Abstract
Study Objectives
Alterations in gut microbiota composition have been associated with several conditions, and there is emerging evidence that sleep quantity and quality are associated with the composition of the gut microbiome. Therefore, this study aimed to assess the associations between several measures of sleep and the gut microbiome in a large, population-based sample.
Methods
Data were collected from participants in the Survey of the Health of Wisconsin from 2016 to 2017 (N = 720). Alpha diversity was estimated using Chao1 richness, Shannon’s diversity, and Inverse Simpson’s diversity. Beta diversity was estimated using Bray-Curtis dissimilarity. Models for each of the alpha-diversity outcomes were calculated using linear mixed effects models. Permutational multivariate analysis of variance tests were performed to test whether gut microbiome composition differed by sleep measures. Negative binomial models were used to assess whether sleep measures were associated with individual taxa relative abundance.
Results
Participants were a mean (SD) age of 55 (16) years and 58% were female. The sample was 83% non-Hispanic white, 10.6% non-Hispanic black, and 3.5% Hispanic. Greater actigraphy-measured night-to-night sleep duration variability, wake-after-sleep onset, lower sleep efficiency, and worse self-reported sleep quality were associated with lower microbiome richness and diversity. Sleep variables were associated with beta-diversity, including actigraphy-measured night-to-night sleep duration variability, sleep latency and efficiency, and self-reported sleep quality, sleep apnea, and napping. Relative abundance of several taxa was associated with night-to-night sleep duration variability, average sleep latency and sleep efficiency, and sleep quality.
Conclusions
This study suggests that sleep may be associated with the composition of the gut microbiome. These results contribute to the body of evidence that modifiable health habits can influence the human gut microbiome.
Keywords: sleep duration, sleep quality, gut microbiome
Graphical Abstract
Graphical Abstract.
Statement of Significance.
Previous studies have found evidence that sleep duration and quality may alter the gut microbiome. However, these studies have had limited sample sizes and have not reported definitive findings. Therefore, we assessed the association between sleep duration and quality and the richness, diversity, and composition of the gut microbiome. We identified several measures of sleep that were associated with gut microbiome richness, diversity, and composition. This study contributes to the body of evidence that modifiable health habits, such as sleep behaviors, can influence the gut microbiome.
Introduction
The gut microbiome is a complex ecosystem housed in the gastrointestinal tract and plays a major role in human health and disease. Alterations in gut microbiota composition have been associated with several diseases and conditions, including allergies, cancer, autism, obesity, irritable bowel disorder, and type 2 diabetes [1].
Recent studies suggest that many exposures influence the gut microbiome, including infant feeding methods, diet, genetics, and medications, especially antibiotics [2]. There is also emerging evidence that sleep quantity and quality are associated with the composition of the gut microbiome [3–5]. In particular, there may be bidirectional relationships between sleep and the gut microbiome via the gut-brain axis [6]. Experimental studies in rodents suggest sleep restriction and sleep fragmentation lead to decreased microbial richness and diversity [7–9].
Investigations into the relationship between sleep and gut microbial ecology are in their nascent stages. However, a few observational studies have considered whether measures of sleep quality and duration are associated with the richness and diversity of the human gut microbiome. Two small studies suggest that sleep quality and duration may influence gut microbiome composition [4, 10]. In addition, there is growing evidence for bidirectional effects between a sleep-disrupting disorder, obstructive sleep apnea, and gut microbiome composition [5].
Therefore, the objectives of this study were to examine associations of subjectively and objectively assessed sleep metrics with indices of gut microbiome richness and diversity in a general population sample of adults. We hypothesized that better sleep quality and longer sleep duration would be associated with greater richness and diversity of the gut microbiome and increased abundance of beneficial taxa. We tested these hypotheses in a large, population-based sample of Wisconsin residents who participated in the Survey of the Health of Wisconsin (SHOW).
Methods
Data collection
The SHOW is an annual survey of the health status of a randomly selected representative sample of Wisconsin residents and communities that began in 2008. SHOW protocols and informed consent documents were approved by the University of Wisconsin-Madison Health Sciences Institutional Review Board (2013-0251). The study methods have been previously described in detail [11–13]. Briefly, a 2-stage cluster sampling method was used to randomly select households from census block groups. Members of selected houses were enumerated and recruited to participate in an in-home interview, complete questionnaires, and participate in a physical exam at a mobile exam center or clinic.
Stool samples were collected as part of a SHOW protocol from 2016 to 2018; a detailed description of stool sample collection and processing has been published previously [14]. Briefly, stool samples were collected at home, following the 7 days of accelerometry data collection, using a collection kit provided by a SHOW study interviewer. This kit included a stool collection tub, a sterile collection cup, a sterile wooden tongue depressor, gloves, a specimen label, a biohazard bag, and an instruction sheet. Participants were instructed to collect the sample during the 24 hours before their scheduled appointment and to refrigerate the sample until their appointment.
Stool samples were shipped overnight to the Infectious Disease Research Laboratory at the University of Wisconsin, Madison and were frozen at −80°C immediately upon receipt. DNA was extracted for 16S ribosomal RNA-based microbiota sequencing as previously described [14]. Bacterial cells from stool samples were lysed using chemical, heat, and mechanical methods and extracted DNA was purified using a phenol-chloroform-isoamyl wash, followed by NucleoSpin Gen and a PCR clean-up kit (Mcherey-Nagel, Germany). The DNA was normalized to concentrations of 5 ng/μL and amplified using PCR with barcoded primers targeting the 16S V3-V4 region of the 16S rRNA gene. Samples were pooled into a single library and sequenced on an Illumina Miseq (Illumina, Inc., San Diego, CA) using a 2 × 250bp v2 sequencing kit, following manufacturer’s instructions. The raw sequence data was processed using Mothur following a Standard Operating Procedure for MiSeq data [15]. Briefly, overlapping sequences were aligned using the SILVA database (v. 132) [16]. Low-quality reads and chimeras, as detected by UCHIME [17], were removed. Sequences were grouped into operational taxonomic units with a threshold of 97% similarity using the GreenGenes database [18]. Rare operational taxonomic units, which we defined as those with a relative abundance ≤0.001% of the overall OTU count, were removed and samples were rarefied to an even depth of 10 000 reads.
Participants wore an accelerometry device (Actigraph wGT3X-BT unit [Actigraph Corporation, Pensacola, FL]) on their wrist to measure sleep and another on their waist to measure physical activity for 7 days and 6 nights. Data were aggregated into 60-second epochs for wear time validation, scoring, and analysis. Sleep data were scored both manually and automatically. In-bed and out-bed times were identified manually, based on activity recorded by the ActiGraph and paper logs filled out by the participants. The Cole-Kripke algorithm was used to distinguish sleeping and waking periods during the time spent in bed [19]. Physical activity data were measured using a hip accelerometer and processed using Freedson’s cutpoints to estimate total sedentary time and total time spent in moderate to vigorous physical activity during the measurement period [20].
Variables
Exposure variables.
