the effects of aerobic exercise on executive ability in MCI patients, and aerobic exercise exerted a significant effect on improving verbal fluency test (VFT) scores compared with controls (GRADE: very low, low, and moderate). One SR20 did not observe an effect of aerobic exercise on improving TMT scores (GRADE: moderate).

(3) Memory

Three SRs18, 22, 28 analyzed the effects of aerobic exercise on memory in MCI patients. Three SRs all included a subgroup analysis, and only one SR22 showed that aerobic exercise group can significantly improve MCI patients immediate recall and delayed recall ability compared with controls (GRADE: low and low). However, the other two SRs18, 28 did not observe an effect of aerobic exercise on improving immediate recall and delayed recall abilities (GRADE: moderate, moderate, low, and low).

Nineteen evidences of eight SRs of physical exercise were evaluated according to the GRADE system for the quality of evidence. One of them was high quality, seven of them were moderate quality, eight of them were low quality, and three of them were very low quality, which was downgraded due to the RoB, inconsistencies, and imprecision.

Table 4 presents the characteristics of the six SRs30, 31, 32, 33, 34, 35 of cognitive intervention. Across the studies, the sample size varied considerably, ranging from 224 participants to 2,177 participants; the average sample size was 945 participants. One SR was published in 2019, one in 2018, two in 2017, and two in 2016. All SRs included RCTs, four SRs30, 31, 33, 35 used meta-analysis methods, whereas two SRs32, 34 describe their findings by narrative reviews.

The results of the assessment of the methodological quality of the SRs of cognitive intervention are presented in Table 5. Six SRs30, 31, 32, 33, 34, 35 evaluated the efficacy of cognitive interventions, and the SRs were rated as having moderate (n=2),34, 35 low (n=2),32, 33 and very low quality (n=2).30, 31

None of the SRs reported funding sources for the studies included in the SRs, and the results of industry-funded studies may occasionally favor sponsored products. Only two SRs employed an a priori design. Two SRs used narrative reviews to describe their findings and four SRs used meta-analysis. Commonly unreported items were an explanation of the methods used to select the study designs included in the review (n=6, 100%). Three reviews did not assess RoB when interpreting and discussing the results of their reviews.

(1) Global cognitive function

Four SRs30, 31, 32, 35 analyzed the effect of cognitive training on cognitive function in MCI patients. One SR32 showed little effect of memory training on improving global cognitive ability compared with controls, but the other three SRs30, 31, 35 reported a moderately significant effect of computerized cognitive training on improving cognitive function among MCI patients (GRADE: very low, very low, and low).

(2) Memory

Four SRs31, 32, 33, 34, 35 analyzed the effect of cognitive interventions on memory in MCI patients. Two SRs32, 35 revealed moderate and statistically significant effects of cognitive interventions on working memory. Only one SR31 described a significant improvement in the delayed recall ability of participants who received mnemonic training compared with controls (GRADE: moderate). However, another SR33 did not observe an effect of computer cognitive training on improving memory (GRADE: very low).

(3) Attention

Three SRs31,33,35 evaluated the efficacy of cognitive training on attention and showed that cognitive training exerted a statistically significant effect on the attention capacity compared with controls (GRADE: low, moderate, and moderate).

(4) Orientation

Two SRs evaluated the efficacy of cognitive training on orientation. One SR31 reported a statistically significant effect of cognitive training on orientation compared with controls (GRADE: moderate). However, another SR33 did not observe an effect of cognitive training (GRADE: very low).

Ten evidences of four SRs of cognitive intervention were evaluated according to the GRADE system for the quality of evidence. Four of them were moderate quality, two of them were low quality, and four of them were very low quality, which was downgraded due to the RoB of the included literatures, inconsistencies, and imprecision of the treatment effect.

Table 6 presents the characteristics of the four SRs36, 37, 38, 39 of nonpharmacological interventions using traditional Chinese medicine. The effects of acupuncture were only reported in, and the most recent search included studies published throughout 2016.36 The SRs included data from 5 to 18 primary prevention trials with 565 to 1,095 participants. All four SRs performed a meta-analysis and reported safety.

The results of the assessment of the methodological quality of the SRs of nonpharmacological interventions using traditional Chinese medicine are presented in Table 7. Only one SR36 was considered moderate quality and three SRs 37, 38, 39ere considered low quality.

None of the SRs reported funding sources for the studies included in the SRs (n=4, 100%), the presence of an a priori design (n=4, 100%), an explanation of the method used to select the study designs included in the review (n=0, 0%), duplication of selected studies (n=1, 25%), and three reviews did not assess RoB when interpreting and discussing the results of their reviews (n=3, 75%). Additionally, two studies failed to report any potential sources of conflicts of interest (n=2, 50%).

(1) MMSE

Three of four SRs36, 37, 39 compared the efficacy of Western medicine alone and acupuncture combined with Western medicine on MMSE scores, and showed a significant difference in MMSE scores. The combination of acupuncture combined with Western medicine produced a statistically significant decrease in cognitive decline (GRADE: very low, low, and low). In addition, one SR reported a significant effect on the outcome of MMSE scores between the two study groups that received acupuncture and acupuncture combined with aricept39 (GRADE: moderate). Two SRs36, 38 revealed a better MMSE score for patients who received acupuncture treatment for MCI than patients who received nimodipine alone (GRADE: moderate and low), as cognitive function was improved to some extent.

