Another Large Scale Research Study Confirms the Value of the Approach of Population Health Management

And here's another study, this time published in JAMA about Kaiser in Northern California that found that the following five components resulted in an increase of population-based blood pressure control: 

1. "Registry" (which the Disease Management Care Blog says is really a stand-alone database that is outside of the electronic health record);

2. "Control Rates" (which the DMCB figures is really an updated "dashboard" that displays key metrics to administrators and docs that provides feedback and helps keep everyone on the same page);

3. "Guideline" (in reality, it was a campaign to gain provider buy-in consisting of emails, publications, pocket cards, conferences, lectures and decision support);

4. "Medical assistant" follow-up operating under protocol to adjust medications (a.k.a population-based care management)

5. "Single" pill treatment (in other words, keep it simple by using pharmaceuticals that are combined in a single once a day prescription pill).

DMCB readers will not be surprised to know that the registry showed a progressive improvement in BP control (defined as less than 140/90 with the usual HEDIS® caveats) from 43.6% in 2001 to 80.4% in 2009.  Because everyone with hypertension at Kaiser was in the registry, there is no internal comparison group.  However, national and northern California HEDIS® rates for blood pressure control ranged from 55.4% to 69.4%.

While the results are 1) not necessarily generalizable outside of integrated systems like Kaiser (so we don't know for sure that this would work in a network of primary care clinics in Idaho), and 2) may have been influenced by an influx of patients with mild and easy-to-treat hypertension during the campaign), the DMCB is impressed. 

An 80% control rate for hypertension is damn good. 

The DMCB also figures that each of the interventions above are mutually supportive and even synergistic.  The whole is much greater than the sum of its parts.

How to translate this kind of success to networks of independent practices?  The answer, says the DMCB, is population health management: sponsored programs that can be owned by an insurer or a provider network that synergistically identify a population, maintain a data base, create a virtuous cycle of measurement and adjustment, get the doctors on board, deploy care managers and are smart about the pharmacy benefit.

If your a PHM service provider, vendor, consultant or stakeholder, the DMCB suggests this is one of those research papers you should bookmark, quote and aspire to.

Image from Wikipedia

The Electronic Health Record (EHR) On-Line Portal Increases Hospitalization Rates



"Hi doc! I used my on-line
portal to make an appointment!"

Hey there Accountable Care Organization executive.

You're probably willing to continue to commit millions of dollars toward an electronic health record (EHR) coupled to an online patient portal.  That's because you've been told by your leadership team that electronic consumer empowerment, patient-provider communication and the substitution of efficient two-way messaging for costly face-to-face visits will increase quality, reduce expenses, generate shared savings and guarantee that your life-sized portrait will be prominently displayed in your flagship hospital's lobby.

Well, after you've read a just-published JAMA research study by Ted Palen, Colleen Ross, David Powers and Stanley Xu, you may want to tell your administrative assistant to cancel that appointment with the portrait artist.

The article's title is Association of Online Patient Access to Clinicians and Medical Records With Use of Clinical Services.

How the study was done:

Kaiser Permanente Colorado added "MyHealthManager" (MHM) to their EHR in May 2006. MHM allows patients to view tests, records, problem lists as well as care plans, schedule appointments, request refills and message their doctors. By June of 2009, over 375,000 Kaiser patients had signed up for MHM. Of those, about 45% had used the system at least once.  Of this number, Kaiser researchers pulled the records of 44,321 persons who had been continuously enrolled in the Kaiser system for at least two years. 

This group was retrospectively matched to a control group of Kaiser patients who had not signed up for MHM.  The authors did this through "propensity matching." This found a similar number of patients, based on age, gender, race, number of chronic illnesses and baseline office visits who, using logistic regression analytics, appeared to be the type of patient who would otherwise sign up for MHM.

The results:

Compared to non-MHM patients, the MHM experienced an increase in hospitalization rates (20 per thousand patients) and emergency room visits (11 per thousand).  In other words, for every hundred patients, the on-line portal seemed to lead to 2 extra hospitalizations and 1 extra ER visit. Both differences were statistically significant.

There were also increases in the number of office visits (.7 per patient per year), telephone calls (.3 per patient per year) and after-hour clinic visits (18.7 per thousand patients per year).

Caveats:

The authors correctly point out that this study is not perfect.  Retrospective propensity matching is not as good as a randomized clinical trial; it's possible that the patients who self-selected for MHM were already realtively more interested in or likely to increase their use of health care services.  Results at Kaiser may not apply elsewhere.

Implications:

Despite the limitations, this study should be a wake-up call for those who believe EHR portals is a savings panacea.  By increasing access to on-line services, physicians and patients may paradoxically use the system to address concerns that otherwise wouldn't come to medical attention.  In other words, the EHR portal exacerbates the classic health care economics problem of supplier-induced demand.

Image from Wikipedia

What The Electronic Health Record (EHR) Can, and Cannot, Do

"The widespread use of electronic health records (EHRs) in the United States is inevitable. EHRs will  improve caregivers' decisions and patients' outcomes. Once patients experience the benefits of this technology, they will demand nothing less from their providers."

"... lives can be saved, outcomes of care improved, and costs reduced by transforming the healthcare system through the appropriate use of IT and management systems."

While the breathless exuberance if the EHR cabal continues to spin on, the Disease Management Care Blog ran into two real world studies from strong institutions that tell it what the EHR can, and cannot, do.

Briefly.... 

