Association Between Obesity and Recent and Ever Suicide Attempt Among Adults in the United States

doi:10.21203/rs.3.rs-1942483/v1

Download PDF

Research Article

Association Between Obesity and Recent and Ever Suicide Attempt Among Adults in the United States

https://doi.org/10.21203/rs.3.rs-1942483/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Purpose

The purpose of this cross-sectional study was to investigate whether obesity is associated with suicide attempt in a nationally representative sample of U.S. adults.

Methods

From the 2004–2005 National Epidemiologic Survey of Alcohol and Related Conditions Wave 2 data (n = 34,653 adults), survey responses were extracted on suicide attempt (ever, recent [≤ 2 years]), demographic (age, sex, race/ethnicity), mental disorders (internalizing, personality, substance abuse), and psychosocial factors (e.g., abuse, trauma, sexual identity). Current body mass index (BMI) categorized participants as obese (≥ 30 kg/m²) or non-obese (< 30 kg/m²). Odds ratio [95% confidence interval] of suicide attempt by obesity status was calculated by logistic regression, while adjusting for mental health, demographic, and psychosocial confounders.

Results

Ever and recent suicide attempt prevalence was 3.4% and 3.1%, respectively, and the obesity prevalence 28%. Compared to non-obese adults, obese adults were 1.43 [1.20–1.69] times and 1.40 [0.90–2.18] times more likely to have ever, and recently, attempted suicide, respectively. Among the obese, middle-age (36–55 years) adults were 0.64 [0.31–0.92] times less likely to have ever-attempted suicide than older (> 55 years) adults. Only 15% of the obesity’s association with lifetime suicide attempt was mediated by mental illnesses.

Conclusion

Current-obesity is associated with lifetime suicide attempt among U.S. adults, even after adjustment for mental health, demographic, and psychosocial confounders. Further research on the temporality and causality of this association is needed.

Obesity

Suicide

Attempt

Adult

United States

In the United States (U.S.), suicide is among the top-10 leading causes of death[1] and a growing public health crisis, killing one American every 11 minutes. During 1999–2018, suicide incidence increased by 35%, reaching the highest level since the Great Depression and World War II[2]. Though adolescent suicide has attracted much attention in the media and the scientific community, the vast majority of suicide deaths occur among adults[3].

The cause of increasing adult suicides remains unclear. Although mental illness is widely recognized as one of the strongest risk factors for suicide[4–6], data suggest that the current trend may not be driven by an increase in mental illness. During 2012–2017, the prevalence of mental illness and substance abuse in the U.S. remained stable or decreased, whereas suicide deaths and attempts increased across all age groups[7]. Furthermore, National Violent Death Reporting System 2003–2017 data revealed that over half of all adult suicides occurred in those without known mental illness[8].

These data indicate that factors other than mental illness may be driving the suicide trend. Some hypothesized that the concurrent obesity epidemic, now affecting 20% of adolescents and 40% of adults in the U.S.[9], might contribute to suicides. Indeed, obesity has been repeatedly shown to be an independent risk factor of suicide behaviors among adolescents[10–12]. In adults, however, the association between obesity and suicide has been less clear. Previous epidemiological studies in adults found conflicting results: no association[13], a non-linear association[14], or an inverse association/correlation[15–17]. Other studies, including a recent systematic review, have reported a positive association. These conflicting results suggest a complex relationship, which may depend on the timing suicide attempts with respect to the obesity onset, as well as potential confounding, modifying, and mediating effects. Thus, the primary aim of this study was to determine if obesity is associated with lifetime history and recent suicide attempt in U.S. adults. Secondary aims were to determine if these associations are modified by age, sex, and race; and if mental illness mediated the obesity–suicide association.

Study Design

This cross-sectional study examined 2004–2005 Wave 2 data from the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) to test the following three hypotheses: (1) obesity is associated with suicide attempt in the U.S. adults; (2) this associations is modified by age, sex, and race/ethnicity; and (3) mental illness partially mediates the effect of obesity on suicide attempt. 2004–2005 NESARC is appropriate, because it is one of the largest publicly-available behavioral and mental-health health dataset on a representative sample of the U.S. adult population.

Participants

Participants comprised a representative sample (N = 34,653) of the U.S. adult population (age ≥ 18 years). The survey employed multistage cluster sampling of households in all 50 States via telephone interviews. Details of the survey were described previously[18]. We excluded participants who answered “unknown” or did not provide an answer to suicide-related questions.

Measures

Main outcome variables. The two dependent variables were self-reported ever-attempted suicide (1 = Yes/0 = No) and self-reported recent suicide attempt in the past two years (1 = Yes/0 = No).

Main predictor variables. The main independent variable was body mass index (BMI [kg/m²]) at the time of survey, calculated as a continuous variable from self-reported height (feet, inches) and weight (pounds), converted to meter and kilogram units, respectively. Then participants were categorized as currently having obesity (1 = > 30 kg/m²) vs. no obesity (0 = under 30 kg/m²) per World Health Organization (WHO) BMI criteria.

Covariates. Self-reported questionnaires on demographics and other variables potentially related to mental health, including social determinants of health, were extracted from the NESARC database as potential confounders. All variable type and coding as originally collected by NESARC are listed in Supplemental Table 1.

We categorized age into three groups: young (18–35 years), middle-aged (36–55), and older (> 55). Race/ethnicity was classified into five categories: non-Hispanic White; non-Hispanic Black; American Indian/Alaska Native; Asian/Native Hawaiian/Other Pacific Islander; Hispanic. Household income level was classified into four categories: <$25,000; $25,000-$49,999; $50,000-$99,999; ≥$100,000. Self-perceived health was classified as: excellent or very good; good; fair; poor. Mental illnesses were classified into three categories: internalizing, personality, substance use/abuse/dependence disorders. All other questionnaires were of binary (Yes/No) responses, such as family history of suicide, ever in active combat, ever sexually assaulted, physical abused, and neglected by parents.