(2) MoCA

Only one SR38 showed an increase in the MoCA score for patients who received acupuncture at multiple sites on the head combined with cognitive training for MCI compared with patients subjected to cognitive training alone (GRADE: low). Importantly, cognitive function was improved to some extent.

All four SRs37, 38, 39 provided detailed descriptions of adverse events. The adverse events of acupuncture may include fainting during treatment, errhysis at the needle sites, and minor hematoma due to a shorter time of applying local pressure at the acupoint, but these adverse effects did not impact the treatment for MCI. Meanwhile, the adverse events of nimodipine therapy may include gastrointestinal reactions and mild headache.

Seven evidences of four SRs of nonpharmacological interventions using traditional Chinese medicine were evaluated according to the GRADE system for the quality of evidence. Two of them were moderate quality, four of them were low quality, and one of them was very low quality, which was downgraded due to the RoB of the included literatures, inconsistencies, and imprecision of the treatment effect.

Table 8 presents the characteristics of the seven SRs40, 41, 42, 43, 44, 45, 46 of dietary interventions, and included data from 900 to 84,481 participants. Of these SRs, four examined Mediterranean diets; one analyzed vitamin B supplementation; and two assessed tea, coffee, and caffeine. One SR was published in 2017, two in 2016, two in 2015, and two in 2014. Only one SR explicitly included RCTs, three SRs40, 42, 46 explicitly included cohort studies, two SRs41, 44 included both cohort studies and longitudinal studies, and one SR43 included case–control studies, longitudinal and cross-sectional. Four SRs40, 42, 45, 46 used meta-analysis methods, whereas three SRs41, 43, 44 describe their findings by narrative reviews.

The results of the assessment of the methodological quality of the SRs of dietary interventions are presented in Table 9. Three SRs40, 45, 46 were considered to have moderate quality and majority of the included SRs41, 42, 43, 44 were of low and very low quality.

Commonly unreported items were an explanation of the methods used to select the study designs included in the review (n=7, 100%), reporting conflicts of interest of included studies (n=7, 100%), the presence of an a priori study design (n=6, 86%), the lack of a comprehensive literature search strategy and the inclusion of grey literature (n=4, 71%), duplication of selected studies or data extraction (n=2, 29%), and a satisfactory technique for assessing the RoB was not appropriately performed in four reviews (57%).

(1) Cognitive effects of vitamin B supplementation on patients with MCI

Only one SR45 analyzed the effects of multivitamin B supplements on people with MCI, and the results showed that had moderate beneficial on memory. However, no significant differences in general cognitive function, executive function, and attention were observed (GRADE: low, moderate, moderate, and low).

(2) Cognitive effects of a Mediterranean diet on patients with MCI

A Mediterranean diet is characterized by the consumption of large amounts of vegetables, fruit, cereals, legumes, and unsaturated fatty acids, small amounts saturated fatty acids and meat, small-to-moderate amounts of dairy products, moderate-to-large amounts of fish and regular but moderate consumption of alcohol.40, 41 At present, some articles have reported a positive effect of a Mediterranean diet on MCI.

One SR40 included nine articles, and follow-up period ranged from 2.2 to 12 years. The Mediterranean diet score (category of high vs. low) was significantly associated with the incident risk of cognitive disorders (GRADE: moderate). However, the Mediterranean diet score (category of median vs. low) was not significantly associated with the risk of developing cognitive disorders (GRADE: low). In addition, there was no significant linear association between the Mediterranean diet and the incident risk of all types of cognitive disorders41. Singh B46 and colleagues reported that highest MeDi tertile had a 33% lower risk of cognitive impairment than the lowest MeDi score tertile (GRADE: moderate). In addition, another SR41 also revealed higher adherence to a Mediterranean diet can slower rate of cognitive decline and reduced conversion to Alzheimer’s disease. The specific cognitive domains that improved as the MeDi score increased were memory, executive function, and visual constructs. Furthermore, another SRs44 showed that better adherence to a Mediterranean diet can reduce cognitive decline.

(3) Cognitive effects of coffee, tea, and caffeine on patients with MCI

Eleven prospective studies, including 29,155 participants, were included in the SRs.42 The dose– response analysis did not show an association between increased coffee intake and cognitive decline (GRADE: low) or cognitive impairment (GRADE: low). In addition, another SR43 included some cross-sectional, case–control, and longitudinal population-based studies that evaluated the long-term effects on brain function and provided some evidence supporting protective effects of tea, coffee, and caffeine used on late-life cognitive impairment/decline, but the association lacked a distinct dose–response association.

Nine evidences of four SRs of dietary intervention were evaluated according to the GRADE system for the quality of evidence. Four of them had moderate quality and five of them had low quality. GRADE evidence from observational studies was initially described as low-quality evidence, and observational studies generally had not been upgraded due to lack the upgrade factors.