No Relationship Between Primary Care and Deaths from Coronary Artery Disease?

Listen in on the average health care Conference, Meeting or Grand Rounds with health care patriarchs or D.C. brainiacs speaking about "primary care," and it's likely that you'll hear at least one PowerPoint slide's worth of praise for PCPs. It reminds the Disease Management Care Blog of what it's read about about the hollow admiration of the Saudis for their hardy Bedouins, the Norwegians for their rugged Laplanders and the Ben and Jerrys for their Vermont dairy farmers. Sure, these special folks may deserve special regard, but no one actually wants to live in a desert, above the Arctic Circle or close to a smelly barn. All the special attention is of small comfort us PCPs who just can't seem to get any respect.

And now this study isn't helping at all. Titled "Association of Features of Primary Health Care with Coronary Heart Disease Mortality," it was published in the November 10 issue of JAMA. Louis Levene, Richard Baker, John Bankart and Kamlesh Khunti accessed ALL of England's health data on mortality, patient characteristics and primary care delivery to explore the relationship between heart disease deaths and the local availability of primary care. This was an impressive study involving 54.3 million Brits being cared for in 152 of their regional "trusts." Regression analysis was used to determine if there was any statistical correlation between deaths from heart disease versus race, socioeconomic status (a "multiple deprivation" index), tobacco use, weight, diabetes, high blood pressure, number of physicians, staff and practices and pay for performance bonuses that were awarded to the docs.

Based on common wisdom, most persons would think that all of the above factors are independently and statistically significantly correlated with death rates. Thousands of DMCB readers know otherwise. That's because they know primary care is not the panacea for all that ails health care and that the DMCB likes to look at published research that goes contrary to the common wisdom.

It turns out that only five characteristics were associated with heart attack death. The risks were 1) being white, 2) having a low socioeconomic status, 3) using tobacco and 4) being diabetic. Having high blood pressure detected (this is not the same as having high blood pressure: patients can have hypertension that goes undetected) was associated with less risk.

What turned out to have no impact was availability or the quality of primary care. According to the authors:

"Neither provision of primary health care as indicated by the numbers of physicians or staff per 100 000 population, nor clinical performance as reflected by the quality and outcomes framework indicator scores predicted mortality in any year."

This should give pause to the universal assumption that manufacturing and then parachuting more PCPs into areas of the U.S. with a high burden of coronary artery disease and showering them with pay for performance dollars will save lives.

As an aside, that was also the conclusion of this recently published RAND review paper from Health Affairs titled "Primary Care: A Critical Review of the Evidence On Quality and Cost of Care." The authors state:

"....a recent survey of primary care physicians found that those in the highest-spending regions of the Dartmouth Atlas of Health Care were significantly more likely than those in the lowest-spending regions to report more aggressive use of discretionary visits, tests, and interventions. In other words, care by primary care physicians in high-cost areas is not the same as care delivered by primary care physicians in low-cost areas. This finding ....suggests that adding more primary care physicians in regions such as South Florida may increase mortality rates. Such analyses offer evidence that adding more primary care providers in high-spending areas could have deleterious effects if local provider cultures and other system-level characteristics are not simultaneously reoriented."

Of course, there are other studies on the topic that say otherwise and England isn't the same as the United States. Time will tell.

One last point: the good news about the JAMA paper is that the detection of high blood pressure seems to pay off. If population-based detection and pursuit of hypertension is what counts and it's independent of the availability of primary care physicians, there are other population-based solutions available. It's called disease management. Examples are here and here.

Image from Wikipedia

Hypertension, In-Home BP Monitoring, Web-Based Reporting, Phamacists and Press Releases: Read the Studies for Yourself

Sometimes the Disease Management Care Blog fancies itself as a virtual "journal club" that simultaneously combines updates on medical research with healthy levels of skepticism about the underlying science. It's a useful way to think through much of the spin that can clutter breaking news about medical advances. Given our mainstream media's spotty track record when it comes to reporting and interpreting medical news, wary journal club approaches are needed more than ever when it comes to uncovering the facts and only the facts.

Case in point? Check out this Kaiser - Microsoft press release touting a "50%" improvement in blood pressure control thanks to "in-home BP monitors and web-based reporting tools that connect clinicians and patients via the internet." The study of "348 patients" was presented at an American Heart Association meeting.

