- Email: [email protected]
Reduction of Environmental Tobacco Smoke Exposure Among Asthmatic Children: A Controlled Trial
Reduction of Environmental Tobacco Smoke Exposure Among Asthmatic Children: A Controlled Trial* Melbourne F. Hovel!, Ph.D. , M .P.H. ; Su san B. Melt zer, M .P.H.; Joy M. Zakarian , M .P.H .; Dennis R. Wahlgren , M .A .; Jen nifer A . Emerson, M .P.H.; C. Richard Hofstetter, Ph.D.; Brian P. Leaderer, PhD ., M .P.H .; Eli O. Melt zer, MD. ; Robert S. Zeiger, MD. , PhD. ; Richard D . O'Connor, MD .; Mary M. Mulvihill, PhD.; and Catherine ]. Atkins, Ph.D. Study objective: This randomized clinical trial tested a behavioral medicine program designed to reduce asthmatic children's exposure to environmental tobacco smoke (ETS) in the home. Design: Families were randomly a ssigned to an experimental preventive medicine counseling group, a monitoring control group, or a usual tr eatment control group. Families were measured six times over 1 year. Participants: Ninety-one families were recruited from four alIergy clinics. Intervention: The experimental gro up received a 6-month series of counseling sess ions designed to decrea se ETS expo sure. This group a lso monitored sm oking, exposure, and children 's asthma symptoms. The monitoring gro up did not receiv e counseling and the usual treatment control group rec eived outcome measures onl y. Measurements and results: Parents reported the daily number of cigarettes children were expose d to during the week preceding interviews. A nicotine air monitor and con struct vali dity analysis confirmed the valid ity of exposur e reports. Exposure to th e parent's cigarettes in
recent Environme nta l Prot ection Agenc y rep ort A sum marize d decad es of resear ch linking passive
smoking with negative health outcomes.' Special popul at ions, such as asthma tic child ren , ar e at elevated risk for the harmful effects of environme ntal tobacco smo ke (ETS). Environme ntal tobac co smoke is estima ted to exace rbate the symptoms of approx"From the Ce nte r for Behavioral Epide miology, Grad uate School of Public Hea lth, San Diego Stat e Universitv (Drs. Hovell, Hofstett er, Mulvihill, and Atkins; and Mss Meltzer, Zakarian, and Emerson, and Mr Wa hlgren ); the John B. Pierce Fou ndation Laboratory and Depa rtmen t of Epidemiolog y and Public Health , Yale University School of Medicin e, New Haven, Conn (Dr. Lead ere r), the Allergy and Asthm a Medical Group and Research Cen ter, San Diego, and Department of Ped iatrics, Universit y of Californi a, San Diego, School of Medicine (Dr. Meltzer); the Depar tment of Allergy , Kaiser Perm anente Medica l Cen ter , San Diego, and Departm ent of Pediatrics, Universitv of Ca lifornia , San Diego, School of Medicine (Dr. Zeiger); and"the Di vision of Asth!TIa, Allergy and Clinica l Immu nology, Shar p Rees-Stealy Medical Group , and Departm ent of Ped iatrics, University of Ca liforn ia, San Diego, School of Medici ne (Dr. O'Connor). This research was supported by funds provided by the Cigarette and Tobacco Surtax Fun d of the Stat e of Ca lifornia thr ough the Tobacco-Related Disease Research Program of the University of Ca lifornia, gran t l RT509. Manuscript received October 2 1, 1993; revision accepted Decembe r 28.
440
th e home decreased for alI groups. The experimental group attained a 79 percent decrease in children's ETS exposure, compared with 42 percent for the monitoring control and 34 percent for the usual treatment control group. Repeated-measures analysis of variance resulted in a significant (F([10,350]=1.92, p
imat ely 20 percent of thi s country's 2 to 5 million asthmatic children and is a major aggravating factor in approximately 10 percent.' Passive smoking has been positively associated with increased prevalence of asthma, with incr eased frequenc y of medication use, severity of asthma symptoms, acute exacerbations and respiratory tract inf ections, and with reduced pulmonary fun ction in asthmatic chlldre n.P'! Given th e demonstrated associations between ETS exposure and negative health effects and the high probability that asthmatic child ren are exposed to ETS in the short and long term , it is imperative that clin ical and public health int er ventions be developed to reduce ETS exposure to asthmatic child ren . This study was a randomized controlled tri al of a pr eviously deveioped -- beh avioral coun seling program for reduction of ETS exposur e in the home among asthmatic childre n living with par ent s who smoke. M ETH OD S
Design Stratified by medical center, a three-group repeated measures Tobacco Smoke Exposure in Asthmatic Children (Ho
Table I-Schedule of Measures and Counseling Sessions Across Groups Environmental Monitor
Peak FlowMeter
Child's Symptom
Paren t's Smoking
Group
FE Vl
Interview
Diar y
D iary
Counseling
Experimental/counseling Monitoring control Usual treatm ent control
X X X
X X X
X X X
X X X
X
X
X
X
X
design was used . Nin et y-on e families with at least one parent wh o smoked ciga re ttes and a child (6 to 17 years) with asthma wer e randomly assigne d to an expe rime ntal/counseling , a monitoring control, or a usual tr eatment contro l group. F amilies in eac h gro up provid ed m easur es of children 's ciga re tte smo ke exposure and health sta tus and par ents' smo king rat es at a baselin e hom e visit, at a pr einter vention clinic visit, at tw o clin ic visits during interven tion (2 months and 6 months), a nd at two clinic visits during follow -up (9 months and 12 months). Two-week monitoring intervals pr eced ed clinic visits. All families receiv ed standard medical care. Tw elve famili es dropped out of the stud y during th e l-year enr ollme n t period. Eight refused further participation, two lost custody of the asthma tic child , one moved out of the ar ea, and one child was institutionaliz ed . Thus, final analyses wer e com ple ted on 79 families. Tabl e I outlines the study measur es acco rd ing to group assignme nt. . For each family, one smokin g parent and one asth matic ch ild wer e selec ted as th e " targe t par ent " and " ta rge t ch ild ," respectively. While othe r famil y member s participated to vary ing d egree s in th e study , most measur es wer e obtained from th e target parent and child. Wh en possible, informa tion reg arding the othe r
parent in a famil y was ob tain ed directly fr om th at paren t.
Subjects Th e mean age of tar get childr en was 10.77 (50 = 2.98) yea rs. More th an two th irds (70 per cent ) were whi te , whil e 9 per ce nt wer e black , 7 percent wer e La tin o, 4 per cent we re Asian, and 10 per cent were member s of othe r eth nic classificati ons. Mor e th an half (57 percent ) of th e target child re n wer e ma le. Th e mean age of targe t par ents was 37.96 (50 = 5.82) yea rs. Seven ty -seven perce nt wer e white, 6 per cent wer e black, 6 per cent wer e Latino, 8 percent were Asian , and 3 per cent we re oth er. Th e majority (67 percen t) of target parents wer e female (Table 2 ).
