Effects of 15:9 time-restricted eating on cognitive function and anthropometric parameters among the elderly with mild cognitive impairment: an assessor-blinded, pilot randomized controlled trial

Abstract

Background

The effects of 15:9 time-restricted eating (TRE) on cognitive function and anthropometric parameters among the elderly with mild cognitive impairment (MCI) requires further investigation.

Objective

This assessor-blinded, pilot randomized controlled trial aimed to examine the effects of 12 weeks of 15:9 TRE on cognitive function and anthropometric parameters in older adults with MCI.

Methods

Forty-six eligible participants (mean age 74.20 ± 5.09 years) were recruited from a community elderly service center in Chengdu, Sichuan, China, and were randomly allocated to either the TRE group (n = 23) or the control group (n = 23). The TRE group followed 15:9 TRE for 12 weeks and the control group maintained eating ad libitum without any modifications. The primary outcome were changes in the Montreal Cognitive Assessment (MoCA) and the auditory verbal learning test-Huashan version (AVLT-H) and their subitems. The secondary outcome were changes in body weight, body mass index (BMI), waist circumference, calf circumference, and adherence. All outcomes, except for adherence, were measured consistently at baseline, 6 weeks, and 12 weeks.

Results

There was a significant improvement in the MoCA total score (from baseline to 6-week: β = 1.22, 95% CI: 0.33 to 2.10, P = 0.007; from baseline to 12-week: β = 1.70, 95% CI: 0.46 to 2.94, P = 0.007) and the attention score of MoCA (from baseline to 6-week: β = 0.65, 95% CI: 0.15 to 1.15, P = 0.011) in the TRE group compared with the control group. The AVLT-H recognition score (AVLT-H-REC-S) was significantly improved in the TRE group compared with the control group (from baseline to 12-week: β = 2.44, 95% CI: 0.63 to 4.24, P = 0.008). There was no significant difference in body weight, BMI, waist circumference, or calf circumference between the two groups. The mean adherence to TRE was more than 5 days per week in the TRE group.

Conclusions

Short-term improvements in MoCA total score and AVLT-H-REC-S were observed among elderly individuals with MCI following a 12-week 15:9 TRE. However, no significant effect of TRE on anthropometric parameters was found. These findings should be interpreted with caution due to the limited sample size. Larger-sample studies are needed to further clarify the underlying mechanisms by which TRE affects cognitive function in the elderly MCI population.

Trial registration

This trial was registered at the China Clinical Trial Registration Center on 5 November 2021, with registration number ChiCTR2100052766.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-025-06788-7.

Background

Dementia is a prevalent neurodegenerative disease among the elderly, and currently, there is no effective treatment available. Previous studies have reported that the estimated prevalence of dementia in China was 6.0% in 2020, which is comparable to the rates observed in most regions worldwide (5.5%–7.0%) [1]. Moreover, from 2019 to 2050, a significant increase in the number of individuals living with dementia is anticipated, with projections indicating a total of 152.8 million cases globally by 2050 [2]. For the elderly, maintaining normal cognitive function during aging is critical for ensuring a better health-related quality of life [3–5]. Once individuals progress to the dementia stage, various burdens can adversely affect caregivers’ physical, psychological, social, and financial well-being [6, 7]. It is widely agreed that targeting modifiable risk factors during the stage of mild cognitive impairment (MCI), which is recognized as the reversible phase preceding dementia, may contribute to delaying or slowing the progression of cognitive decline [8]. For example, proper nutrition, including adherence to a healthy and balanced diet, is recommended to mitigate cognitive decline and reduce the risk of developing dementia [9].

Intermittent fasting (IF), a widely recognized dietary approach for health, has demonstrated positive effects on cognitive function in older adults living with neurodegeneration [10–13]. Time-restricted eating (TRE), a diet form of IF, has attracted increasing attention from researchers as its core principle of restricting the daily eating window rather than limiting total caloric intake which may improve participant adherence. It has been proven that TRE can activate underlying mechanisms that mitigate markers of aging and neurodegenerative diseases, thereby delaying cognitive decline [14, 15]. Performing TRE could lower the risk of cognitive impairment among older adults [16]. In addition, the efficacy, safety, and feasibility of TRE in older adults with overweight or obesity have been reported, highlighting that TRE might be more suitable for the elderly compared to other forms of IF [17]. In this context, TRE shows the potential to be a non-pharmacological intervention for reversing MCI or mitigating the progression of dementia that is worthy of further investigation. However, only a limited number of interventional studies have examined the effects of TRE on cognitive function in older adults which did not limit the baseline cognitive status [17, 18]. Furthermore, there are few original studies on TRE with cognitive evaluation as the primary outcome, particularly among the elderly with MCI. Therefore, the effect of TRE on cognitive function among the elderly with MCI requires further investigation.

In this study, we conducted an assessor-blinded, pilot randomized controlled trial that implemented 15:9 TRE for 12 weeks among the elderly with MCI. The primary outcome was the change in cognitive function. Additionally, we also recorded adherence and compared changes in body weight, BMI, waist circumference, and calf circumference between the two groups.

Methods

Study design

There is a lack of randomized controlled trials regarding this topic. This research was designed as an assessor-blinded, pilot randomized controlled trial for a more comprehensive RCT in the future, and can provide a reference for subsequent original trials to calculate the sample size [19]. It was conducted at a community elderly service center in Chengdu, Sichuan Province, China, which serves elderly individuals living independently. This study adhered to the Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.

Participants and sample size

Mild Cognitive Impairment (MCI) was diagnosed according to the Petersen criteria [20] by experienced clinical neurologists. The Petersen criteria were established by a multidisciplinary, international expert group as the benchmark enrollment criteria for clinical trials and have become a widely recognized and adopted standard among researchers globally [21–23]. Diagnosis was based on: (1) subjective cognitive complaints reported by the patient and/or an informant, corroborated by clinical judgment; (2) objective evidence of impairment (≥ 1.5 standard deviations below age- and education-adjusted norms) in one or more cognitive domains (memory, language, visuospatial, or executive function), as assessed by the Mini-Mental State Examination (MMSE); (3) preserved or only minimally impaired complex instrumental activities of daily living assessed by the Instrumental Activities of Daily Living (IADL) scale, with maintenance of independence in basic activities and (4) no dementia.

