CLINICAL RESEARCH. Accepted: June 2003

Transcription

1 CLINICAL RESEARCH Comparison of Self-Reported Smoking and Urinary Cotinine Levels in a Rural Pregnant Population Geraldine R. Avidano Britton, JoAnne Brinthaupt, Joyce M. Stehle, and Gary D. James Objectives: (a) to assess the accuracy of selfreported smoking status in pregnant women from rural and small metropolitan statistical areas who stated they were smokers at the onset of pregnancy, (b) to compare the characteristics of these women who selfreported a nonsmoker status at the first prenatal visit with those who reported a smoker status, and (c) to determine the characteristics that predict self-reported smoking status and positive/negative urinary cotinine assays. Setting: Seven private obstetric offices serving rural and small metropolitan statistical areas in upstate New York. Participants: A convenience sample of 94 pregnant women who stated they were smokers at the onset of their pregnancies. Their mean age was 23 years and mean level of education was 11.9 years; 95% were White, 65% were single, and 65% were Medicaid-funded. Design: Descriptive correlational design. At the first prenatal visit, the sensitivity and specificity of smoking and nonsmoking status were determined by comparing self-reports of smoking status with urinary cotinine assays. Data were also analyzed for relationships among demographic variables and for predictors of smoking status and urinary cotinine. Results: The discordance rate between selfreports of smoking status and urinary cotinine assays at the first prenatal visit for the total sample was 16.6%, significant at p <.001, chi-square = 27.80, df = 1. Based on biochemical assays of > 200 ng/ml of cotinine indicating active smoking, 34.7% of women who denied smoking (specificity of 65.3%) and 10.4% of women who stated that they smoked (sensitivity of 89.5%) inaccurately reported their status (significant at p <.001). The number of cigarettes smoked per day was positively correlated with age, gravidity, parity, and number of smokers in the household. Gravidity and the number of smokers in the household were significant predictors of positive self-report of smoking status and of positive urinary cotinine assay. Conclusion: These results substantiate the unreliability of self-report on smoking status in the pregnant population and in women who are recent quitters. Such findings have implications for clinical practice, such as (a) changes are necessary in how the prenatal interview assesses past and present tobacco use, (b) pregnant smokers who are multigravidae and who live with other smokers need more tailored interventions, and (c) more research is needed on how selfreport can be ethically and efficiently validated. JOGNN, 33, ; DOI: / Keywords: Cotinine Pregnancy Rural Selfreport Smoking cessation Accepted: June 2003 Epidemiologic studies of smoking status and adverse pregnancy outcomes often rely solely on maternal self-report. Likewise, in clinical practice, smoking cessation interventions are targeted only to pregnant women who verbally present themselves as current smokers. Previous research, however, has demonstrated varying degrees of concordance between self-report and various biochemical markers of nicotine intake. The preferred biochemical marker of smoking status validation is the measurement of cotinine in human fluids, such as urine (Markovic et al., 2000; 306 JOGNN Volume 33, Number 3

2 O Connor et al., 1992), blood (Ford, Tappin, Schluter, & Wild, 1997), and saliva (Boyd, Windsor, Perkins, & Lowe, 1998; Gebauer, Kwo, Haynes, & Wewers, 1998), with higher concordance rates for self-reported smoking noted in the analyses of urine samples (Markovic et al., 2000). Cotinine is a major metabolite of nicotine and is considered the best measure of nicotine consumption (Perez-Stable, Benowitz, & Marin, 1995). It has a half-life of approximately 20 hours, superior sensitivity and specificity, and is not altered by environmental variables (Albrecht, Reynolds, Salamie, & Payne, 1999). Sensitivity is defined as the probability of positive test results confirming self-reported positive smoking status, whereas specificity is the probability of negative test results confirming self-reported nonsmoking status (Shaffer, Eber, Hall, & Vanderbilt, 2000). A meta-analysis of 26 published reports of 51 comparisons in nonpregnant populations suggests that selfreports of smoking have high levels of sensitivity and specificity (Patrick et al., 1994). Similar results have also been