Recruitment Families wer e recruited from four of the largest pe d iatric a llergy medica l ce n ters in San Diego, Scree n ing forms we re com pleted during regul ar office visits. To be included in th e study, the ch ild had to require asthma medicati on a nd to be ex posed to a t least one ciga re tte per d ay b y th eir parent. Th ose meeting th e inclusion criteria were con tac ted b y a research assista nt a nd invited to participate in a study of th e effect s of tobacco exposure
Tab le 2- Group Distrib utions f or Se lec ted Dem ographic Variables Group l,l n (%)
Group 2,1 n (%)
Group 3,1 n (%)
10 (11.0) 21 (23.1)
15 (16.5) 13 (14.3)
14 (15.4) 18 (19.8)
21 (23.1) 10 (11.0)
20 (22.0) 8 ( 8.8)
20 (22.0) 12 (13.2)
17 (18.9) 14 (15.6)
22 (24.4) 6 ( 6.i)
24 (26.7) i ( i. 8)
20 (22.2) 11 (12.2)
25 (2i .8) 3 ( 3.3)
25 (2i .8) 6 ( 6.i)
21 (23.1) 10 (11.0)
23 (25.3) 5 ( 5.5)
24 (26.4) 8 ( 8.8)
2 ( 2.2) 12 (13.2) 16 (Ii .6) 1 ( 1.1)
2 ( 2.2) 6 ( 6.6) 19 (20.9) 1 ( 1.1 )
2 ( 2.2) 10 (11.0) 15 (16.5) 5 ( 5.5)
Overall Mean (SO)
Group 1 Mean (SO)
Group 2 Mean (SO)
Group 3 Mean (SO)
10.77 (2.98) 37.96 (5.82) 2.84 (5.58)
11.16 ( 2.9i ) 39.39 ( 6.82) 3.42 ( 8.82)
10.82 ( 3.13) 3i .00 ( 5.82) 2.il ( 3.31)
10.34 ( 2.90) 37.41 ( 4.54) 2.38 ( 2.25)
Varia ble*t Child gend er Fema le Male Parent gender Female Male Child ethnicity Whil e Nonwh ite Par ent cthnicily Whit e Nonwhite Parent marit al status Marr ied Not marri ed § Parent education < High school graduate High school gra duate Some college College grad uate Variable"]
Child age Par ent age Parent num ber of lifetim e attemp ts to quit smoking
"Pearson's X2 analyses for categorica l variables and ANOVA for continuous variables showed no statistically significant gro up differences (all p>0.05). tn =91 for all varia bles except child and parent ethnicity (n = 90). [C roup I : experimental/counseling ; group 2: monitoring control; group 3: usual treatm ent control. §Marri ed includes cohabitation. Not married includes single, di vorced , separa ted, and widowed . CHEST 1106/ 21 AUGUST, 1994
441
on the hea lth of thei r child . All fam ilies wer e told that they wou ld receive incremen tal payment s totali ng $60 for attend ing clinic sessions and for provid ing information th at could help in th e unde rstand ing of th e relati onship be tween passive smo ke exposure and asthm a. T hey we re not told that th e study would invo lve an int ervention to red uce ETS exposure. All wer e provide d with a description of the possible risks and benefits and all voluntee rs signed informed consents. A sequential recruitment procedure was employed until 91 families wer e recruited ac ross all clinic sites. Th e study was approve d by th e San Diego Sta te University Com mittee on Prot ecti on of Hum an Subjects .
Procedures Experim enta l/Co unse ling Group: Both the target par ent and ch ild atte nded a series of cou nseli ng sessions over 6 m onths design ed to reduce th e child 's exposure to pa re nta l smo king. Between cou nseling sessions, d iari es wer e used during the 2-wee k monitoring periods preced ing clinic visits to record par ent smo king and child ETS exposure patterns, children 's peak flowmeter readings, and children 's sym ptoms. Th ese dat a wer e used to recommen d cha nges in parent al smoking patt ern s that could reduce ETS exposure for the child. Beh avior modificati on techniqu es such as stim ulus con trol, sha ping , per sonal fee d bac k, an d conti nge ncy contracting were inco rpo ra ted in a program tailored for eac h fa mil y.13,14 M onitorin g Control Group : To determine the d iffer ent ial effec ts of monit oring and counseling, a " monitoring con trol" gro up was utili zed. Th ese fam ilies were asked to em ploy the same moni toring measur es th at were used by th e counseling group, but neither review of records nor counseling to red uce ETS exposure was pro vided . Usual Treatmen t Control Group: Th e control gro up was asked to participat e in th e sa me seq ue ntia l clin ic visits as the oth er tw o groups, but th ey were not asked to monit or ciga re tte use, passive exposure, or child re n's sym ptoms . No counseling program was provi de d. Thi s group d id record child ren 's peak flowm eter readings.
in the home, it was not ab le to selectively m easur e th e room when the ch ild was pr esent or to sam ple the child 's exp osure to nico tine outside of the roo m. In addi tion, th e beh a vioral counseling pr ogra m was direct ed toward reducin g th e child ren's exposure, not to redu cin g tobacco use in one parti cul ar room . Th us, th e monitor was not conside red sufficiently sensitive for use as a n out com e measur e.
Statistical An alyses Using the monitorin g control and usua l tr eatm ent contr ol groups only, test-r etest reliabilit y was in vestigat ed for child re n's exposure , par ents' smo king rat es, the environme ntal monit or , and FEY I by intr aclass correlatio ns.J7 Measur es for test-r et est ana lyses we re obtained from the prei ntervention a nd 2-mo nth clinic visit inter views. To deter m ine th e validi ty of pa re nt-re ported exposure a nd smo king rat es, Ken dall 's ta u-b was used as a non-param etric test of association betw een these va ria bles and the envi ronmen ta l monitor valu e.l'' Th e valid ity of par ent s' rep ort s of their child ren 's exposure to all ciga re tte smo ke in the hom e was exam ine d using constru ct validity techniques. Associations betw een va riables having hyp oth esized positive, negative, and null relati onshi ps with th e child 's report ed exposur e to all sources of cigare tte smo ke in the hom e wer e tested with Ken dall's tau-b. All valid ity tests were perform ed for pr einter ve nt ion, 2-mo nth, a nd 6-mon th measures. T o identify d ifferen ces a mo ng groups in repor ted ETS exposure, smo king rate, th e environ me nta l m onit or value, and ch ild ren 's sym ptoms over tim e, repeated measur es ana lyses of varia nce (AND YA) was em ploye d. A priori defined pairwise AND YAs were used to detect significant d iffer en ces between grou ps. Du e to violati ons of sphe ricity , all within-subjects F statistics were com pared with ad justed critical va lues (H uynhFeldt epsilon ad justment) an d significance levels ar e report ed acco rdi ng ly.19 Unad justed deg rees of freed om a rc reported . All a priori pair wise com parisons wer e cond uc ted as one-tailed di rectional tests. Ana lyses wer e perform ed using a co mpute r soft wa re package (SPSS-PC+ ).20 R ESULTS
Measures of Tobacco Exposure
Reliabil ity and Validity
Parent Reports: The primary dependent varia bles co llected at base line a nd a t clinic inte rv iews were the child 's exposure to ETS from th e tar get par ent a nd from all person s who smo ked in th e hom e. Exposur e to parental smo king at hom e was segrega ted into two components: exposure on typi cal workdays a nd non workday s. Th e target a nd th e othe r parent (if ther e was one ) wer e asked : " How man y ciga rette s did you smo ke per workday/ nonworkda y at hom e during the past week ?" Th ey also were asked: " Of th ose ciga rettes, how man y was (child's nam e) exposed to?" Fin all y, parents were asked : " Please estima te the number of cigarettes to which (child 's name) was exposed from others living in the home and those who visited ." Answers wer e used to calculate the paren t's mean dail y smoking ra te a nd the child 's mean dail y exposure rat e to par ent's and to all smo kers' ciga rettes in the hom e. Envi ronmental M onit or: An env iron me nta l monit or was used with all thr ee gro ups to valid at e par ent-rep ort ed exposure and smoking ra tes . Th e monit or was a 37-mm -dia me te r polystyr en e cassette contai ning a teflon- coated glass fiber filter (Emfab TX 40h120WW , Pallflex Corp, Pu tn am , Co nn) satura ted with an aq ueo us solutio n of 4 percent sod ium bisulf at e a nd 5 percent etha nol a nd dried . Th e monit or is based on passive d iffusion of nicotin e to a filte r that is ana lyzed by gas chro ma tog ra ph y 15 and has bee n shown to be bo th reliab le and valid.!" F ilter preparat ion was cond ucted a t th e John B. Pier ce Labo ra tor y (Ne w Haven , Conn). The mo nitor was placed in a room whe re grea test exposure was repor ted at the init ial hom e inter view , and remain ed for the 2 week s precedin g eac h clinic visit. W hile the monitor provide d a 2-week continuous measure of nicotin e level in one room
Reliability: The test-r etest reliability coefficients for parent reports of child ren' s exposure ranged from 0.63 to 0,84 across reported measures on workdays , nonworkdays , and composite measures. The coefficients for par ents' self-reported smoking rates ranged from 0,60 to 0,88. The environme nta l monitor's test-retest reliability was 0.72. Th e test-ret est correlation for th e FEY 1 was 0.96 . Validity: Exposure to th e target par ent's smoking on a workday was significantly correlated (taub=0.15 to 0.25, p<0.05) with th e ETS monitor at each of th e three monitoring peri ods tested . The correlation coefficient between th e monitor and reported exposur e on nonworkdays was significant only at 6 months (tau-b=0.23, p
442
Tobacco Smoke Exposure inAsthmatic Children (Hovell et all
Table 3-Group Means (SD) for Smoking Rat e, Exposure Rate, and En vironmental Monitor by Visit n° Smoking rate (1\0. of ciga rettc s/ day )j
Gro up !
