mcbs.Rmd

# Medicare Current Beneficiary Survey (MCBS) {-}

[![License: GPL v3](https://img.shields.io/badge/License-GPLv3-blue.svg)](https://www.gnu.org/licenses/gpl-3.0) <a href="https://github.com/asdfree/mcbs/actions"><img src="https://github.com/asdfree/mcbs/actions/workflows/r.yml/badge.svg" alt="Github Actions Badge"></a> 

The monitoring system for Medicare enrollees in the United States on topics not available in the program's administrative data, such as out of pocket expenditure and beneficiary satisfaction.

* Survey and supplemental tables with one row per sampled individual, although downloadable datasets not linkable.

* A complex sample survey designed to generalize to all elderly and disabled individuals with at least one month of program enrollment during the calendar year.

* Released annually as a public use file since 2015.

* Conducted by the Office of Enterprise Data and Analytics (OEDA) of the Centers for Medicare & Medicaid Services (CMS) through a contract with NORC at the University of Chicago.

---

## Recommended Reading {-}

Four Example Strengths & Limitations:

✔️ [Respondents interviewed up to three times annually across four years](https://edit.cms.gov/Research-Statistics-Data-and-Systems/Research/MCBS/Downloads/MCBS_Tutorial.pdf#page=21)

✔️ [More than 1.2 million interviews since inception](https://mcbs-interactives.norc.org/about/mcbs/)

❌ [Some respondents designate a proxy to interview on their behalf](https://www.census.gov/fedcasic/fc2022/pdf/2C_carpenter.pdf)

❌ [Facility residents included, but not in public use file](https://www.cms.gov/files/document/mcbs-advanced-tutorial-using-community-and-facility-data.pdf#page=7)

<br>

Three Example Findings:

1. [Among all Medicare beneficiaries, 7 percent reported having problems paying a medical bill in 2021](https://www.medpac.gov/document/march-2024-report-to-the-congress-medicare-payment-policy/).

2. [Between 1999 and 2017, Medicare beneficiaries with diabetes faced higher out-of-pocket costs](https://doi.org/10.2337/dc20-2708).

3. [In 2020, healthcare expenditure for non-fatal falls was $80 billion, the majority paid by Medicare](https://doi.org/10.1136/ip-2023-045023).

<br>

Two Methodology Documents:

> [MCBS Methodology Report](https://data.cms.gov/sites/default/files/2024-10/SFPUF2022_DUG.pdf)

> [MCBS Advanced Tutorial on Weighting and Variance Estimation](https://www.cms.gov/files/document/mcbs-advanced-tutorial-weighting-and-variance-estimation.pdf)

<br>

One Haiku:

```{r}
# old, or disabled
# access to medical care,
# utilization
```

---

## Download, Import, Preparation {-}

```{r eval = FALSE , results = "hide" }
tf <- tempfile()

this_url <- "https://www.cms.gov/files/zip/cspuf2021.zip"

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

unzipped_files <- unzip( tf , exdir = tempdir() )

mcbs_csv <- grep( '\\.csv$' , unzipped_files , value = TRUE )

mcbs_df <- read.csv( mcbs_csv )

names( mcbs_df ) <- tolower( names( mcbs_df ) )
```

### Save Locally \ {-}

Save the object at any point:

```{r eval = FALSE , results = "hide" }
# mcbs_fn <- file.path( path.expand( "~" ) , "MCBS" , "this_file.rds" )
# saveRDS( mcbs_df , file = mcbs_fn , compress = FALSE )
```

Load the same object:

```{r eval = FALSE , results = "hide" }
# mcbs_df <- readRDS( mcbs_fn )
```

### Survey Design Definition {-}
Construct a complex sample survey design:

```{r eval = FALSE , results = "hide" }
library(survey)

mcbs_design <-
	svrepdesign(
		weight = ~cspufwgt ,
		repweights = 'cspuf[0-9]+' ,
		mse = TRUE ,
		type = 'Fay' ,
		rho = 0.3 ,
		data = mcbs_df
	)
```

### Variable Recoding {-}

Add new columns to the data set:
```{r eval = FALSE , results = "hide" }
mcbs_design <-
	update(
		
		mcbs_design ,

		one = 1 ,
		
		csp_age =
			factor( 
				csp_age , 
				levels = 1:3 , 
				labels = 
					c( 
						'01: younger than 65' ,
						'02: 65 to 74' ,
						'03: 75 or older'
					)
			) ,
 		
		two_or_more_chronic_conditions = as.numeric( csp_nchrncnd > 1 ) ,

		csp_sex = factor( csp_sex , labels = c( 'male' , 'female' ) )
	)
```

---

## Analysis Examples with the `survey` library \ {-}

### Unweighted Counts {-}

Count the unweighted number of records in the survey sample, overall and by groups:
```{r eval = FALSE , results = "hide" }
sum( weights( mcbs_design , "sampling" ) != 0 )

svyby( ~ one , ~ csp_age , mcbs_design , unwtd.count )
```