The inclusion criteria for eligible participants in this study were as follows: (1) a diagnosis as MCI stage without dementia; (2) age 60 years or older; (3) body mass index (BMI) of ≥ 18.5 kg/m²; (4) no self-care deficits (Modified Barthel Index score ≥ 60 and number of disability items in the instrumental assessment of daily living skills (IADL) ≤ 2); (5) no depression (the score of the Geriatric Depression Scale < 10) [24]; (6) no risk of malnutrition (the score of the Mini Nutritional Assessment Short Form (MNA-SF) ≥ 12); (7) no risk of sarcopenia (the score of the Simple Five item Scoring Scale for Sarcopenia (SARC-F) + calf circumference < 11) [25, 26]; (8) self-reported being otherwise healthy and (9) voluntary participation with informed consent. The exclusion criteria included: (1) a diagnosis of dementia; (2) participants with other diseases that may affect cognition, such as a history of stroke, diabetes, etc.; and (3) a habit of overeating or reversing day and night. Participants who exhibited overeating during follow-up would be excluded.

The sample size for this study is determined based on an estimate of the expected number of those who agree to participate and the minimum number required to assess the actual purpose of the trial. Therefore, a sample size of 20 per group of 40 cases will be included, which is more than the minimum number recommended for a pilot study [27, 28].

Randomization and masking

Two investigators who received uniform training were responsible for recruitment. A research assistant, who was not involved in the study, generated randomization sequences using an online randomization tool. Block randomization was employed, consisting of a total of 8 blocks: the first 7 blocks each containing 6 participants, and the final block containing 4 participants. This method ensured an equal number of participants in each group and prevented both participants and researchers from identifying the group assignment order (ratio = 1:1). The research assistant placed a labeled card indicating the random group allocation (1 for experimental group, 2 for control group) into a sequentially numbered, sealed opaque envelope. The allocation process and other pertinent information were not disclosed to the investigators.

The outcome assessors responsible for conducting the cognitive assessments were blinded to recruitment, randomization, and group management, and they were instructed not to ask the participants about their dietary patterns during the assessment process. Two independent research assistants with more than five years of experience in behavioral trial monitoring assessed intervention adherence. All outcome assessors and research assistants were prohibited from discussing the intervention program or participant assignments with any relevant individuals and were solely responsible for data collection until the completion of all statistical analyses.

Intervention programs

Throughout the 12-week duration, all participants in the TRE group were permitted to eat ad libitum from 8:00 AM to 5:00 PM, followed by a fasting period from 5:00 PM until 8:00 AM the next day. During the ad libitum window, participants were allowed to eat without calorie restriction, and only drinking water was allowed during the fasting window. The instructions for the TRE group were as follows: “You are allowed to eat whatever you wish between 8:00 AM and 5:00 PM. However, from 5:00 PM until 8:00 AM the following day, you may only drink water. Please try to adhere to this plan every day for the next 12 weeks. A researcher will check in with you daily to record your compliance with this plan.” Participants in the control group continued to eat ad libitum and received the following instructions: “You will maintain the three structured meal times provided by the center as before (breakfast at 7:30 AM, lunch at 11:30 AM, and dinner at 5:30 PM) and keep eating ad libitum without change. A researcher will check in with you daily to record your compliance with this plan.” To maximize the consistency of dietary intake, all participants were provided with three standardized meals by the community senior services center. The main difference lay in the dietary plan: the TRE group consumed all food (including standardized and any self-prepared meals) within a specified time window, while the control group continued their habitual ad libitum eating (including the three provided meals).

Outcomes and measurement

The primary outcomes include two cognitive tests as follows: (a) the Montreal Cognitive Assessment (MoCA) Beijing Version and its subitems (visuospatial & executive, naming, attention, language, abstraction, memory, and orientation). The MoCA total score indicates the global cognitive function which is calculated by summing the scores of the seven subitems. (b) the Auditory Verbal Learning Test Huashan version (AVLT-H) consists of AVLT-H short-term delayed recall score (AVLT-H-SR-S), AVLT-H long-term delayed recall score (AVLT-H-LR-S) and AVLT-H recognition score (AVLT-H-REC-S). Secondary outcomes include body weight, BMI, waist circumference, calf circumference, and adherence. Monitoring body weight indicator is to detect nutritional risks during the intervention period according to the GLIM criteria (5% weight loss over 12 weeks indicates malnutrition and will trigger the dietary adjustment) [29]. All outcomes, except for adherence, were measured consistently at baseline, 6 weeks, and 12 weeks.

Assessment of cognitive function

MoCA can assess different cognitive domains including attention, execution, memory, language, visual structure skills, conceptual thinking, computation, and orientation. It takes about 15 minutes to complete the MoCA test and the maximum score is 30. Participants who got ≤ 25 points were included and those who got ≥ 26 points were considered as having normal cognition and were excluded. To minimize measurement errors, the same version of the scale was consistently used, with score adjustments applied based on educational attainment. Specifically, one point was added to the total score for participants with ≤ 12 years of education [30]. MoCA has a high sensitivity and specificity in identifying MCI (A-level recommendation), with a sensitivity of 92% and a specificity of 85% [31]. AVLT-H is currently the most commonly used scale to check episodic memory, which is divided into three parts including short-delayed recall, long-delayed recall, and recognition. Memory impairment is the core symptom of amnesia MCI, and the sensitivity and specificity of the word learning test for identifying normal elderly and amnesic MCI are 73% and 71%, respectively [32]. The same version of the Chinese Beijing edition of MoCA and the Chinese Huashan Edition of AVLT were used across all assessments.

