Risk Factors Associated with Suicide Attempts in U.S High School Students:
A Replication Study

Catalina Cañizares. MSc; Mark J. Macgowan. Ph.D.; Gabriel Odom. Ph.D., Th.D.

Agenda

  • Background
  • The Original Study
  • Research Aims
  • Methods
  • Results
  • Discussion

Background

  • Any self-reported thoughts of engaging in suicide-related behaviors (O’Carroll et al. 1996)
    • Considering
    • Planning suicide

Code 
library(tidyYRBS)
library(srvyr)
library(tidyverse)

data("clean_yrbs_2019")

suicide_2019 <- 
  clean_yrbs_2019 %>% 
  dplyr::select(weight, stratum, psu, suicide_considered, 
                suicide_planned, suicide_attempts, suicide_injury)

suicide_2019_srv <-
  suicide_2019 %>%
  srvyr::as_survey_design(
    ids     = psu,
    weights = weight,
    strata  = stratum,
    nest    = TRUE
  )


suicide_total <- 
  suicide_2019_srv %>%
  summarise(total = survey_total()) %>% 
  pull(total) %>% 
  scales::comma()
  
suicide_considered_2019 <- 
  suicide_2019_srv %>% 
  group_by(suicide_considered) %>%
  summarise(proportion = survey_mean()) %>% 
  dplyr::filter(suicide_considered == TRUE) %>% 
  pull(proportion) %>% 
  scales::percent()

suicide_considered_total <- 
  suicide_2019_srv %>% 
  group_by(suicide_considered) %>%
  summarise(total = survey_total()) %>% 
  dplyr::filter(suicide_considered == TRUE) %>% 
  pull(total) %>% 
  scales::comma()

suicide_planned_2019 <- 
  suicide_2019_srv %>% 
  group_by(suicide_planned) %>%
  summarise(proportion = survey_mean()) %>% 
  dplyr::filter(suicide_planned == TRUE) %>% 
  pull(proportion) %>% 
  scales::percent()

suicide_planned_total <- 
  suicide_2019_srv %>% 
  group_by(suicide_planned) %>%
  summarise(total = survey_total()) %>% 
  dplyr::filter(suicide_planned == TRUE) %>% 
  pull(total) %>% 
  scales::comma()

suicide_attempts_2019 <- 
  suicide_2019_srv %>% 
  group_by(suicide_attempts) %>%
  summarise(proportion = survey_mean()) %>% 
  dplyr::filter(suicide_attempts == TRUE) %>% 
  pull(proportion) %>% 
  scales::percent()

suicide_attempts_total <- 
  suicide_2019_srv %>% 
  group_by(suicide_attempts) %>%
  summarise(total = survey_total()) %>% 
  dplyr::filter(suicide_attempts == TRUE) %>% 
  pull(total) %>% 
  scales::comma()

  • Suicide is the third leading cause of death among 15-19 year-olds (CDC 2020)

  • According to the most recent data from the Youth Risk Behavior Survey (N = 13,677)

    • 18% (2,527) students nationwide reported suicide ideation

    • 15% (2,117) students has made a suicide plan

    • 7% (1,018) has attempted suicide at least one time in their lifetime

  • Suicide ideation and suicide attempts are the most commonly reported mental health crises among youth (Standley 2020)

Code 
library(tidyYRBS)
library(geomtextpath)
library(tidyverse)

data("hs_suicide")
data("hs_demographics")

the_data <- left_join(hs_demographics, hs_suicide)

# Weights
the_data_weights <- the_data |>
  srvyr::as_survey_design(
    ids=PSU,
    weights=weight,
    strata=stratum,
    nest = TRUE
  )

# Preparing the data for the ggplot

attempts <- the_data_weights %>%
  mutate(
    suicide_attempts = case_when(
      suicide_attempts == 0 ~ FALSE,
      suicide_attempts %in% 1:6 ~ TRUE,
      TRUE ~ NA
    )
  ) %>%
  group_by(year, Sex) %>%
  summarise(
    prevalence = mean(suicide_attempts, na.rm = TRUE),
    n = n()
  ) %>% 
  filter(!is.na(Sex))

