Tobacco Documents Online

Childhood Asthma and Passive Smoking Urinary Cotinine as a Biomarke; of Exposure-' RODNEY EHRLICH, MEYER KATTAN, JAMES GODBOLD, DEBORAH S. SALTZBERG, KATHERINE T. GRIMM, PHILIP J. LANDRIGAN, and DAVID E. LILIENFELD Introduction Despite a number of epidemiologic in- vestigations, the relation between child- hood asthma and passive smoking re- mains uncertain. Prospective studies of general populations have failed to dem- onstrate an increased incidence of diag- nosed asthma among the children of smokers, although an increase in parent- reported wheezing is apparent (1-5). Some cross-sectional studies have dem- onstrated an association between paren- tal smoking and asthma (6, 7) or wheez- ing (8-10); others have failed to find such relations (11-13). In studying a group of asthmatic children, Murray and Morri- son showed that maternal smoking in- creases the severity of the disease and bronchial hyperreactivity (14-16). An in- crease in the number of emergency room visits among asthmatic children from smoking households has been demon- strated in one study (17). The question of whether passive smoking triggers acute attacks of asthma has not yet been spe- cifically addressed. Exposure misclassification may be one reason for the inconsistency among epidemiologic studies. The studies con- ducted thus far have relied on question- naire measurement of passive smoking, which may inadequately reflect the child's dose of environmental tobacco smoke. In general, exposure misclassification reduces the chances of observing a differ- ence in asthma between exposed and un- exposed children. The aim of this study was to test two hypotheses: first, passive smoking is a risk factor for the asthmatic state, and second, recent passive smoke exposure acts as a trigger of actue attacks of asth- ma. To provide a more objective deter- mination of exposure to tobacco smokq we measured cotinine in the urine of these children. Cotinine is a metabolite of nico- tine, with an elimination half-life of about 20 to 40 h(18). Among nonsmok- SUMMARY Tn epw the r.4tbn.hly ta.tw.~n k>. amaYmy and aatMns. wami.qsaas.d (1) wM1hM paaahm amCMrlp wta men PrNa1Mt Ynona aatllnMltC tttan CCmtol Chadn,l YM (2) MMth- trr aepMltle t0 mnaEcY Mnda rr hlgnN In eNtM Y6a1ta thars In nunaeub aNtttms. T1tttM 9qYptr rw,a raendt" InIO a f~M-eenlsW attdy' T2 mYp aatMMtk Ct1110nn trCm the asMSqaner rya/n (ER), 35 nenaeuM Nmmatle eMtdnn bem the aathma eanle, and 121 eeMwt e1111Mn hem the ER. SOth qYMfWnttalrv rW Wh1Yry CottnM,MNMIMIM ntelC IeeR) rnaA uled to aarM•N patral" amDlt. In9. twma of CCR a]0 nymq wese ua.d a Idsatdty eMMn eapoaed M Isons.. k/.an CCR wae aNe compubd.lloub.nd nenacuH a.1Mn.ac eMadmr had aaMlMr pr@wN~ nr paW" emepnp at hwn.. acut. ua.. Mwwd a NpMr mrn CQtthas nen.euY e.~OuttN, wa n<t apnnkant. In comprtnp all ast/MrNtk to euntrul ehaMM,, Mnek/nq by the maMrnal canlqtwr vu nlon p,m. lent amnnq rmmMlc eh6dnn (udda ratio OR - aq 66% CI 1.1, SA). TMa w een6sm.d by CCN > lo np/ma (OR . 1.9, 96%CI 1.04, 735) rrd br aM dt7MMSean sn.us CGtt (49J vanw 26J rs9/m0, p. 0.06). WR eoneludt, that amuklnp by the rtraqrna eanplw is aaaeoMtM ivhh ctlnlealsy N9nlt1- rant a.thme In clslltl,an. We eould not ahur thal It Ia e trlapar of aaRa Ylnnu .Itaeta. AR /IeV R!a/M m61lra;lN.fM-SN . ~ ; ing children, it has been shown to be cor- related with maternal smoking and with the number of smokers at home (18, 19). Mettrods A case-control study was conducted from Oc- tober 1988 through April 1989 in a New York City medical center. The study population comprised inner city children aged 3 to 14 yr attending the pediatric emergency room (ER) or the pediatric asthma clinic at the hospital. The ER functions as both a walk-in clinic and an emergency room. Cases of acute asthma were ascertained from children presenting to the ER on week- days. The definition of a case of actue asth- ma required (1) a physician diagnosis in the ER of acute airflow obstruction requiring bronchodilator therapy, and (2) at least