Assessment of anthropometric parameters

All participants used the same weight scale (model XMTZC04HM), which accurately measures weight to two decimal places. A soft measuring tape (with a minimum scale of 1 mm) was employed to measure and record both waist and calf circumferences. The specific measurement procedures were as follows: (a) body weight: participants were weighed using the provided scale while wearing light clothing and standing barefoot, before breakfast in the morning. (b) waist circumference: this was measured at the midpoint between the lower border of the ribs and the apex of the iliac spine, with the tape positioned horizontally around the waist while the participant stood. (c) calf circumference: this was measured at the thickest part of the calf while the participant was standing. (d) BMI: this was calculated using the formula: BMI (kg/m²) = body weight (kg)/height² (m²).

Adherence record

Two additional research assistants, who were not involved in the study, were tasked with recording daily adherence to the intervention plans for both groups over the 12 weeks. Each morning before the first meal, the research assistants asked and recorded the dietary adherence of participants from the previous day. Complete adherence to the prescribed dietary protocol was recorded as “Adhered”, while any deviation was marked as “Non-adhered”. Furthermore, for participants who were lost to follow-up, their adherence would be recorded as ‘failure to comply with the intervention plan’ from the date of loss until the conclusion of the intervention.

Statistical analysis

We used RStudio and SPSS (version 26.0) for the statistical analysis. Independent t-tests and χ² tests were conducted to compare the descriptive data of the TRE group and control group at baseline. All continuous variables were tested for normality using Shapiro-Wilk tests. For between-group comparisons, independent t-tests were applied to normally distributed data, and Mann-Whitney U tests to non-normally distributed data. Descriptive data were presented as means with standard deviations (SD) or frequencies with percentages (%). Participants who were lost to follow-up during the trial would be assigned their baseline scores for the outcome.

The generalized evaluation equation (GEE) was utilized to evaluate the differences in the cognitive scores and anthropometric parameters including MoCA total score, subitems scores of MoCA, AVLT-H-SR-S, AVLT-H-LR-S, AVLT-H-REC-S, body weight, BMI, waist circumference and calf circumference at baseline (T0), 6-week (T1), and 12-week (T2). We used robust standard errors and the working correlation structure specified as exchangeable. The GEE model of cognitive scores was adjusted for gender, age, education, and baseline scores. The GEE model of anthropometric parameters was adjusted for baseline scores. We used Bonferroni correction (two-sided alpha: 0.05/3 = 0.017) to adjust for multiple comparisons. All analyses adhered to the intention-to-treat (ITT) principle.

Results

Participant characteristics at enrollment

This trial screened 268 individuals and enrolled a total of 46 eligible participants (mean [SD] age, 74.20 [5.09] years; 34 [73.91%] female), as illustrated in Fig. 1. After the recruitment, the mean BMI of all eligible participants was overweight (25.76 ± 3.64 kg/m²) according to the classification criteria of WHO [33]. Participants were randomly allocated to either the TRE group or ad libitum control group (n = 23/group). One participant in the TRE group was lost to follow-up due to an unexpected illness in the second week. The remaining 45 participants included 22 in the TRE group and 23 in the control group, both of whom completed the 12-week follow-up. All participants were included in the ITT analysis. Baseline characteristics were comparably distributed between the TRE group and the control group (Table 1).

Fig. 1.

Participant enrollment according to the CONSORT 2010 flow diagram

Table 1.

Demographic characteristics of the participants at baseline (T0)

	Total	TRE group (n = 23)	CG group (n = 23)	Effect size	P-value
Age (year)	74.20 (5.09)	74.04 (5.09)	74.35 (5.20)	−0.200 a	0.842
MoCA	18.76 (4.21)	18.43 (4.47)	19.09 (4.01)	−0.521a	0.605
Gender†
Female	34 (73.91)	16 (69.60)	18 (78.30)	0.451 b	0.502
Male	12 (26.09)	7 (30.40)	5 (21.70)
Education (year)	9.24 (3.15)	9.35 (3.42)	9.13 (2.93)	0.232 a	0.818
Marital status
Married	27 (58.70)	13 (56.52)	14 (60.87)	1.391 b	0.238
Widowed	19 (41.30)	10 (43.48)	9 (39.13)
BMI (kg/m2)	25.76 (3.64)	25.94 (2.78)	25.59 (4.39)	0.317 a	0.753
Body weight (kg)	62.30 (9.28)	62.20 (7.43)	62.41 (10.99)	−0.076 a	0.940
Height (m)	1.55 (0.07)	1.55 (0.06)	1.56 (0.08)	−0.749 a	0.458
Waist circumference (cm)	92.84 (10.31)	92.63 (10.34)	93.04 (10.52)	−0.134 a	0.894
Calf circumference (cm)	34.51 (2.88)	33.89 (2.79)	35.13 (2.89)	−1.478 a	0.146

Abbreviations: BMI Body mass index, CG Control group, TRE Time-restricted eating

Data are presented as mean (SD) or frequency (%)

^a t value of the independent t-test

^b χ² value of the χ² test

Protocol adherence

The entire intervention duration was 12 weeks (84 days). Regarding adherence of the TRE group, the mean frequency that adhering to TRE was more than five days per week and it increased to more than six days per week from the tenth week onward (Fig. 2). Only 2 participants (8.70%) completed the entire diet program consecutively and 20 participants (86.96%) adhered to TRE for at least five days per week from 7-week to 12-week. Among those who did not complete the entire diet program consecutively, the reasons were either discomfort at the beginning (n = 12) or lack of motivation (n = 8). No adverse effects associated with the TRE diet were observed during the intervention period. The control group remained eating ad libitum without change throughout the 12-week duration.

Fig. 2.

The weekly adherence report of the TRE group

Effects of intervention on cognitive function and episodic memory

MoCA total score increased in the TRE group while it decreased in the control group at both T1 and T2 (Table 2; Fig. 3). GEE showed the effects of 15:9 TRE on global cognitive function from T0 to T1 (β = 1.22, 95% CI = 0.33, 2.10; P = 0.007) and the T1 to T2 (β = 1.70, 95% CI = 0.46, 2.94; P = 0.007), with a significant difference (Table 3). Regarding the subitem scores of MoCA, the Attention score of the TRE group increased compared with the control group at T1 (Table 2; Fig. 3), and it showed a significant improvement for the intervention compared with the control group from T0 to T1 (β = 0.65, 95% CI = 0.15, 1.15; P = 0.011) (Table 3). AVLT-H-REC-S increased in the TRE group while it decreased in the control group at both T1 and T2 (Table 2; Fig. 4). The recognition showed significantly greater improvement for the TRE compared with the control group from T0 to T2, as the AVLT-H-REC-S increased by 2.44 scores on average (95% CI = 0.63, 4.24; P = 0.008) (Table 4).

