Improving automated case finding for ectopic pregnancy using a classification algorithm.

BACKGROUND: Research and surveillance work addressing ectopic pregnancy often rely on diagnosis and procedure codes available from automated data sources. However, the use of these codes may result in misclassification of cases. Our aims were to evaluate the accuracy of standard ectopic pregnancy codes; and, through the use of additional automated data, to develop and validate a classification algorithm that could potentially improve the accuracy of ectopic pregnancy case identification. METHODS: Using automated databases from two US managed-care plans, Group Health Cooperative (GH) and Kaiser Permanente Colorado (KPCO), we sampled women aged 15-44 with an ectopic pregnancy diagnosis or procedure code from 2001 to 2007 and verified their true case status through medical record review. We calculated positive predictive values (PPV) for code-selected cases compared with true cases at both sites. Using additional variables from the automated databases and classification and regression tree (CART) analysis, we developed a case-finding algorithm at GH (n = 280), which was validated at KPCO (n = 500). RESULTS: Compared with true cases, the PPV of code-selected cases was 68 and 81% at GH and KPCO, respectively. The case-finding algorithm identified three predictors: ≥ 2 visits with an ectopic pregnancy code within 180 days; International Classification of Diseases, 9th Revision, Clinical Modification codes for tubal pregnancy; and methotrexate treatment. Relative to true cases, performance measures for the development and validation sets, respectively, were: 93 and 95% sensitivity; 81 and 81% specificity; 91 and 96% PPV; 84 and 79% negative predictive value. Misclassification proportions were 32% in the development set and 19% in the validation set when using standard codes; they were 11 and 8%, respectively, when using the algorithm. CONCLUSIONS: The ectopic pregnancy algorithm improved case-finding accuracy over use of standard codes alone and generalized well to a second site. When using administrative data to select potential ectopic pregnancy cases, additional widely available automated health plan data offer the potential to improve case identification.

Dust and chemical exposures, and miscarriage risk among women textile workers in Shanghai, China.

INTRODUCTION: To investigate possible associations between miscarriage and occupational exposures in the Shanghai textile industry. METHODS: A retrospective cohort study of miscarriages among 1752 women in the Shanghai textile industry was conducted. Reproductive history was self-reported by women and occupational work histories were collected from factory personnel records. Occupational exposures were assigned by linking work history information to an industry-specific job-exposure matrix informed by factory-specific textile process information and industrial hygiene assessments. Estimates of cotton dust and endotoxin exposure were also assigned. Odds ratios (OR) and 95% CI were estimated by multivariate logistic regression, with adjustment for age at pregnancy, educational level, smoking status of the woman and her spouse, use of alcohol, and woman's year of birth. RESULTS: An elevation in risk of a spontaneously aborted first pregnancy was associated with exposure to synthetic fibres (OR 1.89, 95% CI 1.20 to 3.00) and mixed synthetic and natural fibres (OR 3.31, 95% CI 1.30 to 8.42). No increased risks were observed for women working with solvents, nor were significant associations observed with quantitative cotton dust or endotoxin exposures. Associations were robust and similar when all pregnancies in a woman's reproductive history were considered. CONCLUSIONS: Occupational exposure to synthetic fibres may cause miscarriages, and this possibility should be the subject of further investigation.

Ascertainment of intimate partner violence using two abuse measurement frameworks.

OBJECTIVE: To estimate prevalence of intimate partner violence (IPV) according to two abuse ascertainment tools, and agreement between the tools. METHODS: 2504 women randomly selected from a health maintenance organization were asked about IPV exposure in their most recent intimate relationship using five questions on physical and sexual abuse, and fear due to partner's threats and controlling behavior from the Behavioral Risk Factor Surveillance Survey (BRFSS) and 10 questions from the Women's Experience with Battering (WEB) scale. IPV prevalence was estimated according to the BRFSS and WEB, and the proportion of women who were WEB+/BRFSS+, WEB-/BRFSS-, WEB-/BRFSS+, and WEB+/BRFSS-. RESULTS: In their most recent relationship, 14.7% of women reported abuse of any type on the BRFSS versus 7.0% on the WEB scale. In direct comparisons of the WEB and BRFSS questions, a higher percentage of abused women reported any IPV on the five BRFSS questions (88.4%) compared to the 10 WEB questions (42.0%). However, both the BRFSS and WEB identified some women as abused that would have been missed by the other instrument. CONCLUSIONS: Intimate partner violence prevalence depends on how women are asked about abuse. Resources permitting, more than one abuse ascertainment strategy (for example, both the BRFSS and WEB questions) should be tried in order to broadly identify as many women as possible who interpret themselves as abused.