  

ggplot(attempts, aes(year, prevalence, label = Sex, color = Sex)) +
  geom_smooth(alpha = 0.1, size = 0) +
  geom_textline(hjust = .40, size = 10) +
  scale_color_manual(values = c("#4e2d86", "#24bccb")) + 
  theme_minimal(base_size = 28) +
  theme(legend.position = "none") +
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
  scale_x_continuous(breaks = seq(1990, 2020, 2)) +
  theme(axis.text.x = element_text(angle = 90)) +
  theme(panel.background = element_rect(fill = "#f5fafc",
                                colour = "#f5fafc")) +
  theme(plot.background = element_rect(fill = "#f5fafc", colour = "#f5fafc")) +
  scale_y_continuous(lim=c(.0, .20),
                     breaks = seq(0, 1, 0.05),
                     labels = scales::percent) +
  labs(y="Suicide Attempts Prevalence", x="",
       title="Youth Prevalence of Suicide Attempts by Sex",
       caption = "Data from: YRBS, 1990-2019, tidyYRBS")

Code 
library(tidyYRBS)
library(geomtextpath)
library(tidyverse)

data("hs_suicide")
data("hs_demographics")

the_data <- left_join(hs_demographics, hs_suicide)

# Weights
the_data_weights <- the_data |>
  srvyr::as_survey_design(
    ids=PSU,
    weights=weight,
    strata=stratum,
    nest = TRUE
  )

# Preparing the data for the ggplot

planned <- the_data_weights %>%
  group_by(year, Sex) %>%
  summarise(
    prevalence = mean(suicide_planned, na.rm = TRUE),
    n = n()
  ) %>% 
  filter(!is.na(Sex))

  

ggplot(planned, aes(year, prevalence, label = Sex, color = Sex)) +
  geom_smooth(alpha = 0.1, size = 0) +
  geom_textline(hjust = .40, size = 10) +
  scale_color_manual(values = c("#4e2d86", "#24bccb")) + 
  theme_minimal(base_size = 28) +
  theme(legend.position = "none") +
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
  scale_x_continuous(breaks = seq(1990, 2020, 2)) +
  theme(axis.text.x = element_text(angle = 90)) +
  theme(panel.background = element_rect(fill = "#f5fafc",
                                colour = "#f5fafc")) +
  theme(plot.background = element_rect(fill = "#f5fafc", colour = "#f5fafc")) +
  scale_y_continuous(lim=c(.0, .20),
                     breaks = seq(0, 1, 0.05),
                     labels = scales::percent) +
  labs(y="Suicide Ideation Prevalence", x="",
       title="Youth Prevalence of Suicide Ideation by Sex",
       caption = "Data from: YRBS, 1990-2019, tidyYRBS")

Past studies

Evaluated 66 studies from 2015 to 2019

  • Internal risk factors:
    • Ineffective coping
    • Poor lifestyle
    • Disturbed sleep
  • External risk factors
    • Family history of mental health
    • Poor interactions in the family

Evaluated 67 population-based longitudinal studies

  • A history of previous suicidal thoughts and behaviors
  • Family history of mental disorders
  • Physical and psychological abuse

  • Evaluated 365 longitudinal studies of the past 50 years of research
  • Risk factors have been homogeneous over time
    • Demographic characteristics
    • Internalizing psychopathology
    • Prior history of suicide attempts
    • Externalizing psychopathology
    • Social factors

Do these risk factors remain significant over time?

The Original Study

Objective:

  • Identify risk factors associated with suicide attempts in adolescents at the school-based national level, and to determine whether these differences were gender related

Method:

  • The 2001 School-based Youth Risk Behavior Survey (YRBS)

  • The study population consisted of 13,601 high school students in grades 9–12

  • Outcome: “During the past 12 months, did you make a plan about how you would attempt suicide?”

  • Predictors: Backward stepwise regressions

  • Multivariate logistic regressions

Results

Significant predictors of suicide attempts:

  • Gender
  • Asian race/ethnicity
  • Threatened or injured with a weapon
  • Physical fights in the past 30 days
  • Being abused by boyfriend/girlfriend
  • Forced sexual intercourse
  • Being depressed every day for more than two weeks
  • Alcohol consumption
  • Using hallucinogenic drugs or inhaling chemicals to get high
  • Being offered an illegal drug at school
  • Being obese or underweight
  • Anorexic bulimic behavior

Results

Significant predictors for female

  • Threatened or injured with a weapon
  • Being depressed every day for more than two weeks
  • Smoking cigarettes
  • Alcohol consumption
  • Inhaling chemicals to get high
  • Being obese or underweight
  • Anorexic bulimic behavior
  • Asian
  • Physical fights in the past 30 days
  • Forced sexual intercourse
  • Trying to quit smoking
  • Being offered an illegal drug at school