obtained in a study of pregnant women (Klebanoff, Levine, Clemens, DerSimonian, & Wilkins, 1998) in which 94.9% of women who denied smoking and 87.0% of women who stated they smoked reported their status correctly (p. 259). Of note, however, is that the latter study was conducted on data collected between 1959 and 1966, before public awareness campaigns especially directed to pregnant women (Markovic et al., 2000). Studies of more recent data have questioned the validity of self-report regarding smoking status in this population and reported significant misclassification rates, with sensitivity values ranging from 86.2% to 86.5%, and specificity values from 62% to 92.6% (average value: 75.2%) (Albrecht et al., 1999; Boyd et al., 1998; Ford et al., 1997; Markovic et al., 2000; Perez- Stable et al., 1995; Walsh, Redman, & Adamson, 1996). The highest discrepancy in specificity was found in a study of pregnant adolescents (Albrecht et al., 1999). The integrity of self-reported data, therefore, varies according to population and the social context in which the data are collected. Four factors influence the individual patient s responses to questions about smoking status: characteristics of the individual respondents, the method and setting of the encounter, cognitive demands imposed by the question, and the motivation of the respondent as mediated by the social desirability of the subject of inquiry (Shaffer et al., 2000, pp ). The phenomenon of smokers misclassifying themselves as nonsmokers has given rise to the terms self-deception, deceivers, and discordance. As evidenced in the aforementioned studies, the self-deception rate among pregnant women is higher than the general population. It is also higher in patients who are former smokers or trying to quit and is influenced by the intensity of the antismoking environment (Ford et al., 1997). With increasing education regarding the hazards of smoking during pregnancy, patient denial of smoking may represent a socially desired response rather than true nonsmoker status. This study demonstrates that self-report of smoking status among a pregnant population is an inadequate indicator of smoking cessation compared with urinary cotinine. Study Purpose Most of the data concerning smoking cessation and pregnancy have been collected on population samples from large urban and suburban areas. To date, the behavior of rural women in this regard has rarely been systematically evaluated. Therefore, the purpose of this study was to (a) determine the sensitivity and specificity of smoking or nonsmoking status by comparing self-reports of smoking status with urinary cotinine assays of women from rural and small metropolitan statistical areas who stated they were smokers at the onset of pregnancy, (b) compare the characteristics of these women who selfreported a nonsmoker status at the first prenatal visit with those who reported a smoker status, and (c) determine the characteristics that predict self-reported smoking status and positive or negative urinary cotinine levels. Methods The participants in this study were women in the control group subsample of a larger ongoing smoking cessation trial called Smoke Free Baby & Me. The overall purpose of this trial is to assess the tobacco use behaviors of rural and semirural pregnant women in western New York State and northern Pennsylvania and the impact of patient education by nurses on these behaviors. Data for the present study were collected between 1999 and 2000 from private obstetric offices in Chemung County and also serving Schuyler, Steuben, Tioga, and northern Pennsylvania counties. All patients receiving prenatal care in these offices deliver at a 250-bed hospital that provides the only obstetric services for Chemung County. All prenatal patients were screened for eligibility. Participants who met the following criteria were enrolled: confirmed pregnancy, self-reported smoker at the onset of pregnancy, and less than 16 weeks gestation. Written consent was obtained after verbal and written explanations May/June 2004 JOGNN 307

