A B S T R A C T

Background: Studies examining the adverse effects of smoking during pregnancy commonly use maternal reports. We hypothesized that if an adverse event occurred during pregnancy, women may underreport smoking. This study looked for bias in maternal report of smoking if fetal distress occurs. Methods: Data were collected prospectively from patients attending The MotheRisk Program who smoked during pregnancy, and were categorized by delivery outcome, maternal and neonatal characteristics, and the raw number of cigarettes smoked per day during pregnancy reported at clinic and at follow-up. The difference between these two values was compared. Results: 95 women had uneventful deliveries and 25 had fetal distress. Women who reported fetal distress decreased their report of smoking after delivery compared to their original report during pregnancy, whereas women with an uneventful labour did not (p=0.04). Conclusions: Our results suggest that if an adverse pregnancy outcome occurs, mothers may tend to underreport their cigarette consumption.

Bias in Maternal Reports of Smoking During Pregnancy Associated With Fetal Distress
Matthew Wong, MSc,1,2 Gideon Koren, MD 1 -4

Contexte : Les tudes des effets nfastes du tabac pendant la grossesse emploient communment les dclarations des mres. Nous avons suppos quen cas de raction adverse durant la grossesse, les femmes pourraient avoir tendance minimiser le degr de leur tabagisme. Notre tude visait dtecter un biais dans les dclarations, par les mres, de leur utilisation du tabac en cas de dtresse ftale. Mthode : Nous avons recueilli des donnes des fins prospectives auprs de patientes du programme MotheRisk qui fumaient pendant leur grossesse, puis class ces donnes en fonction du rsultat de laccouchement, des caractristiques de la mre et du nouveau-n, ainsi que du nombre de cigarettes fumes par jour pendant la grossesse, selon les dclarations des mres la clinique et lors du suivi. Nous avons ensuite compar les carts entre ces deux derniers chiffres. Rsultats : 95 femmes ont signal un accouchement sans problme, et 25 un accouchement avec dtresse ftale. Ces dernires ont fait tat dune moindre utilisation de la cigarette aprs laccouchement que durant la grossesse. Pour les femmes ayant accouch sans problme, le degr de tabagisme dclar est rest le mme. Conclusions : Nos rsultats suggrent que lorsquune raction adverse se manifeste lissue dune grossesse, les mres auraient tendance minimiser leur degr de tabagisme. MARCH APRIL 2001

Studies examining the potential adverse fetal effects of smoking during pregnancy commonly rely on maternal self-reports. In the event of negative results, researchers often cite underreporting as a potential source of bias.1 This can have a profound effect upon such studies, particularly evident when maternal reports are compared with a biological marker. Studies comparing biochemical markers of smoking, such as serum or urine cotinine levels, with maternal report of smoking found that between 5 and 15% of women who identified themselves as nonsmokers had levels consistent with active smoking.2-4 The use of biological markers, however, is not always feasible for several reasons including reluctance of patients to give consent, which itself can introduce bias, and in the case of retrospective studies where only patient records are available. Moreover, the existing studies, 2-4 while identifying a reporting bias of smoking dichotomously (i.e., yes versus no), were not designed to address a bias in the number of cigarettes reported. At least two reasons for underreporting have been suggested. First, because of diminishing social acceptance of smoking during pregnancy, mothers may provide false reports of their smoking, believing
1. The Faculty of Pharmacy, University of Toronto, Toronto, ON 2. The Division of Clinical Pharmacology/ Toxicology, The Hospital for Sick Children, Toronto 3. The MotheRisk Program, The Hospital for Sick Children, Toronto 4. The Departments of Pediatrics, Pharmacology, and Medicine, University of Toronto Correspondence and reprint requests: Gideon Koren, The Division of Clinical Pharmacology and Toxicology, The Hospital for Sick Children, 555 University Ave., Toronto, ON, M5G 1X8, Tel: 416813-5781, Fax: 416-813-7562 Supported in part by the Medical Research Council of Canada, and the Seed Grant Program, The Hospital for Sick Children.