Significant predictors for male

  • Threatened or injured with a weapon
  • Being depressed every day for more than two weeks
  • Smoking cigarettes
  • Alcohol consumption
  • Inhaling chemicals to get high
  • Being obese or underweight
  • Anorexic bulimic behavior
  • Driving while intoxicated
  • Carrying a weapon
  • Being abused by boyfriend/girlfriend

Research aims

  1. Identify if the risk factors associated with suicide attempts in adolescents at the school-based national level in 2001 are still relevant for adolescents in 2015, 2017 and 2019.

  2. Identify if the gender related (male and female) differences of risk factors reported by Bae and Colleague (2005) are still relevant for adolescents in 2015, 2017 and 2019

Method

Code 
#|eval: false
# data("hs_district")
# data("hs_demographics")
# data("hs_suicide")
# 
# # hsSuicide_df has clean demographics and suicide data
# hsSuicide_df <- left_join(hs_demographics, hs_suicide, by = "record")
# 
# # To add it to the other variables I need for the model, I have to recode
# #  the record vector from dbl to chr in the original data
# hs_district <-
#   hs_district %>%
#   mutate(record = as.character(record))
# 
# # Created an object that contains the variables I need for the analysis
# V_interest <- c("record","state.x","district.x","year.x","weight.x","stratum.x", "PSU.x", "Sex", "Grade", "race4", "q16", "q17","q22", "suicide_considered","suicide_attempts", "suicide_planned", "is_hopeless","q19", "q41", "q51", "q57", "qhallucdrug", "q67", "q10", "q32", "q39", "q12")
# 
# analysis_attempts_df <-
#   hsSuicide_df %>%
#   left_join(hs_district, by = "record") %>%
#   filter(year.x >= 2013) %>%
#   mutate(across
#   (c(q26, q16, q17, q22, q19, q41, q51, q57, qhallucdrug, q67, q10, q32, q12, Grade, race4), factor)) %>%
#   mutate(across(c(q19, q57, q26, q39), RecodeTF)) %>%
#   mutate(across(c(q51, qhallucdrug), ScalingToBinary40)) %>%
#   mutate(across(q16:q17, ScalingToBinary12)) %>%
#   mutate(across(c(q41,q32), ScalingToBinary30)) %>%
#   mutate(across(c(q10,q12), ScaleToNumber6)) %>%
#   mutate(
#     suicide_attempts = case_when(
#       suicide_attempts == 0 ~ FALSE,
#       suicide_attempts %in% 1:6 ~ TRUE,
#       TRUE ~ NA
#     )
#   ) %>%
#   mutate(
#     q22 = case_when(
#       q22 == 1 ~ FALSE,
#       q22 == 2 ~ FALSE,
#       q22 %in% 3:6 ~ TRUE,
#       TRUE ~ NA
#     )
#   ) %>%
#   mutate(
#     race4 = case_when(
#       race4 == 1 ~ "White", 
#       race4 == 2 ~ "Black or African American", 
#       race4 == 3 ~ "Hispanic/Latino", 
#       race4 == 4 ~  "All Other Races", 
#       TRUE ~ NA_character_
#     )
#   ) %>% 
#   mutate(
#     q67 = case_when(
#       q67 == 1 ~ "Very underweight",
#       q67 == 2 ~ "Slightly underweight",
#       q67 == 3 ~ "About the right weight",
#       q67 == 4 ~ "Slightly overweight",
#       q67 == 5 ~ "Very overweight"
#     )
#   ) %>%
#   select(all_of(V_interest)) %>%
#   rename(
#     Year                = year.x,
#     weight              = weight.x,
#     stratum             = stratum.x,
#     psu                 = PSU.x,
#     injured_weapon      = q16,
#     physical_fight      = q17,
#     hurt_partner        = q22,
#     forced_sexual       = q19,
#     consumed_alcohol    = q41,
#     consumed_inhalant   = q51,
#     offered_drugs       = q57,
#     used_hallucinogenic = qhallucdrug,
#     describe_weight     = q67,
#     car_drunk           = q10,
#     cigarettes          = q32,
#     quit_smoking        = q39,
#     weapon_carrying     = q12, 
#     Race                = race4
#   )
# 
# saveRDS(analysis_attempts_df, "data/clean_combined_data.rds")

  • Youth Risk Behavior Surveillance System (YRBSS)

  • Surveys that monitors health behaviors and experiences among high school students in grades 9–12 attending U.S. public and private schools since 1991 (Underwood et al. 2020)

  • Combined High School-based Youth Risk Behavior Survey (YRBS) from the CDC.