Sleep variables. Objective sleep measures were derived from wrist-worn actigraph data, including sleep duration, within-person night-to-night sleep duration standard deviation, sleep efficiency (percent time spent sleeping between first falling asleep and waking up for the day), wake after sleep onset (WASO, the number of minutes spent awake between first falling asleep and waking up for the day), and average sleep latency (number of minutes between getting into bed and falling asleep). Because actigraphy was recorded for 7 days (6 nights), a weighted average was used for sleep measures including sleep duration, sleep efficiency, WASO, and latency (e.g. {[5 × weeknight mean] + [2 × weekend mean]}/7) to account for participants who wore the actigraphy for a different number of week- and weekend nights. Sleep duration standard deviation was the within-individual standard deviation of each night’s actigraphy-measured sleep duration.
Self-reported sleep data were collected in a questionnaire that participants were given at the end of the in-home interview. These questionnaires were completed and returned to SHOW offices by mail. Respondents reported their typical weekday and weekend sleep duration over the past 30 days, which was used to calculate a weighted nightly sleep duration as (5 × weeknight sleep duration + 2 × weekend sleep duration)/7. Participants also indicated the number of days napped in a typical week in the past month, and whether a physician has ever told them they have sleep apnea, insomnia, or restless legs syndrome. Participants were asked to report the quality of their sleep in the past month as excellent, very good, good, fair, or poor (dichotomized in analyses as 1 = excellent, very good, or good; 0 = fair or poor); participants reported the frequency of four insomnia symptoms, with response options that included never, rarely, sometimes, often, or almost always: trouble falling asleep, waking during the night, waking up too early in the morning, and feeling excessively sleepy (dichotomized in analyses as 1 = sometimes, often, or almost always; 0 = never or rarely). Participants also completed the Epworth Sleepiness Scale [21]. For complete questionnaires, (see Supplementary Table S1). As expected, several of the objective and self-reported sleep measures used in this analysis were significantly correlated (Supplementary Table S2).
Microbiome variables.
Alpha-diversity. Three metrics of alpha diversity and richness were used. Richness, an indicator of the number of distinct species present in the gut, was estimated using Chao1 richness [22]. Diversity, an indicator that incorporates microbiome richness and a measure of evenness of abundance of each species, was estimated with Shannon’s and Inverse Simpson’s diversity [23, 24]. Chao1, Shannon, and Inverse Simpson metrics were calculated in R, using the phyloseq package [25].
Beta-diversity. We used the Bray-Curtis (BC) dissimilarity index to estimate the compositional differences between samples. BC dissimilarity incorporates both the presence and abundance of shared taxa, and ranges from 0 (indicating samples are identical) to 1 (indicating samples do not have any taxa in common).
Covariates.
Medication use was considered a potential confounding factor, as several classes of medication have been found to be associated with changes in gut microbiome compositions. In particular, use of antibiotics is strongly associated with gut microbiota; antibiotic intake appears to initially cause a drop in richness and diversity, which is followed by rapid changes after antibiotic cessation, during which the community returns to a state similar to (though not identical to) its pre-antibiotic configuration [26, 27]. Additionally, some antidepressant medications have been demonstrated to affect the gut microbiome [28]. Therefore, the use of both antibiotics and antidepressants are considered potential confounders and are examined in some of the models. Participants were interviewed about antidepressant prescription medication use over the past 30 days and self-reported medications were compared with the RxNorm database for identification of antidepressants [29]. Antibiotic use in the last year was assessed during the in-home interview.
SHOW participants were also asked about the presence of a variety of healthcare provider-diagnosed disease conations. Here, we examined type 2 diabetes and cardiovascular disease (CVD), collected during the in-home interview, as potential confounding factors.
Participants completed the Diet History Questionnaire II (DHQ), a food frequency questionnaire developed by the National Cancer Institute, which ascertains usual food consumption patterns in the previous 12 months [30]. From this survey, we used Diet*Calc to estimate participants’ typical daily consumption (in grams) of carbohydrates, fats, proteins, dietary fiber, and alcohol [31].
Participants completed a computer-based audio questionnaire that included the Patient Health Questionniare-2, a screener for depression [32]. Participants were considered to have depression if their PHQ-2 score was greater than or equal to 3 (i.e. they responded that they had little interest in doing things or felt down, depressed, or hopeless on at least several days in the past 2 weeks).
We also considered physical activity and sedentary behavior a potential confounder, as previous work in this cohort has found some evidence of a relationship between these variables and gut microbiome composition [33].
All models additionally adjusted for sample age (i.e. the time between when the stool sample was produced and when it was put into storage at −80°C), to account for potential changes that occur during cold storage of stool samples during shipping [34].
Statistical Analysis
First, diet, sleep, and accelerometry variables were examined for outliers and implausible values were set to missing. Next, all missing data in this analysis were imputed using the mice package in R [35]. Day-level values for accelerometer-measured sleep variables were estimated and combined into the weighted averages used in the analysis. Predictors were chosen using mice’s quickpred function, whose steps have been previously detailed [36]. No participants were excluded from the analysis based on missing accelerometry data, as previous studies have shown that imputation improves precision and reduces bias [37]. Prior to imputation, 87% of participants wore accelerometers for the full 7-day protocol and the mean number of days worn was 6.2.
Models for each of the alpha-diversity outcomes were calculated using pooled imputed data and linear mixed effects models with random intercepts to account for the potential inclusion of multiple individuals per household using the lme4 package in R [38]. Associations between each of the actigraphy-measured and self-reported sleep variables and alpha-diversity outcomes were evaluated with separate sets of models, with additional covariates included in each successive model. Model 1 included the covariates age and sex; model 2 added antibiotic duration and antidepressant medications; model 3 additionally included BMI, moderate to vigorous physical activity, and sedentary time; model 4 added dietary fat, dietary fiber, carbohydrates, protein, and alcohol consumption; and model 5 also included physician-diagnosed diabetes and CVD, and depression status.
Next, imputed data were used to perform permutational multivariate analysis of variance (PERMANOVA) tests [39] to estimate whether there was a difference in overall microbiome composition by sleep measures using BC dissimilarity, calculated in vegan [40]. Imputed data could not be pooled for PERMANOVA calculations, and a single imputation was used for each calculation. Each analysis was run on several different imputations to ensure that results were robust to the choice of imputed dataset, but results from one such calculation are presented here.
Sleep measures that were significantly associated with overall microbiome composition in our PERMANOVA analyses were further analyzed using negative binomial models to assess whether they were associated with any individual genera, after removing genera from the dataset comprised of more than 30% zeros to reduce multiple testing and the effect of zero-inflation. These models were adjusted for the full set of covariates (model 5 above) and included an offset of the logged number of microbial counts. The Benjamini–Hochberg procedure was used to adjust for multiple testing, and results were considered significant if pBH ≤ 0.1 [41].
Results
During 2016–2017, 805 individuals participated in the Survey of the Health of Wisconsin, but four were removed from the present analysis because they reported taking antibiotics at the time of stool collection. Of the remaining 801 participants, 727 provided a viable stool sample, and 7 samples were removed during rarefaction due to low read counts, leaving 720 individuals in the final analytical sample. The mean (SD) age of the sample was 55 (16) years; 58% of the sample was women; 83% of the sample was non-Hispanic white, 10.6% was non-Hispanic black, and 3.5% of the sample was Hispanic. Descriptive statistics are presented in Table 1.
