Pesquisa Nacional de Saude (PNS)

Brazil’s health survey, measuring medical conditions, risk behaviors, access to and use of care.

One consolidated table with one row per individual within each sampled household.
A complex sample survey designed to generalize to Brazil’s civilian population.
Released at approximately five year intervals starting in 2013.
Administered by Instituto Brasileiro de Geografia e Estatistica partnered with the Ministério da Saúde.

Please skim before you begin:

Conceitos e métodos
Wikipedia Entry
A haiku regarding this microdata:

# cheer the ministry!
# with each caipirinha, or
# fail sex life module

Download, Import, Preparation

Download and import the dictionary file:

dictionary_tf <- tempfile()

dictionary_url <-
    "https://ftp.ibge.gov.br/PNS/2019/Microdados/Documentacao/Dicionario_e_input_20220530.zip"

download.file( dictionary_url , dictionary_tf , mode = 'wb' )

dictionary_files <- unzip( dictionary_tf , exdir = tempdir() )

sas_fn <- grep( '\\.sas$' , dictionary_files , value = TRUE )

sas_lines <- readLines( sas_fn , encoding = 'latin1' )

Determine fixed-width file positions from the SAS import script:

sas_start <- grep( '@00001' , sas_lines )

sas_end <- grep( ';' , sas_lines )

sas_end <- sas_end[ sas_end > sas_start ][ 1 ]

sas_lines <- sas_lines[ seq( sas_start , sas_end - 1 ) ]

# remove SAS comments
sas_lines <- gsub( "\\/\\*(.*)" , "" , sas_lines )

# remove tabs, multiple spaces and spaces at the end of each string
sas_lines <- gsub( "\t" , " " , sas_lines )
sas_lines <- gsub( "( +)" , " " , sas_lines )
sas_lines <- gsub( " $" , "" , sas_lines )

sas_df <- 
    read.table( 
        textConnection( sas_lines ) , 
        sep = ' ' , 
        col.names = c( 'position' , 'column_name' , 'length' ) ,
        header = FALSE 
    )

sas_df[ , 'character' ] <- grepl( '\\$' , sas_df[ , 'length' ] )

sas_df[ , 'position' ] <- as.integer( gsub( "\\@" , "" , sas_df[ , 'position' ] ) )

sas_df[ , 'length' ] <- as.integer( gsub( "\\$" , "" , sas_df[ , 'length' ] ) )

stopifnot( 
    sum( sas_df[ , 'length' ] ) == 
    ( sas_df[ nrow( sas_df ) , 'position' ] + sas_df[ nrow( sas_df ) , 'length' ] - 1 ) 
)

Download the latest data file:

this_tf <- tempfile()

this_url <-
    "https://ftp.ibge.gov.br/PNS/2019/Microdados/Dados/PNS_2019_20220525.zip"

download.file( this_url , this_tf , mode = 'wb' )

Import the latest data file:

library(readr)

pns_tbl <-
    read_fwf(
        this_tf ,
        fwf_widths( 
            widths = sas_df[ , 'length' ] , 
            col_names = sas_df[ , 'column_name' ] 
        ) ,
        col_types = 
            paste0( ifelse( sas_df[ , 'character' ] , "c" , "d" ) , collapse = '' )
    )

pns_df <- data.frame( pns_tbl )

names( pns_df ) <- tolower( names( pns_df ) )

pns_df[ , 'one' ] <- 1

Save Locally

Save the object at any point:

# pns_fn <- file.path( path.expand( "~" ) , "PNS" , "this_file.rds" )
# saveRDS( pns_df , file = pns_fn , compress = FALSE )

Load the same object:

# pns_df <- readRDS( pns_fn )

Survey Design Definition

Construct a complex sample survey design:

library(survey)

options( survey.lonely.psu = "adjust" )

pns_prestratified_design <-
    svydesign(
        id = ~ upa_pns ,
        strata = ~v0024 ,
        data = subset( pns_df , !is.na( v0028 ) ) ,
        weights = ~v0028 ,
        nest = TRUE
    )

popc.types <-
    data.frame(
        v00283 = as.character( unique( pns_df[ , 'v00283' ] ) ) ,
        Freq = as.numeric( unique( pns_df[ , 'v00282' ] ) )
    )

popc.types <- popc.types[ order( popc.types[ , 'v00283' ] ) , ]

pns_design <-
    postStratify(
        pns_prestratified_design ,
        strata = ~v00283 ,
        population = popc.types
    )