It goes on to say:

At the start of the study, the average systolic blood pressure was 149 mm Hg in the home monitoring group and 145 mm Hg in the usual care group. At six months, patients in the home monitoring group were 50 percent more likely to have their blood pressure controlled to healthy levels compared to the usual care group. Similarly, a significantly greater decrease in systolic blood pressure at six months occurred in the home monitoring group (-21 mm Hg) versus the usual care group (-9 mm Hg).

The DMCB thought this sounded quite impressive, so it pulled the AHA meeting abstracts and went to page 82. According to the report, "A Pharmacist-led, AHA Heart 360 supported Home Blood Pressure Monitoring Program Improves Blood Control in Patients with Uncontrolled Hypertension," the experiment involved a total of 353 (not 348) hypertensive patients who agreed to be randomly assigned to usual care (N=174) or an intervention group (N=179). The intervention really involved a pharmacist-led program that adjusted the patients' medications in response to home BP readings that were uploaded into a personal health record HealthVault.

Six months later, data was available on 125 of the usual care patients and 120 of the intervention patients. 47 of the 125 patients, or 38%, in the usual care group vs. 69 of 120, or 58%, of the intervention patients reached target blood pressure levels. This difference turned out to be statistically significant. After some statistical adjustment, the 20% absolute advantage (58% minus 38%), works out to represent an approximate "50%" relative advantage. In other words, if a Kaiser pharmacist is adjusting your blood pressure pills, you're 1.5 times more likely to get your blood pressure under control compared to usual care.

Or does it? The trial started out with 353 patients but ended with only 245 - which is about a 30% drop out rate. A more conservative calculation would include the drop outs and assume none of them got their blood pressure under control. Accordingly, 47 of 174 or 27% of the usual care group achieved blood pressure control while 69 of the 179 or 38% of the pharmacist led group did so. That's a more modest absolute improvement of 11% or a 1.4 relative likelihood of control. The DMCB went to this web site to do a quick chi square and thinks the results had a more modest yet still statistically significant p value of .03.

So what does the DMCB think?

Never take a press release's word for it: always go to the original report and read it for yourself. In this instance, there were two important features that were not stressed in the press release:

1. The secret sauce behind study was really being "pharmacist led" (the title of the study), not home based monitoring and web-based reporting. In that context, this isn't really all that new: it's abundantly clear that non physicians can improve care quality. To sort out the added value of the Kaiser-Microsoft alliance, we'd need a trial that compares pharmacist-directed care using the old fashioned telephone vs. pharmacists using a web-based personal health record.

2. Learn to look at the numbers: in this instance, a casual read uncovered a 30% drop out rate. By focusing only on the patients that remained in the study at 6 months, the authors cast a more favorable light on their success rates. In addition the "1.5" relative likelihood of improvement is technically correct but of less use to doctors, who have to worry about the patients who may not keep appointments. They're more interested in the 11% absolute improvement rate.

Two New Studies Show Telephonic Disease Management Works

When skeptics think about 'disease management' (DM), they think about distant and nurse-filled cubicle farms that put unsuspecting patients through a speed dial version of education-lite. Plumbing the depths of these telephonic knockoffs, critics have made it abundantly clear that don't like what they see: a pseudoscience that confounds patients and antagonizes physicians. They're fed up with the lack of financial provider incentives, lagging technology and ineffective leadership support. It's so bad, 'how to' articles like this have become necessary to help address the physicians' loss of prestige, influence over patient care and income. Policy makers, academics and organized medicine groups have all agreed that the outrageous vendor fees could be better used for other stuff. Like vaccinating upper-class suburban tots. Or paying for motorized wheelchair scooters for affluent octagenarians. Or increasing primary care physicians' fees.

A pox on disease management you say? Stone them?

Think again. Two important publications in the mainstream peer-reviewed medical literature suggest that traditional telephonic disease management is quite effective.

The first is this article, in which researchers from the University of Pittsburgh report the results from a randomized control trial that compares telephonic "collaborative care' (CC) (N=150) versus usual care (N=152) for fresh heart surgery patients that were discharged from one of seven Pittsburgh area hospitals with a surprisingly common side effect of their treatment: active depression (1).