73 2 3
Exposure to target paren t's ciga rettes (No. of cigarettc s/ workday )§
71 2 3
Exposure to all cigarett es at home (No. cigs/ day )§
Environmental monitor, p.g / m 3
61
Baseline
Prcintervcnli on
2 mo
6 mo
9 mo
12 mo
21.2 (13.1) 20.2 (11.9) 23.4 (12.0) 5.8 (5.4) 5.5 (3.4) 8.0 (9. 1)
21.5 (15.8) 19.3 (10.3) 22.0 (11.4) 3.4 (2.6) 4.2 (2.8) 5.7 (4,8) 5.1 (4.3) 6.9 (5.6) 7.8 (5.0) 1.7 (1.6) 3.0 (3.3) 2.3 (3.0)
18.9 (17.6) 21.2 (12.9) 19.0 (12.7) I.7 (2.1) 3.2 (2.1) 5.3 (4.1) 2.0 (3.2) 6.2 (4.9) 6.6 (4.0) 1.9 (1.8) 3.7 (4.8) 2.1 (2.6)
16.8 (14.1) 17.5 (11.6) 20.4 (13.7) 1.6 (1.9) 3.8 (3. 1) 3.9 (2.7) 2.5 (4.2) 6.7 (5.9) 5.6 (3.7) 2.8 (3.4) 2.2 (1.9) 2.4 (3.5)
15.0 (14.9) 16.8 (11.5) 20.5 (14.4) 1.5 (2.2) 4.1 (2.5) 4.0 (3.2) 2.2 (4.4) 6.4 (4.0) 5.7 (5.2) 1.9 (1.8) 2.8 (3.2) 1.9 (1.9)
14.5 (13.3) 16.5 (1 1.6) 22.0 (14.7) 1.2 (1.5) 3.7 (3.0) 4.6 (3.0) 2.9 (4.8) 7.8 (6.7) 9.5 (5.1) 1.9 (1.8) 2.7 (3.6) 1.4
. . ·11
2
. . ·11
3
. . ·11
65
. . ·11
2
. . ·11
3
. ··11
(1.1)
*n sizes vary du e to missing data for some variables. [ C roup I : experimental/ counseling; grou p 2: monitorin g control; grou p 3: usual treatment control. tp <0.05 for repeated -measures AI\OVA gro up- by-lime effect. §p < O,OI for repeated -measures AI\OV A gro up-by-lime effect. IIExposur e to all cigarett es at home and environmental monitor measures not obta ined at base line.
rep ort s of smoking rat e were statistically significant (p < O.OI) and most were larger than wer e th e correlati ons with parent-report ed exposure . Th ese values ranged from 0.25 to 0.39 for workdays and fr om 0.18 to 0.29 for non workdays. Th e construc t valid ity tests for all ETS exposure in the hom e result ed in 79 percent agree me nt for all constr ucts. The binomial pr obability of this proportion of agr eem ents across tim e occur ring by chance was statistically significant (p < 0.05), dem onstrating convergent, di verg ent, and discr iminant valid ity of th e par ent-reported measures.
Random Assignment Th e three groups were com pa red for gende r, age, and ethnicity of child an d par ent; par ent s' marital status, ed ucation, lifet im e smoking q uit atte m pts, and repo rted level of smo king; and child's baseline ETS exposure . Pear son's x2 analyses for categoric variables and ANOVA for continuous variables showe d no significant diff er ences among groups, thus confirmi ng the success of random assignm ent (Table 2).
Target Paren t Smoking Levels Tar get par ents reported smo king a mean of 21.6
(SO = 12.3) cigar ettes per typi cal day at baselin e. Repeated-measures ANO VA for smoking rat e showed a significant tim e effect (F[5,350]= 7.92, p
Children' s Environmental Tobacco Smoke Exposure and Asthma Sym ptoms Th e child ren's m ean ETS exposure to th eir parents' cigarettes was 6.5 (SO = 6.6) ciga rettes per workday at th e baselin e hom e visit. Figure 1 shows th e mean exposure ra tes by group over tim e. Repeated-measures ANOV As with exposur e to the target parent's CHEST / 106/ 2/ AUGUST, 1994
443
10
BL
INTERVENTION
FOLLOW UP
8 >-
ell
C
... Ql
6 Usual-tx Controls
Co Ul
~
~
ell
.!:l
4
0
Monitoring Controls
2 Experimental Group O+-~--,r----,.--r-,---r----,-........--r-,---r---,.-...,
o
2
3
4
5
6
7
8
Month
ciga rettes in th e hom e on workdays as the dependent var iable yielde d significan t group (F[2,68]= 11.83 , p
9
10
11
12
FIGURE 1. Mean number of the target pa rent's ciga rettes to whic h childr en are exposed on workdays, over time , and for each experime nta l group .
However , at th e final I2-m onth visit , only th e expe rime ntal/counseling gro up susta ine d a decr ease (44 perce nt). The monitoring con trol gro up showed a 14 per cent increase and the usua l treatment contro l group showe d a 22 per cent increase fr om pr einter vention at 12 months. Pair wise comparisons showed th e same pattern of results as for exposure to th e target par en t' s cigarettes, with the expe rimental/cou nseling group showing a significant decr ease over th e monit oring control group (F[4,148]=3. 71, p
(F[I ,55]=4.89, p<0.05) groups reported a significant mean decrease in symptoms compared with the usual treatment control group. Differences between th e counseling and monitoring control groups did not reach significance, and no differences between groups were found after 2 months. Although no group or group-by-time differences were apparent for FEV I , pulmonary function was consistently negatively correlated with ETS exposure at all measurement points (r = -0.13 to -0.37). Wh en the relationship between F EV1 and exposure was adjusted for age and height using hierarchical multiple regr ession, th e partial correlation remained negative but was not statisticall y significant. DISCUSSION
To our knowledge, this is the first fully controlled trial of a behavioral counseling program designed to assist parents in reducing th eir asthmatic childre n's exposure to cigarette smok e. This analysis is the first to include formal tests of the reliability and valid ity of parent rep orts of cigare tte smoking and childre n's ETS exposure . Par ent reports were shown to be both reliable and valid measures of ETS exposure in th e home. Use of random assignment and demonstrably reliable and valid measures greatly increase th e internal validity of this analysis. Families in the exp erimental/counseling group attained a significant and clinically important decr ease in ETS exposure for th eir asthmatic children. This decrease was achieved not only for exposure to th e target par ents ' cigarettes, but also for exposure to all cigarettes smoked in th e hom e. Th e counseling proc edures used in thi s study are well within th e capacity of most clinical assistants found in outpatient clinics or in private practice offices and offer a potentially practical means of extending the therapeutic effectiveness of the physician. This program required littl e or non e of th e ph ysicians ' time. Th e use of the environme ntal monitor to measure ETS in a particular room of the resid enc e provided an objective means of validating th e parents' mor e subjective estimate of th eir child's ETS exposure. Validity was substantiated by formal constru ct validity testing. From th ese results it is concl uded that motivated parents can report th eir child's ETS exposure accurately enough for clinical purposes. It is important to note that their accuracy ma y have been due, in part, to the apparent ability of th e investigators to verify th eir reports by use of th e environme ntal monitor. Clinical use of these measures might in clude such a monitor, at least as a " bogus pipelin e" in order to enha nce the pot ential accuracy of parent reports.P! Also of clinical importance is the fact that no objective substitute for th e parents' reports of ETS ex-
posure could be used in th e administration of th e behavioral counseling program. The parents' reported details regarding the timing, location, and activities taking place when the child was exposed were the foundations for counseled change in lifestyle to reduce ETS exposure. Such detailed information could not be obtained by existing objective measures of ETS exposure. Thus, this intervention depends on accurate parent reporting. Although thi s study demonstrated satisfactory accuracy of parent reports, it can be assumed that parents ar e not able to det ect subtle level s of exposure. Parent reports ar e inh erently limited in th eir sensitivity, increasing th e need for development of accurate object ive measur es of ETS exposur e. To determine relationships betw een disease and ETS exposure, it may be critical to obtain estimates of more subtl e levels of ETS exposure than can be achieved by parent obser vations. Th e environme ntal monitor used in thi s study has the pot ential for a high level of sensitivity . However , it has not yet been constru cted in a sufficie ntly small and durable form to attain such measures for children . Although only a few centime ters in size, the monitor is not small enough to be unobtrusive. It is relatively fragile