An Unsuccessful Cotinine-Assisted Intervention Strategy to Reduce Environmental Tobacco
Smoke
Barbara A.
Exposure During Infancy
Chilmonczyk, MD; Glenn E. Palomaki; George J. Knight, PhD; Josephine Williams; James E. Haddow, MD
Objective.\p=m-\Totest a low-intensity physician's office\p=m-\ based intervention strategy using infant urine cotinine measurements, aimed at reducing infant exposure to environmental tobacco smoke. Design.\p=m-\A randomized intervention trial. Setting.\p=m-\Offices of 28 physicians (pediatricians and family practitioners) and two hospital-based clinics in the Greater Portland, Me, area. Participants.\p=m-\Infants brought to the physician by a parent (usually the mother) for an initial well-child visit. A total of 518 of these infants and their mothers consented to provide information about household smoking habits and to provide a samfor biochemical analysis. ple of the infant's urine \p=m-\ Selection Procedure. From among the consecutive sample of 518 enrolled mother/infant pairs, 103 mothers reported that they smoked 10 or more cigarettes per day, and these were randomized on an individual basis for intervention (52 pairs) or control (51 pairs). Randomization took \s=b\
association between environmental tobacco smoke An (ETS) exposure and respiratory illness in infants is of this well documented.1_4 Given the
strength care providers and information to help
now
evidence, it would be helpful if health
could have access to materials them persuade families with smokers in the household to modify their smoking behavior and thereby reduce ETS exposure to the infant. Only fragmentary published data exist, however, to indicate how successful any given in¬ tervention strategy might be.5 The present pilot study tested compliance with, and effec¬ tiveness of, a physician's office-based intervention protocol that combined a biochemical measurement of a cigarette smoke metabolite (cotinine) in the infant's urine with infor¬ mation about the consequences of ETS exposure. It also pro¬ vided recommendations about how the family can avoid ex¬ posing the infant to ETS. The protocol was designed to min¬ imize disruption to the physician's office routine and was based on a strategy that we had previously tested and found successful in pregnant women.6 The relationship between cotinine levels in the infant's urine and ETS exposure using the sample collection and assay methods employed in this study were already
Accepted for publication From the Division of
October 7, 1991.
Biometry (Mr Palomaki and
Mrs
Williams),
the Prenatal Screening Laboratory (Drs Knight and Haddow), and Clinical Immunology (Dr Chilmonczyk), Foundation for Blood Re-
search, Scarborough, Me. Reprint requests to the Foundation for Blood Research, 190, Scarborough, ME 04070-0190 (Dr Haddow).
PO Box
place at the study center when the infant's urine sample and information about household smoking habits were received. Intervention.\p=m-\The physician telephoned the mother to report the urine cotinine result and to explain its meaning. The physician then signed and sent an individualized form letter to the mother, providing specific recommendations for changing household smoking habits. Results.\p=m-\Follow-up urine cotinine measurements were obtained in 27 (52%) of 52 infants from the intervention group and in 29 (57%) of the 51 controls 2 months later. The mean log ratio of the follow-up to initial urine cotinine measurements was 6% lower in the intervention group than in the control group. This difference was not statistically
significant.
Conclusion. \p=m-\Thelow-intensity intervention strategy did significantly influence infant exposure to environmental tobacco smoke in the household. not
(AJDC. 1992;146:357-360) evaluated in
study population,7
and the cotinine found to be a reliable way to express ETS exposure. Maternal cigarette smoking is the major contributor to infant ETS absorption as defined by infant urine cotinine levels,7 and this relationship provided the rationale for selecting families to be included in the present randomized trial. our
measurements
were
PATIENTS AND METHODS In 1988, pediatricians, family physicians, and two hospitalbased clinics in the greater Portland, Me, area were invited to participate in a randomized trial of smoking intervention; 30 (81%) agreed to participate. Following informed consent, 518 mothers coming with their infants for initial well-baby immuni¬ zation completed a questionnaire and provided a urine specimen from the infant. Methodologie details of this portion of the study have been described elsewhere.7 It was explained to the mothers that the urine sample was to be analyzed to learn more about environmental exposures, but cigarette smoke exposure was not specifically mentioned. Two cotton rolls were placed in the diaper at the beginning of the of¬ fice visit. They were either removed and placed in the glass screw-top specimen vial (if wet) at the end of the office visit or mailed back by the mother if urination had not occurred. This method of urine collection in no way alters the concentration of urine constituents, such as creatinine, and the rate of evapora¬ tion of liquid is negligible.8 Approximately half of the samples were obtained during the office visit. Nine percent of the sam¬ ples were unsatisfactory for analysis, either because of stool contamination or inadequate volume, and in 81% of those cases a second urine sample was successfully obtained. Before assay¬ ing, the urine was transferred to a test tube by compressing the cotton rolls inside a disposable syringe.
Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/20/2015
Table 1.—Standardized and Nonstandardized
Components of the Randomized Intervention Trial Component
Standardized?
Data collection form
Yes Yes
Office Office
Intervention packet to physician Cover letter to physician Interpreted cotinine report Personalized letter to patient
Yes Yes Yes Yes
Project staff Project staff Project staff
Telephone call to patient Making the call Content of the call 'Occasionally, an office nurse acted
Yes No
Physician* Physician*
Informed consent
as
Name:
Agent
Date of Birth/Age:
Doe, Jane J. 27-Aug-1991
Physician: Smith
ID#: 156.00
Date of Collection: 17-Oct-1991
nurse
EXPOSURE TO TOBACCO SMOKE
nurse
Minimal
Results:
Urinary cotinine
Physician
=
415
Significant
Heavy
ng/mL
INTERPRETATION: Cotinine is a metabolic product of nicotine. This urinary cotinine value indicates VERY HEAVY EXPOSURE to environmental cigarette smoke and is associated with an increased risk of bronchitis and pneumonia in the first year of life. this Stepsriskto ofdecrease infant's exposure to cigarette smoke may lead to decreased respiratory problems.
agent.
The mothers who smoked 10 or more cigarettes per day were ran¬ domly assigned by computer on an individual basis to intervention
control groups. This group of women and infants was selected for study because the infants were likely to be most heavily exposed to ETS and, therefore, likely also to show the effects of changes in the smoking habits of those around them. The critical elements in the intervention protocol were standardized, with the exception of the content of the telephone call made to the mother by the phy¬ sician (Table 1). This latter element was left unstructured to allow for variations that might be expected in routine practice. Once a family was selected for intervention, a cotinine measure¬ ment was performed on the urine sample and a report of that co¬ tinine measurement containing an interpretation of its significance in relation to the infant's health (Fig 1) was then returned to the phy¬ sician, along with a personalized letter to the parents to be signed by the physician, outlining ways to reduce their infant's exposure to tobacco smoke. The physician also explained the results to the parents by telephone. A project physician made telephone contact with physicians to guide them through the first intervention and to verify that they had received the intervention materials. The project coordinator then monitored physician participation via repeated visits to verify that office cooperation and compliance with the in¬ or
tervention protocol was satisfactory. The letter sent to the parents recommended the following ways to avoid ETS exposure: (1) Parents and all other persons living in the home who smoke cigarettes, pipes, and/or cigars should smoke only in a room that is as far away from the infant as possible and has outide vents or, better yet, limit all smoking to out of doors. (2) Anyone who takes care of the infant should wash his or her hands after smoking, because chemicals from tobacco smoke can be transmitted to the infant from the hands. (3) If the infant is attending day care, parents should take steps to ensure that there is no smoking going on in the day care home while the infant is in attendance. (4) Finally, when parents take their infants to visit friends, they should ask that they not smoke during the visit. The letter did not stress smoking cessation. A second urine sample was requested by the project coordi¬ nator 2 months later in both the intervention and control groups. These second urine samples were sought directly from the fam¬ ily, rather than through the physician office, and, when the sample was not obtained, the project coordinator recontacted the family in an effort to maximize compliance. All urine samples from the intervention and control groups were assayed promptly, but the results from the controls were not available to investigators until the end of the study. The urine was assayed for cotinine by a competitive radioimin our laboratory, using cotinine antisemunoassay rum raised in rabbits and labeled with iodine 125, as described previously.8,9 The standard curve of the procedure ranged from 0.25 to 12.0 µg/L, with an estimated lower limit of sensitivity of 0.2 µg/L. A dose-response relationship down to 0.3 µg/L has pre¬ viously been demonstrated between serum cotinine and ETS ex¬ posure with this assay.9 The lower limit of 1.0 µg/L was selected for detecting urine cotinine, however, to allow for possible assay variability within the 0.25- to 1.0^g/L range that might result
developed
Fig
1.—A
that
was
sample report of the interpreted cotinine measurement the physician to communicate with the mother.
