Now Reading: Do people who use PHRs use health care more?

Therefore, the findings reported by Palen et al in this issue of JAMA will be sobering for patient portal enthusiasts.

Not really, because I wasn’t under the influence.

About a month ago I was walking with Christine Sinsky, MD, who is a passionate advocate for usability in Electronic Health Records (see her testimony in 2011 here). I told her “5-7 years ago, your work would have made me anxious, because it might prompt people who are pessimistic about the future to not allow our profession (and mostly, our patients) to see the benefits of health information technology while they were just being envisioned.”

Before I discuss the data, I want to celebrate two statements that David Bates and Susan Wells make in their editorial:

The (linked personal health record) appears to be dominating the scene and offers many advantages because it can include specific information from the medical record that is difficult and time consuming for patients to access otherwise.

and

Personal health records are here to stay, and the tethered (linked) architecture appears to offer the most benefits.

Hooray – David (who is truly an expert in the field and who I had the honor of shadowing in 2007) puts to rest the ongoing myths that (a) PHRs can’t be adopted successfully by large numbers of patients and (b) PHRs that are linked to actual physicians are somehow not really PHRs. Kaiser Permanente’s adoption numbers are near 4.1 million at time of this writing, and represent 63% of the eligible population of users.

Now the data – using a retrospective cohort design, colleagues at Kaiser Permanente Colorado found that rates of office visits, telephone encounters, after hours clinic visits, emergency room visits, and hospitalizations were higher for people who used Kaiser Permanente’s My Health Manager compared to similar patients who did not access or use My Health Manager from 2005-2010. They started the clock on the day the patient signed up for My Health Manager for the MHM User groups. They started the clock on the median date of patient enrollment for the Non-MHM users. I have a question into the study authors about this, since signing up for My Health Manager is usually tied to a visit, which could be tied to increased health care usage in general (response from the study author is in the comments below – added 12/6/2012). The same is not true of the date chosen as the start date to measure for the control group. The study team did a thorough job adjusting for age, gender, ethnicity, number of chronic conditions, and office visits.

So the answer to the question in the title of the post, for this group of patients, is yes. Ironically, this may actually quell the fears of the fee for service health care system and encourage it to adopt PHRs more rapidly. Previously, the concern was that PHRs would reduce reimbursement for visit-based care. So they’ll go faster, integrated care is already there, all the patients will win in the end.

There is NOT an answer to the question, “Do PHRs cause people to use health care more?” There is also NOT an answer to the question, “Do PHRs cause people to be more unhealthy?” In fact, previous studies at Kaiser Permanente show that there’s an association with greater health among users of the same PHR (See: Data Graphic: Patient-Physician E-mail improves care | Ted Eytan, MD). Another significant study shows that there’s an association with more care continuity, because people who use PHRs are more likely to stay with their current health care, especially in their first year of health system membership (see: Now Reading: Access to a PHR constitutes highly meaningful use – published data on member retention from Kaiser Permanente | Ted Eytan, MD).

It’s not clear exactly what’s going on, which means that the next best step is probably to dig into the patient stories to find out answers to the many questions posed by the study authors – are these patients more activated to begin with? Are they more unstable medically which is why they sought PHR use in the first place? Are they more or less healthy (able to achieve their life goals) because of the use of the PHR and associated health services? And a question from me – would their utilization be even higher without a PHR? The way to dig in, of course, is using the biggest innovation that PHRs have brought to health care – going beyond the data and listening.

