Quality of Care - 2026-02-19
Donabedian’s paper discussion.
Davis paper (Technology Acceptance Model): Before we deploy a system how can we know how it will be accepted? Intentions Predict Usage -→ Perceived usefulness is super important -→ Perceived ease is second. Note that you have to communicate usefulness to users.
Delone McLean’s Information Systems Success Model: (Information, System, Service) Quality. You need to think of these when you’re building an information system.
Theory of Reasoned Action: You need intent + the ambient environment/social norms to effect behaviour (think vaccines and how these norms have shifted over time).
Theory of Planned Behaviour: If you don’t believe X, you are very unlikely to engage. How do others view you when you don’t engage in behaviour. Self-efficacy (“it’s too hard”): even if you belive X, if you don’t have self-efficacy (“I can do this”) you will not engage in behaviour.
There’s this commonality of “Intention to Act” —→ Behaviour.
The key here is that you need to think about these structures/frameworks (or others) when you design interventions. They can really help you.
CSCW: informatics systems are designed for individual physicians. This is not how people work! PEople are embedded in teams! Informatics systems need to account for this teamwork. CDS’es target individuals, and this is ineffective. There are several tiers: Political, Institutional, Large Groups (Ward), Small Groups (Patient Care Team). Clinical decision-making is not 100% individual! There are a lot of people involved.
You make a distinction between routine workflows and exceptions.
We work on the assumption that clinicians only want the best for their patients. May be true but there are several motivating factors (e.g. promotions). It is important to understand incentive structures.
Your solution MUST align with clinical workflow. The timing is imporant. When you show them something matters. If you show them a solution after they’ve made a decision, you’re dead in the water wrt. adoption. Before might make you luckier in deployment. How good your model is won’t matter.
Another thing not addressed in a lot of informatics solutions: awareness. People need to be aware of what other people are working on. Focus (domain of one’s activity), Nimbus (outward projection of one’s activities), Awareness is Focus & Nimbus. Nimbus: how do you know what someone’s doing? Sitting in office and on laptop: are you busy? Are you just browsing TikTok? Are you deep into reading a paper? Can you be interrupted?
Coeira’s Computation-Conversation: When do you even deploy technology? Key concepts are common ground (what info do you and another share?), grounding (how do you acquire common ground?): solid ground (established prior), shifting ground (common ground on-demand).
The idea here is to distinguish “conversation” tasks from “information” tasks. Guidelines, for example, can be encoded in information systems. You might be better off with talking to people when dealing with exceptions.
TODO:
- Prediction versus forecasting