Improving disclosure of smoking by pregnant women Patricia Dolan Mullen, DrPH,. Joseph P. Carbonari, EdD: Ellen R. Tabak, PhD,. and Marianna C. Glenday, MPH' Houston, Texas Smoking is a major modifiable risk factor in pregnancy, and low-cost interventions have been developed and tested in diverse populations of pregnant smokers. Successful intervention depends on identification, however, and nondisclosure can be a problem. This randomized study compared rates of disclosure with two response formats-multiple choice, in which the patient is able to describe herself as having "cut down," and the usual history question, "Do you smoke?," in which she is forced to answer simply "yes" or "no." Each format was tested in both oral and written channels with a multiethnic adult prenatal population (n = 1078) entering care in a multispecialty group. Study results indicate that the multiple choice question improved disclosure, regardless of channel (oral versus written), by 40%. This effect was observed across racial and ethnic groups. Biochemical tests of urine samples from reported nonsmokers indicated smoking in only 3%. Eleven percent of the "nonsmokers" in the experimental groups refused consent for the urine test, however, and many of these were probably smokers. (AM J OSSTET GVNECOL 1991 ;165:409-13.)
Key words: Prenatal risk assessment, prenatal history taking, validation of self-reported smoking status
Smoking is a major modifiable risk factor in pregnancy, and recent randomized trials of low-cost counseling and self-help material tailored to pregnancy have documented increased cessation in diverse populations of pregnant smokers.'" Clinically significant decreases in rates of smoking, 12 to 14 percentage points over usual care, have been achieved. Application of these intervention methods in obstetric practices, however, depends on effectively identifying smokers. Most physicians now ask about patients' smoking habits!,5 and as a practical matter they must rely on simple verbal inquiry, Biochemical validation is costly, and several of the inexpensive tests are subject to substantial error in classifying individuals. Thus it is of importance that the results of recent studies evaluating the veracity of responses to verbal inquiry indicate a wide range in nondisclosure rates, from 4% to 45%.6.9 The higher rates occur in contexts where patients perceive smoking as especially undesirable. lo This is a likely perception for women in prenatal care, as awareness of the health risks to the fetus increases in the population. From the Center for Health Promotion Research and Development, School of Public Health, University of Texas Health Science Center at Houston,a and the Department of Psychology, University of Houston,' Supported by Michael Wolf, Project Officer, the American Heart Association, Texas Affiliate (grant #88G-258) , Dr, Tabak was supported by Postdoctoral Training Grant (T32-HL07555, Dr, Lawrence W, Green, Principal Investigator). Received for publication July 25, 1990; revised January 16, 1991; accepted January 28, 1991, Reprint requests: Dr. Patricia Mullen, Associate Professor of Behavioral Sciences, School of Public Health, University of Texas Health Science Center, Box 20186, Houston, TX 77225, 611 /28265
The study reported here originated in a populationbased study of a prenatal smoking cessation intervention in a health maintenance organization! The investigators drew on social science methods for obtaining sensitive information 11 to develop an effective question for identifying smokers. The resulting question appeared to be advantageous, but it was not tested experimentally against alternatives. In this paper we report the results of a randomized study comparing the relative effectiveness of practical methods for identifying pregnant smokers against the traditional methods for eliciting smoking status. Material and methods Design. The study used a randomized, fully crossed 2 x 2 factorial design with two response formats ("yes" or "no" and multiple choice) and two channels of questioning (oral and written) and applied the four combinations to all eligible women on their first prenatal visit. Research literature from sociology and the previous clinical trial experience suggested that channel and response format are crucial dimensions in eliciting sensitive information. The same neutrally worded question was used in the four experimental conditions (Fig. 1). The usual history question was represented as the dichotomous response format, whether through an oral or written channel. We selected the multiple choice format (Fig. I) because of its success in the pilot study and its options ("I smoke every once in a while" and " ... I have cut down since becoming pregnant") that offer the woman a means of portraying herself in a more positive light. Multiple choice may be better suited for
409
410
Mullen et al.
Which of the following statements best describes your cigarette smoking: Would you say: 1. I smoke regularly now-about the same amount as before finding out I was pregnant. 2. I smoke regularly now, but I've cut down since I found out I was pregnant. 3. I smoke every once in a while. 4. I have quit smoking since finding out I was pregnant. 5. I wasn't smoking around the time I found out I was pregnant, and I don't currently smoke cigarettes.