This brings me back to my walk with Christine. I agree with David and Susan when they say in their editorial that it’s unlikely that studies where patients are randomly given (or denied) accesss to their health information online will be conducted. Happy to be told we are beyond that. People who worked to create PHRs created them to improve the use of health care, so that it would be safer, listen better than it does, be more health focused instead of health care focused. We’ll study that as we give patients what they want. I used to hear the term “turning visits into clicks,” which is really a healthcare centric view of what we’re here for. Interestingly, before I read this study I had the opportunity to visit with Anna Lisa-Silvestre, Vice President of Online Services for Kaiser Permanente (@ALSilvestre), in Washington, DC, and here’s what I tweeted:

I believe this is possible, I am happy that we’ve reached critical mass in being able to study and listen better to our patients, and I’m optimistic about the future. The video below, released with the study, touches on some of these themes below. Thanks for the excellent work, Institute for Health Research.

9 Comments

Hi Ted –Thoughtful post.  I’d be interested to know whether patientswho use PHRs use healthcare services differently from those who don’t. Forexample, the woman in the video uses e-mail to ask her MD basic maintenancequestions (“should I take a vitamin?”), so perhaps she is able to focus herin-person visits on other topics, whether more “serious” ones (“I’m nervousabout X or Y–or have experienced the following syptoms…”), or ones thatbenefit more directly from an in-person component (like a demonstration of howto use an inhaler, or do a particular exercise or exam)? Ultimately myquestion is related to your question re whether PHR-users arehealthier as a result of PHR use. Lots more learning ahead, I hope!Thanks,Lygeia

Ted, One critical factor in the high utilization of the Kaiser PHR (60% +) is that you have a closed network of providers and almost all the care occurs within the Kaiser system. Thus, patients only have to access one, integrated patient portal.
However, Kaiser is atypical. About 5% of the U.S. population is in a highly integrated system similar to Kaiser.
The more typical situation is a “leaky” network of providers — patients go across systems to doctors and hospitals with few boundaries. The Medicare ACO model is built on the premise of patient choice being more important than network and referral management.I’m already hearing anecdotes “My mom has 5 patient portals with 5 different doctors”. This non-integrated, tethered PHR clearly is suboptimal and IMHO is destined for failure.

NOTE: This is a verbatim copy of a comment made on another post in error, intended for this post; this comment made by http://twitter.com/VinceKuratis.

Ted, One critical factor in the high utilization of the Kaiser PHR (60% +) is that you have a closed network of providers and almost all the care occurs within the Kaiser system. Thus, patients only have to access one, integrated patient portal.
However, Kaiser is atypical. About 5% of the U.S. population is in a highly integrated system similar to Kaiser.
The more typical situation is a “leaky” network of providers — patients go across systems to doctors and hospitals with few boundaries. The Medicare ACO model is built on the premise of patient choice being more important than network and referral management.
I’m already hearing anecdotes “My mom has 5 patient portals with 5 different doctors”. This non-integrated, tethered PHR clearly is suboptimal and IMHO is destined for failure.

Hi Vince,
Thanks for taking a look and your opinion may be humble, but it is important, as is your comment. Allow me to add perspective and see what happens…
My reaction to:
“I’m already hearing anecdotes “My mom has 5 patient portals with 5 different doctors” is:
1. “Wow.” Your observation confirms how far we’ve come. What a great problem to have after 12 years of hearing reason after reason why PHRs will fail. 
2. Challenging question: “Relative to what – having no information from any doctor at all?” See #1 above, and also https://www.tedeytan.com/2012/11/09/11881. Which failure is worse, let me know, and be prepared to defend your answer to Regina :).
My reaction to the critical factors is that Kaiser Permanente, as an integrated health systemis actually not closed, with many many members receiving care at places like the VA, and in this part of the US, at non-Kaiser Permanente hospitals. 
The critical factor is the connection to people rather than a connection to data. When I say “connection”, I mean to a trusted person who answers your secure email on time, provides you with your health information, and uses a portal as a tool to involve you and your caregivers in your care, with involvement defined by you.
If you look closely, this is the common feature of successful PHRs, and there are many out there today – as I mention above, David Bates thinks the linked approach is the way to go.
I think we can’t get where you want to be until we are where you are right now, which is why your comment is so important. If the information is never available, people won’t be able to envision integrating it. Before we get there, though, there are a lot of things PHRs are going do do to change health care,
Ted
(how did I do, is YHO changed at all?)