In the first study, Kaiser Permanente of Colorado initiated an interactive voice response outbound telephone campaign to eligible patients that, according to its EHR, had not undergone screening for colon cancer. The call lasted 5 minutes and used the telephone keypad to solicit additional information and patient preferences about screening. If the call was not completed, a mail-in screening packet was sent to the patient with a reminder letter 4 weeks later. In addition to the IVR campaign itself, there were other provider as well as staff educational and promotional campaigns, individualized physician reports and quality incentive programs. Patients with known high risk (for example, a history of polyps), removal from the screening program by request of their doctor or evidence of prior screening were not included. 58,440 patients got the IVR.  Their mean age was 59 years and 53% were female.

Results: 45% of those exposed to the IVR eventually completed the screening.

In the second study, the Veterans Administration "VistA" EHR was mined to find patients who had received at least 2 separate prescriptions for a cholesterol lowering drug during a prior 9 month baseline period that was followed by an "absent" prescription.  If a "past due date" was found, it was assumed that the patient was having a problem with medication compliance. Of an initial 1000 "past due" patients, closer examination of the records revealed that 176 had gotten the drug.  Of the 824 left over, 95 had died and 17 had entered nursing homes which were supplying the medicine. Patients were then surveyed about their medication use and 302 indicated that they had been told by their physicians to stop the medicine, had had side effects, had switched to alternative treatments or had gotten their medicine from outside the VA system.

Results:  After all this, it turned out that only 20% of the patients who, according to the EHR, had prematurely stopped their cholesterol medicine, were correctly identified.

The most generous interpretation of both studies is that something is better than nothing.  The DMCB agrees but points out that having an EHR doesn't automatically lead to 100% cancer screening rates or 100% medication compliance.

In the colon cancer study, the authors had to turn to a free standing IVR system instead of using tools resident in the ER (physician prompts, automated patient letters) to achieve their cancer screening success.  The EHR was necessary, but not sufficient.

In the cholesterol drug study, researchers had to expend a lot of time and energy to clean up the EHR's data base. While 20% of patients at apparent risk will eventually benefit, finding these patients has gone from pulling individual paper charts to pulling individual patient data.

Improve decisions and outcomes?  Save lives? Reduce costs?  You be the judge. 

Kaiser Permanente's Lesson for the Disease Management Community

The disease management care blog found this blog posting of an interesting letter from George Halvorson, CEO of Kaiser Permanente. In it, he celebrates Kaiser’s unique information systems that have “all of the data about the provision of care and about the cost of care.” He points out this capability was responsible for important research published in the American Journal of Obstetrics and Gynecology on the association between caffeine consumption and miscarriage. He takes not a small amount of pleasure in noting that the findings of that study were reported in many news outlets worldwide.

According to George (that’s how he signed the letter) just what is it about the Kaiser information systems that make them so special? He gushes that it is its comprehensiveness, thanks to patients getting all their care at Kaiser. This secret sauce contrasts with the “splintered” care in other settings, where patients are unnecessarily victimized by unconnected clinics and hospitals.

So, the ever curious disease management blog went to the published study (hats off to AJOG for making it fully available on-line) to give it a once-over. It turns out the study was not an exercise in culling and correlating extant caffeine and miscarriage data from the user-friendly KP information ether. The authors’ success was the result of hard work, done the old fashioned way, surveying volunteers (with a 43% refusal rate) about their caffeine intake one interview at a time. And yes, while the ready availability of a single clinical data repository aided the researchers, the statistically significant association between caffeine intake and miscarriage:

a) could have also been extracted from a splintered electronic health record and

b) depended not only on the databases and the medical records but on “contacting participants whose outcomes could not be determined by using the previous 2 methods.” Those would be the folks in the Kaiser System getting splintered care.

While it is the nature of health care CEOs to spin, the disease management blog thinks Dr. Halvorson (assuming the letter is completely reproduced) could have mentioned the lead author Xiaoping Weng and colleagues Roxana Odouli and De-Kun Li by name, congratulated them for their hard work and not implied that the information systems yielded up the association after the researchers keystroked Esc-F1.

That being said, George has a point. The study was significantly aided by the Kaiser registries. This has a huge lesson for the disease management community. I believe these companies’ multi-state multi-payor insurance claims and clinical data bases rival Kaiser’s. In fact, they probably exceed it. Toss in their IVR and telephonic-based access to millions of persons and the possibilities become astonishing. The Carl Saganesque access to billions and billions of 1s and 0s make it possible for them to discover hitherto unknown, important and possibly life-saving associations.

Want a small example? At a prior DMAA meeting, the disease management blog sat in on a presentation about the use of disease management data as well as other public information to better identify persons at risk for diabetes. A fascinating finding was the discovery of a strong association between diabetes and ownership of a minivan. Yup, a minivan. It makes sense when you think about it: exclude persons with an age and gender consistent being a soccer mom, and the remainder are older individuals with an abdominal girth that is best accommodated by the height and style of the “captain’s chair” typical of that style of automobile. Is this life saving? While it doesn’t rise to the level of miscarriages, asking patients about the presence of a minivan may be the next best thing if you can’t get BMI or waist circumference data, lead to a strong recommendation to see a doctor and….you get the picture.

The difference is that Weng, Odouli and Li did us a great service by getting their insight into the public domain. In contrast, the relative silence from the disease management companies is deafening. Either that or their work is not getting the attention it deserves.

More on the implications of this tomorrow.