The CC nurses provided 'psychoeducation' in the intervention group that increased awareness of depression treatment options. Backed-up by a psychiatrist/internist team, the nurses also facilitated the patients' treatment decisions. The article includes a description of the CC nurses' roles could have been written as a job description by any of the current for-profit disease management vendors:

'Adheres to evidence-based treatment protocols, supports patients with timely education about their illness, considers patients' prior treatment experiences and current preferences, teaches self-management techniques, actively involves primary care physicians in their patients' care through regular exchanges of real-time information, proactively monitors treatment responses and suggests adjustments when indicated, and facilitates co-management or transfer of care to local mental health specialists when patients do not respond to treatment, have clinically complicated cases, or upon request by the patient or primary care physician.'

Using an approach that is quite similar to any typical disease management vendor program, patients were telephoned every other week for two to four months with calls lasting 15-45 minutes. This was followed by a 'continuation phase' with a call every one to two months. Eight months later, various mood tests showed that the CC group had a greater and statistically significant improvement in psychological well being compared to the usual care group. In looking at the graphs from the study, the Disease Management Care Blog was unable to discern any meaningful difference in the overall rehospitalization rate, though it looks (no 'p' value was reported) like rehospitalizations for cardiovascular disease were considerably lower in the CC patients. CC women were also more likely to being taking antidepressants.

The second is this article, where the Duke University primary care clinics randomly assigned 636 patients to one of four treatments: 1) a telephonic bimonthly 'behavioral intervention' that used the patients' perceived risks, memory ability, literacy, educational level and the quality of the doctor-patient relationship to tailor engagement in the DASH diet (N=160) 2) just a home blood pressure (BP) monitoring device (N=158), 3) both education and a BP device (N=159) or 4) neither (N=159). Two years later, there was an absolute 11% increase in the proportion of patients that had blood pressure under control vs. 7.6% in the blood pressure cuff group vs. 4.3% in the phone call only group. There was no impact on health care costs (2).

First of all, the Disease Management Care Blog thinks both studies are an affirmation of what the mainstream DM vendors have been doing for years. While post-heart surgery depression hasn't been a topic of research, telephonic-based DM for depression in other settings has been shown to have considerable merit. As for hypertension, managed care insurers have known for years that BP control in primary care settings is not what it should be. In response, many DM vendors are selling patient engagement programs that promote the DASH diet with or without blood pressure monitors. Based on the Duke study, it would appear that the managed care organizations can expect and have achieved better blood pressure control with hypertension DM.

Secondly, aha, you ask, but are we getting our money's worth? Neither study 'saved money.' If the cost of the nurses was included, both interventions described above would probably be rated as money losing. While that may be technically 'true,' a) neither study followed patients for a sufficient period of time - it can take longer for a pay-off to accrue, b) commercial DM vendors are much better at identifying, targeting and successfully managing the high risk patients with a higher likelihood of excess costs, c) the interventions above were just for depression or hypertension; modern DM vendors are able to fold in additional care management interventions for other co-morbid conditions that can lead to hospitalization or increase costs and d) maybe, just maybe, the ultimate purpose of DM is not to save money but to increase quality of care at a price point that yields the greatest bang for the dollar. In other words, if depression or hypertension is better treated, maybe it's worth it to pay for it. Stick with usual primary care and you get what you pay for.

Thirdly, critics may point out that both studies above originated in physician-owned, operated and led settings. Fair enough, says the DMCB, but it also knows that primary care physicians in large health care systems are not necessarily more loyal to the 'home office' than any external vendor . In fact, close reading of both studies fails to show that the UPMC or Duke nurses were really all that different from any other external care management initiative. The DMCB doesn't believe the location/ownership of the nurses is what's important. Rather, it's what they do and which patients they do it to.

The DMCB has pointed out for years that telephonic disease management is an important option in the suite of services for caring for populations with chronic illness. It's nice to see that there are now two studies that confirm that perspective.

1. Rollman, B, Herbeck Belnap B, LeMenager MS, Mazumdar S, Houck PR, Counihan PF et al: Telephone-Delivered Collaborative Care for Treating Post-CABG Depression: A Randomized Controlled Trial. JAMA 2009;302(19):2095-2103

2. Bosworth HB, Olsen MK, Grubber JM, Neary AM, Orr MM, Pwers BJ et al: Two Self-management Interventions to Improve Hypertension Control. A Randomized Trial. Ann Intern Med 2009 151(10):687-695

USPSTF: Blood Pressure Trumps Blood Glucose & the Implications for Diabetes and Hypertension Disease Management

Persons with blood pressure readings consistently greater than 135/80 should be screened for diabetes mellitus every three years, or so says the U.S. Preventive Services Task Force. When this headline first crossed the Disease Management Care Blog’s twin flat screen monitors, it thought the rationale was simple.