one prrvious episode of physician-diagrtosed acute asthma as reported by the accompanying adult. A second case group, consisting of children whose asthma was not acute, was recruited from all children aged 3 to 14 yt attending the asthma clinic during the period of study- These children all had a history of episodic or chronic airflow obstruction requiring some form of bronchodilator therapy. Any child (1) who had suffered an attack of acute uth- ma resulting in a visit to a doctor or school absence during the previous 2 wk or (2) who required treatment during that visit to the c ic was eacluded. The control group comprised children tendingthe ER during the period of the srr with any presentation other than acute a! ma. Theaccnmpanyingadult was intervin about demographic characteristics (age child, sex, ethnic group, parental occupati and years of schooling completed by parer pets, recent history of asthma and other ness (including recent upper respiratory fection), use of asthma medication, smoking habits of the maternal careg (mother or other primary c aregiver ) and - (ReceivMinorigrnel/ormAudusr29, ffl90a t . .evised Jorm lu)Y lJ, 1991) I From the Division of En~ironmental an< cupational Medicine. txpartment of Comm Medieine, and the Jack and Lucy Clark Dt i, ment or Pediatrics, Mount Sinai Medical C. New York, New York. ' SupportedinpanbyGram Nos. ROIESI and RAOOPr1 fromthe National Insdtuwes of F and a pant from the Charles A. Dana Found and Mafesnal and Chitd Health Grant MCJ 0( ' Cortdpondence and requests for rr should be addressed to Rodney Ehrlich. Department of Community Medicine. l:m• or Capetown Medical School, Obser.a,ory South Africt. 594 2505493837 1

C,IILalraaa AlrN11A aMa VaSa1Ve 1Ya1UMa er household members. Gas stove use was not included because a pilot study found gas use to be almost universal in this poptilation. Chil- dren aged 10 to 14 yr were taken aside by the interviewer, who inquired whether the child had ever tried smoking, and if so, whether he or she had smoked recently. Chart review was undertaken to exclude subjects who did not meet study criteria and to deterntine asth- ma medication. An upper respiratory infection (URI) was defined as the occurrence during the previ- ous week of any twa of a list of four symp- toms. These included (1) a runny nose not "ttsual" nor "frequent" for that child (to dis- tinguish acute infecton from allergic symp- toms), (2) sneezing, also not usual nor tre- quent. (3) sore throat, and (4) sore or discharg- ing ears. Alternatively, if the adult reported fever in the child or a cold in another house- hold member in the previous week, only one af the four symptoms was needed to define a URf. Noncompliance was defined as having missed one or more doses of asthma medica- tion during the previous week if the child was on a prescribed daily regimen. An index of socioeconomic status (SES) was computed from the occupational category and years of education of the parent(s) with whom the child resided (Hollingshead AB. Four factor index of social sutus- Department of Sociol- ogy, Yale University, 1975). Parents or guar- dians were classified into three categories: (1) unskilled, semiskilled, or not formally em- ployed; (fl) skilled, clerical, or sales: or (IIn technical, professional, or business. The 6 months of the study were classified by average monthly temperature (recorded by the New York Meteorological Service) into cool months (October, November, and April) or cold months (Dettmber, February, and March). The smoking status of the maternal caregiver (current, ex, or never) and of each household member (current or not) was re- corded. Number of cigarettes smoked daily was coded into four intervals (i to 5, 6 to 15, 16 to 20, and > 20). The total mmmber of ciga- reues smoked datly by all household smokers was expressed as a continuous variable by summing representative numbers for each of the four intervals (2, 10. 