Ideal Medical Practices

Hospitals, Practice Administrators and Clinicians: You Gotta Learn to Love Patient Ratings – Health Affairs Blog

You are increasingly being held accountable for the outcomes of the health care you deliver. Pay for performance; shared savings in ACOs; public report cards…the list of strategies to monitor and measure the effects of your efforts is lengthening. Many of you seem dismayed by the increased weight accorded to the patient experience of care ratings embedded in most of these programs.  Here’s why you should embrace them: The care you deliver cannot improve our health outcomes or even maintain passable ones without the knowledgeable, active participation of us patients and our families.

via healthaffairs.org

There is good evidence linking patients' experience of care to important outcomes.  Here are a couple of examples: 

• Hospitalized Medicare beneficiaries who can't tell who is running the show and perceive conflicting messages from clinicians are much more likely to be readmitted within 30 days than those reporting good experiences. (Boulding 2011)
• People receiving care in practices with good access, less wasted time, and better communication are more likely to report that treatments have helped reduce problems with pain & emotions, have improved markers of chronic conditions, and reduced use of the hospital & emergency room. (Wasson 2006)

We think that if we adhere to evidence-based guidelines our patients will achieve excellent outcomes, but if we also know that if patients experience delays getting a timely appointment they are less likely to follow up for needed care (Lacy 2004), and people who experience in-office waits and delays are less likely to follow through on appropriate care recommendations (Ling 2006).

Clinician adherence to evidence-based guidelines only goes so far and all the best health information technology for disease tracking is of limited use if the health care team fails to communicate effectively with people who struggle to manage their conditions.  Before we write people off as "non-compliant" we can and should ask ourselves if we have inadvertantly created a series of hurdles and barriers that negatively impacts our patients.

These fundamental attributes of primary care are a good starting point:

1. Access:  how easy is it for people to get a timely appointment?  The ultimate answer comes from those you serve - don't just rely on administrative data.
2. Person-focused care over time:  what percent of your patients say that they can identify a 'go-to' medical person?
3. Comprehensiveness: Recognizing that people must manage their own conditions the overwhelming majority of time, what percent of those with chronic conditions report that they understand what to do and have the confidence to manage?

We showed that it is possible to implement tools to measure and address these issues even in solo independent primary care practices. (Moore 2007)

I agree with Jessie Gruman that we should embrace patient experience data as a powerful addition to our understanding of system performance as well as insight into why some people struggle with effective self-managment.

L Gordon Moore

 

Boulding W, Glickman S, Manary M, Schulman K, Staelin R.  Relationship between patient satisfaction with inpatient care and hospital readmission within 30 days  Am J Manag Care 2011 17(1): 41-48

Lacy, N., Pullman, A. Reuter, M., Lovejoy, B. Why we don’t come: Patient perceptions on No-shows. Annals of Family Medicine 2004;2:541-545.

Ling BS, Klein WM, Dang Q.  Relationship of communication and information measures to colorectal cancer screening utilization: results from HINTS.  Journal of Health Communication. 11 Suppl 1:181-90, 2006.

Moore LG, Wasson JH.  The Ideal Medical Practice Model: Maximizing Efficiency, Quality, and the Doctor-Patient Relationship.  Family Practice Management September 2007 pp. 20-24.

Wasson, J. H., Johnson, D. J., Benjamin, R., Phillips, J., & MacKenzie, T. A. Patients report positive impacts of collaborative care. Journal of Ambulatory Care Management, July-September 2006 29(3), 199–206

Strong and weak indicators of quality and outcomes: not what you'd expect

The prevailing quality paradigm starts with the simple observation that we can observe gaps in care when we create an evidence based guideline and assess observed care against expected.  One study says that we meet these expectations about 56% of the time.[1]

This observation leads to the seemingly logical conclusion that we can aggregate gaps in care as strong indicator of quality.  This is important as we assume that more of this quality will lead to improved outcomes.  We know from other studies that quality and cost are out of whack in the U.S. and health care costs are sucking the life blood out of employers, further exacerbating dismal job growth.Because we desperately want improved quality, we must have more of this gap closing work.

The prevailing quality paradigm has a number of flaws.  It is expensive, overly focused on process measures, and may not predict the outcomes we want and need.