Table 1.
Selected Descriptive Statistics of Study Participants, N = 720
| Participant characteristics | Mean (SD) or N (%) | Range |
|---|---|---|
| Age, years | 54.9 (16.1) | 18.0–94.0 |
| Sex (Male, Female, % Female) | 303, 417 (58%) | |
| Race/ethnicity | ||
| Non-Hispanic white | 595 (83%) | |
| Non-Hispanic black | 76 (10.6%) | |
| Hispanic | 25 (3.5%) | |
| Other | 24 (3.3%) | |
| BMI, kg/m2 | 30.7 (7.6) | 17.1–66.7 |
| Diabetes (Yes, No, % Yes) | 107, 613, 14.9% | |
| Cardiovascular disease (Yes, No, % Yes) | 85, 635, 11.8% | |
| Depression (Yes, No, % Yes) | 57, 663, 7.9% | |
| Antidepressant use (Yes, No, % Yes) | 214, 506, 29.7% | |
| Total sedentary time (minutes) | 3,545.0 (902.4) | 815–6316 |
| Carbohydrate intake (g) | 225.8 (137.3) | 13.1–1106.8 |
| Protein intake (g) | 74.9 (47.5) | 3.3–526.9 |
| Fat intake (g) | 76.1 (44.9) | 4.3–310.6 |
| Alcohol intake (g) | 8.7 (17.2) | 13.1–186.0 |
| Microbiome measures | Mean (SD) or N (%) | Range |
| Shannon index | 3.2 (0.5) | 0.0–4.4 |
| Inverse Simpson index | 14.2 (6.3) | 1.0–42.9 |
| Chao1 | 175.6 (53.1) | 3.0–56.00 |
| Sleep measures | Mean (SD) or N (%) | Range |
| Average sleep duration (minutes) | 489.6 (64.6) | 229.9–722.0 |
| Night-to-night sleep duration variability (minutes) | 72.3 (37.8) | 2.4–244.9 |
| Average WASO (minutes) | 50.1 (26.1) | 0–179.3 |
| Average sleep latency (minutes) | 3.0 (1.0) | 0–7.8 |
| Average sleep efficiency (minutes) | 89.2 (5.0) | 66.4–100.0 |
| Self-reported sleep duration (hours) | 7.2 (1.2) | 3.8–10.9 |
| Epworth Score | 6.8 (4.2) | 0–23 |
| Days napped | 2.2 (2.5) | 0.0–8.0 |
| Sleep quality | ||
| Excellent | 39 (5.4%) | |
| Very good | 139 (19.3%) | |
| Good | 285 (39.6%) | |
| Fair | 196 (27.2%) | |
| Poor | 61 (8.5%) | |
| Sleep apnea (Yes, No, % Yes) | 108, 612 (15.0%) | |
| Insomnia (Yes, No, % Yes) | 22, 698 (3.1%) | |
| Restless leg syndrome (Yes, No, % Yes) | 32, 688 (4.4%) | |
| Trouble falling asleep | ||
| Never | 129 (17.9%) | |
| Rarely | 220 (30.6%) | |
| Sometimes | 215 (29.9%) | |
| Often | 110 (15.3%) | |
| Almost always | 46 (6.4%) | |
| Woke up during the night | ||
| Never | 104 (14.4%) | |
| Rarely | 205 (28.5%) | |
| Sometimes | 241 (33.5%) | |
| Often | 113 (15.7%) | |
| Almost always | 57 (7.9%) | |
| Woke up too early | ||
| Never | 156 (21.7%) | |
| Rarely | 204 (28.3%) | |
| Sometimes | 220 (30.6%) | |
| Often | 94 (13.1%) | |
| Almost always | 46 (6.4%) | |
| Felt excessively sleepy | ||
| Never | 68 (9.4%) | |
| Rarely | 203 (28.2%) | |
| Sometimes | 273 (37.9%) | |
| Often | 133 (18.5%) | |
| Almost Always | 43 (6.0%) | |
After filtering of chimeras, low-quality reads, and sequences of incorrect length, there were 23 788 688 reads, with an average of 32 632 reads per sample. These filtered reads belonged to 6645 unique taxa. After rarefaction and removal of rare taxa, there are 871 unique taxa remaining.
Alpha-diversity
To assess whether measures of sleep quantity and quality were associated with overall gut microbiome richness and diversity, we used linear mixed effects models to test the association between sleep measures and Chao1 richness, Inverse Simpson’s diversity, and Shannon’s diversity with random intercepts to account for clustering of participants within households. We used nested models, progressively adjusting for additional covariates that we identified as potential confounders. Across all models, greater night-to-night sleep duration variability was associated with decreased richness and diversity (Tables 2–4). For example, in the fully adjusted model, an increase of 1 minute of night-to-night sleep duration variability was associated with a decrease of 0.14 (SE 0.06) in Chao1 richness, a decrease of 0.01 (SE 0.007) in Inverse Simpson’s diversity, and a decrease of 0.001 (SE 0.0005) in Shannon’s diversity. Greater actigraphy-measured WASO and lower sleep efficiency were also associated with lower species richness and diversity, across all models. Increased WASO was associated with a 0.2 (SE 0.08) decrease in Chao1 richness in models 1–3 and increased average efficiency was associated with an increase of 1.1 (SE 0.4) in Chao1 richness in models 1 and 2. WASO was inversely associated with Inverse Simpson and Shannon’s diversity, though the findings were borderline significant (p < 0.1) (Tables 3 and 4). Neither actigraphy-measured sleep duration nor sleep latency was associated with richness or diversity (Tables 2–4).
Table 2.
