Overreliance on bronchodilators as a risk factor for life-threatening asthma

OBJECTIVE: To assess the potential impact on the risk of life-threatening asthma of current recommendations in pharmacotherapy, which emphasize the early use of steroids and the avoidance of beta-agonist overuse.

Jc 1980 ii 1987.129 patients-cas lllll etc identilics qui avaicnt subi un L;pisodc lL1sth111c l"auil nu prcsque fatal.ct 129 patients-tcmoins ctaient apparies aux cas pour I' ilge, ht durfr du risque ct la gravitc de I' a,thmc Qugc par le bcsoin d'hospitalisatinn causee par l' asthme dans les deux annces prcccdcntes ).METHODES : Dcux cliniL•iens (Jill p:1sse L ' ll revue le lrailellll' lll qui avail etc prcsnil it ccs 258 .,ujcts pendant lcs dell\ A STIIMA MORllllJITY .\NDMORTA LIT Y HA Y E BEEN JN.   creasing in a nu111hn or countries despite recent sei cn-Jilic• advances tlwt have i111proved the understanding of this chronic lung condition ( 1-3).The treat ment or ast hma has been identified as a pntcnt ial contributing fa ctor to this increasing morbidity and mortality \4-6 ).In the past.phannacological treat111enl or asthma rocused nn hrnnchodilator therapy.hut the more recent recognitinn that chrnnic .1.~thmainvolvl's a characteristic inflammatory response in the airways has changed the therapeutic management or asthma {7 -\J ).This greater emphasis on anli-i ntl ammatory therapy and decreased reliance on bronchodilators alone is reflected in the revised guidelines for the lrcalmcnl o f asthma disseminated to the medical community in written reports :is we ll as scientiriL• journals ( I 0-13 ).

PATIENTS AND METHODS
Overall study design: As previously repmlcd ( 14 ).a nestcc. .l case-control design was used to cxami11L' the relationship between the use of bela-agonists :rnd lhL' risk of' death and nc:ll' de:1th lrnm asthma.Briclly.linked rll':dth insurance databases of' the province or Saskatchewan wnc used to identif'y a cohort of' 12.301 patients who were dispensed I 0 or more prescriptions for an :1sth111a rnediL•ation hL' l ween January I. 1978 and April 30.I l)87 .These mediL•ations were not avai lable without a prescription and a di spensing record was required for thl' pharrnaL•ist tii 11l' dirl'L'tly reimbursed by the government which pays fur all prescribec. .l rncdications.Subjects had lo be between the ages of five and 5-t years during the study period (January I. ll)XO to April 30.1987) lo be consiclerec..l al risk.Prcdetcrrnined criteria were used lo identily L•ase p:1til'nls who died ol asthma during the study period H-4 patients) or, lor those who su1'\'i vcd.experienced a near-fauil asthmatic episode (X5 patients ).A ncar-l'al,il e vent was defin ec. .l as intubation.mechanical ventil ation .m PC02 greater than ..is mmHg during a hospit:ili zation f'or asthma.Up to ei ght co ntrols were selected from ,1mong the n ilmrt of 12.30 I ,md were matched to each c;i.-;c patiL'nt on the basis of age at entry into the cohort.date or entry.the lll"l'UITCllCe of' a hw;pi tali1;1[ion rrn a. sthma in lhl' prior [WO year~.region of' reside nce and rece ipt or soc i,il assistance al any time during the st udy.In addition .the controls were l"L'l)Uircd lo ha ve bee n al risk for the llllll"OlllL' al the time of the cvrnt in the case p,tticnl.rcJ'errL•d to as the index Cl in ical information was obtained for all 129 cases and their 655 matched controls from two sources -hospital medical records and physician questionnaires ( 15).First, hospitalizations with a primary discharge diagnosis of asthma during the two years preceding the index event or matching index date (not the hospitalization which might have ocrnrred at the time of the life-threatening episode) were identified from the computerized databases.Abstractors, blind to the study hypothesis and unaware of the case/control status of the subjects, visited hospital medical record departments and.using a standard form.obtained information from the patients' charts on precipitating factors, presenting respiratory symptoms and their durati on.past history.si gns of a life-threatening atlack, lung function and arterial blood gases.Second, the computerized out-patient physician datafile was used to identify physician visits by study subjects.Eligible visits were those in which a general practitioner.pediatrician, internist.allergist or respirologist had seen the stu<ly subjects in the two ye ars before the index date or event.Questionnaires were hand delivered by research personnel to these physic ians and we re subsequently picked up on completi on.Clinical information similar to that gathere<l from hospital medical records was sought from the ph ys icians.Ethical approval was obtaine<l from the arrrorriate agencies of Saskatchewan Health .'Visits were those made during the two years before the index event or date.Specialists included pulmonary medicine, allergology, pediatrics and internal medicine; t In which there was a primary or secondary diagnosis of asthma viously i<lcntified, as described ,ibove.From the controls who were matched to each case.the first control per case was selected in whom similar c linica l information (ie, hospit al rccor<l information) was or was not available.This was undertaken to ensure that cases and controls were classifie<l accor<ling to comparable information.A total of 258 subjects WLTC lherefore included in the analysis.Because controls were listed by sex with !'cmalcs appL~aring first.they were more likely to be selected as controls for the presen t analysis.

