Proceedings of a workshop on near fatal asthma

Concern has been expressed about rising asthma morbidity and mortality, although the latter appears to have declined recently. A reasonable surrogate for fatal asthma is an episode of near fatal asthma (NFA). The etiology of episodes of NFA appears to be multifactorial. Features that would characterize asthma patients at risk of NFA have been difficult to define but have included psychosocial barriers. environmental exposures, inadequate or inappropriate physician and/or patient responses to deteriorating asthma and, in particular, overreliance on symptomatic bronchodilator therapy. The association between fatal asthma and NFA with beta-agonist use has been controversial, with it being argued that high use of beta-agonists reflects severity of asthma as opposed to being causal. Studies in the laboratory and ambulatory care setting suggest that regular compared with as-required use of beta-agonists is associated with worsening in asthma control. Although a reduced perception of dyspnea has been identified in some asthma patients, it is not universally present in those with NFA. Retrospective data suggest that hyperinflation of the thorax, as judged by total lung capacity, may be a useful marker for subjects at risk of NFA. Future studies should better characterize these risk factors and develop management strategies (both therapeutic and educational) that might reduce the risk of subjects experiencing episodes of NFA and, by extension, reducing the continued unacceptable mortality associated with asthma.

T HFlff IS IN REA SING CONCERN OVER TII E RISI NG RATES of a\thrna monality ( 1,2).Patients wiih near fatal asthma (NFA) represent an impo rtant gro up of.ubjects that can hl' l'valuated to de fine ri sk fa ctors for fatal asthma.To rev iew thl' curre nt data re lated to NFA , the Respiratory Heal th Network or Centres of Excellence and the Laboratory Centre for Disease Control (LCDC ), wi th significant contributions from a num be r of pharmaceutical companies, hosted a workshop on NFA that was he ld in the Montreal Chest Institute o n April 28 and 29, 1994.The objectives of this workshop we re to rev ie w the current e pidem iology of asthma in Canada, as wd l as global tre nds in asthma morbidity and mort ality .In addition.the publi shed li te rature on NFA was revie wed and spec ific ri sk factors were evaluated in more detail.T he oppnrtllnity was also taken to present preliminary results from a number or studies on near fatal and fata l asthma from th e Prairie provinces .New Ze aland and Vancouver.It was also hoped tktt, following a review of the curre nt know ledge wit h regard to NFA. future stud ies can be planned in Canada to evaluate the pathophys iological mec hani sms associated w ith the fi nal common pathway leading to near fa tal and fatal asthma.to identify thro ug h prospecti ve epide m iol ogica l surveillance studies ri sk factors for NFA and, finally .to deve lop interve ntions that might reduce the docume nted high risk of subjects who have an episode of NFA from progress ing to a futurl' N FA attack and.more particularly, death .

FATAL AND NEAR FATAL ASTHMA REVIEW
Dr J Mark FitzGerald reviewed the c urren t li terature and gave a sy nopsis of docume nted ris k factors fo r NFA , usually de fined as a seve re life-threaten ing asthma attack re 4uiring mechanical ventilation or in te nsive treatmen t in the presence of ;1euk hype rca pnic respiratory fail ure .The mu lt ifac torial as pec ts of a n episode of NFA were hi ghlighted with a complex int eraction among the patient .therapeut ic and enviro nme ntal fac tors, and the intervention or lac k of interve ntion of physic ians (3).The documrnted high ri sk of a subsequent episode o f NFA or death follow ing an episode of mec hanical wntilation for NFA was stressed (4.5).These and othe r issues were further developed in more detai l by subsequent speake rs.

CANADIAN ASTHMA MORTALITY AND MORBIDITY: AN EPIDEMIOLOGICAL REVIEW IN CANADA
Dr Fel ix I .irev iewed the recent trends in asthma morbidity and morta lity in Canada.T he rising trend of asth ma monali ty and morbidity in Canada since the 1970s has raised much conce rn among health professionals.asthma pati<'nts and th e gene ral public (6,7).This tre nd is particularly worrying because it occurred despite continu ing medical advances in the understanding and treatment of asthma.Studi es have ;il,o shown th at many as thma deaths and hospita lizations are pot,:nti ally preve ntable (8.lJ).
This presentation rev ie wed the important epidemiological data on asthma mortality and morbidity in Canada from 196 1 to 199 I, and updated information provided in previous re-

Mortality :
A revie w of data from I% I shows that a signi Iicant rise in asthma mortality in Canada occurred principally in the age group IO to 34 years (Figure I), in both sexes .In this age grou p, ast hma mortality started to rise in 1971.By 1981.its age-standa rdized asthma mortality rate had risen to 2.7 times its 1971 level.T here followed a plateau in rates that lasted most of the 1980s.In 1991.the age-standardized mortality rate fo r this age group was still 1.9 times its 1971 le vel.
During the recent pe riod ol 1986 to 199 1. ast hma morta lity in the I() to .14 yl'ar age group was hi ghest in Alberta and Saskatchewan.
Hospitalization: Hospi tal admission rates for asthma (Fig- ur<' 2) in Ca nada started to rise sharply in the mid-I 970s.and reached peak levels in 1986 to 1988.T hereafter, it has shown some small decline (latest available data are from 1990).The ris ing trend of asthma hospitalization was contri buted mainly by the increase in the O to .14 year age group (Figure 2).3) , whose latL•st asthma hospital separation rates were 4. 6 and J.6 times their respective l 97 I rates.In the under 35 year age group, males consistently have a higher (by 33%) hospital separation rate than their kmalc counterparts .and Prince Edward Island, New Brun.,wickanJ Nova Scotia had the highest asthma separation rates in recent years ( I •)86 to 1990).MorhiJity.in terms or annual hospital-days due to asthma.however, showed no net ri.,ing trend hct ween 1971 and 1990.ll1is is because the rise in asthma hospital-days in the younger (unJn 25 years) age group was ofl\ct by the decrease in the older (over 25 years) age groups (Figure 4).Prince Edward Island, New B111nswiek and Nova Scotia again had the highest asthma hospital-days rates ( 1986 to 1990).P revalence: The Canada Health Survey ( 1978 to 1979) estimated asthma (self-reported) prt'v alencc in Canada to he 2.4% (2.3 % in the population over 15 years old}.Suhscquently, various provincial surVL'ys yieldeJ prevalence rates between 2.5 % and 5.3 %.Recently.results or the General Social Survey ( 199 I ) ( IO) showed an asthma prevalence or 6';.'.. among C anadians aged 15 years anJ abnve.l'rcvalenn: was equal in males and fem;iles; highest (l)c;{, ) among l .'ito24-year-o ld., : and highest in Quebec (7 %}.However. it is unclear how much the apparent increase between I lJ78 and I LJY I (from 2.3'k to ()';n can be attrihutcd to 'diagnostic tra ns fer ' from other ob.,tructive pulmonary diseases.
Thus. over the past two decaJes, asthma mortality and hospitali1:ation trends in Canada have generally undergone two phases -• a rising phase in the 1970s, follmVL'd hy a plateau phase in the 1980s.These changes affected 111;1inly the population under 35 years old.Obviously.the inneascd asthma mortality and morhidity have haJ a s ignificant adverse impact on the health and well-being of CanaJians.as well a., on the health care system.Although there was some decrease in asthma mortality and morbidity from I 989 to l 99 I. it would be premature to assume that this signifil's the hcginning of a decreasing trend.