Linear Mixed Effects Model Estimates of the Association Between Sleep and Chao1 Richness With Random Intercepts to Account for Clustering of Participants by Household
| Sleep variable | Model 1 estimate (SE) | Model 2 estimate (SE) |
Model 3 estimate (SE) |
Model 4 estimate (SE) |
Model 5 estimate (SE) |
|---|---|---|---|---|---|
| Actigraphy measures | |||||
| Average sleep duration | −0.03 (0.03) | −0.03 (0.03) | −0.02 (0.03) | −0.02 (0.03) | −0.02 (0.03) |
| Night-to-night sleep duration variability | −0.20 (0.05)*** | −0.20 (0.05)*** | −0.18 (0.05)*** | −0.15 (0.06)*** | −0.14 (0.06)** |
| Average WASO | −0.2*** (0.08) | −0.2*** (0.08) | −0.2** (0.08) | −0.1* (0.08) | −0.1* (0.08) |
| Average latency | 0.02 (2.1) | 0.2 (2.1) | 0.07 (2.0) | −0.3 (2.0) | −0.4 (2.0) |
| Average efficiency | 1.1*** (0.4) | 1.1*** (0.4) | 0.9** (0.4) | 0.7* (0.4) | 0.6 (0.4) |
| Self-reported measures | |||||
| Self-reported sleep duration (hrs) | 0.5 (1.8) | 0.3 (0.18) | −0.07 (1.7) | −0.1 (1.7) | −0.2 (1.8) |
| Epworth score | −0.3 (0.5) | −0.1 (0.5) | 0.09 (0.5) | 0.3 (0.5) | 0.3 (0.5) |
| Days napped | −0.9 (0.8) | −0.7 (0.8) | −0.2 (0.8) | 0.2 (0.8) | 0.2 (0.8) |
| Sleep qualitya | −11.5*** (4.1) | −10.8*** (4.2) | −8.8** (4.1) | −7.0* (4.1) | −7.0* (4.1) |
| Sleep apnea (ref = No) | −9.5* (5.7) | −7.7 (5.7) | −1.4 (5.9) | −0.05 (5.9) | 0.7 (5.9) |
| Insomnia (ref = No) | −1.6 (11.5) | −0.4 (11.5) | −1.7 (11.4) | 2.4 (11.6) | 2.8 (11.8) |
| Restless legs (ref = No) | −18.6* (9.8) | −15.8 (9.9) | −11.3 (9.9) | −7.6 (10.0) | −6.9 (9.9) |
| Trouble falling asleepb | −6.5 (4.0) | −6.0 (4.0) | −4.9 (4.0) | −4.3 (4.0) | −4.5 (4.0) |
| Woke up during the nightb | −8.0** (4.0) | −7.7* (4.0) | −6.6 (4.0) | −5.7 (4.0) | −6.0 (4.0) |
| Woke up too earlyb | −5.4 (4.0) | −5.1 (4.0) | −4.2 (3.9) | −2.7 (4.0) | −2.1 (4.0) |
| Felt excessively sleepyb | −3.9 (4.1) | −2.3 (4.2) | −1.1 (4.1) | −0.2 (4.1) | −0.6 (4.1) |
* p < 0.1, ** p < 0.05, *** p < 0.01.
Model 1 included the covariates age, sex, and sample age; model 2 added antibiotic duration and antidepressant medications; model 3 additionally included BMI, MVPA, and sedentary time; model 4 added dietary fat, dietary fiber, carbohydrates, protein, and alcohol consumption; and model 5 also included diabetes, CVD, depression.
aParticipant responses “excellent,” “very good,” and “good” were condensed into one category, and “fair” and “poor” another to create a binary variable. Reference = excellent/good/very good.
bParticipant responses “almost always,” “often,” and “sometimes” were condensed into one category, and “rarely” and “never” another to create a binary variable. Reference = never/rarely.
Table 4.
Linear Mixed Effects Model Estimates of the Association Between Sleep and Shannon’s Diversity With Random Intercepts to Account for Clustering of Participants by Household
| Sleep variable | Model 1 estimate (SE) |
Model 2 estimate (SE) |
Model 3 estimate (SE) |
Model 4 estimate (SE) |
Model 5 estimate (SE) |
|---|---|---|---|---|---|
| Actigraphy measures | |||||
| Average sleep duration | −0.0003 (0.0003) | −0.0003 (0.0003) | −0.0002 (0.0003) | −0.0002 (0.0003) | −0.0002 (0.0003) |
| Night-to-night sleep duration variability | −0.002 (0.0005)*** | −0.002 (0.0005)*** | −0.002 (0.0005)*** | −0.001 (0.0006)** | −0.001 (0.0005)** |
| Average WASO | −0.001* (0.0008) | −0.001* (0.0008) | −0.001 (0.0008) | −0.0009 (0.0008) | −0.0007 (0.0008) |
| Average Latency | 0.002 (0.02) | 0.004 (0.02) | 0.004 (0.02) | −0.0007 (0.02) | −0.002 (0.02) |
| Average Efficiency | 0.007* (0.004) | 0.007* (0.004) | 0.006 (0.004) | 0.004 (0.004) | 0.003 (0.004) |
| Self-reported measures | |||||
| Self-reported sleep duration (hrs) | −0.005 (0.02) | −0.006 (0.02) | −0.007 (0.02) | −0.008 (0.02) | −0.01 (0.02) |
| Epworth score | 0.0008 (0.005) | 0.002 (0.005) | 0.003 (0.005) | 0.004 (0.005) | 0.006 (0.005) |
| Days napped | −0.01 (0.008) | −0.01 (0.008) | −0.008 (0.009) | −0.004 (0.009) | −0.003 (0.009) |
| Sleep qualitya | −0.06 (0.04) | −0.06 (0.04) | −0.05 (0.04) | −0.04 (0.04) | −0.03 (0.04) |
| Sleep apnea (ref = No) | −0.009 (0.06) | 0.009 (0.06) | 0.04 (0.06) | 0.05 (0.06) | 0.06 (0.06) |
| Insomnia (ref = No) | 0.12 (0.1) | 0.1 (0.1) | 0.1 (0.1) | 0.2 (0.1) | 0.2* (0.1) |
| Restless legs (ref = No) | −0.07 (0.1) | −0.05 (0.1) | −0.02 (0.1) | 0.02 (0.1) | 0.03 (0.1) |
| Trouble falling asleepb | −0.06 (0.04) | −0.06 (0.04) | −0.05 (0.04) | −0.05 (0.04) | −0.05 (0.04) |
| Woke up during the nightb | −0.04 (0.04) | −0.03 (0.04) | −0.03 (0.04) | −0.02 (0.04) | −0.02 (0.04) |
| Woke up too earlyb | −0.005 (0.04) | −0.001 (0.04) | 0.003 (0.04) | 0.01 (0.04) | 0.02 (0.04) |
| Felt excessively sleepyb | 0.04 (0.04) | 0.06 (0.04) | 0.06 (0.04) | 0.07 (0.04) | 0.07* (0.04) |
* p < 0.1, ** p < 0.05, *** p < 0.01.
Model 1 included the covariates age, sex, and sample age; model 2 added antibiotic duration and antidepressant medications; model 3 additionally included BMI, MVPA, and sedentary time; model 4 added dietary fat, dietary fiber, carbohydrates, protein, and alcohol consumption; and model 5 also included diabetes, CVD, and depression status.
aParticipant responses “excellent,” “very good,” and “good” were condensed into one category, and “fair” and “poor” another to create a binary variable. Reference = excellent/very good/good.
bParticipant responses “almost always,” “often,” and “sometimes” were condensed into one category, and “rarely” and “never” into another to create a binary variable. Reference = never/rarely.
Table 3.