sent to
from differences in urine composition and pH. Consequently, all values below 1.0 µg/L in the present study were assigned a value of less than 1.0 µg/L for purposes of analysis. Within- and between-assay coefficients of variability of 5.1% and 7.8% were obtained for a quality control serum with a value of 3.7 µg/L, the midrange of typical cotinine values found in the urine of nonsmokers. Creatinine measurements were not per¬ formed in the urine samples because an earlier study demon¬ strated that use of an individual's cotinine-creatinine ratio did not improve discriminatory power significantly.10 Between-group comparisons were made with a two-sided Student's t test for continuous variables (after appropriate trans¬ formations) and 2 analysis for categorical variables using a level of significance of .05.
RESULTS The cooperation of physicians' offices with all aspects of enrollment and intervention was documented to be uniformly satisfactory throughout the study. The study coordinator noted that a physician would occasionally
experience difficulty
in
making telephone
contact. In
those cases, the office staff would persist until contact was made and then send out the letter. From among the 518 enrolled infant-mother pairs, 103 women (20%) reported smoking 10 or more cigarettes per day and thereby satisfied the criteria for inclusion in the randomized trial. Figure 2 displays infant urine cotinine levels in urine samples from those 52 intervention and 51 control infant-mother pairs and compares them with similarly obtained urine cotinine levels from 305 house¬ holds in which ETS exposure was reported not to occur. Further data on this population are available elsewhere.7 Cotinine concentration was 10 µg/L or higher in 99 (96% ) of 103 infants selected for randomization, as opposed to 27 (9%) of 305 infants with no reported ETS exposure. Table 2 compares selected characteristics of the 52 intervention and the 51 control group women. Ages of infants at en¬ rollment were similar between the two groups, as were ini¬ tial urine cotinine levels. The mothers' ages, education lev¬ els, and numbers of cigarettes smoked were also similar. In 66 (64%) of the 103 households, a second person also smoked cigarettes, usually the infant's father (91%). None of these differences was statistically significant. Because study subjects were randomized individually, the inter¬ vention and control groups were distributed throughout all participating offices and clinics. Interpreted reports of
Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/20/2015
the urine cotinine levels were transmitted to mothers in the intervention group within 10 days of urine collection, together with the informational materials. Follow-up urine samples were then sought from all of the interven¬ tion and control subjects 2 months later, and the cotinine measurements were compared with the first measure¬ ments to assess the extent to which intervention led to a decrease in absorption of ETS products. Figure 3 contains data from the 29 control (57% ) and 27 in¬ tervention (52%) infants who provided second urine sam¬ ples. Values were highly skewed in both groups, requiring that logarithmic transformation be performed prior to anal¬ ysis. Each circle in the figure represents the ratio of the co¬ tinine measurement in the follow-up sample to the initial co1000
tinine measurement multiplied by 100, and the solid circles indicate that the infant was breast-fed. The infant was clas¬ sified as being breast-fed if the mother was breast-feeding at onset of the trial. After removal of the infants who were breast-fed, the mean log ratio multiplied by 100 was 2.05 in the intervention group, compared with 2.17 in the controls. Although the log ratio was 6% lower following inter¬ vention, the difference was not significant (i 1.15, =
=.26; pooled Student's t test on log10 transformed ratios). Follow-up cotinine levels in two of the breast-fed infants in the intervention group were strikingly lower, but it was not determined whether the mothers had stopped breast-feeding in the interim. Compliance in providing follow-up urine samples was relatively low in both groups in spite of vigorous efforts from the study Table 2.—Comparison of Selected Parameters of the Mothers in the Intervention Group With Controls*
100-
M
10" " i!i«~ 1.0-
Control
Intervention
Group
Group
No. of mothers
51
52
No of households with other smokers Mean (SD) infants' age
31
35
1.8(0.7)
1.7(0.5)
23.8 (5.1)
24.7 (4.1)
11.8(1.6)
12(1.5)
19 (9)
18 (7)
at
0.1
Intervention
Control
No Household
Exposure
Fig 2.—Urine cotinine levels in the 103 infants whose mothers smoked 10 or more cigarettes per day. These infants were randomly assigned to an intervention group (52 infants) or control group (51 infants). An additional 305 infant urine cotinine levels from house¬ holds with no reported exposure to tobacco smoke are displayed for comparison. The horizontal line at 10 v-g/L represents the demarcation point chosen to define significant environmental to¬ bacco smoke exposure.
