Inclusion criteria were the following: (1) peer-reviewed publications, (2) available in English, (3) cross-sectional and cohort/observational studies, (4) older adults ages 60 and older with or without cognitive impairment, (5) measurement of sedentary behavior at baseline, (6) objective cognition outcomes acquired by a validated assessment measure, (7) brain health outcomes measured via structural (volume, thickness, surface area, diffusion tensor imaging) and/or functional (fMRI, functional connectivity, cerebral blood flow) neuroimaging.

Exclusion criteria were the following: (1) study not available in English, (2) Intervention studies unless sedentary behavior and cognition/brain health are reported cross-sectionally at baseline, (3) studies that do not examine the associations of sedentary behavior with a cognition or brain health outcome, (4) participants younger than 60 years old, (5) participants with reported psychopathology or neurological disorders (e.g., depression, Parkinson’s Disease) other than Alzheimer’s disease.

A search was conducted using EBSCOhost (MEDLINE, Academic Search Premier), Ovid (PsycINFO), ProQuest (PSYCArticles), PubMed, and Sedentary Behavior Research Database (SBRD) databases. The initial search began on December 8, 2023, and therefore all articles published prior to this date were eligible to be included in the search. Following this, reference lists from pertinent studies, reviews, and meta-analyses were manually searched by study authors (MAG, JW) for studies that may have not been captured in the original search. This search strategy was developed and pre-registered prior to beginning the search by MAG, JW, and SG.

An identical search strategy was applied to each database and included the following: (“older adults” OR “geriatrics” OR “aging” OR “seniors” OR “elderly” OR “healthy aging” OR “MCI” OR “dementia” OR “Alzheimer’s disease) AND (“sedentary behavior” OR “inactivity” OR “sitting” OR “low activity”) AND (“neuroimaging” OR “brain volume” OR “brain change” OR “MRI” OR “white matter” OR “connectivity” OR “PET” OR “cerebral blood flow” OR “cortical” OR “cognition” OR “memory” OR “thinking”). These search terms were developed by MAG, SG, and JW.

Eligibility was assessed using the previously discussed criteria (see Eligibility Criteria). Following the initial search, study titles and abstracts were reviewed for preliminary determination of eligibility (i.e., sedentary behavior studies in older adults) by MAG. Studies that met the initial eligibility were then reviewed in detail to make a final decision on eligibility (MAG, JW, AA, CC). The corresponding authors were contacted when the data presented was insufficient to be able to determine final eligibility. When discrepancies assessment arose, MAG and JW discussed the studies until a consensus was reached.

A data extraction form was developed using Microsoft Excel and was consistent with the Cochrane Consumers and Communication Data Extraction Template (Ryan and Hill, 2019). Several authors (AA, CC, AV, JW, MAG) extracted the data, and discussions regarding data collection procedures (i.e., inclusion of demographic information, classification of outcomes) were routinely conducted via bi-weekly group meetings. All extracted data was verified by a second rater. The list of variables to be coded included the following: age, sex, education level, cognitive status (e.g., normal cognition, mild cognitive impairment, dementia), type of sedentary behavior (e.g., self-report, objectively measured, etc.), cognitive outcome (e.g., global cognition, memory, executive functions, processing speed, etc.), brain health outcome (e.g., volume, thickness, white matter integrity, functional connectivity, etc.), and bias in individual studies. For cohort studies where baseline sedentary behavior was related to cognition or brain health outcomes over time, follow-up time was extracted.

In order to assess risk of bias within individual studies, the NIH Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies was used¹, which uses 14 criteria to evaluate the methodological quality of cohort longitudinal and cross-sectional studies. Only one criterion: “Were the outcome assessors blinded to the exposure status of participants?” was not evaluated, as it was not applicable for observational studies with no status manipulation.

The initial search results from MEDLINE, PsycInfo, PsycArticles, Academic Search Premier, PubMed, and Sedentary Behavior Research Database returned 3,062 records. After removal of duplicates, 2,019 unique records remained. These 2,019 unique entries were initially screened using title and abstract review, and 1,688 records were excluded because based on this screen. Following this, 331 records were assessed for full eligibility using the criteria listed above (see Eligibility Criteria), and 298 records were excluded. No new records were included from the manual reference search. Final inclusion was 33 studies. This information is presented in a flowchart (see Figure 1) based on the PRISMA template (Page et al., 2021).

