Depression by sociodemographic characteristcs

Youths (12 - 17 years old)

  1. Youth depression by age category:
data %>% 
  select(catag6, ymdeyr) %>% 
  drop_na(ymdeyr) %>% 
  mutate(ymdeyr = ifelse(ymdeyr == 2, 0, 1),
         catag6 = recode(catag6, "1" = "12-17 years old",
                         "2" = "18-25 years old",
                         "3" = "26-34 years old",
                         "4" = "35-49 years old",
                         "5" = "50-64 years old",
                         "6" = "65+ years old"
                         )) %>% 
  group_by(catag6) %>% 
  summarise(
    prevalence_of_past_year_depression = sum(ymdeyr, na.rm = TRUE) / n() * 100) %>% 
  knitr::kable(digits = 2)
catag6 prevalence_of_past_year_depression
12-17 years old 16.2
  • The prevalence of past year major depressive episode for youth is 16.2%.
  1. Youth depression by gender:
data %>% 
  select(irsex, ymdeyr) %>% 
  drop_na(ymdeyr) %>% 
  mutate(ymdeyr = ifelse(ymdeyr == 2, 0, 1),
         irsex = recode(irsex, "1" = "Male",
                         "2" = "Female")) %>% 
  group_by(irsex) %>% 
  summarise(
    prevalence_of_past_year_depression = sum(ymdeyr, na.rm = TRUE) / n() * 100) %>% 
  knitr::kable(digits = 2)
irsex prevalence_of_past_year_depression
Female 23.62
Male 9.18
  • The prevalence of past year major depressive episode for youth is highest among female.
  • Female youth depression is more than twofold of male youth depression.
  1. Youth depression by race:
data %>% 
  select(newrace2, ymdeyr) %>% 
  drop_na(ymdeyr) %>% 
  mutate(ymdeyr = ifelse(ymdeyr == 2, 0, 1),
         newrace2 = recode(newrace2, "1" = "NonHisp White",
                         "2" = "NonHisp Black",
                         "3" = "NonHisp Native Am",
                         "4" = "NonHisp Native HI",
                         "5" = "NonHisp Asian",
                         "6" = "NonHisp more than one race",
                         "7" = "Hispanic"
                         )) %>% 
  group_by(newrace2) %>% 
  summarise(
    prevalence_of_past_year_depression = sum(ymdeyr, na.rm = TRUE) / n() * 100) %>% 
  knitr::kable(digits = 2)
newrace2 prevalence_of_past_year_depression
Hispanic 17.65
NonHisp Asian 13.91
NonHisp Black 11.61
NonHisp more than one race 20.61
NonHisp Native Am 17.09
NonHisp Native HI 13.33
NonHisp White 16.42
  • The prevalence of past year major depressive episode for youth is highest among those who identified them to have more than one race.
  • The lowest is among Non-Hispanic Black.
  1. Youth depression by family income:
data %>% 
  select(income, ymdeyr) %>% 
  drop_na(ymdeyr) %>% 
  mutate(ymdeyr = ifelse(ymdeyr == 2, 0, 1),
         income = recode(income, "1" = "Less than $20,000",
                         "2" = "$20,000 - $49,999",
                         "3" = "$50,000 - $74,999",
                         "4" = "$75,000 or More"
                         )) %>% 
  group_by(income) %>% 
  summarise(
    prevalence_of_past_year_depression = sum(ymdeyr, na.rm = TRUE) / n() * 100) %>% 
  knitr::kable(digits = 2)
income prevalence_of_past_year_depression
$20,000 - $49,999 16.03
$50,000 - $74,999 19.37
$75,000 or More 15.36
Less than $20,000 15.68
  • The prevalence of past year major depressive episode for youth is highest among those with total family income range $50,000 - $74,999.

Adults (18 + years old)