20, and 25, respectively). A urine specimen was collected from each child at the time of the interview. Within 2 h it was deep frozen until transfer to the labo- ratory. Urinary cotinine concentration was de- termined by competitive inhibition radioim- munoassay using rabbit cotinine antiserum and tritiated cotinine (20). To adjust for the effect of variable dilution on the spot con- centration of cotininq urinary creatinine was measured and the cotinine/creatinine ratio CCR) was calculated. Because the frequency distribution of the CCR values was highly nonnormal, being skewed to the right,.CCR values were ana- lyzed in two ways. First, Henderson and col- leagues (21) reported that a cutn ff CCR level of 30 ng/mg identifies children exposed at home with a high degree of sensitivity (80%) and specificity (100a7a). This level was there- fore used to categorize subjeets into exposed and unexposed. Second, logarithmic trans- formation of CCR produced a bimodal dis- tribution with one peak at the zero or non- detectable level and a log normal distribution of the remaining values. Owen and DeRouen (22) have shown that a function proposed by Aitchison (the Aitchison estimator) provides best estimates of the mean and variance of such a distribution. These sutistics were esti- mated for grouped CCR values and a z test applied to differences between groups. The role of possible confounding or effect- modifying variables, such as agC se:, and eth- nicity, were examined by stratified analysis. (Effect modification refers to significant vari- ation of the odds ratio with different levels of a third variable, such as sex. For example, TAatE t ACUTE VERSUS NONACUTE ASTHMA: SAMPLE CHAMCTERISTICS Factnr Acute Asthm. (n . rn Monacute Asthma tn . 35) Aqn, yr 7.0 (3-14) r.9 (3-t4) Sax (mala), % 67 60 Ethniciry, % Hiapanie 60 66 Amran-Amsrican 37 26 Otnet 3 9 SES.%' 1 W Be u 18 20 m ,9 s. URI. %- 69 47t Prevroua uae of ER lor aeuta umma, % 97 96t Any prN,ous ana0danp at aatnma Uinlc. % 65 toof Daily astnma madieation, % 36 da1 Misa.d any tlew in pravious wMks-. % 16 14 s.. t.n ta abnn.cn r OOa~ ru,o lOn/ - P. S f f,1, 5.670" O.fIJ- c-aot. fF~~000 the association between passive smoking arr asthma might be observed among boys bi not among girls). Multivariatc analysis w: performed using the BMDP statistical sot ware for logistic regression via SAS (SAS f i stitute Inc, Release 5.18, March 1989). This study was approved by rhe institution review board of the hospital, and informt consent was obtained for each participan Rnults A total of 271 parents or guardians we approached in the ER and asthma clii ic, of whom 244 (90a1o) gave informt consent. Response rates by group we acute asthma (88%.), asthma clinic (98% and ER control (88%). Fourteen ast matic and two control subjects were r jected on chart review for failing to me study criteria (eg., no previous asthn among acute asthmatic cases or an z tack in the prior 2 wk among clinic case This left 228 children in the study: ' acute asthmatic ehildren, 35 childr, from the asthma clinic, and' 121 ER co trol children. Urine was not obtain from 14 of these, leaving 214 for cotini analysis. The mother was the study 1 spondent for 181 children (79010); for t remainder the father, grandmother, aunt provided the information- For 18 of children, someone other than the bi logic mother was the primary caregiv usually the grandmother. In 55wo the ! ther did not live at home. Among the cc trols, 35074 had respiratory diagnol identified on chart review (including e nose, and throat), 8e/a "viral syndrotc and 12% trauma or soft tissue diagnos 45079 had other medical diagnoses (s dominal, eyes, skin, neurologic, othet including some with no clear diagno7 Acute Versus Nonacure Asthmatic Children Demographic and medical features acute asthma and nonacute asthma displayed in table I. The two groups w similar in age, sex, and SES. Afric American children were overrepresen in the acute group. Recent URI • markedly more common among acute asthmatic children, with an o ratio (OR) of 2.5 (95% confidence in val, 1.1. 5.6). With regard to pattern of medical c most of the children in both groups - viously made use of the ER for at asthma. A smaller percentage (65% the acute asthmatic children previoattended the asthma clinic Among tt children on a daily medication regin there was no difference in the proport; missing one or more doses in the pr ous week- Howevcr, the asthma cl 2505493838 1