Sample Size

A priori power calculations were performed to ensure sufficient sample size for the primary analysis of obesity–suicide association. During 2004–2005 NESARC Wave 2, the population prevalence of obesity in the U.S. was 32.2%[19]. The prevalence rate of suicide attempts (95% confidence interval [95% CI] in the preceding period (2001–2002) was 2.4% (2.2–2.6%)[20]. Assuming the prevalence odds ratio (POR) of attempts between the obese and non-obese groups is clinically significant (i.e., POR > 2)[21], the sample size required to detect this difference with power of 80% and a of 0.05 is 2,380. If we use the more conservative POR based on the 95% CI lower bounds, and a lower prevalence of obesity (20%), the required sample size is 3,363. Since Wave 2 NESARC has sample size of 34,653 and a population prevalence of at least one mental health disorder was 42.1% (i.e., n = 14,601/20,052 with/without mental illness), this dataset far exceeds the required sample size to conduct stratified analysis of obesity–suicide association in participants with and without mental illness. We had less than 3% of “unknown” or blank response, which were discarded since our precision was far above the required threshold.

Statistical Analysis

Study population characteristics were summarized using descriptive statistics of each variable as mean and standard deviation (SD) for continuous variables, frequency for categorical variables. Bivariate statistics via t-test, Chi-square or Fisher’s were employed to test for the association between the dependent variable and covariates. Adjusted Prevalence Odds Ratio (aPOR) from a multivariate logistic regression model for binary suicide outcome (ever- and recent-attempt) was our primary statistical metric employed to assess the association between suicide behavior and obesity. In this model, we adjusted for confounder variables selected from the demographic and mental health-related variables listed in Table 1 by the p-value method[22].

To assess whether age, sex, or race/ethnicity modify the obesity–suicide association, we created new 2-way, 3-way, and 4-way interaction terms (e.g., Obese*Sex, Obese*Sex*Race, and Obese*Sex*Age*Race) and added them to the above model one by one and assessed the goodness-of-fit via the likelihood ratio test. Based on these results, the multivariable models (for ever- and recent-attempt) above were modified by adding the significant interaction terms. The prediction confidence intervals for the aPOR of the two suicide attempts were calculated for the obese group, as well as for each obese subgroup (by age, sex, race/ethnicity, and mental illness).

To assess whether mental illness partially mediates the association of obesity with suicide attempt, we followed Baron and Kenny’s approach modified by Iacobucci et al.[23] as shown in Figure 1. The ratio of the indirect effect to the total effect (RIT) was used to determine the effect size of significant mediation effect using the following formula: RIT = Indirect effect / Total effect and tested for statistical significance by Sobel test[24].

All statistical data analyses were performed using STATA v15.0 (STATA Corporation, College Station, TX) using survey weights for the multistage cluster design and inverse probability sampling of NESARC. For statistical inferences, p < 0.05 were considered statistically significant, and 95% CIs were calculated for point estimates.

Sensitivity analysis

Sensitivity analysis was performed using surrogate outcomes (suicide ideation, thoughts about death), as well as for different thresholds for WHO weight categories (i.e., overweight [BMI: 25.0–29.9], class I obesity [BMI: 30.0–34.9], class II obesity [BMI: 35.0–39.9], and class III obesity [BMI: ≥40]), for BMI treated as a continuous variable, and for age treated as a continuous variable and as quartiles. For mediation analysis, the Zhao approach[25] was used to assess whether the effect of obesity on the suicide attempt was partially mediated by mental illnesses.

Participant characteristics

Table 1 summarizes the participant characteristics. The mean age of the participants (N = 34,653) was 49 years (SD = 17). Women were slightly more represented than men (52.08% vs. 47.92%, respectively). Prevalence of obesity was 28.15% (17.35%, 6.20%, and 4.60% WHO class I, II, and III obesity, respectively). Most participants identified as heterosexual (98.52%), non-Hispanic White (70.91%), perceived themselves as healthy (84.32%), and had no known mental illness (57.99%). The majority (99.36%; n = 34,432) responded to the ever-attempted suicide question, and 3.40% (n = 1,265) of those who responded answered “yes”. Less than a quarter (21.19%; n = 7,400) of the participants responded to the recent suicide attempt question, and 3.12% (n = 241) of those answered “yes”.

Table 1

Descriptive sample characteristics, 2004–2005 Wave 2 NESARC (N = 34,653)
	N		Weight %
Sociodemographic characteristics
Age (mean, SD*)	49.06 (17.30)
Age
Young adults (18-35yrs)	8744		27.43%
Middle-aged adults (36-55yrs)	14177		40.25%
Older adults (> 55yrs)	11732		32.22%
Sex
Male	14564		47.92%
Female	20089		52.08%
Sexual identity
Heterosexual	33598		98.52%
Gay or lesbian	335		0.86%
Bisexual	242		0.62%
Race/ethnicity
Non-Hispanic White	20161		70.91%
Non-Hispanic Black	6587		11.05%
Hispanic and other**	7905		18.04%
Race/ethnicity
Non-Hispanic White	20161		70.91%
Non-Hispanic Black	6587		11.05%
Non-Hispanic American Indian/Alaska Native	578		2.19%
Non-Hispanic Asian/Native Hawaiian/Other Pacific Islander	968		4.27%
Hispanic, any race	6359		11.58%
Household income level
<$25,000	10947		26.41%
$25,000-$49,999	9815		27.80%
$50,000-$99,999	9606		30.86%
≥$100,000	4285		14.93%
Self-perceived current health status
Excellent/very good	19166		58.59%
Good	9272		25.73%
Fair Poor	4566 1602		11.74% 3.94%
Family history of suicide death or attempt
Yes	1326		3.91%
No	33327		96.09%
Ever in active combat
Yes	1438		4.55%
No	33031		95.45%
Sexually assaulted
Yes	3328		8.74%
No	31087		91.26%
Physically abused by partner
Yes	2573		6.12%
No	31858		93.87%
Neglected by parent
Yes	1195		3.11%
No	33235		96.89%
History of physical abuse
Yes	1320		3.52%
No	33121		96.48%
Mental illness status
Yes	14601		42.01%
No	20052		57.99%
Internalizing disorders
Yes	8427		23.20%
No	26226		76.80%
Personality disorders
Yes	4967		13.55%
No	29686		86.45%
Substance abuse disorders
Yes	7736		23.69%
No	26917		76.31%
Ever-attempted suicide
Yes	1265		3.40%
No	33167		96.60%
Suicide attempt in the past two years
Yes	241		3.12%
No	7147		96.88%
Suicide ideation in the past two years
Yes		1284	17.35%
No		6104	82.65%
Body mass index (mean, SD*)		27.81 (6.03)
Obesity index
No obesity (BMI < 30)		24351	71.82%
Obesity (BMI ≥ 30)		10302	28.18%
Body mass index
Underweight (BMI ≤ 18.5)		499	1.35%
Normal weight (18.5 < BMI < 25)		11593	34.72%
Overweight (25 < BMI < 30)		12259	35.74%
Obese Class I (30.0 < BMI < 35)		6190	17.35%
Obese Class II (35 < BMI < 40)		2337	6.20%
Obese Class III (BMI ≥ 40)		1775	4.60%
* SD = standard deviation. **Other = Asian American/Pacific Islander and Alaskan Native/American Indian. Note: Complete case analysis was performed.

