Johansen ME, Gold KJ, Sen A, Arato N, Green LA. A National survey of the treatment of hyperlipidemia in primary prevention. JAMA Intern Med. Published online March 11, 2013. doi:10.1001/jamainternmed.2013.2797
eTable. Demographics of Participating Physicians
eFigure 1. Vignette 3: 40-Year-Old Male With Hypertension and an LDL of 180
eFigure 2. Vignette 4: 50-Year-Old Female With an LDL of 180
Johansen ME, Gold KJ, Sen A, Arato N, Green LA. A National Survey of the Treatment of Hyperlipidemia in Primary Prevention. JAMA Intern Med. 2013;173(7):586–588. doi:10.1001/jamainternmed.2013.2797
Author Affiliations: Departments of Family Medicine (Drs Johansen, Gold, Sen, Arato, and Green), Obstetrics & Gynecology (Dr Gold), and Biostatistics, School of Public Health (Dr Sen), University of Michigan, Ann Arbor; and Department of Family Medicine, University of Alberta, Edmonton, Alberta, Canada (Dr Green).
The majority of statin use in the United States is for primary prevention: that is, in patients without established coronary heart disease (CHD). The evidence of mortality benefit in this population is inconclusive.1,2 The patient's baseline risk is critical in determining the risk-benefit ratio of statins.3
Little is known about physician decision making regarding the use of statins in primary prevention. Previous studies investigating treatment of hyperlipidemia focused on adherence to various guidelines.4,5 We investigated physicians' prescribing strategies in relationship to baseline risk of CHD.
We sent an anonymous and voluntary written survey to 750 physicians selected randomly from a nationally representative sample of US physicians from the American Medical Association Physician Masterfile. The sample consisted of an equal number of family medicine, cardiology, and general internal medicine physicians. Inclusion criteria were physicians (doctor of medicine or doctor of osteopathic medicine degree) who had seen adult patients with hyperlipidemia in an outpatient clinic within the last 12 months. Three waves of letters were sent with an initial $2 cash incentive.
The survey contained 6 vignette-style questions involving patients without CHD and different baseline risks, for whom a physician might consider treatment of hyperlipidemia (Table). All risk factors were stated, and for all patients, the high-density lipoprotein cholesterol level was 50 mg/dL (to convert cholesterol to millimoles per liter, multiply by 0.0259) and triglyceride level was 150 mg/dL (to convert to millimoles per liter, multiply by 0.0113). Vignettes 3 to 6 describe patients who had attempted lifestyle modifications prior to consideration of medications. Vignettes 3 and 4 queried how many similar patients would need to be treated with a statin to prevent a death at 5 and 20 years and given choices of 1 to 10, 11 to 20, 21 to 50, 51 to 100, 101 to 500, and 501 or more.
All analyses were conducted on fully deidentified data using IBM SPSS Statistics 19 software (SPSS Inc). A logistic regression was run for prescribing vs not prescribing with specialty and sex as independent variables. Inclusion of other physician demographic factors did not substantially affect the model. A linear regression model was fitted for number needed to treat, with specialty and sex as independent variables. We compared prescribing for vignette 1 vs 2, 3 vs 4, and 5 vs 6 using the McNemar test. The study was approved by the University of Michigan institutional review board.
Of 750 surveys, 289 were returned, with 202 usable and meeting inclusion criteria. There were 90 family medicine (44.6%), 59 cardiology (29.2%), and 53 (26.2%) internal medicine responses. Eighty-seven surveys were unusable, of which 22 were returned to sender unopened, 7 did not meet eligibility criteria, and 58 were returned incomplete. The usable return rate was 30.5% (202 of 663). For respondent demographics please see the eTable.
In vignettes 1 and 2, a diabetic woman with different low-density lipoprotein cholesterol (LDL-C) levels (120 vs 88 mg/dL) was recommended a statin therapy more often in the patient with the higher LDL-C level (94.0% vs 40.2%) (P < .001).
In vignettes 3 and 4 involving low-risk patients with an LDL-C level of 180 mg/dL, significantly more health care providers treated the 40-year-old man with well-controlled hypertension (88.9%) compared with the 50-year-old women (73.5%) (P < .001). The number needed to treat (NNT) was identified as being lower in the man compared with the woman at both 5 and 20 years (P < .001). Health care providers reported a significantly lower perceived NNT for outcomes at 20 years compared with 5 years in both the male and female patient (P < .001). Distributions of NNT responses are shown in eFigure 1 and eFigure 2.
In vignettes 5 and 6, a 75-year-old man (LDL-C level of 140 mg/dL) was compared with a 50-year-old woman (LDL-C level of 145 mg/dL). Both patients had hypertension and used tobacco. Respondents recommended similar treatment rates (86.6% vs 88.9%).
We compared responses by clinical specialty, but there were no consistent differences.
Our study investigated physician prescribing intentions and outcome improvement beliefs regarding the use of statins in the treatment of hyperlipidemia in primary prevention of CHD. This study was unique because we did not compare management with current guidelines but instead asked physicians to make choices regarding treatment and their perception of effect on patient outcomes for clinical scenarios. We found that physicians consider medication for patients (vignettes 3, 4, and 6) with low Framingham risk scores (≤5%), for whom available evidence does not support outcome benefit.1,6 Seventy-three percent of respondents even recommended treating a 50-year-old woman with an LDL-C level of 180 mg/dL, though one-fourth of those practitioners identified an NNT for mortality of more than 500 at 5 years (eFigure 2). Furthermore, though there is clear evidence that most diabetic patients benefit from statins even if their LDL-C level is below goal,7,8 most respondents tended not to treat a diabetic patient if their LDL-C level was below the threshold set by guidelines. Overall, our study suggests that physicians may not adequately consider a patient's cardiovascular risk when prescribing statins in primary prevention.
Correspondence: Dr Johansen, Department of Family Medicine, University of Michigan, 1018 Fuller St, Ann Arbor, MI 48104-1213 (firstname.lastname@example.org).
Published Online: March 11, 2013. doi:10.1001/jamainternmed.2013.2797
Author Contributions:Study concept and design: Johansen, Gold, and Green. Acquisition of data: Johansen and Arato. Analysis and interpretation of data: Johansen, Gold, Sen, and Green. Drafting of the manuscript: Johansen, Sen, Arato, and Green. Critical revision of the manuscript for important intellectual content: Johansen, Gold, and Sen. Statistical analysis: Johansen, Sen, and Green. Obtained funding: Johansen and Green. Administrative, technical, and material support: Johansen and Arato. Study supervision: Johansen, Gold, Arato, and Green.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by grant G1001RR from the American Academy of Family Physicians.
Previous Presentation: These data were presented as a poster at the North American Primary Research Group meeting; November 21, 2011; Banff, Alberta, Canada.