Linear Mixed Effects Model Estimates of the Association Between Sleep and Inverse Simpson’s Diversity With Random Intercepts to Account for Clustering of Participants by Household
| Sleep variable | Model 1 estimate (SE) |
Model 2 estimate (SE) |
Model 3 estimate (SE) |
Model 4 estimate (SE) |
Model 5 estimate (SE) |
|---|---|---|---|---|---|
| Actigraphy measures | |||||
| Average sleep duration | −0.003 (0.004) | −0.003 (0.004) | −0.003 (0.004) | −0.003 (0.004) | −0.004 (0.004) |
| Night-to-night sleep duration variability | −0.02 (0.007)** | −0.02 (0.007)*** | −0.02 (0.007)** | −0.01 (0.007)** | −0.01 (0.007)* |
| Average WASO | −0.01 (0.009)* | −0.01 (0.009)* | −0.01 (0.009) | −0.01 (0.009) | −0.007 (0.009) |
| Average latency | 0.02 (0.2) | 0.03 (0.2) | 0.03 (0.2) | −0.02 (0.2) | −0.05 (0.2) |
| Average efficiency | 0.07 (0.05) | 0.07 (0.05) | 0.06 (0.05) | 0.04 (0.05) | 0.02 (0.05) |
| Self-reported measures | |||||
| Self-reported sleep duration (hrs) | 0.04 (0.2) | 0.03 (0.2) | −0.0009 (0.2) | −0.02 (0.2) | −0.05 (0.2) |
| Epworth Score | −0.02 (0.06) | −0.008 (0.06) | 0.003 (0.06) | 0.02 (0.06) | 0.04 (0.06) |
| Days napped | −0.1 (0.1) | −0.1 (0.1) | −0.09 (0.1) | −0.04 (0.1) | −0.02 (0.1) |
| Sleep qualitya | −0.9 (0.5)* | −0.8 (0.5)* | −0.7 (0.5) | −0.6 (0.5) | −0.5 (0.5) |
| Sleep apnea (ref = No) | −0.002 (0.7) | 0.2 (0.7) | 0.5 (0.7) | 0.6 (0.7) | 0.8 (0.7) |
| Insomnia (ref = No) | 2.1 (1.4) | 2.2 (1.4) | 2.1 (1.4) | 2.4* (1.4) | 3.0** (1.4) |
| Restless legs (ref = No) | 0.3 (1.2) | 0.06 (1.2) | 0.2 (1.2) | 0.7 (1.2) | 0.8 (1.2) |
| Trouble falling asleepb | −0.9* (0.5) | −0.8* (0.5) | −0.8* (0.5) | −0.8* (0.5) | −0.7 (0.5) |
| Woke up during the nightb | −0.4 (0.5) | −0.3 (0.5) | −0.3 (0.5) | −0.2 (0.5) | −0.2 (0.5) |
| Woke up too earlyb | −0.2 (0.5) | −0.1 (0.5) | −0.08 (0.5) | −0.01 (0.5) | 0.2 (0.5) |
| Felt excessively sleepyb | 0.3 (0.5) | 0.4 (0.5) | 0.5 (0.5) | 0.5 (0.5) | 0.5 (0.5) |
* p < 0.1, ** p < 0.05, *** p < 0.01.
Model 1 included the covariates age, sex, and sample age; model 2 added antibiotic duration and antidepressant medications; model 3 additionally included BMI, MVPA, and sedentary time; model 4 added dietary fat, dietary fiber, carbohydrates, protein, and alcohol consumption; and model 5 also included diabetes, CVD, depression.
aParticipant responses “excellent,” “very good,” and “good” were condensed into one category, and “fair” and “poor” another to create a binary variable. Reference = excellent/good/very good.
bParticipant responses “almost always,” “often,” and “sometimes” were condensed into one category, and “rarely” and “never” another to create a binary variable. Reference = never/rarely.
Worse self-reported sleep quality was significantly associated with lower Chao1 richness (Table 2). The group that self-reported a physician diagnosis of sleep apnea had lower Chao1 richness than those without a self-reported diagnosis in model 1 (adjusted for age and sex only) and self-reported restless leg syndrome was associated with lower Chao1 richness (Table 2). None of the self-reported sleep measures was consistently associated with Inverse Simpson or Shannon measures of diversity across multiple models.
Beta-diversity
Next, we aimed to assess whether overall microbiome composition, estimated using the BC dissimilarity index, differed significantly based on actigraphy-measured or self-reported sleep quality or quantity. We found that BC dissimilarity differed significantly by night-to-night sleep duration variability (p = 0.001), sleep latency (p = 0.01), and sleep efficiency (p = 0.04), self-reported sleep quality (p = 0.03), and napping at least once per week (p = 0.05) (Table 5). Among these factors, night-to-night sleep duration variability explained the most variability in microbiome composition (R2 = 0.004).
Table 5.
Results of Several PERMANOVA Tests Estimating Whether There was Clustering by Sleep Quality and Duration, Using Bray-Curtis Dissimilarity
| Sleep predictor variable | R 2 (p-value) |
|---|---|
| Actigraphy measures | |
| Average sleep duration | 0.002 (0.1) |
| Night-to-night sleep duration variability | 0.004*** (0.001) |
| Average WASO | 0.002 (0.1) |
| Average latency | 0.003** (0.01) |
| Average efficiency | 0.002** (0.04) |
| Self-reported measures | |
| Self-reported sleep duration (hrs) | 0.001 (0.8) |
| Epworth score | 0.002 (0.3) |
| Days napped | 0.003** (0.01) |
| Sleep qualitya | 0.002** (0.03) |
| Sleep apnea (ref = No) | 0.002* (0.05) |
| Insomnia (ref = No) | 0.001 (0.8) |
| Restless legs (ref = No) | 0.002 (0.1) |
| Trouble falling asleepb | 0.001 (0.5) |
| Woke up during the nightb | 0.002 (0.2) |
| Woke up too earlyb | 0.002 (0.2) |
| Felt excessively sleepyb | 0.002 (0.2) |
* p < 0.1, ** p < 0.05, *** p < 0.01.
aParticipant responses “excellent,” “very good,” and “good” were condensed into one category, and “fair” and “poor” another to create a binary variable. Reference = excellent/very good/good.
bParticipant responses “almost always,” “often,” and “sometimes” were condensed into one category, and “rarely” and “never” another to create a binary variable. Reference = never/rarely.
Taxa
Finally, we explored whether the abundance of any taxa differed by sleep measures that were identified in the PERMANOVA analyses (i.e. night-to-night sleep duration variability, average latency, average efficiency, sleep quality, and days napped). We found that several microbial taxa were significantly associated with objectively measured sleep latency and sleep efficiency, as well as self-reported sleep quality. Increased night-to-night sleep duration variability was associated with an increase in an unclassified genus from the phylum Firmicutes (pBH = 0.05). Longer average sleep latency was associated with higher abundances of sequences belonging to the genera Coprococcus and Clostridium and a reduced abundance of sequences from the family Mogibacteriaceae (pBH = 0.04, 0.04, and 0.06, respectively). Higher sleep efficiency was associated with a greater abundance of sequences in the genera Subdoligranulum and Adlercreutzia (pBH = 0.03, both). Better sleep quality was associated with lower abundance of sequences from the family Christensenellaceae (pBH = 0.05) (Table 6).
Table 6.