Bivariate analysis

Tables 2 and 3 show bivariate relationship between demographics, mental illnesses, traumatic events, and current-obesity status by the ever- and recent-suicide attempt. For ever-attempted suicide (Table 2), there were statistically significant differences between suicide vs. non-suicide groups by obesity status, BMI as a continuous variable, and WHO weight categories (all p < 0.001), as well as by sociodemographic variables (i.e., age, sex, sexual identity, race, household income level; all p < 0.001). There were also statistically significant differences by all mental illness categories (internalizing, personality, substance abuse; all p < 0.001) and all types of traumas (family history of suicide attempt/death, ever in active combat, sexual harassment, physical abuse, and neglect; all p < 0.001).

For recent suicide attempt (Table 3), no statistically significant difference was found in obesity status, BMI as a continuous variable, or WHO weight categories (P = 0.09, 0.21, and 0.69, respectively). However, there were statistically significant differences by age, sexual identity, all mental illness categories (internalizing, personality, and substance abuse; all p < 0.001) and some but not all types of trauma (family history of suicide attempt, sexual harassment, physical abuse, and neglect; p < 0.001).

Table 2

Association between sociodemographic characteristics, mental illness, traumatic events, and obesity by ever-attempted suicide status (Yes/No), (n = 1,265)a
	Ever attempted suicide (n = 1,265) n(%)	Never attempted suicide (n = 33,167) n(%)	X², t-test or F test (Design based)	P value^b
Age (mean, SE*)	41.70 (0.54)	48.36 (0.15)	141.01	< 0.001
Age
Young adults (18-35yrs)	410 (1.2%)	8292 (26%)	40.50	< 0.001
Middle-aged adults (36-55yrs)	611 (1.6%)	13484 (39%)
Older adults (> 55yrs)	244 (0.58%)	11391 (32%)
Sex
Male	363 (1.1%)	14110 (47%)	53.55	< 0.001
Female	902 (2.3%)	19057 (50%)
Sexual identity
Heterosexual	1154 (3.1%)	32412 (95%)	96.77	< 0.001
Gay or lesbian	39 (0.08)	296 (0.77%)
Bisexual	52 (0.12%)	190 (0.5%)
Race/ethnicity
Non-Hispanic White	750 (2.4%)	19295 (69%)	0.34	0.71
Non-Hispanic Black	214 (0.34%)	6337 (11%)
Other (Hispanic, Asian, American Indian)	301 (0.61%)	7535 (17%)
Race/ethnicity
Non-Hispanic White	750 (2.4%)	19295 (69%)	4.16	0.0222
Non-Hispanic Black	214 (0.34%)	6337 (11%)
Non-Hispanic American Indian/Alaska Native	47 (0.14%)	524 (2%)
Non-Hispanic Asian/Native Hawaiian/Other Pacific Islander	27 (0.08%)	920 (4.1%)
Hispanic, any race	227 (0.38%)	6091 (11%)
Household income level
<$25,000	594 (1.4%)	10256 (25%)	31.60	< 0.001
$25,000-$49,999	339 (0.94%)	9420 (27%)
$50,000-$99,999	246 (0.82%)	9318 (30%)
≥$100,000	86 (0.27%)	4173 (15%)
Family history of suicide behavior (attempt or death)
Yes	149 (0.42%)	1176 (3.5%)	100.14	< 0.001
No	1116 (3%)	31991 (93%)
Family history of suicide death
Yes	37 (0.12%)	623 (1.9%)	6.59	0.0126
No	1224 (3.3%)	32492 (95%)
Family history of suicide attempt
Yes	142 (0.4%)	1107 (3.3%)	107.95	< 0.001
No	1109 (3%)	31977 (93%)
Ever in active combat
Yes	38 (0.1%)	1400 (4.5%)	4.2091	0.04427
No	1226 (3.3%)	31746 (92%)
Sexually assaulted
Yes	612 (1.6%)	2710 (7.1%)	809.03	< 0.001
No	651 (1.8%)	30394 (89%)
Physically abused by partner
Yes	429 (1%)	2139 (5.1%)	589.07	< 0.001
No	836 (2.4%)	30978 (92%)
Neglected by parent
Yes	223 (0.57%)	967 (2.5%)	774.28	< 0.001
No	1035 (2.5%)	32152 (94%)
History of physical abuse
Yes	280 (0.77%)	1037 (2.7%)	583.20	< 0.001
No	983 (2.6%)	32088 (94%)
Mental illness status
Yes	1075 (2.9%)	13475 (39%)	845.17	< 0.001
No	190 (0.52%)	19692 (57%)
Internalizing disorder
Yes	852 (2.2%)	7546 (21%)	847.29	< 0.001
No	413 (1.2%)	25621 (76%)
Personality disorder
Yes	721 (1.9%)	4230 (12%)	1050.07	< 0.001
No	544 (1.5%)	28937 (85%)
Substance abuse disorder
Yes	610 (1.7%)	7097 (22%)	186.70	< 0.001
No	655 (1.7%)	26070 (75%)
Self-perceived current health
Excellent or very good	471 (1.3%)	18608 (57%)	53.50	< 0.001
Good	364 (1%)	8859 (25%)
Fair	277 (0.68%)	4260 (11%)
Poor	153 (0.36%)	1430 (3.5%)
Body mass index (mean, SE*)	28.93 (0.29)	27.53 (0.04)	23.69	< 0.001
Body mass index categories
Normal weight (18.5–24.9)	378 (1.1%)	11143 (34%)	8.54	< 0.001
Underweight (< 18.5)	21 (0.05%)	475 (1.3%)
Overweight (25-29.9)	364 (0.96%)	11838 (35%)
Obese class I (30.0-34.9)	263 (0.7%)	5906 (17%)
Obese class II (35.0-39.9)	117 (0.31%)	2212 (5.9%)
Obese class III ≥ 40.0	122 (0.29%)	1593 (4.2%)
Obesity index
Non-obese (BMI < 30)	763 (2.1%)	23456 (70%)	38.11	< 0.001
Obese (BMI ≥ 30)	502 (1.3%)	9711 (27%)
^a Complete case analysis.
^b P values derived from t-test, chi-square test, and Fisher’s exact test, as applicable.
* SE = standard error.