3 were given to potential participants. The study protocol was approved by the institutional review boards of the participating medical center and university. Data on a number of demographic factors were collected during the first prenatal interview. These included age, race, highest year in school completed, marital status, type of health insurance, number of smokers in household, gravidity, parity, gestational age at enrollment, and number of cigarettes smoked per day. Smoking status over the course of the study was monitored using self-reports and blind urinary cotinine assays. At the first prenatal visit, nurse providers asked potential participants the questions, Do you smoke? Did you smoke at the beginning of this pregnancy? If either response was affirmative, they were asked, Are you smoking now? When did you quit? and How many cigarettes do you smoke per day? Responses to these questions as well as demographic information were recorded on a flow sheet kept in each patient s chart. The interview was integrated into the first prenatal visit intake protocol. Urinary cotinine levels were measured in urine samples collected at prenatal visits using an enzyme multiplied immunoassay technique (EMIT) (Quest Diagnostics, Teterboro, NJ). Cotinine concentrations greater than 200 ng/ml indicate active use of nicotine-containing products (Quest Diagnostics, 2001) and were reported as positive (> 200 ng/ml) or negative (< 200 ng/ml). All analyses of data were performed using SPSS for Windows 9.0. Continuous variables were summarized by the mean and standard deviation, and categorical variables by percentages. Chi-square tests were used to evaluate the discordance in reported smoking status and positive and negative cotinine concentrations. T tests and one-way ANOVA (analysis of variance) were used to compare continuous patient characteristics by smoking status (yes/no) and cotinine concentration (positive/negative). Pearson correlations were used to assess associations in continuous patient characteristics. Finally, stepwise logistic regression analyses were used to assess the associations between subject characteristics and reported smoking status and urinary cotinine concentrations, first in the total sample and then in the subgroup of patients who stated they did not smoke at the first prenatal visit. Characteristics included in these analyses were age, education level, gravidity, parity, number of smokers per household, cigarettes smoked per day, marital status, ethnicity, and insurance status. Results Ninety-four participants were enrolled in the study. All participants self-reported that they smoked at the beginning of pregnancy. At their first prenatal visit, 69 (74.19%) stated that they were still smoking, and 24 (25.8%) denied smoking (data were missing for 1 participant). Demographic characteristics of these patients are presented in Table 1. The mean age of the total sample was 23 years, and the highest year in school completed was Approximately 95% of the patients were White, 65% were single, and 63% had Medicaid insurance. These study participants reported that they smoked an average of 5.9 cigarettes a day. Patients who were self-reported smokers at the first prenatal visit had a mean age of years and a mean school completion of years. Nearly 99% were White, 59.4% were single, and 62.3% had Medicaid insurance. In contrast, the self-reported nonsmokers had a mean age of years and years of school completion, 83.3% were White, 83.3% were single, and 65.2% had Medicaid insurance. Analysis of data from 90 patients who had both selfreport of smoking status and urinary cotinine concentrations showed that the total misclassification rate was 16.6% (15 participants at p <.001). Specifically, 24 women self-reported a nonsmoking status at the first prenatal visit. Eight out of 23 women (missing data on 1) had positive cotinine assays > 200 ng/ml. This discordance rate is 34.7% (significant at p <.001), yielding a specificity of 65.3% for self-report. There were 69 women who stated that they were smoking at the first prenatal visit; 60 had positive urinary cotinine assays > 200 ng/ml. Seven had negative assays (missing data on 2). This translated into a discordance rate of 10.44% (significant at p <.001), yielding a sensitivity of 89.5%. Univariate analysis revealed that several correlations among patient characteristics, self-reported smoking status, and urinary cotinine concentrations were significant. The number of cigarettes smoked per day was positively correlated with age (r =.217, p =.036, n = 93), gravidity (r =.379, p <.001, n = 91), parity (r =.365), p <.001, n = 91), and number of smokers in the household (r =.234, p =.038, n = 79). Gravidity was greater among women who reported smoking during pregnancy (t = 3.06, df = 89, p =.003, n = 91) and among those with positive urinary cotinine concentrations (t = 2.60, df = 86, p =.011, n = 88). The number of smokers in the household was also greater among women who reported smoking during pregnancy (t = 3.79, df = 77, p <.001), and among women with a positive cotinine concentration (t = 2.13, df = 74, p =.037). Finally, the results of stepwise logistic regression showed that gravidity was inversely related to smoking status (relative risk =.464, CI [confidence interval] = , p =.007) in the total sample. That is, the high- 308 JOGNN Volume 33, Number 3