Fig.!. Multiple choice question.
written delivery, although in a pretest all women were easily able to recall the response options after oral presentation. The logical order of the multiple choice options is as given in Fig. 1. Although it is theoretically possible to reverse the present order, beginning the list with "I don't smoke" might program the woman's thinking to select the first and easiest alternative, thereby interfering with her consideration of a different and more honest answer. The reading level of the question was fifth grade, as rated on the Flesch-Kincaid Readability Test. In all four experimental conditions, we combined the smoking history question with other questions about personal behavior to form a presumed gradient of sensitivity. Thus use of automobile safety belts preceded the smoking question, and questions about alcohol and illicit drugs followed it. The format of the alcohol question was always identical to the smoking question to maintain a plausible flow of questions, and the ordering of the four questions in each set was constant. Within each of the four clinic sites all women seen on a given day were randomly assigned to one of the four experimental conditions. This clustering of one to 10 women who might be seen in one day's intake was to assure integrity of application and to fit the project into the clinic routine as smoothly as possible. Population. All women ::=: 18 years old were eligible for the study if they were able to speak and read English, were free of mental and sensory handicaps, and made their first prenatal visit at one of four multispecialty group satellite clinic locations during a 17month period (September 1988 through February 1990). Ability to read and speak English was established with a single question, "Do you speak and read English?" We excluded potential subjects if a sensory handicap or evidence of mental retardation or mental illness was found in the course of the history and education session with the nurse educator. Procedures. The experiment was a part of the first visit for prenatal care after pregnancy had been con-
August 1991 Am J Obstet Gynecol
Table I. Sociodemographic and obstetric history profile for eligible obstetric patients
Variable
Consent for prenatal form Smokers Nonsmokers Race White Black Hispanic Other Age (yr) 40 Married Gravidity 0 1 ~2
Parity 0 1 ~2
No. of maternal health problemst 0 1 ~2
No. of fetal or infant health problemst 0 1 ~2
Total (n = 947)* (%)
88.8 89.5 88.7 50.1 32.6 14.7 2.7 6.6 57.0 34.7 1.7 82.5 25.8 35.0 39.4 43.5 37.2 19.3 75.5 20.4 4.1 73.5 22.7 3.8
*Of these, 121 did not consent for use of their prenatal form. tMaternal index score ranges from 0 to 6 and is the number of problems in this pregnancy: chronic hypertension, diabetes, heart disease, hemorrhage, pregnancy-induced hypertension, pulmonary disease. tFetal-infant index score ranges from 0 to 8 and is the number of problems in previous pregnancy: mental handicap, neonatal death, neurologic disorder, prematurity, small for gestational age, sudden infant death syndrome, spontaneous abortion or ectopic pregnancy, stillbirth.
firmed. Prenatal intake visits were scheduled on a single day of the week. At this visit all women saw a specially trained nurse educator to give their medical history, discuss clinic procedures, and receive advice about diet, smoking, alcohol and other drug use, and warning signs of problems in the pregnancy. The sessions lasted approximately 40 minutes. The patient then gave urine and blood samples. Over the course of the study, 19 nurse educators conducted sessions in the four clinics, with a range of two to 203 sessions per nurse, with five nurses doing fewer than 25 sessions and four doing 99 or more each. The experimental questions were embedded in the history questions and were substituted for usual ques-
Improved disclosure of smoking in pregnancy 411
Volume 165 Number 2
Table II. Disclosure of current smoking by experimental condition Dichotomous
Multiple choice
%
I
No.
%
J
Total No.
%
Oral Written
14.34 13.60
36 / 251 37/272
8.56 9.73
22 /257
29/298
11.4 Il.6
TOTAL
13.95
73/523
9.20
511555
1l.5
tions about smoking, alcohol, and illicit drugs. The history already contained both oral and written components, so the experimental variation was easily integrated. Self-report of sensitive behavior can be influenced by the knowledge that an objective measure ofthe same behavior will be used to verify the report. 12 Thus only after the history and education components of the session were completed and before being directed to the laboratory were all eligible women asked to give written consent for copies of their prenatal history forms to be used anonymously by the university-based researchers. Those women who said they were not smoking currently also were asked for written consent to conduct an additional test on the urine sample they would be giving at the laboratory. The same procedure was used for all four groups so that knowledge of the test did not influence response to the smoking history questions. The test was described as (1) anonymous and (2) "a test for by-products of exposure to cigarette smoke in pregnant women." This procedure was approved by the human subjects protection committees at the university and the clinic. Except for a study number, identifying information was removed from the history forms and urine samples before being transferred to the researchers. Validation of self-report. The urine samples of all women who reported nonsmoking and who gave consent for the test were analyzed for presence of a major nicotine metabolite, cotinine. This assessment was by enzyme immunoassay, IS with the laboratory blinded as to subjects' group assignment. The sensitivity of this test is 93%, with a specificity of 95% for serum in pregnant women. I4 All recommended precautions were observed in labeling, handling, storing, and shipping the urine samples. Forty-eight samples were cross-validated with liquid gas chromatography analysis, a more sensitive and specific procedure. 15. 16 There was 100% agreement on both positive and negative samples, with a cutoff of ;::::50 ng/mL Analysis. We analyzed the data by means of a logistic regression modeL The primary dependent variable was self-reported smoking on either the dichotomous question or a "yes" response to any of the first three categories on the multiple choice question. The indepen-
1
No.