Table 3

Association between sociodemographic characteristics, mental illness, traumatic events, and obesity by recent suicide attempt (n = 241)a
	Recently attempted suicide (n = 241) n(%)	Not recently attempted suicide (n = 7,147) n(%)	X², t-test or F test (Design based)
Age (mean, SE*)	38.15(1.37)	46.65(0.3)	37.81	< 0.001
Age
Young adults (18-35yrs)	89 (1.3%)	1896 (28%)	8.75	< 0.001
Middle-aged adults (36-55yrs)	126 (1.5%)	3157 (42%)
Older adults (> 55yrs)	26 (0.25%)	2094 (27%)
Sex
Male	73 (1%)	2206 (35%)	0.21	0.65
Female	168 (2.1%)	4941 (62%)
Sexual identity
Heterosexual	209 (2.8%)	6884 (95%)	30.53	< 0.001
Gay or lesbian	8 (0.07)	95 (1.2%)
Bisexual	18 (0.2%)	73 (0.93%)
Race/ethnicity
Non-Hispanic White	121 (2%)	4190 (69%)	3.14	0.04
Non-Hispanic Black	36 (0.27%)	1358 (11%)
Other (Hispanic, Asian, American Indian)	84 (0.86%)	1599 (17%)
Race/ethnicity
Non-Hispanic White	121(1.99%)	4190 (69.37%)	1.63	0.17
Non-Hispanic Black	36 (0.27%)	1358 (10.69%)
Non-Hispanic American Indian/Alaska Native	10 (0.17%)	158 (2.89%)
Non-Hispanic Asian/Native Hawaiian/Other	9 (0.13%)	156 (3.29%)
Hispanic, any race	65 (0.55%)	1285 (10.65%)
Household income level
<$25,000	132 (1.5%)	2812 (34%)	2.41	0.08
$25,000-$49,999	58 (0.75%)	2025 (28%)
$50,000-$99,999	38 (0.72%)	1647 (25%)
≥$100,000	13 (0.01%)	663 (11%)
Family history of suicide behavior (attempt or death)
Yes	30 (0.36%)	379 (5.2%)	6.24	0.015
No	211 (2.8%)	6768 (92%)
Family history of suicide death
Yes	6 (0.09%)	153 (2.2%)	0.06	0.8
No	232 (3%)	6963 (95%)
Family history of suicide attempt
Yes	30 (0.36%)	360 (5%)	7.6	< 0.01
No	206 (2.7%)	6739 (92%)
Ever in active combat
Yes	9 (0.1%)	219 (3.3%)	0.0011	0.974
No	231 (3%)	6915 (94%)
Sexually assaulted
Yes	108 (1.4%)	1326 (17%)	35.14	< 0.001
No	132 (1.7%)	5790 (80%)
Physically abused by partner
Yes	82 (0.92%)	1023 (12%)	23.85	< 0.001
No	158 (2.2%)	6100 (85%)
Neglected by parent
Yes	50 (0.58%)	498 (6.2%)	46.755	< 0.001
No	189 (2.5%)	6623 (93%)
History of physical abuse
Yes	58 (0.71%)	505 (6.3%)	50.58	< 0.001
No	182 (2.4%)	6617 (91%)
Mental illness status
Yes	236 (3%)	5345 (73%)	5.71	0.01
No	5 (0.075%)	1802 (24%)
Internalizing disorder
Yes	223 (2.9%)	4574 (62%)	74.59	< 0.001
No	18 (0.25%)	2573 (35%)
Personality disorder
Yes	164 (2.1%)	2031 (26%)	60.25	< 0.001
No	77 (0.99%)	5116 (70%)
Substance abuse disorder
Yes	144(2%)	2230(32%)	26.24	< 0.001
No	97(1.2%)	4917(65%)
Self-perceived current health
Excellent or very good	58 (0.8%)	3037 (44%)	6.89	< 0.001
Good	72 (0.98%)	2041 (28%)
Fair	62 (0.7%)	1368 (17%)
Poor	48 (0.64%)	700 (8.4%)
Body mass index (mean, SE*)	29.06 (0.81)	28.06 (0.11)	1.56	0.21
Body mass index categories
Normal weight (18.5–24.9)	72 (0.96%)	2327 (34%)	0.53	0.69
Underweight (< 18.5)	5 (0.05%)	121 (1.5%)
Overweight (25-29.9)	69 (0.86%)	2258 (30%)
Obese class I (30.0-34.9)	49 (0.68%)	1371 (18%)
Obese class II (35.0-39.9)	24 (0.33%)	584 (7.4%)
Obese class III ≥ 40.0	22 (0.25%)	486 (5.8%)
Obesity index
Non-obese (BMI < 30)	146 (1.9%)	4706 (66%)	2.88	0.09
Obese (BMI ≥ 30)	95 (1.3%)	2441 (31%)
^a Complete case analysis.
^b P values derived from t-test, chi-square test, and Fisher’s exact test, as applicable.
* SE = standard error.

Multivariate analysis

Tables 4 and 5 show multivariable logistic regression for association between current obesity status and ever-attempt and recent attempt. We evaluated two types of models, one with obesity as binary variable (BMI 30 kg/m² as cutoff) and the other with obesity as an ordinal variable using WHO weight categories. For ever-attempted suicide (Table 4), there was a statistically significant association with the current obesity status, with obese adults having POR of 1.43 (95% CI, 1.20–1.69) times higher odds of having ever attempted suicide, compared with non-obese adults, after adjusting for all significant covariates identified in bivariate analysis above. Other covariates were also independently associated, with significantly higher odds in young and middle age groups (vs. older age), female sex, bisexuality, race, and history of abuse and mental illness. When current-obesity status was subdivided into WHO weight categories (Model 1), all three obesity classes (I, II, and III) had significantly higher POR of 1.39, 1.37, and 1.55, respectively.