  1. Adult depression by age categories:
data %>% 
  select(catag6, amdeyr) %>% 
  drop_na(amdeyr) %>% 
  mutate(amdeyr = ifelse(amdeyr == 2, 0, 1),
         catag6 = recode(catag6, "1" = "12-17 years old",
                         "2" = "18-25 years old",
                         "3" = "26-34 years old",
                         "4" = "35-49 years old",
                         "5" = "50-64 years old",
                         "6" = "65+ years old"
                         )) %>% 
  group_by(catag6) %>% 
  summarise(
    prevalence_of_past_year_depression = sum(amdeyr, na.rm = TRUE) / n() * 100) %>% 
  knitr::kable(digits = 2)
catag6 prevalence_of_past_year_depression
18-25 years old 15.47
26-34 years old 10.99
35-49 years old 8.51
50-64 years old 6.21
65+ years old 3.43
  • The highest prevalence of past year major depressive episode for adult is among those aged 18 - 25 years old.
  • And the prevalence decreases with age.
  1. Adult depression by gender:
data %>% 
  select(irsex, amdeyr) %>% 
  drop_na(amdeyr) %>% 
  mutate(amdeyr = ifelse(amdeyr == 2, 0, 1),
         irsex = recode(irsex, "1" = "Male",
                         "2" = "Female")) %>% 
  group_by(irsex) %>% 
  summarise(
    prevalence_of_past_year_depression = sum(amdeyr, na.rm = TRUE) / n() * 100) %>% 
  knitr::kable(digits = 2)
irsex prevalence_of_past_year_depression
Female 12.99
Male 7.84
  • The prevalence of past year major depressive episode for adult is also highest among female. However, the difference is not as large as that of the youth.
  1. Adult depression by race:
data %>% 
  select(newrace2, amdeyr) %>% 
  drop_na(amdeyr) %>% 
  mutate(amdeyr = ifelse(amdeyr == 2, 0, 1),
         newrace2 = recode(newrace2, "1" = "NonHisp White",
                         "2" = "NonHisp Black",
                         "3" = "NonHisp Native Am",
                         "4" = "NonHisp Native HI",
                         "5" = "NonHisp Asian",
                         "6" = "NonHisp more than one race",
                         "7" = "Hispanic"
                         )) %>% 
  group_by(newrace2) %>% 
  summarise(
    prevalence_of_past_year_depression = sum(amdeyr, na.rm = TRUE) / n() * 100) %>% 
  knitr::kable(digits = 2)
newrace2 prevalence_of_past_year_depression
Hispanic 8.99
NonHisp Asian 6.79
NonHisp Black 7.71
NonHisp more than one race 16.20
NonHisp Native Am 11.99
NonHisp Native HI 5.86
NonHisp White 11.70
  • The prevalence of past year major depressive episode for adult is highest among those who identified them to have more than one race.
  • The lowest is among Non-Hispanic Native Hawiian.
  1. Adult depression by education: (We are not considering youth education in this case.)
data %>% 
  select(eduhighcat, amdeyr) %>% 
  drop_na(amdeyr) %>% 
  mutate(amdeyr = ifelse(amdeyr == 2, 0, 1),
         eduhighcat = recode(eduhighcat, "1" = "Less high school",
                         "2" = "High school grad",
                         "3" = "Some coll/Assoc",
                         "4" = "College graduate"
                         )) %>% 
  group_by(eduhighcat) %>% 
  summarise(
    prevalence_of_past_year_depression = sum(amdeyr, na.rm = TRUE) / n() * 100) %>% 
  knitr::kable(digits = 2)
eduhighcat prevalence_of_past_year_depression
College graduate 8.85
High school grad 10.28
Less high school 7.82
Some coll/Assoc 13.23
  • The prevalence of past year major depressive episode for adult is highest among those with some college degrees.
  1. Adult depression by family income:
data %>% 
  select(income, amdeyr) %>% 
  drop_na(amdeyr) %>% 
  mutate(amdeyr = ifelse(amdeyr == 2, 0, 1),
         income = recode(income, "1" = "Less than $20,000",
                         "2" = "$20,000 - $49,999",
                         "3" = "$50,000 - $74,999",
                         "4" = "$75,000 or More"
                         )) %>% 
  group_by(income) %>% 
  summarise(
    prevalence_of_past_year_depression = sum(amdeyr, na.rm = TRUE) / n() * 100) %>% 
  knitr::kable(digits = 2)
income prevalence_of_past_year_depression
$20,000 - $49,999 11.39
$50,000 - $74,999 10.55
$75,000 or More 8.12
Less than $20,000 14.13
  • The prevalence of past year major depressive episode for adult is highest among those with total family income range less than $20,000.
  1. Adult depression by marital status:
data %>% 
  select(irmarit, amdeyr) %>% 
  drop_na(amdeyr) %>% 
  mutate(amdeyr = ifelse(amdeyr == 2, 0, 1),
         irmarit = recode(irmarit, "1" = "Married",
                         "2" = "Widowed",
                         "3" = "Divorced or Separated",
                         "4" = "Never Been Married"
                         )) %>% 
  group_by(irmarit) %>% 
  summarise(
    prevalence_of_past_year_depression = sum(amdeyr, na.rm = TRUE) / n() * 100) %>% 
  knitr::kable(digits = 2)
irmarit prevalence_of_past_year_depression
Divorced or Separated 12.19
Married 6.51
Never Been Married 13.93
Widowed 7.67
  • The prevalence of past year major depressive episode for adult is highest among who have never been married.
  • And is lowest among those who is married.