  • tidyYRBS

Code 
#|echo: false

attempts_df<- readRDS("data/clean_combined_data.rds") %>% 
 mutate(Race = factor(Race)) %>% 
  mutate(Race = relevel(Race, ref = "White")) %>%
  mutate(Sex = factor(Sex)) %>% 
  mutate(Sex = relevel(Sex, ref = "Male"))

# This function transforms the Data Frame into a survey object

# yrbs_df <-
#   attempts_df %>%
#   srvyr::as_survey_design(
#     ids     = psu,
#     weights = weight,
#     strata  = stratum,
#     nest    = TRUE)
# 
# # N weighted
# total_weight <- 
#   yrbs_df %>% 
#   summarise(N = survey_total()) %>% 
#   select(N) %>% 
#   pull() %>% 
#   comma()
# 
# saveRDS(total_weight, "data/total_weight.rds")
# 
# # Sex weighted
# 
# female <- 
#    yrbs_df  %>% 
#   group_by(Sex) %>%
#   summarise(N = survey_total()) %>% 
#   filter(Sex == "Female") %>% 
#   select(N) %>% 
#   pull() %>% 
#   comma()
# 
# saveRDS(female, "data/female.rds")
# 
# male <- 
#    yrbs_df  %>% 
#   group_by(Sex) %>%
#   summarise(N = survey_total()) %>% 
#   filter(Sex == "Male") %>% 
#   select(N) %>% 
#   pull() %>% 
#   comma()
# 
# saveRDS(male, "data/male.rds")
# 
# # Proportion of suicide attempts
# 
# suicide_attempts_df <- 
#   yrbs_df  %>% 
#   group_by(suicide_attempts) %>%
#   summarise(proportion = survey_mean(),
#             total = survey_total()) %>% 
#   dplyr::filter(suicide_attempts == TRUE) %>% 
#   pull(proportion) %>% 
#   scales::percent()
# 
# saveRDS(suicide_attempts_df, "data/suicide_attempts.rds")
# 
# # Proportion of suicide ideation
# 
# suicide_ideation_df <- 
#   yrbs_df  %>% 
#   group_by(suicide_planned) %>%
#   summarise(proportion = survey_mean(),
#             total = survey_total()) %>% 
#   dplyr::filter(suicide_planned == TRUE) %>% 
#   pull(proportion) %>% 
#   scales::percent()
# 
# saveRDS(suicide_ideation_df, "data/suicide_ideation.rds")

n_weighted <- readRDS("data/total_weight.rds")

suicide_attempt <- readRDS("data/suicide_attempts.rds")

suicide_ideation <- readRDS("data/suicide_ideation.rds")

  • The total weighted sample for the Combined YRBS High School Dataset is 4,751,549 cases since 2015 until 2019

  • The real sample used for the models is 1,602,357 due to missing data

  • From these, 817,700 are female, and 784,700 are male

  • The proportion of students who reported attempting suicide in this data is 8%

  • The proportion of students who reported thinking suicide in this data is 7%

Outcome:

(Q27) During the past 12 months, did you make a plan about how you would attempt suicide?

(Q28) During the past 12 months, how many times did you actually attempt suicide?

Predictors

  • Sex, Race

  • Being injured by a weapon in school, Physical fighting, Physical dating violence, Sexual violence, Carrying a weapon to school

  • Consuming alcohol, Inhalant drugs, Being offered drugs in school, Smoking Cigarettes

  • Driving while consuming alcohol

  • Feeling sad or hopeless

  1. Selected the predictors using Bae and Collegues (2005)

  2. Converted the data frame into a survey object

  3. Multiple Logistic regression with year as a fixed effect

Code 
# Created an object that contains the variables I need for the analysis

V_interest <- c( "q17","q22","suicide_considered","suicide_attempts", "suicide_planned", "is_hopeless","q19", "q41", "q51", "q57", "qhallucdrug", "q67", "q10", "q32", "q39", "q12")


analysis_weighted_attempts_ls <- 
  survey::svydesign(
  id = ~psu,
  weights = ~weight,
  strata = ~stratum ,
  nest = TRUE,
  survey.lonely.psu = "adjust",
  data = attempts_df
)