GLOBAL TRENDS IN ASTHMA MORBIDITY ANO MORTALITY
Dr Malcolm Sears reviewed the data related to global trends in asthma morbidity and mortality.Dr Scars cited a study by Baum an ( I I), who published an ovcrviL•w of studies in Australia from 1970 to 1990, showing the prevalence of diagnosed asthma in primary schoolchildren increased from less than 5% to over 20% during this period.Supportin° data suggested this was at least in pa11 a true increase in prevalence and not simply a change in diagnostic labelling.Simil ar data have been reported among 17-year-old males recruited into the Israel defe nce fo rces ( 12) as well as subjects recruited into the Finnis h defence forces ( 13 ), In the United States, Ge r.,tman cl al ( 14) reported increased asthma hospitalization., in five-lo 14-year-olJs in the Mic hi g an Medicaid population: increases occurred particu la rly in five-to nine year-olds with a slight increase in IO-lll 14-year-olds.In Australia.Kun cl al ( l 5} reporte d an increased proportion of emerge ncy room visits or hospitali1:alions due to asthma.In the Un ited Kingdom.hospital discharges and deaths from asthma increased approxim~1tcly twofold from 1979 to 1985 ( 16), while most other respiratory diagnoses remained stahk or decreased.
An examination or aJmissions and deaths from chronic hronchitis, bronchiti ., unspecified , emph ysema and a.,lhma data indicated that the increase in asthma deaths was not simply e xplained by diag nostic transfer hctween asthma and other diseases of airflow obstruction.Dr Scars outlined problems with mortality stati stics, wh ich inl'l11dc the true mortality from asthma together with effects from false positives and false negat ive certifications.anJ indicateJ that in the age group five to .14 years.lhe accurac y of Jiagnosis and certification of asthma deaths i., nearly IO(Vii ( 17.18}.Within thi., age group, asthma mortality has increased over the past two decades in several countries.including Eng land and Wales, western Gennany.Australia, Japan, Ne w Zealand and the Un ited States.It was indicated again that th e increase may have reached a plateau over the past few years and may now be decreasing.In the United States significant ethnic Jiffer- ences.with hi gher mortality among blacks than whi tes am.l si gnificant regional differences with highe r mortality in urban compared with rural or outlying areas , have been documented ( 19).
Dr Sears brie fly discussed the issue of beta-agon ist treatment.and asthma mortality and the data rel ating to thi s issue werL'.further expanded upon by Dr Pierre Ernst (see be low) (20)(21)(22).A recent substantial decline in asthma mortality since 1989 in New Zea land was reported (personal communication).It was noted that the beta-agonist fenoterol was fi rst linked wi th excess mortality in 1989 (20).Suissa et al (22) attri buted the 1989 to 199 1 decline in mortality to a steady downward trend that had been ev ident since the peak of the epi demic in 1981, but the initial downward trend was small and mortality reached a plateau from 198 3 to 1988 .The subseq uent abrupt downward trend in 1989 through 1991 was assoc iated with restriction and withdrawal of fenoterol from the drug tariff in New Zealand.In conjunction with this reduction in mortality , there has been a substantial reduction in morbid ity from asthma as measured by almost a 50% decrease in hospital admissions in fi ve-to 54-year-o lds, and a 75 % dec rease in intensi ve care admissions (personal communicati on).The decline in both mortali ty and hosp ital admissions indicates that the former is not due to better hospital access or emergency care, but rather to decreased severity of disease.T rends away from frequ ent bcta-agonist use towards more appropriate use of anti-infl ammatory therapy were felt to be in part responsible fo r the slight downward trends seen recently in other countries.

PSYCHOSOCIAL FACTORS IN NFA
Dr Paul G reenberger outlined reasons for patient noncompliance, identifyi ng psychiatric disorders, psychological condit ions.envi ronmental and social factors, cognitive overload such as in the geriatric patient with potentially fatal asthma, ineffective physician-patient communication or cri sis planning, and poor patient acceptance of their asthma and the associated patient responsibilities req uired to prevent acute life-threatening a thma (23.24).Dr G reenberge r noted that from a grou p of 21 patients .aged 45 years or less, who were intubated during acute life-threaten ing asthma.a third had left the allergy service within a year (23) indicati ng the poor compliance with care in this group.Three of these seven patients died from asthrna in that year.Dr Greenberger outli ned the effects of psychosocial problems as the y related to the patient with severe asthma, and these are li sted in Table  (24).He noted that such factors can force a change in priorit ies such that the patient with chronic severe asthma does not accept achieving effective management of asthma as a personal priority.Practically.it may be difficult to obtain adequate amounts of medications, arrange transportation and follow-up with appropriate trained phys icians to work to prevent the next hospitalization.Alternatively.physician facto rs have contributed to patients meeting criteria for potentia lly fatal asthma in that the physicians fail to understand ambul atory manage ment of severe asthma, fail to understand the se verity of ind ividual cases or learn to practise bizarre forms of asthma management.
To test the hypothes is that many cases of fa tal asthma were attributed to potentially preventable causes.an investigation of asthma deaths in 39 cases aged 45 years and less was carried out (25 ).The authors.surprisingly, found that 14 of 23 (60.8% ) cases where toxicol ogica l detem1inations for illicit drugs were performed were positive, suggesting an unexpected confound ing fac tor of substance abuse.In discussion it was felt that such leve ls of substance abu e that might be present in inner city areas in the United States may not be relevant to asthma care in Canada.