Table 2.

MoCA and AVLT-H in the baseline (T0), 6-week (T1) and 12-week (T2)

	T0			T1			T2
	TRE	CG	P-value	TRE	CG	P-value	TRE	CG	P-value
MoCA
Visuospatial/Executive b	1.83 (1.56)	1.87 (1.91)	0.839	1.87 (1.52)	1.87 (1.91)	0.726	2.17 (1.64)	1.57 (1.50)	0.216
Naming b	2.87 (0.63)	2.70 (0.82)	0.730	3.04 (0.82)	2.70 (1.02)	0.321	2.78 (0.67)	2.65 (0.83)	0.750
Attention b	4.39 (1.34)	4.52 (1.08)	0.910	4.43 (1.53)	3.91 (1.04)	0.155	4.22 (1.48)	4.09 (1.08)	0.533
Language b	1.22 (0.90)	1.61 (0.78)	0.122	1.17 (0.94)	1.61 (0.78)	0.096	1.43 (0.95)	1.43 (0.84)	0.991
Abstraction b	1.17 (0.78)	1.39 (0.72)	0.328	1.52 (0.73)	1.35 (0.65)	0.376	1.43 (0.84)	1.43 (0.79)	0.898
Memory b	1.78 (1.78)	1.83 (1.27)	0.883	2.13 (1.55)	2.22 (1.20)	0.770	2.39 (1.88)	2.22 (1.57)	0.604
Orientation b	5.09 (1.28)	5.13 (1.25)	0.961	5.04 (1.22)	5.04 (1.30)	0.936	5.13 (1.22)	5.17 (1.27)	0.893
MoCA total score a	18.35 (4.59)	19.04 (4.08)	0.636	19.22 (4.63)	18.70 (4.17)	0.515	19.57 (5.37)	18.57 (4.10)	0.384
AVLT-H
AVLT-H-SR-S b	3.35 (2.66)	4.61 (2.89)	0.141	4.13 (2.56)	4.74 (2.56)	0.451	4.80 (2.87)	4.87 (2.80)	0.833
AVLT-H-LR-S b	2.13 (2.62)	3.52 (3.23)	0.122	3.61 (2.78)	4.09 (2.68)	0.527	4.04 (3.08)	4.22 (2.98)	0.816
AVLT-H-REC-S b	17.43 (3.51)	18.43 (4.34)	0.319	18.09 (3.49)	18.22 (4.27)	0.791	18.48 (3.80)	17.04 (5.25)	0.514

Abbreviations: AVLT-H The auditory verbal learning test-Huashan version, AVLT-H-SR-S AVLT-H short-term delayed recall score, AVLT-H-LR-S AVLT-H long-term delayed recall score, AVLT-H-REC-S AVLT-H recognition score, BMI Body mass index, CG Control group, MoCA Montreal cognitive assessment, TRE Time-restricted eating.

Data are presented as mean (SD).

^a The independent t-test was used.

^b The Mann-Whitney U test was used.

Fig. 3.

Change in MoCA and subitems in baseline, 6-week and 12-week. A Changes of MoCA total score over 12 weeks in two groups. B Changes of Visuospatial & Executive over 12 weeks in two groups. C Changes of Naming over 12 weeks in two groups. D Changes of Attention over 12 weeks in two groups. E Changes of Language over 12 weeks in two groups. F Changes of Abstraction over 12 weeks in two groups. G Changes of Memory over 12 weeks in two groups. H Changes of Orientation over 12 weeks in two groups

Table 3.

GEE analysis of differences between baseline (T0), 6-week (T1) and 12-week (T2) in MoCA and its subitems

