- Email: [email protected]
Association of Alcohol With Fatal Snowmobile Accidents
Correspondence
Reply
Association of Alcohol With Fatal Snowmobile Accidents
Brian H Rowe, MD, MSc
To the Editor.
Myocardial Infarction Induced by
valuable contributions to injury epidemielegy. 1,2 In such a study, con-
Association of Alcohol With Fatal Snowmobile Accidents Peter Cummings, MD, MPH
Case-control studies have yielded
Migraine Therapy Sushil George, MD
trols should be drawn from the same
Robert Vannozi, MD, FACC
underlying population at risk for the outcome as the cases. 3 A control
Submental Injection of Naloxone AngeIo A Salvucci, Jr, MD, FACEP Marc Eckstein, MD Angel L Iscovich, MD, FACEP
should represent those who might have become a case in the study had the outcome of interest developed in them. 4 Thus, if we wish to know whether alcohol is associated with the risk of death while driving a
Gila Monster Envenomation Mark F Miller
snowmobile in Ontario, we would like the cases to be those who died while driving a snowmobile in the
Vision Loss in Polyarterids Nodosa
study area. The controls would be
William F Deegan III, MD
drawn from the pool of all snowmo-
Copyright © by the American College
bile drivers. Rowe and colleagues5 attempted
of Emergency Physicians.
to estimate the relative risk of death from a snowmobile crash associated with alcohol use. Their controls were
not selected to sample the experience of all snowmobile drivers. Instead, they measured the prevalence of alcohol use among fatally injured car and motorcycle drivers. This control group is inappropriate because it tells us nothing about the use of alcohol among the population of interest--living snowmobile drivers. If dead snowmobile drivers can be compared with dead car drivers in a study of the relative risk for death in an alcohol-related snowmobile accident, as was done by Rowe, then it should be reasonable to compare dead car drivers with dead snowmobile drivers to measure the association between alcohol use and fatal car crashes. Using the data of Rowe et al would lead to the conclusion that alcohol use protected car drivers from crashes; the OR would be the inverse of the estimate given by Rowe et al, 1/4.3=.23. Emergency physicians will recognize that this result is not correct, and this insight may help them appreciate that the control group selected by Rowe and colleagues is not a useful one.
Guidelines for Letters Annals welcomes correspondence, including observations, opinions, corrections, very brief reports, and comments on published articles. Letters to the editor will not be accepted if they exceed three double-spaced pages, with a maximum of 10 references. Two double-spaced copies must be submitted. They should not contain abbreviations. Letters must be signed and include a postscript granting permission to publish. Financial associations or other possible conflicts of interest should always be disclosed. Letters discussing an Annals article should be received within 6 weeks of the article's publication. Annals acknowledges receipt of letters with a postcard, and correspondents are notified by postcard when a decision is made. Published letters will be edited and may be shortened. Unpublished letters wili not be returned.
MAY 1995
25:5
ANNALS OF EMERGENCY MEDICINE
An example of how Rowe and colleagues might have sampled the exposure experience of snowmobile users is illustrated by a recent study of alcohol use and drowning. 6 The study team actually visited the location where each patient drowned and administered alcohol breath tests to control subjects.
Peter Cummings, MD, MPH Department of Epidemiology University of Washington Seattle, Washington i. Thompson RS, Rivara FP, Thompson DC: A case-control study of the efjectiveness of bicycle safety helmets. N Engl J Med 1989;320:1361-1367. 2. Kellerman AL, Rivara FP, Somes G, et al: Suicide in the home in relation to gun ownership. N EEglJ Med 1992;327:467472. 3. Wacholder S, McLaughlin JK, Silvennan DT, et al: Selection of controls in case-control studies. I. Principles. Am J Epidemiol 1992;135:1019-1028. 4. Rothman KJ: Modern Epidemiology. Boston: Little, Brown, 1986, pp 64-68. .5. Rowe B, Milner R, Johnson C, et al: The association of alcohol and night driving with fatal snowmobile trauma: A case-control study. Ann EmergMeal 1994;24:842-848. 6. Smith GS, Houser-Hadley J: Risk factors for drowning: A case-control study. Abstracts of the 122nd Annual Meeting of the American Public Health Association, 1994, p 323.