and subject to contamination by children 's normal pla y and curiosity, and it is not likely that children could be trained to wear th e monitor continuously in its pr esent form . Thes e limitations restrict th e monitor to mor e limited sampling strategies such as those used in th e pr esent study. By limiting th e monitor to th e room wher e th e most exposure to th e child was reported , sam piing th e child's ETS exposure was maximized . Howe ver , for many families, th e majority of the child's exposure occurred outside thi s one room . Thus, th e environmental monitor was not likely to det ect th e majority of thc child's ETS exposure or change in exposure. In addition , once counseling began , th e parents in th e couns elin g group ma y ha ve decr ea sed th eir child' s exposure by arranging for the child to wat ch television in ano the r room or by oth erwise sepa ra ting th e child from th e plac e of pot ential exposur e. For th ese famili es, th e filter no longer accurately reflected th e child's exposure, but only th e amount of tobacco use in the selected room . These sampling limitations precluded th e use of th e environme nta l monitor as an outcome measure. Th e mean monitor valu es depict relatively low ETS levels and a very small decrease across tim e for all groups, whil e neither the group nor the groupby-time effects reached significance . These result s suggest that the ETS level in th e select ed room decreased during th e cours e of study. However , no inferenc es can be drawn from these result s regarding the changes in the child re n's ETS exposure. CHEST 1 106/ 21 AUGUST, 1994
445
Exploratory analyses of sym ptoms and pulmonary function pro vided encouraging evide nce of possible therapeutic effects of th e behavioral counseling program. Children assigned to the experimental/counseling group and th e monitoring control group report ed significant decreases in overall symptoms at th e 2-month measur em en t point wh ich conclude d th e most intensive couns eling period. The decr ease was gr eat est for th e exper imental/counseling group. These results suggest that health benefits ma y be at tained from th e counse ling program and even from self-monitoring alone , and support replication with a larg er sample in order to provide sufficient power to discover oth er possible health benefits related to decreased ETS exposure among asthmatic children . Although th is analysis was not design ed to achieve smoking cessation among parents, a notable order ed diff erenc e among groups in quit rate was observed . The gr eatest number of qu its occurred in th e expe rimental/couns eling group, follow ed by th e monitoring control and the usual tr eatment control groups. Th ese d ifferenc es d id not reach significa nce. However, if th ese rat es wer e replicated with a larger sam ple, smoking cessation might be found to be an added ben efit of the program . Of course , since thes e parents reported previous quit attempts, future assessments should exam ine maintenance. The sample included in th ese analyses was attained from four asthma specialty services in San Diego , and as such provided modest heterogen eity with respect to types of medical care recei ved . Though sam ples' sizes did not permit clinic-specific analyses, th e use of th ese different services sugge sts that th ese results may be generalizabl e to similar services prov ided to working class and predominantly, although not exclusively, white fam ilies. Additional resear ch will be needed to determine wh ether similar counseling procedures are as effective with minority or lower socioeconomic status families. Finally, these ana lyses were limited to a I- year follow-up. Based on pr evious behavioral research, it is likely that th e expe rime ntal/ counseling group's reduced ETS exposure will not be sustained once the int ervention program is completely removed. The actual level of recidivism needs to be det ermined and it may be necessary to develop additional int erv entions th at could be implem ent ed by the ph ysician or clinical staff to sustain th e decreased ETS exposure att ain ed immediately following counselin g. The fact that th e expe rimental/counseling group did sustain a large decr ease in exposure at 9 and 12 months, when monitoring but no formal counseling was provided , ind icates that minimal intervention ma y be suffic ient for mai nt enance. Th ese results justif y follow-up analyses an d furt he r exploration of affordabl e techniques that could support ma int enance of th erapeu 446
tic effects . The significant reduction in ETS exposure an d th e promising results reg arding smoking cessation seen in th is sample of high-risk asthmatic families ma y hav e broad application and result in important pu blic health benefit s. REFERENCES
1 US Environme ntal Protection Agency. Respirat ory health effects of passive smoking: lung cancer and other disorders. EPAj600j6-90j006F. Washington, DC: Office of Health and Environmenta l Assessment , 1992 2 Dekker C, Dales R, Bartlett S, Brunekr eef B, Zwancnburg H. Childhood asthma and the indoor environment. Chest 1991; 100:922-26 3 Weitzman M, Gortmaker S, Walker DK, Sobol A. Maternal smoking and childhood asthma. Pediatrics 1990; 85:505-11 4 Murray AB, Morrison BJ. Effcct of passive smoking on asthmatic children who have and who ha ve not had atopic derm atitis. Chest 1992; 101:16-8 5 McConnochie KM, Roghm ann KJ. Wheezing at 8 and 13 years: changing imp ortance of bronchiolitis and passive smoking. Ped iatr Pulmonol 1989; 6:138-46 6 Harlap S, Davies AM. Infant adm issions to hospita l and maternal smoking. Lancet 1974; 1:529-32 7 Menon P, Rand o RJ, Stankus RP, Salvaggio JE, Lehrer SB. Passive cigarette smoke-challengc studies: increase in bronchial hyperac tivity. J Allergy Clin Imm unol 1992; 89:560-66 8 Murray AB, Morrison BJ. Th e effect of cigarette smoke from the mother on bronchial responsiveness and severity of symptoms in children with asthma. J Allergy Clin lmmunol 1986; 77:575-81 9 Pedreira FA, Guandolo VL, Feroli EJ, Mella MW, Weiss IP. Involuntary smoking and incidence of respiratory illness during the first year of life. Pediatri cs 1985; 75:594-97 10 Weiss ST, Tager lB, Munoz A, Speizer FE. Th e relationship of respiratory infectio ns in early childhood to the occur rence of increased levels of bronchial responsiveness and atrophy. Am Rev Respir Dis 1985; 131:573-78 11 Chilmonczyk BA, Salmu n LM, Mega thlin KN, Neve ux LM, Palomaki GE, Knight GJ, et al. Association between exposure to environme ntal tobacco smoke and exacerbations of asthm a in children. N Engl J Med 1993; 328:1665-69 12 Meltzer S, Hovell M, Meltzer E, Atkins C, de Peyster A. The impact of parental counseling for redu cing tobacco smoke exposure in children with asthma. J Asthm a 1993; 30:391-400 13 Speigler M, Guevremont D. Contemporary beha vior therapy. Pacific Grove, Calif: Brooks Cole Publishing Co, 1993 14 Elde r J, Geller S, Hovell M, Mayer J. Motivating health beha vior. Albany, NY: Delmar Publishers, 1994 15 Hammond SK, Leaderer BP. A diffusion monitor to measure exposure to passive smoking. Environ Sci Tee 1987; 21:494-97 16 Leaderer BP, Hamm ond SK. Evaluation of vapor -phase nicotine and respirable suspended particle mass as markers for environmental tobacco smoke. Environ Sci Tee 1991; 25:770-77 17 Snedecor GW, Cochran WG. Statistical methods. 6th ed . Ames, Iowa: Iowa State University Press, 1967 18 Blalock HM Jr. Social statistics. 2nd ed. New York: McGrawHill, 1979 19 H uynh H, Feldt LS. Estimation of the box correction for degrees of freedom from sam pie da ta in randomized block and split-plo t designs. J Ed Stat 1976; 1:69-82 20 Norusis MJ. SPSSj PC +4. 0 base manual. Chicago: SPSS Inc, 1990 21 Murra y DM, O'Connell CM, Schmid LA, Perry CL. Validation of smoking by self-report by adolescents: a re-examination of bogus pipeline procedures. Addict Behav 1987; 12:7-15 Tobacco Smoke Exposure in Asthmatic Children (Hovell et al)
recent Environme nta l Prot ection Agenc y rep ort A sum marize d decad es of resear ch linking passive
