Whose Evidence?

“You choose the way you die.” David C. White, MD, PhD.

Due to a recent change in circumstances, I have a new health insurance plan. I am looking forward to meeting my new primary care provider with all the mixed emotions of a participant in an arranged marriage. Although I do not know my new PCP, I know something of the incentives that drive her. With a few notorious exceptions, physicians are highly motivated to do the right thing for their patients. The right thing is the intervention that will result in an improvement in the indicators that doctors use to predict quality and length of life. The provisions of Meaningful Use, Clinical Quality Measures, and Physician Quality Reporting System are attempts to codify that basic beneficence. These programs specify objectives and measures that are derived from scientific studies and consensus statements. For example, if my doctor diagnoses high blood pressure, she will strive to achieve and maintain a blood pressure under 140/90. She will be rewarded by financial incentives, or the escape of financial penalties. Along with lifestyle changes, the medications she chooses will be partially based on my race.

Where does evidence come from? Peer-reviewed medical journals contain various types of articles whose conclusions merit various levels of trust. Curiously, the hierarchy of medical evidence is itself in dispute. The term “gold standard” is sometimes applied to the randomized controlled trial (RCT). In a RCT, the study design, null hypothesis, and statistical analyses are set prior to the beginning of the study. The null hypothesis is a statement that is constructed in anticipation of being knocked down. Usually, it states that there will be no difference between groups and/or that the treatment under study will make no difference. The researchers aim to demonstrate findings that would be very unlikely, were the null hypothesis true. In other words, researchers seek a statistical basis for rejecting the null hypothesis. Participants (subjects) are randomly assigned to treatment groups, and sometimes to sham-treatment groups. Neither subjects nor researchers should be able to discern to which group an individual belongs, until the code is cracked at the close of the study. Alternatively, subjects are stratified based on certain characteristics and subjected to the same treatment. At times, a study is prematurely terminated if the researchers determine that one group is at risk of significant harm.

In contrast to the RCT, the consensus statement is an opinion hammered out by a group of authorities in the field. These individuals are well-versed in the relevant research, but unavoidably influenced by their clinical experience as well. Once respected, expert opinion is now relegated to the bottom or near-bottom of the hierarchy. The body of accumulated evidence can also be summarized in a meta-analysis, wherein several RCTs of the same condition are discussed and subjected to statistical analysis as a large pool of data. Unfortunately, methodological differences among studies make this process difficult. Nevertheless, some consider the meta-analysis as the 24-carat gold standard, of higher quality than even the individual RCT.

Despite its position in the hierarchy, the RCT and its relatives have limitations. A good write-up includes an acknowledgement of those limitations. The probability of a study making an impact, or of seeing publication at all, is strongly correlated with the degree to which it challenges the prevailing wisdom. This phenomenon contributes to the public perception that experts are constantly changing their minds. Of necessity, the popular media must report studies that capture their readers’ interest, and create headlines and summaries that hold that interest. Key points in the study may be sacrificed thereby. From the study itself, researcher bias or outright deception may temporarily contaminate the body of evidence. Fortunately, findings must be replicated to ensure long-term acceptance.

On a more basic level, no one can anticipate, much less control, for all the variables. The perfect study would consist of an infinite number of groups, each composed of 1 person. Endometriosis was once thought to be limited to thin, white, nervous, childless women. Racial profiling in medicine, now evidence-based, still has its pitfalls. What is race?  In a homogeneous society, race-based evidence would have little relevance. In a mobile heterogeneous society, a patient is mixed-race until DNA data prove otherwise. As it happens, DNA data can provide a more fine-grained basis for decision-making. Subjects can be stratified based on known genetic make-up.

My North American culture prides itself on rugged individualism. Everyone wants to believe he or she is an exception to the rule. Patients complain about “one-size-fits-all” prescriptions, and doctors inveigh against “cook-book medicine.” However, we don’t have to abandon evidence in favor of anecdote. Personalized medicine, based an individual’s genome, may be the solution. For example, testing for BRCA mutations has enabled a small percentage of women and a smaller percentage of men to minimize their risks of breast and other cancers. Unfortunately, the vast majority of those destined to develop breast cancer are negative for BRCA mutations. Genetic differences in estrogen metabolism can guide risk reduction for that majority. Knowledge of Apolipoprotein E4 status may motivate lifestyle changes that could delay onset of dementia. A test for LPA-Aspirin Genotype can help a person decide whether to take daily low-dose aspirin.

If we choose the way we die, we also choose the way we live. Faulty short-term decisions affect not only lifespan, but also the quality of life and the agony of death. Doctors are motivated to herd their patients into the boxes prescribed by governmental and professional organizations. Patients may value the stakes differently. At any rate, more individualized evidence will give us the pathways to better choices.


Why Don’t We Get Engaged?

