Communicating Scientific Data to Heterogeneous Patient Populations – Health Numeracy

June 2011

June 2011 Presentation


Those who missed last week’s Journal Club were deprived of the verbal sparring between Polites/Lewis in the left corner and Theodoro/Griffey in the right corner. We’ve provided a recap of the fireworks, along with an revised and improved version of the Power Point slides provided at Journal Club last Thursday for those who could not attend. For those who want to read more about this topic, there is a list of references at the end of the attached Power Point slides. Thanks to Mary Politi and Kim Kaphingst from the Department of Surgery. Both provided useful background information and a body of literature to begin discussing health numeracy among our heterogeneous ED populations. Mary has also tuned me into the fact that EM-specific research (still unpublished) has begun at the Mayo Clinic with chest pain risk-stratification (Erik Hess) so we’ll be seeing more of this topic in coming years. thought that this article from ACEP News was apropos to this discussion. Physician: Patient Involvement Key To Realizing Full Benefits Of Health Reform. In an op-ed for USA Today (6/8), primary-care physician Kevin Pho, MD, writes, “Team-based care can improve quality of care by emphasizing preventive medicine, improving communication between providers, and facilitating better management of chronic disease.” In fact, “pilot studies show provider teams save money through better coordination, which reduces duplicate procedures and expensive trips to the emergency department.” Pho concludes, “We will realize the full benefits of health reform only if patients are informed and involved every step of the way.” Shared Decision Making Communicating with Patients in the Emergency Department Goal: Take the diagnostic or therapeutic effect size (sensitivity/specificity, number needed to treat) and describe at least one technique to verbally or graphically illustrate the data for both the low and high health numeracy population. Use the recommended article for each class assignment below. These objectives are not easy and very little has been published about what communication techniques work in ED (or any) patients. Nonetheless, this is why we conduct Journal Club and maintain an awareness of research findings. If we fail to incorporate patient preferences into our clinical decision making, then we are not practicing Evidence Based Medicine as originally described by David Sackett and Gordon Guyatt. As our Washington University group discussed patient-communication skills, several legitimate concerns were raised – all of these concerns have been lumped into 5 categories and are summarized below. When ED leaders are contemplating whether or not to invest the time and resources to develop mechanisms for ensuring that patients have the opportunity to understand test- and treatment-related effect sizes, the following issues should not be ignored, although readers should realize that none of the critics could provide any supporting literature to back the alternative suppositions and no alternative ideas (other than what clinicians are already doing) were offered.

Do we need to develop pictorial decision aids to facilitate patient comprehension for every decision that we make in the busy ED? Not all diagnostic/therapeutic questions are equally important for patients or providers; some minor issues (for example chest tube versus needle aspiration of pneumothorax) are probably not worth investing the resources to develop educational patient materials. [no evidence to support] How then does one label a test or treatment as less important on the information hierarchy for EM?

  • “As the test or treatment in question become more complicated (or) has secondary impacts (such as testing for PE or applying results of stress tests in low risk ACS patients) then the need to talk and inform the patient goes up — they need to be more involved in the decision making. Figuring out how to best explain data risk stratification to (such) patients become critical. Same for long term chronic treatments that have little immediate impact (which is not what we usually do in the ED).”
  • As a result of this critique, one of the original scenarios (pneumothorax needle aspiration) below will not be reviewed because the stats showed near equivalence for the the 2 options and all admissions for the chest tube group – so visual representation of these numbers is unnecessary and redundant. Emergency Medicine currently has no consensus (or any data) on what clinical topics clinicians find most important and amenable for sharing detailed information with patients as part of shared decision-making. Therefore, one future direction of health literacy/health numeracy research must be to define a finite number of conditions that the majority of EP’s accept as sufficiently important to merit enhanced communication efforts. The nomenclature, methods, and internal/external validity of such lists have yet to be developed.

Do emergency physicians have time to explain complicated research information upon which medical decisions are based? In the ED where (physician) job security depends upon patients per hour and RVU’s-generated, researchers and patient communication advocates need to develop methods that do not slow physicians or require equipment/materials not routinely available. Pre-developed patient instruction sheets that can be printed from the electronic medical record for a handful of important clinical situations (see #1 above) might make more sense. Another suggested alternative to better inform patients without slowing physician workflow was individual viewing stations (TVs) in ED rooms where health care providers can call up videos for the patients that detail the semi-elective procedures that the patients are about to undergo (lumbar punctures, central lines, operations, etc). These videos could be standardized nationally and professional produced (CSI or other medical drama TV show quality graphics and descriptions). And videos could be developed at a variety of age and education levels. [no evidence to support]

What methods should emergency physicians use to convey complicated medical research to patients who may not have a sufficient level of health numeracy to understand statistical interpretations? “Simple verbal explanations are better than the [pictorial] examples given for the more difficult concepts we were trying to explain…And we need to be careful to avoid information overload.” Pictorial depictions of the data may not be any more effective than qualitative descriptors (low risk, high risk) and we only have a few minutes to convey complicated ideas that many physicians do not understand.