!ABLE 2 ACUTE VERSUS NONACUTE ASTHMA: EXPOSURE VARIABLES Factor Acute A.mma Nona(Yta Aatnme (n . )2) (n - 36) Any amukV at Iqm.. % 53 57 oailv clpamtM Dy all lmok.ra' m.an S SO (atanEVd Ov.ilrion) 77 x 11.8 10.7 x 14.6 Mat.rnal C4aQnhr Culront/y anl0kp. % AO 51- CCR > 30 nymp. % - J8 391 M.an CCR. np1mp, %t 48.2 x 99.3 38.5 s 74.11 „upans A0.9x 13.3 31.4 z 14.7 Atnc.n-Amnican 57.At28.9 59.2s<1.7 Omsruq.0.6101l.1Ut.o -0z 1 OGOa uun - 0.9 i0.3a 2 191- G- 08 t Anc/N. ,rMis/p,Mtqn: M1 MFTNOOS fu-O.aa TABLE 3 ASTHMA VERSUS CONTROL: SAMPLE CHARACTERISTICS FaOtcr Asthma (n . 107) Control (n - 12/) Ava. n 7.3 (3-4) 7.5(}1q S.s 4nw), % 62 59 Elhnicly. % 19ap.mc 62 72 AhiCan-ATNrCan 34 28' om« 5 t SES, %r 1 64 66 n 19 19 111 /L 1s URI. %T 62 62 Mpntn, % Octop.rlNewmn.r/Aprll (Caol) 55 40t D.c.mWNF.EruvylMarM (coW) 45 60 t See vn for wfinn,on. je-0us. TABLE 4 ASTHMA VERSUS CONTROL EXPOSURE VARIABLES Factor Auhma (n . ro) Comr01 (n . 12I) Any lmokp at (Wms, % 54 51 Oaily cpu.nn ty atll amokua. m.an : 5o 8.7 z 12.8 6.1 z 10.3 MMamal ws9nM finek.s, % i4 28' CCR ~ 30 nymp, % 38 25T M.an CCR- ng/mg= 43.6 s 87 7 25.8 z 46.54 - 000. 161m - 20 0 1. 1.). C- 0-00. toR-+.9(1 01,a15),p.o.M T Mlcmoon tr.nfbrmnqn. N. MFi,qn3 t P - o.oe. group had a much larger proportion on such a daily regimen. There was no differ- ence between the groups in pet owner- ship or month of recruitment. The passive smoke exposure of the two groups is compared in table 2. There was no significant difference in general household smoking. Smoking by the maternal caregiver was more common in the nonacute group. There was no differ- ence in the proportions of children ex- posed at home as defined by CCR levels at or above 30 ng/mg. The mean CCR was nonsignificantly greater in the acute group (46.2 ng/mg) than among the nonacute children (38.5 ng/mg). Because the two asthma groups were sitnilar with regard to dentographic char- atxeristics, smoking prevalences, and past ER use, they were combined into a single asthmatic group for comparison with the control group. Asthmatic Versus Control Groups The asthmatic and control groups are similar with respect to age, sex, SES, and recent URI, as shown in table 3. There was a significant difference in month of recruitment, with asthmatics enrolled in higher proportion than control subjeces in the cool months of October, Novem- ber, and April compared with the cold months. There was also a greater propor- tion of African-American children among those with asthma. There was no difference in ownership of household pets. Comparing smoking variables (table 4), there was no significant difference in the proportions having any smokers at home or in daily cigarette consumption by all smokers. The maternal caregiver, however, was much morr likelyto smoke among the asthmatic group (OR = 2.0). This was confirmed by the differences in CCR, whether defined cazegorically (OR = 1.9) or quantitatively (mean 43.6 ver- sus 25.8 ng/mg). When analysis was restricted to those children (n = 181) whose maternal care- giver was their biologic mother, the same association between maternal smoking status and asthma was found. This was so whether maternal smoking was de- fined as (1) current smoking by the bio- logic mother (OR =1-9 (95% confidence interval ]-), 3.6)), (2) current or exsmok- ing (OR = 2.0 (1.1, 3.8)], or (3) smoking in pregnancy (OR = 1.9 (1.1, 3.5)1. Ethnicity and month of recruitment were examined as potential confounders. African-American children had a slight- ly higher mean CCR (38.9 ± 9-4 ng/mg) than Hispanic children (32.6 t 5-]6 ng/ml), and they showed no significant difference on the categorical CCR mea- sure [OR = (0.7-2.5)j• Regarding month of