smoking with negative health outcomes.' Special popul at ions, such as asthma tic child ren , ar e at elevated risk for the harmful effects of environme ntal tobacco smo ke (ETS). Environme ntal tobac co smoke is estima ted to exace rbate the symptoms of approx"From the Ce nte r for Behavioral Epide miology, Grad uate School of Public Hea lth, San Diego Stat e Universitv (Drs. Hovell, Hofstett er, Mulvihill, and Atkins; and Mss Meltzer, Zakarian, and Emerson, and Mr Wa hlgren ); the John B. Pierce Fou ndation Laboratory and Depa rtmen t of Epidemiolog y and Public Health , Yale University School of Medicin e, New Haven, Conn (Dr. Lead ere r), the Allergy and Asthm a Medical Group and Research Cen ter, San Diego, and Department of Ped iatrics, Universit y of Californi a, San Diego, School of Medicine (Dr. Meltzer); the Depar tment of Allergy , Kaiser Perm anente Medica l Cen ter , San Diego, and Departm ent of Pediatrics, Universitv of Ca lifornia , San Diego, School of Medicine (Dr. Zeiger); and"the Di vision of Asth!TIa, Allergy and Clinica l Immu nology, Shar p Rees-Stealy Medical Group , and Departm ent of Ped iatrics, University of Ca liforn ia, San Diego, School of Medici ne (Dr. O'Connor). This research was supported by funds provided by the Cigarette and Tobacco Surtax Fun d of the Stat e of Ca lifornia thr ough the Tobacco-Related Disease Research Program of the University of Ca lifornia, gran t l RT509. Manuscript received October 2 1, 1993; revision accepted Decembe r 28.
440
th e home decreased for alI groups. The experimental group attained a 79 percent decrease in children's ETS exposure, compared with 42 percent for the monitoring control and 34 percent for the usual treatment control group. Repeated-measures analysis of variance resulted in a significant (F([10,350]=1.92, p
imat ely 20 percent of thi s country's 2 to 5 million asthmatic children and is a major aggravating factor in approximately 10 percent.' Passive smoking has been positively associated with increased prevalence of asthma, with incr eased frequenc y of medication use, severity of asthma symptoms, acute exacerbations and respiratory tract inf ections, and with reduced pulmonary fun ction in asthmatic chlldre n.P'! Given th e demonstrated associations between ETS exposure and negative health effects and the high probability that asthmatic child ren are exposed to ETS in the short and long term , it is imperative that clin ical and public health int er ventions be developed to reduce ETS exposure to asthmatic child ren . This study was a randomized controlled tri al of a pr eviously deveioped -- beh avioral coun seling program for reduction of ETS exposur e in the home among asthmatic childre n living with par ent s who smoke. M ETH OD S
Design Stratified by medical center, a three-group repeated measures Tobacco Smoke Exposure in Asthmatic Children (Ho
Table I-Schedule of Measures and Counseling Sessions Across Groups Environmental Monitor
Peak FlowMeter
Child's Symptom
Paren t's Smoking
Group
FE Vl
Interview
Diar y
D iary
Counseling
Experimental/counseling Monitoring control Usual treatm ent control
X X X
X X X
X X X
X X X
X
X
X
X
X
design was used . Nin et y-on e families with at least one parent wh o smoked ciga re ttes and a child (6 to 17 years) with asthma wer e randomly assigne d to an expe rime ntal/counseling , a monitoring control, or a usual tr eatment contro l group. F amilies in eac h gro up provid ed m easur es of children 's ciga re tte smo ke exposure and health sta tus and par ents' smo king rat es at a baselin e hom e visit, at a pr einter vention clinic visit, at tw o clin ic visits during interven tion (2 months and 6 months), a nd at two clinic visits during follow -up (9 months and 12 months). Two-week monitoring intervals pr eced ed clinic visits. All families receiv ed standard medical care. Tw elve famili es dropped out of the stud y during th e l-year enr ollme n t period. Eight refused further participation, two lost custody of the asthma tic child , one moved out of the ar ea, and one child was institutionaliz ed . Thus, final analyses wer e com ple ted on 79 families. Tabl e I outlines the study measur es acco rd ing to group assignme nt. . For each family, one smokin g parent and one asth matic ch ild wer e selec ted as th e " targe t par ent " and " ta rge t ch ild ," respectively. While othe r famil y member s participated to vary ing d egree s in th e study , most measur es wer e obtained from th e target parent and child. Wh en possible, informa tion reg arding the othe r
parent in a famil y was ob tain ed directly fr om th at paren t.
Subjects Th e mean age of tar get childr en was 10.77 (50 = 2.98) yea rs. More th an two th irds (70 per cent ) were whi te , whil e 9 per ce nt wer e black , 7 percent wer e La tin o, 4 per cent we re Asian, and 10 per cent were member s of othe r eth nic classificati ons. Mor e th an half (57 percent ) of th e target child re n wer e ma le. Th e mean age of targe t par ents was 37.96 (50 = 5.82) yea rs. Seven ty -seven perce nt wer e white, 6 per cent wer e black, 6 per cent wer e Latino, 8 percent were Asian , and 3 per cent we re oth er. Th e majority (67 percen t) of target parents wer e female (Table 2 ).
Recruitment Families wer e recruited from four of the largest pe d iatric a llergy medica l ce n ters in San Diego, Scree n ing forms we re com pleted during regul ar office visits. To be included in th e study, the ch ild had to require asthma medicati on a nd to be ex posed to a t least one ciga re tte per d ay b y th eir parent. Th ose meeting th e inclusion criteria were con tac ted b y a research assista nt a nd invited to participate in a study of th e effect s of tobacco exposure
Tab le 2- Group Distrib utions f or Se lec ted Dem ographic Variables Group l,l n (%)
Group 2,1 n (%)
Group 3,1 n (%)
10 (11.0) 21 (23.1)
15 (16.5) 13 (14.3)
14 (15.4) 18 (19.8)
21 (23.1) 10 (11.0)
20 (22.0) 8 ( 8.8)
20 (22.0) 12 (13.2)
17 (18.9) 14 (15.6)
22 (24.4) 6 ( 6.i)
24 (26.7) i ( i. 8)
20 (22.2) 11 (12.2)
25 (2i .8) 3 ( 3.3)
25 (2i .8) 6 ( 6.i)
21 (23.1) 10 (11.0)
23 (25.3) 5 ( 5.5)
24 (26.4) 8 ( 8.8)
2 ( 2.2) 12 (13.2) 16 (Ii .6) 1 ( 1.1)
2 ( 2.2) 6 ( 6.6) 19 (20.9) 1 ( 1.1 )
2 ( 2.2) 10 (11.0) 15 (16.5) 5 ( 5.5)
Overall Mean (SO)
Group 1 Mean (SO)
Group 2 Mean (SO)
Group 3 Mean (SO)
10.77 (2.98) 37.96 (5.82) 2.84 (5.58)
11.16 ( 2.9i ) 39.39 ( 6.82) 3.42 ( 8.82)
10.82 ( 3.13) 3i .00 ( 5.82) 2.il ( 3.31)
10.34 ( 2.90) 37.41 ( 4.54) 2.38 ( 2.25)
Varia ble*t Child gend er Fema le Male Parent gender Female Male Child ethnicity Whil e Nonwh ite Par ent cthnicily Whit e Nonwhite Parent marit al status Marr ied Not marri ed § Parent education < High school graduate High school gra duate Some college College grad uate Variable"]
Child age Par ent age Parent num ber of lifetim e attemp ts to quit smoking
"Pearson's X2 analyses for categorica l variables and ANOVA for continuous variables showed no statistically significant gro up differences (all p>0.05). tn =91 for all varia bles except child and parent ethnicity (n = 90). [C roup I : experimental/counseling ; group 2: monitoring control; group 3: usual treatm ent control. §Marri ed includes cohabitation. Not married includes single, di vorced , separa ted, and widowed . CHEST 1106/ 21 AUGUST, 1994
441
on the hea lth of thei r child . All fam ilies wer e told that they wou ld receive incremen tal payment s totali ng $60 for attend ing clinic sessions and for provid ing information th at could help in th e unde rstand ing of th e relati onship be tween passive smo ke exposure and asthm a. T hey we re not told that th e study would invo lve an int ervention to red uce ETS exposure. All wer e provide d with a description of the possible risks and benefits and all voluntee rs signed informed consents. A sequential recruitment procedure was employed until 91 families wer e recruited ac ross all clinic sites. Th e study was approve d by th e San Diego Sta te University Com mittee on Prot ecti on of Hum an Subjects .