# Logistic model

model_1 <- (svyglm(
  suicide_planned ~
    Year + Sex + Race + `Injured by a Weapon`  + `Physical Fighting` + `Physical Dating Violence` +  `Sexual Abuse` +`Sad and Hopeless` + 
    `Alcohol Use` + `Inhalant Drugs` + `Offered Drugs in School`  +   Weight,
  family = binomial,
  design = analysis_weighted_attempts_ls
))

Results General model

Table 1. Multiple Logistic Regression Model Controlling by Sex for Suicide Ideation
Characteristic N OR1 95% CI1 p-value
4-digit Year of survey 1,602,357 1.00 0.98, 1.01 0.5
Sex
Male 784,656 — —
Female 817,700 1.32 1.21, 1.44 <0.001
Race
White 345,120 — —
All Other Races 140,098 1.32 1.14, 1.52 <0.001
Black or African American 387,014 0.99 0.88, 1.11 0.8
Hispanic/Latino 730,123 0.85 0.77, 0.94 0.001
Injured by a Weapon
FALSE 1,530,497 — —
TRUE 71,860 1.53 1.36, 1.72 <0.001
Physical Fighting
FALSE 1,295,140 — —
TRUE 307,217 1.22 1.11, 1.35 <0.001
Physical Dating Violence
FALSE 1,534,979 — —
TRUE 67,378 1.56 1.38, 1.77 <0.001
Sexual Abuse
FALSE 1,496,160 — —
TRUE 106,197 1.94 1.73, 2.17 <0.001
Sad and Hopeless
FALSE 1,125,482 — —
TRUE 476,874 6.33 5.84, 6.87 <0.001
Alcohol Use
FALSE 1,166,500 — —
TRUE 435,856 1.24 1.15, 1.33 <0.001
Inhalant Drugs
FALSE 1,515,150 — —
TRUE 87,207 1.81 1.60, 2.06 <0.001
Offered Drugs in School
FALSE 1,182,156 — —
TRUE 420,200 1.49 1.39, 1.61 <0.001
Weight
About the right weight 823,303 — —
Slightly overweight 420,467 1.41 1.29, 1.54 <0.001
Slightly underweight 214,168 1.27 1.13, 1.43 <0.001
Very overweight 87,525 2.33 2.01, 2.70 <0.001
Very underweight 56,892 2.36 1.96, 2.84 <0.001
1 OR = Odds Ratio, CI = Confidence Interval

  • Adolescents who were injured with a weapon have an odds of thinking about suicide that is 1.53 times higher compared to individuals who were not.

  • Adolescents who have been physically abused by a partner have an odds of thinking about suicide that is 1.56 times higher compared to those who have not been physically abused.

  • Adolescents who were sexually abused have an odds of thinking about suicide that is 1.94 times higher compared to adolescents who have not

  • Adolescents who feel sad or hopeless have an odds of thinking about suicide that is 6.33 times higher compared to individuals who were not.

  • Adolescents exposed to inhalant drugs have an odds of developing suicide ideation that is 1.81 times higher.

  • Adolescents that consider their body very overweight or very underweight have an odds of thinking about suicide that is 2.3 times higher than the ones who report normal weight.

Results - Comparing models

Code 
ggplot(conf_general, 
       aes(x = reorder(varnames, OR), y = (OR), 
           ymin = low, ymax = high, colour = origen)) +
  scale_color_manual(values = c("#4e2d86", "#24bccb")) + 
  geom_pointrange(alpha = 0.5) +
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
  theme(panel.background = element_rect(fill = "#f5fafc",colour = "#f5fafc")) +
  theme(legend.key = element_rect(fill = "#f5fafc")) +
  theme(plot.background = element_rect(fill = "#f5fafc", colour = "#f5fafc")) +
  labs(y = "Odd Ratio CI",
       title="Comparing Odd Ratios CI", 
       x = "",
       color = "Study") +
  geom_hline(yintercept=1, linetype="dashed") +
  coord_flip() +
 scale_y_continuous(breaks = seq(0, 7, 1),lim = c(0,7))

