Thorax 1990;45:645-648 645

for cases, hospitalisation for controls). The asthma about the principles and practice of improvement in data gathering methods does treatment for this common condition. LETTERS TO not offset the persisting principal conceptual, methodological, and execution problem of WALTER 0 SPITZER THE EDITOR the original study-that is, inadequacy in the Department ofEpidemiology and Biostatistics, classification and adjustment for asthma McGill University,

Montreal, Canada Thorax: first published as 10.1136/thx.45.8.645 on 1 August 1990. Downloaded from severity and the likely confounding which probably results. They also repeated a serious A SONIA BUIST error by dissociating the time when Department ofMedicine, severity Division ofPulmonary Medicine, was measured from the time when exposure Oregon Health Sciences University, was classified. In this regard we do not agree Portland, Oregon, USA that medications noted at admission or dis- 1 Crane J, Pearce N, Flatt A, et al. Prescribed charge are a valid proxy for "chronic drug fenoterol and death from asthma in New usage." Zealand, 1981-1983: case-control study. Case-control study of prescribed The investigators have not addressed an Lancet 1989;i:917-22. fenoterol and death from asthma in New important alternative explanation for their 2 Buist AS, Burney PGJ, Feinstein AR, et al. Fenoterol and fatal asthma [letter]. Lancet Zealand, 1977-81 findings-namely, that sicker asthmatic 1989;i:1071. patients tend to be switched from other 3 National Heart Lung and Blood Institute. Data Earlier this year we and others commented on medications to fenoterol and that sicker asth- fact sheet: asthma statistics. Bethesda, a case-control study by Crane and others'2 Maryland: NHLBI Education Programs matic patients are more likely to die than Information Center. suggesting that fenoterol was associated with those with less severe disease. Fenoterol was risk ofdeath in patients with severe asthma in marketed in New Zealand as a medication to New Zealand. We did not accept the validity be tried when control of the patient is other- AUTHORS' REPLY The letter by Professors ofthe first report because of several problems wise difficult. Spitzer and Buist consists almost entirely ofa in methods and analysis. The most serious The contrasting relative risks for non- repetition of the criticisms, made by a larger problems were: (1) ambiguity about the Europeans and Europeans, men and women, group ofBoehringer Ingelheim reviewers,' of underlying clinical question; (2) poor stan- and those under age 20 and over age 20 our first New Zealand case-control study. dardisation of data gathering from cases and deserve comment. An appropriate analogy These criticisms have already been answered controls; and (3) inappropriate classifications might be the dilemma which faces inves- at length.2 It is surprising that the hypothesis of severity of asthma leading to inadequate tigators of aviation accidents. Does one that our findings are due to fenoterol being adjustment for severity as a confounder. We impute the accident to the aircraft and its prescribed to those with more severe asthma now comment on a second case-control study manufacturer or was it operator (pilot) error? has been raised once again, and even des- by the same group (March 1990;45:170-5). The data of this case-control study (and the cribed as "equally tenable." No substantive The second case-control study explores a earlier one) do not allow us to set aside a evidence has been presented in support ofthe possible relation between asthma medications second important competing hypothesis: hypothesis. The evidence in fact is almost and risk of death in asthmatic patients aged fenoterol (or any inhaled bronchodilator) is entirely against it. But as it has been raised 5-45 years in New Zealand during the period risky when prescribed in the context of a again it is necessary for us to re-examine the 1977-81. The cases were 58 asthma deaths substandard quality of care or when there is available evidence. (ICD 493) ofpatients who had been admitted poor adherence to an appropriate therapeutic We reviewed advertisements in clinical to hospital for asthma within 12 months of regimen. This may be the situation, for journals at the time of fenoterol's intro- death. For each case four asthmatic controls example, in the United States, where blacks duction into New Zealand, and can find no discharged the same year were selected at have a death rate for asthma three to four evidence that it was "marketed in New random from the same hospital and matched times that of whites3 and asthma mortality in Zealand as a medication to be tried when