For recent suicide attempt (Table 5), no statistically significant association with the current obesity status was found with POR of 1.40 (95% CI, 0.90–2.18), whereas young and middle age groups (vs. older age), bisexuality, and mental illnesses were independently associated with higher odds of recent attempts (Model 2). Subdividing the obesity status into WHO weight categories (Model 1) did not significantly change the results.

Table 4

Odds of ever-attempted suicide by demographic characteristics and body mass index category (n=34,037)^a
	Model 1^b		Model 2 (Fully Adjusted Model)^c
	POR^d (95% CI)	P value	POR^d (95% CI)	P value
Obesity Index
Non-obese (BMI < 30)			ref
Obese (BMI ≥ 30)			1.43 (1.20–1.69)	<0.0001
Age
Young adults (18-35yrs)	1.58 (1.21–2.06)	0.001	1.58 (1.22–2.07)	0.001
Middle-aged adults (36-55yrs)	1.42 (1.10–1.81)	0.007	1.42 (1.11–1.82)	0.005
Older adults (> 55yrs)	ref		ref
Sex
Male	0.78 (0.62–0.98)	0.033	0.78 (0.62–0.98)	0.031
Female	ref		ref
Sexual identity
Heterosexual	ref		ref
Gay or lesbian	1.34 (0.70–2.54)	0.374	1.34 (0.69–2.58)	0.377
Bisexual	2.59 (1.54–4.36)	0.001	2.61 (1.55–4.39)	<0.0001
Race/ethnicity
Non-Hispanic White	ref		ref
Non-Hispanic Black	0.60 (0.46–0.79)	0.001	0.61 (0.46–0.79)	0.001
Non-Hispanic American Indian/Alaska Native	0.91 (0.37–2.20)	0.838	0.91 (0.38–2.22)	0.839
Non-Hispanic Asian/Native Hawaiian/Other Pacific Islander	0.97 (0.78–1.20)	0.799	0.97 (0.78–1.19)	0.745
Hispanic, any race	0.89 (0.77–1.03)	0.104	0.88 (0.76–1.02)	0.107
Household income level
<$25,000	1.47 (1.11–1.96)	0.008	1.47 (1.11–1.96)	0.009
$25,000-$49,999	1.15 (0.86–1.52)	0.339	1.14 (0.86–1.51)	0.345
$50,000-$99,999	1.18 (0.90–1.53)	0.22	1.17 (0.90–1.53)	0.224
≥$100,000	ref		ref
Family history of suicide behavior (attempt or death)
Yes	1.77 (1.19–2.61)	0.005	1.77 (1.19–2.61)	0.005
No	ref		ref
Sexually assaulted
Yes	3.75 (2.93–4.79)	0.005	3.75 (2.93–4.78)	<0.0001
No	ref		ref
Physically abused by partner
Yes	1.79 (1.33–2.42)	< 0.0001	1.79 (1.33–2.42)	<0.0001
No	ref		ref
Neglected by parent
Yes	1.27 (0.92–1.77)	0.145	1.27 (0.92–1.76)	0.146
No	ref		ref
History of physical abuse
Yes	2.32 (1.75–3.08)	< 0.0001	2.32 (1.74–3.07)	<0.0001
No	ref		ref
Internalizing disorder
Yes	2.03 (1.62–2.52)	< 0.0001	2.03 (1.75–3.07)	<0.0001
No	ref		ref
Personality disorder
Yes	3.69 (2.99–4.55)	< 0.0001	3.68 (2.98–4.55)	<0.0001
No	ref		ref
Substance abuse disorder
Yes	1.61 (1.27–2.03)	< 0.0001	1.60 (1.27–2.03)	<0.0001
No	ref		ref
Self-perceived current health
Excellent or very good	ref		ref
Good	1.32 (1.05–1.65)	< 0.0001	1.32 (1.05–1.66)	0.019
Fair	1.59 (1.21–2.11)	< 0.0001	1.61 (1.22–2.12)	0.001
Poor	2.34 (1.57–3.47)	< 0.0001	2.35 (1.59–3.48)	<0.0001
Body mass index categories
Normal weight (18.5–24.9)	ref
Underweight (< 18.5)	0.78 (0.40–1.49)	0.442
Overweight (25-29.9)	0.99 (0.79–1.25)	0.994
Obese class I (30.0-34.9)	1.39 (1.05–1.84)	0.021
Obese class II (35.0-39.9)	1.37 (0.99–1.90)	0.061
Obese class III ≥ 40.0	1.55 (1.06–2.28)	0.026
^a Complete case analysis
^b Model 1: BMI categories as main exposure, adjusted for age, sex, sexual identity, race/ethnicity, household income, family history of suicide attempt or death, sexually assaulted, abused by partner, neglected, abused by parent, internalizing disorder, personality disorder, substance abuse disorder, self-perceived current health. ^c Model 2: Full model, obesity index as main exposure, adjusted for age, sex, sexual identity, race/ethnicity, household income, family history of suicide attempt or death, sexually assaulted, abused by partner, neglected, abused by parent, internalizing disorder, personality disorder, substance abuse disorder, self-perceived current health. ^d POR: Prevalence odds ratio from logistic regression.

