Rothman: Oral contraceptives and stroke.

In pubh: A Toolbox for Public Health and Epidemiology

Oral contraceptives and stroke.

Description

A case-control study of oral contraceptives and stroke in young women with presence or absence of hypertension. Cases represent thrombotic stroke and controls are hospital controls. The group of no hypertension includes normal blood pressure (<140/90 mm Hg) and borderline hypertension (140-159/90-94 mm Hg). Hypertension group includes moderate hypertension (160-179/95-109 mm Hg) and severe hypertension (180+/110+ mm Hg). This data has been used as an example of join exposure by Rothman for measuring interactions (see examples).

Usage

Rothman

Format

A labelled tibble with 477 rows and 3 variables:

stroke

Thrombotic stroke, factor with levels "No" and "Yes".

oc

Current user of oral contraceptives, factor with levels "Non-user" and "User".

ht

Hypertension, factor with levels "No" (<160/95 mm Hg) and "Yes".

Source

Collaborative Group for the Study of Stroke in Young Women (1975) Oral contraceptives and stroke in young women. JAMA 231:718-722.

Rothman, KJ (2002) Epidemiology. An Introduction. Oxford University Press.

Examples

require(dplyr, quietly = TRUE)
data(Rothman)

Rothman %>%
  select(stroke, oc, ht) %>%
  tbl_summary() %>%
  cosm_sum()

mhor(stroke ~ ht / oc, data = Rothman)

## Model with standard interaction term:
model1 <- glm(stroke ~ ht * oc, data = Rothman, family = binomial)
glm_coef(model1)

## Model considering join exposure:
Rothman$join <- 0
Rothman$join[Rothman$oc == "Non-user" & Rothman$ht == "Yes"] <- 1
Rothman$join[Rothman$oc == "User" & Rothman$ht == "No"] <- 2
Rothman$join[Rothman$oc == "User" & Rothman$ht == "Yes"] <- 3
Rothman$join <- factor(Rothman$join, labels = c(
  "Unexposed", "Hypertension", "OC user",
  "OC and hypertension"
))

require(sjlabelled, quietly = TRUE)
Rothman$join <- set_label(Rothman$join, label = "Exposure")

Rothman %>%
  select(stroke, join) %>%
  cross_tbl(by = "stroke")

model2 <- glm(stroke ~ join, data = Rothman, family = binomial)
glm_coef(model2)

pubh documentation built on Nov. 14, 2023, 1:08 a.m.