- Email: [email protected]
No Gag Rule for Intubation
CORRESPONDENCE
ture an estimate of 100 lightningrelated deaths and 300 injuries annually in the United States. He then cites a 1976 National Oceanic and Atmospheric Administration (NOAA) publication in suggesting, "Most authorities agree that the numbers are much larger. "1 The difficulties in tabulating lightningrelated deaths and injuries are quite different and deserve further comment. Several unrelated data sources (meteorologic, vital and health statistics, and fire) provide information on death and injury resulting from lightning in the United States. The only source that specifically tracks lightning-caused death and injury is the monthly publication Storm Data from the NOAA National Climatic Data Center, which relies heavily on newspapers fer information on casualties due to lightning. The annual averages of lightning deaths and injuries reported to NOAA from 1959 through 1992 were 92 and 264, respectively.2 Lopez et al3 found that NOAA captured only 72% of the lightning-related deaths recorded on Colorado death certificates and only 58% of the patients with lightning-related injuries discharged from Colorado hospitals from 1988 to 1991. The shortfall in NOAA reports of patients with lightning injuries treated and discharged from outpatient settings could not ever be calculated in the absence of statewide outpatient data. The most comprehensive counts of lightning deaths are provided by national vital-statistics reports. These are prepared annually by the National Center br Health Statistics from data supplied by state health departments, which are drawn from death certificates submitted by medical examiners. Between 1980 and 1991, deaths att!ibuted to lightning by NCHA averaged 87 per year, ranging from a h!gh of 100 in 1982 to a low of 75 in 19;89and 19914 The continuing long-term decline in such deaths is reflected in the average of 92 for the first sik years of this period and the aver&ge of 83 for the second six yearsi Baker et al5 noted that from 1930 t 9' 1980, the death rate from lightning showed one of the largest decreases for any injury (from .29 per 1001000 in 1930 to .04
OCTOBER 1994
26:4
per 100,000 in 1980). The decrease mostly reflects improvements in electrical systems and a decrease in the population of farmers, although a change in vital-statistics coding methods in the early 1950s is partly responsible and improved treatment may have reduced fatalities. Because vital-statistics reports quantify injury-related deaths with respect to a single underlying cause, such statistics do not include all lightning-related deaths. A more complete profile can be obtained from medical-examiner reports, which list all contributing causes. When Duclos and Sanderson 6 compared NCHS data for t972-1984 with medical-examiner reports from North Carolina, they found that 8 of 64 North Carolina medical-examiner reports including a lightning-related diagnosis (an additional 14%) did not appear in the NCHS database. Duclos and Sanderson6 did not address possible systematic reasons for the omission of these eight cases. Some such omissions may reflect deaths resulting from structure fires started by lightning. Weigel 1 had hypothesized in 1976 that because 2% of all fires were started by lightning, 2% of all fire deaths might be similarly ascribed to lightning. If these deaths were all coded as fire deaths rather than lightning deaths by medical examiners, this would have increased total lightning deaths by 70%, from the 124 reported by NCHS in 1973 to 211. In fact, occupants of a residence ignited by lightning who are not struck themselves are rarely trapped by the fire. The National Fire Protective Association (NFPA), however, has estimated only 14 deaths in fires started by lightning and 90 lightning-related fire injuries per year to civilians (nonfirefighters) for the period 1989-19937 and an average of four firefighter deaths per year since 1977. Lightning-related fire deaths therefore represented only .3% of total fire deaths as represented by the NFPA. If all these deaths were coded as fire (E890-99) rather than as lightning (E907)in vital-statistics reports, the total number of lightning-related deaths would increase by about 20%, to about 100 per year.