Variables	β (95% CI)	S.E.	Wald’s χ2	P-value
MoCA total score (possible range 0–30)
Group (TRE)†	−0.70 (−3.15, 1.76)	1.25	0.31	0.579
T1‡	−0.35 (−1.04, 0.34)	0.35	0.98	0.322
T2‡	−0.48 (−1.19, 0.23)	0.36	1.74	0.188
Group (TRE) × T1⸹	1.22 (0.33, 2.10)	0.45	7.27	0.007
Group (TRE) × T2⸹	1.70 (0.46, 2.94)	0.63	7.19	0.007
Visuospatial & Executive (possible range 0–5)
Group (TRE)†	0.19 (−0.69, 1.07)	0.45	0.18	0.676
T1‡	0.22 (−0.02, 0.06)	0.02	1.02	0.312
T2‡	0.22 (−0.37, 0.41)	0.20	0.01	0.913
Group (TRE) × T1⸹	004 (−0.04, 0.13)	0.04	1.05	0.307
Group (TRE) × T2⸹	0.65 (−0.10, 1.41)	0.38	2.88	0.090
Naming (possible range 0–3)
Group (TRE)†	0.217 (−0.18, 0.61)	0.20	1.16	0.281
T1‡	0.09 (−0.12, 0.29)	0.11	0.68	0.411
T2‡	−0.07 (−0.22, 0.09)	0.08	0.70	0.402
Group (TRE) × T1⸹	0.17 (−0.24, 0.59)	0.21	0.69	0.407
Group (exp) × T2⸹	−0.04 (−0.35, 0.26)	0.16	0.08	0.780
Attention (possible range 0–6)
Group (TRE)†	−0.13 (−0.82, 0.56)	0.35	0.14	0.710
T1‡	−0.61 (−1.02, −1.20)	0.21	8.35	0.004
T2‡	−0.44 (−0.85, −0.02)	0.21	4.23	0.040
Group (TRE) × T1⸹	0.65 (0.15, 1.15)	0.26	6.53	0.011
Group (TRE) × T2⸹	0.26 (−0.34, 0.86)	0.30	0.74	0.391
Language (possible range 0–3)
Group (TRE)†	−0.28 (−0.72, 0.17)	0.22	1.50	0.221
T1‡	−0.22 (−0.06, 0.02)	0.02	1.02	0.312
T2‡	0.22 (−0.22, 0.26)	0.12	0.03	0.857
Group (TRE) × T1⸹	−0.04 (−0.13, 0.04)	0.04	1.05	0.307
Group (TRE) × T2⸹	0.39 (−0.07, 0.85)	0.24	2.77	0.096
Abstraction (possible range 0–2)
Group (TRE)†	−0.01 (−0.39, 0.36)	0.19	0.01	0.940
T1‡	0.15 (−0.02, 0.32)	0.09	3.08	0.080
T2‡	0.15 (−0.04, 0.34)	0.10	2.46	0.117
Group (TRE) × T1⸹	0.39 (0.07, 0.71)	0.16	5.72	0.017
Group (TRE) × T2⸹	0.22 (−0.16, 0.59)	0.19	1.29	0.256
Memory (possible range 0–5)
Group†	0.01 (−0.80, 0.83)	0.42	0.01	0.972
T1‡	0.37 (0.13, 0.61)	0.12	9.41	0.002
T2‡	0.50 (0.13, 0.87)	0.19	7.01	0.008
Group (TRE) × T1⸹	−0.04 (−0.52, 0.43)	0.24	0.03	0.857
Group (TRE) × T2⸹	0.22 (−0.52, 0.96)	0.38	0.33	0.564
Orientation (possible range 0–6)
Group (TRE)†	−0.03 (−0.71, 0.65)	0.34	0.01	0.933
T1‡	−0.07 (−0.27, 0.14)	0.10	0.40	0.530
T2‡	0.04 (0.15, 0.23)	0.10	0.20	0.654
Group (TRE) × T1⸹	0.04 (−0.36, 0.45)	0.21	0.04	0.834
Group (TRE) × T2⸹	−1.81 (−0.38, 0.38)	0.19	0.01	0.923

Abbreviations: MoCA Montreal cognitive assessment, TRE Time-restricted eating

The covariates are the baseline score, gender, age and education

^† Reference is control group

^‡ Reference is T0 (baseline)

^⸹ Reference is group (CG) × T0 (baseline)

Fig. 4.

Changes in AVLT-H in baseline, 6-week and 12-week. A Changes of AVLT-H-SR-S over 12 weeks in two groups. B Changes of AVLT-H-LR-S over 12 weeks in two groups. C Changes of AVLT-H-REC-S over 12 weeks in two groups

Table 4.

GEE analysis of differences between baseline (T0), 6-week (T1) and 12-week (T2) in AVLT-H

Variables	β (95% CI)	S.E.	Wald’s χ2	P-value
Short-Delayed Recall
AVLT-H-SR-S (possible range 0–12)
Group (TRE)†	−0.68 (−2.17, 0.81)	0.76	0.81	0.369
T1‡	0.46 (0.13, 0.78)	0.16	7.68	0.006
T2‡	0.80 (0.21, 1.30)	0.25	10.27	0.001
Group (TRE) × T1⸹	0.65 (0.04, 1.27)	0.32	4.28	0.038
Group (TRE) × T2⸹	1.09 (0.16, 2.02)	0.48	5.22	0.022
Long-Delayed Recall
AVLT-H-LR-S (possible range 0–12)
Group (TRE)†	−0.68 (−2.23, 0.86)	0.79	0.75	0.388
T1‡	1.02 (0.54, 1.50)	0.25	17.40	< 0.001
T2‡	1.30 (0.64, 1.97)	0.34	14.77	< 0.001
Group (TRE) × T1⸹	0.91 (−0.01, 1.84)	0.47	3.76	0.053
Group (TRE) × T2⸹	1.22 (−0.07, 2.50)	0.65	3.46	0.063
Recognition
AVLT-H-REC-S (possible range 0–24)
Group (TRE)†	−1.00 (−3.23, 1.23)	1.14	0.77	0.380
T1‡	−0.22 (−0.78, 0.35)	0.29	0.57	0.451
T2‡	−1.39 (−2.88, 0.10)	0.76	3.35	0.067
Group (TRE) × T1⸹	0.87 (0.08, 1.66)	0.40	4.70	0.030
Group (TRE) × T2⸹	2.44 (0.63, 4.24)	0.92	7.00	0.008

Abbreviations: AVLT-H-SR-S AVLT-H short-term delayed recall score, AVLT-H-LR-S AVLT-H long-term delayed recall score, AVLT-H-REC-S AVLT-H recognition score, TRE Time-restricted eating

The covariates are the baseline score, gender, age and education

^† Reference is the control group

^‡ Reference is T0 (baseline)

^⸹ Reference is group (CG) × T0 (baseline)

Effects of intervention on anthropometric parameters

The mean percentage change of body weight in the TRE group did not reach 2% from T0 to T2 (Fig. 5). Body weight, BMI and calf circumference decreased in the control group from T0 to T2 (Table 5; Fig. 5). In addition, waist circumference decreased in the TRE group while it increased in the control group from T0 to T2 (Table 5; Fig. 5). The GEE showed no significant differences in body weight, BMI, waist circumference, or calf circumference (P > 0.017) (Table 6).

Fig. 5.

Mean percentage changes in anthropometric parameters. A Mean percentage changes in BW, BMI, WC and CC from baseline to 6 weeks. B Mean percentage changes in BW, BMI, WC and CC from baseline to 12 weeks

Table 5.