them to be more socially acceptable to the researchers.5 Second, in the instance of an adverse event during pregnancy or delivery, women may underreport a behaviour that they associate with these adverse events. This has never previously been demonstrated empirically. It is well accepted that smoking during pregnancy puts the fetus at an increased health risk. Studies have clearly illustrated the increased risk for lower birthweight, perinatal mortality and delivery complications.6,7 These risks are often explained to women who smoke during their pregnancy by health professionals, and these women are often cited as being aware of these risks prior to counselling.8 The objective of the present study was to investigate whether there is bias in maternal report of the amount of cigarette consumption if fetal distress is diagnosed. Fetal distress is described as a cluster of clinical situations, including oxygen deprivation (e.g., presence of meconium, low Apgar score), heart rate abnormality or biochemical disturbances (e.g., fetal acidemia). 9-11 Fetal distress presented at birth often requires medical interventions, and these infants may suffer long-term sequelae.12 Although fetal distress is lacking an accurate clinical definition, most mothers are aware of it and report it. METHODS The original data on smoking status were collected prospectively from patients seen between 1988 and 1997 in clinic by a physician through The MotheRisk Program (The Hospital for Sick Children, Toronto, Ontario, Canada) a counselling service for women with medicinal, chemical, illicit drug or other exposures in pregnancy. The majority of patients are from a middle to upper socioeconomic category,

CANADIAN JOURNAL OF PUBLIC HEALTH 109

BIAS IN MATERNAL REPORTS OF SMOKING

TABLE I Additional Information Provided by Mothers Reporting Fetal Distress

Comment No. of Occurrences Emergency Caesarian Section Performed 6 Babys Heart Rate was Low 5 Cord Around Neck 4 Babys Heart Rate Elevated 3 Induced Labour 3 Baby Admitted to NICU 2 No Additional Comments* 2 Decreased Oxygen to Fetus 1 Presence of Meconium 1 Baby Aspirated Meconium 1 Baby was Blue 1 Baby Received Oxygen 1 * 2 women did not elaborate on the circumstances of Fetal Distress Not mutually exclusive

representing the diverse ethnic background found in the greater Toronto area. During the clinic visit, maternal characteristics including age, gravidity, parity, and previous spontaneous or therapeutic abortions were collected as were details of underlying medical condition and previous pregnancy outcomes. Detailed reports of patient exposures during pregnancy were made ascertaining time of exposure, dose and frequency of use, where applicable. These included exposures that the patient had come to clinic for specifically, as well as other exposures including cigarettes, alcohol and illicit drugs. After collection of all data, patients were explained the concept of a baseline risk that exists in every pregnancy for an infant to have a major birth defect without any teratogenic exposure. Through critical evaluation of the current medical literature, patients were then informed of the potential risk (if any) to the fetus as a result of these exposures. Treatment for any substance use issues, or other therapeutic intervention occurs through referral, or direction back to the primary physician. Documentation of the counselling with specific literature references is then forwarded to the physician caring for the woman and also to the patient directly if requested. As part of the research program, patients seen in clinic may be followed up with a telephone interview within two years after their clinic visit to confirm exposure details and to inquire about pregnancy outcome. For the present study, patients were selected based on three criteria: 1) Live
110

birth with completion of the follow-up interview, 2) documented delivery details, and 3) documented details of maternal smoking behaviour in both clinic and follow-up files (a value of zero could occur in either the clinic or follow-up file, but zero in both files by definition was a nonsmoker). Exclusion was based on documentation in the files of intentional decrease or cessation of smoking (where the patient record documented an alteration in behaviour, e.g., cut down or quit/quitting), or where details of smoking were not complete for clinic or followup information (e.g., 16 cigarettes per day at clinic vs. response of only yes at follow-up). During follow-up interviews, patients were asked if there was an event of Fetal Distress during delivery and the interviewer gave examples summarized in Table I. The mothers response to this question was documented in a narrative. Patients were categorized into one of two groups based on their report of Fetal Distress or Uneventful Delivery. These two groups were then compared regarding maternal characteristics at clinic, number and nature of exposures (teratogenic, unknown, nonteratogenic), use of alcohol or illicit drugs, presence of maternal illness, and incidents of fetal distress in a previous delivery. Neonatal characteristics at follow-up including gestational age, birthweight and presence of major malformations, and time between clinic visit and follow-up were also compared. Comparisons between the two groups were done using the Students t-test, MannWhitney Rank Sum Test or Chi-square as appropriate. The primary endpoint of interest was the difference in the reported daily number of cigarettes in the first trimester between the first interview (real time) versus the second (post partum) interview. The raw number of cigarettes smoked per day reported at clinic, at follow-up and the difference between the two values (follow-up value minus clinic value) were compared between women reporting Fetal Distress versus Uneventful Delivery using the Mann-Whitney Rank Sum Test. The numerical difference in number of cigarettes per day reported in the second versus the first interview was then categorized as

