HIMSS14…. brought me to the highest of highs and the lowest of lows. As in HIMSS past, I found myself surrounded by people genuinely committed to fixing our defunct healthcare system — and the population health management group-chant was louder than ever.
At the same time, I died a little inside as I witnessed healthcare execs smoking, eating gluttonous amounts of salt-sugar-fat, and struggling to walk long distances in heels that would give podiatrists nightmares. Is pop health mgmt. the solution to all of our human failings? Probably not, but more IT might just keep some diabetics compliant, risk better managed, and doctors more aware of care delivery variation.
As I continue to focus on how analytics is being adopted as part of pop health strategies, here are some of the learnings that came out of my HIMSS14 trip:
A common analytics/pop health tech stack is emerging. The larger HIT vendors are now offering more comprehensive pop health solutions, as detailed in my CAPH Tech Stack diagram. These consist of (1) A data integration piece (2) Performance Management Analytics (3) Care Management Workflow apps.
Lots of small vendors are implementing pieces of this stack, e.g. they specialize in data integration, quality measures, predictive analytics, care management, … On everyone’s mind is how the best-of-breed approach vs. single vendor solution will shake out in the long run.
Vendors that are missing a piece of the puzzle want to fill this gap quickly, e.g. analytics-heavy vendors want to offer care management apps on top; care management vendors are looking for a strong analytics partner.
Data integration is hard to do and even more difficult to explain to the market. I continue to talk to anyone who cares about how challenging clinical data integration is at this point in time. Analytics vendors are in a precarious position where both the data sources underneath them and the business rules on top are ever-changing.
However, I noticed that vendors have all but given up on marketing their data integration capabilities (of course there are a few exceptions, e.g. Health Catalyst, Forward Health Group), with a few still saying that they “take care of the plumbing”. Most vendors, however, are positioning their offerings and messaging in terms of app-level capabilities, e.g. care management and physician benchmarking and seemingly ignoring data integration, or at least not talking about it.
Risk-management still means very different things across payer & provider lines. Prospective, claims-based risk scoring/risk adjustment systems have been employed by payers for many years. When I asked physician leaders and clinical-centric vendors about risk, they tended to discuss LACE for readmissions, or ways of figuring out who will be high risk in the future based on clinical variables and patient-reported outcomes. (I especially heard about the following 3 variables as being crucially important: living alone, chronic condition co-morbid with depression, and zip code). (Editor’s note: And a fourth, education level).
These clinical-centric folks generally knew little about and distrusted claims-based risk models — disregarding them as only prospective-based and lacking crucial variables.
Combining clinical + claims data: as fuzzy as ever. I liked to ask vendors how they are combining clinical and claims into the longitudinal patient record. What is the source of truth? For example, is a diabetic defined by the ICD-9 code from adjudicated claims, or meds/labs/problem list from clinical data? What if there is a discrepancy?
There are only a few options here: (1) claims data is source of truth, enrich this with clinical outcomes data; (2) clinical data is source of truth, possibly use claims to verify quality of clinical data; (3) keep clinical and claims separate, maybe feed clinical data into claims-based risk models; (4) Maintain duplicate data and do whatever the heck the client wants.
Well, I heard (4) over and over again. This claims-clinical issue is a good example of just how heterogeneous and to a certain extent immature, end-user needs are at this point in time, leading vendors to build highly customizable systems that require services-intensive engagements.
A few dissenters are coming out against care management workflow tools. Caradigm has just laboriously built a care management workflow tool with Geisinger Health Plan. Other vendors (Optum, McKesson) have been offering payer-based care management workflow tools for a long time and are repositioning and rebuilding these for the provider market. The thinking goes that by standardizing care management into a series of steps against a common care plan, there will be less variation and fewer FTEs will be required.
Smaller vendors without the $$$ or inclination to build out humongous workflow tools are asserting that these tools are already obsolete and still require too many FTEs (e.g. Phytel, Explorys). One common idea is that if we get good enough at predicting risk and automating patient outreach, then workflow tools will no longer be needed. I don’t envision care management workflow to be shelved anytime soon but am definitely not ruling out the anti-workflow camp long-term.
Ladies & men alike hated walking in their shoes. Wow, I wasn’t prepared for how many people had read my barefoot post — who then proceeded to complain about how much their feet were killing them (ladies especially felt they had no choice but to wear high heels). Next year, I hope to see others join me in wearing a barefoot-like shoe so we can discuss pop health without the distraction of foot pain.