BET A-AGONIST AND ASTHMA MORT ALITV -OUTSTANDING QUESTIONS AND IMPLICATIONS FOR FUTURE STUDIES
Dr Pierre Ernst reviewed the literature on the association between the use of inhaled beta-agonists and deaths from as thma.Th is association was originally described in the mid-I 960s when a temporal relationship between increasing saks of isoproterenol and an excess of asthma deaths was found in young people in England and Wales (26).The assoL•iation was ecological only; that is, there was uo evidence that subjects dy ing from asthma were actually using betaagonists.For this reason no conclusions could be made as to causal ity.A similar association between sa les of knoterol and asthma deaths in Ne w Zealand was suggested in J lJKlJ (27): ho wever, the comparison group in this study comprised subjects prescribed sa lbutamo l, and it was therefore impossible to discern whethe r an associ ation with life-threatening asthma was present for most drugs in this class or was restricted to fenoterol.
The Saskatchewan Asth ma Epidemiology Project (SAEP) identified a cohort of 12,30 I suhjects aged five to 54 years dispe nsed 10 or more prescriptions for asthma drugs be tween 1980 and 1987 (28).Among a nested case control sample of I 29 cases of Ii fc -threateni ng asthma and 655 control patie nts , the re was a strong dose-response relati onship between increasing use of inhaled beta-agonists and the risk of fatal and NFA , with a mcm• than 20-fold innease in risk for subjects dispensed more than two canisters per month of either salbu-tamol or fenotc rol.An analysis of the complete cohort, which allowed calculation of ahsolutL' rates nf asthma lkath, strongly suggested that the excess in risk of NFA and (kath was largely confined to subjects dispensed more than 1.5 canisters per month (29,JO) .
The nonex peri mental design of the SAEP makes it impossible to di stinguish a causal association from one that results from the way in which asthma medications were prescribed.Dr Ernst noted that it seemed plausible that subjects who u •e large amounts of bcta-agonists have more severe asthma, and that it is the severity of the disease rather than the exposure to these drugs that dete m1inecl th e excess in risk of life-threatening asthma.In an attempt to shed light on this matter.clinical indicators of asthma severity we re e xam ined in the case control sample from the SAEP (29).Wh ile several clinical eleme nts were found to be associated with an excess ri sk of fatal and NFA. for e xampk a hi story of loss of consciousness or of attacks precipitated hy food, adjustments for such independe nt risks did not alter the assoc iation between heta-agonists and the risk of a lire-threateni ng e ve nt.The ava ilable clinical data were sparse.howeve r, so that disease severity was un likely to have been accounted for completely.Unce rtai nty as to the nature of the associ ati on between li fe-threatening asthma and he ta-agon ist use does not prevent use of the Saskatc hewan data to identify patients at risk of an adve rse event based on the ir pattern of use of these llledications.While it is readily apparent that users of more than two canisters per month are at g reatest risk overall.a rea nalysi s of patterns of use among subjects using one or more cani sters per month of beta-agonist found that the g reatest increase in risk is seen among those with increasing use and that increasing use is a stronger marker of risk than the actual number of canisters dispensed (30).
There are various lines of ev idence suggesting that bet aagonist use is more than just a marker of asthma seve rity.The seminal paper by Sears and colleagues (21) provi ded evidence that patie nts using bcta-agonist bronchocl ilators on a regular basis had worse control of their asthma and a shorter time to exacerbation (31 ) than subjec ts who used th c-ir hronchodilator on an intermittent or as-needed basi s.Th is work, which needs to he confirmed in other se ttings, suggests that be ta-agon is ts used regularly may make asthma more severe and thus increase the risk of near fatal attacks.This incrcase in asth ma severity may result from an increase in airways responsiveness (32).Use of bcta-agonists has the potential to make attacks of asthma more precipitou s.Whi le de laying thl' Workshop on near fata l asthma hronchncon strictivc response-, be ta-agoni sts do not appear to limit the degree of bronchoconstriction once it starts and may actually make the foll in lung function more prec ipitous (3 3).Regular use of beta-agoni sts may be associated with a loss of their protect ive effect against induced bronchoconstriction to methac holine (34) or to AMP (35) and, mo rc irnportant, it a ppea rs to in c rease the bronchoconstrictive response lo allergen (36).While be ta-ago ni sts ce rtainly havedetec table cardiac and metabolic effects (37), these have not been convincingly linked to life-threa te ning events.On the contrary, there is good evide nce that NFA attacks are clue to respiratory failurt' without significant cardiac phenome na (38 ) .
Long -acting inhaled beta-agoni st bronchoclilators have been available for a number of years in other countries and became available in Canada in late 1994.The ir role rema ins somewhat uncertain given the concerns e xpressed above abo ut regu la r use of shorter-acti ng agents.T he effi cacy of the new long-acting pre parations in reducing sym ptoms and improving lung fun c tion is impress ive (39).No adverse e ffec ts on asthma control have been described as ye t, despite c lose observati on fo r up to a year (40).Li fe -threatening attacks of ast hma are rare eve nts , howeve r, and studies of a large number of patients may be required to demonstrate safe ty.J\n attem pt at suc h a study was recently published and found a threefo ld, though not statistically significant.exec s of asthma deaths among approximately 15.000 pat ients dispensed the long-acting agent salmcterol compared with subjects taking the short-acting salbutamol four times a day (41 ).A more interesting comparison would he regular salmetcrol versus a short-acting be ta-ago ni st used on an as-needed basis.

HOW GENERALIZABLE ARE STUDIES IN NFA TO FATAL AND SEVERE ASTHMA?
Dr Nestor Molfino rev ie wed the studies of NFA, noti ng that the majority of stuclie on near fatal and fatal asthma have been epidem iolog ical in nature.characterizing patients at risk of dying from asthma and investigating th e measures being taken by patients.relatives a nd health care workers.The major conclusions of such studies are as follows: • patients dyi ng from fatal asthma and patients who suffer from NFA see m to come from the same population: • the time elapsed to medica l care.and the quality of such care.are cruL•ial: • there appt' arL:d to be imprope r therapeutic measures taken in emergency circumstances in ~eneral because of underestimation of th e severity of the episode of ac ute asthma.
A review of the rccL' llt literature led to the conclusion that patients who die from asthma do so because they do not reach the emergency department expeditiously, and based on thi s hypothes is many authors haw studied NFA cpisodcs as a surTogate for fatal as thma (42) .In th is re gard, Molfino et al (JX) examined the IllL'l'ha nisms by which patients rni gh r die during acute e xacerhations of asthma hy studying In patients who arrived at the hospital in respiratory arrest or in whom it de veloped within 20 mins artl'.rthei r admi ssion .These patients had charactcristiL•s similar to those described earlier for patients al high risk of death from asthma.They were found 10 have marked hypercapnia (mean ± SD) 97.1 ±3 1.1 mm Hg and acidosis pll 7.0 1± 1 I before mechanical ventilation wa s begun.In additi on .four patients had hypokalcmia on admission.although for the total group the mean potassium was in the low normal rangL' (3.4±(U mmnl/L).Despite the marked severe respiratory acidosis, no patient had a serious cardiac arrhythmia during resuscitation manoeuvres or during hospi -lali1a1ion.Although one patient had atrial fibri llation and another had relative sinus bradycardia.both arrhythmias reverted to sinus rhythm after the initiation of vcntilatinn wi th hypcroxic gas mixtures.Based on these findings .ii wa.s concluded that.al least in this group of patients, the near-faral nature of the exacerbations was the result of severe asphyxia rather than cardiac arrhythmias.Given the severity of their clinical and biochemica l d;tta on admission.these patients would probably have died if they had not been treated aggressively.Thus, they rL•prese nt a subgroup of patien ts who die from asthma; this is true for otlwr studies (38,(43)(44)(45)(46)(47)(48)(49)(50)(51).
The mechanisms hy which asphyxia ckvelops in se vere asthma arc probably mullifactori;1I: a cmnhinatinn nf acertain degree ol' airway lahilit y. the severity of the pre-existing obstruction.the magnitude of the .,timulus applied, and the patient' s ability to respond 10 the alterations of the airway geometry (52-5 5).Rece ntly, Kikuchi ct al (56) reported that patients suffering from NFA have reduced chemosensi tivity to hypox ia and blunted pcrn•ption of dyspnea.
It appears reasonable to conclude from the studies that pat ients suffering from NFA and fatal asthma arc derived from the same populat ion of individuals.and that a better understa nding of n<'ar fatal events m:1y lead to better prevention of fatal episodes of asthma .