### Weighted Counts {-}
Count the weighted size of the generalizable population, overall and by groups:
```{r eval = FALSE , results = "hide" }
svytotal( ~ one , mcbs_design )

svyby( ~ one , ~ csp_age , mcbs_design , svytotal )
```

### Descriptive Statistics {-}

Calculate the mean (average) of a linear variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
svymean( ~ pamtoop , mcbs_design )

svyby( ~ pamtoop , ~ csp_age , mcbs_design , svymean )
```

Calculate the distribution of a categorical variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
svymean( ~ csp_sex , mcbs_design )

svyby( ~ csp_sex , ~ csp_age , mcbs_design , svymean )
```

Calculate the sum of a linear variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
svytotal( ~ pamtoop , mcbs_design )

svyby( ~ pamtoop , ~ csp_age , mcbs_design , svytotal )
```

Calculate the weighted sum of a categorical variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
svytotal( ~ csp_sex , mcbs_design )

svyby( ~ csp_sex , ~ csp_age , mcbs_design , svytotal )
```

Calculate the median (50th percentile) of a linear variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
svyquantile( ~ pamtoop , mcbs_design , 0.5 )

svyby( 
	~ pamtoop , 
	~ csp_age , 
	mcbs_design , 
	svyquantile , 
	0.5 ,
	ci = TRUE 
)
```

Estimate a ratio:
```{r eval = FALSE , results = "hide" }
svyratio( 
	numerator = ~ pamtoop , 
	denominator = ~ pamttot , 
	mcbs_design 
)
```

### Subsetting {-}

Restrict the survey design to household income below $25,000:
```{r eval = FALSE , results = "hide" }
sub_mcbs_design <- subset( mcbs_design , csp_income == 1 )
```
Calculate the mean (average) of this subset:
```{r eval = FALSE , results = "hide" }
svymean( ~ pamtoop , sub_mcbs_design )
```

### Measures of Uncertainty {-}

Extract the coefficient, standard error, confidence interval, and coefficient of variation from any descriptive statistics function result, overall and by groups:
```{r eval = FALSE , results = "hide" }
this_result <- svymean( ~ pamtoop , mcbs_design )

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

grouped_result <-
	svyby( 
		~ pamtoop , 
		~ csp_age , 
		mcbs_design , 
		svymean 
	)
	
coef( grouped_result )
SE( grouped_result )
confint( grouped_result )
cv( grouped_result )
```

Calculate the degrees of freedom of any survey design object:
```{r eval = FALSE , results = "hide" }
degf( mcbs_design )
```

Calculate the complex sample survey-adjusted variance of any statistic:
```{r eval = FALSE , results = "hide" }
svyvar( ~ pamtoop , mcbs_design )
```

Include the complex sample design effect in the result for a specific statistic:
```{r eval = FALSE , results = "hide" }
# SRS without replacement
svymean( ~ pamtoop , mcbs_design , deff = TRUE )

# SRS with replacement
svymean( ~ pamtoop , mcbs_design , deff = "replace" )
```

Compute confidence intervals for proportions using methods that may be more accurate near 0 and 1. See `?svyciprop` for alternatives:
```{r eval = FALSE , results = "hide" }
svyciprop( ~ two_or_more_chronic_conditions , mcbs_design ,
	method = "likelihood" )
```

### Regression Models and Tests of Association {-}

Perform a design-based t-test:
```{r eval = FALSE , results = "hide" }
svyttest( pamtoop ~ two_or_more_chronic_conditions , mcbs_design )
```

Perform a chi-squared test of association for survey data:
```{r eval = FALSE , results = "hide" }
svychisq( 
	~ two_or_more_chronic_conditions + csp_sex , 
	mcbs_design 
)
```

Perform a survey-weighted generalized linear model:
```{r eval = FALSE , results = "hide" }
glm_result <- 
	svyglm( 
		pamtoop ~ two_or_more_chronic_conditions + csp_sex , 
		mcbs_design 
	)

summary( glm_result )
```

---

## Replication Example {-}

This example matches the weighted total from the [2021 Data User's Guide: Cost Supplement File Public Use File](https://www.cms.gov/files/document/2021mcbsdatausersguidecsfilepuf.pdf):

```{r eval = FALSE , results = "hide" }
stopifnot( round( coef( svytotal( ~ one , mcbs_design ) ) , 0 ) == 59040948 )
```

---

## Analysis Examples with `srvyr` \ {-}

The R `srvyr` library calculates summary statistics from survey data, such as the mean, total or quantile using [dplyr](https://github.com/tidyverse/dplyr/)-like syntax. [srvyr](https://github.com/gergness/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](https://cran.r-project.org/web/packages/srvyr/vignettes/srvyr-vs-survey.html) details the available features. As a starting point for MCBS users, this code replicates previously-presented examples:

```{r eval = FALSE , results = "hide" }
library(srvyr)
mcbs_srvyr_design <- as_survey( mcbs_design )
```
Calculate the mean (average) of a linear variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
mcbs_srvyr_design %>%
	summarize( mean = survey_mean( pamtoop ) )

mcbs_srvyr_design %>%
	group_by( csp_age ) %>%
	summarize( mean = survey_mean( pamtoop ) )
```