Methods for management classification:
For each case and matched control a dossier of information was assembled.For the fatal case patients this included a copy of the death certificate, as well as autopsy and coroner reports when available.If the death occurred in hospital.a copy or the hospital di -•charge summary associa ted with the event was also obtained.For the ne ar-fatal case patients.a rnpy of the hospital <lischarge summary associated with the near-fatal episode was retrieved.In addition to the inflirmation surroundi ng 'the event'.additional clinical information was obtained from the hospital medical records for each case an<l matc hed control who were hospitalized ror an asthma-related episode during the two ye,irs before the index event.as described above.Detailed ex posure information regarding mc<lication <lispenscd each month fo r the 12-month period bcl'orc the index event or matching inde x date was also obtained for each case patient and matched control from the Prescrirtion Drug D ata Base uf Saskatchewan Health.One unit or an inhaled bctaagonist was equ ivalent lo the <lispcnsing or one mctrc<l <lusc inhaler (M D I).For other medications a unit rcprcsente<l the dispensing of one prescription, usually a one-month supply.For the physician review, beta-agonists were identified by class only.an<l not by in<livi<lual generic names .
Each of the dossiers for the cases and their matched control was reviewed in<lcpen<lently by two cliniciaus experienced in the treatment of asthma .On the basis of the information available the clinicians classified the management or each subj ect as either 'incompatible• or •at least partially nnnpatible' with the current emphasis in pharmacotherary.These classifications were then collated to determine the number of cases and con1rols dassific<l as •incompatible' as judged by both reviewers, as well as the number classified as 'incompatible' as judged by al least one or the reviewers.No allempl was made to have the two clinicians judge the adequacy of pharmacotherapy in an equivalent way, and their assessments ,vcre cllTied out without knowledge of the other clinician• s judgement.Only general instructions were given as to the factors lo he considered.These were the use of high or increasing doses of hronchodilators, e specially inhaled hcta-agonists. the undcrusc of inflammatory drugs and the use of c ontraindicated drugs.especially in relation to clinical events such as hospitalitations.It was felt that greater sensitivity to inadequacies in pharmacotherapy might he obtained by keeping the clinical evaluations independent in this way.though al the price of a less adequate demonstration of the validity of such a clinical assess ment.
T o va lidate the suhjecti ve assessments made by the two clinicians and to provide a more objective measure or the qua! ity of pharmacotherapy.an asthma management score was developed based on c urn.:nt pharmacological treatment guidelines (Table I) .This was considered necessary since the clinicians could not be blinded to the case control status.ie. the occurre nce of a life-threatening event, and such knowledge may have resulted in their be ing more critical of the management in this g roup.The potential range for the management score was O lo I 0. with O reflecting pharmacolog ical treatment most compatihlc with current guidelines, and 10 indicating treatment considered least compatihle.T his management setm . .' was applied lo each of the cases and c1mtrols by another member of the research tean1.who was hlind to the case and control status of each subject, using the same information the clinicians had reviewed.The weight given to Can Respir J Vol 2 No 1 Spring 1995 Overreliance on bronchodilators in asthma the different elements of the score are somewhat arbitrary.This is necessarily so.since it is dillicult lo equate the prescription of a contraindicated medica tion .for example, a beta-blocker.which occurred in one case, with a given level or overuse of beta-agonists.However. the weights g iven for the different levels of use of heta-agonisls and anti-in fla mmatory agents are compatible with the dose-response relationships observed in the cohort of 12.JO I subjects with asthma.Analysis: The frequency of management incompatible with cu rrent pharmacotherapy.as classified by the two clinicians.was compared in the cases and controls using the Mc Ncmar l . .' .
x-stat1st1c lor matched pairs.Till' 111c;111 asthma manage ment score was calculated for cases and controls.and the paired t test was used to determine whether there was a significant difference in the mean scores.In determining the association between e xposure to fcnoterol and albutcrol and management classified as incompatible.an unmatched odds ratio and 95 % confidence interval were calculated.This was undertaken since the outcome of interest was now the management classification in cases and controls crnnhined.