A high prevalence of undiagnosed diabetes among persons with hypertension should be the cause for alarm. Diabetes mellitus is present in about 9% of the U.S. population, its prevalence increases to more than 20% among persons aged greater than 65 years and is also associated with a high body mass index. Since these risk factors tend to cluster together, it makes sense to look for the condition, right? Earlier diagnosis would lead to earlier treatment and prevention of the complications of diabetes, right?

That’s not the case. Instead, the USPSTF logic is reversed and twofold. The first is that the presence of diabetes alters the preferred blood pressure to less 135/80, even if it means starting a lifetime of daily pills. The second is that it makes a difference as to which medicines are used. So it’s not a matter of diagnosing and then treating diabetes, it’s a matter of managing the blood pressure differently.

Don’t be surprised. Several studies including the U.K. Prospective Diabetes Study (UKPDS) and the Hypertension Optimal Treatment (HOT) Study demonstrated a strong link between blood pressure control and macrovascular complications, such as heart attack and stroke, among persons with diabetes. Up to 80% of persons with diabetes will die of macrovascular disease, so blood pressure control is a priority. In fact, it is probably more important than blood glucose control. To give you a sense of this, the very valuable “Number Needed to Treat” (NNT) blood pressure calculation, provided in this excellent Annals article, is an impressive 23 or less for all cause mortality, stroke or heart attack over ten years. Thiazide diuretics or ACE inhibitors (and frequently both, combined with other medications if necessary) should be used initially, versus other first line agents.

Note that there are no prospective studies that show control of the blood glucose level in diabetes changes the incidence of heart attack or stroke. Rather, blood sugar control is correlated with fewer “microvascular” complications, such as damage to the sensory nerves in the feet or kidney disease.

The DMCB wonders how this new recommendation would work for physicians in primary care settings:

Scenario 1: During the course of every one-on-one visit with patients, the physician tries to remember that a blood pressure reading that normally isn’t considered “high” should prompt a check of those past blood tests in the back of the chart and to order a screening test for diabetes, in addition to all the other things that need to fit into a 15 minute office visit. If really well organized, have a flow sheet in the front of the chart. Anyone without an appointment loses out.

Scenario 2: Have a standing order for the office nurse to review the charts’ labs and arrange for a screening test for diabetes if, during the course of the intake, the blood pressure is more than 135/80. Anyone without an appointment loses out.

Scenario 3: During the course of every one-on-one visit with patients, the physician gets annoying prompts from that new EHR that a blood pressure reading that normally isn’t considered “high” is high and a lack of any labs under the 'results' tab should prompt an order for a screening test for diabetes click here, in addition to all the other aspirin, cholesterol screening, mammogram and immunizations prompts – click heres that clutter the screen during the 15 minute office visit. Anyone without an appointment loses out.

Scenario 4: Physician fires up that new and improved EHR registry and uses some if-then branching programming logic to extract everyone with a mean of > 135 OR >80 over three visits in the two BP fields AND absent diagnosis of diabetes (look up the ICD 9) codes AND absent qualifying blood test over the last 365 days x2. Generate form letter to all patients meeting criteria and “blow in” the name and address from the demographic data fields. Blow in a screening lab order on hundreds of patients. The physician and the office staff deal with each one at a time when patients start calling with questions and when the physician needs to actually see the folks with evidence of diabetes.

Scenario 5: Physician tells the office manager or nurse to deal with Scenario 4. They tell the physician (s)he needs to contact the EHR vendor and find out how much it will cost to have this ad-hoc programming done.

Scenario 6: Physician awaits the arrival of a newly developed HEDIS measure for the number of persons with blood pressure > 130/85 (denominator) who have a screening blood glucose level (numerator). (S)he resists the flaky letter from the managed care organization listing patients, many of whom are not recognized, that the MCO believes meet criteria for measurement or intervention so that they can get NCQA accreditation. There is a change of mind when the flaky letter Ver. 2 outlines the terms of a new P4P initiative linked to this measure.