58/ 508 66 / 570 124/1078
dent variables were response format, dichotomous or multiple choice, and channel of question, oral or written. Interaction of the two also was tested. An alternative measure of the extent of smoking disclosure was developed from the giving or withholding of informed consent to allow a urine analysis. Those patients who said they were not smoking but refused to give consent were classified as disclosures. This measure excluded those patients who admitted to smoking and thus potentially identified another population, i.e., those who would not disclose smoking behavior regardless of the format and channel of questioning. Results
Of 1206 pregnant women seen for care during the study period, 1078 (89.4%) were eligible for the study. The two major reasons for ineligibility were being < 18 years old (37%) and not speaking and reading English (24%). Sociodemographic and obstetric history profiles are given for each experimental group and for the total sample, excluding 121 women who did not give consent for use of their prenatal history form (Table I). The sample can be described as representing white, black, and hispanic ethnic and racial groups, married, having a mean age of 28 years, with one or fewer living children, and with the large majority having no maternal or previous fetal or infant health problems. The nurses frequently omitted questions about maternal and paternal schooling, and thus these characteristics are not known. Random assignment appears to have yielded equivalent groups on measured variables. The results of the analysis regarding self-reported smoking showed disclosure was affected by format rather than channel of questioning. More women reported smoking as a function of being asked using the multiple choice format than being asked using the dichotomous format (p < 0.014). The channel of questioning, oral or written, did not result in different levels of self report, and there was no interaction of format and channeL Table II provides the frequencies of reported smoking related to format. The multiple choice format elicited a five percentage point higher (13.96% versus 9.19 %) positive smoking response (odds ratio, 1.60; 95% confidence interval, 1.10 to 2.33). There were no
412 Mullen et al.
differences among ethnic or racial groups. Within the multiple choice question, distribution of responses indicating current smoking was as follows: cut back for the pregnancy, 61.6%; smoke every once in a while, 21.9%; and smoke the same amount as before pregnancy, 16.4%. The second phase of the study tested withholding of consent for the additional urine test as a measure of nondisclosure in response to the questions about smoking. The same categorical model was applied to the urine consent measure for 954 self-reported nonsmokers. No interaction effects or main effects were found for format and channel. The overall rate of nonconsent was 11.0%. The urine test results indicated smoking in fewer than 3% of those who gave consent for the test.
Comment The 40% improvement in disclosure found in this study is statistically and clinically significant. It was observed across three ethnic and racial groups in a clinical context where patients gave their responses without being influenced by knowledge of the research. Also, the questions were tested by interviewers with a range of backgrounds and interviewing skills. The lack of difference between the oral and written formats suggests that the written form could be used by any health care provider. Because the "demand" to give a desirable answer to a physician may be higher than to a nurse and because of the time savings, we recommend use of the written format. It is probable that the current association of urine testing with screening for illicit drugs produced some of the nonconsent for the urine tests. Nevertheless, many of the 11 % who did not give consent for the urine test undoubtedly are smokers. This group of unidentified smokers suggests the limits on the sensitivity of the history-taking method tested here. The difference in disclosure that we found, however, would mean that over 55,000 additional pregnant smokers could be identified annually by simple changes in history taking. These identification methods did not elicit a defensive reaction by the woman, as would other means such as nonvoluntary biochemical testing, and the multiple choice question may give more information about the patient's smoking habit. The response option most frequently selected by current smokers, having cut down since the pregnancy began, appears to reflect the true behavior in many women early in pregnancy. For this group and others who had not really cut down, this response offers a means of displaying a partially positive image to the prenatal caregiver. Thus the additional women identified are expected to be as amenable to intervention as others. Such intervention should be a high priority in prenatal care not only to help prevent low birth weight '7 . 18 but also
August 1991 Am J Obstet Gyneco1
to reduce exposure to environmental smoke by infants '9 . 20 and to reduce smoking among younger women whose rates of smoking make them a high priority target for smoking cessation efforts!'·21 We acknowledge the contributions of Mary Ann Richardson and Carol Busby of the Center for Health Promotion Research and Development to project management and manuscript development. From the Kelsey-Seybold Clinics the following individuals supported the conduct of the project: Dr. George Coale, Elene Heyer, Carolyn Hayball, Barbara Erwin, Ms. Vivian Malone, Rene Anderson, Pamela Andrews, Anna Bean, Debra Carpenter, Lisa Kelley, Ruthann Koza, Barbara Krohn, Jeanne Lauda, Anna Perez, and other nurse educators. Dr. Jeffrey Wilkins, West Los Angeles Veterans Administration Medical Center, conducted the urine validation tests, and the Clinical Biochemistry Branch, Centers for Disease Control, contributed the biochemical analyses (Dr. Francis W. Spierto, Supervisory Research Chemist). REFERENCES 1. Windsor RA, Cutter G, Morris J, et al. The effectiveness of smoking cessation methods for smokers in public health maternity clinics: a randomized trial. Am J Public Health 1985;75: 1389-92. 2. Ershoff DH, Mullen PD, Quinn VP. A randomized trial of a serialized self-help smoking cessation for pregnant women in an HMO. AmJ Public Health 1989;79:182-7. 3. Mayer JP, Hawkins B, Todd R. A randomized evaluation of smoking cessation interventions for pregnant women at a WIC clinic. AmJ Public Health 1990;80:76-8. 4. Sachs DPL. Treatment of cigarette dependency: what American pulmonary physicians do. Am Rev Respir Dis 1984; 129: 10 10-3. 5. Mullen PD, Tabak ER. Patterns offamily physicians' counseling practices for health habit modification. Med Care 1989;27:694-704. 6. Li VC, Coates TJ, Spielberg LA, et al. Smoking cessation with young women in public family planning clinics: the impact of physician messages and waiting room media. Prev Med 1984;13:477-89. 7. Richmond R, Webster I. Evaluation of general practitioners' use of a smoking intervention programme. Int J Epidemiol 1985; 14:396-401. 8. Sanders D, Fowler G, Mant D, et al. Randomized controlled trial of anti-smoking advice by nurses in general practice. J R ColI Gen Pract 1989;39:273-6. 9. Cummings SR, Coates TJ, Richard RJ, et al. Training physicians in counseling about smoking cessation: a randomized trial of the "Quit for Life" program. Ann Intern Med 1989;110:640-7. 10. Strecher VJ, Becker MH, Clark NM, Prasada-Rao P. Using patients' descriptions of alcohol consumption, diet, medication compliance, and cigarette smoking: the validity of self-reports in research and practice. J Gen Intern Med 1989;4: 160-6. II. Bradburn NM, Sud man S, et al. Improving interview method and questionnaire design: response effects to threatening questions in survey research. San Francisco: Jossey-Bass, 1979. 12. Bauman KE, Dent CWo Influence of an objective measure of self-reports of behavior. J Appl Psychol 1982;67: 623-8. 13. Bjercke R.I, Cook G, Langone .IL. Comparison of monoclonal and polyclonal antibodies to cotinine in nonisotopic
Improved disclosure of smoking in pregnancy
Volume 165 Number 2
14.