Procedures Experim enta l/Co unse ling Group: Both the target par ent and ch ild atte nded a series of cou nseli ng sessions over 6 m onths design ed to reduce th e child 's exposure to pa re nta l smo king. Between cou nseling sessions, d iari es wer e used during the 2-wee k monitoring periods preced ing clinic visits to record par ent smo king and child ETS exposure patterns, children 's peak flowmeter readings, and children 's sym ptoms. Th ese dat a wer e used to recommen d cha nges in parent al smoking patt ern s that could reduce ETS exposure for the child. Beh avior modificati on techniqu es such as stim ulus con trol, sha ping , per sonal fee d bac k, an d conti nge ncy contracting were inco rpo ra ted in a program tailored for eac h fa mil y.13,14 M onitorin g Control Group : To determine the d iffer ent ial effec ts of monit oring and counseling, a " monitoring con trol" gro up was utili zed. Th ese fam ilies were asked to em ploy the same moni toring measur es th at were used by th e counseling group, but neither review of records nor counseling to red uce ETS exposure was pro vided . Usual Treatmen t Control Group: Th e control gro up was asked to participat e in th e sa me seq ue ntia l clin ic visits as the oth er tw o groups, but th ey were not asked to monit or ciga re tte use, passive exposure, or child re n's sym ptoms . No counseling program was provi de d. Thi s group d id record child ren 's peak flowm eter readings.
in the home, it was not ab le to selectively m easur e th e room when the ch ild was pr esent or to sam ple the child 's exp osure to nico tine outside of the roo m. In addi tion, th e beh a vioral counseling pr ogra m was direct ed toward reducin g th e child ren's exposure, not to redu cin g tobacco use in one parti cul ar room . Th us, th e monitor was not conside red sufficiently sensitive for use as a n out com e measur e.
Statistical An alyses Using the monitorin g control and usua l tr eatm ent contr ol groups only, test-r etest reliabilit y was in vestigat ed for child re n's exposure , par ents' smo king rat es, the environme ntal monit or , and FEY I by intr aclass correlatio ns.J7 Measur es for test-r et est ana lyses we re obtained from the prei ntervention a nd 2-mo nth clinic visit inter views. To deter m ine th e validi ty of pa re nt-re ported exposure a nd smo king rat es, Ken dall 's ta u-b was used as a non-param etric test of association betw een these va ria bles and the envi ronmen ta l monitor valu e.l'' Th e valid ity of par ent s' rep ort s of their child ren 's exposure to all ciga re tte smo ke in the hom e was exam ine d using constru ct validity techniques. Associations betw een va riables having hyp oth esized positive, negative, and null relati onshi ps with th e child 's report ed exposur e to all sources of cigare tte smo ke in the hom e wer e tested with Ken dall's tau-b. All valid ity tests were perform ed for pr einter ve nt ion, 2-mo nth, a nd 6-mon th measures. T o identify d ifferen ces a mo ng groups in repor ted ETS exposure, smo king rate, th e environ me nta l m onit or value, and ch ild ren 's sym ptoms over tim e, repeated measur es ana lyses of varia nce (AND YA) was em ploye d. A priori defined pairwise AND YAs were used to detect significant d iffer en ces between grou ps. Du e to violati ons of sphe ricity , all within-subjects F statistics were com pared with ad justed critical va lues (H uynhFeldt epsilon ad justment) an d significance levels ar e report ed acco rdi ng ly.19 Unad justed deg rees of freed om a rc reported . All a priori pair wise com parisons wer e cond uc ted as one-tailed di rectional tests. Ana lyses wer e perform ed using a co mpute r soft wa re package (SPSS-PC+ ).20 R ESULTS
Measures of Tobacco Exposure
Reliabil ity and Validity
Parent Reports: The primary dependent varia bles co llected at base line a nd a t clinic inte rv iews were the child 's exposure to ETS from th e tar get par ent a nd from all person s who smo ked in th e hom e. Exposur e to parental smo king at hom e was segrega ted into two components: exposure on typi cal workdays a nd non workday s. Th e target a nd th e othe r parent (if ther e was one ) wer e asked : " How man y ciga rette s did you smo ke per workday/ nonworkda y at hom e during the past week ?" Th ey also were asked: " Of th ose ciga rettes, how man y was (child's nam e) exposed to?" Fin all y, parents were asked : " Please estima te the number of cigarettes to which (child 's name) was exposed from others living in the home and those who visited ." Answers wer e used to calculate the paren t's mean dail y smoking ra te a nd the child 's mean dail y exposure rat e to par ent's and to all smo kers' ciga rettes in the hom e. Envi ronmental M onit or: An env iron me nta l monit or was used with all thr ee gro ups to valid at e par ent-rep ort ed exposure and smoking ra tes . Th e monit or was a 37-mm -dia me te r polystyr en e cassette contai ning a teflon- coated glass fiber filter (Emfab TX 40h120WW , Pallflex Corp, Pu tn am , Co nn) satura ted with an aq ueo us solutio n of 4 percent sod ium bisulf at e a nd 5 percent etha nol a nd dried . Th e monit or is based on passive d iffusion of nicotin e to a filte r that is ana lyzed by gas chro ma tog ra ph y 15 and has bee n shown to be bo th reliab le and valid.!" F ilter preparat ion was cond ucted a t th e John B. Pier ce Labo ra tor y (Ne w Haven , Conn). The mo nitor was placed in a room whe re grea test exposure was repor ted at the init ial hom e inter view , and remain ed for the 2 week s precedin g eac h clinic visit. W hile the monitor provide d a 2-week continuous measure of nicotin e level in one room
Reliability: The test-r etest reliability coefficients for parent reports of child ren' s exposure ranged from 0.63 to 0,84 across reported measures on workdays , nonworkdays , and composite measures. The coefficients for par ents' self-reported smoking rates ranged from 0,60 to 0,88. The environme nta l monitor's test-retest reliability was 0.72. Th e test-ret est correlation for th e FEY 1 was 0.96 . Validity: Exposure to th e target par ent's smoking on a workday was significantly correlated (taub=0.15 to 0.25, p<0.05) with th e ETS monitor at each of th e three monitoring peri ods tested . The correlation coefficient between th e monitor and reported exposur e on nonworkdays was significant only at 6 months (tau-b=0.23, p
442
Tobacco Smoke Exposure inAsthmatic Children (Hovell et all
Table 3-Group Means (SD) for Smoking Rat e, Exposure Rate, and En vironmental Monitor by Visit n° Smoking rate (1\0. of ciga rettc s/ day )j
Gro up !