Anthropometric parameters in the baseline (T0), 6-week (T1) and 12-week (T2)

	T0			T1			T2
	TRE	CG	P-value	TRE	CG	P-value	TRE	CG	P-value
Body weight	62.20 (7.43)	62.41 (10.99)	0.940	60.79 (7.96)	62.07 (10.97)	0.930	61.17 (6.73)	61.90 (11.01)	0.475
BMI	25.94 (2.78)	25.59 (4.39)	0.753	25.37 (3.21)	25.46 (4.37)	0.701	25.52 (2.62)	25.39 (4.42)	0.307
Waist circumference	92.63 (10.34)	93.04 (10.52)	0.894	92.70 (8.49)	93.19 (10.54)	0.861	90.93 (8.29)	93.39 (10.67)	0.388
Calf circumference	33.89 (2.79)	35.13 (2.89)	0.146	33.97 (3.00)	35.07 (2.92)	0.216	33.79 (2.86)	34.87 (2.55)	0.182

Abbreviations: BMI Body mass index, CG Control group, TRE Time-restricted eating

Data are presented as mean (SD). The independent t-test was used

Table 6.

GEE analysis of differences between baseline (T0), 6-week (T1) and 12-week (T2) in anthropometric parameters

Variables	β (95% CI)	S.E.	Wald’s χ2	P-value
Body weight
Group (TRE)†	−0.74 (−5.99, 4.51)	2.68	0.08	0.783
T1‡	−0.87 (−1.61, −0.13)	0.38	5.31	0.021
T2‡	−0.77 (−1.28, −0.25)	0.26	8.47	0.004
Group (TRE) × T1⸹	−1.07 (−2.52, 0.37)	0.74	2.12	0.145
Group (TRE) × T2⸹	−0.51 (−1.53, 0.51)	0.52	0.96	0.326
BMI
Group (TRE)†	0.13 (−1.95, 2,21)	1.06	0.02	0.901
T1‡	−0.35 (−0.66, −0.04)	0.16	4.91	0.027
T2‡	−0.31 (−0.52, −0.10)	0.11	8.03	0.005
Group (TRE) × T1⸹	−0.43 (−1.03, 0.18)	0.31	1.91	0.167
Group (TRE) × T2⸹	−0.21 (−0.63, 0.22)	0.21	0.92	0.337
Waist circumference
Group (TRE)†	−0.41 (−6.31, 5.48)	3.01	0.02	0.891
T1‡	0.15 (−0.48, 0.78)	0.32	0.21	0.644
T2‡	0.35 (−0.62, 1.32)	0.50	0.49	0.482
Group (TRE) × T1⸹	−0.08 (−1.73, 1.56)	0.84	0.01	0.921
Group (TRE) × T2⸹	−2.04 (−3.90, −0.19)	0.95	4.67	0.031
Calf circumference
Group (TRE)†	−1.14 (−2.69, 0.42)	0.79	2.06	0.151
T1‡	0.01 (−0.20, 0.22)	0.11	0.01	0.952
T2‡	−0.18 (−0.56, 0.19)	0.19	0.92	0.339
Group (TRE) × T1⸹	0.14 (−0.28, 0.56)	0.21	0.45	0.503
Group (TRE) × T2⸹	0.16 (−0.59, 0.90)	0.38	0.17	0.681

Abbreviations: BMI Body mass index, TRE Time-restricted eating

The covariates are the baseline values

^†Reference is the control group

^‡Reference is T0 (baseline)

^⸹Reference is group (CG) × T0 (baseline)

Discussion

This study examined the efficacy of 15:9 TRE for 12 weeks among MCI participants and found improvement in the MoCA total score and the AVLT-H-REC-S after the intervention. No adverse events were reported during the intervention. However, there were no significant improvements in body weight, BMI, waist circumference, and calf circumference.

Adherence in the TRE group

We assessed adherence in the TRE group by aggregating the number of days participants followed the TRE diet program, whether continuously or intermittently. Twenty participants (86.96%) adhered to the TRE regimen for at least five days per week during the later follow-up phase (from 7-week to 12-week), which aligns with findings from a previous study [34]. No participant engaged in overeating as a result of the dietary change, suggesting that adapting to eating within a specified time window became easier as the study progressed [35]. The adherence trend observed in this study suggests that compliance is associated with participants’ acceptance and adaptation to the TRE. It is reported that the mean adherence rate of early TRE (eating time between 6:00 am and 4:00 pm) was 81.4% and maintaining daily adherence posed a challenge for participants [36]. Long-term adherence to TRE may prove more difficult for participants, particularly those experiencing it for the first time, as the sustained sensation of hunger can be a significant obstacle. To enhance adherence, interventions could be customized to align with individual preferences and circumstances, or the frequency of adherence could be adjusted to intermittent or alternate days [37].

In addition, the participants in this study were elderly people with MCI and overweight but no sarcopenia. Participants with higher compliance might be more likely to change their previous dietary behaviors through the TRE intervention, then establish a dietary schedule more aligned with circadian rhythms and develop dietary health awareness during this period. An observational study reported the lowest mortality rate in adults with normal BMI who engaged in a higher level of physical activity along with a healthy diet, suggesting that a healthy diet and lifestyle might be the best choice for promoting longevity [38]. For the MCI population who are overweight, future studies could further combine TRE with other physical exercises to explore more deeply whether this combination can improve quality of life and promote longevity, which is of great significance for achieving healthy aging.

The effects of 15:9 TRE on cognitive function among MCI

Our results indicated that 12-week 15:9 TRE might be associated with the improvement of the MoCA total score from baseline to the end of the intervention compared to the control group (β = 1.70, 95% CI 0.46 to 2.94, p = 0.007). In contrast to the findings of Anton et al., the total score of MoCA for the 4-week TRE group was higher than the baseline, however, this difference did not reach statistical significance [18]. Similarly, 64 older adults with normal weight implemented TRE for 6 weeks, only slight improvements in walking activity and glucose tolerance were observed, with no significant changes in cognitive function [17]. Notably, the baseline cognitive status of participants in both above studies was unclear, and specific subitem scores from the cognitive assessments were not reported. Furthermore, the mean BMI of the participants was overweight in this study and they might already have unhealthy dietary habits. Participants of the TRE group might be more likely to show active responses to the intervention that they first experienced, driven by their health awareness, which might affect the assessment results of cognitive function [39].