recruitment, there was no difference in CCR~ between the cool and cold months, whether measured categorically or quantitatively. On enter- ing CCR (4, 30 ng/mg), month, and eth- nicity simultaneously into a logistic regression model, the association be- tween CCR and asthma was altered only slightly, increasing the odds ratio from 1.9 to 2.0, Boys showed a stronger association be- tween maternal smoking and CCR (;a 30 ng/mg) and their asthma than did girls. The differences were not statistically sig- nificant, however. To remove the influence of extreme values, the analysis was repeated cxclud- ing the three CCR outliers greater than 200 ng/mg. The results were essentially unchanged. The analysis was repeated using only acute asthmatic subjects as the case group (table 5). The pattern was similar to that of table 4, except that the odds ratios were slightly smaller, and no longer significant at the 0.05 level, for the comparison of 2505493839 1

-ABLE - ACUTE ASTHMA VERSUS CONTROL EXPOSURE VARIA6lES FaCtor Acute Asmma Controi (n • 721 (n . 12r) Any fmokp at home. % -sa-- 51 Daily nqereffe by a11 smokers, meen z SO 7.7 - 118 8 6., ¢ 10.3 MalVnal OafeQiver stllokes. % 40 26- CCR a 30 n¢/mg % 3B 25t Mean. CCR. nyrmQ= a6.2 e 98.3 25.fi n 46.6t ' Oddf rYa •1 7 (0 a2. 3 15). P- O.Oe toCe.n,w. 1 a10.9r.J.W1.D -0.07 t Ailcrrnon v.narCm+uqn. aw MEinopG tD•Aii maternal smoking and CCR measured categorically. The mean difference in CCR between acute asthmatic and con- trol children was 46.2 versus 25.8 ng/mg, was also non-significant. The association of CCR (2 30 ng/mg) with questionnaire measures of exposure was computed. CCR was most strongly associated with the maternal caregiver's smoking status [OR = 11.9 (6.3, 22.3)]. Association with smoking by household smokers other than the maternal care- giver was lower [OR = 3.4 (1.3, 8.7)]. To examine further whether smoking by the maternal caregiver was mainly a sur- rogate for the total number of smokers in the home, the correlation of smoking by maternal caregiver with number of other smokers was calculated. There was little correlation (Spearman's coefficient = 0.1; p = 0.13). CCR thus accorded most closely with current smoking by the maternal caregiver as an independent source of passive smoking by the child. None of the children questioned ad- mitted to smoking. Active smoking by the child could be a confounder of the association between passive smoking and asthma. If recent, it should account for CCR levels at the high end of the distri- bution. These possibilities were explored by examini ng the charaeteristics of t hose children with high CCR values (> 100 ng/mg) (table 6). The ages of these chil- dren were mostly at the low end of the age rang0. making it highly unlikely that active smoking explains most of these values. Funher, they were evenly divided between cases and controls, so that even if all were attributable to active smoking the error would not be systematic. Oiscusalon We found that passive smoking is as- sociated with clinically significant child- hood asthma in a sample of children drawn from an inner city population of mainly Hispanic and African-American children using a hospital's ambulatory care services. The comparability of the groups needs to be considered. Controls in our study were drawn from the pediatric ER and can be regarded as sampling the popula- tion of children who use the ER. There TABLE 6 SUBJECTS WITH COTININE/CREATININE RAT/O > ,00 NCa/M0 Age Matemal Care9wer COR tyr) SmO4N other STOkera at Nome Astnma 745 3 No Yes 335 5 Yes Yes 270 4 No No 238 3 No NO 186 5 Yee NO le0 3 Yes Yes 118 9 Yes Yea Control e66 3 Yea No 232 5 No No 136 7 Yes No 130 S Yes Yes 128 a No Yes 112 12 No NO 102 9 Yes Yes '.terC no confollndln2 ,7enlJgray^nK .1:1ferences between control and asthmatic children. The asthma clinic subjects. al- though not drawn from the ER. prevl- ously used the ER in 8607o of cases, mak- ing it unlikely that they differed markedly from acute asthmatic or control children in their pattern of