having increased (positive value), remained the same (zero value), or decreased (negative value) and were compared using the Chi-squared test. RESULTS From 2,432 patients seen in clinic, 379 patients were identified with smoking reported, and 205 were selected for research-related follow-up. Seventeen patients refused the follow-up interview and 56 were lost to follow-up. In total, 132 cases remained which met the inclusion criteria. However, in 7 cases patients expressly mentioned that they had quit, and in 2 cases mentioned successfully decreasing their smoking. In 2 cases data were not sufficiently quantified to be analyzed and in 1 case the patient told the interviewer that she had memory problems. After exclusions, there were 120 eligible cases. Of these 120 women, 95 reported an uneventful delivery and 25 reported an event following the definition of Fetal Distress. These women provided details of the events in narratives summarized in Table I. Maternal characteristics at clinic between women who had an uneventful delivery and those who had Fetal Distress were not significantly different (Table II), including maternal age, gravidity, parity, and spontaneous or therapeutic abortions. The mean number of exposures to teratogenic or nonteratogenic agents did not differ. Use of illicit drugs or alcohol during pregnancy also did not differ between the two groups, nor did the proportion of women who had a chronic illness, including psychiatric illness. Finally, there were no significant differences in the proportion of women who had an event of Fetal Distress during a previous pregnancy. Other than the selection criteria of Fetal Distress, pregnancy outcome characteristics at follow-up also did not differ significantly between the two groups as shown in Table III. There was no difference between the two groups in the proportion of infants born with major malformations, in gestational age at birth, or birthweight. In addition, there was no significant difference in the number of months that had elapsed between clinic
VOLUME 92, NO. 2

REVUE CANADIENNE DE SANT PUBLIQUE

BIAS IN MATERNAL REPORTS OF SMOKING

TABLE II Maternal Characteristics at Clinic

Uneventful Delivery n=95 Maternal Age* 29.14.9 Gravidity 2 [1-7] Parity 1 [0-4] SA 0 [0-6] TA 0 [0-2] No. Exposures for Clinic 1 [1-7] No. Exposures Teratogenic 0 [0-1] No. Exposures Unknown 0 [0-3] No. Exposures Nonteratogenic 1 [0-7] Use of Illicit Drugs 11 (11.6%) Use of Alcohol 37 (38.9%) Maternal Illness 72 (75.8%) Psychiatric Illness 27 (28.4%) Prev. Fetal Distress 0 (0%) * Values expressed as mean S.D. Values expressed as median with range Students t-test Mann-Whitney Rank Sum Test Chi-squared Fetal Distress n=25 31.04.9 1 [1-6] 0 [0-5] 0 [0-2] 0 [0-2] 2 [1-8] 0 [0-1] 0 [0-2] 1 [0-8] 3 (12%) 14 (56%) 18 (72%) 6 (24%) 1 (4%) P value 0.09 0.16 0.12 0.82 0.85 0.61 0.99 0.65 0.81 0.77 0.19 0.90 0.85 0.47

TABLE III Follow-up Characteristics

Major Malformations Gestational Age (wk) Birthweight (g)* Time to Follow-up (mos) * Uneventful Delivery n=95 3 (3.2%) 40 [35-42] 3271546 16 [7-35.5] Fetal Distress n=25 0 (0%) 40 [34-42] 3250565 20 [2-36.5] P value 0.86 0.50 0.86 0.31

(p=0.32) in the proportion of mothers who at follow-up declared themselves to be non-smokers (decreased to zero cigarettes). There was however a statistically significant difference between the two groups in the change of report for cigarette consumption during pregnancy at clinic versus follow-up after pregnancy. That is, mothers who experienced Fetal Distress in their babies reported significantly less smoking during pregnancy at follow-up than during their initial clinic visit. This difference was categorized, shown in Table V. The two groups were significantly different in terms of the change in the report of number of cigarettes per day at follow-up with respect to whether it had increased, remained unchanged or decreased. These results indicate that mothers who had events of fetal distress during delivery were significantly more likely to decrease their subsequent report of smoking during pregnancy compared to mothers who had uneventful deliveries, who did not change their reports. DISCUSSION The problem of underreporting in epidemiological research poses a threat to the validity of a study. Our results suggest that in studies of adverse outcomes during pregnancy, mothers tend to underreport their smoking. Because the study groups had similar characteristics and outcomes, it is highly probable that the adverse pregnancy outcome led to the reporting bias. Moreover, women with uneventful pregnancies did not change their reported number of cigarettes (median change=0), indicating that because of the chronic and stable nature of smoking, there is no problem of recall per se.13 The event of fetal distress may have in fact improved maternal recall. In a study examining bottled water consumption during pregnancy, women who had spontaneous abortions had a more accurate recall of their water intake when compared to women with uneventful outcomes. 14 Again, this is not an issue of recall per se, but rather of recall bias. The present study has two major advantages over previous attempts to characterize reporting bias: This was a prospectively