There was no significant improvement among all subitems in the TRE group compared to the control group from baseline to 12 weeks. However, the attention score (β = 0.65, 95% CI −0.34 to 0.86, p = 0.011) of the TRE group was improved from baseline to 6-week compared to the control group. Few studies have reported the impact of TRE on different cognitive domains currently. A systematic review reported that the Mediterranean-style diet could improve executive function and visual constructs [40], while similar effects were not observed in this trial, which might be associated with the heterogeneity of dietary patterns, intervention duration, and measurement instruments. While existing evidence indicates that cognitive improvement is a potential benefit of long-duration intermittent fasting and TRE [10, 16], it still needs further study regarding the mechanisms and pathways affecting important cognitive functional domains.

The effects of 15:9 TRE on episodic memory among MCI

AVLT-H-REC-S (β = 2.44, 95% CI 0.63 to 4.24, p = 0.008) was significantly improved compared with the control group after 12 weeks of TRE. An original trial showed significant improvement in episodic memory among healthy older adults after a year of the AU-AGE diet [41]. However, its measurement instrument and dietary intervention were different from this trial. Some researchers proposed that TRE can regulate circadian rhythms, thereby triggering autophagy mechanisms that lead to cognitive function improvements [42, 43]. On a molecular level, TRE may exert neuroprotective effects, enhancing neurogenesis and synaptic plasticity, which could delay cognitive decline and mitigate neurodegenerative diseases by upregulating neurotrophic factors and protein chaperones, while decreasing levels of pro-inflammatory cytokines [44]. However, there is a paucity of original research specifically investigating the effect of TRE on memory function among individuals with MCI. Some trials have explored other short-term fasting diets, such as a very low-carbohydrate diet or continuous calorie restriction, with results indicating that lower calorie intake could benefit memory function in the short term [11, 12, 45]. Nevertheless, the effective negative energy balance required to improve memory function in the MCI population remains to be determined. Furthermore, regarding TRE, it is still unclear whether the neuroprotective effect is achieved through the regulation of circadian rhythms or an unconscious reduction in total calorie intake during the fasting window.

Notably, the results of this pilot study could not establish a causal relationship between 12-week TRE and the improvement in cognitive scores in older adults with mild cognitive impairment. Future research should consider the following two critical points. Firstly, the physiological mechanisms by which TRE affects cognitive function in older adults with mild cognitive impairment remain undetermined and require further in-depth exploration. Some possible underlying pathways should be focused on, such as modulating circadian rhythms, reducing systemic inflammation, and improving brain-derived neurotrophic factor (BDNF), which may shift the results from association to causality [46, 47]. Secondly, whether the way TRE affects cognitive function differs from that of simple caloric restriction remains unknown and can be considered in the study design.

The effects of 15:9 TRE on anthropometric parameters among MCI

Body weight and BMI

In this study, the body weight and BMI of the TRE group decreased more over time compared to the control group, however, no significant difference was observed between the two groups. In line with the findings of Lowe’s study, TRE was not more effective for weight loss than ad libitum eating, which contrasts with previous evidence [48–50]. The initiation and conclusion times of the fasting window, the duration of the fasting period, and the degree of calorie restriction might be critical factors contributing to the discrepancies among trials. It seemed that short-term TRE did not trigger overeating, indicated by the changes in weight loss observed during the intervention. As the trial progressed, the body might gradually adapt to the TRE diet and the mean body weight even slightly increased from 6-week to 12-week. Although no participant reached the ≥ 5% malnutrition threshold which supports the safety of TRE in MCI over 12 weeks, the longer-term monitoring is warranted [29].

On the one hand, whether implementing TRE intervention in the elderly population with MCI who are overweight improves cognitive function through weight loss (even if no significant effect of TRE on body weight was found) requires further physiological mechanism studies regarding the association between body composition and cognitive function. On the other hand, it remains unknown whether being overweight is a protective factor for maintaining cognitive function, since the relationship between obesity and cognitive function in older adults is complex. Some studies indicate that individuals with higher BMI experience lower executive function and poorer cognitive outcomes [51–53], while others argue that obesity may enhance cognitive performance and mitigate the risk of cognitive decline [54, 55]. In conclusion, while weight management is crucial for adults living with obesity [56], there remains insufficient evidence to evaluate the risk-benefit ratio of implementing TRE among the elderly with high BMI. Beyond weight loss, recent research has demonstrated that TRE could optimize metabolic indicators and reduce the risk of cardiovascular and endocrine diseases in adults who are living with overweight or obesity [57–59]. Nevertheless, controversy persists, as some scholars have proposed the ‘obesity paradox’, suggesting that a higher BMI may lower the risk of death from cardiovascular disease [60], while others contest this view [61–63].

Waist circumference and calf circumference

Waist circumference in the TRE group decreased over time during the intervention duration, which might be attributed to a reduction in body weight. Consistent with recent evidence, no significant difference in waist circumference was observed after TRE [64]. As an important anthropometric parameter, waist circumference indirectly reflects the accumulation of abdominal fat. However, the association between waist circumference and cognitive function still needs further study. An original study reported that a larger waist circumference was associated with a better MMSE (Mini-Mental State Examination) score, suggesting that central obesity might have a protective effect on cognitive function in some elderly groups [65]. On the contrary, a longitudinal study found a negative association that the risk of cognitive impairment rises by 63% (OR = 1.63) for every one standard deviation increase in waist circumference, which might be achieved through pathological pathways such as diabetes and inflammation [66]. Although no significant difference was observed in waist circumference after TRE treatment in this trial, the association between changes in waist circumference and improvements in cognitive function should go beyond simply comparing group differences and warrants further investigation. Specifically, the dynamic relationship between abdominal fat and cognitive function warrants further monitoring to clarify the pathways by which TRE affects cognitive function in future studies.

No significant change in the calf circumference was observed in the TRE group during the trial. This finding aligns with an original study that indicated that short-term TRE did not reduce lean body mass [49]. Muscle mass is crucial for the mobility and independence of elderly individuals who are at high risk of frailty and sarcopenia [67]. As a simple and easily measurable parameter, calf circumference can indirectly reflect calf muscle mass [68]. Mechanism research indicates that reduction of calf circumference is associated with an increased risk of cognitive impairment (OR = 1.14–2.21) [69, 70]. This study found that short-term TRE might not lead to a significant reduction in calf circumference, which would provide evidence for further investigation about the suitability and feasibility of TRE for the elderly. However, additional evidence is required to evaluate and integrate the long-term effects of TRE on lean body mass.

