HW2

Author

Drew Schaefer

I will be using the NHIS linked mortality data.

The event variable is death, which in this data set is indicated with a 1 for the MORTSTAT variable.

Tentatively, for this analysis I will use death within 5 years of the survey but also the entire survey (just to see what the differences are).

The censoring indicator is whether the event (death) did not take place, which means the MORTSTAT variable has a value of 2.

In this preliminary analysis the grouping variable will be sex (this is dichotomous). In the coming weeks I will decide on something more substantive.

Following mortality patterns, males should have a lower survival probability than females (at least until old age). This is shown to be true in the survival probability graphs.

SEX=1 is male and SEX=2 is female

---
title: "HW2"
format: html
editor: visual
author: "Drew Schaefer"
code-tools: TRUE
---

I will be using the NHIS linked mortality data.

The event variable is death, which in this data set is indicated with a 1 for the MORTSTAT variable.

Tentatively, for this analysis I will use death within 5 years of the survey but also the entire survey (just to see what the differences are).

The censoring indicator is whether the event (death) did not take place, which means the MORTSTAT variable has a value of 2.

In this preliminary analysis the grouping variable will be sex (this is dichotomous). In the coming weeks I will decide on something more substantive.

Following mortality patterns, males should have a lower survival probability than females (at least until old age). This is shown to be true in the survival probability graphs.

```{r load libraries, include=FALSE}
library(tidyverse)
library(ipumsr)
library(survival)
library(ggsurvfit)
library(muhaz)
```

```{r read in NHIS data, include=FALSE}
nhis <- read_ipums_ddi("nhis_00002.xml")
  
nhis_dat <- read_ipums_micro(nhis) 

nhis_dat <- haven::zap_labels(nhis_dat)
```

```{r start cleaning data, echo=FALSE}
nhis_dat <- nhis_dat %>%
  filter(MORTELIG == 1)

nhis_dat <- nhis_dat %>%
  mutate(death_age = ifelse( MORTSTAT ==1, 
                             MORTDODY - (YEAR - AGE), 
                             2019 - (YEAR - AGE)), 
         d.event = ifelse(MORTSTAT == 1, 1, 0))

age_fit <- survfit(Surv(death_age, d.event) ~ 1, 
                   data = nhis_dat)

age_fit %>%
  ggsurvfit() +
  add_confidence_interval(type = "ribbon") +
  add_quantile() 
```

SEX=1 is male and SEX=2 is female

```{r, echo=FALSE}

sex_age_fit <- survfit(Surv(death_age, d.event) ~ SEX, 
                   data = nhis_dat)

sex_age_fit %>%
  ggsurvfit() +
  add_confidence_interval(type = "ribbon") +
  add_quantile()
  
```

```{r, echo=FALSE}
nhis_dat <- nhis_dat %>%
  mutate(death_time = ifelse( MORTSTAT ==1, 
                             MORTDODY - YEAR , 
                             2019 - YEAR ), 
         d5.event = ifelse(MORTSTAT == 1 & death_time <= 5, 1, 0))



time_fit <- survfit(Surv(death_time, d5.event) ~ 1, 
                  conf.type = "log",
                   data = nhis_dat)

time_fit %>%
  ggsurvfit()+
  xlim(0, 1) +
  add_censor_mark() 
```