Results of Negative Binomial Models Estimating the Relationship Between Sleep Measures and the Abundance of Microbial Taxa
| Genus | Direction | p | p BH | |
|---|---|---|---|---|
| Night-to-night sleep duration variability | Unclassified, phylum Firmicutes | ↑ | 0.002 | 0.05 |
| Average latency | Coprococcus | ↑ | 0.002 | 0.04 |
| Clostridium | ↑ | 0.003 | 0.04 | |
| Unclassified genus, family [Mogibacteriaceae] | ↓ | 0.005 | 0.06 | |
| Average efficiency | Subdoligranulum | ↑ | 0.002 | 0.03 |
| Adlercreutzia | ↑ | 0.001 | 0.03 | |
| Sleep qualitya | Unclassified genus, family Christensenellaceae | ↓ | 0.002 | 0.049 |
Models were adjusted for age, sex, sample age, antibiotic duration, antidepressant medications, BMI, MVPA, sedentary time, dietary fat, dietary fiber, carbohydrates, protein, and alcohol consumption, diabetes, CVD, and depression status.
aParticipant responses “excellent,” “very good,” and “good” were condensed into one category, and “fair” and “poor” another to create a binary variable. Reference = excellent/very good/good.
Discussion
In a large, community-based sample of adults, we found that measures of sleep are associated with gut microbiome alpha-diversity, beta-diversity, and taxa relative abundance. Though few other studies have investigated these associations, there are some studies that have reported similar findings. For example, one observational study of 26 males (mean [SD] age 22 [3 years]) reported associations between sleep quality and gut microbiome richness and diversity, as assessed by fecal swab sample [4]. Gut microbiome richness, Shannon’s diversity, and Inverse Simpson’s diversity were each correlated with sleep efficiency. Shannon’s diversity was significantly correlated with WASO, and Inverse Simpson’s diversity was significantly associated with total sleep time. Though we did not find any significant associations with total sleep time, these study’s results are otherwise consistent with our findings. In contrast, another previous study assessed the proportion of each phylum of microbiota from fecal samples and evaluated sleep quality with the global score on the Pittsburgh Sleep Quality Index. In this study of 37 adults, ages 50 to 85 years, worse sleep quality was significantly correlated with lower proportions of two phyla, Verrucomicrobia and Lentisphaerae, suggesting that sleep quality may influence microbiome composition [10]. Although the present study was focused on differences in abundance of genera and did not explore whether phyla-level taxonomy differed by sleep, the genera identified as differentially abundant by night-to-night sleep duration variability, average latency, average efficiency, and sleep quality did not belong to the phyla Verrucomicrobia nor Lentisphaerae. This difference may be due to the wider age range represented in our study, as the previous study was focused on older adults.
In general, we found that adverse sleep measures (e.g. greater WASO) were inversely associated with alpha-diversity, while positive sleep measures (e.g. greater sleep efficiency) were associated with increases in alpha-diversity. Though defining a healthy gut microbiome remains elusive, there is emerging consensus that greater alpha diversity is associated with better health outcomes, including better metabolic and mental health outcomes [42–44].
In this study, we found that the abundance of a genus from the family Christensenellaceae was inversely associated with sleep quality and the abundance of a genus from the Mogibacteriaceae family was positively associated with sleep latency. Previous studies of Christensenellaceae and Mogibacteriaceae have found that these families are closely linked with metabolic health, where higher abundance is associated with reduced body mass index [45–47]. Similarly, we found that increased average sleep efficiency was associated with increased abundance of Subdoligranulum, whose abundance has been inversely associated with childhood BMI [48]. Finally, we found that Clostridium was positively associated with sleep latency. Previous studies have found that one Clostridium species decreased following weight loss [49], and that Clostridium was lower in association with greater exercise [33, 50]. Given the known link between sleep and obesity risk [51, 52], this study provides preliminary evidence that alterations to the gut microbiome may mediate the relationship between sleep and obesity. However, further mediation studies are needed to verify this hypothesis.
In this study, we additionally found that increased sleep latency was associated with higher relative abundance of Coprococcus, a butyrate-producing bacteria [53]. Coprococcus is considered a core constituent of the healthy gut microbiome, is increased in healthy adults in comparison to children and older adults, and has been associated with increased health and quality of life [54]. This finding is somewhat counter-intuitive, as we expected that increased sleep latency would be associated with worse sleep quality and negatively impact the composition of the gut microbiome. While this study adjusted for participant age, this finding could be due to residual confounding by age, where older participants may have worse sleep and lower Coprococcus abundance. Additionally, a limitation of 16S sequencing is its primary function as a census of microbes; it cannot indicate transcriptional activity of microbes [55]. Therefore, future analyses should incorporate fecal metabolomics to provide additional context on transcriptional activity to this counter-intuitive finding.
Our study has several strengths, including its large, population-based sample and use of objective measures of sleep. Additionally, this study was able to incorporate several a priori confounders including antibiotic and antidepressant medication use, diet, physical activity, and presence of a number of chronic diseases, all of which may be important confounders of the relationship between sleep and the gut microbiome. However, this study also has limitations which are important to consider. This cross-sectional analysis cannot provide insight into the direction of the association between poorer quality and quantity of sleep and changes to the gut microbiome. While experimental studies can theoretically help to answer the question, the existing studies of experimental sleep curtailment in humans are very small and have not resulted in definitive findings. Even more importantly, our study assessed sleep for one week (objective measures) and 1-month (subjective measures). Therefore, we were unable to examine long-term impacts of poor or short sleep on the gut microbiome. In addition, this study included some individuals from the same household. While we accounted for household-clustered participants in our alpha-diversity analyses, we were not able to account for the nested nature of the data in our beta-diversity or taxa analyses. Additionally, though this study was able to account for the use of antidepressants and antidiabetic drugs, we were not able to classify the use of all potentially confounding medications. Sleep medications may confound the relationship between sleep measures and the gut microbiome. While there is not yet evidence that sleep medications in particular impact the gut microbiota, several types of drugs have been associated with gut microbiome composition [56] and future studies should consider the use of sleep aids. Finally, this study was conducted in a population of primarily non-Hispanic White individuals, which may limit its generalizability to other populations. Future studies should focus on improved representation among additional racial and ethnic groups.
Conclusions
This population-based study of adults suggests that sleep may be associated with the composition of the gut microbiome. These results contribute to the body of evidence that modifiable health habits can shape the human gut microbiome.
Supplementary Material
Contributor Information
Elizabeth A Holzhausen, Department of Integrative Physiology, University of Colorado, Boulder, CO, USA.
Paul E Peppard, Department of Population Health Sciences, University of Wisconsin, Madison, WI, USA.
Ajay K Sethi, Department of Population Health Sciences, University of Wisconsin, Madison, WI, USA.
Nasia Safdar, Department of Medicine and the William S. Middleton Memorial Veterans Hospital, University of Wisconsin, Madison, WI, USA.
Kristen C Malecki, Division of Environmental and Occupational Health Sciences, School of Public Health, University of Illinois Chicago, Chicago, IL, USA.
Amy A Schultz, Department of Population Health Sciences, University of Wisconsin, Madison, WI, USA.
Courtney L Deblois, Department of Bacteriology, University of Wisconsin, Madison, WI, USA.
Erika W Hagen, Department of Population Health Sciences, University of Wisconsin, Madison, WI, USA.