mo
level, µg/Lt
38 (36)
*AII of the women smoked 10 tGeometric mean (log10 SD).
cigarettes
44 (0.44) or more
per
day.
1000-
1000-
ico-
intervention,
Mean (SD) mothers' age, y Mean (SD) mothers' education, y Mean (SD) No. of cigarettes smoked by mother per day Infant cotinine
&o
vP^0o0
_8_CL · o
o
uO#
"S
Intervention
Group " -
100
Initial Urine Cotinine,
\ig/L
I 1 I I I
j100
Initial Urine Cotinine,
500
pg/L
Fig 3.—Relationship between the initial and follow-up urine cotinine measurements in infants whose mothers were advised to avoid ex¬ posing them to environmental tobacco smoke and in control infants. Each circle represents the ratio of the follow-up to the initial cotinine measurement multiplied by 100. A logarithmic scale was used because of the wide range and skewness of the cotinine distribution. The horizontal line at 100 represents no change in cotinine concentration between the initial and follow-up measurements. Circles above the line indicate higher concentrations in the follow-up sample; circles below the line, lower concentrations; and solid circles, infants who were breast-fed.
Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/20/2015
Table 3.—Selected Parameters of the Mothers Who Complied With the Request for a Follow-up Urine Sample Compared With Those Who Did Not Comply
Compilers
Noncompliers
No. of mothers
56
47
No. of households with other smokers Mean (SD) infants' age at
39
28
intervention,
mo
Mean (SD) mothers' age, y Mean (SD) mothers' education, y Mean (SD) No. of cigarettes smoked by mothers per day Infant cotinine
level, µ& *
'Geometric
mean
1.84(0.60)
1.64(0.57)
24.9 (5.0)
23.4 (4.1)
12.3(1.6)
11.3(2.0)
17.7 (7.8)
20 (8.0)
41 (0.42)
41
(0.38)
(log10 SD).
center, and, therefore, a comparison of the variables was again carried out, this time between the 56 compilers and 47 noncompliers (Table 3). The two groups appeared similar except for the mothers' levels of education, which was lower in the group of noncompliers (P= .01). COMMENT The present intervention protocol provided the mother with a laboratory measurement that underscored actual of tobacco smoke products into the infant's absorption and offered recommendations for reducing ETS system exposure that might realistically be carried out without smoking cessation. This strategy was not successful in al¬ tering family behavior sufficiently to affect subsequent urine cotinine levels significantly. This contrasts with a similar study carried out by us during pregnancy in which the mother was provided with an interpreted measure¬ ment of her serum cotinine level and urged to stop smok¬ ing.6 A significant percentage of those pregnant women complied, as reflected in the subsequent serum cotinine levels and an increase in the mean birth weight of their infants. In retrospect, it is possible that smoking cessation (at least for the mother) should have been one of the goals of the present study as well. A mother's smoking con¬ tributes more to her infant's ETS absorption than any other household source,711 and the extent to which she alters her smoking habit is therefore likely to have the greatest impact on reducing ETS absorption. One published study that attempted to reduce infant exposure to ETS by suggesting parental smoking cessa¬ tion was, however, unsuccessful.5 In the event that the mother cannot stop smoking but still wishes to avoid ex¬ posing her infant to ETS, it may be necessary not only to smoke in a separate room but actually to smoke in a space that is separately ventilated.11 One drawback to requiring the mother to smoke in a separate location from the infant might be that the infant is left unattended, and it might be counterproductive to encourage this type of behavior. That concern provides further support for protocols aimed at smoking cessation, rather than relocation. The present pilot study was small but provided insights that may be useful in future activities involving smoking reduction. The relatively low compliance with providing follow-up urine samples, for example, might be largely