ENVIRONMENTAL FACTORS RELATED TO
ASTHMA MORBIDITY Dr Sverre Veda! rev iewed the association between as thma morbidity and the e nvironme nt.Dr Veda! noted that air contaminants and asthma both exhibit :1 tem pora l variability , which suggests that environmental air contam inants and asthma may be lin ked.Aeroallcrgen., also m:1y he causes of asthma and factors that influence the course of as thma because of the marked differenu's in prevalence of acroal lergcn sensitization bet ween asthmatiL• and nonas1hrna1ic subjects (57).Sensitivity l!11he house dust 1111te is in many sc11ings the most comrnnn aeroa llergen sens itivity in asthm,11ics.with SL'nsitization to outdoor pollens and In cats also being common.However, based on data from the National Health and Nutriti on Exam Survey in the United States, sensitization lo fun gal allergens.in particular alternaria.may hl' the most common acrnallergen se nsitivity in asthmatics in many other selli ngs (58).
Based on recent data showing that avoidance of allergens during the first yearo!'lifccan reduce the incidence of asthma by the firs t year (59).and data showing :tn association between aeroalle rgc n exposure :11 the first year and the likcli-hood of havin g asthma al age IO yea rs (60).allergens arl' likely to play ;1 role in causing asthma.Several studies also provide strong evidence that continued ex posure to allergens influences the course of asthma.Data Crom Barcelona, Spain (61) and Crom the American mi dwest (4h) show that rnrrl'llt acroallergen exposure can also result in VL'ry severe adverse effects in asthmatics, including respi ratory l"ailure.Studies in which risk f;1ctors for adverse effects in aslhmaliL•s arc studied using nonasthmatic control gro ups arc less infnrma1ivc in implicating these risk factors in NFA than when ,111 asthma control group is used.Another desi gn problem that compromises the ability o r a study to reac h conclusive findin gs is the lack of st ati stical power.For exampl e. in studying NFA.not only is it difficult to enrol enough subjects from one investigating centre, but if interest is in study ing the risk of acroallergen exposu re in the subset who arc specifically sens iti zed to that allergen.statistical power is eve n l'urlhn compromised.
Meteorologica l fac tors clearly have an 111lh1ence on ast hmatics.\Vhcreas the exerc ise-induced lall in lt•wl or lung function is exacerbated by low tempe rature and low hun1idit y, it is less clear in cliniL:al and epidemio log ical studi c. \ that low temperatu re and low humidity have similar associations with asthma exacerlx11ions (62 ).In fact, in sorne instances both extremes of temperature arc adversely associated wit h asthma course: hu midity has variable effec ts depending on ambient temperatu re.
There is also a large body of data to support the notion that ambient and indoor conce ntrations of air pollution adversely affect asthma .The acute effec t of 01.one exposure on level of lunµ function has convincingly been shown to be due to a reduction in the ability to take a deep breath (lhat is, a reduction in inspiratory capac ity) rather than being an effe ct llll airways (h.1).It is not su rprising.then.that the acu te cllecl llf o;nne nn lun" fun ction has nnt been shown to allc t asthmatics more adverse ly than nonaslhmalic subjects in exposure chamber studies (64).However.e pilkmi ological studies have shown that subjects with asthma arc adversel y affected by o:1one ex posure, in term s of both asthma exacerbations and hospital izations (65.66).Another aspect of wone e ffects has recently been sugges ted hy work show ing that response to spec ific allergen exposurl• in se nsiti 7ecl as thm atil•s is aggravated hy preceding oz.one exposure ( f,7).
The adverse effects of inhalable parliL•ks on asthm:llics arc caus ing great conce rn.A recent study rron1 Sealllc found ;111 association between daily changes in inha lahle particle concentration and cmeroenc y room visits for asth ma at particle concentrations that are observab le in any urban area in the United States or Canada (68).It appears that the adverse e ffec ts associated with inhalable pa11iclcs arc large ly limited to p;1rticles gene rated by combustion (69).There has also been recent interest in the adverse effects associated wit h acid aerosols.A recent study of a pan<'i of asthmatics in Denver found that the assoc ial ion he twee n ambiL•nl air pol -I utan ts and asthma L'xaenhalions was strongest I'm acid aero-• sols (70).
Ci1vcn that the majlirity of our time is spent indoor.,, it i, li kely that indoor concentrations of ai r pollutants can also adversely affect asthmatics.Exposure of asthmatic children to environmental tobacco smoke has been shown to res ult in reduced levels of lung function and increased bronchial hyperresponsiveness (7 l ).Recent ly.expos ure to environmental tobacco smoke in children as assessed by urinary cotinine concentr::itions has been shown to be associated with exacerbations of asthma (72).
It is likely that any exacerbation in an asthmatic is caused by several fact ors acting togethL'r.Env ironmental factors such as indoor and outdoor acroallergens and indoor and outdoor air contaminants are among a group of factors that can help to trig"cr exacerbations of asthma.

VANCOUVER NFA STUDY
Dr Mark Turner outli ned the prel iminary data from a prospecti ve evaluation of 99 patients admitted with acute asthma over a 20-month period to tlK Vancou ver General Hospital and St Paul's Hospital (73), Vancouver.Subjects gave a detailed history of their asthma , recent exposures to possi ble triggers, treatme nt undertaken and compliance.Measurements of airflow obstruction were recorded, and while in hosp ital the pat ients had skin prick testing and blood draw n for serum immunoglobulin E. Following discharge a home visit was made, samples of bedroom and mattress dust were collected for house dust mite analysis, and air samples were collected for analysis of moulds.Subjects were defined as cases if they had respiratory fa ilure (PC02 greater than 45 mmHg) or mechanical vent ilation (controls were all other asthma patients admitted to hospital during the same peri od) .Subjec ts were between the ages of l 5 and 55 years .
There were 20 cases and 79 controls ident ified.There were no deaths .The cases were more li kel y to have been ventilated previously (P=0.000I ).There was no difference in the prescript ion and compliance with inhaled co11icosteroids.There was also no difference in the numbers ingesting prednisone before admiss ion.Cases also did not have higher bedroom concentrations of dust mite or cat allergen than ast hma patients who we re controls (74 ).
Thi s study again confirms.hut for the first time rn a prospective manner.that a history of previous mechani ca l ventilation for asthma is a significant ri sk factor for NFA .The use of inhaled medications was not identified as a risk factor.The data also suggest that asthmatics with NFA arc not exposed to higher conccnrrations of home all ergens than are asthmatics hospitalized without NFA .Dr Turner noted that the sample size was sma ll and a largL•r prospective study would be rc4uirecl to categorize better ri sk factors and potential interventions to prevent hospitalizations and episodL's of NFA.

