National Survey on Drug Use and Health (NSDUH)

The primary survey to measure of prevalence of substance use and its correlates in the United States.

One table with one row per sampled respondent.
A complex survey designed to generalize to civilian, non-institutional americans aged 12 and older.
Released periodically since 1979 and annually since 1990.
Administered by the Substance Abuse and Mental Health Services Administration.

Please skim before you begin:

2021 National Survey on Drug Use and Health (NSDUH): Public Use File Codebook
2021 National Survey on Drug Use and Health (NSDUH): Methodological Summary and Definitions
A haiku regarding this microdata:

# drinking and thinking
# about your first time, were you
# smoking and joking?

Download, Import, Preparation

Download and import the national file:

zip_tf <- tempfile()

zip_url <-
    paste0(
        "https://www.datafiles.samhsa.gov/sites/default/files/field-uploads-protected/" ,
        "studies/NSDUH-2021/NSDUH-2021-datasets/NSDUH-2021-DS0001/" ,
        "NSDUH-2021-DS0001-bundles-with-study-info/NSDUH-2021-DS0001-bndl-data-r_v3.zip"
    )
    
download.file( zip_url , zip_tf , mode = 'wb' )

nsduh_rdata <- unzip( zip_tf , exdir = tempdir() )

nsduh_rdata_contents <- load( nsduh_rdata )

nsduh_df_name <- grep( 'PUF' , nsduh_rdata_contents , value = TRUE )

nsduh_df <- get( nsduh_df_name )

names( nsduh_df ) <- tolower( names( nsduh_df ) )

nsduh_df[ , 'one' ] <- 1

Save Locally

Save the object at any point:

# nsduh_fn <- file.path( path.expand( "~" ) , "NSDUH" , "this_file.rds" )
# saveRDS( nsduh_df , file = nsduh_fn , compress = FALSE )

Load the same object:

# nsduh_df <- readRDS( nsduh_fn )

Survey Design Definition

Construct a complex sample survey design:

library(survey)

nsduh_design <- 
    svydesign( 
        id = ~ verep , 
        strata = ~ vestr_c , 
        data = nsduh_df , 
        weights = ~ analwt_c , 
        nest = TRUE 
    )

Variable Recoding

Add new columns to the data set:

nsduh_design <- 
    update( 
        nsduh_design , 
        
        one = 1 ,
        
        health = 
            factor( 
                health , 
                levels = 1:5 , 
                labels = c( "excellent" , "very good" , "good" ,
                    "fair" , "poor" )
            ) ,
            
        age_first_cigarette = ifelse( cigtry > 99 , NA , cigtry ) ,
        
        age_tried_cocaine = ifelse( cocage > 99 , NA , cocage ) ,

        ever_used_marijuana = as.numeric( ifelse( mjever < 4 , mjever == 1 , NA ) ) ,
        
        county_type =
            factor(
                coutyp4 ,
                levels = 1:3 ,
                labels = c( "large metro" , "small metro" , "nonmetro" )
            )
            
    )

Analysis Examples with the `survey` library

Unweighted Counts

Count the unweighted number of records in the survey sample, overall and by groups:

sum( weights( nsduh_design , "sampling" ) != 0 )

svyby( ~ one , ~ county_type , nsduh_design , unwtd.count )

Weighted Counts

Count the weighted size of the generalizable population, overall and by groups:

svytotal( ~ one , nsduh_design )

svyby( ~ one , ~ county_type , nsduh_design , svytotal )

Descriptive Statistics

Calculate the mean (average) of a linear variable, overall and by groups:

svymean( ~ age_first_cigarette , nsduh_design , na.rm = TRUE )

svyby( ~ age_first_cigarette , ~ county_type , nsduh_design , svymean , na.rm = TRUE )

Calculate the distribution of a categorical variable, overall and by groups:

svymean( ~ health , nsduh_design , na.rm = TRUE )

svyby( ~ health , ~ county_type , nsduh_design , svymean , na.rm = TRUE )