on age. The key findings in this second study the age group 5-34 is now primarily a control of the patient is difficult." Professors http://thorax.bmj.com/ were that the odds ratio (relative risk) for problem ofinner city ethnic minority popula- Spitzer and Buist provide no reference for asthma death in all patients prescribed tions. We now believe that in the United this claim, and it is unlikely that fenoterol inhaled fenoterol was 1-99; for patients pres- States this is primarily a reflection of poorer could have gained a 30% market share if it cribed three or more categories of asthma access to health care ofhigh quality and poor had been targeted at such a small and specific drugs 2-98; for patients with a previous adherence to therapeutic regimens. group. We can also find no substantive admission for asthma in the past 12 months The fact that two other well established evidence that fenoterol was selectively 3-91; and for patients prescribed oral corti- pharmaceuticals, oral corticosteroids and prescribed to more severe asthmatic patients costeroids at the time of admission 5-83. In a theophylline, had increased risks was not (within the population of recently hospital- group ofpatients with the most severe asthma adequately discussed by the authors. As they ised asthmatics on which our studies are (defined by a previous admission for asthma point out, both of these categories of drugs based). Most importantly, the increase in the on October 2, 2021 by guest. Protected copyright. during the past 12 months and prescribed oral were prescribed at discharge to virtually all relative risk for fenoterol when our analyses corticosteroids at time of admission) the those with severe asthma, and the confidence are restricted to those with the most severe relative risk of death for those prescribed limits for the relative risk estimates were asthma effectively refutes the confounding by inhaled fenoterol was 9-82. Of particular therefore very wide. Why they choose to severity hypothesis. This point has already note, the authors reported an odds ratio of5-2 attribute the increased risks for theophyllines been made by one group of epidemiologists among non-Europeans, compared with 1-2 and oral corticosteroids to chance while who were commissioned by Boehringer (NS) for caucasians only. Also of note, the choosing to accept the increased relative risk Ingelheim to review our first study, and risk was greater in men (2.77) than women for fenoterol as a reflection of its toxicity is who reached different conclusions from (1-53) and higher in persons under 20 (4 0) not at all clear. Professors Spitzer and Buist.3 than those over 20 (1-3). The investigators We conclude that, for the second study as Professors Spitzer and Buist have. also concluded that "These findings add further well as for the first, problems in the design of suggested that "sicker patients tend to be support to the hypothesis that inhaled the study and the way in which the data were switched from other medications to feno- fenoterol increases the risk of death in analysed make it impossible for us to agree terol", but provide no data to support this patients with severe asthma." with the authors' narrow interpretation poin- claim. This actually indicates that a hospital In the first case-control study the inves- ting to only one primary explanation of the admission for asthma is a good marker of tigators confused the question of whether findings. As with the previous study, we asthma which is perceived to be severe fenoterol has an acute toxic effect when used believe that the results are consistent with enough to require changes in medication. during an acute attack with the separate several hypotheses which are equally tenable. More importantly, it means that the con- question of whether long term chronic use The investigators have in our opinion failed founding by severity hypothesis can be tested increases the risk of death. This confusion to give due weight to these alternative by examining changes in medication resulting occurred in part because drug information for hypotheses in their discussion and con- from such an event. cases came from general practitioners but for clusions. They have nevertheless stimulated We have tested this hypothesis with the the controls from hospital medical records. In and challenged the scientific community to data for the controls in our most recent case- the second study the investigators focused on take a closer look at the disturbing possibility control study.4 Each of the controls had two the question of chronic use of fenoterol by that good drugs when poorly used may be hospital admissions for asthma over a 12 collecting data on drug exposure for cases and potentially harmful. This underscores the month period, and we have examined the 420 controls from records pertaining to the hos- urgent need for better education of both admissions in the 210 controls for which all pital admission before an index event (death health professionals and individuals with the relevant data were available. There were 646 Letters to the Editor