use of the ER for acute illnesses. Among children using the ER with di- ~ agnoses other than acute asthma, a large proportion present with respiratory symptoms. Passive smoking has been shown to be associated with acute respi- ratory infection in younger children and chronic respiratory symptoms in older children (23). Our control group there- fore probably had more passive smoke exposure than would be found in a com- parable group of community controls. If so, we would be less likely to observe a difference in passive smoking between control and asthmatic children in this study. The fact that an effect was none- theless found strengthens its validity. Smoking by the maternal caregiver was the exposure variable most strongly as- sociated with theasthmatic state. There is now evidence from a number of studies that it is maternal smoking that is im- portant in predicting the risk or severity of asthma or wheezing (4-7, 9, 14-16). In our study we were unable to distin- guish among current or past smoking by maternal caregiver or smoking in preg- nancy by the biologic mother, as these measures were closely intercorrelated and all significantly associated with the child's asthmatic status. However, we confirmed that this is a direct effect of maternal smoking rather than-a reflec- tion of the number of smokers in the household. We were unable to show an effect of passive smoke exposure on the precipi- tation of acute asthmatic attacks. For this purpose, we distinguished children visit- ing the ER with an acute attack at the time of recruitment from children with presumed similar asthmatic conditions who were not acute. We found no dif fer- ', ence between the acute and nonacute j groups when the CCR was used as a cate- ' gorical measure (OR = 0.9), and self- ~ reported smoking by the caregiver was , actually more common among the non- , acute group (OR = 0.6). Using CCR as ~, a continuous variable, an elevation in the acute group was non-significant (46.2 ver- sus 38.5 ng/mg). The power of this second part of the study to show a twofold excess of ex- posure among acute asthmatic subjects ~ 2505493840 1

•f as less than 50a'o. Howe\rr, tonjecture =hat a larger sample might have shown ,uch a significant positive association inust be balanced against what was ob- served: no difference at all between the two groups with regard to cotinine mea- sttred categorically, and a negative associ- ation between maternal smoking status and acute asthma. There were also some differences be- :ween the two asthmatic groups, which :nay have made it more difficult to show s positive association, Among the acute :ases, only 65% had at some time previ- 3usly attended the asthma cGnic Further, 3nly 34%p were on daily asthma medica- :ion at the time of recruitment compared :o g0%u of the asthma clinic cases. It is iherefore possible that children attend- ~.ng the asthma clinic have more severe isthma. If greater severity of asthma is itself associated with passive smoking ;14-16), the asthma clinic gmup may have had more passive smoke exposure to be- gin with. This would make it more diffi- ,ult to show an elevated CCR in the acute 3sthma group even if they were subject :o recent increases in exposure. It is pos- ;ible that a real effect was thereby ob- ;cured. An alternative possibility is that :reatment suppresses the effect of pas- iive smoking. In such a case, the trigger- .ng effect of such exposure may be evi- 9ent only in comparing acute cases with nonacute cases among children not on -egular medication. Our numbers were .oo small to explore this further. Use of cotinine as a biomarker of ex- ~osure enabled us to validate the report- od smoking status of the maternal care- ;iver and to demonstrate nicotine ab- ;orption by the child. In addition, it 7rovided an exposure measure free of in- erviewer bias. The use of cotinine raised methodo- ogic questions that need to be resolved. 