Funding
Funding for the Survey of the Health of Wisconsin was provided by the Wisconsin Partnership Program PERC Award (233 PRJ 25DJ). EAH’s time was supported by a Eunice Kennedy Shriver National Institute of Health and Human Development grant to the Center for Demography and Ecology at the University of Wisconsin-Madison (T32 HD007014) and a Transdisciplinary Training Grant in Sleep and Circadian Research (T32 HL149646). KCM’s time was supported by a National Institute on Aging grant (R01 AG061080) and a National Institute on Aging grant to the Center of Demography of Health and Aging at the University of Wisconsin-Madison (P30 AG17266).
Disclosure Statement
Financial disclosure: none. Nonfinancial disclosure: none.
Data Availability
The data underlying this article were provided by the Survey of the Health of Wisconsin and were used with permission. Data can be requested from the Survey of the Health of Wisconsin at their website, show.wisc.edu.
References
- 1. Clemente JC, Ursell LK, Parfrey LW, Knight R.. The impact of the gut microbiota on human health: An integrative view. Cell. 2012;148:1258–1270. doi: 10.1016/j.cell.2012.01.035 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Wen L, Duffy A.. Factors influencing the gut microbiota, inflammation, and type 2 diabetes. J Nutr. 2017;147:1468S–1475S. doi: 10.3945/jn.116.240754 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Benedict C, Vogel H, Jonas W, et al. Gut microbiota and glucometabolic alterations in response to recurrent partial sleep deprivation in normal-weight young individuals. Mol Metab. 2016;5:1175–1186. doi: 10.1016/j.molmet.2016.10.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Smith RP, Easson C, Lyle SM, et al. Gut microbiome diversity is associated with sleep physiology in humans. PLoS One. 2019;14:e0222394. doi: 10.1371/journal.pone.0222394 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Cai Y, Juszczak HM, Cope EK, Goldberg AN.. The microbiome in obstructive sleep apnea. Sleep. 2021;44(8). doi: 10.1093/sleep/zsab061 [DOI] [PubMed] [Google Scholar]
- 6. Sgro M, Kodila ZN, Brady RD, Reichelt AC, Mychaisuk R, Yamakawa GR.. Synchronizing our clocks as we age: The influence of the brain-gut-immune axis on the sleep-wake cycle across the lifespan. Sleep. 2022;45(3). doi: 10.1093/sleep/zsab268 [DOI] [PubMed] [Google Scholar]
- 7. Poroyko VA, Carreras A, Khalyfa A, et al. Chronic sleep disruption alters gut microbiota, induces systemic and adipose tissue inflammation and insulin resistance in mice. Sci Rep. 2016;6(35405):1–11. doi: 10.1038/srep35405 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Zhang SL, Bai L, Goel N, et al. Human and rat gut microbiome composition is maintained following sleep restriction. Proc Natl Acad Sci USA. 2017;114:E1564–E1571. doi: 10.1073/pnas.1620673114 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Bowers SJ, Vargas F, González A, et al. Repeated sleep disruption in mice leads to persistent shifts in the fecal microbiome and metabolome. PLoS One. 2020;15:e0229001. doi: 10.1371/journal.pone.0229001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Anderson JR, Carroll I, Azcarate-Peril MA, et al. A preliminary examination of gut microbiota, sleep, and cognitive flexibility in healthy older adults. Sleep Med. 2017;38:104–107. doi: 10.1016/j.sleep.2017.07.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Nieto FJ, Peppard PE, Engelman CD, et al. The Survey of the Health of Wisconsin (SHOW), a novel infrastructure for population health research: Rationale and methods. BMC Public Health. 2010;10(785):785. doi: 10.1186/1471-2458-10-785 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Malecki KMC, Nikodemova M, Schultz AA, et al. The survey of the health of wisconsin (SHOW) program: An infrastructure for advancing population health. Front Public Health. 2022;10(818777):818777. doi: 10.3389/fpubh.2022.818777 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Schultz, A. A., Malecki Kristen MC, Holzhausen Elizabeth A, et al. The Population-based Microbiome Research Core: A longitudinal infrastructure for assessment of household microbiome and human health research. medRxiv. 2021. doi: 10.1101/2021.11.22.21266369 [DOI]
- 14. Eggers S, Malecki KM, Peppard P, et al. Wisconsin microbiome study, a cross-sectional investigation of dietary fibre, microbiome composition and antibiotic-resistant organisms: rationale and methods. BMJ Open. 2018;8:e019450. doi: 10.1136/bmjopen-2017-019450 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD.. Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the MiSeq Illumina sequencing platform. Appl Environ Microbiol. 2013;79:5112–5120. doi: 10.1128/AEM.01043-13 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Quast C, Pruesse E, Yilmaz P, et al. The SILVA ribosomal RNA gene database project: Improved data processing and web-based tools. Nucleic Acids Res. 2013;41:D590–D596. doi: 10.1093/nar/gks1219 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Edgar RC, Haas BJ, Clemente JC, Quince C, Knight R.. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics. 2011;27:2194–2200. doi: 10.1093/bioinformatics/btr381 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. DeSantis TZ, Hugenholtz P, Larsen N, et al. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl Environ Microbiol. 2006;72:5069–5072. doi: 10.1128/AEM.03006-05 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Cole RJ, Kripke DF, Gruen W, Mullaney DJ, Gillin JC.. Automatic sleep/wake identification from wrist activity. Sleep. 1992;15:461–469. doi: 10.1093/sleep/15.5.461 [DOI] [PubMed] [Google Scholar]
- 20. Freedson PS, Melanson E, Sirard J.. Calibration of the Computer Science and Applications, Inc accelerometer. Med Sci Sports Exerc. 1998;30:777–781. doi: 10.1097/00005768-199805000-00021 [DOI] [PubMed] [Google Scholar]
- 21. Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep. 1991;14:540–545. doi: 10.1093/sleep/14.6.540 [DOI] [PubMed] [Google Scholar]
- 22. Chao A. Species estimation and applications. in Encyclopedia of Statistical Sciences (John Wiley & Sons, Ltd, Hoboken, New Jersey, USA: ). 2005;12:7907–7916. [Google Scholar]
- 23. Simpson EH. Measurement of diversity. Nature. 1949;163:688–688. doi: 10.1038/163688a0 [DOI] [Google Scholar]
- 24. Peet RK. Relative diversity indices. Ecology. 1975;56:496–498. doi: 10.2307/1934984 [DOI] [Google Scholar]
- 25. McMurdie PJ, Holmes S.. phyloseq: An R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One. 2013;8:e61217. doi: 10.1371/journal.pone.0061217 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Caporaso JG, Lauber CL, Costello EK, et al. Moving pictures of the human microbiome. Genome Biol. 2011;12:R50. doi: 10.1186/gb-2011-12-5-r50 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Dethlefsen L, Relman DA.. Incomplete recovery and individualized responses of the human distal gut microbiota to repeated antibiotic perturbation. Proc Natl Acad Sci USA. 2011;108(suppl 1):4554–4561. doi: 10.1073/pnas.1000087107 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Macedo D, Filho AJMC, Soares de Sousa CN, et al. Antidepressants, antimicrobials or both? Gut microbiota dysbiosis in depression and possible implications of the antimicrobial effects of antidepressant drugs for antidepressant effectiveness. J Affect Disord. 2017;208:22–32. doi: 10.1016/j.jad.2016.09.012 [DOI] [PubMed] [Google Scholar]
- 29. Nelson SJ, Zeng K, Kilbourne J, Powell T, Moore R.. Normalized names for clinical drugs: RxNorm at 6 years. J Am Med Informat Assoc. 2011;18:441–448. doi: 10.1136/amiajnl-2011-000116 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Diet History Questionnaire II (DHQ II) for U.S. and Canada | EGRP/DCCPS/NCI/NIH.https://epi.grants.cancer.gov/dhq2/. Accessed December 5, 2023.