Values expressed as mean S.D. Values expressed as median with range Students t-test Mann-Whitney Rank Sum Test Chi-squared

TABLE IV Maternal Self-Report of Smoking

Uneventful Delivery n=93 No. Cig/d reported at Clinic No. Cig/d reported at Follow-up Difference in No. Cig/d reported at Follow-up vs. Clinic 10 [0-40] 10 [0-40] 0 [-24-+30] Fetal Distress n=25 10 [0-30] 10 [0-25] -4 [-10-+15] P value 0.28 0.71 0.04

Values expressed as median with range Mann-Whitney Rank Sum Test

TABLE V Changes in Self-Report of Maternal Smoking

Change in No. Cig/d Reported at Follow-up vs. Clinic Increased Same Decreased Chi-squared Uneventful Delivery n=95 24 (25.3%) 34 (35.8%) 37 (38.9%) Fetal Distress n=25 5 (20%) 3 (12%) 17 (68%) P value

0.02

visit and follow-up between the two groups. Details of maternal self-report of smoking are shown in Table IV. There was no
MARCH APRIL 2001

significant difference in the number of cigarettes smoked per day during pregnancy reported at clinic between the two groups. There was also no significant difference

CANADIAN JOURNAL OF PUBLIC HEALTH 111

BIAS IN MATERNAL REPORTS OF SMOKING

collected cohort, and at the time of the initial interview, women had significant incentives to be very open about their smoking habits. Also, assessment of changes in the number of reported cigarettes was possible. A potential limitation of this study is generalizability. This research examined differences in reports relative to an initial reported value. Reliability of the initial report, and the extent of subsequent underreporting may differ between populations. Patients attending the MotheRisk program are highly motivated. It is assumed that the majority of these women provide complete and accurate information, in order to receive a more accurate risk assessment. In a population with no incentive for accurate reporting, the initial report could be less reliable. The extent of underreporting occurring after fetal distress may depend on a number of variables. Maternal guilt may result in an attempt to conceal behaviour, or may encourage complete divulgence. With our data, some of the incentives for maternal truthfulness may be lost postpartum, and it is unknown whether the postpartum report in our study population would be more or less reliable than a report in another population. The results of this study reinforce the need to obtain biological markers of expo-

sures during pregnancy. However, while biological markers can help distinguish smokers from nonsmokers, they may not be adequate for the detection of changes in consumption, as nicotine undergoes pharmacokinetic changes during pregnancy.15 Further studies in the area of underreporting should be undertaken to determine if there is some predictive value that can be gained from these results. Many studies categorize smoking behaviour as light versus heavy, based on a value of 10 or more cigarettes per day. Four cigarettes per day could indeed affect that categorization and bias study results. If there is a consistent pattern of underreporting or determinants of underreporting, this would be important information in an attempt to improve the understanding of the maternal fetal toxicology of tobacco smoke. REFERENCES
1. Jedrychowski W, Whyatt RM, Cooper TB, et al. Exposure misclassification error in studies on prenatal effects of tobacco smoking in pregnancy and the birth weight of children. J Exposure Analysis & Environment Epidemiol 1998;8(3):347-57. 2. Klebanoff MA, Levine RJ, Clemens JD, et al. Serum cotinine concentration and self-reported smoking during pregnancy. Am J Epidemiol 1998;148(3):259-62. 3. Murray RP, Connett JE, Lauger GG, Voelker HT. Error in smoking measures: Effects of intervention on relations of cotinine and carbon monoxide to self-reported smoking. The Lung Health Study Research Group. Am J Public Health 1993;83(9):1251-57.