RESU LTS
C haracteristics of study subjects are presented in Table 2.While the case patients and the controls were similar with respect lo age.consequent lo the way controls were selected a larger percentage of the cases were males (54'-4,) than in the cuntrnls ( 12'/, ).\Vhcn e xamined in the cohort of 12.' .10 I patients as a whole.sex was not associated with outcome ( life-threatening episode of asthma).Furthermore.in the study sample, there was no sex difference in the type ll!' quantity of pharrnacotherapy received .The present authors arc.therefore.confident that the sex difference bt'lween cases and controls Jiu not bias the results presented here.The case patients used the services or physicians and were hospitalized more frequently than the control. .The frequencies uf management classified as incumpalible \Vith the current emphasi s in phannacolherapy.as determined hy the two clinician reviewers , arc presented in Tahk .3.A significa ntly higher proportion of cases (suhjeels who had died of asthma.or experienced a near-fatal asthmatic episode ) than controls from each matched pair were classified as receiving tre atment incompatible with current guidelines hy al least one of the two clinician reviewers (P<().()()I).Fur the situation in which holh the clinicians classified the subject as recei v ing treatment incompatible with current guidelines, the di ffe rence between the cases and contrnls did not achieve s ta tistical sig nificance (P>0 .05),at least in part clue to the small number of subjects cl assified in this way.Note that the 63 cases and 26 controls el assified as havi ng incompatible treatment hy at least one clinician included the 15 cases and IO controls classified as such hy both clinicians.
The result., or the asthma management score that was applied lo each nl' the case and contrnl subjects revealed a score ranging from O lo 8 for cases.and O lo 7 for controls.
with IO being the maximum score possihle.The cases had a statisticall y significant higher mean score (3 ..5) than did the with a higher score rcl'lectivc or plrnrn1acological treatment less compatihlc with current phannacothcrapy.The corn> spondrncc hetwecn the classification or treatment hy clinician rcviL'W and that or the asthma rn,magemcnt score is pro vided in T,1hk 4. The average scores arc consiste nt with the clinicians' classification -a hi gher rne,1n score (rellect ivc of pharmacological treatment less co111p,1tihle with current guidelines) corresponding to subjects who were• class ified as incompatihle with current guidelines hy hoth clinician ,rvicwers.or hy one alone compared with suhjccts whose treatment was cla.,siried as at least partially compatible with the current emphasis in pharrn(tcotherapy. Table 5 examines the relationsh ip between the li ke lihood ora suhjcct•s treatment heing classified as incompatible with current pharmacothcrapy by both clinicians and whether they had been dispensed l'cnotcrnl rather than albuterol in the prior 12 months.The small number or subjects precludes achieving statistical significance.The trend towards a positive association het\wcn being dispen sed l'cnoteml and being classified as overreliant on bronchlldilators as opposed to the possible negative associalilln hct ween alhutcrol and appropriateness of drug therapy suggests.however.that these two inhaled bcta-agoni sts were used in patie nts in whom the quality of therapy was dif!crcnt.It may also suggest, in accordance with lhe results of Scars (5). that prescribing rcnoterul may worsen the severity of the di sease.which in turn would lead to a greater need rnr.or reliance on .b••onchodilators.