3ther studies have shown that cotinine s measurableS sometimes at high levels, n children with no reported exposure at 3ome (19). We found this also. We made Jse of the findings of Henderson and col- eagues (21), who found that a cutoff level >f 30 ng/mg optimally distinguished chil- iren exposed to tobacco smoke at home measured by air nicotine concentration md cigarette butts saved) from those inexposed. We used cotinine to measure degree of 'ecent exposure among acute and non- icute asthma. This presupposes that coti- 3ine levels reflect the intensity of passive ;moking. Because of interindividual lifferences in metabolism, however, the variation in connme among children (or a given exposure may be considerable it also remains to be confirmed whether urinary cotinine levels in an individual child are stable over time so that a single measure reflects "average" exposure, or whether they are sufficiently sensitive to changes over and above this background exposure level to detect short-term ("peak") increases. Henderson's group, doing repeated measures, found stable urinary CCR levels over a period of 4 wk (21). The correlation coefficient found between average log CCR and average home nicotine concentration was 0.68. In contrast. Coultas and coworkers re- ported a wide variation in urinary CCR over a period of about 11 wk; their corre- lation coefficient between CCR and am- bient nicotine was 0.15 (24). In view of the difficulties posed by these conflict- ing data, the hypothesis concerning acute exacerbation of asthma by environmen- tal tobacco smoke needs a prospective study of asthmatic children, linking acute exacerbations of asthma to variations in exposure based both on repeated mea- sures of cotinine and on some measure of environmental exposure. We conclude that passive smoking by the mother or other maternal caregiver is associated with the asthmatic state among children. Given our observed odds ratio around 2.0, the high prevalence of both parental smoking and asthma makes this association a public health problem of considerable impact in this population. We could not show that recent eleva- tions in exposure to tobacco smoke trig- gered attacks of asthma requiring visits to the emergency room. Lack of statisti- cal power, differences between acute and nonacute cases in medication use, and limitations in using a single cotinine mea- sure may explain this finding rather than a true lack of effect. If our finding is val- id, however, it may be because the mech- anism of effect of maternal smoking on asthma is through increasing bronchial responsiveness in the child rather than by triggering bronchospasm. A number of studies have shown that bronchial re- sponsiveness among asthmatic children is greater if the mother smokes (14-16, 25, 26). Such a mechanism is also com- patible with the finding of Evans and col- leagues (17) that the number of ER visits for acute asthma is increased if there is a smoker in the household. In contrast to these findings, those of general popu- lation studies have yet to clearly demon- strate an association between bronchial responv•enaso :n <hfwren an:i parei:a: smoking (2'., 26). The clinical implications of this study are clear. Maternal smoking in the house- holds of asthmatic children in this popu- lation is all too common. Reduction of this potentially important risk factor should be the target of clinicians and health educators working with the fami- lies of these asthmatic children. Acknowledgment The writers thank Michele Marcus, Ph.D., and Nancy 1. Haley, Ph.D., for their helpful ad- vicq Caryn M. Axelrad, MS., for perform- ing the cotininc assays, Eva Chan, M.S., for assistance with the data management, and Kent Pena and Alex Echevarria, who conduct- ed the interviews. Reterencea 1. tseder SR. Corkhill RT. Irwig LM. Holland WW. Influence of family fac[ors on asthma and wheezing during the first five years of life Br J Prev Soc Med 1976; 30113-8. 2. Horwood L, Fergusson DM, Shannon FT. So- cial and familial factors in the development of ear- ly childhood uthma. Pediatrics 1985: 79:859-68. 