Wednesday, May 30, 2012

Book Review: Between the Lines

Every year for the past several years, I have served as a faculty group leader for a course in "Evidence-Based Medicine" taught to first and second-year medical students. The course aims to provide students with basic tools to navigate the medical literature that we hope they will retain during their training and use to answer clinical questions long after they enter practice. Unfortunately, the course is a low priority for students in their preclinical years, and many of the epidemiology and statistics concepts we teach are far more advanced than what they will need to know as physicians.
It was a pleasure, then, to read Dr. Marya Zilberberg's Between the Lines: Finding the Truth in Medical Literature, a rare book that bridges the gulf between medical publications and the real world of practicing clinicians. Zilberberg, a physician and noted health services researcher who blogs at Healthcare, etc., distills her expertise from two decades of teaching evidence-based medicine into a concise text that is accessible not only to medical students and other health professionals, but to journalists and educated laypersons who want to look past the latest sensational headlines to uncover what we actually know about sickness and health. The book's conversational tone makes the reader feel as if Zilberberg herself is in the room giving a one-on-one tutoring session.

The book is divided into two parts: "Context" and "Evaluation." The first part was my favorite, containing a collection of short essays with provocative titles such as "Beware of What Seems Too Good to Be True" and "Assume a Spherical Cow." Here Zilberberg exposes the faulty reasoning behind certain health care beliefs shared by much of the general public and a good number of clinicians, as well. For example, a screening test that is touted as being highly sensitive for the condition it detects still may not be worth undergoing, depending on how common (or uncommon) the disease is and how many false positive results it generates.

The second part follows a more standard format for a book on the medical literature, moving logically through a traditional hierarchy of study designs and threats to the interpretation of study results. Zilberberg's writing is clear and straightfoward, and key points are helpfully highlighted in accompanying figures and tables. I recommend this book highly to all students of evidence-based medicine, regardless of occupation or professional degree.

Monday, May 21, 2012

"The best [colorectal cancer screening] test is the one that gets done"

Last week was a busy one for cancer screening. I could choose to criticize the Centers for Disease Control and Prevention's premature guidelines to screen all baby boomers for hepatitis C (which can lead to cirrhosis and liver cancer) or the less-than-sound recommendations of several major cancer organizations to screen present or former heavy smokers between ages 55 and 74 for lung cancer with CT scans. (As I've pointed out before, there are at least 4 good reasons not to reflexively follow the latter advice.) But instead, I would like to explore the irony that another screening test that has conclusively been shown to result in more health benefits than harms is on the verge of becoming extinct in the U.S. That test is screening for colorectal cancer with flexible sigmoidoscopy.

Flexible sigmoidoscopy is an uncomplicated office procedure that requires no anesthesia and was once commonly taught to, and performed by, thousands of family doctors and general internists in adults over the age of 50. But over the past decade, gastroenterologists have done a magnificent job convincing primary care physicians and the American public that colonoscopy is the "gold standard" for colorectal cancer screening. As a result, a recent national survey found that while 55 percent of respondents reported receiving a screening colonoscopy within the past 10 years, only 1.3 percent of respondents reported being screened recently with flexible sigmoidoscopy.

That's too bad, because not only does screening colonoscopy cost a lot more money, it hasn't ever been shown to be more effective than screening sigmoidoscopy. In fact, screening colonoscopy has never even been tested in a randomized trial, and may never be. In contrast, yesterday the lead researchers of the National Cancer Institute's Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (which previously found that prostate and ovarian cancer screening don't save lives) reported that having a flexible sigmoidoscopy every 3 to 5 years reduced deaths from colorectal cancer by 26 percent, even though nearly half of the control group underwent screening at least once. That's a big deal, since colorectal cancer is the third-leading cause of cancer death among U.S. women, and the second-leading cause of cancer death in men under age 75.

In an accompanying editorial, Dr. John Inadomi urges clinicians to reconsider flexible sigmoidoscopy as a preferred cancer screening strategy for three reasons: 1) it's more evidence-based than screening colonoscopy; 2) patients are more likely to be screened if multiple options are available; 3) screening resources are limited, and efficiency matters. He concludes:

Where does this leave us with regard to screening flexible sigmoidoscopy? First, it should be acknowledged that flexible sigmoidoscopy reduces colorectal-cancer incidence and mortality for the portion of the colon that it is designed to examine. Next, high-quality evidence must show the superiority of colonoscopy over other screening tests before we dismiss the use of flexible sigmoidoscopy and fecal occult-blood testing, both of which have randomized, controlled trials supporting their benefit. Especially critical are data that confirm the ability of colonoscopy to reduce mortality from proximal cancers. Finally, patient preferences for screening tests should be identified and respected — in this case, the best test is the one that gets done.

Wednesday, May 16, 2012

Electronic health records: medical progress, not panacea

Yesterday, the family medicine residency program where I serve as a faculty preceptor "went live" with their new electronic health record. They posted a sign at the front desk that read in part: "Pardon Our Progress," as if we were starting a major construction project - which in a way, we were. Instead of wading through stacks of unruly paper charts, my colleagues and I logged on to a sleek online portal via laptop computers to review and sign residents' progress notes. Thanks to months of meticulous preparation and the presence of onsite technical support, the day went relatively smoothly for physicians and patients.