4 TABLE 1 Baseline Characteristics of Study Cohorts Self-Reported Self-Reported Total Sample Smokers 1st visit Non-Smokers 1st Mean (SD) Mean (SD) Mean (SD) Age (years) (4.97) n = (5.48) n = (2.84) n = 24 Education (years) (1.62) n = (1.71) n = (1.25) n = 24 Gravidity 2.41 (1.38) n = (1.45) n = (.85) n = 24 Parity.86 (1.08) n = (1.13) n = (.88) n = 24 Smokers/household 1.5 (.74) n = (.57) n = (.97) n = 20 Cigarettes per day 5.96 (5.75) n = (5.27) n = 69 0 Insurance: Total n n = 94 n = 69 n = 23 Private 30.4%, n = %, n = %, n = 7 HMO 4.3%, n = 4 4.3%, n = 3 4.3%, n = 1 Medicaid 63%, n = %, n = %, n = 15 Self pay/none 2.2%, n = 2 2.9%, n = 2 0% Marital status: Total n n = 94 n = 69 n = 24 Married 26.9%, n = %, n = %, n = 4 Single 65.6%, n = %, n = %, n = 20 Divorced 4.3%, n = 4 5.8%, n = 4 0% Separated 3.2%, n = 3 4.3%, n = 3 0% Ethnicity: Total n n = 94 n = 69 n = 24 White 94.6%, n = %, n = %, n = 20 Black 2.2%, n = 2 1.4%, n = 1 4.2%, n = 1 Hispanic 2.2%, (n = 2) 0% 8.3%, n = 2 Other 1.1%, (n = 1) 0% 4.2%, n = 1 er the number of pregnancies, the less likely the patient had a negative smoking status. Likewise, the number of cigarettes smoked per day was also inversely related to urinary cotinine (relative risk =.719, CI = , p =.0025). The higher the number of cigarettes, the less likely the patient had a negative urinary cotinine assay. No other predictors were found in the total sample or in the subgroup. High discordance rates indicate that the initial prenatal interview fails to detect a significant proportion of smokers. Discussion The results of this study suggest that self-report of smoking status among a sample of rural and semirural pregnant women was a poor indicator of smoking cessation, compared with the biochemical standard of urinary cotinine. Findings on the specificity of smoking status were substantially lower than rates reported in the literature (65.5% vs. 75.2%). In addition, the discordance rate of 34.7% among women in this study who reported a nonsmoking status approached the highest rate of 38%, which was reported for pregnant adolescents (Albrecht et al., 1999). One reason for these high rates could be that the sample consisted of women who were smokers at the beginning of pregnancy, and, therefore, were recent quitters. These data substantiate findings from the CARDIA study that discordance rates were higher in former smokers (Wagenknecht, Burke, Perkins, Haley, & Friedman, 1992). Lending further validity to data on specificity is this current study s high sensitivity rate for urinary cotinine. The rate of 89.5% exceeded the average rate of 86.3% in previous research. The findings that women who report smoking at the beginning of pregnancy are predominately White (95%), single (65%), and have Medicaid insurance (63%) are similar to other studies (Britton, 1998). Those studies May/June 2004 JOGNN 309

5 reported that a woman was more likely to smoke during pregnancy if she was White (Milham & Davis, 1991), single or separated (Frost et al., 1994; Madeley, Gillies, Power, & Symonds, 1989), and received publicly funded prenatal care (Adams et al., 1992; Frost et al., 1994). Univariate analysis showed that there were two significant factors associated with positive self-report of smoking and positive urinary cotinine at the first prenatal visit: multigravidity and the number of smokers in the household. Significant positive correlations between multigravidity, multiparity, and number of smokers in the household, with the number of cigarettes smoked per day also corroborated previous research, which has shown that a woman who was a multigravida was more likely to continue smoking during pregnancy (Stockbauer & Ladd, 1991) and had a partner who smokes (Madeley et al., 1989). The incentive for pregnant smokers to quit smoking may be diminished if they have had previous successful births and delivered seemingly healthy infants. In clinical practice, pregnant smokers who are multiparae often remark that they are not concerned about the deleterious effects of smoking because I smoked with my other pregnancies and my babies were fine. The stress of