“I agree, Dr. Xyz, the patient does bear a striking resemblance to Rodney Dangerfield.” As a medical student, I read this actual entry in a medical chart with stunned disbelief. I still wonder how the patient would have reacted. In those days, patients had no access to their records. It is now past time to start showing them some respect.

Of all the pieces in the healthcare puzzle, patient engagement is arguably the most crucial. Without it, innovation counts for naught. Without patient understanding and buy-in, we labor in vain; medical and surgical devices, miracle drugs, evidence-based practice, genomic medicine, and health information technology serve only to increase costs, rather than life-span or quality. Physicians condemn the “noncompliant” patient; CMS recognizes that compliance is impossible without engagement. Accordingly, several of the Stage 2 Meaningful Use Core Objectives target patient-specific education, clinical summaries, and the ability to access health information online. But, there’s more: at least 5% of patients must demonstrate actual use of the available online information.(1)

In a Wall Street Journal article, Dr. Hardeep Singh recommends several fixes for the costly and dangerous phenomenon of misdiagnosis. “Finally,” he suggests,”patients must start keeping good records of each meeting with a doctor, bringing the information to subsequent medical appointments and following up with the physician if their condition doesn’t improve.”(2) “We’re not expecting the patients to have the expertise of the doctors,” says Dr. Richard Schilsky in a Medscape interview, “But the patients have a lot of expertise about themselves. . . The patient should be encouraged to convey that information to their healthcare teams.”(3) The triumvirate of Electronic Health Record (EHR), Personal Health Record (PHR), and patient portal is a potential solution. Patients can record symptoms, medication reactions, visits to providers, and other health information in the PHR. Through the portal, they can download information from the provider’s EHR into their PHR, or upload information into the EHR. At least, that is the dream configuration. Unfortunately, real systems seldom live up to their potential.

It is difficult to obtain a current estimate of PHR penetration. In 2012, Paul Cerrato quoted a figure of 10%. He suggested possible reasons for the low acceptance rate, exemplified by the demise of Google Health. Google’s explanation was providers’ unwillingness to release data. Cerrato countered that people just don’t care enough about their health. Online rejoinders blamed poor interface with EHRs, poor usability in general, technophobia among the elderly, inconvenience of logging in, inability of patients to understand medical language, fear of loss of privacy and misuse of data, and insufficient financial incentive. (4) “Google’s PHR was built on a universal exchange language, but those with which its creator wished it to exchange data were not,” wrote Nicholas Terry. Google “failed to convince its users that the product had any . . . value to them.” (5)

I think patients can learn to love being participants in their own healthcare. Laymen commonly search for diagnoses on the internet. My friends upload data from their fitness devices to Facebook. It shouldn’t be hard to use a portal to exchange information from EHR to PHR, or in the other direction. Do you know anyone who looks forward to filling out the same answers on different medical forms, visit after visit? Isn’t it easier to remember one more password? So, how do we tear down these barriers to online communication?

The ability to download a copy of the patient encounter or test result is not sufficient. Data do not equal information does not equal knowledge. If my downloaded ANA result is positive, does that mean I have lupus? Will I die in pain? What can I do? Standardized languages for electronic exchange of structured data do not ensure communication. The first day of medical school, students receive a list of medical terms to memorize. After that day, they are never the same. A radical culture shift erases their memory of not knowing. The transformation means more than “you say collarbone, I say clavicle.” Communication with patients is akin to international diplomacy. Say “chronic disease.” A doctor thinks diabetes; a patient thinks AIDS. The common phrase “Patient complains of chest pain.” might evoke the response, “I’m not a complainer, doc!” To a patient, the boilerplate “Genital exam was unremarkable.” may sound worse than a comparison to Rodney Dangerfield.(6) What happens to these notations when the patient exercises the right to “correct” the record? From the other direction, when a patient proudly uploads exercise data or anxiously notes a breast lump, who is there to receive and interpret the news? In due course, all of this data will pass to the next provider, along with encounter notes including “pertinent negatives”, meaning a list of things the patient does not have. No matter how interoperable we become, excessive, redundant data create walls as impenetrable as the “silos” we decry today.

If we are to accord due respect to our patients and our colleagues, we must ensure that a parallel human infrastructure is laid down along with the Nationwide Health Information Network. True patient engagement is not possible without explaining, prioritizing, and humanizing all the available data.

(1) Eligible Professional’s Guide to Stage 2 of the EHR Incentive Program, Centers for Medicare and Medicaid Services, January 1, 2014

(5) Terry, Nicolas P. (2013) “Information Technology’s Failure to Disrupt Health Care,” Nevada Law Journal: Vol. 13: Iss. 3, Article 6.

(6) A comedian who complained he got “no respect”.