LESSONS FROM THE NEW ZEALAND CASE
CONTROL STUDY OF NFA Dr John Kolbe outlined the rationale fo r an NFA study from New Zea land: hL• agreed that, although NFA is an important entity in its own right , it may also be regarded as a proxy for asthma death.Data collected in the contex t of NFA Can Respir J Vol 2 No 2 Summer 1995 Workshop on near fatal asthma have cons iderable advantages over those collected in relat ion to asthma death.Obviously in the latter situation, data have to be collected from fri ends and rel atives.who are pa rticularly likel y to experience recall bias .or fro m case notes where incompleteness and inaccuracy of data are a concern .Obta ining data di rec tly fro m both cases and controls is espec ial ly val uable in tem1s of deta ils of events in the 24 h before the index attack, identification of psychosocial fac tors, assessment of the quali ty of medical care (both acute and long te1m) and assessment of practical self-management knowledge.Campbell et al (75 ) have shown considerable di screpancy between the in formation prov ided by persons sustai ni ng an NFA attack and their close acquaintance• even in terms of med icat ion use.limitation of ac tiv ities and simple conventi ona l measures of asthma morbidity: there were also discrep-,111cies between the prescribed medicat ions as listed by the patient and those recorded in tlw hospital record at the tinw of an acute admission for asthma (unpubli shed data) .
Direct inte rvie w also allows for the collec tion of comp lete detai led information using instruments specifica lly addressin g clearly defined variab les of in terest.In asthma , psychosoc ial factors may operate via effec ts on symptom reporting, compliance, self-manage ment behaviou r. health-seeking behaviour and me thods of use of the heal th system, response to educational and other interventions.and may affect quality of medical care.
Dr Kolbe and colleagues are attempt ing to in vest igate the complex relationships among psychological factors, soc ioeconom ic factors.quali ty of med ical care, disease seve rity and asthma death and near death in a case control study: cases are those admi tted to an intensive care un it with NFA. and controls are patients admitted to genera l wards wit h acute asthma.Nonnative data for the instrumen ts are also being obtained from a random sample of adult community-based asthm atics.Preliminary data from the initial 124 subjects, not ,tnalyzed in a case control fom1at, sho w that those admitted to hospi tal in Auck land with acute asthma have the foll owing: • severe asthma on admiss ion (pH 7. 32±0.l 7. PaC02 54.1 ±39.1 mmHg, n=75) but short hosp ita l stay (J.75±2.75days): • acute preci pitous asthma in only a few -only 7% stated that they had an attack lasting less than 6 h: • indicators of good ongoing medical care -96% had a regular family doctor.80% had been presc ribed inhaled corticosteroids, 87% owned a peak flow meter, 78 % stated th::it the family doc tor measured their peak !low 01 asked about their readings, 49% had their metered dose inhaler technique chec ked in the prev ious 12 months.52% had a written •action plan• and 98% of those who had tried (n=62) had experienced no difficulty obta ining an urgent appoi ntment with their family doctor: • hi gh morbidity from as thma -33% and 73 % had a previous intensive care unit and hosp ital adm ission.respecti ve ly, fo r acute asthma, 40% had a hospital admission in the previous year.and 60% indicatL'd that they had moderate to severe interference with sleep and/or exercise intolerance: • poor !L'vels of practical asthma .,elf-managementas assessed by responses to scenarios describing the rapid and slow onset of scYl'rL~ lik -threateni11g attacks of asthma (76 ): • severe economic disad vantages -48 % had experienced financial di tlicultics in the previous year, 4:F/r.were in paid employment hut 17'/r had looked for work for more than one month: for 30'4-the house hold's only income was a social security bL~ne fit (compan:d with 12'k for New Zealand as a whok) anJ 14% indicateJ that concerns about costs had a Jetrimental effect on the management of the index attack (primary health care is suhsidi1cd, but not frc.c, in New Zealand, and in the past few years chargL'S fllr atlL'nJing publiL• hospitals have been introduced): • high levels of clinically significant anxiety (52%) bnt lower levels of clinically significant dqm' ssion ( 15%): • high levels of ge nnal social support, but in 43% this is deficient in astluna-specifiL• terms anJ in 22cyo there had been moderate lll severe conflict between the patient and their primary social support person; • a high number (.l 1±2.4) of significant life events in the previous 12 months.
Following rcrngnition of the •epidemic• or asthma deaths in New Zealand in the late I ()70s.a number of eJucational and management initiati ves were introduced (77).Subsequently there was a progress ive anJ swaaineJ decline in asthma mortality during th1.: 1980s and 1.:arly 1990s and more recently a decline in various morbidity indi1.:cs(77).Currently.dcspire achieving high lewis of indicators that would be generally accepted as imfa•ating good quality of ongoing asthma management (78,79), patient., admit ted to hospital with acnte asthma had considerable chronic asthma morbidity.The data strongly suggested that psychological.social and L'conomic factors may be playing major roks.
In New ZL•alanJ further r1.:ductions in asthma morbidity and mortality arc unlikely to result l'rom additional Cllnwntional educational and management interventions.lniti:1liVL'.S such as community-based asthma clinics staffed by ethnic health care workers have been prospectively cvaluateJ (80).However, the health rroblems of these asthmatic patie nts wil 1 not he comprel11::•n.,ivclyaddressed unless the soci al. economic and psychological factors that limit the affordability , availability and accessahility lll.health care are also ad-Jressed.To not rccogniz.c or understand these influences, along with certain cultural issues, means that the health care services provided will not be effectively used.

PRAIRIE PROVINCES FATAL ASTHMA STUDY
Ms S,mmne Tough outlined the design and implementation of the epidemiological component of the Prairi e Prov-incL'S Fatal Asthma stuJy (PPAS).The presentation focused on the design, logistics and administration uf the project and 120 the challenges associateJ with est ablishing and running this multicentred trial.The study initially was pilot-tested for Oill' year in Albe rta to determine fea sibility.
The objectives of the PPAS arc to Jeterm inc risk factors associated with mortality among as thmatics and to determine till' risk of a fatal outcome among those experiencing acute asthmatic attacks.The case control study has two epidemiological control groups and a pathology control group.Cases are ide ntified as those J)L'rsons aged l'iw to 50 years dying from asthma in Saskatchewan, Manitoba ;1nd Al berta.Suspected asthma fat al ities are reported by medical investigators and rural examiners, pathologist and emergency department physicians in each province through a 24 h 1-800 number.In the event that a case is not reported, the Division of' Vital Statistics in each province notifies the PPAS collrdinator or an asthma case on receipt of thl' death cert i ficatc.
Within a month of the fatality the family is invit ed to participate by complet ing a questionnaire and allowing access to pathology specimens l'or study purposes.The gri eving process leads to variability in questionnaire return, but the study coordinator keeps in contact with next-of-ki n and may offer to do a personal or telephone interview.Controls identified duri ng the same period include subjects who have attL•mlcd an eme rgency de partment with acute asthma :,swell as unmatched community controls identified through r:11100111 digit dialling.
Dr John Butt outli ned preliminary pathology results from the Prairie provinces asthma mortality study group (81 ).In this study the lungs of 18 adults (six asthmatics dying from :lL'LLtc asthma.six ast hmatics dying or other causes and six deaths not related to respiratory disease) were removed at au topsy and fixed by bronchial and vascular instillation of 2. 5% phosphate-bu ffe red glutaralde hyJc at 20 cm H20 pressure.Nine samples were taken at L'qually spaced intervals along the main anterior and posterior basal segme ntal Immchi of the left low.:rlobe.Paraffin sections were L'lastic trichrome stained.Projected areas of the bronchial lumen (Alp) anJ smooth muscle (Amp) were determined by point counting.and the length of the base ment membrane by inter .. section counting.The ex panded cross-sec tional area (Acx) lll' the bronchial lumen i11 each section was calculated.Normalized lo the Alp, Amp in the a~thma death group showed a Jramatic peak in the middle third of both segmental bronchi that was significantly diffe rent from nonasthrnatic control bronchi.The peak was lacking in the nonfatal asthma control grour and in nonasthmatics.When normali1cd to Aex. however, the peak in /\mp was much less pronounced.The results suggest that while smooth muscular hyperpl as ia does exist i11 severe asthma, its magnitud1.: and pattern or di stribution is profound ly influenced by the way in which the data arc viewed.