15. 16.
17.
18.
and isotopic immunoassays. J Immunol Methods 1987; 96:239-46. Haddow JE, Knight GJ, Palomaki GE, Haddow PK. Estimating fetal morbidity and mortality resulting from cigarette smoke exposure by measuring cotinine levels in maternal serum. In: Scarpelli DG, Migaki G, eds. Transplacental effects on fetal health. New York: Alan R Liss, 1988;289-300. Jarvis MJ, Tunstall-Pedoe H, Feyerbrand C, Vesey C, Saloojee Y. Comparison of tests used to distinguish smokers from nonsmokers. AmJ Public Health 1987;77:1435-8. Jacob P, Wilson M, Benowitz NL. Improved gas chromatographic method for the determination of nicotine and cotinine in biologic fluids. J Chromatogr 1981; 222:61-70. Department of Health and Human Services. Caring for our future: the content of prenatal care: a report of the Public Health Service Expert Panel on the Content of Prenatal Care. Washington DC: United States Department of Health and Human Services, 1989. Department of Health and Human Services. Guide to
19. 20. 21. 22.
23. 24.
clinical preventive services: a report ofthe U.S. Preventive Services Task Force. Washington DC: United States Department of Health and Human Services, 1989. American Academy of Pediatrics. Involuntary smokinga hazard to children. Pediatrics 1986;77:755-7. Department of Health and Human Services. The health consequences of involuntary smoking. Washington DC: United States Government Printing Office, 1986. Department of Health and Human Services. The health consequences of smoking for women. Washington DC: United States Government Printing Office, 1980. Department of Health and Human Services. The health consequences of smoking: chronic obstructive lung disease. Washington DC: United States Government Printing Office, 1982. Department of Health and Human Services. The health consequences of smoking: cancer. Washington DC: United States Government Printing Office, 1982. Leads from the MMWR. State-specific estimates of smoking attributable mortality and years of potential life lostU.S., 1985. JAMA 1989;261:23-5.
Thrombotic thrombocytopenic purpura first seen as massive vaginal necrosis Deborah C. Gallup, MD,. Thomas E. Nolan, MD,. David Martin, PhD, MD: Donald G. Gallup, MD,' and Donald M. Sherline, MD' Augusta, Georgia Thrombotic thrombocytopenic purpura is a hematologic disorder that affects the microcirculation. A 38-year-old woman was first seen with fever, thrombocytopenia, and vaginal bleeding. Pelvic examination revealed massive vaginal necrosis. Hematology conSUltation resulted in agreement with the diagnosis of thrombotic thrombocytopenic purpura. This is the first reported case of thrombotic thrombocytopenic purpura first seen as vaginal necrosis of which we are aware. (AM J OSSTET GVNECOL 1991 ;165:413-5.)
Key words: Thrombotic thrombocytopenic purpura, vaginal necrosis, cocaine Thrombotic thrombocytopenic purpura is a rare, life-threatening disorder characterized by the pentad of fever, microangiopathic hemolytic anemia, thrombocytopenia, neurologic symptoms, and renal dysfunction. The pathogenesis of thrombotic thrombocytopenic purpura, however, remains unclear. In >90% of cases, purpura is the initial manifestation and may be associated with epistaxis, hematuria, gastrointestinal From the Departments of Obstetrics and Gynecology" and Pathology,' The Medical College of Georgia. Received for publication November 16,1990; revised January 23, 1991; accepted February 27, 1991. Reprint requests: Deborah C. Gallup, MD, Department of Obstetrics and Gynecology, CJ-123, Medical College of Georgia, Augusta, GA 30912-3345.