Table 5

Odds of a recent suicide attempt by descriptive characteristics and body mass index category (n=7,285)^a
	Model 1^b			Model 2 (Full Model)^c
	POR^d (95% CI)		P value	POR^d (95% CI)	P value
Obesity index
Non-obese (BMI < 30)				ref
Obese (BMI ≥ 30)				1.40 (0.90–2.18)	0.13
Covariates Age
Young adults (18-35yrs)	3.68 (1.52–8.93)	0.005		3.69 (1.54–8.86)	0.004
Middle-aged adults (36-55yrs)	2.91 (1.20–7.03)	0.018		2.93 (1.22–7.04)	0.017
Older adults (> 55yrs)	ref			ref
Sex
Male
Female
Sexual identity
Heterosexual	ref			ref
Gay or Lesbian	1.14 (0.45–2.86)	0.778		1.15 (0.47–2.81)	0.751
Bisexual	4.29 (2.16–8.52)	< 0.0001		4.26 (2.17–8.34)	< 0.0001
Race/ethnicity Non-Hispanic White	ref			ref
Non-Hispanic Black	0.69 (0.29–1.58)	0.372		0.69 (0.29–1.58)	0.373
Non-Hispanic American Indian/Alaska Native	1.59 (0.37–6.85)	0.529		1.57 (0.34–7.11)	0.55
Non-Hispanic Asian/Native Hawaiian/Other Pacific Islander	1.75 (0.61–5.02)	0.292		1.74 (0.60–5.02)	0.297
Hispanic, any race	1.72 (1.22–2.43)	0.003		1.71 (1.20–2.44)	0.003
Internalizing disorder
Yes	3.27(2.00-5.32)	< 0.0001		3.26 (2.00-5.29)	< 0.0001
No	ref			ref
Personality disorder
Yes	3.24 (1.92–5.47)	< 0.0001		3.23 (1.92–5.43)	< 0.0001
No	ref			ref
Substance abuse disorder
Yes	2.11 (1.31–3.37)	< 0.0001		2.12 (1.32–3.38)	< 0.0001
No	ref			ref
Self-perceived current health
Excellent or very good	ref			ref
Good	1.80 (0.99–3.29)	0.05		1.78 (0.97–3.26)	0.05
Fair	2.01 (1.06–3.79)	0.03		1.98 (1.05–3.73)	0.03
Poor	4.53 (2.39–8.58)	< 0.0001		4.53 (2.39–8.58)	< 0.0001
Body mass index categories
Normal weight (18.5–24.9)	ref
Underweight (< 18.5)	0.88 (0.09–7.85)	0.905
Overweight (25-29.9)	0.92 (0.42–1.96)	0.825
Obese class I (30.0-34.9)	1.39 (0.67–2.90)	0.364
Obese class II (35.0-39.9)	1.33 (0.65–2.71)	0.415
Obese class III ≥ 40.0	1.21 (0.51–2.84)	0.648
^a Complete case analysis.
^b Model 1: BMI categories as main exposure, adjusted for age, sexual identity, race/ethnicity, internalizing disorder, personality disorder, substance abuse disorder, self-perceived current health.
^c Model 2: Full model, obesity index as main exposure adjusted for age, sexual identity, race/ethnicity, internalizing disorder, personality disorder, substance abuse disorder, self-perceived current health.
^d POR: Prevalence odds ratio from logistic regression.

Effect modification

The changes in goodness-of-fit after adding 2-, 3-, and 4-way interaction terms to the multivariable models are shown in Supplemental Table 2. The adjusted prevalence odds of ever-attempted suicide among middle-aged (age 36–55 years) currently obese adults was 0.54 (95% CI, 0.31–0.92) times lower than older (age > 55 years) currently obese adults. The adjusted prevalence odds of ever-attempted suicide was 2.96 (95% CI, 1.40–6.25) times higher among currently obese Asian/Native Hawaiian/Other Pacific Islanders than currently obese non-Hispanic Whites. No statistically significant effect modification was found between current obesity and age, sex, race/ethnicity in relation to recent suicide attempt.

Mediation analysis

Supplemental Table 3 shows the strengths of the associations along the mediation pathways shown in Fig. 1. Mental illness was found to partially mediate the effect of current-obesity on ever-attempted suicide, with a RIT of 0.149, meaning that 15% of the obesity effect on ever-attempted suicide was mediated by mental illness. Similar statistically significant mediation effects were found for each mental illness categories: internalizing disorder (RIT = 0.278), personality disorder (RIT = 0.190), and substance abuse disorder (RIT = 0.067). Mediation analysis results for recent suicide attempts is shown in Supplemental Table 4.

Sensitivity analysis

Sensitivity analysis was performed using surrogate outcome (recent suicide ideation, recent thoughts about death). We found no statistically significant association between current-obesity and recent suicide ideation or recent thoughts about death, after adjusting for demographic and mental health covariates. There was no statistically significant association with any suicide attempts when BMI was treated as a continuous variable. When age was included in the model as a continuous variable, a weak negative association was observed for ever- and recent-suicide attempt with POR of 0.98 (95% CI, 0.97–0.99) and 0.97 (95% CI, 0.95–0.98), respectively, for each year. When age was categorized as quartiles (≤ 35, 35–47, 47–61, 61–90 years), a similar association with suicide behavior was found, with younger quartiles having higher odds of suicide behaviors than the oldest quartile as the reference. In particular, the two youngest quartiles had a POR of 4.80 (95% CI, 1.55–14.82) and 4.55 (95% CI, 1.39–14.88) of recent suicide attempt, respectively. When a different mediation analysis method (Zhao approach) was used, mental illnesses were similarly found to have statistically significant partial mediating effects on ever-attempted suicide.

The incidence of suicide in U.S. adults has been steadily increasing over the past two decades, during which obesity has also declared as a major public health challenge. Results here showed that adults with current obesity are over 40% more likely than adults with normal weight to have ever-attempted suicide, after adjusting for age, sex, race, and mental illnesses. The strength of this association also increased with the obesity severity per WHO obesity classes (I, II, III). Causality of this relationship, however, remains unclear in this cross-sectional study, as temporality of obesity onset and suicide attempt cannot be established. It is possible that the suicide attempts earlier in life could have predated weight gain to BMI > 30 kg/m².

The association between current obesity and recent suicide attempt within past 2-years was similar in effect size but did not reach statistical significance. However, we note that the vast majority of the participants (78.1%) did not respond to the recent-attempt questionnaire, as opposed to only 0.64% non-response to the ever-attempt questionnaire. This raises the concern regarding the representativeness of the sample with respect to recent attempts. Also, the small number of recent attempt (n = 221) compared ever attempt (n = 1,265), suggests that the sample may also be underpowered to draw conclusion regarding recent attempts.

The Wave 2 NESARC cohort allowed the inclusion of important potential demographic, mental illness, and psychosocial confounding variables for analysis which have not been available in prior similar analyses. Consistent with previous studies[26–28], younger age, female sex, mental illnesses, and other psychosocial factors were significantly associated with lifetime history of suicide attempts and that the obesity effect on ever-attempted suicide is also modified by age, sex, and ethnicity. Our findings unfortunately add further to the complexity of suicidality prediction. Extensive previous research has failed to develop sufficiently accurate prediction model, even based on a panel of various suicide risk factors [29] and our study suggests a yet another variable (obesity) to be considered in suicide prediction.

