November 26, 2018
Briefly
- “Rather than assume algorithms will produce better outcomes and hope they don’t accelerate discrimination, we should assume they will be discriminatory and inequitable unless designed specifically to redress these issues.” Lucy Bernholz
- ” Introduction of [a predictive risk screening tool] resulted in a statistically significant increase in emergency hospital admissions and use of other [National Health] services without evidence of benefits to patients or the [National Health Service].” In the academic journal BMJ, so a bit more technical
- Why the NY Times map of the US election results is so good: a Twitter thread
- Stacey Kirk in the Sunday Star-Times on the campaign to get Pharmac to pay for one of the most expensive drugs in the world.
- Interesting interactive in the Herald about quality-of-life and work in NZ cities. It’s very economist in style. That’s true on the good sense that it appreciates high house prices are a signal that lots of people want to live somewhere and low house prices are a signal that lots of people don’t. It’s also true in the bad sense that there some places where not many people want to live, but the people who do live there really like it — and this sort of analysis suppresses that variation in preferences.
- Interesting book on data science and data use: “Data Feminism”
Thomas Lumley (@tslumley) is Professor of Biostatistics at the University of Auckland. His research interests include semiparametric models, survey sampling, statistical computing, foundations of statistics, and whatever methodological problems his medical collaborators come up with. He also blogs at Biased and Inefficient See all posts by Thomas Lumley »
“Introduction of [a predictive risk screening tool] resulted in a statistically significant increase in emergency hospital admissions…”
But their results say this “Across 230, 099 participants, PRISM implementation increased use of health services: emergency hospital admission rates by 1 %…”
1% increase in emergency admissions . Hmmm …and thats statistically significant ?
Having 22 co authors too .
5 years ago
Yes, it’s quite definitely statistically significant. There were 230000 people in the study.
More to the point (which is why confidence intervals are more useful than p-values) it’s very strong evidence that the predictive models didn’t *decrease* admissions or costs by a meaningful amount, as hoped.
I’m not convinced that one author per ten thousand study particiants is a bad ratio, but either way it doesn’t say anything about the results.
5 years ago
The 230,000 people were the patient roll of the 32 general practices in the study which ran for roughly 18 months 183,000 were of a very low risk group ( mean age 36) while only 1100 were of high risk group, mean age 70). Its a much much smaller number for emergency hospital admissions, and 1% of those is how many. less than 10? Maybe more,but of more importance than the 230,000 number.
5 years ago
I was just working with some sf-12 data yesterday and the sf-12 scores of those patients are pretty low even in the low risk group. The changes in sf-12 scores are interesting – PRISM seems to make people feel worse but physically healthier.
I’d expect a service bump with PRISM as you suddenly capture all the people who haven’t previously been captured by self-report (because they didn’t know how to recognise or want to recognise their health issues). Eighteen months seems too small a time to look for changes in mortality rates.
5 years ago
The average number of hospital admissions across all risk groups was 0.17.
5 years ago