Expensive

Delivering evidence based care is a good thing, but the burden of managing the vast array of discrete data elements is beyond the capabilities of the vast majority of EMR vendors in spite of their claims to the contrary.  Data entry is laborious and back end analytics infuriatingly complex, weak, or non-existent. Vendors have jumped into the breach to extract information from electronic records and even dictations but these solutions add significant cost that might not be recuperated in outcomes.

Overly focused on process

With years and years of disease management experience we know that process improvement does not always result in outcomes improvement.  In a time of limited resource we should avoid the added cost and work burden of meaningless measurement.

Weakly predictive of outcomes

Making the case is a recent article from Duke.  Boulding and colleagues used Hospital Compare data to analyze thousands of hospitals, looking at predictors of re-admission.  They found evidence-based gap analysis a weak predictor of outcomes (re-admission in this case).  The strongest predictor was the patients report of good communication.[2]

This study adds to the growing literature that is trying to tell us something:  Patients perception and report of their care is tapping very important attributes of quality.  These attributes appear to have strong correlation with important outcomes.

Good patient experience data can inform hospitals about structural weakness in communication that increases readmission risk.  Patient report on primary care key performance indicators (access, person focused care over time, comprehensiveness, care coordination) can identify structural weakness far upstream.  When we address these structural elements we improve population health in meaningful ways.

This is a measurement burden we can bear. This information is a powerful addition to our understanding of quality and outcomes. The health care industry tends to disparage or only provide lip service to patient-reported outcome measures.  We continue to do so at our and our patients' peril.

 

[1] McGlynn, E. A., Asch, S. M., Adams, J., Keesey, J., Hicks, J., DeCristofaro, A., et al. The quality of health care delivered to adults in the United States.  N Eng J  Med, 2003; 348, 2635–2645.

[2] Boulding W, Glickman S, Manary M, Schulman K, Staelin R.  Relationship between patient satisfaction with inpatient care and hospital readmission within 30 days  Am J Manag Care 2011 17(1): 41-48

Our approach to measuring health care quality is broken

I'm reading through the Massachusetts AG's report on Examination of Health Care Cost Trends.

The report points out many things that are obvious: 

• Payment changes absent substantive changes in care delivery will ultimately fail
• The logical route to improved (and lower cost) care is through care coordination, which is defined in the report as: "care that is managed across care settings, continuous over time, and patient-centered."
  
  • This is a near perfect match with the essential features of primary care:  access, person focused relationship over time, comprehensive services, and coordination of care.
• Current measures of quality do not appear to address the failings of the current system

From page 19 of the report:

"We also reviewed CMS, HEDIS, Mass-DAC, and ACES measures to understand how well providers in Massachusetts deliver care as compared to each other and, where national data is available, how they perform as compared to health care providers nationally. Our review shows that providers in Massachusetts deliver excellent care with little material variation in the quality of care delivered.18    For example, substantially all Massachusetts physician groups performed above the national average on HEDIS process measures. Other measures that we examined, such as CMS hospital process measures, show the same trend: little variation in the measured quality performance of providers, and high quality care from all providers. Based on our review of these measures, there are some differences in provider quality performance and room for improvement in certain areas of performance, but our review does not suggest that any provider performs consistently better or worse than any other."


I paraphrase:  "In spite of the facts that our current market fails to achieve meaningful goals and recognizing that considerable variation exists, our current approach to quality measurement fails to recognize meaningful variation and fails to provide any indication that our system is struggling."

If the current approach to measurement cannot identify variation when variation is known to exist, and if the current approach to quality measurement cannot identify system failure when the system is obviously failing, then the current approach to quality measurement is of very questionable value indeed.

High performing health care is founded on primary care.

Primary care has four cardinal features:
Access
Person-focused relationship over time
Comprehensive care
Care coordination

High performing health care results in better population health outcomes and reduced cost of care by avoiding potentially preventable events

Safran, Wasson and others have demonstrated that people can accurately report their experience with these cardinal features of primary care and both show us great variability between practices and strong indication of system failure:  Wasson tells us that only 35% of a national sample report that they experience the bulk of these positive attributes.

In addition to our national conversation and efforts to change payment, we must talk about the immense momentum behind a failed approach to quality measurement.  Patient experience with the core attributes of primary care appears to be a much more logical starting point.