Results by Sex

Table 2. Logistic Regression Model for Males
Characteristic N OR1 95% CI1 p-value
4-digit Year of survey 657,203 1.01 0.98, 1.04 0.5
Race
White 146,575 — —
All Other Races 59,866 1.30 1.05, 1.61 0.015
Black or African American 144,688 0.91 0.74, 1.11 0.3
Hispanic/Latino 306,072 0.86 0.74, 1.02 0.076
Physical Fighting
FALSE 510,254 — —
TRUE 146,948 1.02 0.84, 1.24 0.8
Sexual Abuse
FALSE 633,350 — —
TRUE 23,853 2.13 1.57, 2.90 <0.001
Offered Drugs in School
FALSE 480,312 — —
TRUE 176,890 1.46 1.30, 1.64 <0.001
Injured by a Weapon
FALSE 628,958 — —
TRUE 28,245 1.89 1.48, 2.41 <0.001
Sad and Hopeless
FALSE 525,280 — —
TRUE 131,922 7.55 6.54, 8.70 <0.001
Smoking Cigarettes
FALSE 628,358 — —
TRUE 28,845 1.38 0.99, 1.91 0.058
Alcohol Use
FALSE 508,160 — —
TRUE 149,042 1.20 1.01, 1.43 0.036
Inhalant Drugs
FALSE 630,364 — —
TRUE 26,839 1.47 1.12, 1.94 0.006
Weight
About the right weight 344,361 — —
Slightly overweight 151,921 1.52 1.29, 1.80 <0.001
Slightly underweight 107,168 1.29 1.09, 1.52 0.003
Very overweight 28,528 2.30 1.72, 3.07 <0.001
Very underweight 25,224 2.25 1.61, 3.13 <0.001
Drinking and Driving 657,203 0.84 0.76, 0.92 <0.001
Carried a Weapon 657,203 1.10 1.05, 1.15 <0.001
Physical Dating Violence
FALSE 639,834 — —
TRUE 17,369 1.18 0.83, 1.67 0.4
1 OR = Odds Ratio, CI = Confidence Interval

  • Male adolescents who were sexually abused have an odds of thinking about suicide that is 2.13 times higher compared to adolescents who have not

  • Male adolescents who were injured with a weapon have an odds of thinking about suicide that is 1.89 times higher compared to individuals who were not.

  • Male adolescents who feel sad or hopeless have an odds of thinking about suicide that is 7.55 times higher compared to individuals who were not.

  • Male adolescents that consider their body very overweight or very underweight have an odds of thinking about suicide that is 2.2 times higher than the ones who report normal weight

Table 3. Logistic Regression Model for Females
Characteristic N OR1 95% CI1 p-value
4-digit Year of survey 693,369 1.00 0.98, 1.02 0.8
Race
White 136,206 — —
All Other Races 61,865 1.26 1.01, 1.56 0.039
Black or African American 165,823 0.98 0.83, 1.15 0.8
Hispanic/Latino 329,474 0.81 0.70, 0.94 0.005
Physical Fighting
FALSE 598,789 — —
TRUE 94,579 1.33 1.15, 1.54 <0.001
Sexual Abuse
FALSE 635,734 — —
TRUE 57,635 1.89 1.64, 2.17 <0.001
Offered Drugs in School
FALSE 526,570 — —
TRUE 166,798 1.53 1.35, 1.72 <0.001
Injured by a Weapon
FALSE 671,314 — —
TRUE 22,055 1.45 1.14, 1.83 0.002
Sad and Hopeless
FALSE 432,496 — —
TRUE 260,873 5.83 5.28, 6.45 <0.001
Smoking Cigarettes
FALSE 667,629 — —
TRUE 25,740 1.37 1.12, 1.69 0.002
Alcohol Use
FALSE 489,179 — —
TRUE 204,190 1.22 1.10, 1.36 <0.001
Inhalant Drugs
FALSE 653,270 — —
TRUE 40,098 1.86 1.55, 2.22 <0.001
Weight
About the right weight 347,743 — —
Slightly overweight 209,300 1.40 1.25, 1.56 <0.001
Slightly underweight 71,692 1.24 1.05, 1.47 0.012
Very overweight 46,333 2.27 1.82, 2.83 <0.001
Very underweight 18,299 2.31 1.76, 3.03 <0.001
Drinking and Driving 693,369 0.84 0.77, 0.90 <0.001
Carried a Weapon 693,369 1.14 1.08, 1.20 <0.001
Physical Dating Violence
FALSE 660,414 — —
TRUE 32,955 1.66 1.39, 1.98 <0.001
1 OR = Odds Ratio, CI = Confidence Interval

  • Female adolescents who were sexually abused have an odds of thinking about suicide that is 1.89 times higher compared to adolescents who have not

  • Female adolescents who were offered drugs in school have an odds of thinking about suicide that is 1.53 times higher compared to adolescents who have not

  • Female adolescents who feel sad or hopeless have an odds of thinking about suicide that is 5.83 times higher compared to individuals who were not.