NFA AND ASTHMA EDUCATION
Dr Loui s-Philippe Boulet revieweJ the role of ast hma education ge nerally.Subjects with a previous NFA episode arc considered at risk of severe.asthma eve nt.including asthma death , as described in studies looking al factors in volvcd in an increased risk of fatal or Nl•'A.Many of these fac tors can be addresseJ by asthma educ11ion and coun se lling; they include sensitizing environmen tal e xposurt'.poor L'Ompliance and lack of• unde rsta nding o f the treatmenL poor asthma conlrol including managemenl or a.,lhma narc-ups.undL'l°l'SI imal ion of asl lrn1a sevnit y. overrel iarJL•c on hronchodi lalors.undcruse or L 'orliL•ostcroids.hlunted perceplion or :1s1hma symploms and psyc hosocial problems.
Ast hma education prog rams arc effective in reducing asthma morbidity in asthmatic patie nts.including those recruited al lhc emergency room following acute ast hma (X2-85) .There is , howe ve r. a need for studies looking al 1hc influence of a sthma education on specific interven tions lo prevent NFA .particularly in pa1ic11t s who previously e xperienced this severe form or asthma.Until such studies become available.it seems mandalory 10 provide to p<tticnls wilh previous life-1hreatening asthma the basic information and training that address the many deficiencies assoc iated wilh NFA .Self-management skills should be improved, reg ular follo w-up ensured and available resources offned.Particular attention should be paid to adolescents anJ people from low socioeconomic classes because they seem more prone lo NFA (86-88).
For patients presenling wilh an antic asthma l'pisode.education ideally begins in the emergency room.or in hospital if the patient is hospitali1.ed;lhe interventions maJe by the different heallh profess ionals should be clearly articulated to provide uni form basic information as well as ongoing reinforcement of th e educational process.The objectives of the educational intervention arc thrccfolJ: cognitive (knO\vledgc) , psychomotor (skills) and psycho-affective (attitudes and behav iour) .Teaching idea lly should be tailored to each individual and e mphasi s placed on practical elements that will help to im pro ve behaviour (83).II is importanl that asthma education programs be carried oul in parallel with lhc use of appropriate anti-inflammatory 1herapy (89), as we ll as allergen avoidance and the use of specific 1herapy before unavoidable allergen exposure.The appropriate use of antiinflammatory therapy is especially im portanl (90) as is provision for ready direct access to health care in lhc event of an aL•ute li fe-threatening allack of asthma (43).

Essentials of an education program:
Basic elernenls have bee n suggested by differe nt ed ucation prog rams.and although lheir usefulness in lhe preve nlion of NFA should be validated.they probably can improve aslhma conlrol and prevent severe as thma episodes in many of 1ltcse subjecls (9).The followin g summari zes 1hc mosl frequently ci ted principles and content of a sthma education programs.
• The goals of the treatment should be clearly undersloml.
• Avoidance or early 1n:: a1men1 of lhc cffec1s of triggers, mainly sensitizing agents such as ,tllergens or indu.\trialsu bs tances.or subslances 10 which lhc pa1ic111 is conside red allergic or in1olcra11L such as foods and sulphilcs, as well as smoking cessation, are mandatory .These lasl measures arc .however.an1111112 the most difficult to implcmcnl.
Can Respir J Vol 2 No 2 Summer 1995

Workshop on near fatal asthma
• The two main features of asthma -bronc hoco11stric1io11 a nJ inflammation -shoulJ be explained in simple lcrms and related to the treatment (bronchodilators and anli -inflammatory agents).
• T he asthmatic patient should understand how lo evaluate aslhrna severity and the crileria of control or lack of con1rol.Peak flow meters arc useful, particularly for those paliL•nts who have poor pcn.:cplion of symplmn severi1y.
• The types, appropriate use and side effects of medic,ttions need to be explained.In palients who require regular use of hronchodilalors.we should insist on regular anti-inflarnma1ory therapy, particularl y steroids.which .,ccm lo reduce markedly the ri sks or fatal aslhrna (lJO).Inhaled 'shorl-acting• beta2-agoni sts (eg, salbutamol.tcrbutaline) arc considered !'or use \m demand' and can be used as another means lo assess asthma control.• Preventive• medical ions such as cromog lycatc nr ncdocromil can also he used hdorc potential allerge n exposure.
• Adequate tech nique of inhaler use should be laugh I and potentia l difficullies detected.
• Most important, lhe paticnl shoulJ be instruclcd 011 how and when 10 change medication according to spec ific crite ria (written action pla n) wilh or without peak expiratory f'low ra te measuremcnl.Delays in 1rcal111L'lll mod ification or medical consultation contribute to the se verity or all,tcks, anJ early intervention should he suggested (43).

Specific problems related to management of patients with NF A:
Many problems related to aslhma managcmenl have been desc ribed in the general populati on of patients with asthma but also specifically in fatal or NFA (90.91 ).In fact, 1hc problems encountered in teaching all asthmatic • arc us ually emphasized in a patient with previous NFk T hey arc ph ysiologica l (symptom perception, m a rk eJ airway hypcrresponsi vcncss ) (92), behavioural (compliance with 1rca1 .. ment.manageme nt of flare-ups) and psydrnsocioeconomic (motivation, family and financial problems , disease acceptance and soci al pressures) (93).Poor pcrceplion or recogni-1ion or aslhma symptoms can possibly be improved hy symptom recording and peak expiratory flow rate measurcmenis (94).Improveme nts in compliance wilh lreatment and follow -up appointme nls can result from a better understanding of lhc principles outlined above and from specific techniques addressi ng this proble m (95 ).Increased access to structured asthma educa1ion should he promoted.but since 'non-attendance• is frcqucnL 1his shoul d be subjccl to specific interven tio ns. to try to change patien1s• a1ti1udcs Inwards their d isease (96).Finally, psychosocial problems a rc frequent and should be recogni1l•d and rne l wilh offers of counselling and, if required.refeITal to a resource person (97).