corrected if physicians' offices were to serve as the collec¬ during follow-up well-child visits. Even with the low compliance in the present study, however, it is unlikely that exclusion bias would mask a true impact of intervention. Characteristics of those who complied were similar to those of the noncompliers. Assuming full par¬ ticipation, the power of this study was designed to be sufficient to detect a between-group difference in the co¬ tinine levels of 10% (as measured with the log ratio). Even with the reduced participation, as defined by the propor¬ tion of second samples provided, the data were adequate to indicate that the response to intervention was poor. One possible explanation of this study's negative find¬ ings might be that participating physicians increased their counseling of all families about ETS exposure as a result of heightened awareness. By randomizing the families within physician practices, however, the study could test the impact of this low-intensity intervention over and above whatever the background office practice might be. It is unlikely that intervention families would receive less of the office's standard counseling than control families. The randomization took place at the study center only after the office visit was complete. In addition, no actual reduc¬ tion in urine cotinine concentration occurred in either of the groups, suggesting that there was little overall impact. Motivation to stop smoking during pregnancy and to avoid exposure to the fetus appears higher than motivation to stop smoking or to alter location of smoking after preg¬ nancy to avoid exposure to the infant. This might be due to a lack of public awareness that the infant's health is adversely affected by ETS. Our study indicates that more effective in¬ tervention strategies will need to be developed if the atmo¬ spheric environment is to improve for infants in households in which the mother or another household member smokes. tion center
This research
was
funded
Foundation, Falmouth, Me.
by grant
4-64515-06 from the Davis
The authors thank the many pediatricians and family physicians and the hospital clinic and their office staffs in the Greater Portland area who contributed much time and effort in patient enrollment for this study. References
Colley JRT, Holland WW, Conkhill RT. Influences of passive parental phlegm on pneumonia and bronchitis in early smoking andLancet. childhood. 1974;2:1031-1034. 2. Fergusson DM, Harwood W, Shannon FT. Parental smoking and respiratory illness in infancy. AJDC. 1980;55:358-361. 3. Rantakallio P. Relationship of maternal smoking to morbidity and mortality of the child up to the age of five. Acta Paediatr Scand. 1.
1978;67:621-631. 4. Yue Chen, Warxian Lu, Shurylang Yu. Influence of passive smok-
ing on admissions
for
respiratory
illness in
early childhood.
Br Med J.
1986;293:303-306. 5. Woodward A, Owen N, Grgurinovich N, Griffith F, Linke
H. Trial of intervention to reduce passive smoking in infancy. Pediatr Pulmonol.
1987;3:173-178. 6. Haddow JE, Knight GJ, Kloza EM, Palomaki GE, Wald NJ. Cotinine-
assisted intervention to reduce smoking and low birthweight delivery. Br JObstet Gynaecol. 1991;98:859-865. 7. Chilmonczyk BA, Knight GJ, Palomaki GE, Pulkkinen AJ, Williams J, Haddow JE. Environmental tobacco smoke exposure during infancy. Am JPublic Health. 1990;80:1205-1208. 8. Ahmad T, Vickers D, Campbell S, Coulthard MG, Pedle RS. Urine collection from disposable nappies. Lancet. 1991;338:674-676. 9. Knight GJ, Wylie P, Holman MS, Haddow JE. Improved I-125 radioimmunoassay for cotinine by selective removal of bridge antibodies. Clin Chem. 1985;31:118-121. 10. Knight GJ, Palomaki GE, Lea DH, Haddow JE. Exposure to environmental tobacco smoke measured by cotinine 125I\x=req-\ radioimmunoassay. Clin Chem. 1989;35:1036-1039. 11. Greenberg RA, Bauman KE, Glover LH, et al. Ecology of passive smoking by young infants. J Pediatr. 1989;114:774-780.
Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/20/2015