  • “Our patients don’t have the chance to review a video of what we told them. Their retention after discharge I’m sure is absolutely horrible. So that’s why I think that the one example that everyone states was too complex was actually the best ( i.e. “If I had 100 of your closest friends in the room and got this CT of your chest and the results came back normal, I would be wrong and 1 of your 100 friends would have a blood clot.” ) I don’t think it makes it any easier by showing them 99 red faces and 1 green one. Is that more exact? Do they count all of the faces?”
  • Those who provided this opinion could find no evidence to support this perspective. In fact, there is a growing body of evidence to the contrary (follow hyperlinks for more details): Houts 2006, Fagerlin 2005, Lipkus 2007, and Fagerlin 2007. Furthermore, our current verbal explanations may be ethically deficient (Moulton 2010) and legally dangerous (Barry 2008). A simple (<6thgrade reading level, pictogram using natural frequencies) that is archived for specific diagnoses on the electronic medical record and reflective of the highest quality/least biased (meta-analyses > RCT > observational trials) research and consistent with existing guidelines (i.e. most of the Washington University Journal Club archives), may protect a physician in a court-of-law if the medical record clearly documents shared decision making with the patient (Barry 2008). However, we need to be cautious because this communication could cut both ways:
  • “I think that in the case of explaining sensitivity/specificity and NNT to our patients – that this is probably best for us to understand more than for us to translate to our patients. If, for example, I told a patient’s family who’s grandmother had a DNI order that if I used BIBAP on 8 people I would prevent intubation in 1 of them (a great NNT) – to a family member that may not sound very good at all. We would then have to try to convince them that this is a good NNT.”

Are there any potential harms to patients by trying to more actively involve them in the decision making process? By attempting to communicate counterintuitive concepts like diagnostic accuracy (which patients usually assume is 100%) and NNT (which patients usually assume is 1), we may unintentionally harm patients by creating unnecessary angst as they worry about misdiagnosis and ineffective treatments that never would have crossed their cognitive threshold if not brought forth by the unwitting physician. Therefore, involving patients in the decision-making process needs to be done on a case-by-case basis, weighing the patients need for information with their capacity to use that information in their medical decision making. (Lipkus 2007, Woloshin 2007, Halvorsen 2007)

Are there any potential benefits to physicians in trying to more actively involve patients in the decision making process? Even if the vast majority of patients cannot understand these concepts regardless of how they are presented, physicians understanding will be enhanced if they learned multiple different ways to express NNT and sensitivity/specificity. In other words, one does not really understand a concept until forced to teach it to somebody else. This Journal Club was essentially an attempt to teach the teachers how to teach – but we need more ideas and objective data about what patients do and do not understand about their medical care in the ED. (Windish 2007, Horowitz 2008)

Sample Case A 68-year-old right-handed woman arrives from home with expressive aphasia and right-sided weakness beginning 2.5 hours prior. The triage nurse astutely initiates the ‘‘stroke pager’’ as the emotional patient and her husband (a construction worker who did not finish high school) are wheeled back to their room. The neurologist arrives in the room at the same time as the emergency physician, quickly communicating with both the patient and her husband to complete the National Institute of Neurological Disorders and Stroke (NINDS) stroke checklist. As laboratory tests are sent and the patient transported across the hall for her emergent cranial computed tomography (CT) scan, the nervous new neurologist and emergency medicine team confirm that no NINDS or ECASS-III thrombolysis exclusion criteria have yet been identified. Laboratory tests and CT imaging are available within 45 minutes, pushing your patient across the 3-hour no-thrombolysis threshold as Neurology and EM concurrently decide to administer tissue plasminogen activator (tPA). However, her husband asks “How well does the ‘clot buster’ administered within 4.5 hours work?” You remember discussing this topic at the July 2009 Journal Club so you quickly review the ECASS-III critical appraisal form for exclusion criteria, Number Needed to Treat, and Number Needed to Harm. Based upon this synopsis which weighed all the available evidence at the time, the best-estimate for NNT is 14 (95% CI 7-255) to result in one patient with a modified Rankin score of 0 or 1 (no deficit or deficit that does not inhibit daily activity) who would otherwise not have had such an outcome. The NNH is 47 (95% CI 39-161). Unfortunately, neither your patient nor her husband understand NNT or NNH and they ask for clarification. Using the data from ECASS-III, you decide to construct a Cates’ plot to depict this data pictorially. To use this site, you need to know the outcome, control event rate (proportion of mRS 0 or 1 in the placebo arm), and odds ratio (preferably adjusted). All of these details are found in Table 3 of the ECASS manuscript. You plug the numbers into the website to produce the following.