4. Walsh RA, Redman S, Adamson L. The accuracy of self-report of smoking status in pregnant women. Addictive Behav 1996;21(5):675-79. 5. Welte JW, Russell M. Influence of socially desirable responding in a study of stress and substance abuse. Alcoholism, Clinical & Experimental Res 1993;17(4):758-61. 6. DiFranza JR, Lew RA. Effect of maternal cigarette smoking on pregnancy complications and sudden infant death syndrome. J Fam Practice 1995;40(4):385-94. 7. Walsh RA. Effects of maternal smoking on adverse pregnancy outcomes: Examination of the criteria of causation. Human Biology 1994;66(6):1059-92. 8. Haslam C, Draper ES, Goyder E. The pregnant smoker: A preliminary investigation of the social and psychological influences. J Public Health Med 1997;19(2):187-92. 9. Hill LM. Diagnosis and management of fetal distress. Mayo Clinic Proceedings 1979;54(12):78493. 10. Parer JT, Livingston EG. What is fetal distress? Am J Obstet Gynecol 1990;162(6):1421-25. 11. Mead M. The diagnosis of foetal distress: a challenge to midwives. J Adv Nurs 1996;23(5):97583. 12. Gilstrap L, Leveno KJ, Burris J, et al. Diagnosis of birth asphyxia on the basis of fetal pH, Apgar score, and newborn cerebral dysfunction. Am J Obstet Gynecol 1989;161:825-30. 13. Feldman Y, Koren G, Mattice K, et al. Determinants of recall and recall bias in studying drug and chemical exposure in pregnancy. Teratology 1989;40(1):37-45. 14. Neutra RR, Swan SH, Hertz-Picciotto J, et al. Potential sources of bias and confounding in environmental epidemiologic studies of pregnancy outcomes. Epidemiology 1992;3(2):134-42. 15. Seaton MJ, Vesell ES. Variables affecting nicotine metabolism. Pharmacology & Therapeutics 1993;60(3):461-500. Received: December 1, 1999 Accepted: September 7, 2000

Tobacco, from page 89 of the tobacco industry in orchestrating the smuggling that ensued have now been brought to light. Lack of progress on cigarette taxes and other key tobacco control measures, such as increased restrictions on smoking in public places and workplaces, is due to industry lobbying, lack of public concern, political ideology, and other political factors. While the role of the state in public health and some of the concerns raised by Fischer and Rehm are worthy of further debate, we hope that such debate will not delay the implementation of measures that are known to be effective. Litigation can
112

play an important role in holding the tobacco industry accountable for its contribution to the continuing epidemic of tobacco-related disease and death. The opinions expressed in this article are those of the authors and not of their respective institutions. REFERENCES
1. Hobbs FM, Pickett W, Ferrence RG, et al. Youth smoking in Ontario 1981-1997: A cause for concern. Can J Public Health 1999;90:80-82. 2. DiFranza JR, Rigotti NA, McNeill AD, et al. Initial symptoms of nicotine dependence in adolescents. Tobacco Control 2000;9:313-19. 3. Colby SM, Tiffany ST, Shiffman S, Niaura RS. Are adolescent smokers dependent on nicotine? A

review of the evidence. Drug and Alcohol Dependence 2000;59 (Suppl 1):S83-S95. 4. Single E, Rehm J, Robson L, Truong MV. The relative risks and etiologic fractions of different causes of death and disease attributable to alcohol, tobacco and illicit drug use in Canada. CMAJ 2000;162:1669-75. 5. Single E, Robson L, Xie X, Rehm J. The economic costs of alcohol, tobacco and illicit drugs in Canada, 2000. Addiction 1998;93:991-1006. 6. Ashley MJ, Boadway T, Cameron R, et al. Actions will speak louder than words: Getting serious about tobacco control in Ontario. A Report to the Minister of Health from her Expert Panel on the Renewal of the Ontario Tobacco Strategy. /Les actes sont plus loquents que les mots : un plan dattaque au tabagisme en Ontario. Rapport prsent la ministre de la sant par son comit dexperts sur la relance de la stratgie antitabac de lOntario. Toronto, Canada: Expert Panel on the Renewal of the Ontario Tobacco Strategy: February 1999. ISBN 0-9686913-0-7 (http:www.camh.net/otru).

REVUE CANADIENNE DE SANT PUBLIQUE

VOLUME 92, NO. 2