ANNALS OF EMERGENCY MEDICINE
Weigel 1 also noted that vital statistics may not identify the lightning origin of fatalities caused by lightning-felled trees and electrocution by lightning-downed power lines. The Colorado study and a review of anecdotal reports in several annual StormDatasummaries suggest that deaths from such secondary effects of lightning are rare. Statistics for nonfatal lightning injuries are far more elusive. Given the dependence of Storm Data on media reports, the NOAA statistic appears to seriously underreport lightning injuries. Federal data systems useful in other injury contexts are not helpful because lightning is not a consumer product and is more than just an occupational hazard. The effect of lightning on hospital admissions can be tabulated only in states that collect data on all hospital discharges and require the external cause of injury {E code) to be recorded. There are few large sets of outpatient injury data and too few lightning injuries for a reliable total incidence statistic to be produced from the National Health Interview Survey. The 1:6 ratio of deaths to injuries in lightning-related fires cannot reliably by projected to all lightning injury. It appears that an estimate of 100 deaths per year from lightning in the United States is currently reasonable. However, the total number of lightning injuries, as Cheringten notes, is probably much larger than the average of 264 per year most recently reported by the National Climatic Data Center. Just how much larger is a matter for conjecture, pending improvement in manifestations of lightning injury described by Cherington and Andrews8 and the likelihood that the threshold of injury survival is rising, Cherington's concern about the absence of a central agency to record detailed information on lightning injuries is well taken. PeterA Brigham,MSW Bum Foundation Philadelphia, Pennsylvania 1. Weigel EP: Lightning: The underrated killer. NOAA reprint 1976;6:2. 2. HoIlifield J: National summary of lightning. 1992. Storm Data 1992;34:12. 3. Lopez RE, Holle RL, Heitkamp TA, et al: The underreporting of lightning
injuries and deaths in Colorado. Bull Am Meteorol Soc I993; 74:2171-2178. 4. Vital Statistics of the United States. Hyattsville, Maryland, National Center for Health Statistics, 1980-1991,
5. Baker SP, O'Neill B, Karpf R5: The Injury Fact Book. Lexington, Massachusetts: DC Heath & Co, 1984, pp 149-152. 6. Duclos PJ, Sanderson LM: An epidemiological description of lightningrelated deaths in the United States. IntJ Epidemio11990;I 9:673-679. 7. National Fire Protection Association: Annual Average of Fires and Losses Reported to US Fire Departments Caused by Lightning, 1989-93. 8. Andrews C: Keraunomedicine: A discipline come of age. Ann Emerg Meal 1995;25:543-545.
No Gag Rule for Intubation To the Editor. Some institutions, some textbooks of emergency medicine 1, and most recently the American College of Emergency Physicians clinical policy on the treatment of poisoned patients [May 1995; 25:571-576], mention the presence of a "gag reflex" in the assessment of the need for airway intervention in patients with depressed alertness. It is stated as though this is of major significance in the decision to intubate. However, this may be yet another dictum passed down that seems logical but explicit discussion of its significance is never undertaken. The implication is that a patient with an intact gag reflex can protect the airway, whereas one in whom the gag reflex is absent requires intubation or that it is at least additive data in the airway decision. Several points argue against using the gag reflex. First, when present, gag reflexes are highly variable among patients and even in the same patient. When sensitized by nausea, a patient whose gag reflex is normally not easily elicited may vomit with minimal provocation. Second, the gag reflex may be absent in a patient who can protect the airway. A small study by Ruffin 2 of 127 neurologically normal patients showed absence of the palatal reflex in 31% and absence of the pharyngeal reflex in
529
CORRESPONDENCE