Calculate the sum of a linear variable, overall and by groups:

svytotal( ~ age_first_cigarette , nsduh_design , na.rm = TRUE )

svyby( ~ age_first_cigarette , ~ county_type , nsduh_design , svytotal , na.rm = TRUE )

Calculate the weighted sum of a categorical variable, overall and by groups:

svytotal( ~ health , nsduh_design , na.rm = TRUE )

svyby( ~ health , ~ county_type , nsduh_design , svytotal , na.rm = TRUE )

Calculate the median (50th percentile) of a linear variable, overall and by groups:

svyquantile( ~ age_first_cigarette , nsduh_design , 0.5 , na.rm = TRUE )

svyby( 
    ~ age_first_cigarette , 
    ~ county_type , 
    nsduh_design , 
    svyquantile , 
    0.5 ,
    ci = TRUE , na.rm = TRUE
)

Estimate a ratio:

svyratio( 
    numerator = ~ age_first_cigarette , 
    denominator = ~ age_tried_cocaine , 
    nsduh_design ,
    na.rm = TRUE
)

Subsetting

Restrict the survey design to individuals who are pregnant:

sub_nsduh_design <- subset( nsduh_design , preg == 1 )

Calculate the mean (average) of this subset:

svymean( ~ age_first_cigarette , sub_nsduh_design , na.rm = TRUE )

Measures of Uncertainty

Extract the coefficient, standard error, confidence interval, and coefficient of variation from any descriptive statistics function result, overall and by groups:

this_result <- svymean( ~ age_first_cigarette , nsduh_design , na.rm = TRUE )

coef( this_result )
SE( this_result )
confint( this_result )
cv( this_result )

grouped_result <-
    svyby( 
        ~ age_first_cigarette , 
        ~ county_type , 
        nsduh_design , 
        svymean ,
        na.rm = TRUE 
    )
    
coef( grouped_result )
SE( grouped_result )
confint( grouped_result )
cv( grouped_result )

Calculate the degrees of freedom of any survey design object:

degf( nsduh_design )

Calculate the complex sample survey-adjusted variance of any statistic:

svyvar( ~ age_first_cigarette , nsduh_design , na.rm = TRUE )

Include the complex sample design effect in the result for a specific statistic:

# SRS without replacement
svymean( ~ age_first_cigarette , nsduh_design , na.rm = TRUE , deff = TRUE )

# SRS with replacement
svymean( ~ age_first_cigarette , nsduh_design , na.rm = TRUE , deff = "replace" )

Compute confidence intervals for proportions using methods that may be more accurate near 0 and 1. See ?svyciprop for alternatives:

svyciprop( ~ ever_used_marijuana , nsduh_design ,
    method = "likelihood" , na.rm = TRUE )

Regression Models and Tests of Association

Perform a design-based t-test:

svyttest( age_first_cigarette ~ ever_used_marijuana , nsduh_design )

Perform a chi-squared test of association for survey data:

svychisq( 
    ~ ever_used_marijuana + health , 
    nsduh_design 
)

Perform a survey-weighted generalized linear model:

glm_result <- 
    svyglm( 
        age_first_cigarette ~ ever_used_marijuana + health , 
        nsduh_design 
    )

summary( glm_result )

Replication Example

This matches the prevalence and SE of alcohol use in the past month from Codebook Table G.2:

result <- svymean( ~ alcmon , nsduh_design )

stopifnot( round( coef( result ) , 3 ) == 0.474 )
stopifnot( round( SE( result ) , 4 ) == 0.0043 )

Analysis Examples with `srvyr`

The R srvyr library calculates summary statistics from survey data, such as the mean, total or quantile using dplyr-like syntax. srvyr allows for the use of many verbs, such as summarize, group_by, and mutate, the convenience of pipe-able functions, the tidyverse style of non-standard evaluation and more consistent return types than the survey package. This vignette details the available features. As a starting point for NSDUH users, this code replicates previously-presented examples:

library(srvyr)
nsduh_srvyr_design <- as_survey( nsduh_design )