6/1/29141
hemorrhage, menorrhagia, and hemoptysis. Fever is noted in all cases.1. 2 The following represents the first known case of thrombotic thrombocytopenic purpura first seen as massive vaginal necrosis of which we are aware.
Case report A 38-year-old gravida 3, para 3 black woman went to a local emergency room with a 3-day history of fever, chills, abdominal pain, vaginal bleeding, and discharge. Examination was remarkable only for mild abdominal tenderness. Pelvic examination was refused because of discomfort. Vital signs were stable. Laboratory data revealed a thrombocytopenia of 42,000/mm 3 • She refused admission. The next day her symptoms worsened and she returned to the emer-
413
Key words: Prenatal risk assessment, prenatal history taking, validation of self-reported smoking status
Smoking is a major modifiable risk factor in pregnancy, and recent randomized trials of low-cost counseling and self-help material tailored to pregnancy have documented increased cessation in diverse populations of pregnant smokers.'" Clinically significant decreases in rates of smoking, 12 to 14 percentage points over usual care, have been achieved. Application of these intervention methods in obstetric practices, however, depends on effectively identifying smokers. Most physicians now ask about patients' smoking habits!,5 and as a practical matter they must rely on simple verbal inquiry, Biochemical validation is costly, and several of the inexpensive tests are subject to substantial error in classifying individuals. Thus it is of importance that the results of recent studies evaluating the veracity of responses to verbal inquiry indicate a wide range in nondisclosure rates, from 4% to 45%.6.9 The higher rates occur in contexts where patients perceive smoking as especially undesirable. lo This is a likely perception for women in prenatal care, as awareness of the health risks to the fetus increases in the population. From the Center for Health Promotion Research and Development, School of Public Health, University of Texas Health Science Center at Houston,a and the Department of Psychology, University of Houston,' Supported by Michael Wolf, Project Officer, the American Heart Association, Texas Affiliate (grant #88G-258) , Dr, Tabak was supported by Postdoctoral Training Grant (T32-HL07555, Dr, Lawrence W, Green, Principal Investigator). Received for publication July 25, 1990; revised January 16, 1991; accepted January 28, 1991, Reprint requests: Dr. Patricia Mullen, Associate Professor of Behavioral Sciences, School of Public Health, University of Texas Health Science Center, Box 20186, Houston, TX 77225, 611 /28265
The study reported here originated in a populationbased study of a prenatal smoking cessation intervention in a health maintenance organization! The investigators drew on social science methods for obtaining sensitive information 11 to develop an effective question for identifying smokers. The resulting question appeared to be advantageous, but it was not tested experimentally against alternatives. In this paper we report the results of a randomized study comparing the relative effectiveness of practical methods for identifying pregnant smokers against the traditional methods for eliciting smoking status. Material and methods Design. The study used a randomized, fully crossed 2 x 2 factorial design with two response formats ("yes" or "no" and multiple choice) and two channels of questioning (oral and written) and applied the four combinations to all eligible women on their first prenatal visit. Research literature from sociology and the previous clinical trial experience suggested that channel and response format are crucial dimensions in eliciting sensitive information. The same neutrally worded question was used in the four experimental conditions (Fig. 1). The usual history question was represented as the dichotomous response format, whether through an oral or written channel. We selected the multiple choice format (Fig. I) because of its success in the pilot study and its options ("I smoke every once in a while" and " ... I have cut down since becoming pregnant") that offer the woman a means of portraying herself in a more positive light. Multiple choice may be better suited for
409
410
Mullen et al.
Which of the following statements best describes your cigarette smoking: Would you say: 1. I smoke regularly now-about the same amount as before finding out I was pregnant. 2. I smoke regularly now, but I've cut down since I found out I was pregnant. 3. I smoke every once in a while. 4. I have quit smoking since finding out I was pregnant. 5. I wasn't smoking around the time I found out I was pregnant, and I don't currently smoke cigarettes.
Fig.!. Multiple choice question.