33%. Kulig et al3 noted that 4 of 18 faculty membersdid not havea gag reflex. Davies4 studied 140 patients, some old and some young, and noted absenceof the gag reflex in 37% In some circumstances patients are drowsy but do not need intubation, despitethe fact that the gag reflex is absent.A study by Homer5 in stroke patientsshowed that the absence(or presence)of the gag reflex is not an independent predictor of aspiration. Third, an intact gag reflex, as well as the more pertinent cough reflex, does not guaranteethat the supine patient will not aspirate. In the Homer study, patients with intact gag reflexes aspirated. In Kulig's series he noted the presenceof the gag reflex in 12 "significantly obtunded" patients, 6 of whom had required intubatien and 3 of whom had aspiration pneumonia. Fourth, is it safe? The gag reflex is the lower end of a spectrum the full manifestation of which is induction of vomiting. It seems counterproductive to stimulate a reflex known to induce vomiting in a patient whose airway protection is in question. Is it not the fear of the gag reflex that prompts us to rapid-sequenceintubate the "furl-stomach" patient? Also, when increasedintracranial pressure is in question, induction of gagging could be detrimental. Except for studies published as letters, there is surprisingly little about this reflex in the literature. Most is in the dentai literature, where a hyperactive gag reflex has been an impediment to dental care. Two studies examined the relationship between the gag reflex and the Glasgow Coma Scale (GCS). Moulton 6 concluded that the gag reflex "may be significantly attenuated at all levels of the Glasgow Coma Scale," including "more conscious patients" (GCSscore of more than 8). He suggestedthat the gag reflex be assessed independently of consciousnessas an indicator of the airway at risk. Although they document the variability of this reflex, we believe his data de not support his conclusion. The "more conscious patients" with depressed reflex may not be at risk and in need of intubation, as Homer's study suggests.5 A study by Char7
530
of 41 patients showed that an initial GCS score of 8 or less had a sensitivity of 90% and a specificity of 95% for predicting the need for intubation. Absenceof the gag reflex on admissionyielded a sensitivity of 70% and specificity of 100%. Unfortunately,we are not told what the physiciansrelied on in their decision to intubate. The gold standard was "assessedby attending physicians as requiring airway protection." If these physiciansbelieved that absenceof the gag reflex meant the patient should be intubated, then all such patients would be intubated, yielding 100% sensitivity. Despite this, it is significant that the presence or absence of the gag reflex added nothing to the GCS score of less than 8 in the determination of the probability of intubatien. In summary, it appears that cerebral function should be the main determinant of the need for intubatien and that the gag reflex is of little value, of some risk, and should not be used. We hope the input of others will clarify this issue and that a thoughtful consensus will be reached.
James CKelb, MD RobertL Gafli,Me, FACEP University Medical Center Jackson, Mississippi 1. May HL (ed): EmergencyMedicine, ed2. 1992, p 1601. 2. Ruffin R, Rachootin P: Gag reflex in disease (letter). Chest 1987;92:1130. 3. Kulig K, Rumack BH, Rosen P: Gag reflex in assessing level of consciousness (letter). Lancet 1982;565. 4. DaviesAE, Kidd D, Stone SP, et al: Pharyngeal sensation and gag reflex in healthy subjects. 1995;345:487-488. 5. Homer J, Massey EW, Riski JE, etal: Aspiration following stroke: Clinical correlates and outcome. Neurology 1988;38:1359- 1362. 6. Moulton C, Penwcook A, Makower R: Relation between Glasgow Coma Scale and the gag reflex. BMJ 1991;303:1240I241. 7. Chan B, Gaudry P, Grattan-Smith TM, et al: The use of Glasgow Coma Scale in poisoning.
In reply: Thank you for your thoughtful and referenced letter discussing the value of a "gag reflex" in the assessment of the risk of airway compromise for poisoning patients. The need to appropriately evaluate the
risk of airway compromiseand to protect patients from avoidable complications is an important aspect of clinical management.As you note, the methodstaught and promoted have not been rigorously studied or evaluated. The Clinical Policies Committeewished to draw attention to the importance of assessing the risk of airway compromise, and we used our experience and understandingof current practice to establish a consensus. We realize that part of the benefit of establishing such positions is that it provides the basis for critical discussion, challenge of traditional practice, and focus for additional clinical investigation. We appreciate your comments and references and will use them in our deliberations during the revision and update process for the policy. Perhaps your letter also will stimulate more research in this area that will provide additional information on the best methods for us to use in clinically assessing and managing patients' airways. The guidelines are scheduled for review and revision every 3 years, although if critical changes occur, requiring more expeditious revision, that would occur. Your participation in the process of establishing guidelines that will evolve and improve over time is greatly appreciated.