  • Female adolescents exposed to inhalant drugs have an odds of developing suicide ideation that is 1.86 times higher.

  • Female adolescents that consider their body very overweight or very underweight have an odds of thinking about suicide that is 2.2 times higher than the ones who report normal weight

Results by Sex - Comparing models

Code 
#|echo: false
girls <- ggplot(conf_girls_df, 
       aes(x = reorder(varnames, OR), y = (OR), 
           ymin = low, ymax = high, colour = origen)) +
  scale_color_manual(values = c("#4e2d86", "#24bccb")) + 
  geom_pointrange(alpha = 0.5) +
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
  theme(panel.background = element_rect(fill = "#f5fafc",colour = "#f5fafc")) +
  theme(legend.key = element_rect(fill = "#f5fafc")) +
  theme(plot.background = element_rect(fill = "#f5fafc", colour = "#f5fafc")) +
  labs(y = "Odd Ratio CI",
       title="Comparing Odd Ratios CI for Girls", 
       x = "",
       color = "Study") +
  geom_hline(yintercept=1, linetype="dashed") +
  coord_flip() +
 scale_y_continuous(breaks = seq(0, 9, 1),lim = c(0,9)) 

boys <- ggplot(conf_boys_df, 
       aes(x = reorder(varnames, OR), y = (OR), 
           ymin = low, ymax = high, colour = origen)) +
  scale_color_manual(values = c("#4e2d86", "#24bccb")) + 
  geom_pointrange(alpha = 0.5) +
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
  theme(panel.background = element_rect(fill = "#f5fafc",colour = "#f5fafc")) +
  theme(legend.key = element_rect(fill = "#f5fafc")) +
  theme(plot.background = element_rect(fill = "#f5fafc", colour = "#f5fafc")) +
  labs(y = "Odd Ratio CI",
       title="Comparing Odd Ratios CI for Boys", 
       x = "",
       color = "Study") +
  geom_hline(yintercept=1, linetype="dashed") +
  coord_flip() +
 scale_y_continuous(breaks = seq(0, 9, 1),lim = c(0,9)) 

gridExtra::grid.arrange(girls, boys, ncol=2)

Results for Suicide Attempts

Multiple Logistic Regression Model Controlling by Sex for Suicide Attempts
Characteristic N OR1 95% CI1 p-value
4-digit Year of survey 1,197,406 1.01 0.99, 1.03 0.4
Sex
Male 577,454 — —
Female 619,951 1.50 1.31, 1.72 <0.001
Race
White 265,990 — —
All Other Races 114,133 1.54 1.30, 1.82 <0.001
Black or African American 258,808 1.45 1.23, 1.72 <0.001
Hispanic/Latino 558,473 1.23 1.08, 1.41 0.002
Physical Fighting
FALSE 987,668 — —
TRUE 209,738 1.41 1.24, 1.60 <0.001
Sexual Abuse
FALSE 1,127,573 — —
TRUE 69,832 2.13 1.86, 2.44 <0.001
Offered Drugs in School
FALSE 890,667 — —
TRUE 306,738 1.17 1.04, 1.30 0.006
Injured by a Weapon
FALSE 1,153,404 — —
TRUE 44,002 1.96 1.57, 2.43 <0.001
Sad and Hopeless
FALSE 840,458 — —
TRUE 356,947 5.01 4.45, 5.64 <0.001
Smoking Cigarettes
FALSE 1,148,171 — —
TRUE 49,234 1.56 1.28, 1.90 <0.001
Alcohol Use
FALSE 877,788 — —
TRUE 319,617 1.19 1.05, 1.35 0.007
Inhalant Drugs
FALSE 1,136,861 — —
TRUE 60,545 1.79 1.47, 2.18 <0.001
Weight
About the right weight 607,059 — —
Slightly overweight 326,860 1.15 1.04, 1.27 0.007
Slightly underweight 159,272 1.05 0.90, 1.22 0.6
Very overweight 68,504 1.70 1.35, 2.13 <0.001
Very underweight 35,709 1.44 1.12, 1.85 0.005
Drinking and Driving 1,197,406 0.94 0.88, 1.01 0.11
Carried a Weapon 1,197,406 1.07 1.03, 1.11 <0.001
Physical Dating Violence
FALSE 1,153,273 — —
TRUE 44,132 1.76 1.47, 2.12 <0.001
1 OR = Odds Ratio, CI = Confidence Interval

  • Adolescents who were sexually abused have an odds of thinking about suicide that is 2.13 times higher compared to adolescents who have not

  • Adolescents who were injured with a weapon have an odds of thinking about suicide that is 1.96 times higher compared to individuals who were not.