child care may also be a deterrent to smoking cessation. Likewise, having other smokers in the household who refuse to curb their smoking habits presents roadblocks to successful prenatal quitting. Many related factors, including the lack of social support, the easy availability of cigarettes, and the trigger of tobacco smoke, may sabotage cessation efforts. Two factors associated with positive self-report of smoking and positive urinary cotinine at the first prenatal visit were multigravidity and number of smokers in the household. Nursing Implications The results of this study also have important implications for the clinical practice of maternal-child health care. High discordance rates indicate that the initial prenatal interview fails to detect a significant proportion of prenatal smokers. That the collection of this study s data took place in a variety of private offices, by different levels of nurses, nurse practitioners, and nurse-midwives, attests to the widespread nature of the problem. If prenatal smoking cessation interventions are not targeted to patients who self-report a nonsmoker status at the first prenatal visit, then nearly 35% of pregnant smokers who self-report a nonsmoker status are being missed. As efforts are increased to make tobacco use during pregnancy more and more socially abhorrent, it is likely that self-report will become even more unreliable. Finally, in light of the current and previous findings, several changes need to be made in how smoking status is ascertained in pregnant women. First, questions regarding smoking need to be extended beyond the single question assessment approach commonly used (Walsh et al., 1996). Focus should be on developing a line of prompts to get to the heart of past tobacco use, taking into consideration the knowledge that former smokers and recent quitters have high self-deception rates. In addition, interview techniques should be used that encourage women to report occasional smoking or very low daily cigarette use (Walsh et al., 1996, p. 678). Second, in addition to self-reports, biochemical validation of smoking status should be integrated into first prenatal visit protocols. Boyd et al. (1998) suggest employing the newly developed cotinine urine dipstick as part of the standard array of tests in order to screen all patients. Last, the profile of the woman who continues to smoke during pregnancy as White, single, and on Medicaid is once again reinforced. It is a reminder to health care professionals that smoking behavior is often shaped by social deprivation and cultural patterning (Dines, 1994). It also signals heightened awareness of the impact that geographic and regional differences have on this patterning. Perhaps rural residence is another demographic marker to be added to the profile. This extrapolation is supported by findings that women who report high levels of social isolation are also at risk for higher tobacco use during pregnancy (McCormick & Wallace, 1990) and that isolation is a key concept in rural sociology (Lee, 1998). Summary In summary, this study further substantiates the unreliability of self-report of smoking status in the pregnant population, especially among recent quitters. It also initiates data collection on the tobacco use of pregnant women residing in rural and small metropolitan statistical areas and suggests that discordance between self-reports of smoking status and urinary cotinine assays is higher in this population. Findings also earmark pregnant smokers who are multigravidas and who live with other smokers as more likely to continue smoking. They, therefore, require more tailored interventions. Before health care providers can intervene, however, more research is needed on how the smoking status of patients can be ethically and efficiently identified. 310 JOGNN Volume 33, Number 3

6 Acknowledgments This study was supported in part by Quest Diagnostics, The Southern Tier Tobacco Awareness Coalition, New York State, and Ross Laboratories. We wish to thank the participating obstetric offices and medical center, and the many nurses, nurse midwives, and nurse practitioners who implemented the Smoke Free Baby & Me Program and assisted with data collection. REFERENCES Adams, M. M., Brogan, D. J., Kendrick, J. S., Shulman, H. B., Zahniser, S. C., & Bruce, F. C. (1992). Smoking, pregnancy, and source of prenatal care: Results from the pregnancy risk assessment monitoring system. Obstetrics and Gynecology, 80(5), Albrecht, S. A., Reynolds, M. D., Salamie, D., & Payne, L. (1999). A