written delivery, although in a pretest all women were easily able to recall the response options after oral presentation. The logical order of the multiple choice options is as given in Fig. 1. Although it is theoretically possible to reverse the present order, beginning the list with "I don't smoke" might program the woman's thinking to select the first and easiest alternative, thereby interfering with her consideration of a different and more honest answer. The reading level of the question was fifth grade, as rated on the Flesch-Kincaid Readability Test. In all four experimental conditions, we combined the smoking history question with other questions about personal behavior to form a presumed gradient of sensitivity. Thus use of automobile safety belts preceded the smoking question, and questions about alcohol and illicit drugs followed it. The format of the alcohol question was always identical to the smoking question to maintain a plausible flow of questions, and the ordering of the four questions in each set was constant. Within each of the four clinic sites all women seen on a given day were randomly assigned to one of the four experimental conditions. This clustering of one to 10 women who might be seen in one day's intake was to assure integrity of application and to fit the project into the clinic routine as smoothly as possible. Population. All women ::=: 18 years old were eligible for the study if they were able to speak and read English, were free of mental and sensory handicaps, and made their first prenatal visit at one of four multispecialty group satellite clinic locations during a 17month period (September 1988 through February 1990). Ability to read and speak English was established with a single question, "Do you speak and read English?" We excluded potential subjects if a sensory handicap or evidence of mental retardation or mental illness was found in the course of the history and education session with the nurse educator. Procedures. The experiment was a part of the first visit for prenatal care after pregnancy had been con-
August 1991 Am J Obstet Gynecol
Table I. Sociodemographic and obstetric history profile for eligible obstetric patients
Variable
Consent for prenatal form Smokers Nonsmokers Race White Black Hispanic Other Age (yr) 40 Married Gravidity 0 1 ~2
Parity 0 1 ~2
No. of maternal health problemst 0 1 ~2
No. of fetal or infant health problemst 0 1 ~2
Total (n = 947)* (%)
88.8 89.5 88.7 50.1 32.6 14.7 2.7 6.6 57.0 34.7 1.7 82.5 25.8 35.0 39.4 43.5 37.2 19.3 75.5 20.4 4.1 73.5 22.7 3.8
*Of these, 121 did not consent for use of their prenatal form. tMaternal index score ranges from 0 to 6 and is the number of problems in this pregnancy: chronic hypertension, diabetes, heart disease, hemorrhage, pregnancy-induced hypertension, pulmonary disease. tFetal-infant index score ranges from 0 to 8 and is the number of problems in previous pregnancy: mental handicap, neonatal death, neurologic disorder, prematurity, small for gestational age, sudden infant death syndrome, spontaneous abortion or ectopic pregnancy, stillbirth.
firmed. Prenatal intake visits were scheduled on a single day of the week. At this visit all women saw a specially trained nurse educator to give their medical history, discuss clinic procedures, and receive advice about diet, smoking, alcohol and other drug use, and warning signs of problems in the pregnancy. The sessions lasted approximately 40 minutes. The patient then gave urine and blood samples. Over the course of the study, 19 nurse educators conducted sessions in the four clinics, with a range of two to 203 sessions per nurse, with five nurses doing fewer than 25 sessions and four doing 99 or more each. The experimental questions were embedded in the history questions and were substituted for usual ques-
Improved disclosure of smoking in pregnancy 411
Volume 165 Number 2
Table II. Disclosure of current smoking by experimental condition Dichotomous
Multiple choice
%
I
No.
%
J
Total No.
%
Oral Written
14.34 13.60
36 / 251 37/272
8.56 9.73
22 /257
29/298
11.4 Il.6
TOTAL
13.95
73/523
9.20
511555
1l.5
tions about smoking, alcohol, and illicit drugs. The history already contained both oral and written components, so the experimental variation was easily integrated. Self-report of sensitive behavior can be influenced by the knowledge that an objective measure ofthe same behavior will be used to verify the report. 12 Thus only after the history and education components of the session were completed and before being directed to the laboratory were all eligible women asked to give written consent for copies of their prenatal history forms to be used anonymously by the university-based researchers. Those women who said they were not smoking currently also were asked for written consent to conduct an additional test on the urine sample they would be giving at the laboratory. The same procedure was used for all four groups so that knowledge of the test did not influence response to the smoking history questions. The test was described as (1) anonymous and (2) "a test for by-products of exposure to cigarette smoke in pregnant women." This procedure was approved by the human subjects protection committees at the university and the clinic. Except for a study number, identifying information was removed from the history forms and urine samples before being transferred to the researchers. Validation of self-report. The urine samples of all women who reported nonsmoking and who gave consent for the test were analyzed for presence of a major nicotine metabolite, cotinine. This assessment was by enzyme immunoassay, IS with the laboratory blinded as to subjects' group assignment. The sensitivity of this test is 93%, with a specificity of 95% for serum in pregnant women. I4 All recommended precautions were observed in labeling, handling, storing, and shipping the urine samples. Forty-eight samples were cross-validated with liquid gas chromatography analysis, a more sensitive and specific procedure. 15. 16 There was 100% agreement on both positive and negative samples, with a cutoff of ;::::50 ng/mL Analysis. We analyzed the data by means of a logistic regression modeL The primary dependent variable was self-reported smoking on either the dichotomous question or a "yes" response to any of the first three categories on the multiple choice question. The indepen-
1
No.
58/ 508 66 / 570 124/1078
dent variables were response format, dichotomous or multiple choice, and channel of question, oral or written. Interaction of the two also was tested. An alternative measure of the extent of smoking disclosure was developed from the giving or withholding of informed consent to allow a urine analysis. Those patients who said they were not smoking but refused to give consent were classified as disclosures. This measure excluded those patients who admitted to smoking and thus potentially identified another population, i.e., those who would not disclose smoking behavior regardless of the format and channel of questioning. Results
Of 1206 pregnant women seen for care during the study period, 1078 (89.4%) were eligible for the study. The two major reasons for ineligibility were being < 18 years old (37%) and not speaking and reading English (24%). Sociodemographic and obstetric history profiles are given for each experimental group and for the total sample, excluding 121 women who did not give consent for use of their prenatal history form (Table I). The sample can be described as representing white, black, and hispanic ethnic and racial groups, married, having a mean age of 28 years, with one or fewer living children, and with the large majority having no maternal or previous fetal or infant health problems. The nurses frequently omitted questions about maternal and paternal schooling, and thus these characteristics are not known. Random assignment appears to have yielded equivalent groups on measured variables. The results of the analysis regarding self-reported smoking showed disclosure was affected by format rather than channel of questioning. More women reported smoking as a function of being asked using the multiple choice format than being asked using the dichotomous format (p < 0.014). The channel of questioning, oral or written, did not result in different levels of self report, and there was no interaction of format and channeL Table II provides the frequencies of reported smoking related to format. The multiple choice format elicited a five percentage point higher (13.96% versus 9.19 %) positive smoking response (odds ratio, 1.60; 95% confidence interval, 1.10 to 2.33). There were no