In reply. Dr Cummings argues that we should have performed a different study, such as attending the scene of the fatal snowmobile crash sites across Ontario and testing other snowmobilers for alcohol intake. The study he suggests would involve standing
7 17
CORRESPONDENCE
along remote snowmobile routes (in the late evening and early morning to match the time of injury) in northern Ontario (approximateaverage winter temperature -10°C), hoping that the occasionalsnowmobiler who happenedby would participate in the study and not simply speed away from the oddball with the clipboard waving him down. It is possible that only the alcoholintoxicated group would be so lacking in judgment as to stop, thereby resulting in a biasedcontrol group. More seriously,although this method of studymay be both logisticallydifficult and ethicallychallenging,we have completed some snowmobile field research,and these biasesare real. In comparative studies, the control group often representsa compromise between what is possible and what is ideal. Moreover, there is rarely one correct control group choice. In this study, consideration was given to the alternatives Cummings outlined, and these alternatives were rejected for a variety of reasons. The cases in this study were males who were fatally injured while driving snowmobiles. The controls were in the category of "drivers," who were eligible for snowmobile operation becausethey had driver's licenses,who were fatally injured while driving other vehicles. We do not believe the comparisons between the groups are illogical, as Cummingsargues. Our study illustrates that excessivealcohol intake is less commonlyassociated with fatal car crashes than with fatal snowmobile crashes. This information tells us that it will be important to target alcohol intake to prevent snowmobiling deaths. It also suggests that the same message about alcohol and road driving, even if maximally effective, is unlikely to have as big an impact in preventing fatal car crashes. Our results support the suggestionthat other factors (eg, driving barriers, conditions on the road, falling asleep at the wheel) may play an important role in the
718
deaths of drivers killed while in cars and on motorcycles. Dr Cummingstrivializes the importance of these findings with the suggestionthat the data demonstrate a protective benefit from alcohol in fatal car crashes. He bases this on discussion of the concept of relative risk. However, the relative odds, a number derived from retrospective comparison of groups of drivers, cannot be taken as a relative risk estimate. Relative risk estimates are appropriate for prospectivecohort studies but are incorrect for retrospective studies such as ours. When drawing comparisons,researchers must decide which comparisons make historic, biologic, and physiologic sense. Clearly,the Cummings argument does not make sense in any of these areas and is a misrepresentation of the data. The power of injury research lies in its ability to identify factors that can be modified to reduce the risk of trauma. In comparingdifferent groups of subjects who were fatally injured white driving vehicles, our work helps identify factors that can be altered to reduce snowmobile mortality. We believe such work is valid and highly worthwhile.
Brian 14Rowe, MO, MSc University of Ottawa Sudbury, Ontario Canada
Myocardial Infarction Induced by Migraine Therapy To the Editor. Coronaryartery spasm is a rare cause of myocardialinfarction in the absence of underlyingatherosclerotic coronary vascular disease, but occasionally, therapeutic drug use may precipitate coronary artery spasm, myocardialischemia,and infarction#,2 We report a patient in whom an acute myocardial infarction developed
within hours of the start of therapy with an isometheptenecombination. A 36-year-old,previouslyhealthy woman presentedto the emergency departmentwith severe retrosternal chest pain. Her sole risk factor for coronary artery disease was a 10pack-year history of cigarette smoking. She led an active lifestyle and denied using oral contraceptivesor recreational drugs. The patient had seen her doctor the preceding day for intermittent headache of 3 weeks' duration; migraine was diagnosed.She was prescribeda combinationof ismetheptene mucate 65 mg, acetaminophen 325 mg, and dichloralphenazone 100 mg (Midrin; Carnrick).The patient took two capsules initially, followed by one capsule every 4 hours, for a total of four capsulesin the 12 hours before admission. Her headache resolved, but she then experienced severe chest pain. Vital signs and clinical examination were normal. ECGon presentation with chest pain showed .5- to 1-mm ST-segmentdepressionin leads V2 to V5 and T-wave inversion in leads V1 and V2. The chest pain was relieved with two sublingual nitroglycerine tablets, and repeat ECG showed normalizationof the STsegment and T-wave changes.Three hours later the pain recurred, but it was less severe. ECG did not show any new changes.The patient was given an IV nitroglycerine infusion and diltiazem 30 mg three times a day, which relieved her pain. Twelve hours after admissionthe patient again had chest pain, which resolved with an increasein the nitroglycerine infusion rate. Intravenousheparin was given and continued for 48 hours. The patient's subsequent hospital stay was uneventful. Routine laboratory investigations were normal, with a serum total cholesterol of 145 mg/dL Urine screening tests were negative for cocaine and amphetamines. The creatine phosphekinaselevel was 179 IU/L (normal, 30 to 70 IU/L) on