Logistic regression and Hypothesis testing

Youths (12 - 17 years old)

  1. We perform a logistic regression on our outcome and exposure on youths while adjusting for other covariates.
    ymdeyr (youth depression) = mrjmon (Past month marijuana Use) + catag6 (age) + irsex (sex) + newrace2 (race) + coutyp4 (region)
    And the resulting odds ratio is shown below:
reg %>% 
  glm(ymdeyr ~ mrjmon + catag6 + irsex + newrace2 + coutyp4, data = ., family = binomial()) %>% 
  broom::tidy() %>% 
  mutate(OR = exp(estimate),
         conf.low = exp((estimate - 1.96 * std.error)),
         conf.high = exp((estimate + 1.96 * std.error))
         ) %>%
  filter(term == "mrjmon1") %>% 
  select(term, OR, conf.low, conf.high) %>% 
  knitr::kable(digits = 3)
term OR conf.low conf.high
mrjmon1 2.673 2.309 3.093
  • The odds of depression is 2.673 (95% CI: 2.309 - 3.093) comparing youth who used cannabis in the past month to those who do not adjusting for age, sex, race, region.
  1. Is the beta for our exposure variable truly different from zero?
    We obtained the p.value from the t.test statistics.
reg %>% 
  glm(ymdeyr ~ mrjmon + catag6 + irsex + newrace2 + coutyp4, data = ., family = binomial()) %>% 
  broom::tidy() %>% 
  filter(term == "mrjmon1") %>% 
  select(term, statistic, p.value) %>% 
  knitr::kable(digits = 4)
term statistic p.value
mrjmon1 13.1759 0
  • As we can see that the t-tests statistic for the term mrjmon1 which is marijuana past month use comparing Yes to No is very large and the resulting p.value is close to zero. Our beta coefficient of the exposure variable is significant.

Adults (18 + years old)

  1. We perform a logistic regression on our outcome and exposure on adults while adjusting for other covariates.
    amdeyr (adult depression) = mrjmon (Past month marijuana Use) + catag6 (age) + irsex (sex) + newrace2 (race) + eduhighcat (education) + coutyp4 (region) + income (income) + irmarit (martial status)
    And the resulting odds ratio is shown below:
reg %>% 
  glm(amdeyr ~ mrjmon + catag6 + irsex + newrace2 + eduhighcat + coutyp4 + income + irmarit, data = ., family = binomial()) %>% 
  broom::tidy() %>% 
  mutate(OR = exp(estimate),
         conf.low = exp((estimate - 1.96 * std.error)),
         conf.high = exp((estimate + 1.96 * std.error))
         ) %>%
  filter(term == "mrjmon1") %>% 
  select(term, OR, conf.low, conf.high) %>% 
  knitr::kable(digits = 3)
term OR conf.low conf.high
mrjmon1 1.935 1.795 2.086
  • The odds of depression is 1.935 (95% CI: 1.795 - 2.086) comparing adults who used cannabis in the past month to those who do not adjusting for age, sex, race, education, region, income, and martial status.
  1. Is the beta for our exposure variable truly different from zero?
    We obtained the p.value from the t.test statistics.
reg %>% 
  glm(amdeyr ~ mrjmon + catag6 + irsex + newrace2 + eduhighcat + coutyp4 + income + irmarit, data = ., family = binomial()) %>% 
  broom::tidy() %>% 
  filter(term == "mrjmon1") %>% 
  select(term, statistic, p.value) %>% 
  knitr::kable(digits = 4)
term statistic p.value
mrjmon1 17.2173 0
  • As we can see that the t-tests statistic for the term mrjmon1 which is marijuana past month use comparing Yes to No is very large and the resulting p.value is close to zero. Our beta coefficient of the exposure variable is significant.

Finding & Summary

  • Depression was distributed unevenly across a wide range of sociodemographic variables.
  • The prevalence of youth depression was higher among female. It is also true for adults.
  • Depression seems to be decreasing with age. The highest prevalence is among those who is youngest.
  • Depression is more prevalent for adults with family income less than $20,000 annually and who have never been married.
  • People who identified them to have more than one race have the highest prevalence of depression among all the race categories for both youth and adults.
  • Depression was significantly more common among those who reported marijuana use in the past month for both youths and adults, adjusting for demographic factors.
  • The odds of depression is 2.673 (95% CI: 2.309 - 3.093) comparing youth who used cannabis in the past month to those who do not adjusting for age, sex, race, region.
  • The odds of depression is 1.935 (95% CI: 1.795 - 2.086) comparing adults who used cannabis in the past month to those who do not adjusting for age, sex, race, education, region, income, and martial status.