24 changes from another beta agonist to of evidence is now in favour of a causal in the doses recommended above cocaine and fenoterol as a result of the admission, and 46 association between fenoterol use and asthma lignocaine appear equally effective. changes in the other direction (most of the mortality. As a result, the New Zealand CHARLES TEALE latter patients were switched to salbutamol). Minister of Health has moved to severely MF MUERS SB PEARSON On the other hand, there were substantial restrict the availability of fenoterol by Pulmonary Function Laboratory changes for other classes of asthma drugs. removing it from the Drug Tariff, and a Killingbeck Hospital, Leeds LS14 6UQ In particular, as a result of the admission, similar policy has now been adopted in Thorax: first published as 10.1136/thx.45.8.645 on 1 August 1990. Downloaded from the proportion of patients prescribed oral Australia. 1 Teale C, Gomes PJ, Muers MF, Pearson SB. Bronchoscopic local anaesthesia: intratracheal corticosteroids increased from 28% to 62%. NEIL PEARCE cocaine and lignocaine appear equally effec- Thus patients were often prescribed prophy- JULIAN CRANE CARL BURGESS tive [abstract]. Thorax 1990;45:333P. lactic medication as a result of their severe RICHARD BEASLEY 2 Reynolds JEF, ed. Martindale. The extra phar- is no evidence that sicker macopoia. 29th ed. London: Pharmaceutical attack, but there Department ofMedicine, Press,1989: 1213-5. patients were switched to fenoterol. School ofMedicine, 3 Kinnear WJM, Reynolds L, Gaskin D, Mac- The comment about substandard care in Wellington, New Zealand farlane JT. Comparison of transcricoid and the United States among blacks is not bronchoscopic routes for administration of relevant to our findings in the Maori. It is local anaesthesia before fibreoptic broncho- 1 Buist AS, Burney PGJ, Feinstein AR, et al. scopy [abstract]. Thorax 1988;34:805P. well known that Maori have a higher asthma Fenoterol and fatal asthma [letter]. Lancet death rate than non-Maori. We have shown 1989;i: 1071. something quite different: that the Maori 2 Pearce NE, Crane J, Burgess C, Beasley R, who use fenoterol have a higher death rate Jackson R. Fenoterol and asthma mortality. AUTHOR'S REPLY We are grateful to Dr Teale than the Maori who do not use fenoterol. The Lancet 1989;i:1 196-7. and colleagues for drawing our attention to implication that this finding is due to 3 Sackett DL, Shannon HS, Browman GW. their study of local anaesthesia for broncho- Fenoterol and fatal asthma [letter]. Lancet scopy, which we had overlooked. The main confounding by ethnicity is nonsense as the 1990;i:46. comparison was made within the one ethnic 4 Pearce NE, Grainger J, Atkinson M, et al. Case- point that we were making was that patient group. control study of prescribed fenoterol and tolerance is dependent on the effectiveness of More generally, are Professors Spitzer and death from asthma in New Zealand, 1977-81. local anaesthesia and two studies now attest to Buist suggesting that the standard ofmedical Thorax 1990;45:170-5. the superiority of the transcricoid route.'2 5 Crane J, Burgess C, Beasley R. Cardiovascular Our impression of the greater effectiveness of care for New Zealand declined so rapidly in and hypokalaemic effects of inhaled salbuta- 1976 (when fenoterol was introduced) that mol, fenoterol and isoprenaline. Thorax cocaine may be the result of our use of a this accounted for a doubling of the mortality 1989;44: 136-40. higher dose (200-300 mg) than that used by in two years? If so, what evidence do they 6 Bearshaw J, MacLean L, Chan-Yeung M. Com- Dr Teale and colleagues in their study (100 have for this? Do they also imply that the parison of the bronchodilator and cardiac mg); it is interesting that a trend towards a effects of hydroxyphenylorciprenline and superiority of cocaine is apparent in their standard of medical care also declined orciprenaline. Chest 1974;65:507-1 1. suddenly in the six countries which had 7 Elwood JM. Prescribed fenoterol and deaths abstract. The question of the safe maximum mortality epidemics in the 1960s when from asthma in New Zealand-second report. dose of topical anaesthetic agents is con- isoprenaline forte was introduced? Further- Wellington: New Zealand Department of troversial and surveys of practice in the UK more, if any bronchodilator can be harmful Health, 1989. suggest the use of more lignocaine at when prescribed in the context of poor bronchoscopy than many authorities recom- medical care, why did New Zealand not see an mend. That this is safe practice is suggested epidemic when salbutamol was introduced? by a study demonstrating serum concentra- that tions well below the toxic range