The role of asthma educators in NFA :
The success of the L'ducation prog ram also depends on lhe quality uf lhc teaching.Many Jcficienci es have been observed in hL'alth educa-tors (98,99).They should ther fo re be offered adequate training to update their kno wledge.deve lop specific allitudes and appropriate teachi ng skills and be g iven bas ic pedagogical tools and methods ( we should ai m at early recog nit ion of this subgroup of patients.ens ure optimal treatment.offe r basic asthma education fo cused on speci fic deficil:'nc ies detec ted.and ensure close follow-up.Increased access to medical serv ices, such as those provi ded hy a self-admission service, has been suggested as a means of preventing ••evere asthma events, although th is approach is not large ly used and may pe rhaps be made up for by the increased availahi lity of ed ucators and physician consultation at earlie r stages o f worseni ng asthma ( 102 ).Further studies should look at means to improve preventive measures (smoking.allergies.etc).compliance and attendance for foll ow-up .

PATHOPHYSIOLOGICAL ASPECTS OF NFA
Dr Peter Mack km introduced and chaired the workshop proceedi ngs.which dealt with the pathophys iological aspects of NFA .
The re is consensus that a major pathophysio logical abnormality in asthma is excessive airway narrowi ng.Normal subjects are protec ted from thi s by a plateau on the closeresponse c urve so that increasing doses of a smooth muscle agonist fail to produce increasing responses ( I 06-l 08).Because a plateau is present.regardless of the means o f measuring the response (forced expiratory volume in l s I FEY I I I 1071.maxim um and partial expiratory flow vo lume curves I lOS.109 1. ai rway resistance I I IOI).an I regardless of the agonist or combi nations of' agonists used ( l 07.I 08.l l I). it is usm11l y take n to indicate supramaximal stimul ation .T hus, the concept has arisen that suscl'ptihility to fata l or NFA ari ses from loss of the mechan ism(s) that normally limit the degree of airway narrowing and that may protect less severe asthmatics from near fatal e pisodes.
Assessment of hyperreactivity: Drs W Gi bbons and A Sharma po inted out that the us ual met hod of assess ing bronchial hyperreact ivity , the PC20 or PD20, co uld not detect excess ive airway narrow ing because the independe nt variable was the response (a 20% fall in FEY 1) whi le the dependent variable was the dose that produced this response.For a test to detect excess ive airway narrowing.the response would have to be the dependent variable.T hey proposed an altemat ive test, namel y the fa ll in forced vital capac ity (FYC) that occurred at the do e of agon i •t prod ucing a 20% fa ll in FEY t.T he rationa le for such a tes t is that the fall in FYC retlects an increase in residua l vol ume due. to airway closure.T hey s how ed in asthma ti cs tha t the FYC decreased in a dosedependent manner and was no t corre lated wi th the fal l in FEY 1/FYC, ie, some asthmatics had a decrease in thL' f'rnmer with little change in the latter.and vice versa.There was no correlation bet ween PC20 and the fall in FYC at the PC20.Thus the two tests measure d ifferent parameters.To the extent that FYC decreases due to airway closure (wh ich re presents the limit to excess iv airway narrowing) and that a rapid ly decreasing FYC wit h increasing dose represents a rapidl y increas ing degree of airway c losure. the test mi ght prove use fu l in detecting asth matics at ri sk for excess ive airway narrowing and near fatal attacks .
In normal subjects given hi gh dose (2 56 mg/ml ) methacholine challenges.G ibbons and Sharma measured inspiratory pulmonary resistance (RL) at different static transpu lmonary pressures (PL) as the response.T hey showed that the maxi mum increase in RL. at a PL of 10 cmH20 was almos t negligi ble.whereas at a PL of 3 cm H20 the plateau was abo lished, confim1ing the fi nd ings of Ding ct al ( 110) in humans.T he fact that asthmatics with near fatal attacks are hype rin tlated ind icates that the mechanism that protects normal subjec ts from any significant degree of airway narrowing at a PL of IO cm H20 i aboli heel; ast hmatics breathing near to tal lu ng capac ity behave li ke nonnal subjects breathing at a PL of 3 cm H20.Effect of lung elastic r ecoil : Drs Irvi n, Liu and Brugman pointed out that a mechan ism that would allow for excessive airway narrow ing in asthmatics is loss o f lung e lastic recoil pressure (Pel).They had no ticed a num ber of stero id--dependent.asthmatic children who , follow ing aggress ive treatmen t and normal izat ion of ai rflo w. were still persi sten tly hypcrinfl ated.These patients were also pe rce ived to be at high ri sk .Therefore, they pe rformed a study to characte ri ze these patients better and to address the postulate that patients with reduced e lastic recoil are more prone to NFA e pisodes.
Pe l was measured in the stud y group and compared with that in age-matched, severely asthmatic children without hype rin flat ion.Pe l was assessed by static transpulmonary pressu re-volume curves of the lungs .Press ure-volume c urv es were subjected to curve-fitting analysis with a single exponentia l function ( l 12).Asthmatic ch ildren with hyperinrtation were fo und to have sign ificantly decreased Pe l.Of note, many prev ious ly identifi ed risk fac tors for NFA e pisodes were not significan tly di ffe rent between the two grou ps ol st~verely as thmatic children.The most irnpurtant finding of' this .,tudy w:1., that the group with reduced Pel were charac-tcri1.edby episodes of ex acerbations 01• asthma with a signiticanl incidence of loss of consciousness.respiratory failure and requirement for mechanical ventilation .
Dr Irvin and colleagues concluded that a suhsl'l or severe asthmatic children with persiste nt hyperinfl ation :ll'c dis tingui shed by pressure -volume curves e xhibiting a marked loss in recoil and an increase in complian ce.This group is also characteri 1.ctl by more ditlicult-to-cuntrol asthma and hy a greater inc idenL'l' of respiratory failure.Thus.airw:1ys hyperreac tivity coupkd with reduced lung ela stic ity ma y represent a situation or ri sk of near fatal attacks .
Airway smooth muscle contraction: Dr Mad.lcrn anal y1ed the load on airway smooth muscle when airways narrow durin g hnmehocon strictiun .The re are three com ponents to thi s load: lhc elast ancc of till' airway wall: Pel; and airway parenchymal interdepe nde nce.All three of lhese dccrease as lung volume dec reases.He showed that at Pel of 20 :111d 10 cm H20 (whe re Ors Gibbon s and Shanna fo und littl e increase in RL with high dose rnl'lhacholine challe nge) the load was large and little narrowing would be predi cted , based on kno wn maxi mum active force-length c haracteristics or smooth muscle.At Pel or 5 and 2 cmH20 it was more than an order or magnitude less than at Pel of 20 cm H20.It is know n that the pl ateau is :1holishcd in normal subjects when Pel decreases from 5 to 2 cmt•l20 ( 110).whereas in as th matiL• subjects it is abolished al hi ghn valucs of Pe l.Peri bronchial in flammation might account !'or !his because.first.it wo uld dimini sh pe ribronchi al press ure .effectivel y lowering local elasti c recoil pressure: seccmd.lhe rcductiun in pe ribronchial prcssurc would allow lhe airway to recoil inward lo a more L 'Olllplianl pan of its arca-lransmural pressure relationship: and third.it would decrease air wa y-parc nc hyn1;tl int erdepe ndence .