412 Mullen et al.
differences among ethnic or racial groups. Within the multiple choice question, distribution of responses indicating current smoking was as follows: cut back for the pregnancy, 61.6%; smoke every once in a while, 21.9%; and smoke the same amount as before pregnancy, 16.4%. The second phase of the study tested withholding of consent for the additional urine test as a measure of nondisclosure in response to the questions about smoking. The same categorical model was applied to the urine consent measure for 954 self-reported nonsmokers. No interaction effects or main effects were found for format and channel. The overall rate of nonconsent was 11.0%. The urine test results indicated smoking in fewer than 3% of those who gave consent for the test.
Comment The 40% improvement in disclosure found in this study is statistically and clinically significant. It was observed across three ethnic and racial groups in a clinical context where patients gave their responses without being influenced by knowledge of the research. Also, the questions were tested by interviewers with a range of backgrounds and interviewing skills. The lack of difference between the oral and written formats suggests that the written form could be used by any health care provider. Because the "demand" to give a desirable answer to a physician may be higher than to a nurse and because of the time savings, we recommend use of the written format. It is probable that the current association of urine testing with screening for illicit drugs produced some of the nonconsent for the urine tests. Nevertheless, many of the 11 % who did not give consent for the urine test undoubtedly are smokers. This group of unidentified smokers suggests the limits on the sensitivity of the history-taking method tested here. The difference in disclosure that we found, however, would mean that over 55,000 additional pregnant smokers could be identified annually by simple changes in history taking. These identification methods did not elicit a defensive reaction by the woman, as would other means such as nonvoluntary biochemical testing, and the multiple choice question may give more information about the patient's smoking habit. The response option most frequently selected by current smokers, having cut down since the pregnancy began, appears to reflect the true behavior in many women early in pregnancy. For this group and others who had not really cut down, this response offers a means of displaying a partially positive image to the prenatal caregiver. Thus the additional women identified are expected to be as amenable to intervention as others. Such intervention should be a high priority in prenatal care not only to help prevent low birth weight '7 . 18 but also
August 1991 Am J Obstet Gyneco1
to reduce exposure to environmental smoke by infants '9 . 20 and to reduce smoking among younger women whose rates of smoking make them a high priority target for smoking cessation efforts!'·21 We acknowledge the contributions of Mary Ann Richardson and Carol Busby of the Center for Health Promotion Research and Development to project management and manuscript development. From the Kelsey-Seybold Clinics the following individuals supported the conduct of the project: Dr. George Coale, Elene Heyer, Carolyn Hayball, Barbara Erwin, Ms. Vivian Malone, Rene Anderson, Pamela Andrews, Anna Bean, Debra Carpenter, Lisa Kelley, Ruthann Koza, Barbara Krohn, Jeanne Lauda, Anna Perez, and other nurse educators. Dr. Jeffrey Wilkins, West Los Angeles Veterans Administration Medical Center, conducted the urine validation tests, and the Clinical Biochemistry Branch, Centers for Disease Control, contributed the biochemical analyses (Dr. Francis W. Spierto, Supervisory Research Chemist). REFERENCES 1. Windsor RA, Cutter G, Morris J, et al. The effectiveness of smoking cessation methods for smokers in public health maternity clinics: a randomized trial. Am J Public Health 1985;75: 1389-92. 2. Ershoff DH, Mullen PD, Quinn VP. A randomized trial of a serialized self-help smoking cessation for pregnant women in an HMO. AmJ Public Health 1989;79:182-7. 3. Mayer JP, Hawkins B, Todd R. A randomized evaluation of smoking cessation interventions for pregnant women at a WIC clinic. AmJ Public Health 1990;80:76-8. 4. Sachs DPL. Treatment of cigarette dependency: what American pulmonary physicians do. Am Rev Respir Dis 1984; 129: 10 10-3. 5. Mullen PD, Tabak ER. Patterns offamily physicians' counseling practices for health habit modification. Med Care 1989;27:694-704. 6. Li VC, Coates TJ, Spielberg LA, et al. Smoking cessation with young women in public family planning clinics: the impact of physician messages and waiting room media. Prev Med 1984;13:477-89. 7. Richmond R, Webster I. Evaluation of general practitioners' use of a smoking intervention programme. Int J Epidemiol 1985; 14:396-401. 8. Sanders D, Fowler G, Mant D, et al. Randomized controlled trial of anti-smoking advice by nurses in general practice. J R ColI Gen Pract 1989;39:273-6. 9. Cummings SR, Coates TJ, Richard RJ, et al. Training physicians in counseling about smoking cessation: a randomized trial of the "Quit for Life" program. Ann Intern Med 1989;110:640-7. 10. Strecher VJ, Becker MH, Clark NM, Prasada-Rao P. Using patients' descriptions of alcohol consumption, diet, medication compliance, and cigarette smoking: the validity of self-reports in research and practice. J Gen Intern Med 1989;4: 160-6. II. Bradburn NM, Sud man S, et al. Improving interview method and questionnaire design: response effects to threatening questions in survey research. San Francisco: Jossey-Bass, 1979. 12. Bauman KE, Dent CWo Influence of an objective measure of self-reports of behavior. J Appl Psychol 1982;67: 623-8. 13. Bjercke R.I, Cook G, Langone .IL. Comparison of monoclonal and polyclonal antibodies to cotinine in nonisotopic
Improved disclosure of smoking in pregnancy
Volume 165 Number 2
14.