Study characteristics for cognition (Table 1) and brain health (Table 2) were presented separately. Detailed study information including sample size, sample demographics (e.g., sex, age, education), cognitive status, and sedentary behavior are presented in Table 1 (Iso-Markku et al., 2018; Wu et al., 2020; Vance et al., 2005; Heisz et al., 2015; Steinberg et al., 2015; Rosenberg et al., 2016; Edwards and Loprinzi, 2017a; Edwards and Loprinzi, 2017b; Ku et al., 2017; Vásquez et al., 2017; Čukić et al., 2018; Wanigatunga et al., 2018; García-Hermoso et al., 2018; Folley et al., 2019; Zlatar et al., 2019; Amagasa et al., 2020; Burzynska et al., 2020; Suzuki et al., 2020; Maasakkers et al., 2020; Kurita et al., 2022; Chen et al., 2022; Gerten et al., 2022; Silva-Fernandes et al., 2022; Zhou et al., 2022; Shuai et al., 2023; Major et al., 2023; Han et al., 2023) and Table 2 (Arnardottir et al., 2016; Bronas et al., 2019; Zlatar et al., 2019; Burzynska et al., 2014; Engeroff et al., 2018; Dion et al., 2021; Machida et al., 2022). Two included studies had overlapping participants (Edwards and Loprinzi, 2017a; Edwards and Loprinzi, 2017b).

The 33 included studies resulted in a total of 43,577 participants (M = 1,321, SD = 2,375, range = 18–10,450). The sample displayed some variability in age (M = 73, SD = 5, range = 65–88) and was, on average, gender balanced (% female; M = 56%, SD = 9%, range 21–72%). Of the 28 studies that reported education level, 57% reported a majority completing high school or higher, while 43% reported a majority of the sample completing less than high school. Only 23 studies explicitly reported on the cognitive status of the sample. Most studies utilized cognitively healthy samples (n = 14), and nine studies included participants with mild cognitive impairment (MCI) and/or dementia.

Regarding measurement of sedentary behavior, a majority of studies (n = 20) used purely objective measures, followed by subjective measures (n = 10), and then combination of both objective and subjective methods (n = 3). Of the studies that used an objective measure of sedentary behavior and reported the device location, the majority of studies used a hip-worn device placement (n = 16), with one study using a wrist-worn device and one study using a thigh-worn device (n = 5 did not put wear location but hip is suspected based on device type). The most common devices utilized were versions of the Actigraph (GT3X, GT3X+, GTM1; n = 14), and the majority of studies focused on total sedentary time using <100 counts per minute (n = 13). In studies that utilized a subjective measure of sedentary behavior (n = 13), all studies included participant self-report of sedentary behavior. Of these 13 studies that utilized some sort of self-report, some (n = 6) used a validated measure with several questions (e.g., Sedentary Behavior Questionnaire (Rosenberg et al., 2010)), while the rest utilized 1–3 individual questions (n = 7).

Among studies that reported a cognitive outcome (n = 27), the most commonly used assessment was a single-domain cognitive measure or composite scores (e.g., executive function composite; n = 11), followed by a cognitive screening tool (e.g., Mini Mental Status Exam; [MMSE]) (n = 10), and finally, fewer studies included more comprehensive neuropsychological batteries (n = 6). A majority of the studies with cognitive outcomes (n = 15/27) reported a negative association between sedentary behavior and cognitive performance, indicating that greater sedentary time was associated with worse cognitive performance. Five studies reported positive associations between sedentary behavior and cognitive performance, and seven studies reported no or mixed associations between sedentary behavior and cognitive performance.

Of the studies that reported brain structure and function outcomes (n = 7), the most common outcome measured was white matter hyperintensities (WMH; n = 3), followed by hippocampal volume (n = 2), and functional neuroimaging outcomes (n = 2). All brain imaging studies reported scanner strength: most studies utilized a 3 T scanner (n = 5) while two studies utilized a 1.5 T scanner. Five of the brain health studies reported negative associations between sedentary behavior and measures of brain health, while two studies reported no association between sedentary behavior and hippocampal volume.

Most studies were cross-sectional (n = 27/33) with only 1/6 longitudinal studies belonging to the brain health category. A majority of the longitudinal studies (n = 5/6) reported negative associations between sedentary behavior and cognitive/brain health outcomes. The mean follow-up time for longitudinal studies was 3.6 years. One study (Zlatar et al., 2019) contained both cognitive and brain health outcomes, and those results are presented separately in both tables. Two studies with cognitive outcomes utilized participants from the same sample (Edwards and Loprinzi, 2017a; Edwards and Loprinzi, 2017b).

The risk of bias (ROB) assessment revealed consistent, reasonable methodological standards for the majority of studies, with 73% of the studies rated overall as “good” and 27% of the studies scoring “fair” taking into account their methodological rigor. Nearly all studies included clear statement of the research objective (100%), defined the study population (88%), recruited subjects from similar populations (88%), defined independent (97%) and dependent variables (100%), examined different levels of exposure of the independent variable (100%), and adequately addressed confounding variables (100%). Across all studies, lower endorsed ROB categories included providing sample size justification (21%) and obtaining participation rate above 50% of the total eligible sample (42%). In addition, due to most of the studies being cross-sectional, time-dependent categories such as assessing exposure more than once (15%) or providing sufficient time frame to observe associations (18%) were less frequent. ROB metrics are depicted in detail in Figure 2.