admission and rose to 562 IU/L 6 hours later. The creatine phosphokinase level peakedat 677 IU/L at 12 hours before returning to normal at 30 hours. Flectrophoresisconfirmed elevationof the MB fraction to 10.3%. The lactate dehydrogenase level was 327 IU/L (normal,313 to 618 IU/L) on admissionand remained in the normal rangeduringthe first 24 hoursbut roseto 707 IU/Lat 30 hours. Assay of lactate dehydrogenase isoenzymesshowed elevation of LDH1, and the LDH1/LDH2ratio was 1:17. These results confirmed an acute myocardial infarction. An echocardiogramon day 3 showed no valvular lesionsor wall motion abnormalities. The ejection fraction was normal. Coronaryangiographyon day 8 showed no narrowingto suggest coronaryatherosclerosis.The patient has remainedfree of pain since then. This patient presentedwith chest pain highly suggestiveof cardiac pain within 12 hours of taking an isometheptenecombination. She had mild ischemicchanges on her initial ECGthat reversedwith nitroglycerine. However, serial cardiac enzymestudies confirmed an acute non-Q-wave myocardialinfarction. The time sequenceand the normal coronary angiogram lead us to believe that the patient had coronary artery spasm induced by isometheptone severe enough to cause a non-Q-wave myocardialinfarction. Midrin is a combination of isometheptene, dichloralphenazone, and acetaminophen.Neither dichloralphenazone nor acetaminophen is known to have cardiovascular effects at therapeutic doses. Isometheptenemucate is an unsaturated aliphatic amine that acts by constricting cranial and cerebral arteries.3 It is consideredpossibly effective in the treatment of migraine bythe FDA and is recommendedby the manufacturerfor relief of tension and vascular headaches. Caution is advised when patients have hypertensionor peripheral or cardiovasculardisease.3
ANNALS OF EMERGENCY MEDICINE
25:5
MAY 1995
Reply
Association of Alcohol With Fatal Snowmobile Accidents
Brian H Rowe, MD, MSc
To the Editor.
Myocardial Infarction Induced by
valuable contributions to injury epidemielegy. 1,2 In such a study, con-
Association of Alcohol With Fatal Snowmobile Accidents Peter Cummings, MD, MPH
Case-control studies have yielded
Migraine Therapy Sushil George, MD
trols should be drawn from the same
Robert Vannozi, MD, FACC
underlying population at risk for the outcome as the cases. 3 A control
Submental Injection of Naloxone AngeIo A Salvucci, Jr, MD, FACEP Marc Eckstein, MD Angel L Iscovich, MD, FACEP
should represent those who might have become a case in the study had the outcome of interest developed in them. 4 Thus, if we wish to know whether alcohol is associated with the risk of death while driving a
Gila Monster Envenomation Mark F Miller
snowmobile in Ontario, we would like the cases to be those who died while driving a snowmobile in the
Vision Loss in Polyarterids Nodosa
study area. The controls would be
William F Deegan III, MD
drawn from the pool of all snowmo-
Copyright © by the American College
bile drivers. Rowe and colleagues5 attempted
of Emergency Physicians.
to estimate the relative risk of death from a snowmobile crash associated with alcohol use. Their controls were
not selected to sample the experience of all snowmobile drivers. Instead, they measured the prevalence of alcohol use among fatally injured car and motorcycle drivers. This control group is inappropriate because it tells us nothing about the use of alcohol among the population of interest--living snowmobile drivers. If dead snowmobile drivers can be compared with dead car drivers in a study of the relative risk for death in an alcohol-related snowmobile accident, as was done by Rowe, then it should be reasonable to compare dead car drivers with dead snowmobile drivers to measure the association between alcohol use and fatal car crashes. Using the data of Rowe et al would lead to the conclusion that alcohol use protected car drivers from crashes; the OR would be the inverse of the estimate given by Rowe et al, 1/4.3=.23. Emergency physicians will recognize that this result is not correct, and this insight may help them appreciate that the control group selected by Rowe and colleagues is not a useful one.