Scenario 7: The physician discusses this with the Medical Home trained staff and instructs them to work with the disease management vendor, who has the mojo to contact everyone meeting criteria after they’ve remotely accessed the electronic patient files in a HIPAA compliant way. Patients meeting criteria are contacted with letters, IVR and eventually live nurses who remotely arrange testing. Patients who have the labs done have been coached, and those with a fasting blood glucose > 126 (have diabetes) and a blood pressure > 135/80 are seen by your Medical Home staff and medications are started and adjusted using a standing order protocol that is safe and effective. Physician is in the meantime seeing sick people, but available if the protocol isn’t working.

Hypertension in Persons Older Than 80 Years: Time for a Disease Management Program

If you believe Li Ching-Yun really made it to the age of 256 years, you can pretty much assume that he didn’t have high blood pressure. He wouldn’t have lived that long and he sure wouldn’t have been able to father so many children. Compared to the elderly in the U.S., he’s the exception; almost three quarters of persons over the age of 65 has hypertension. Because persons who are among the very old (in their 80’s) are the among the fastest growing group in the United States, more are making it into their 80's with high blood pressure. How should they be treated so that they can also lead the happy boomer life of Mr. Li?

The Disease Management Care Blog knows many physicians are reluctant to treat the ‘very’ elderly (often defined as age > 80 years) for hypertension. While clinical inertia and ‘elder’ bias may be playing roles, the DMCB suspects docs have a healthy respect for medication side effects that could violate the 'first do no harm' adage. What’s more, there has been some science supporting the notion that in the very elderly, treating high blood pressure results in trade-offs. The bible of hypertension treatment guidelines, JNC VII, is also silent on how to treat this group of individuals.

Enter the May 1 New England Journal of Medicine. The lead article was a huge international multi-centre (if it were in the U.S., the ever-urbane DMCB would need to describe it as multi-center) prospective, blinded, placebo control study that randomly allocated very elderly hypertensives with a blood pressure > 160 systolic to treatment versus no treatment. First-line treatment was a less commonly used water pill (diuretic) called indapamide. If that didn’t get the patient to a target blood pressure of less than 150 systolic and 80 diastolic, escalating doses of an ACE inhibitor were added. Two years later, data from over 3800 patients showed treatment beat placebo with a 21% ‘relative’ reduction in the occurrence of any death, 64% relative risk of heart failure and 30% relative reduction in the risk of stroke. Side effects were minimal: ‘Only five… events... were classified by the local investigator as possible having been due to the trial medication.’

The DMCB says it may be time to think about a population-based disease management initiative aimed at the very elderly hypertensives. Using the trusty ‘number needed to treat’ (NNT) calculation, the DMCB notes the treatment group had a death-from-any-cause rate of 47.2 per 1000 patient years, versus 59.6 in the placebo group. That’s 4.7 per 100 patient-years vs. 5.9 or really a difference of 1.2%. To calculate the NNT, take 1 divided by the reduction in absolute risk, which in this case is 83. So, the death rate will decrease by just over 1%. You have to treat 83 persons to save one life. The ‘any cardiovascular event’ rates (stroke, heart attack or heart failure) went from 5.1 to 3.4 per 100 patient-years or 1.7%; the NNT is 58. While the NNT numbers are not nearly as impressive as other widely accepted interventions, they are still respectable, especially when the treatment appeared to be so well tolerated and cardiovascular event rates in the very elderly are otherwise high.

In the opinion of the DMCB, the study protocol lends itself quite nicely to a population-based approach. The exclusion criteria not only included a generic contraindication (relying on physician judgment), there was a list of other diagnoses (for example, dementia) that precluded treatment. Once the patient was in the protocol, the steps that followed were very easily implemented outside of the usual time consuming face-to-face physician visits. Standing orders are well within reach that could be followed by non-physician health care professionals with or without home monitoring. Given the low baseline rate of treatment and low expense of the medications (and assuming ACE inhibitors are interchangeable), an efficient disease management program could, based on these data, readily reduce the admission rate from cardiovascular events and demonstrate a ‘return on investment’ or a beneficial impact on trend. Medicare Advantage programs take note. In fact, so should our colleagues in running the Medicare program.

The growing population of Mr. Li's, his wives and children would also appreciate it also.