tedeytan  Ted, Thanks for getting my comment into the right blog post :)Thanks also for your good words and your great blog post exploring these important issues.
As a community we don’t have the right terminology in health care to understand where portals will fit in — we need to be able to analogize to other industries that have gone through platform adoption trial and error. Many lessons to learn here.I have my own blog post on this topic in the works, but I’ll provide a quick preview.Let me first ask a question and bring in some new terminology. Will patient portal platforms have a tendency toward single-homing or multi-homing?Many platforms tend toward single-homing, i.e, we prefer to have one platform “home” as users. Examples: your ISP, your cell phone carrier, your cell phone hardware, cable TV. There is no reason you couldn’t use more than one platform, but we tend to optimize on one platform. The benefits of a second platform are minimal and the costs are high.Other platforms are amenable to multi-homing, i.e, switching costs are low and our workflow can be consistent with using multiple platforms. Examples: credit cards, magazines, securities brokerages, network TV.Consumers will have TENDENCIES to single home or multihome on different types of platforms, based on their evaluation of cost/benefits. These tendencies have important implications for platform evolution in different industry segments.I bet we’d agree quickly that physicians have a strong tendency toward single-homing — they want ONE platform to integrate workflow for all patients, not different platforms for different insurance plans, ACOs, whatever. I can’t imagine a physician saying that 5 EHR platforms are better than none.Patients are a little trickier…and we really don’t know the final answer.So let’s analogize to a social networking platform. Facebook works as a single-homing social network platform because ALL your friends are there — you don’t have to spread your personal workflow across multiple platforms. Network effects are strong and switching costs are prohibitive.If your friends were spread across 5 different platforms, would it even make sense to bother with using a social networking platform? My answer is no — because creating value in the use case is dependent on single homing on ONE platform, and the use case JUST DOESN”T WORK across multiple platforms.Similarly, my take is that patients will not bother with 5 platforms for 5 physicians/hospitals. Consider a patient with multiple chronic conditions — who would bother to collect AND integrate data from multiple platforms? Extracting value from the use case is dependent on single homing on one platform.I have to admit that you might be right — that multiple patient portals MIGHT be better than NO patient portals — but my read is that the tendency here strongly will be toward single homing. Thus, I view the evolution of multiple patient portals as a detour toward the final answer (the optimal model) of one patient centric portal.I’ll suggest that Kaiser has achieved this end of one patient centric portal. It works for the patient — even across care providers as you point out with the VA example. Kaiser and VA wisely have agreed to have interoperable data exchange. It works because you have one patient portal platform in one highly integrated delivery network.The open/closed issue needs to be looked at on multiple levels. You (Kaiser) are very open when it comes to data sharing and integration, but I’ll still maintain that your provider network is relatively closed — at least as compared to most of the US. Again, I have another blog post in the works with more detail.I’m still left scratching my head over the studies you write about. The inconsistent results are hard to explain and for now probably cause more confusion and slow down adoption.

VinceKuraitis tedeytan Vince,
I think the blog post(s) you’re working on are going to be useful and I look forward to them. They build on the current less-than-ideal situation you describe, which is way-more-ideal than what you might have commented in 2002 on this issue :).
Re: slowing/quickening adoption, and inconsistent results, again, in my post above I talk about the results to be measuring and what our patients want. With that in mind, we’ll be able to explain them eventually and I’m not sure adoption is going to be slowed down. 
Glass 3/4 fully yours,
Ted

Adding a response I received from Ted Palen, MD, the study’s primary author, regarding my question about the control group:
“We did not link the index date to a peak of clinic activity, we did match between year of index date and date of kp.org activation between the two cohorts. But your suggestion is a good one. The study raises many questions and I have a lot more investigations to do, stay tuned.Ted E. Palen”

Ted Eytan, MD