Measuring and rewarding what matters - Rich Keller from Treo Solutions describes the value of the patient voice

There was a good article in the May 2010 issues of Health Affairs written by Christine Bechtel and Debra Ness from the National Partnership for Women and Families. They commissioned a number of focus groups to assess patients' thoughts on what was most important when considering how care was delivered. The elements they reference are ones that fundamentally define a medical home. They were able to classify the attributes of care that patients seek into 4 categories:

1. Whole person care. Treat the person – not the parts. This means understanding things about the patient that can't be described using an ICD 9 code; e.g. life situation, home environment, personal preferences.
2. Coordination and communication. Patients and care givers want someone to coordinate their care and coach them through the system. Basically, patients want a general contractor responsible for overseeing all aspects of their health.
3. Patient support and empowerment. Patients want to feel aligned and in partnership with their physician. They want to be part of the decision making process.
4. Ready access. This is just basic customer service. Short wait times, email access, appointment availability, etc.

Link to the full blog post here.

Dreaming of the ideal practice - IMP article in Feb 10 Medical Economics

The February 10, 2011 edition of Medical Economics published an article about our work.

Here's a link to the article.

Funny title the journal editors chose.  I guess that given our current payment and policy environment the ideal is still just a dream, but the pressure to truly change cost and outcomes may help make that dream a reality.

 

What do the data tell us about quality measurement? Improvement is more likely with aligned incentives.

Herb Fillmore (Director of Research at Treo Solutions) was looking at the HEDIS data set again.  He figured that on average there would be very little variation in process measures because we all pretty much agree on them (things like "percent of patients appropriately referred for colo-rectal cancer screening").  He figured that process measures should be pretty easy to nail and quality improvement ought to focus on the greater expected variation in hitting outcome measure targets (things like "Percent of patients with diabetes with HbA1c less than 8.").

He was surprised that the variation in process measures exceeded that of the outcome measures. He drilled into the process measures he found process measures with very low variation (prescribing the right asthma meds for kids, ordering tests for patients with diabetes, etc.) and others with high variation (chlamydia screening, alcohol and drug dependency interventions, etc).  [Take a look at Herb's blog post on this.]

What struck me as I reviewed his bifurcated list was the different each process meant to a busy office.  Prescribing a med is fast.  So is ordering a blood test.  Grab the paper, fill out the form, off you go.  Most of the low variation process measures fell into what I'd call a 'low work impact' category.

Achieving high rates of delivery of care on these lower work burden process measures rests on implementation of reliable gap analysis, embedding the gap analysis and intervention into the work flow and training the office staff, coupled with reliable and regular measurement.  Once past the initial learning curve these processes are pretty simple.

On the other hand, the highly variable process measures struck me as leading to potentially significant workflow burden.  Screening for alcohol or chemical dependency might not eat up a lot of time, but working the bureaucratic hurdles to get someone into treatment is a huge burden for lots of different reasons.  Even after the initial learning curve in a practice, each event could significantly ramp up the work flow.

If we want to tackle more the more complex issues we need a payment and policy environment aligned with the work.  This is especially true when we focus on the entire population of care in a practice or group: rather than credit quality for achieving high scores on a handful of metrics that may only touch a fraction of the population we must address the underlying practice functions that support or impede patient outcomes:  access to care, effective communication (continuity and support of self-efficacy), etc.

These underlying issues impact the entire population served in the practice and thus have the greatest potential impact on overall quality and cost of care.  HEDIS provides no insight into these underlying issues so we must look to more innovative approaches like the patient experience data one can obtain through Dartmouth's HowsYourHealth.org engine.

Why is it so hard to get primary care practices to change?

There is a good deal of talk about support for primary care.  That support is demonstrated in AHRQ grants targeting interventions that support independent small and medium size practices, in a blizzard of Medical Home initiatives, and more.  There is also a good deal of concern that these initiatives may prove disappointing unless we can overcome the glacial pace of change in primary and fractional improvements observed.

What can we do to accelerate primary care transformation?  Hypotheses abound.  Some point to disappointingly incremental results of the Learning Collaborative model, others to the lack of aligned payment, still others to overly broad 'change package' that fail to promote fundamental changes. 

Many things have to go right if primary care is to make the transition. 