15. 16.
17.
18.
and isotopic immunoassays. J Immunol Methods 1987; 96:239-46. Haddow JE, Knight GJ, Palomaki GE, Haddow PK. Estimating fetal morbidity and mortality resulting from cigarette smoke exposure by measuring cotinine levels in maternal serum. In: Scarpelli DG, Migaki G, eds. Transplacental effects on fetal health. New York: Alan R Liss, 1988;289-300. Jarvis MJ, Tunstall-Pedoe H, Feyerbrand C, Vesey C, Saloojee Y. Comparison of tests used to distinguish smokers from nonsmokers. AmJ Public Health 1987;77:1435-8. Jacob P, Wilson M, Benowitz NL. Improved gas chromatographic method for the determination of nicotine and cotinine in biologic fluids. J Chromatogr 1981; 222:61-70. Department of Health and Human Services. Caring for our future: the content of prenatal care: a report of the Public Health Service Expert Panel on the Content of Prenatal Care. Washington DC: United States Department of Health and Human Services, 1989. Department of Health and Human Services. Guide to
19. 20. 21. 22.
23. 24.
clinical preventive services: a report ofthe U.S. Preventive Services Task Force. Washington DC: United States Department of Health and Human Services, 1989. American Academy of Pediatrics. Involuntary smokinga hazard to children. Pediatrics 1986;77:755-7. Department of Health and Human Services. The health consequences of involuntary smoking. Washington DC: United States Government Printing Office, 1986. Department of Health and Human Services. The health consequences of smoking for women. Washington DC: United States Government Printing Office, 1980. Department of Health and Human Services. The health consequences of smoking: chronic obstructive lung disease. Washington DC: United States Government Printing Office, 1982. Department of Health and Human Services. The health consequences of smoking: cancer. Washington DC: United States Government Printing Office, 1982. Leads from the MMWR. State-specific estimates of smoking attributable mortality and years of potential life lostU.S., 1985. JAMA 1989;261:23-5.
Thrombotic thrombocytopenic purpura first seen as massive vaginal necrosis Deborah C. Gallup, MD,. Thomas E. Nolan, MD,. David Martin, PhD, MD: Donald G. Gallup, MD,' and Donald M. Sherline, MD' Augusta, Georgia Thrombotic thrombocytopenic purpura is a hematologic disorder that affects the microcirculation. A 38-year-old woman was first seen with fever, thrombocytopenia, and vaginal bleeding. Pelvic examination revealed massive vaginal necrosis. Hematology conSUltation resulted in agreement with the diagnosis of thrombotic thrombocytopenic purpura. This is the first reported case of thrombotic thrombocytopenic purpura first seen as vaginal necrosis of which we are aware. (AM J OSSTET GVNECOL 1991 ;165:413-5.)
Key words: Thrombotic thrombocytopenic purpura, vaginal necrosis, cocaine Thrombotic thrombocytopenic purpura is a rare, life-threatening disorder characterized by the pentad of fever, microangiopathic hemolytic anemia, thrombocytopenia, neurologic symptoms, and renal dysfunction. The pathogenesis of thrombotic thrombocytopenic purpura, however, remains unclear. In >90% of cases, purpura is the initial manifestation and may be associated with epistaxis, hematuria, gastrointestinal From the Departments of Obstetrics and Gynecology" and Pathology,' The Medical College of Georgia. Received for publication November 16,1990; revised January 23, 1991; accepted February 27, 1991. Reprint requests: Deborah C. Gallup, MD, Department of Obstetrics and Gynecology, CJ-123, Medical College of Georgia, Augusta, GA 30912-3345.
6/1/29141
hemorrhage, menorrhagia, and hemoptysis. Fever is noted in all cases.1. 2 The following represents the first known case of thrombotic thrombocytopenic purpura first seen as massive vaginal necrosis of which we are aware.
Case report A 38-year-old gravida 3, para 3 black woman went to a local emergency room with a 3-day history of fever, chills, abdominal pain, vaginal bleeding, and discharge. Examination was remarkable only for mild abdominal tenderness. Pelvic examination was refused because of discomfort. Vital signs were stable. Laboratory data revealed a thrombocytopenia of 42,000/mm 3 • She refused admission. The next day her symptoms worsened and she returned to the emer-
413
