Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Why Are Westerners Unhappy with their Health In...

Peter Higgins
September 13, 2019

Why Are Westerners Unhappy with their Health Insurance?

A presentation on teaching R with real data, and finding new insights by learning new R packages with real data. Also NCQA data on health insurance satisfaction, and why people in western US states are less satisfied with their health insurance

Peter Higgins

September 13, 2019
Tweet

More Decks by Peter Higgins

Other Decks in Education

Transcript

  1. MICHIGAN MEDICINE Why People in the Western US Are Dissatisfied

    with their Health Insurance Peter D.R. Higgins, Megan McLeod, and Jeffrey Berinstein University of Michigan @ibddoctor Slide deck: https://speakerdeck.com/higgi13425
  2. MICHIGAN MEDICINE Teaching • I teach MDs how to analyze

    clincal data in R (tidy-biased) • Start with toy medical datasets • Move to real medical data as soon as possible • Learn with datasets that learners care about • Build in reproducible research practices early
  3. MICHIGAN MEDICINE Why Bother to Teach MDs? Tommy Francis, MD

    Enrolled 1.8 million children Analyzed results with punch cards John Snow, MD Developed early version of geom_sf Removed pump handle Mona Hanna-Atisha, MD, MPH Wrangled data from EMR Compared lead levels
  4. MICHIGAN MEDICINE Learning R Can Lead to New Insights •

    I wanted to • Learn rvest for harvesting data • Learn geom_sf for mapping • My patients are frequently unhappy with their health insurance • They have to select a health insurance plan each October • They have little information on which to base a selection ON HOLD WITH INSURANCE CO FOR 3 HOURS, STILL NO ANSWER
  5. MICHIGAN MEDICINE Learning R with Real Data • Found the

    NCQA site: National Committee for Quality Assurance reports health insurance consumer satisfaction • Each private insurance plan rated on a 0-5 scale each year • Website http://healthinsuranceratings.ncqa.org/2018/ • 150 sub-websites, for 3 insurance types x 50 states, i.e. this link http://healthinsuranceratings.ncqa.org/2018/search/Commercial/AL This link provides ratings for private(Commercial vs. Medicaid and Medicare) insurance plans in Alabama
  6. MICHIGAN MEDICINE link <- “http://healthinsuranceratings.ncqa.org/2018/” for (i in 1:length(state_abbs)) {

    for (j in 1:length(instype)) { healthins <- read_html(paste0(link, "/", instype[j], "/", state_abbs[i])) onetable<- as.data.frame(healthins %>% html_nodes("table") %>% .[[3]] %>% html_table(fill=T)) if (i=1 & j =1) { fulltable <- onetable } else { fulltable <- bind_rows(fulltable, onetable) • Some cleaning of missing ratings, names of plans, etc. to get a clean dataframe • group_by() %>% summarize() to get the mean Commercial insurance plan satisfaction by state • Now map it Looping with rvest
  7. MICHIGAN MEDICINE Mapping with geom_sf() map_2018 <- consumersat_2018 %>% ggplot(aes(fill

    = consumer_satisfaction_rating, color=consumer_satisfaction_rating)) + geom_sf() + scale_fill_viridis_c() + #fill scale theme_minimal(base_size = 10) + #base font size coord_sf(xlim = c(-60, -125), ylim = c(20, 50), datum = NA) + #sets limits, turns off lat/long grid labs(title = "2018", fill = 'Rating')
  8. MICHIGAN MEDICINE Multiple Years of Insurance Satisfaction That is surprising.

    And surprisingly consistent over time. Why are Westerners dissatisfied with Commercial health insurance? pubmap1 <- ggpubr:: ggarrange(map_2015, map_2016, map_2017, map_2018, ncol=2, nrow=2, common.legend = TRUE, legend="bottom") pubmap1
  9. MICHIGAN MEDICINE Outlier Health Insurance Plans • After controlling for

    state and insurance family Top Ten Over-performers Bottom Ten Under-performers
  10. MICHIGAN MEDICINE So Why Are Westerners Dissatisfied? • State level

    explanatory variables needed • State Health Facts website https://www.kff.org/statedata/ • All downloadable
  11. MICHIGAN MEDICINE Take Home Points • Learn new skills with

    new packages in R • Practice with real data that are important to you • You might learn something new and useful
  12. MICHIGAN MEDICINE Data, Code, and Preprint Manuscript • OSF: Open

    Science Foundation repository https://osf.io/xcbe4/ • Preprint: MedRxiv https://www.medrxiv.org/content/10.1101/19004762v2