This study is the first study to evaluate mediators of the obesity effect on suicide behaviors. It is widely known that mental illness is associated with both suicide behaviors and obesity[30–32]. Therefore, mental illness may be an important confounder when assessing the obesity–suicide relationship. We found that only 15% of the total obesity effect on ever-attempted suicide was mediated by mental illnesses. While this result requires careful interpretation as causal relationships among obesity, mental illness, and suicide remain unclear, our finding suggests that obesity-associated mental illness alone cannot explain the observed obesity–suicide association, which in turn suggests that other non-psychiatric factors may be driving this association. For example, obesity is associated with several biological states implicated in suicidality, such as high levels of serotonin (neurotransmitter in the brain)[33], endocrine dysregulation in the hypothalamic-pituitary-adrenal axis,[34]and serum inflammatory markers such as C-reactive protein[35, 36].

Our effect modification analysis did not find significant sex differences in the obesity-suicide attempt association. However, we found that being middle-aged weakened this association and being Asian/Native Hawaiian/Other Pacific Islander strengthened this association. These results however underscore the complexity of the obesity-suicide relationship and prompts the need for further studies in populations of different age, ethnic, and cultural compositions.

There are several limitations of this study. As previously mentioned, temporality of the development of obesity and suicide attempt cannot be determined by this cross-sectional study, and the causality cannot be inferred. It also remains unclear whether current obesity is associated with current or future suicide attempt risk. As a secondary analysis, this study is also limited by the specific design issues of the parent study. First, Wave 2 NESARC was a survey of living individuals and did not include individuals deceased by suicides. Those surveyed who attempted suicide were therefore survivors of prior suicide attempts and may represent a biased sample of those with suicide behaviors. Although data are nationally representative, they were drawn from a household-based survey. Those adults not living in a household (e.g., homeless, institutionalized adults) may be at a higher risk for suicide than those represented in these surveys. Only self-reported obesity and mental illnesses were available in NESARC, thus under-reporting is a concern. In fact, we observed that the majority the participants did not respond on the recent suicide attempt questionnaire. The rate of under-reporting may differ by age, sex, and race because the social and cultural stigma may differ between generations and racial groups.

In summary, in a nationally representative Wave 2 NESARC adult cohort, we found a statistically significant association between obesity and lifetime history of suicide attempt, after adjusting for confounders including age, sex, and mental illnesses. Further studies may be needed to establish temporality and causality of this association.

Research Ethics Approval

The UT Southwestern Human Research Protection Program has determined that secondary analyses of these de-identified data are exempt from human subjects review.

Conflict of Interest

No funds, grants, or other support was received. All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or non-financial interest in the subject matter or materials discussed in this manuscript.

Contributors

E. Iwatate conceptualized the study, obtained the dataset, conducted data analyses, drafted the manuscript. F. Atem assisted E. Iwatate on data analyses, the Methods section and the Results section. E. Jones, J. Hughes, and T. Yokoo assisted in writing the Introduction and Discussion sections of the manuscript. S. Messiah supervised and worked with E. Iwatate on all aspects of manuscript development. All authors reviewed and approved the final version for submission.