Earl E Smith III, Me, FACEP Chairman,ACEPClinicalPolicies Committee
Admission Criteria for Stroke To the Editor. Dr Hennemanand Dr Lewis have made a valuable contribution to the emergency medicine literature with their study of admission criteria for victims of stroke or TIA [April 1995;25:458-463].However, the grouping of all such victims for TIA may not be justified. Although a recent American Heart Association special report concluded that the decision to hospitalize a TIA patient should be individualized1, two other recent papers have actually constructed algorithms based on the fact that some subsets of TIA
patients are at considerablyless risk of complications. 2,3 In general, patients with more than four TIAs in the 2 weeks before admission, a cardiac source, severe deficit, or multiple TIAs over a period of a day increasing in severity (the socalled crescendo TIA) may be at higher risk for stroke and probably should be admitted. Amaurosis and fewer than five TIAs, the last one more than 2 weeks before presentation, pose lesser risk, and patients with such signs may be discharged pendingfurther outpatient testing. Although no one has suggested it in these papers, there may be a role for observationand expedited testing in an observationfacility akin to those used in chest pain settings for some of the TIA cases that may fall in the gray zone. Henneman and Lewis had only five TIAs in their study that met their criteria for admission justification, and one of these was for "another diagnosis." This small sample size precludes conclusions. However, of these, two actually had adverse events. It would be interesting to know what these events were and how these TIAs fit in the categories mentionedabove.
JB Leibman,MD Departmentof EmergencyMedicine Albany MemorialHospital Albany, New York 1. Feinberg WM, et al: Guidelines for the management of TIAs. Circulation 89:2950-2965. 2. Frey JL: Decision making in TIA and stroke. Barrow Institute Quarterly 1994;10:22-29. 3. Brown RD, etal: TIA and minor ischemic stroke: An algorithm for evaluation and treatment. Mayo Clin Proc 1994;69:1027-1039.
To the Editor. Henneman and Lewis are to be commended for a most appropriate article. However, I think further review of their data might alter their conclusion. I cannot determine from their data how many of the "net medically justified" admissions for nonhemarrhagic stroke or TIA resulted in "complications." Furthermore, of those patients with complications, how many had better outcomesas a result of admission?What if these patients had been dischargedafter
ANNALS OF EMERGENCY MEOICLNE 26:4
OCTOBER 1995
ture an estimate of 100 lightningrelated deaths and 300 injuries annually in the United States. He then cites a 1976 National Oceanic and Atmospheric Administration (NOAA) publication in suggesting, "Most authorities agree that the numbers are much larger. "1 The difficulties in tabulating lightningrelated deaths and injuries are quite different and deserve further comment. Several unrelated data sources (meteorologic, vital and health statistics, and fire) provide information on death and injury resulting from lightning in the United States. The only source that specifically tracks lightning-caused death and injury is the monthly publication Storm Data from the NOAA National Climatic Data Center, which relies heavily on newspapers fer information on casualties due to lightning. The annual averages of lightning deaths and injuries reported to NOAA from 1959 through 1992 were 92 and 264, respectively.2 Lopez et al3 found that NOAA captured only 72% of the lightning-related deaths recorded on Colorado death certificates and only 58% of the patients with lightning-related injuries discharged from Colorado hospitals from 1988 to 1991. The shortfall in NOAA reports of patients with lightning injuries treated and discharged from outpatient settings could not ever be calculated in the absence of statewide outpatient data. The most comprehensive counts of lightning deaths are provided by national vital-statistics reports. These are prepared annually by the National Center br Health Statistics from data supplied by state health departments, which are drawn from death certificates submitted by medical examiners. Between 1980 and 1991, deaths att!ibuted to lightning by NCHA averaged 87 per year, ranging from a h!gh of 100 in 1982 to a low of 75 in 19;89and 19914 The continuing long-term decline in such deaths is reflected in the average of 92 for the first sik years of this period and the aver&ge of 83 for the second six yearsi Baker et al5 noted that from 1930 t 9' 1980, the death rate from lightning showed one of the largest decreases for any injury (from .29 per 1001000 in 1930 to .04
OCTOBER 1994
26:4