By leaving paper behind, we looked forward to eliminating inconvenience and errors associated with lost charts and illegible or missing documentation. But the most important reason for the switch to an electronic health record was the unspoken presumption that it would allow us to provide better preventive and chronic care for patients. At the end of the afternoon session, I asked the residents how strong the evidence is that practices with electronic health records actually improve their quality of care.

The answer, it turns out, is not very strong at all. In an editorial published in the May 15th issue of American Family Physician, I review the small number of studies that have evaluated the effect of electronic clinical decision support systems (CDSSs) on processes and outcomes of preventive care. Whether the goal was to improve immunization or behavioral counseling rates, electronic health records have had, at best, modest effects:

In summary, the evidence is far from conclusive that EHRs and CDSSs improve preventive care processes and outcomes in primary care settings. The small number of mostly nonrandomized studies makes it hard to determine whether changes in physicians' behaviors were the result of implementing CDSSs, or if other factors were responsible. Also, the most promising studies to date were performed in large practices of employed physicians, rather than in small physician-owned practices. Finally, all but a few studies measured only guideline adherence, rather than patient-oriented health outcomes. To be worth the investment, EHR-enabled CDSSs must ultimately be shown to not only improve processes of preventive care, but also reduce morbidity and mortality and improve quality of life.

Similarly, a study published in this month's issue of the Annals of Family Medicine found that in a group of 42 similar primary care practices in the Northeast, those using EHRs were less likely than those without EHRs to meet three diabetes care quality measures (hemoglobin A1c, LDL cholesterol, and blood pressure), and that the gap did not narrow after 3 years.

So what are the chances that our residency's substantial investment (and the U.S. government's billions of dollars of incentives for physicians and hospitals to install and demonstrate "meaningful use" of electronic health records) will ultimately pay off for patients? The key to success for integrated health systems such as Kaiser Permanente and the Mayo Clinic has been to use the data from EHRs to manage population health. Rather than the traditional model of treating diabetes one patient at a time, for example, "panel managers" (registered nurses or other non-physician health professionals) can reach out to patients outside of the office visit and make sure that they are receiving recommended care. Who will pay these managers outside of the Kaisers and Mayo Clinics remains a largely unanswered question. The bottom line, though, is that it's not enough to just collect electronic data. For EHRs to transform primary care, we need to be able to use the data in new and creative ways, improving the health of large groups of patients - and eventually, entire communities - at the same time.

Friday, May 11, 2012

Effective health care for children with autism spectrum disorders

A recent report from the Centers for Disease Control and Prevention found that the prevalence of autism spectrum disorders (ASDs), estimated at 1 in 110 children in a 2010 American Family Physician article, may now have risen as high as 1 in 88. Previous AFP Community Blog posts have discussed potential explanations for the continuing increase in autism diagnoses, from the phenomenon of "diagnosis shift" to increased screening for ASDs at well-child visits, a controversial practice.

Although the etiology of ASDs remains unknown, there is evidence to support some treatments for affected children. In the May 1st issue of AFP, Dr. Corey Fogleman launched our "Implementing AHRQ Effective Health Care Reviews" series by summarizing key points from an Agency for Healthcare Research and Quality-sponsored review of the effectiveness, benefits, and harms of therapies for core and associated symptoms of ASDs in children two to 12 years of age. The review found that the antipsychotic drugs risperidone and aripiprazole reduce challenging behaviors in children with ASDs, but are associated with significant adverse effects. Also, intensive one-on-one behavioral interventions appear to improve outcomes if begun before four years of age.

The AHRQ review's conclusion that there is insufficient evidence to assess the benefits and harms of other treatments for ASD-associated repetitive behaviors was supported by a recent study published in Pediatrics. Dr. Melisa Carrasco and colleagues analyzed published and unpublished data on selective serotonin receptor inhibitors (SSRIs) and initially found that SSRIs were modestly helpful in reducing repetitive behaviors in children with ASDs. However, after they adjusted for the effect of publication bias (i.e., the tendency for trials showing a benefit to be published while those showing no benefit are not), the improvement was no longer statistically significant. This study illustrated how difficult it is for even the highest-quality reviews to determine what constitutes effective health care for patients when important data are unavailable for review.


The above post was first published in the AFP Community Blog.

Friday, May 4, 2012

Job Posting: Primary Care Health Policy Fellow

The Department of Family Medicine at Georgetown University School of Medicine is currently seeking qualified applicants for its one-year fellowship in Primary Care Health Policy. This is a unique, full-time program that combines experiences in scholarly research, faculty development, and clinical practice. Fellows have the opportunity to interact with local and federal policymakers in Washington, D.C. and pursue original research projects with experienced mentors at the Robert Graham Center for Policy Studies in Family Medicine and Primary Care. They will join a dynamic group of faculty (including me) at one of the flagship departments for urban family medicine on the East Coast. Past Health Policy Fellows have gone on to hold leadership positions in federal health agencies, community health organizations, and academia. Applicants should be graduates of an accredited residency program in family medicine or expect to graduate in 2012. Please e-mail me at for additional information.