Next, you hit “calculate” and get the following for the placebo arm:

And the tPA arm

You also need to carefully explain to your patient what outcome is being measured. In this case, the difference between a modified Rankin scale of 0 (no deficit) or 1 (minimal deficit) and all other outcomes might be the ability to work and/or care for oneself independently at home. Of course, the “benefits” of tPA are only half the story. What about the risks? One can use the same ECASS-III manuscript and the Cates plot website to produce a pictorial depiction of the risk. Using the NINDS definition* of intracranial hemorrhage (which offered the most dangerous portrayal of tPA in this setting), the following Cates plots are produced. *In the NINDS definition, a hemorrhage was considered symptomatic if it had not been seen on a previous CT scan but there was subsequently either a suspicion of hemorrhage or any decline in neurologic status. To detect intracranial hemorrhage, CT scans were required at 24 hours and 7 to 10 days after the onset of stroke and when clinical findings suggested hemorrhage.

The patient and her husband are satisfied with this pictorial description of the ECASS-III data and consent to tPA. Shortly thereafter, she leaves the ED for the Neuro-intensive care unit without any clinical improvement or deterioration. Your Cases PGY I (Needle aspiration of atraumatic pneumothorax – use PGY IV article) The tall, thin patient notes sudden onset sharp right-sided chest pain beginning 3-hours ago. He appears comfortable and has completely normal vital signs (120/80, 64, 14, 36.8, 100% on room air) without any evidence of tension pneumothorax. Chest x-ray identifies a large right-sided pneumothorax without hemothorax. As you prepare to insert a chest tube, your patient asks if there are ANY other management options available. When you mention that needle aspiration is one possibility, your patient asks for more details. PGY II (Diagnosis of PE – use PIOPED II the PGY III article) Ms. Z is a 27 year old patient with 18-hours of pleuritic left-sided chest discomfort in the mid-clavicular line at ribs 4-6 with associated dyspnea, but no cough, fever, edema, or visible rash/trauma. The pain is not reproducible and her ECG is normal sinus rhythm without any old ones for comparison. She takes birth control pills (hence her non-low risk PERC score), is not currently pregnant, and has no family history of venous thromboembolism. You are about to reflexively order a PE protocol CT, but Ms. Z has been reading about the long term health risks of CT-related radiation exposures in Time magazine. She asks you to explain exactly how accurate this test is. Step 1: How ACCURATE is CT to diagnose PE?

Step 2: What does this less than 100% accuracy mean for my chances of having PE? Answer: It depends on whether you have a “negative” (no PE) CT scan or a “positive” (PE identified) CT scan. Note that the Fagan Nomogram is meant to illustrate to the physician and patient how the post-test probability has been determined on a logarithmic scale. It is meant to be accompanied by the verbal description at left (“In other words…”) to help patients conceptualize these numbers using natural frequencies.

PGY III (Steroids to prevent early migraine recurrence – use the PGY IV paper) Approximately ten hours ago a 26 year old female noted the gradual onset of her typical right retro-orbital pulsating headache pattern following scotomata indicating that her migraines were back for another visit. She has no other past medical history including no malignancies, polycystic disease, prior strokes or known aneurysms. She had an unremarkable head CT at age 18, about 2-years after her migraines ensued. She notes no recent head trauma nor do you note any physical evidence of occult trauma. Your physical exam is unremarkable including no evidence of jolt accentuation of the headache. While evaluating her electronic medical record, you note that she has had three prior ED evaluations for migraine headache. Each time she has responded to intravenous dopaminergic agents, but she reports that her headaches recurred twice before following discharge from the ED. She asks whether any prescription medication might reduce the risk of headache recurrence once she is discharged. You remember that steroids may reduce early recurrence and pull up the data from the Journal Club to describe the risks and benefits for her.

PGY IV (Indications for Head CT After Blunt Trauma – use the PGY II paper) After volunteering for a Joplin Missouri tornado relief disaster medical team, you are dispatched without delay to serve in a field hospital in downtown Joplin. The local hospital has not yet reopened to serve the public. You are on night duty in the field hospital when several patients are brought to you from a collapsed bingo parlor, all with head injuries. The patients include: a Canadian who does not appear intoxicated; an American, who appears intoxicated; and a Mormon, who hasn’t been drinking. The American demands a CT, but the 30 year old Canadian wants to know exactly why he needs a head CT. He has a GCS 15, no open or basilar skull fracture, has not been vomiting, and has no amnesia to the event. He does have a deep 3 cm laceration to his forehead that is not actively bleeding. Step 1: How Accurate is the Canadian Head CT Rule to rule-in or rule-out a significant head injury?

Step 2: What does this less than 100% accuracy mean for my chances of having a significant head injury? Answer: It depends on whether you have “negative” (low risk) or “positive” (non-low risk) Canadian Head CT Rule findings. Note that the Fagan Nomogram is meant to illustrate to the physician and patient how the post-test probability has been determined on a logarithmic scale. It is meant to be accompanied by the verbal description at left (“In other words…”) to help patients conceptualize these numbers using natural frequencies.

Journal Club


June 2011