• We need an approach that invites the participation of all kinds of practices.  Participation costs (time away from patients, travel expense, etc) must be set very low if we are going to include the full spectrum of primary care practices including solo and rural.  The Ideal Medical Practices project demonstrated one path with group learning and practice coaching by conference calls plus online resources, all tied together with a simple measurement strategy addressing the fundamental changes required.  Group learning can break out of major urban centers and meet practices closer to home.  Regional learning can be complemented with conference call and online support.
• Assistance must address the heterogeneous nature of practices.  Coaching support, peer mentoring, and improvement pathways can be nuanced to address developmental stage, baseline capacity and infrastructure, variation in population needs.  Using a measurement system based on the fundamental features of high performing primary care can identify practices in need of further assistance.
• Launch projects in supportive environments:  Teaching transformative change without aligned payment is a waste of everybody's time.  At best we can prepare participants for a future when payment will support the work, but the full work of effective primary care is ruinous in our current payment environment.  Payment aligned with the work is necessary but not sufficient: the financial incentives must flow all the way to the front lines of work. If the aligned payment dissapears into health system general funds and has no impact on staffing and support of the work in the primary care office it is a wasted effort.
• Improve the signal to noise ratio of the measurement paradigm: Flying in the face of numerous studies we have adopted a measurement system holding primary care practices accountable for outcomes for which they are sometimes only fractionally responsible.  The primary care team is not responsible for the agriculture bill supporting terrible food choices, nor is it responsible for the epidemic of obesity and our society's willingness to restrict access to health care services based on socioeconomic status.  We play a role but often not the dominant role in a diabetic person's A1c level.  Why should we beat up practices with bad quality scores when the reason the A1c is because patients can no longer afford to take their meds because they've lost their jobs?  We can and should measure practice behaviors that predict good outcome: access, relationship over time, comprehensive services, care coordination (all easily and inexpensively measured through patient experience of care).

Primary care practices want to change.  Our professional ethics are aligned with the work of effective primary care.  Change happens when the forces are aligned with the desired results.

Physician Groups Use of Data from Patient Experience Surveys

Patient reported experience of care can be accurately measured and can identify important aspects of practice performance.(1)  Patients who report good experience with key attributes are more likely to have good outcomes and cost the health system less.(2)  Key attributes include: good access to care, good communication, support for self-management, good continuity, care coordination.

I love that we're seeing patient experience in more publications but I'm a bit perplexed by this article:

M. W. Friedberg, G. K. SteelFisher, M. Karp et al., "Physician Groups' Use of Data from Patient Experience Surveys," Journal of General Internal Medicine, published online Dec. 17, 2010.

It is an interesting observational study looking at how physician groups in Massachusetts used patient experience data to inform quality improvement.

The study tells us that financial incentives for patient experience measures predicts work on patient experience.  No surprise, but it is good to confirm widely held belief in studies.

The summarizing conclusion seems to be a circular arguement.  Here's the statement:

Groups with more integrated organizational models were especially likely to engage in group-wide improvement efforts, and all groups facing financial incentives based on patient experience reported improvement efforts of some kind.

Here's the authors' definition of integrated organizational models:

For this study, integrated medical groups were defined as those in which most decisions about policies, staffing, and resources were made by a group manager or management team.

So the authors have concluded that group wide activity is predicted by having an organization with a group leader.

Should we push independent or loosely affiliated groups into 'integrated' groups with central control?  I don't think we can make the call based on this study.  Community Care of North Carolina succeeds in improved population outcomes without resorting to centralized practice administration.

We mustn't mistake the means with the end.  The goal is improved outcomes and bending the cost curve, it is not 'centralized administration of practices' unless and until we have definitive studies linking centralized administration with improved outcomes.

(1) Safran DG. Defining the future of primary care: what can we learn from patients? Ann Intern Med. 2003;138:248–55.

(2) Wasson, J. H., Johnson, D. J., Benjamin, R., Phillips, J., & MacKenzie, T. A. Patients report positive impacts of collaborative care. Journal of Ambulatory Care Management, July-September 2006 29(3), 199–206.

Firsthand knowledge and personal relationships: the keys to better outcomes and lower cost of care

A large study of diabetes management just published in Health Affairs (1) provides convincing support for effective primary care.  Rather than the typical US disease management approach - hiring third party disease management companies to work around the health system - the Germans worked through the patients' primary care physicians. 