per 100,000 in 1980). The decrease mostly reflects improvements in electrical systems and a decrease in the population of farmers, although a change in vital-statistics coding methods in the early 1950s is partly responsible and improved treatment may have reduced fatalities. Because vital-statistics reports quantify injury-related deaths with respect to a single underlying cause, such statistics do not include all lightning-related deaths. A more complete profile can be obtained from medical-examiner reports, which list all contributing causes. When Duclos and Sanderson 6 compared NCHS data for t972-1984 with medical-examiner reports from North Carolina, they found that 8 of 64 North Carolina medical-examiner reports including a lightning-related diagnosis (an additional 14%) did not appear in the NCHS database. Duclos and Sanderson6 did not address possible systematic reasons for the omission of these eight cases. Some such omissions may reflect deaths resulting from structure fires started by lightning. Weigel 1 had hypothesized in 1976 that because 2% of all fires were started by lightning, 2% of all fire deaths might be similarly ascribed to lightning. If these deaths were all coded as fire deaths rather than lightning deaths by medical examiners, this would have increased total lightning deaths by 70%, from the 124 reported by NCHS in 1973 to 211. In fact, occupants of a residence ignited by lightning who are not struck themselves are rarely trapped by the fire. The National Fire Protective Association (NFPA), however, has estimated only 14 deaths in fires started by lightning and 90 lightning-related fire injuries per year to civilians (nonfirefighters) for the period 1989-19937 and an average of four firefighter deaths per year since 1977. Lightning-related fire deaths therefore represented only .3% of total fire deaths as represented by the NFPA. If all these deaths were coded as fire (E890-99) rather than as lightning (E907)in vital-statistics reports, the total number of lightning-related deaths would increase by about 20%, to about 100 per year.
ANNALS OF EMERGENCY MEDICINE
Weigel 1 also noted that vital statistics may not identify the lightning origin of fatalities caused by lightning-felled trees and electrocution by lightning-downed power lines. The Colorado study and a review of anecdotal reports in several annual StormDatasummaries suggest that deaths from such secondary effects of lightning are rare. Statistics for nonfatal lightning injuries are far more elusive. Given the dependence of Storm Data on media reports, the NOAA statistic appears to seriously underreport lightning injuries. Federal data systems useful in other injury contexts are not helpful because lightning is not a consumer product and is more than just an occupational hazard. The effect of lightning on hospital admissions can be tabulated only in states that collect data on all hospital discharges and require the external cause of injury {E code) to be recorded. There are few large sets of outpatient injury data and too few lightning injuries for a reliable total incidence statistic to be produced from the National Health Interview Survey. The 1:6 ratio of deaths to injuries in lightning-related fires cannot reliably by projected to all lightning injury. It appears that an estimate of 100 deaths per year from lightning in the United States is currently reasonable. However, the total number of lightning injuries, as Cheringten notes, is probably much larger than the average of 264 per year most recently reported by the National Climatic Data Center. Just how much larger is a matter for conjecture, pending improvement in manifestations of lightning injury described by Cherington and Andrews8 and the likelihood that the threshold of injury survival is rising, Cherington's concern about the absence of a central agency to record detailed information on lightning injuries is well taken. PeterA Brigham,MSW Bum Foundation Philadelphia, Pennsylvania 1. Weigel EP: Lightning: The underrated killer. NOAA reprint 1976;6:2. 2. HoIlifield J: National summary of lightning. 1992. Storm Data 1992;34:12. 3. Lopez RE, Holle RL, Heitkamp TA, et al: The underreporting of lightning
injuries and deaths in Colorado. Bull Am Meteorol Soc I993; 74:2171-2178. 4. Vital Statistics of the United States. Hyattsville, Maryland, National Center for Health Statistics, 1980-1991,
5. Baker SP, O'Neill B, Karpf R5: The Injury Fact Book. Lexington, Massachusetts: DC Heath & Co, 1984, pp 149-152. 6. Duclos PJ, Sanderson LM: An epidemiological description of lightningrelated deaths in the United States. IntJ Epidemio11990;I 9:673-679. 7. National Fire Protection Association: Annual Average of Fires and Losses Reported to US Fire Departments Caused by Lightning, 1989-93. 8. Andrews C: Keraunomedicine: A discipline come of age. Ann Emerg Meal 1995;25:543-545.