with topical Professors Spitzer and Buist suggest doses in presumably we have challenged the scientific community excess of 500 mg,' when because only a proportion ofthe administered with the possibility that good drugs dose is actually absorbed through mucus used poorly may be potentially harmful. We Local anaesthesia for fibreoptic bron- membranes. The recommended doses quoted http://thorax.bmj.com/ have not. The scientific community already choscopy by Dr Teale are unrealistically low and the knows this. Rather we have suggested that a Dr AC Davidson and colleagues (March British National Formulary recommends a poorly selective beta, agonist, which is more 1990;45:239) were impressed with the local maximum topical dose of 3 mg/kg for potent than salbutamol' but available by anaesthesia produced by a transcricoid injec- the of cocaine. We now aim to measure serum metered dose inhaler at twice dose tion of 4-6 ml 5% cocaine (200-300 mg). cocaine levels to ensure that the doses we salbutamol, may have been responsible for an They went on to state that they were unaware employ do not produce toxicity. An internal epidemic of asthma deaths in young people of a formal comparison of cocaine and audit of complications in nearly 1000 bron- with severe asthma in New Zealand. lignocaine as local anaesthetics during choscopies has, however, shown no evidence Rather than repeatedly raising the same bronchoscopy but recommended that other of cocaine induced neurological or cardiac on October 2, 2021 by guest. Protected copyright. criticisms ofour work, Professors Spitzer and centres consider changing to the transcricoid toxicity. Buist, or Boehringer Ingelheim, should instillation of cocaine for fibreoptic bron- AC DAVIDSON inform the scientific community of the foll- choscopy. R LEACH owing: (1) Why was fenoterol marketed as a In a double blind, randomised study of 60 NT BATEMAN "forte preparation" (200 pg/puff compared patients we recently compared the local St Thomas's Hospital, London SEI 7EH with salbutamol at 100 ug/puff) when it was anaesthetic effects of intratracheal injections known to be more potent and to have greater of lignocaine (4 ml of 4%: 160 mg) with 1 Kinnear WJM, Reynolds L, Gaskin D, Mac- cardiac effects than other commonly available cocaine (4 ml of2-50%: 100 mg)'. Local anaes- Farlene JT. Comparison of transcricoid and beta agonists? (2) Why was the nebuliser thesia was assessed by numbers of coughs, bronchoscopic routes for administration of formulation made available in New Zealand operator acceptability, and patient discom- local anaesthesia before fibreoptic bronchos- five times the concentration used in in all areas cocaine scored copy [abstract]. Thorax 1988;43:805P. (5 mg/ml) fort; only slightly 2 Hay J, Clague J, Nisar M, Earis JE. Local Canada (1 mg/ml)? (3) Why was fenoterol better than lignocaine (for example, there was anaesthesia for fibreoptic bronchoscopy [abs- never licensed in the United States? (4) Why a mean of eight coughs per procedure with tract]. Thorax 1989;44:890P. was fenoterol marketed as a highly selective cocaine compared with 11 with lignocaine), 3 Gove RI, Wiggins J, Stableforth DE. A study of the use of ultrasonically nebulised lignocaine beta, agonist when Boehringer Ingelheim's none of the differences achieving statistical for local anaesthesia during fibreoptic bron- own funded experiments had indicated that significance. The impression of Dr Davidson choscopy. Br J Dis Chest 1985;79:49-59. in the clinical situation it was no more selec- and colleagues of the superiority of cocaine tive than orciprenaline,6 the poorly selective may reflect their use of a dose two to three agent it was designed to replace? times higher than in our study. It has been In conclusion, we concur with the recommended that no more than 1-0-1-5 mg/ independent report commissioned by the kg cocaine should be applied to mucous BCG vaccination of schoolchildren in New Zealand Health Department that the membranes in adults, while others have sug- England and Wales evidence proposed by the Boehringer Ingel- gested a maximum dose as low as 50 mg'; the heim reviewers in favour of the confounding use of higher doses may increase the risks of One aspect of discontinuing routine BCG hypothesis is indirect, circumstantial, and toxicity. not discussed by Drs V H Springett and I considerably subjective.7 Although further We agree with Kinnear et aP that trans- Sutherland (February 1990;45:83-8) is the research would clearly be valuable we also cricoid injections of local anaesthetic are well possible increase in mortality which may agree with the New Zealand Health Depart- tolerated and produce effective local anaesth- result. ment's conclusion that the current balance esia for fibreoptic bronchoscopy. When used Using published data" from the Office