They achieved impressive results: lower costs and fewer lives lost as well as a host of other improved process and outcome metrics (lower hospitalization, improved rates of eye and foot exam, etc).

According to the study team this "...physicians’ firsthand knowledge of, and personal relationships with, their patients to promote adherence to treatment goals and self-management..." is the essence of the intervention.

One of the things I find most compelling in this study is the evidence that patient experience is important and informative. Good outcomes follow when patients say they get care that eliminates waits and delays, has good communication, takes their personal issues into account, etc.  Our measurement paradigm is heavily skewed away from this essential source of information, a flaw we must correct.

"Enrolled patients were more likely to receive patient-centered, structured, and collaborative care; the largest differences were in the domains of follow-up and coordination, goal setting and tailoring, and problem solving and contextual scale. This contextual scale is from PACIC. It asks, for example, questions on how doctors or nurses consider patients’ cultural values in therapy planning and make sure that care plans are adapted to patients’ daily lives."

"While US Medicare invested in regional pilots that differ in their structure of care delivery and may use disease management vendors, German health plans decided on an approach with a heavy emphasis on quality assurance and the primary care physician as the program manager. The emphasis is on educating both the patient and the care provider. Characteristics of care considered desirable in a patient-centered medical home, such as coordination, integration, timeliness, efficiency, and effectiveness as well as the patient-centeredness of care, improved markedly."

 We can replicate this work in the US and because of how much more we pay for things here the savings potential is much larger.

To replicate this work we need:

• A payment model that supports the full scope of excellent primary care:  how about a case mix adjusted population based payment to primary care?
• A measurement model that accurately reflects 'firsthand knowledge and personal relationships': NCQA's PPC-PCMH doesn't go anywhere near this stuff.  How about HowsYourHealth.org or the Patient Assessment of Chronic Illness Care instrument?

(1) Stock SAK, Drabik A, Buscher G, et al.  German Diabetes Management Programs Improve Quality of Care and Curb Costs.  Health Affairs 29, 12(2010): 2197-2205

Small and solo practices can work together to prove and improve quality and participate in ACOs

The Ideal Medical Practices project used Dartmouth Professor John Wasson's HowsYourHealth.org tool to measure key attributes of primary care.  This broad denominator approach gave us meaningful data down to the level of individual clinician - a problem considered insurmountable by many of the current medical home projects.

The individual practices could compare their results against a large (200K+) database and even sort by burden of illness, indicators of socio-economic status, etc.  Practice participants were able to use an online curriculum to close any gaps they might find and collaborate in group learning through conference calls and listserv.

Contrary to those who think that solo and small practices are incapable of delivering high quality care, we found that:

• Solo and small practices would volunteer if engagement were made simple and the cost of engagement reduced to acceptable thresholds
• Meaningful outcome improvement was tied to active participation
• Individual clinician data were meaningful and actionable
• Data could be aggregated so that we could look at low-incidence issues of interest at the population level

We were in essence a practice-based research network with pooled data.  We could expand on this approach to understand the impact of our work on congestive heart failure or any other condition whose incidence can only be tracked for populations aggregated across many practices.

We could become an Accountable Care Organization able to demonstrate meaningful outcomes.

The necessary ingredients are:

• A large enough group of practices willing to work together on a common quality platform
• A payer willing to be flexible in the MEANS to the end. 

Our path to high quality primary care is different and we believe more compelling than the approaches gathering most of the press.  We are willing to be held to high standards but are not willing to pursue endless trivia that is tangential to high quality primary care.

To be specific: NCQA recognition is not the goal.  Meaningful use criteria are not the goal.  The goal is to improve population health, improve the experience of care, and in so doing achieve a lower cost per capita or a meaningful reduction in health care cost trends.

A group of practices serving a large enough population could be held accountable to the rate of 'preventables' (ED use, hospitalization, re-hospitalization), to total cost trends, etc.  A payer can use claims data as the overall assessment tool.   The group could use the HowsYourHealth tool as engine for individual practice accountability and improvement.

The outcomes are important.  Be flexible in the approach.  Allow inventive means.

 

P.S.  I'll be talking with Gregg Masters of ACO Watch today on blogtalk radio.  11A PST.

 

Listen to internet radio with ACOwatch on Blog Talk Radio