  • Adolescents who feel sad or hopeless have an odds of thinking about suicide that is 5.01 times higher compared to individuals who were not.

  • Adolescents exposed to smoking have an odds of developing suicide ideation that is 1.56 times higher.

  • Adolescents exposed to inhalant drugs have an odds of developing suicide ideation that is 1.79 times higher.

  • Adolescents who have been physically abused by a partner have an odds of thinking about suicide that is 1.76 times higher compared to those who have not been physically abused.

Conclusions

  • Similar to the results found by Bae and colleagues (2005), most risk factors contributed significantly to suicide ideation and attempts, with ORs within the same confidence interval.

  • The findings in this study identify persistent risk factors that should be a part of suicide prevention programs in high schools nationwide.

  • There are differences in risk factors for suicide ideation and suicide attempts.

  • There are differences in risk factors between females and males for suicide ideation

  • For males:
    • Being injured by a weapon
  • For females
    • Being offered drugs and consuming inhalant drugs

Limitations

  • The outcome of interest used by Bae and Collegues (2005) is really measuring suicide ideation not suicide attempts

  • Some of the items measured in 2001 are not measured in 2015, 2017 and 2019

    • Vomited to control weight
    • Fasting to control weight
    • Exercised to control weight

  • Amount of missing data

    • Hallucinogenic Drugs
    • Quitting smoking cigarettes

  • Dataset was not designed for measuring suicide morbidity related risk factors

Future Research

  • Fit models informed by feature selection from other statistical techniques such as LASSO (least absolute shrinkage and selection operator method)

  • Evaluate if the effect of risk factors change through the years.

  • Follow cohorts of adolescents who report suicide attempts to assess predictors of suicide completion

  • Use methods with better predictability performance and multiple interactions

References

tidyYRBS

Bae, S., R. Ye, S. Chen, P. A. Rivers, and K. P. Singh. 2005. “Risky Behaviors and Factors Associated with Suicide Attempt in Adolescents.” Journal Article. Arch Suicide Res 9 (2): 193–202. https://doi.org/10.1080/13811110590904034.
CDC. 2020. “America’s Health Rankings Analysis of CDC WONDER, Multiple Cause of Death Files, United Health Foundation.” Web Page. https://www.americashealthrankings.org/explore/health-of-women-and-children/measure/teen_suicide/state/ALL.
O’Carroll, P, A Berman, R W Maris, E K Moscicki, B L Tanney, and M M Silverman. 1996. “Beyond the Tower of Babel: A Nomenclature for Suicidology.” Journal Article. Suicide Life Threat Behav 26 (3): 237–52.
Silverman, Morton M., Alan L. Berman, Nels D. Sanddal, W. O’Carroll Patrick, and Joiner Thomas E. 2007. “Rebuilding the Tower of Babel: A Revised Nomenclature for the Study of Suicide and Suicidal Behaviors: Part 2: Suicide-Related Ideations, Communications, and Behaviors.” Suicide & Life - Threatening Behavior 37 (3): 264–77. https://www.proquest.com/scholarly-journals/rebuilding-tower-babel-revised-nomenclature-study/docview/224871202/se-2.
Standley, Corbin. 2020. “Expanding Our Paradigms: Intersectional and Socioecological Approaches to Suicide Prevention.” Death Studies 46 (February). https://doi.org/10.1080/07481187.2020.1725934.
Underwood, Michael, Nancy Brener, Jemekia Thornton, William A. Harris, Leah N. Bryan, Shari L. Shanklin, Nicholas Deputy, et al. 2020. “Overview and Methods for the Youth Risk Behavior Surveillance System — United States, 2019.” Journal Article. MMWR, 1–10. https://doi.org/ http://dx.doi.org/10.15585/mmwr.su6901a1external i.