3. Fergusson DM, Horwood LI. Parental smok- ing and respiratory illness during childhood: a six- year longitudinal study. Pediatr Pulmonol 1985; 1:99-106. 4. McConnoNMie KM, RogJsmann KJ. Breast feed- ing and maternal smoking as predictors of wheaz- ing in children age 6 to 10 yean. Pedietr Pulmonol 1986; 2:260-9. 5. Neuspid DR. Rush D, Butler N R, rt af. Paren- W smoking and post-infancy wheezing in children: a prospective cohort study. Am J Public Health 1989; 79:168-71. 6. Grotmaker SL. Walker DK, lacobs FH, Ruch- Rou H. Parental smoking and the risk of child- hood anhna Am 1 Public Health 1982; 72:574-9. 7. Weitzmap M. Gortmaker S. Walker. D. Sobol MA. Maternal smoking and childhood- asthma. Pediatrics 1990; 85:505-u. 8. Weiss ST. Tager iB, Spelzer FE. Rome, B. Perv slstent wheeze. Its relation to respiratory illness, cigarette smoking, and level of pulmonary func- tion in a population sample of children. Am Rev Respir Dis 198o- 122:697-707. 9. Ware JH, Dockery Dw, Spiro A. 111. Speizer FE, Ferris BG, Jr. Passive smoking, gas cooking, and respintory health of children living in six cit- ies. Am Rev Respir Dis 1984; 129:366-74. 10. BurtfsfM CM, Higg4u MW. Keller J& Howart WF, Butler WJ, Higgaas ITT. Pusive smoking in childhood. Respiratory conditions and pulmonary function in Teeumseh, Michigan. Am Rev Rcspir Dis 1986; 133:966-73. 11. Lebowitz MD, Burrows B. Respintor,v symp- toms related to smoking habits of family adults. Chet 1976; 69:48-50. 12. Schflling RSF, t<taf AD, Hui SL. Beck IB• Schoenberg JB, Bouhuys A. Lung function respi- ratory disease aad smoking in families. 4m I Epidcnsiol 1977; 106:274-83. 13. Schenker MB, Samn 1M, Speizer FE. Risk factors for childhood asplratory disease The eF feet of host facton and home environmental e.- posure. Am Rtv Respir Dis 198): 12E:1038-43. 2505493841 I

j4_ Murny AB. ?somson Bl. The effect of oga- rette smoke from the mother on bronchial respon- sivaness and sevcrity of symptoms in children with asthma. J ABergyCBn Immunol 1986; 77:575-81. I5, Murray A73, Morrison BI. Passive smoking apd the seasonal difference of severity of asthma in children. Chat 1988; 94:701-8. 16. Murray AB. Morrison 81. Passive smoking by asthmatics: its peater effen on boys than on girls and on older than on younger children. Pedi- atrics 1989: 84:451-9. - 17. Evans D, Levison MJ. Feldman CH, ef aL The impact of passive smoking on emergency rmm visits o f urban cIW dren with asthma- Am Rev Respir Dis 1987; 135:367-72. Ig. U.S. Department of Health and Human 5er- vices. The health consequences of involuntary smoking: a report of the Surgeon Genenl. Washing- ton, DC. Goverr.menr PrnungOfOce, i986. PuD- lication No. DHHS ICDCI 9'-g398. 19. Coultas DR Howard CA. Peakc GT, Skipper fll. Sarnet 1 N. Salivary cosinine levels and involun- rary tobacco smo ke ezposurc in children antl adWts in New Mexico. Am Rev Respir Dis 1987; 136:305-9. 20. Langone 1J. Gjika HB. Van Vanukis H. Nico- tinc and its meubolites: ndiolmmunoassays for nicotine and cotinine. Biochemistry 1973: 12:5025-30. 21. Henderson FW. Reid HF. Morris R. et al. Home air nicotine levels and urinary cneinina ez- cretion in preschool children. Am Rev Respir Dis 1989; 140:197-201. 22. Owen WJ. DeRoucn TA. Estimation of the mean for log normal data conuining zeroes and Ieft~censored values, with applications to the mea- surement of worker ezposure to air contaminann. 3romevncs 1Y80: ?5.'0--;9 23. Fielding JE Phcnow f;J. Hcalm <irec's st .c 1oluntarv smoking. }' Engl I ?1ea :998 319:1452-60. 24. Coulus DB. Samet IM.McCarthy IF. Spcn gler ID. Variability of measures of eaposure ro en WronmenW tobacco smoke in the home. Am Re• Respir Dis 1990; 142:602-6. 25. O'Connor GO, Weiss 5T, Tager 1, Speizer FE The effect of passive smoking on pulmonarv func tion and nonspecific bronchial responsiveness rt a population-based sample of children and younl adults. Am Rev Respir Dis 1987; 135:800.n. 26. Martinex FD. Antognoni G. Mach F, er ol Parental smoking enhances bronchial r<sponstvc ness in nine-year old children. Am Rev Respir Dt. 1988: 138:518-23.