Go Deeper

It’s not difficult for a hiring manager to figure a candidate’s age. High school graduation dates, years of experience, linguistic styles, photos in social media, and many other clues abound. If you are using these as short-cuts to delete your way through an inbox clogged with resumes, you will never know how much your company has lost. Perhaps you fear that someone outside your target hiring demographic will retire at an inopportune time or develop health problems limiting productivity. The older employee may mesh poorly with the company culture. Your concerns are valid. However, consider this: a 50-year-old with 30 years of experience has more than likely made and learned from twice as many mistakes as a 35-year old with 15 years of experience. A digital semi-native who has followed technology since the 1970s may have deeper knowledge than a native born in 1987. Work ethic and ability to profit from constructive criticism vary greatly across individuals, but years of brutally honest feedback have a tendency to harden one’s core and thicken one’s skin. Interview techniques should help you distinguish the individual who has been honed into a workplace superhero.

I have never made age a criterion for my hires, and I have never had to fire anyone over 50. Other companies’ losses have been my gains. One 60-something brought with her enough creativity, wisdom, skill, and compassion to turn the place around. Don’t miss this opportunity to add value to your company.


“Over my dead body.” That was one surgeon’s reaction to electronic medical charting. And he is only one of many.

Health information technology is built on the premise that healthcare can become more cost-efficient, timely, safe, effective, equitable, and patient-centered by harnessing technology for the benefit of the patient. From my perspective, patients are cautiously optimistic. Physicians on the other hand, are frustrated, furious, and frightened. They are frustrated at inability to navigate through patient charts, furious at lost orders, and frightened that patient care will suffer when the focus shifts from the patient to the computer.

The goal of health information technology (HIT) is to build a nationwide health information network (NHIN). Patient health information, such as chronic conditions, medications, allergies, medical history, surgical history, and family history, could be instantly and securely accessed when and where it is needed. HIT means never having to say, I’m sorry, I don’t remember my meds, or, I’m sorry, but I refuse to fill out any more forms! It means never having to depend on mailed or faxed patient records. The NHIN will be constructed of many components. The building blocks that doctors feel forced to play with are EHR (electronic health records), CPOE (computerized physician order entry), and CDSS, (clinical decision support systems).

EHR, also known as EMR for electronic medical records, holds the promise of seamless exchange of legible data between providers. The downsides include increased time and effort required to write a note. Logging into the system, opening up a patient’s electronic chart, navigating through the sections, and entering the data to fit an electronic master’s prissy requirements, all take more time than opening a paper chart and jotting down a note by hand. Insufficient bandwidth can further slow the process. As reimbursement is steadily cut, doctors must see 4-5 patients per hour in order to maintain financial viability. However, the demands of electronic charting siphon time from the ultimate purpose of the patient visit. In addition to the cost of the system itself, physicians experience loss of income (“productivity”) from the extra time required to enter patient information. Furthermore, the ease of cutting and pasting generates long, repetitive notes. For any future encounter, the time required to glean information from today’s note will only add to the total time unavailable to the patient as a person. Worst of all, erroneously-entered verbiage becomes truth by the process of repetition from one note to the next.

Computerized physician order entry is fraught with hazards and frustrations. Though orders are now legibly and directly transmitted, some of the safeguards are gone. In the past, nurses and clerks who questioned physician orders served as backstops for errors. Clinical decision support is designed to replace this function by alerting physicians to allergies and drug interactions, but it is difficult to build nuance into a database. Frequent and irrelevant computer alerts cause “alert fatigue” and distract from the concentration providers need to make decisions. Repeated demands to override the alerts and justify the overrides divert more cognitive energy from the task at hand. Computer-facilitated errors do occur. Sometimes a human has to look at the big picture to see the obvious, such as a tightly-packed drop-down list causing an adult dose to be ordered for a child. There is many a slip twixt the list and the lips.  During the transition from paper to electronic information systems, unanticipated situations have led to patient deaths.

Physician resistors point to the poor usability of HIT systems. Most technology is inherently appealing. We didn’t have to be incentivized to buy a cellphone or a tablet. In contrast, physicians are cajoled and compelled to use systems that are cumbersome for them and potentially dangerous to their patients. Any Medicare provider who is not on board by 2015 will see a cut in income. Participatory design has been used to create usable products in other areas of technology, but user participation in HIT design is low. Physicians cannot spare the time to participate in usability studies. Healthcare organizations cannot afford to scrap their multimillion dollar investments. Some software engineers think physician-led design is the worst thing that could happen! However, many people on both sides of this divide are dedicated to bridging the gap.

Resistors in electrical circuits can be used to lower voltage, produce heat and/or light. Physician resistors have a valuable role in the metamorphosis from a paper system to an electronic infrastructure. The free-text message they send is clear: we have a long way to go! Even so, we are learning rapidly. Once a danger is recognized, it is immediately addressed. Progress from blood-letting to robotic surgery took two hundred years. Refinements in health information technology are making it safer and more usable every day.