73 2 3
Exposure to target paren t's ciga rettes (No. of cigarettc s/ workday )§
71 2 3
Exposure to all cigarett es at home (No. cigs/ day )§
Environmental monitor, p.g / m 3
61
Baseline
Prcintervcnli on
2 mo
6 mo
9 mo
12 mo
21.2 (13.1) 20.2 (11.9) 23.4 (12.0) 5.8 (5.4) 5.5 (3.4) 8.0 (9. 1)
21.5 (15.8) 19.3 (10.3) 22.0 (11.4) 3.4 (2.6) 4.2 (2.8) 5.7 (4,8) 5.1 (4.3) 6.9 (5.6) 7.8 (5.0) 1.7 (1.6) 3.0 (3.3) 2.3 (3.0)
18.9 (17.6) 21.2 (12.9) 19.0 (12.7) I.7 (2.1) 3.2 (2.1) 5.3 (4.1) 2.0 (3.2) 6.2 (4.9) 6.6 (4.0) 1.9 (1.8) 3.7 (4.8) 2.1 (2.6)
16.8 (14.1) 17.5 (11.6) 20.4 (13.7) 1.6 (1.9) 3.8 (3. 1) 3.9 (2.7) 2.5 (4.2) 6.7 (5.9) 5.6 (3.7) 2.8 (3.4) 2.2 (1.9) 2.4 (3.5)
15.0 (14.9) 16.8 (11.5) 20.5 (14.4) 1.5 (2.2) 4.1 (2.5) 4.0 (3.2) 2.2 (4.4) 6.4 (4.0) 5.7 (5.2) 1.9 (1.8) 2.8 (3.2) 1.9 (1.9)
14.5 (13.3) 16.5 (1 1.6) 22.0 (14.7) 1.2 (1.5) 3.7 (3.0) 4.6 (3.0) 2.9 (4.8) 7.8 (6.7) 9.5 (5.1) 1.9 (1.8) 2.7 (3.6) 1.4
. . ·11
2
. . ·11
3
. . ·11
65
. . ·11
2
. . ·11
3
. ··11
(1.1)
*n sizes vary du e to missing data for some variables. [ C roup I : experimental/ counseling; grou p 2: monitorin g control; grou p 3: usual treatment control. tp <0.05 for repeated -measures AI\OVA gro up- by-lime effect. §p < O,OI for repeated -measures AI\OV A gro up-by-lime effect. IIExposur e to all cigarett es at home and environmental monitor measures not obta ined at base line.
rep ort s of smoking rat e were statistically significant (p < O.OI) and most were larger than wer e th e correlati ons with parent-report ed exposure . Th ese values ranged from 0.25 to 0.39 for workdays and fr om 0.18 to 0.29 for non workdays. Th e construc t valid ity tests for all ETS exposure in the hom e result ed in 79 percent agree me nt for all constr ucts. The binomial pr obability of this proportion of agr eem ents across tim e occur ring by chance was statistically significant (p < 0.05), dem onstrating convergent, di verg ent, and discr iminant valid ity of th e par ent-reported measures.
Random Assignment Th e three groups were com pa red for gende r, age, and ethnicity of child an d par ent; par ent s' marital status, ed ucation, lifet im e smoking q uit atte m pts, and repo rted level of smo king; and child's baseline ETS exposure . Pear son's x2 analyses for categoric variables and ANOVA for continuous variables showe d no significant diff er ences among groups, thus confirmi ng the success of random assignm ent (Table 2).
Target Paren t Smoking Levels Tar get par ents reported smo king a mean of 21.6
(SO = 12.3) cigar ettes per typi cal day at baselin e. Repeated-measures ANO VA for smoking rat e showed a significant tim e effect (F[5,350]= 7.92, p
Children' s Environmental Tobacco Smoke Exposure and Asthma Sym ptoms Th e child ren's m ean ETS exposure to th eir parents' cigarettes was 6.5 (SO = 6.6) ciga rettes per workday at th e baselin e hom e visit. Figure 1 shows th e mean exposure ra tes by group over tim e. Repeated-measures ANOV As with exposur e to the target parent's CHEST / 106/ 2/ AUGUST, 1994
443
10
BL
INTERVENTION
FOLLOW UP
8 >-
ell
C
... Ql
6 Usual-tx Controls
Co Ul
~
~
ell
.!:l
4
0
Monitoring Controls
2 Experimental Group O+-~--,r----,.--r-,---r----,-........--r-,---r---,.-...,
o
2
3
4
5
6
7
8
Month
ciga rettes in th e hom e on workdays as the dependent var iable yielde d significan t group (F[2,68]= 11.83 , p
9
10
11
12
FIGURE 1. Mean number of the target pa rent's ciga rettes to whic h childr en are exposed on workdays, over time , and for each experime nta l group .
However , at th e final I2-m onth visit , only th e expe rime ntal/counseling gro up susta ine d a decr ease (44 perce nt). The monitoring con trol gro up showed a 14 per cent increase and the usua l treatment contro l group showe d a 22 per cent increase fr om pr einter vention at 12 months. Pair wise comparisons showed th e same pattern of results as for exposure to th e target par en t' s cigarettes, with the expe rimental/cou nseling group showing a significant decr ease over th e monit oring control group (F[4,148]=3. 71, p
(F[I ,55]=4.89, p<0.05) groups reported a significant mean decrease in symptoms compared with the usual treatment control group. Differences between th e counseling and monitoring control groups did not reach significance, and no differences between groups were found after 2 months. Although no group or group-by-time differences were apparent for FEV I , pulmonary function was consistently negatively correlated with ETS exposure at all measurement points (r = -0.13 to -0.37). Wh en the relationship between F EV1 and exposure was adjusted for age and height using hierarchical multiple regr ession, th e partial correlation remained negative but was not statisticall y significant. DISCUSSION
To our knowledge, this is the first fully controlled trial of a behavioral counseling program designed to assist parents in reducing th eir asthmatic childre n's exposure to cigarette smok e. This analysis is the first to include formal tests of the reliability and valid ity of parent rep orts of cigare tte smoking and childre n's ETS exposure . Par ent reports were shown to be both reliable and valid measures of ETS exposure in th e home. Use of random assignment and demonstrably reliable and valid measures greatly increase th e internal validity of this analysis. Families in the exp erimental/counseling group attained a significant and clinically important decr ease in ETS exposure for th eir asthmatic children. This decrease was achieved not only for exposure to th e target par ents ' cigarettes, but also for exposure to all cigarettes smoked in th e hom e. Th e counseling proc edures used in thi s study are well within th e capacity of most clinical assistants found in outpatient clinics or in private practice offices and offer a potentially practical means of extending the therapeutic effectiveness of the physician. This program required littl e or non e of th e ph ysicians ' time. Th e use of the environme ntal monitor to measure ETS in a particular room of the resid enc e provided an objective means of validating th e parents' mor e subjective estimate of th eir child's ETS exposure. Validity was substantiated by formal constru ct validity testing. From th ese results it is concl uded that motivated parents can report th eir child's ETS exposure accurately enough for clinical purposes. It is important to note that their accuracy ma y have been due, in part, to the apparent ability of th e investigators to verify th eir reports by use of th e environme ntal monitor. Clinical use of these measures might in clude such a monitor, at least as a " bogus pipelin e" in order to enha nce the pot ential accuracy of parent reports.P! Also of clinical importance is the fact that no objective substitute for th e parents' reports of ETS ex-
posure could be used in th e administration of th e behavioral counseling program. The parents' reported details regarding the timing, location, and activities taking place when the child was exposed were the foundations for counseled change in lifestyle to reduce ETS exposure. Such detailed information could not be obtained by existing objective measures of ETS exposure. Thus, this intervention depends on accurate parent reporting. Although thi s study demonstrated satisfactory accuracy of parent reports, it can be assumed that parents ar e not able to det ect subtle level s of exposure. Parent reports ar e inh erently limited in th eir sensitivity, increasing th e need for development of accurate object ive measur es of ETS exposur e. To determine relationships betw een disease and ETS exposure, it may be critical to obtain estimates of more subtl e levels of ETS exposure than can be achieved by parent obser vations. Th e environme ntal monitor used in thi s study has the pot ential for a high level of sensitivity . However , it has not yet been constru cted in a sufficie ntly small and durable form to attain such measures for children . Although only a few centime ters in size, the monitor is not small enough to be