Strengths and limitations

Our study has several important strengths. In contrast to previous research, we specifically controlled for cognitive function at baseline and reported the subitem scores from cognitive tests. The adherence to the 15:9 TRE, as previously discussed, may provide new insights into the feasibility of implementing TRE among elderly individuals with MCI. Additionally, we analyzed the effects of TRE on anthropometric parameters, which may provide more evidence of its effects on body composition in the elderly for future investigations.

This trial has several limitations. Firstly, although this study initially aimed to recruit MCI participants with a BMI ≥ 18.5 kg/m² as many as possible to facilitate subsequent subgroup analysis, the BMI of the majority of eligible participants recruited was overweight. These individuals were retained due to the limited number of eligible participants available. Therefore, the findings of this study may only present the overweight subgroup of MCI participants and cannot be generalizable to the broader elderly population with other BMI ranges. Secondly, the relatively small sample size may limit statistical power and increase the risk of Type I and Type II errors. However, this is an inherent and acceptable characteristic of a pilot trial, as its primary purpose is not to test efficacy but rather to assess feasibility and to estimate effect size and variability for planning future definitive randomized controlled trials. In this context, our sample size exceeded the conventional recommendations for pilot studies [27, 28]. This study may provide preliminary data and practical guidance for future research. In addition, considering the statistical power, the sample size of this study is insufficient to support the subgroup analysis. Future research with a larger sample size can try to conduct subgroup analysis by BMI, age, gender, and education. Thirdly, the absence of participant blinding due to the specificity of the behavioral intervention may inevitably introduce the risk of performance bias. The TRE group might be influenced by expectations of improved health status (placebo effect) or awareness of being observed (Hawthorne effect), which might contribute to improved cognitive testing scores. Although the outcome assessors were blinded, we could not completely rule out this possibility. Fourthly, the same version of the MoCA and AVLT-H was used at all time points (baseline, 6 weeks, 12 weeks), which might result in retest effects and manifest as an improvement of cognitive assessment [71, 72]. However, the main finding of this study was based on the comparison of the difference in changes between the TRE group and the control group, rather than a simple intra-group time point comparison. The retest effect would theoretically affect all participants of the two groups equally. If the observed cognitive improvement is mainly driven by the retest effect, there should be no significant difference in the extent of improvement between the two groups, or only a very small difference [73]. However, a slight decline was observed in cognitive scores of the control group, while the intervention group exhibited a slight increase after the 12-week TRE. This might indicate that the retest effect was not obvious in the cognitive assessment of the two groups. The more significant improvement in the TRE group compared to the control group is more likely attributed to the effect of the intervention itself. Fifthly, the outcome indicators adopted in this study are limited and future research should integrate multi-dimensional indicators, such as fat mass, lean body mass and biomarker indicators (e.g., glucose/insulin metabolism, inflammatory markers, BDNF, or gut hormones) to more accurately reveal the biological mechanisms of the association between overweight and cognitive function. Sixthly, the lack of formal monitoring of daily total caloric and nutrient intake is an important limitation. Although we made every effort to ensure consistent food intake between both groups, TRE might have inadvertently led to a reduction in total energy intake in the TRE group. Given that previous studies have shown caloric restriction itself can have beneficial effects on cognitive function [10–13], we cannot determine whether the observed cognitive improvements are solely attributable to the timing of food intake, an energy deficit, or synergistic effects. Future studies should incorporate detailed dietary records to differentiate between the effects of the timing of food intake and caloric restriction, and to elucidate their underlying mechanisms. Seventhly, the results of this study were only based on short-term changes and maintaining long-term effects remains a challenge. Furthermore, the effects of TRE on cognitive function over an extended period still require further investigation.

Conclusion

This original pilot trial found short-term improvements in MoCA score and AVLT-H-REC-S among elderly individuals with MCI and no significant change in anthropometric parameters after 12-week 15:9 TRE. However, these findings should be interpreted with caution due to the limited sample size and generalizability. Larger-sample investigations are needed to further clarify the mechanisms by which TRE affects the cognitive function of elderly patients with MCI, including the potential mediating role of changes in circadian rhythm, body composition, or biomarkers.

Abbreviations

AVLT-H

The auditory verbal learning test-Huashan version

AVLT-H-SR-S

AVLT-H short-term delayed recall score

AVLT-H-LR-S

AVLT-H long-term delayed recall score

AVLT-H-REC-S

AVLT-H recognition score

BW

Body weight

BMI

Body mass index

CG

Control group

CC

Calf circumference

GEE

General estimating equation

IF

Intermittent fasting

ITT

Intention to treat

MoCA

Montreal cognitive assessment

TRE

Time-restricted eating

WC

Waist circumference

Authors’ contributions

The authors’ responsibilities were as follows: JW and FW wrote the paper. MHS and HXC performed the statistical analysis and interpretation of the data. JW, QQL, LL, XQX, CXH and LSX conducted the research. MHS, LL and XQX performed the formal analysis. XYH designed the research and supervision. All authors have read and approved the final manuscript.

Funding

This work was supported by grants from the West China Nursing Discipline Development Special Fund Project, Sichuan University (Grant HXHL20015), and the research project on the nursing discipline of the Journal of Chinese Medical Association (CMAPH-NRP 2022007).

Data availability

Please contact the corresponding authors to discuss the availability of the datasets used and/or analyzed during the current study.

Declarations

Ethics approval and consent to participate

All participants signed an informed consent form before participation. The study was approved by the Ethics Committee on Biomedical Research, West China Hospital of Sichuan University (Approval number: 2021 − 1020) and registered in the China Clinical Trial Registration Center (Registration number: ChiCTR2100052766, Registration date: 05/11/2021). The trial is in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.