Guidelines for Letters Annals welcomes correspondence, including observations, opinions, corrections, very brief reports, and comments on published articles. Letters to the editor will not be accepted if they exceed three double-spaced pages, with a maximum of 10 references. Two double-spaced copies must be submitted. They should not contain abbreviations. Letters must be signed and include a postscript granting permission to publish. Financial associations or other possible conflicts of interest should always be disclosed. Letters discussing an Annals article should be received within 6 weeks of the article's publication. Annals acknowledges receipt of letters with a postcard, and correspondents are notified by postcard when a decision is made. Published letters will be edited and may be shortened. Unpublished letters wili not be returned.
MAY 1995
25:5
ANNALS OF EMERGENCY MEDICINE
An example of how Rowe and colleagues might have sampled the exposure experience of snowmobile users is illustrated by a recent study of alcohol use and drowning. 6 The study team actually visited the location where each patient drowned and administered alcohol breath tests to control subjects.
Peter Cummings, MD, MPH Department of Epidemiology University of Washington Seattle, Washington i. Thompson RS, Rivara FP, Thompson DC: A case-control study of the efjectiveness of bicycle safety helmets. N Engl J Med 1989;320:1361-1367. 2. Kellerman AL, Rivara FP, Somes G, et al: Suicide in the home in relation to gun ownership. N EEglJ Med 1992;327:467472. 3. Wacholder S, McLaughlin JK, Silvennan DT, et al: Selection of controls in case-control studies. I. Principles. Am J Epidemiol 1992;135:1019-1028. 4. Rothman KJ: Modern Epidemiology. Boston: Little, Brown, 1986, pp 64-68. .5. Rowe B, Milner R, Johnson C, et al: The association of alcohol and night driving with fatal snowmobile trauma: A case-control study. Ann EmergMeal 1994;24:842-848. 6. Smith GS, Houser-Hadley J: Risk factors for drowning: A case-control study. Abstracts of the 122nd Annual Meeting of the American Public Health Association, 1994, p 323.
In reply. Dr Cummings argues that we should have performed a different study, such as attending the scene of the fatal snowmobile crash sites across Ontario and testing other snowmobilers for alcohol intake. The study he suggests would involve standing
7 17
CORRESPONDENCE
along remote snowmobile routes (in the late evening and early morning to match the time of injury) in northern Ontario (approximateaverage winter temperature -10°C), hoping that the occasionalsnowmobiler who happenedby would participate in the study and not simply speed away from the oddball with the clipboard waving him down. It is possible that only the alcoholintoxicated group would be so lacking in judgment as to stop, thereby resulting in a biasedcontrol group. More seriously,although this method of studymay be both logisticallydifficult and ethicallychallenging,we have completed some snowmobile field research,and these biasesare real. In comparative studies, the control group often representsa compromise between what is possible and what is ideal. Moreover, there is rarely one correct control group choice. In this study, consideration was given to the alternatives Cummings outlined, and these alternatives were rejected for a variety of reasons. The cases in this study were males who were fatally injured while driving snowmobiles. The controls were in the category of "drivers," who were eligible for snowmobile operation becausethey had driver's licenses,who were fatally injured while driving other vehicles. We do not believe the comparisons between the groups are illogical, as Cummingsargues. Our study illustrates that excessivealcohol intake is less commonlyassociated with fatal car crashes than with fatal snowmobile crashes. This information tells us that it will be important to target alcohol intake to prevent snowmobiling deaths. It also suggests that the same message about alcohol and road driving, even if maximally effective, is unlikely to have as big an impact in preventing fatal car crashes. Our results support the suggestionthat other factors (eg, driving barriers, conditions on the road, falling asleep at the wheel) may play an important role in the
718
deaths of drivers killed while in cars and on motorcycles. Dr Cummingstrivializes the importance of these findings with the suggestionthat the data demonstrate a protective benefit from alcohol in fatal car crashes. He bases this on discussion of the concept of relative risk. However, the relative odds, a number derived from retrospective comparison of groups of drivers, cannot be taken as a relative risk estimate. Relative risk estimates are appropriate for prospectivecohort studies but are incorrect for retrospective studies such as ours. When drawing comparisons,researchers must decide which comparisons make historic, biologic, and physiologic sense. Clearly,the Cummings argument does not make sense in any of these areas and is a misrepresentation of the data. The power of injury research lies in its ability to identify factors that can be modified to reduce the risk of trauma. In comparingdifferent groups of subjects who were fatally injured white driving vehicles, our work helps identify factors that can be altered to reduce snowmobile mortality. We believe such work is valid and highly worthwhile.