DISCUSSION
In response to increasing asthma morbidity and mortality in many parts of the world.vario us nrgani1ations have brou ght together ex perts in the treatment of asthma to develop guidelines for asthma 111anagemL:11t in the hope of favourably int'luencing prognosis in this L'OlllllHHl disorder ( I 0-13 ).An important aspect of these guide!incs is the greater emphasis placed on early use or anti-inl'lammatury therapy and less relianee on hronchodi lat or therapy.This current emphasis in phannacothcrapy is based on several long term studies recently demonstrating the efficacy or inhaled steroids in decreasing asthma morbidity ( 9. I()-I 8 ).suggestive cvidcnCL' that regular use of brnnchodilators may actually make asthma worse (5. (1.19-21) and the strong a •sociation between mcrusc llr inhaled beta-ago ni .sls ,llld li fe -threatening asthma ( 1-J.).In an attempt to est imate thL~ impact on asthma morhit.!ity that might result from this change in emphasis in the phann:1cothcrapy of asthma .we evaluated the drug therapy received by asthmatic subjects who had CXPL'rienccd a lire-threatening atlad.and compared it with subjects of the same age and asthma severity who had not experienced a lite-threatenin g ep isode.Nearly half' of the cases nr life threatrning asthma were judged by at least one clinician experienced in the treittlllL' nl or aslh111,1 to have received treatment that did not at least partially meet L'lllTL'ntly rcco111rnended the rapy , compared with 20% of study subjects who did not experience such a life-threatening event.The 1110,t com mon pattern observed was the increasi ng use of various bronchodilator.sllVLT time without any dispensing of prcvcnli vc or anti-inrlam matory therapy ( 22 ).
To rende r the evaluation of therapy more objective.we LiL'vclopcd a management score that was then applied to the treatment regime ns by an observer not ex perienced in the treatment of asthma .The score was formulated to re llcct the hypothesis that excess ive therapy with inhaled bcta-agonist bronchodilators without umum1itant use of anti-inflammatory therapy is ill -advised.Again.according to this score.cases of life-threa tening asthma dcrnonslrntcd greater diverge nce from current 1' L'co111111cndations than subjects who had not experienced such attacks.;\s one would L'XJ)L'cL since bolh the score and the clinical a.-;scssment arc based 011 current standards of therapy, the score is higher among subjects in whom clinicians thought the therapy did not meet these standards.This provides so me val idation of'thc clinical assessments.
We and others (23) have reported a stronger association between fenotcrol, as opposed to alhutcrol.use and fata l asthma.Such a di ffcrc nce mi ght be explained i r subjects who were presc ribed feno terol were at a greater baseline risk of an adverse even t.Inasm uch as not being di spensed therapy that rellects current recommendations is a marker or increased risk.subjects dispen sed fenotcrol appear to be at grea ter risk.Such channelling or therapy to different risk groups has been l'uund for fcnotcrol in another population (24).Our cvidcncL'.
for such di llcrcntial prescribing can only be cons idL'1wl as pre liminary given the low statistica l power availabl e for such an analysis.T his study was based un lhc L'xamination or computer and clinical records of inror111ation collected in the past.This li mits the amuunt or detail available on nonpharmaeological therapy as well as on environmental risk factors for life-threatening :1sthma.If such !'actors were associated with the pharmacotherapy di spen sed in a different way in patients with a life-threatening epi sode or asthma than in those without such a history.these factors could partly explain the relationships dcsnibed.

CONCLUSIONS
The rc~ults presented show that cases ex pe riencing fatal or near-fatal asthma were more likely than controls to have rece ived in adeq uate therapy as judged by l'IIITL'nt treatment guidelines (spec ifically less anti-inflammatory and more hera-agonist treatment).From thi ., evidenc e. we suggest thal less rel iance on bronchrnJilators and greate r use of antiinrlammatory therary may have a positive impact tlll ,1sthma morbidity.
The subjects were the 129 cases of fatal and nc::ir-l'atal asthma pre-

TABLE 1 Asthma management score Criteria
1 10 'Includes oral corticosteroids, inhaled corticosteroids, and inhaled cromolyn; t Includes beta-blockers.sedatives and parasympathomimetics date.A tolal of 655 controls were matched to the 129 case patients .