"Centers for Disease Control and Prevention. Leading causes of death reports, 1999–2019." https://wisqars.cdc.gov/fatal-leading (accessed 05/01/2022.)
H. Hedegaard, S. C. Curtin, and M. Warner, "Increase in Suicide Mortality in the United States, 1999–2018," NCHS data brief, no. 362, pp. 1–8, Apr 2020. [Online]. Available: https://www.ncbi.nlm.nih.gov/pubmed/32487287.
"Centers for Disease Control and Prevention, National Center for Injury Prevention and Control. Web-based Injury Statistics Query and Reporting System (WISQARS)." (accessed 05/01/2022.
L. Bradvik, "Suicide Risk and Mental Disorders," Int J Environ Res Public Health, vol. 15, no. 9, Sep 17 2018, doi: 10.3390/ijerph15092028.
S. Bachmann, "Epidemiology of Suicide and the Psychiatric Perspective," Int J Environ Res Public Health, vol. 15, no. 7, Jul 6 2018, doi: 10.3390/ijerph15071425.
G. Arsenault-Lapierre, C. Kim, and G. Turecki, "Psychiatric diagnoses in 3275 suicides: a meta-analysis," BMC Psychiatry, vol. 4, p. 37, Nov 4 2004, doi: 10.1186/1471-244X-4-37.
"Mental Health America. 2019 Report on Mental Health in America." https://mhanational.org/issues/state-mental-health-america. (accessed 05/01/2022.
S. T. Stone DM, Fowler KA, et al., "Vital Signs: Trends in State Suicide Rates - United States, 1999–2016 and Circumstances Contributiing to Suicide – 27 States, 2015," MMWR Morb Mortal Wkly Rep 2018;67:617–624. doi: http://dx.doi.org/10.15585/mmwr.mm6722a1, 2018. [Online]. Available: https://www.cdc.gov/mmwr/volumes/67/wr/mm6722a1.htm#References.
C. M. Hales, C. D. Fryar, M. D. Carroll, D. S. Freedman, and C. L. Ogden, "Trends in Obesity and Severe Obesity Prevalence in US Youth and Adults by Sex and Age, 2007–2008 to 2015–2016," JAMA, vol. 319, no. 16, pp. 1723–1725, Apr 24 2018, doi: 10.1001/jama.2018.3060.
M. H. Zeller, J. Reiter-Purtill, T. M. Jenkins, and M. B. Ratcliff, "Adolescent suicidal behavior across the excess weight status spectrum," Obesity (Silver Spring), vol. 21, no. 5, pp. 1039–45, May 2013, doi: 10.1002/oby.20084.
D. K. Eaton, R. Lowry, N. D. Brener, D. A. Galuska, and A. E. Crosby, "Associations of body mass index and perceived weight with suicide ideation and suicide attempts among US high school students," (in English), Archives of pediatrics & adolescent medicine, vol. 159, no. 6, pp. 513–519, Jun 2005, doi: DOI 10.1001/archpedi.159.6.513.
T. Minor, M. M. Ali, and J. A. Rizzo, "Body Weight and Suicidal Behavior in Adolescent Females: The Role of Self-Perceptions," J Ment Health Policy Econ, vol. 19, no. 1, pp. 21–31, Mar 2016. [Online]. Available: https://www.ncbi.nlm.nih.gov/pubmed/27084791.
S. Stack and D. Lester, "Body mass and suicide risk," Crisis, vol. 28, no. 1, pp. 46 – 7, 2007, doi: 10.1027/0227-5910.28.1.46.
G. R. Dutton, L. P. Bodell, A. R. Smith, and T. E. Joiner, "Examination of the relationship between obesity and suicidal ideation," Int J Obes (Lond), vol. 37, no. 9, pp. 1282–6, Sep 2013, doi: 10.1038/ijo.2012.224.
K. J. Mukamal and M. Miller, "BMI and risk factors for suicide: why is BMI inversely related to suicide?," Obesity, vol. 17, no. 3, pp. 532-8, Mar 2009, doi: 10.1038/oby.2008.538.
M. S. Kaplan, B. H. McFarland, and N. Huguet, "The relationship of body weight to suicide risk among men and women: results from the US National Health Interview Survey Linked Mortality File," The Journal of nervous and mental disease, vol. 195, no. 11, pp. 948 – 51, Nov 2007, doi: 10.1097/NMD.0b013e3181594833.
K. J. Mukamal, E. B. Rimm, I. Kawachi, E. J. O'Reilly, E. E. Calle, and M. Miller, "Body mass index and risk of suicide among one million US adults," Epidemiology, vol. 21, no. 1, pp. 82 – 6, Jan 2010, doi: 10.1097/EDE.0b013e3181c1fa2d.
B. F. Grant, F. S. Stinson, D. A. Dawson, S. P. Chou, W. J. Ruan, and R. P. Pickering, "Co-occurrence of 12-month alcohol and drug use disorders and personality disorders in the United States: results from the National Epidemiologic Survey on Alcohol and Related Conditions," Arch Gen Psychiatry, vol. 61, no. 4, pp. 361–8, Apr 2004, doi: 10.1001/archpsyc.61.4.361.
C. L. Ogden, M. D. Carroll, L. R. Curtin, M. A. McDowell, C. J. Tabak, and K. M. Flegal, "Prevalence of overweight and obesity in the United States, 1999–2004," JAMA, vol. 295, no. 13, pp. 1549-55, Apr 5 2006, doi: 10.1001/jama.295.13.1549.
E. Baca-Garcia et al., "Suicidal ideation and suicide attempts in the United States: 1991–1992 and 2001–2002," Mol Psychiatry, vol. 15, no. 3, pp. 250-9, Mar 2010, doi: 10.1038/mp.2008.98.
D. L. Streiner and G. R. Norman, "Mine is bigger than yours: measures of effect size in research," Chest, vol. 141, no. 3, pp. 595–598, Mar 2012, doi: 10.1378/chest.11-2473.
Z. Bursac, C. H. Gauss, D. K. Williams, and D. W. Hosmer, "Purposeful selection of variables in logistic regression," Source Code Biol Med, vol. 3, p. 17, Dec 16 2008, doi: 10.1186/1751-0473-3-17.
D. Iacobucci, "Mediation analysis and categorical variables: The Final Frontier," Journal of Consumer Psychology, vol. 22, no. 4, pp. 582–594, 2012.
M. E. Sobel, "Asymptotic Confidence Intervals for Indirect Effects in Structural Equation Models," Sociological Methodology, vol. 13, pp. 290–312, 1982, doi: https://doi.org/10.2307/270723.
X. Zhao, et al, "Reconsidering Baron and Kenny: Myths and Truths about Mediation Analysis," Journal of Consumer Research, vol. 37, pp. 197–206, 2010.
M. K. Nock et al., "Cross-national prevalence and risk factors for suicidal ideation, plans and attempts," Br J Psychiatry, vol. 192, no. 2, pp. 98–105, Feb 2008, doi: 10.1192/bjp.bp.107.040113.
E. K. Moscicki, "Gender differences in completed and attempted suicides," Ann Epidemiol, vol. 4, no. 2, pp. 152–8, Mar 1994, doi: 10.1016/1047-2797(94)90062-0.
M. K. Nock and R. C. Kessler, "Prevalence of and risk factors for suicide attempts versus suicide gestures: analysis of the National Comorbidity Survey," J Abnorm Psychol, vol. 115, no. 3, pp. 616 – 23, Aug 2006, doi: 10.1037/0021-843X.115.3.616.
J. C. Franklin et al., "Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research," Psychol Bull, vol. 143, no. 2, pp. 187–232, Feb 2017, doi: 10.1037/bul0000084.
T. M. Rajan and V. Menon, "Psychiatric disorders and obesity: A review of association studies," J Postgrad Med, vol. 63, no. 3, pp. 182–190, Jul-Sep 2017, doi: 10.4103/jpgm.JPGM_712_16.
F. S. Luppino et al., "Overweight, obesity, and depression: a systematic review and meta-analysis of longitudinal studies," Arch Gen Psychiatry, vol. 67, no. 3, pp. 220–9, Mar 2010, doi: 10.1001/archgenpsychiatry.2010.2.
C. Emmer, M. Bosnjak, and J. Mata, "The association between weight stigma and mental health: A meta-analysis," Obes Rev, vol. 21, no. 1, p. e12935, Jan 2020, doi: 10.1111/obr.12935.
R. J. Wurtman and J. J. Wurtman, "Brain serotonin, carbohydrate-craving, obesity and depression," Obes Res, vol. 3 Suppl 4, pp. 477S-480S, Nov 1995, doi: 10.1002/j.1550-8528.1995.tb00215.x.
B. B. Kahn and J. S. Flier, "Obesity and insulin resistance," J Clin Invest, vol. 106, no. 4, pp. 473 – 81, Aug 2000, doi: 10.1172/JCI10842.
H. M. Gibbs, L. Davis, X. Han, J. Clothier, L. A. Eads, and R. Caceda, "Association between C-reactive protein and suicidal behavior in an adult inpatient population," J Psychiatr Res, vol. 79, pp. 28–33, Aug 2016, doi: 10.1016/j.jpsychires.2016.04.002.
X. Chen et al., "Increased C-reactive protein concentrations were associated with suicidal behavior in patients with depressive disorders: a meta-analysis," Psychiatry Res, vol. 292, p. 113320, Oct 2020, doi: 10.1016/j.psychres.2020.113320.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Association Between Obesity and Recent and Ever Suicide Attempt Among Adults in the United States

Status:

Version 1

Abstract

Purpose

Methods

Results

Conclusion

Figures

Introduction

Methods

Study Design

Participants

Measures

Sample Size

Statistical Analysis

Sensitivity analysis

Results

Participant characteristics

Bivariate analysis

Multivariate analysis

Effect modification

Mediation analysis

Sensitivity analysis

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1