No Gag Rule for Intubation To the Editor. Some institutions, some textbooks of emergency medicine 1, and most recently the American College of Emergency Physicians clinical policy on the treatment of poisoned patients [May 1995; 25:571-576], mention the presence of a "gag reflex" in the assessment of the need for airway intervention in patients with depressed alertness. It is stated as though this is of major significance in the decision to intubate. However, this may be yet another dictum passed down that seems logical but explicit discussion of its significance is never undertaken. The implication is that a patient with an intact gag reflex can protect the airway, whereas one in whom the gag reflex is absent requires intubation or that it is at least additive data in the airway decision. Several points argue against using the gag reflex. First, when present, gag reflexes are highly variable among patients and even in the same patient. When sensitized by nausea, a patient whose gag reflex is normally not easily elicited may vomit with minimal provocation. Second, the gag reflex may be absent in a patient who can protect the airway. A small study by Ruffin 2 of 127 neurologically normal patients showed absence of the palatal reflex in 31% and absence of the pharyngeal reflex in
529
CORRESPONDENCE
33%. Kulig et al3 noted that 4 of 18 faculty membersdid not havea gag reflex. Davies4 studied 140 patients, some old and some young, and noted absenceof the gag reflex in 37% In some circumstances patients are drowsy but do not need intubation, despitethe fact that the gag reflex is absent.A study by Homer5 in stroke patientsshowed that the absence(or presence)of the gag reflex is not an independent predictor of aspiration. Third, an intact gag reflex, as well as the more pertinent cough reflex, does not guaranteethat the supine patient will not aspirate. In the Homer study, patients with intact gag reflexes aspirated. In Kulig's series he noted the presenceof the gag reflex in 12 "significantly obtunded" patients, 6 of whom had required intubatien and 3 of whom had aspiration pneumonia. Fourth, is it safe? The gag reflex is the lower end of a spectrum the full manifestation of which is induction of vomiting. It seems counterproductive to stimulate a reflex known to induce vomiting in a patient whose airway protection is in question. Is it not the fear of the gag reflex that prompts us to rapid-sequenceintubate the "furl-stomach" patient? Also, when increasedintracranial pressure is in question, induction of gagging could be detrimental. Except for studies published as letters, there is surprisingly little about this reflex in the literature. Most is in the dentai literature, where a hyperactive gag reflex has been an impediment to dental care. Two studies examined the relationship between the gag reflex and the Glasgow Coma Scale (GCS). Moulton 6 concluded that the gag reflex "may be significantly attenuated at all levels of the Glasgow Coma Scale," including "more conscious patients" (GCSscore of more than 8). He suggestedthat the gag reflex be assessed independently of consciousnessas an indicator of the airway at risk. Although they document the variability of this reflex, we believe his data de not support his conclusion. The "more conscious patients" with depressed reflex may not be at risk and in need of intubation, as Homer's study suggests.5 A study by Char7
530
of 41 patients showed that an initial GCS score of 8 or less had a sensitivity of 90% and a specificity of 95% for predicting the need for intubation. Absenceof the gag reflex on admissionyielded a sensitivity of 70% and specificity of 100%. Unfortunately,we are not told what the physiciansrelied on in their decision to intubate. The gold standard was "assessedby attending physicians as requiring airway protection." If these physiciansbelieved that absenceof the gag reflex meant the patient should be intubated, then all such patients would be intubated, yielding 100% sensitivity. Despite this, it is significant that the presence or absence of the gag reflex added nothing to the GCS score of less than 8 in the determination of the probability of intubatien. In summary, it appears that cerebral function should be the main determinant of the need for intubatien and that the gag reflex is of little value, of some risk, and should not be used. We hope the input of others will clarify this issue and that a thoughtful consensus will be reached.