unobtrusive. It is relatively fragile and subject to contamination by children 's normal pla y and curiosity, and it is not likely that children could be trained to wear th e monitor continuously in its pr esent form . Thes e limitations restrict th e monitor to mor e limited sampling strategies such as those used in th e pr esent study. By limiting th e monitor to th e room wher e th e most exposure to th e child was reported , sam piing th e child's ETS exposure was maximized . Howe ver , for many families, th e majority of the child's exposure occurred outside thi s one room . Thus, th e environmental monitor was not likely to det ect th e majority of thc child's ETS exposure or change in exposure. In addition , once counseling began , th e parents in th e couns elin g group ma y ha ve decr ea sed th eir child' s exposure by arranging for the child to wat ch television in ano the r room or by oth erwise sepa ra ting th e child from th e plac e of pot ential exposur e. For th ese famili es, th e filter no longer accurately reflected th e child's exposure, but only th e amount of tobacco use in the selected room . These sampling limitations precluded th e use of th e environme nta l monitor as an outcome measure. Th e mean monitor valu es depict relatively low ETS levels and a very small decrease across tim e for all groups, whil e neither the group nor the groupby-time effects reached significance . These result s suggest that the ETS level in th e select ed room decreased during th e cours e of study. However , no inferenc es can be drawn from these result s regarding the changes in the child re n's ETS exposure. CHEST 1 106/ 21 AUGUST, 1994
445
Exploratory analyses of sym ptoms and pulmonary function pro vided encouraging evide nce of possible therapeutic effects of th e behavioral counseling program. Children assigned to the experimental/counseling group and th e monitoring control group report ed significant decreases in overall symptoms at th e 2-month measur em en t point wh ich conclude d th e most intensive couns eling period. The decr ease was gr eat est for th e exper imental/counseling group. These results suggest that health benefits ma y be at tained from th e counse ling program and even from self-monitoring alone , and support replication with a larg er sample in order to provide sufficient power to discover oth er possible health benefits related to decreased ETS exposure among asthmatic children . Although th is analysis was not design ed to achieve smoking cessation among parents, a notable order ed diff erenc e among groups in quit rate was observed . The gr eatest number of qu its occurred in th e expe rimental/couns eling group, follow ed by th e monitoring control and the usual tr eatment control groups. Th ese d ifferenc es d id not reach significa nce. However, if th ese rat es wer e replicated with a larger sam ple, smoking cessation might be found to be an added ben efit of the program . Of course , since thes e parents reported previous quit attempts, future assessments should exam ine maintenance. The sample included in th ese analyses was attained from four asthma specialty services in San Diego , and as such provided modest heterogen eity with respect to types of medical care recei ved . Though sam ples' sizes did not permit clinic-specific analyses, th e use of th ese different services sugge sts that th ese results may be generalizabl e to similar services prov ided to working class and predominantly, although not exclusively, white fam ilies. Additional resear ch will be needed to determine wh ether similar counseling procedures are as effective with minority or lower socioeconomic status families. Finally, these ana lyses were limited to a I- year follow-up. Based on pr evious behavioral research, it is likely that th e expe rime ntal/ counseling group's reduced ETS exposure will not be sustained once the int ervention program is completely removed. The actual level of recidivism needs to be det ermined and it may be necessary to develop additional int erv entions th at could be implem ent ed by the ph ysician or clinical staff to sustain th e decreased ETS exposure att ain ed immediately following counselin g. The fact that th e expe rimental/counseling group did sustain a large decr ease in exposure at 9 and 12 months, when monitoring but no formal counseling was provided , ind icates that minimal intervention ma y be suffic ient for mai nt enance. Th ese results justif y follow-up analyses an d furt he r exploration of affordabl e techniques that could support ma int enance of th erapeu 446
tic effects . The significant reduction in ETS exposure an d th e promising results reg arding smoking cessation seen in th is sample of high-risk asthmatic families ma y hav e broad application and result in important pu blic health benefit s. REFERENCES
1 US Environme ntal Protection Agency. Respirat ory health effects of passive smoking: lung cancer and other disorders. EPAj600j6-90j006F. Washington, DC: Office of Health and Environmenta l Assessment , 1992 2 Dekker C, Dales R, Bartlett S, Brunekr eef B, Zwancnburg H. Childhood asthma and the indoor environment. Chest 1991; 100:922-26 3 Weitzman M, Gortmaker S, Walker DK, Sobol A. Maternal smoking and childhood asthma. Pediatrics 1990; 85:505-11 4 Murray AB, Morrison BJ. Effcct of passive smoking on asthmatic children who have and who ha ve not had atopic derm atitis. Chest 1992; 101:16-8 5 McConnochie KM, Roghm ann KJ. Wheezing at 8 and 13 years: changing imp ortance of bronchiolitis and passive smoking. Ped iatr Pulmonol 1989; 6:138-46 6 Harlap S, Davies AM. Infant adm issions to hospita l and maternal smoking. Lancet 1974; 1:529-32 7 Menon P, Rand o RJ, Stankus RP, Salvaggio JE, Lehrer SB. Passive cigarette smoke-challengc studies: increase in bronchial hyperac tivity. J Allergy Clin Imm unol 1992; 89:560-66 8 Murray AB, Morrison BJ. Th e effect of cigarette smoke from the mother on bronchial responsiveness and severity of symptoms in children with asthma. J Allergy Clin lmmunol 1986; 77:575-81 9 Pedreira FA, Guandolo VL, Feroli EJ, Mella MW, Weiss IP. Involuntary smoking and incidence of respiratory illness during the first year of life. Pediatri cs 1985; 75:594-97 10 Weiss ST, Tager lB, Munoz A, Speizer FE. Th e relationship of respiratory infectio ns in early childhood to the occur rence of increased levels of bronchial responsiveness and atrophy. Am Rev Respir Dis 1985; 131:573-78 11 Chilmonczyk BA, Salmu n LM, Mega thlin KN, Neve ux LM, Palomaki GE, Knight GJ, et al. Association between exposure to environme ntal tobacco smoke and exacerbations of asthm a in children. N Engl J Med 1993; 328:1665-69 12 Meltzer S, Hovell M, Meltzer E, Atkins C, de Peyster A. The impact of parental counseling for redu cing tobacco smoke exposure in children with asthma. J Asthm a 1993; 30:391-400 13 Speigler M, Guevremont D. Contemporary beha vior therapy. Pacific Grove, Calif: Brooks Cole Publishing Co, 1993 14 Elde r J, Geller S, Hovell M, Mayer J. Motivating health beha vior. Albany, NY: Delmar Publishers, 1994 15 Hammond SK, Leaderer BP. A diffusion monitor to measure exposure to passive smoking. Environ Sci Tee 1987; 21:494-97 16 Leaderer BP, Hamm ond SK. Evaluation of vapor -phase nicotine and respirable suspended particle mass as markers for environmental tobacco smoke. Environ Sci Tee 1991; 25:770-77 17 Snedecor GW, Cochran WG. Statistical methods. 6th ed . Ames, Iowa: Iowa State University Press, 1967 18 Blalock HM Jr. Social statistics. 2nd ed. New York: McGrawHill, 1979 19 H uynh H, Feldt LS. Estimation of the box correction for degrees of freedom from sam pie da ta in randomized block and split-plo t designs. J Ed Stat 1976; 1:69-82 20 Norusis MJ. SPSSj PC +4. 0 base manual. Chicago: SPSS Inc, 1990 21 Murra y DM, O'Connell CM, Schmid LA, Perry CL. Validation of smoking by self-report by adolescents: a re-examination of bogus pipeline procedures. Addict Behav 1987; 12:7-15 Tobacco Smoke Exposure in Asthmatic Children (Hovell et al)