As mentioned above, most of the studies in this systematic review contained cognitive outcomes (vs brain health outcomes). The majority of studies indicated that there was a negative association between sedentary behavior and cognition (Iso-Markku et al., 2018; Heisz et al., 2015; Steinberg et al., 2015; Edwards and Loprinzi, 2017a; Edwards and Loprinzi, 2017b; Ku et al., 2017; Wanigatunga et al., 2018; García- Hermoso et al., 2018; Folley et al., 2019; Suzuki et al., 2020; Chen et al., 2022; Gerten et al., 2022; Zhou et al., 2022; Shuai et al., 2023; Han et al., 2023), while the rest of the studies reported mixed or no effect (Wu et al., 2020; Vásquez et al., 2017; Čukić et al., 2018; Zlatar et al., 2019; Amagasa et al., 2020; Maasakkers et al., 2020; Silva-Fernandes et al., 2022), or positive effects (Vance et al., 2005; Rosenberg et al., 2016; Burzynska et al., 2020; Kurita et al., 2022; Major et al., 2023). A majority of the studies assessed cognition using a cognitive screening tool or a more limited cognitive battery. Of the five studies reporting positive associations between sedentary behavior and cognition, two studies utilized self-report of sedentary behavior, two utilized self-report and objective measurement, and only one used objective measurement. Five studies were longitudinal (Ku et al., 2017; Folley et al., 2019; Maasakkers et al., 2020; Shuai et al., 2023; Major et al., 2023). Of these five, most found a negative association between sedentary behavior and cognition (3/5), while 1 study found no association, and 1 found a positive association (see Table 1).

Seven studies investigated the associations of sedentary behavior with brain structure and function measured via magnetic resonance imaging (MRI) and magnetic resonance spectroscopy (MRS). Specifically, observational studies reported that greater sedentary behavior was associated with increases in WM damage (Bronas et al., 2019), deterioration of microstructure organization (Burzynska et al., 2014), decrease in regional CBF (Zlatar et al., 2019), and reduction in functional network connectivity (Dion et al., 2021) in older adult samples. Furthermore, a longitudinal study found that high sedentary behavior at baseline was associated with reduction in WM volume at 5-year follow-up (Arnardottir et al., 2016). In contrast, no associations of sedentary behavior with hippocampal volume were observed. In terms of sedentary behavior measurements, all the studies used objective measurement of sedentary behavior using accelerometry, except one study (Bronas et al., 2019), which used self-reported measurement of sedentary behavior (see Table 2).

Our systematic review of the literature suggests that accelerometry is the most popular method of sedentary behavior measurement. The majority of the studies, especially in the most recent studies, assessed the associations of sedentary behavior with cognition (n = 15) utilizing an accelerometer. This is likely because accelerometry technology and validation for use of sedentary behavior in older adulthood has improved (Heesch et al., 2018), and this methodology is becoming more easy to implement. A minority of studies utilized self-report measures only (n = 9), mostly with validated questionnaires, or combined self-report and accelerometry methods (n = 3). Surprisingly, almost all studies except for one that examined brain health outcomes utilized an accelerometer. It is important to note that self-report has some important considerations, as older adults, particularly with cognitive impairment, pay not be as accurate in their reporting (VandeBunte et al., 2022).

Studies that used accelerometry varied in the devices used, the placement of devices, and the data processing cutpoints applied to classify sedentary level cut-offs, which has well-documented implications for the accuracy of capturing activity level, particularly in older adults (Schrack et al., 2016), and using wrist-worn devices (Wu et al., 2020). Compared to placing the accelerometer device on the wrist or hip, thigh device placement is considered the gold standard for the measurement of sedentary behavior because it captures positional information and postural changes (Kozey-Keadle et al., 2011). Thigh-worn devices were only utilized in one study reviewed in this investigation (Čukić et al., 2018). Future research should consider how device type, processing methodology, and placement in interpretation of findings. In addition, as physical activity and sedentary behavior are interrelated but separate behaviors (Dogra and Stathokostas, 2012), study results may differ based on whether physical activity level was adjusted for in analyses. As the field evolves in identifying best practices, future work might consider the importance of developing standardization procedures.

Many studies with self-report of sedentary behavior differed in the way sedentary behavior was categorized, and this could be a potential explanation for why results were not always consistent. Studies that examined specific domains of sedentary behavior may be better equipped to disentangle these discrepancies, as cognitively stimulating sedentary activities may be less detrimental or even helpful to cognition. For example, Major et al. (2023). found that higher amounts of sedentary reading time was positively associated with cognition, while participants who increased their TV watching time in particular, had significantly lower global cognition. Shuai et al. (2023) found that longer screen watching and playing cards was related to better global cognition, while other forms of sedentary behavior that did not involve screen time or playing games were associated with worse global cognition. In addition, not only did the category of sedentary behavior appear to sometimes be differentially related to cognition, but also perhaps the type as well. For example, Chen et al. (2022) found that sedentary time that accumulated in prolonged bouts, but not total sedentary time, was inversely associated with cognitive orientation ability among older adults. These examples demonstrate the importance of specifying the type of sedentary behavior as another important methodological consideration.