- 31. DHQ II Diet*Calc Software | EGRP/DCCPS/NCI/NIH.https://epi.grants.cancer.gov/dhq2/dietcalc/. Accessed December 5, 2023.
- 32. Kroenke K, Spitzer RL, Williams JBW.. The Patient Health Questionnaire-2: Validity of a two-item depression screener. Med Care. 2003;41:1284–1292. doi: 10.1097/01.mlr.0000093487.78664.3c [DOI] [PubMed] [Google Scholar]
- 33. Holzhausen EA, Malecki KC, Sethi AK, et al. Assessing the relationship between physical activity and the gut microbiome in a large, population-based sample of Wisconsin adults. PLoS One. 2022;17:e0276684. doi: 10.1371/journal.pone.0276684 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Holzhausen EA, Nikodemova M, Deblois CL, et al. Assessing the impact of storage time on the stability of stool microbiota richness, diversity, and composition. Gut Pathog. 2021;13(75):75. doi: 10.1186/s13099-021-00470-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Buuren S. van, Groothuis-Oudshoorn K: Multivariate imputation by chained equations in R. J Stat. Softw. 2011;45:1–67. [Google Scholar]
- 36. van Buuren S, Boshuizen HC, Knook DL.. Multiple imputation of missing blood pressure covariates in survival analysis. Stat Med. 1999;18:681–694. doi: [DOI] [PubMed] [Google Scholar]
- 37. Catellier DJ, Hannan PJ, Murray DM, et al. Imputation of missing data when measuring physical activity by accelerometry. Med Sci Sports Exerc. 2005;37:S555–S562. doi: 10.1249/01.mss.0000185651.59486.4e [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Bates D, Mächler M, Bolker B, Walker S.. Fitting linear mixed-effects models using lme4. J Stat Softw. 2015;67:1–48. [Google Scholar]
- 39. Anderson, M.J. Permutational Multivariate Analysis of Variance (PERMANOVA). in Wiley StatsRef: Statistics Reference Online 1–15 (John Wiley & Sons, Ltd, 2017, Hoboken, New Jersey: ). doi: 10.1002/9781118445112.stat07841 [DOI] [Google Scholar]
- 40. Oksanen, J., et al. Vegan community ecology package version 2.5-7 https://www.researchgate.net/publication/346579465_vegan_community_ecology_package_version_25-7_November_2020. Accessed December 5, 2023 [Google Scholar]
- 41. Benjamini Y, Hochberg Y.. Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Stat Soc Ser B Methodol. 1995;57:289–300. [Google Scholar]
- 42. Lloyd-Price J, Abu-Ali G, Huttenhower C.. The healthy human microbiome. Genome Med. 2016;8(51):51. doi: 10.1186/s13073-016-0307-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Zouiouich S, Loftfield E, Huybrechts I, et al. Markers of metabolic health and gut microbiome diversity: Findings from two population-based cohort studies. Diabetologia. 2021;64:1749–1759. doi: 10.1007/s00125-021-05464-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Järbrink-Sehgal E, Andreasson A.. The gut microbiota and mental health in adults. Curr Opin Neurobiol. 2020;62:102–114. doi: 10.1016/j.conb.2020.01.016 [DOI] [PubMed] [Google Scholar]
- 45. Waters JL, Ley RE.. The human gut bacteria Christensenellaceae are widespread, heritable, and associated with health. BMC Biol. 2019;17(83):83. doi: 10.1186/s12915-019-0699-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Goodrich JK, Waters JL, Poole AC, et al. Human genetics shape the gut microbiome. Cell. 2014;159:789–799. doi: 10.1016/j.cell.2014.09.053 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Oki K, Toyama M, Banno T, Chonan O, Benno Y, Watanabe K.. Comprehensive analysis of the fecal microbiota of healthy Japanese adults reveals a new bacterial lineage associated with a phenotype characterized by a high frequency of bowel movements and a lean body type. BMC Microbiol. 2016;16(284):284. doi: 10.1186/s12866-016-0898-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Houtman TA, Eckermann HA, Smidt H, de Weerth C.. Gut microbiota and BMI throughout childhood: The role of firmicutes, bacteroidetes, and short-chain fatty acid producers. Sci Rep. 2022;12(3140):3140. doi: 10.1038/s41598-022-07176-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Santacruz A, Marcos A, Wärnberg J, et al.; EVASYON Study Group. Interplay between weight loss and gut microbiota composition in overweight adolescents. Obesity (Silver Spring). 2009;17:1906–1915. doi: 10.1038/oby.2009.112 [DOI] [PubMed] [Google Scholar]
- 50. Motiani KK, Collado MC, Eskelinen J-J, et al. Exercise training modulates gut microbiota profile and improves endotoxemia. Med Sci Sports Exerc. 2020;52:94–104. doi: 10.1249/MSS.0000000000002112 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51. Ogilvie RP, Patel SR.. The epidemiology of sleep and obesity. Sleep Health. 2017;3:383–388. doi: 10.1016/j.sleh.2017.07.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52. Beccuti G, Pannain S.. Sleep and obesity. Curr Opin Clin Nutr Metab Care. 2011;14:402–412. doi: 10.1097/MCO.0b013e3283479109 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53. Pryde SE, Duncan SH, Hold GL, Stewart CS, Flint HJ.. The microbiology of butyrate formation in the human colon. FEMS Microbiol Lett. 2002;217:133–139. doi: 10.1111/j.1574-6968.2002.tb11467.x [DOI] [PubMed] [Google Scholar]
- 54. Notting F, Pirovano W, Sybesma W, Kort R.. The butyrate-producing and spore-forming bacterial genus Coprococcus as a potential biomarker for neurological disorders. Gut Microbiome. 2023;4:e16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55. Zierer J, Jackson MA, Kastenmüller G, et al. The fecal metabolome as a functional readout of the gut microbiome. Nat Genet. 2018;50:790–795. doi: 10.1038/s41588-018-0135-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56. Maier L, Typas A.. Systematically investigating the impact of medication on the gut microbiome. Curr Opin Microbiol. 2017;39:128–135. doi: 10.1016/j.mib.2017.11.001 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data underlying this article were provided by the Survey of the Health of Wisconsin and were used with permission. Data can be requested from the Survey of the Health of Wisconsin at their website, show.wisc.edu.