James CKelb, MD RobertL Gafli,Me, FACEP University Medical Center Jackson, Mississippi 1. May HL (ed): EmergencyMedicine, ed2. 1992, p 1601. 2. Ruffin R, Rachootin P: Gag reflex in disease (letter). Chest 1987;92:1130. 3. Kulig K, Rumack BH, Rosen P: Gag reflex in assessing level of consciousness (letter). Lancet 1982;565. 4. DaviesAE, Kidd D, Stone SP, et al: Pharyngeal sensation and gag reflex in healthy subjects. 1995;345:487-488. 5. Homer J, Massey EW, Riski JE, etal: Aspiration following stroke: Clinical correlates and outcome. Neurology 1988;38:1359- 1362. 6. Moulton C, Penwcook A, Makower R: Relation between Glasgow Coma Scale and the gag reflex. BMJ 1991;303:1240I241. 7. Chan B, Gaudry P, Grattan-Smith TM, et al: The use of Glasgow Coma Scale in poisoning.
In reply: Thank you for your thoughtful and referenced letter discussing the value of a "gag reflex" in the assessment of the risk of airway compromise for poisoning patients. The need to appropriately evaluate the
risk of airway compromiseand to protect patients from avoidable complications is an important aspect of clinical management.As you note, the methodstaught and promoted have not been rigorously studied or evaluated. The Clinical Policies Committeewished to draw attention to the importance of assessing the risk of airway compromise, and we used our experience and understandingof current practice to establish a consensus. We realize that part of the benefit of establishing such positions is that it provides the basis for critical discussion, challenge of traditional practice, and focus for additional clinical investigation. We appreciate your comments and references and will use them in our deliberations during the revision and update process for the policy. Perhaps your letter also will stimulate more research in this area that will provide additional information on the best methods for us to use in clinically assessing and managing patients' airways. The guidelines are scheduled for review and revision every 3 years, although if critical changes occur, requiring more expeditious revision, that would occur. Your participation in the process of establishing guidelines that will evolve and improve over time is greatly appreciated.
Earl E Smith III, Me, FACEP Chairman,ACEPClinicalPolicies Committee
Admission Criteria for Stroke To the Editor. Dr Hennemanand Dr Lewis have made a valuable contribution to the emergency medicine literature with their study of admission criteria for victims of stroke or TIA [April 1995;25:458-463].However, the grouping of all such victims for TIA may not be justified. Although a recent American Heart Association special report concluded that the decision to hospitalize a TIA patient should be individualized1, two other recent papers have actually constructed algorithms based on the fact that some subsets of TIA
patients are at considerablyless risk of complications. 2,3 In general, patients with more than four TIAs in the 2 weeks before admission, a cardiac source, severe deficit, or multiple TIAs over a period of a day increasing in severity (the socalled crescendo TIA) may be at higher risk for stroke and probably should be admitted. Amaurosis and fewer than five TIAs, the last one more than 2 weeks before presentation, pose lesser risk, and patients with such signs may be discharged pendingfurther outpatient testing. Although no one has suggested it in these papers, there may be a role for observationand expedited testing in an observationfacility akin to those used in chest pain settings for some of the TIA cases that may fall in the gray zone. Henneman and Lewis had only five TIAs in their study that met their criteria for admission justification, and one of these was for "another diagnosis." This small sample size precludes conclusions. However, of these, two actually had adverse events. It would be interesting to know what these events were and how these TIAs fit in the categories mentionedabove.
JB Leibman,MD Departmentof EmergencyMedicine Albany MemorialHospital Albany, New York 1. Feinberg WM, et al: Guidelines for the management of TIAs. Circulation 89:2950-2965. 2. Frey JL: Decision making in TIA and stroke. Barrow Institute Quarterly 1994;10:22-29. 3. Brown RD, etal: TIA and minor ischemic stroke: An algorithm for evaluation and treatment. Mayo Clin Proc 1994;69:1027-1039.
To the Editor. Henneman and Lewis are to be commended for a most appropriate article. However, I think further review of their data might alter their conclusion. I cannot determine from their data how many of the "net medically justified" admissions for nonhemarrhagic stroke or TIA resulted in "complications." Furthermore, of those patients with complications, how many had better outcomesas a result of admission?What if these patients had been dischargedafter
ANNALS OF EMERGENCY MEOICLNE 26:4
OCTOBER 1995