Variable Recoding

Add new columns to the data set:

pns_design <- 
    update( 
        pns_design , 

        medical_insurance = ifelse( i00102 %in% 1:2 , as.numeric( i00102 == 1 ) , NA ) ,
        
        uf_name =
        
            factor(
            
                as.numeric( v0001 ) ,
                
                levels = 
                    c(11L, 12L, 13L, 14L, 15L, 16L, 17L, 21L, 22L, 23L, 24L, 25L, 
                    26L, 27L, 28L, 29L, 31L, 32L, 33L, 35L, 41L, 42L, 43L, 50L, 51L, 
                    52L, 53L) ,
                    
                labels =
                    c("Rondonia", "Acre", "Amazonas", "Roraima", "Para", "Amapa", 
                    "Tocantins", "Maranhao", "Piaui", "Ceara", "Rio Grande do Norte", 
                    "Paraiba", "Pernambuco", "Alagoas", "Sergipe", "Bahia", "Minas Gerais", 
                    "Espirito Santo", "Rio de Janeiro", "Sao Paulo", "Parana", "Santa Catarina", 
                    "Rio Grande do Sul", "Mato Grosso do Sul", "Mato Grosso", "Goias", 
                    "Distrito Federal")
                    
            ) ,

        age_categories = factor( 1 + findInterval( c008 , seq( 5 , 90 , 5 ) ) ) ,

        male = as.numeric( v006 == 1 )
        
    )

Analysis Examples with the `survey` library

Unweighted Counts

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

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

svyby( ~ one , ~ uf_name , pns_design , unwtd.count )

Weighted Counts

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

svytotal( ~ one , pns_design )

svyby( ~ one , ~ uf_name , pns_design , svytotal )

Descriptive Statistics

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

svymean( ~ e01602 , pns_design , na.rm = TRUE )

svyby( ~ e01602 , ~ uf_name , pns_design , svymean , na.rm = TRUE )

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

svymean( ~ c006 , pns_design )

svyby( ~ c006 , ~ uf_name , pns_design , svymean )

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

svytotal( ~ e01602 , pns_design , na.rm = TRUE )

svyby( ~ e01602 , ~ uf_name , pns_design , svytotal , na.rm = TRUE )

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

svytotal( ~ c006 , pns_design )

svyby( ~ c006 , ~ uf_name , pns_design , svytotal )

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

svyquantile( ~ e01602 , pns_design , 0.5 , na.rm = TRUE )

svyby( 
    ~ e01602 , 
    ~ uf_name , 
    pns_design , 
    svyquantile , 
    0.5 ,
    ci = TRUE , na.rm = TRUE
)

Estimate a ratio:

svyratio( 
    numerator = ~ p00104 , 
    denominator = ~ p00404 , 
    pns_design ,
    na.rm = TRUE
)

Subsetting

Restrict the survey design to individuals that exercise three or more days per week:

sub_pns_design <- subset( pns_design , p035 %in% 3:7 )

Calculate the mean (average) of this subset:

svymean( ~ e01602 , sub_pns_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( ~ e01602 , pns_design , na.rm = TRUE )

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

grouped_result <-
    svyby( 
        ~ e01602 , 
        ~ uf_name , 
        pns_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( pns_design )

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

svyvar( ~ e01602 , pns_design , na.rm = TRUE )

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

# SRS without replacement
svymean( ~ e01602 , pns_design , na.rm = TRUE , deff = TRUE )

# SRS with replacement
svymean( ~ e01602 , pns_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( ~ medical_insurance , pns_design ,
    method = "likelihood" , na.rm = TRUE )

Regression Models and Tests of Association

Perform a design-based t-test:

svyttest( e01602 ~ medical_insurance , pns_design )

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

svychisq( 
    ~ medical_insurance + c006 , 
    pns_design 
)

Perform a survey-weighted generalized linear model:

glm_result <- 
    svyglm( 
        e01602 ~ medical_insurance + c006 , 
        pns_design 
    )

summary( glm_result )

Replication Example

This example matches Estimando totais of gross monthly income from the official PNSIBGE R package:

total_renda <- svytotal( ~ e01602 , pns_design , na.rm = TRUE )
stopifnot( round( coef( total_renda ) , 0 ) == 213227874692 )
stopifnot( round( SE( total_renda ) , 0 ) == 3604489769 )

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 PNS users, this code replicates previously-presented examples:

library(srvyr)
pns_srvyr_design <- as_survey( pns_design )