Phenylpropanolamine contained in cold remedies and risk of hemorrhagic stroke B.W. Yoon, MD, PhD; H.J. Bae, MD, PhD; K.S. Hong, MD, PhD; S.M. Lee, PhD; B.J. Park, MD, PhD; K.H. Yu, MD, PhD; M.K. Han, MD, PhD; Y.S. Lee, MD, PhD; D.K. Chung, MD, PhD; J.M. Park, MD, PhD; S.W. Jeong, MD, PhD; B.C. Lee, MD, PhD; K.H. Cho, MD, PhD; J.S. Kim, MD, PhD; S.H. Lee, MD, PhD; and K.M. Yoo, MD, PhD; for the Acute Brain Bleeding Analysis (ABBA) Study Investigators* Abstract—In this study, we sought to elucidate whether phenylpropanolamine (PPA) in cold remedies (small and divided doses) increases the risk of hemorrhagic stroke (HS). PPA exposure significantly increased the risk, and the risk was much higher in women. In women, linear trends were also found in recency, duration, and dosage of PPA exposure. PPA contained in cold remedies increases the risk of HS, particularly in women. NEUROLOGY 2007;68:146–149 An association between phenylpropanolamine (PPA) and hemorrhagic stroke (HS) has been suggested, 1-4 and a recent case– control study (Hemorrhagic Stroke Project [HSP]) demonstrated that PPA con- tained in appetite suppressants was an independent risk factor for HS in young and middle-aged women. 5 However, this study was confined to individuals aged 18 to 49 years, and an effect of PPA of a small and divided from (usually used as cold remedies) was not investigated. In 2001, the use of PPA as an ingredient of cough and cold drugs was permitted in many countries, including Korea. 6 In March 2002, under the decision of the Central Pharmaceutical Affairs Council of Ko- rea, the Acute Brain Bleeding Analysis (ABBA) study group was organized to investigate the risk of PPA-containing cold remedies for HS. Methods. The ABBA study is a multicenter case– control study in which the cases were recruited from 33 hospitals in Korea between October 2002 and March 2004. HS was defined as sub- arachnoid hemorrhage and intracerebral hemorrhage diagnosed based on clinical symptoms and the corresponding diagnostic tests including brain images (CT/MRI). The eligibility criteria included age of 30 to 84 years, absence of brain lesions that would increase the risk of hemorrhage, absence of stroke history or trauma, and an ability to complete an interview within 30 days after the stroke onset. We screened all of the patients with HS admitted to the participating hospitals. Together, we recruited two age- and sex- matched controls (hospital and community controls). Hospital con- trols were selected among patients who visited the same institution, and community controls were recruited from siblings, friends, or neighbors of the cases. We used a highly structured questionnaire, and the interview- ers and subjects were kept blind to the main hypothesis. The index date was defined as the calendar day when symptoms of stroke occurred for the first time. The subjects were asked to recall all kinds of medications taken during the 2 weeks before the index date. To verify the medication information, we tried to ob- tain the prescriptions. When prescriptions were unavailable, the subjects were asked to bring medication packages. If these pack- ages were unavailable, we asked the subjects to provide the exact names of drugs and select their medications from a drug-photo book. Only patients who had verified exposures to PPA-containing drugs were counted as positive. The sample size was calculated based on the following assump- tions: one-tailed significance, 0.05; power, 0.8; the smallest odds ratio (OR) for detection, 2.0; PPA-exposure rate in control, 2.0% (derived from the exposure rate in the HSP 5 ). Under a 1:2 match- ing strategy, 875 cases and 1,750 controls were required. Exposure to PPA was defined as administration of PPA within 14 days before the index date. Our primary question was whether the exposure to PPA was associated with the increased risk for HS. For secondary analyses, the window and duration of exposure were defined as intervals from the last exposure to the index date and as days of taking PPA-containing drugs. Missing variables other than PPA exposure were imputed to their mean or median values in subgroups assorted by age and sex with preference of values to support the null hypothesis. We estimated ORs and 95% CIs using conditional logistic models for matched sets. For adjust- ment, variables with p  0.1 on the comparison between cases and *See appendix for a complete list of ABBA Study Investigators. From the Department of Neurology, Seoul National University Hospital, Seoul National University College of Medicine (B.W.Y., S.H.L.); Department of Neurology, Seoul National University Bundang Hospital, Seoul National University College of Medicine (H.J.B., M.K.H., J.S.K.); Department of Neurology, Ilsan Paik University Hospital, Inje University (K.S.H.); Department of Preventive Medicine, Seoul National University College of Medicine (S.M.L., B.J.P.); Department of Neurology, Hallym University College of Medicine (K.H.Y., B.C.L.); Department of Neurology, Seoul National University Boramae Hospital, Seoul National University College of Medicine (Y.S.L.); Department of Neurology, Kyungpook National University Hospital (D.K.C.); Department of Neurology, Eulji General Hospital, Eulji University School of Medicine (J.M.P.); Department of Neurology, Dongguk University International Hospital, Dongguk University College of Medicine (S.W.J.); Department of Neurology, Chonnam National University Hospital, Chonnam National University College of Medicine (K.H.C.); and Department of Neurology, Kosin Medical College Hospital (K.M.Y.), Republic of Korea. Supported by the Korean Food and Drug Association. Disclosure: The authors report no conflicts of interest. Received March 20, 2006. Accepted in final form September 27, 2006. Address correspondence and reprint requests to Dr. Byung-Woo Yoon, Department of Neurology, Seoul National University Hospital, 28 Yongon-dong, Jongno-gu, Seoul, 110-744, Republic of Korea; e-mail: bwyoon@snu.ac.kr 146 Copyright © 2007 by AAN Enterprises, Inc.