Original Article: Top 20 Research Studies of 2017 for Primary Care Physicians
Issue Date: May 1, 2018
See additional reader comments at:https://www.aafp.org/afp/2018/0501/p581.html
To the Editor: Drs. Ebell and Grad have provided an excellent summary of patient-oriented evidence in their article and should be commended for their efforts to concisely update busy family physicians. As physicians, we agree with many of the practice recommendations and encourage all physicians to read the article as a starting point for discussions and reflections on their own best practices.
A review of publications to inform patient care requires consideration of the quality and scope of the study and previous studies on the same topic. Overlooking problems with study quality can create overconfidence in study results. Overestimating the scope of the study may lead to incorrect generalizations beyond the study topic. Not placing the study in context of previous studies risks overweighing spurious results that may run counter to a larger body of equal-quality or superior-quality evidence. We are concerned that in the presentation of clinical questions and bottom-line answers for “Cardiovascular Disease and Hypertension,” some of these considerations are missing.
Study 1, an investigation on blood pressure measurements using a bare arm vs. a shirt sleeve, is presented with the clinical question, “What is the best way to measure blood pressure?”1 This question is beyond the scope of the actual article. A more accurate clinical question, such as “How accurate is measuring blood pressure with clothing compared to bare arm?” would help reduce the risk of incorrect generalization by readers.
Study 2 raises concerns of study quality and missing background information.2 This is a nonrandomized observational evaluation of automated 30-minute office blood pressure (OBP30) measurements. Although the results of the study are impressive, the study authors noted several limitations, including a lack of comparison arm, lack of randomization, and potential bias from regression to the mean. Stating the study results without providing context of the study quality could make readers overconfident about the effectiveness of the OBP30 compared with the superior-quality evidence of other ambulatory or automated methods of office blood pressure measurement. A bottom-line answer reflective of the study quality and the background of previous studies might include, “This study was limited by lack of randomization or control, and further study would be of benefit before comparing OBP30 with other validated methods such as ambulatory blood pressure monitoring.”
We hope raising these concerns will better inform readers and highlight some of the challenges in critical appraisal.
In Reply: Thank you for your comments. We fully agree that interpretation of individual studies requires a careful evaluation of study quality and a sense of where it fits in the large body of research. As a group, we have written nearly 5,000 POEMs (patient-oriented evidence that matters) over the past 20 years and recognize that any critical appraisal must balance completeness and context with clarity and conciseness. Each published POEM has separate sections that describe the “Study design,” “Population studied,” “Funding source,” “Allocation concealment” (for randomized controlled trials), and a 200- to 300-word “Synopsis” that provides an objective assessment of the context, results, and any biases. Readers can read the full POEMs summaries of the top 20 research articles here: https://www.aafp.org/afp/poems2017.
However, the article summarizing the top 20 research articles of 2017 as determined by Canadian physicians included only an edited “Clinical question” and “Bottom line.” The other sections were necessarily omitted given space constraints. For the POEM about study 1, the original POEM title was “Bare Arm Best for BP Measurement” and would have been read before the clinical question.
Regarding the second study, the full POEM clearly highlighted potential limitations, including the statement that “It remains to be seen whether this result occurs in other settings or whether patient outcomes are improved,” implying the need for a randomized trial. We suspect that this study was selected by Canadian Medical Association members as one of the most relevant studies of the year in light of the SPRINT study that used a similarly time-consuming approach to measuring blood pressure1 and found that the mean of those six measurements was 23/12 mm Hg lower than the initial office blood pressure. The larger message that blood pressures measured after a period of rest in the office are lower than those taken immediately is an important one, and clinicians who choose to apply the SPRINT trial results in their practices should make sure to measure blood pressure the same way.