comparison of saliva cotinine, carbon monoxide levels, and self-report as indicators of smoking cessation in the pregnant adolescent. Journal of Addictions Nursing, 11(3), Boyd, N. R., Windsor, R. A., Perkins, L. L., & Lowe, J. B. (1998). Quality of measurement of smoking status by selfreport and saliva cotinine among pregnant women. Maternal and Child Health Journal, 2(2), Britton, G. A. (1998). A review of women and tobacco: Have we come such a long way? Journal of Obstetric, Gynecologic, and Neonatal Nursing, 27, Dines, A. (1994). What changes in health behavior might nurses logically expect from their health education work? Journal of Advanced Nursing, 20, Ford, R. P. K., Tappin, D. M., Schluter, P. J., & Wild, C. J. (1997). Smoking during pregnancy: How reliable are maternal self reports in New Zealand? Journal of Epidemiology and Community Health, 51, Frost, F. J., Cawthorn, M. L., Tollestrup, K., Kenny, F. W., Shrager, L. S., & Nordlund, D. J. (1994). Smoking prevalence during pregnancy for women who are and women who are not Medicaid-funded. American Journal of Preventive Medicine, 10(2), Gebauer, C., Kwo, C. Y., Haynes, E. F., Wewers, M. E. (1998). A nurse-managed smoking cessation intervention during pregnancy. Journal of Obstetric, Gynecologic, and Neonatal Nursing, 21, Klebanoff, M. A., Levine, R. J., Clemens, J. D., DerSimonian, R., & Wilkins, D. G. (1998). Serum cotinine and selfreported smoking during pregnancy. American Journal of Epidemiology, 148(3), Lee, H. J. (Ed.). (1998). Conceptual basis for rural nursing. New York: Springer. Madeley, R. J., Gillies, P. A., Power, F. L., & Symonds, E. M. (1989). Nottingham Mothers Stop Smoking Project baseline survey of smoking in pregnancy. Community Medicine, 11(2), Markovic, N., Ness, R. B., Cefelli, D., Grisso, J. A., Stahmer, S., & Shaw, L. M. (2000). Substance use measures among women in early pregnancy. American Journal of Obstetrics and Gynecology, 183, McCormick, M. C., & Wallace, C. Y. (1990). Factors associated with smoking in low-income women pregnant women: Relationship to birth weight, stressful life events, social support, health behaviors and mental distress. Journal of Clinical Epidemiology, 43(5), Milham, S., & Davis, R. L. (1991). Cigarette smoking during pregnancy and mother s occupation. Journal of Occupational Medicine, 33(4), O Connor, A. M., Davies, B. L., Dulberg, C., Buhler, L., Nadon, C., McBride, B. H., et al. (1992). Effectiveness of a pregnancy smoking cessation program. Journal of Obstetric, Gynecologic, and Neonatal Nursing, 21(5), Patrick, D. L., Cheadle, A., Thompson, D. C., Diehr, P., Koepsell, T., & Kinne, S. (1994). The validity of selfreported smoking: A review and meta-analysis. American Journal of Public Health, 84(7), Perez-Stable, E. J., Benowitz, N. L., & Marin, G. (1995). Is serum cotinine a better measure of cigarette smoking than self-report? Preventive Medicine, 24, Quest Diagnostics. (personal communication, January 22, 2001). Shaffer, H. J., Eber, G. B., Hall, M. N., & Vanderbilt, J. (2000). Smoking behavior among casino employees: Self-report validation using plasma cotinine. Addictive Behaviors, 25(5), Stockbauer, J. W., & Ladd, G. H. (1991). Changes in the characteristics of women who smoke during pregnancy. Public Health Reports, 106, Wagenknecht, L. E., Burke, G. L., Perkins, L.L., Haley, N. J., & Friedman, G. D. (1992). Misclassification of smoking status in the CARDIA study: A comparison of self-report with serum cotinine levels. American Journal of Public Health, 82, Walsh, R. A., Redman, S., & Adamson, L. (1996). The accuracy of self-report of smoking status in pregnant women. Addictive Behaviors, 21(5), Geraldine R. Avidano Britton, RN, is a full-time doctoral student and research assistant at the Decker School of Nursing, Binghamton University, Binghamton, NY. She is a former assistant professor of nurse education at Elmira College, Elmira, NY, and a former coordinator of the Southern Tier Tobacco Awareness Coalition of New York State. JoAnne Brinthaupt, CNP, is a nurse practitioner with Ivy Obstetrics and Gynecology Associates, Elmira, NY. Joyce M. Stehle, MSN, is a former Clinical Nurse Specialist in perinatal care at Arnot Ogden Medical Center, Elmira, NY. Gary D. James, PhD, is a research professor of nursing and director of The Institute for Primary and Preventive Health Care, Binghamton University, Binghamton, NY. Address for correspondence: Geraldine (Gerri) Britton, Skyline Drive, Corning, NY 14830; May/June 2004 JOGNN 311