Brian 14Rowe, MO, MSc University of Ottawa Sudbury, Ontario Canada
Myocardial Infarction Induced by Migraine Therapy To the Editor. Coronaryartery spasm is a rare cause of myocardialinfarction in the absence of underlyingatherosclerotic coronary vascular disease, but occasionally, therapeutic drug use may precipitate coronary artery spasm, myocardialischemia,and infarction#,2 We report a patient in whom an acute myocardial infarction developed
within hours of the start of therapy with an isometheptenecombination. A 36-year-old,previouslyhealthy woman presentedto the emergency departmentwith severe retrosternal chest pain. Her sole risk factor for coronary artery disease was a 10pack-year history of cigarette smoking. She led an active lifestyle and denied using oral contraceptivesor recreational drugs. The patient had seen her doctor the preceding day for intermittent headache of 3 weeks' duration; migraine was diagnosed.She was prescribeda combinationof ismetheptene mucate 65 mg, acetaminophen 325 mg, and dichloralphenazone 100 mg (Midrin; Carnrick).The patient took two capsules initially, followed by one capsule every 4 hours, for a total of four capsulesin the 12 hours before admission. Her headache resolved, but she then experienced severe chest pain. Vital signs and clinical examination were normal. ECGon presentation with chest pain showed .5- to 1-mm ST-segmentdepressionin leads V2 to V5 and T-wave inversion in leads V1 and V2. The chest pain was relieved with two sublingual nitroglycerine tablets, and repeat ECG showed normalizationof the STsegment and T-wave changes.Three hours later the pain recurred, but it was less severe. ECG did not show any new changes.The patient was given an IV nitroglycerine infusion and diltiazem 30 mg three times a day, which relieved her pain. Twelve hours after admissionthe patient again had chest pain, which resolved with an increasein the nitroglycerine infusion rate. Intravenousheparin was given and continued for 48 hours. The patient's subsequent hospital stay was uneventful. Routine laboratory investigations were normal, with a serum total cholesterol of 145 mg/dL Urine screening tests were negative for cocaine and amphetamines. The creatine phosphekinaselevel was 179 IU/L (normal, 30 to 70 IU/L) on
admission and rose to 562 IU/L 6 hours later. The creatine phosphokinase level peakedat 677 IU/L at 12 hours before returning to normal at 30 hours. Flectrophoresisconfirmed elevationof the MB fraction to 10.3%. The lactate dehydrogenase level was 327 IU/L (normal,313 to 618 IU/L) on admissionand remained in the normal rangeduringthe first 24 hoursbut roseto 707 IU/Lat 30 hours. Assay of lactate dehydrogenase isoenzymesshowed elevation of LDH1, and the LDH1/LDH2ratio was 1:17. These results confirmed an acute myocardial infarction. An echocardiogramon day 3 showed no valvular lesionsor wall motion abnormalities. The ejection fraction was normal. Coronaryangiographyon day 8 showed no narrowingto suggest coronaryatherosclerosis.The patient has remainedfree of pain since then. This patient presentedwith chest pain highly suggestiveof cardiac pain within 12 hours of taking an isometheptenecombination. She had mild ischemicchanges on her initial ECGthat reversedwith nitroglycerine. However, serial cardiac enzymestudies confirmed an acute non-Q-wave myocardialinfarction. The time sequenceand the normal coronary angiogram lead us to believe that the patient had coronary artery spasm induced by isometheptone severe enough to cause a non-Q-wave myocardialinfarction. Midrin is a combination of isometheptene, dichloralphenazone, and acetaminophen.Neither dichloralphenazone nor acetaminophen is known to have cardiovascular effects at therapeutic doses. Isometheptenemucate is an unsaturated aliphatic amine that acts by constricting cranial and cerebral arteries.3 It is consideredpossibly effective in the treatment of migraine bythe FDA and is recommendedby the manufacturerfor relief of tension and vascular headaches. Caution is advised when patients have hypertensionor peripheral or cardiovasculardisease.3
ANNALS OF EMERGENCY MEDICINE
25:5
MAY 1995