He calculated the degree of pcrihrunchial i11lb111mation !hat would decrease the load by an amount equi valent to dcc reas ing Pel from 5 to 2 c111 H20.He found that an inc reasL• in airway wa ll thickness by a l'aL'lor or 3 al a Pel o f S c111H20 would dc n ease pe ribronchial pressure from -5 to -2 cmH 20 ;111d would ,nimic a dccrcase in lung volume from normal funcli onal rL•sidual capacily lo that ;1t a Pel of 2 cmH20 .Wilh this degree or peribronch ial inflamn,ation. the c hange in pc ribronchial press ure pe r unit change in Pel would be less than normal.;\t a Pel of '20 cm H20.pe ribronchial pressu re would be --11 cm H20.whereas at a Pe l or O it would he posi1i w .
Thus, pe ribnmchial inlbmmatiun woul d no t on ly decrease the load.it would diminish the slrc tch on airway smooth mu scle pro vided by a dl:cp inflation .reducing the slope of the relat ionship bdwecn pcrihronchial and pleural pressure from an assumed v:tluc 111' I 111 a value less than I.This mi ght accounl for Colchalch •s ( 113) obse rvations that RL ck11 1gcd less with Pe l in asthma lh ;m i11 eithe r chronic obstructive pulmonary di sease or normal subjects.Effect of increased lung volume: Dr Sol Pcnnull em pha-.,i Led the risk s imposed by breathing at high lung v11lumcs.Using a mechani cal analog ue of airway cl osure to produce hypcrinll ation i11 anestheti zed dogs. he and hi ., colkag uL'S found that an average inc rease of 0.84 L for 30 mins dL•creased minute ventilation hy 27'/r and increased arterial PC02 from an average o f 40 mmH g to an averagc of 73 mmHg.Thi s was acL•omplished without any airway n • .11ruwing.Continuous positive airway pressure (C PAP) restored mi nute ventilation and decreased arterial ?CO2.In humans breathin g llll a similar analogue of :1irway l'losurc to impose hyperinlbt ion he nolcd marked increases in esophageal pressure swin gs with inspiratory pressures falling to -20 lo •-.\0 cm H20 and expiratory musck reeruitmenl.He reported that breathing at 2 L above functi onal re sidu al capac ity.while ve ry distress ing, could be tolerated for prolonged lll'ri ods.I fow l:ver, greater degrees of hype rinflati on were fo und to be intolerable.He suggested that CPAP mi ght he :1 tool to determine how much respiratory dist ress is dUL' lo dynam ic hyperinflation and how mud1 i., due to increase in lhc fl owresisti ve work of brealhin l) .
Ors O'Donnell and Lougheed repo rted 0 11 the results of an experiment si milar to that suggested by Permutl .Following rnethacholine challenges in aslhmatics they assessed thl: sensation of effort caused by dynamic hyperinfl ation by applying CPAP, and compared it with thc se nsalions engendl'rl'd by airway narrow ing, which they relieved by inspi ratory pressure support.They fo und thal CPA P relieved the sense of effort more than inspiratory pressure support.
Dr P Sliwinski also reported a study in which dynamic hype rinflation was imposed on norma l su bjec ts by ex piring through a Sta rl ing Resistor.wh ich limited expiratory fl ow to 0. 3 L/s while rebrealhing carbon diox ide.T idal volume tkcreasecl in response to the combined e ffect of carbon dioxide and L 'Xpiratory t1ow limitation while res piratory freq ue nc y was increased at any given leve l of end -tidal PC02 comparl'd wi th the response in the absence of expiratory t1ow li mitation.This inc rease in res piratory frequency agorav ated the dynami c hypcrinlhtion and drove lhe lung vo lume at which tidal breat hing look place ahow the normally allainab le tolal lung capacity.There was mass ive but ine ffec tive rcnuitment of expiralory muscles, a substan tial degree or loss or lu ng rl'eoil and an inspiratory PL swing that must have bee n close lo the maximu m that 1he ins piratory muscles could gencrall• .He suggested thal this was a reason:1blc model of vent ilatory pump functiun in a severe asthrnalic attack: that it prod uced loss of el;1 st:mce and thus mimicked the loss of rL•coi I frequentl y obse rved in aslhma: lhat.if palie nts we re forced lo nmlinue to breathe in this way.respiratory muscle fati gue and acute respiratory l'ailurL ' would surely ensue.This ex periment in normal subjects may mimic the events or a near l'a1al ;1ttack of asthma.Prediction of impending severe asthma: Ye1ina.Kenyon.Oli w nstein and Macklem addressed the issue of pred ict ing a scvne at tack of asthma.They pointed out 1hat sei smologists could predi ct 1hc likel ihood of an eart hquake of a given energy over a dd innl time pe riod in a given region by the equal ion E = 1/f' where f is the f1n1ue11cy of earthquakes of a give n energy (E) over a given time period in thl: region in qucslion and x i~ a constanl.In a rq :irn1 where earthqua kes are 12:J Cllllllllon.x is high and nm be used to predict the probability that an earthquake of a given size is likely to occur in the region in ques tion over a giwn time period.The same a pproach can be used to predict freque ncy and severity of avalanches or of stcx•k market t1uctuations.Following Yezina•s suggestion, Kenyon analy1cd diurnal variations in peak expiratory flow over a six-week period in thrcL' asthmatics.T he plot of log severity versus lug frequency was linear with a slope of --0.83 and a L'OITelation r uf 0.8 .If the slope of this rdatiunship is high in a given ast hmatic.the ch;111ccs of a severe attack are greater than in an asthmatic in whom the slope is lower.This raises the possibility that thL' n:btiunship between frequency and severity of asthmatic attacks follows similar relationships as frequ e ncy and power or e arthquakes.avalanches and s tock market fluctuatiuns, ;ind that this can be assessed by variations in simpk 1rn:asureme11ts of lung function measured over a relati vcly short period or time.Thus. it may be possi ble tu predict which asthmatics arc at risk for fatal or NFA altacks , without being able to predict when they woulJ occur.
There was general agreement that further research is necessary to predict which asthmatic is at risk to develop a near fatal allack.

TABLE 2 Factors complicating appropriate management of asthma
( 105 ) O I).They should be able to offer the patie nt pertinent and speci fic data.adapted to hi s or he r specific condition.severitylevel. degree of underst,mdi ng, expectat ions about the disease.immediateneeds .fearsandprejudices.Idea lly , close relat ives should be invo lved.Patient understand ing of basic notions and appropriateness of behavimrr should be evaluated in fo llow-up visits.Traini ng programs for health educators will help t() improw un ifo rmity and quality of L'ducati onal interventions.Collaboration among general pract itioners.spccialisband other hea lth professi onal s is esse ntial to the development of an e ffective program.Early intcrvrntion.withimprovedaccess to health care, should be espL•cial ly emphasized ( l 02 ).W hat r emains to be done'?In the past decade, ,asthma management has been rl'viewed and guidelines have emphasized ast hma preve ntion .particularlyfrom structur d asthma education ( l 03, l 04 ).However.mostevaluations of asthma eclue,1tio11 ha ve involved patients witho ut previous NFA e pisodes( 105 ).These interventions should be further stud ied in thi s 'at risk ' group, to help us to identify wh ich are the most important com ponents and best methods to reduce as thmarelated mortalit y and morbidity in those patients.Until then.