Delivery method and self-reported postpartum general health status among primiparous women.

Despite nearly four million deliveries in the United States each year, minimal information exists on unintended health consequences following childbirth, particularly in relation to delivery method. The purpose of this study was to assess the association between method of delivery and the general health status, sexual, bowel and urinary functioning of primiparous women as measured at 7 weeks postpartum. Data from the Statewide Obstetrical Review of Quality System (StORQS) Survey of Maternity Care in Washington State were analysed. Participants included all primiparous women with a delivery of a singleton infant discharged alive between August and December 1991 from 10 non-federal short-stay hospitals who responded to the StORQS Survey of Maternity Care (n = 971). The main outcome measures included the modified Medical Outcomes Study 36-Item Short-Form Health Survey and self-reported sexual, bowel and urinary functioning. At 7 weeks postpartum, women who had caesarean or assisted vaginal deliveries reported significantly lower postpartum general health status scores than women with unassisted vaginal delivery. Additionally, women with assisted vaginal delivery reported significantly worse sexual, bowel and urinary functioning. Our results suggest that more careful attention to the postpartum general health and sexual functioning of women with caesarean and assisted vaginal delivery may be merited.

Risk of uterine rupture during labor among women with a prior cesarean delivery.

BACKGROUND: Each year in the United States, approximately 60 percent of women with a prior cesarean delivery who become pregnant again attempt labor. Concern persists that a trial of labor may increase the risk of uterine rupture, an uncommon but serious obstetrical complication. METHODS: We conducted a population-based, retrospective cohort analysis using data from all primiparous women who gave birth to live singleton infants by cesarean section in civilian hospitals in Washington State from 1987 through 1996 and who delivered a second singleton child during the same period (a total of 20,095 women). We assessed the risk of uterine rupture for deliveries with spontaneous onset of labor, those with labor induced by prostaglandins, and those in which labor was induced by other means; these three groups of deliveries were compared with repeated cesarean delivery without labor. RESULTS: Uterine rupture occurred at a rate of 1.6 per 1000 among women with repeated cesarean delivery without labor (11 women), 5.2 per 1000 among women with spontaneous onset of labor (56 women), 7.7 per 1000 among women whose labor was induced without prostaglandins (15 women), and 24.5 per 1000 among women with prostaglandin-induced labor (9 women). As compared with the risk in women with repeated cesarean delivery without labor, uterine rupture was more likely among women with spontaneous onset of labor (relative risk, 3.3; 95 percent confidence interval, 1.8 to 6.0), induction of labor without prostaglandins (relative risk, 4.9; 95 percent confidence interval, 2.4 to 9.7), and induction with prostaglandins (relative risk, 15.6; 95 percent confidence interval, 8.1 to 30.0). CONCLUSIONS: For women with one prior cesarean delivery, the risk of uterine rupture is higher among those whose labor is induced than among those with repeated cesarean delivery without labor. Labor induced with a prostaglandin confers the highest risk.

First-birth cesarean and placental abruption or previa at second birth(1).

OBJECTIVE: To assess the association between first-birth cesarean delivery and second-birth placental abruption and previa. METHODS: We conducted a population-based, retrospective cohort analysis using data from the Washington State Birth Events Record Database. The study cohort included all primiparas who gave birth to live singleton infants in nonfederal short-stay hospitals from January 1, 1987, through December 31, 1996, and who had second singleton births during the same period (n = 96,975). Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated for placental abruption or previa at second births associated with first-birth cesareans. RESULTS: Among our study cohort, abruptio placentae complicated 11.5 per 1000 and placenta previa 5.2 per 1000 singleton deliveries at second births. In logistic regression analyses adjusted for maternal age, women with first-birth cesareans had significantly increased risk of abruptio placentae (OR 1.3, 95% CI 1.1, 1.5), and placenta previa (OR 1.4, 95% CI 1.1, 1.6) at second births, compared with women with prior vaginal deliveries. CONCLUSION: We found moderately increased risk of placental abruption and previa as a long-term effect of prior cesarean delivery on second births.

Cesarean delivery and postpartum mortality among primiparas in Washington State, 1987-1996(1).

OBJECTIVE: To examine the association between delivery method and mortality within 6 months of delivery among primiparas. METHODS: We conducted a population-based, retrospective cohort analysis using statewide, maternally linked birth certificate, hospital discharge, and death certificate data. The present cohort was all primiparas who gave birth to live-born infants in civilian hospitals in Washington State from January 1, 1987 through December 31, 1996 (n = 265,471). Odd ratios (OR) and 95% confidence intervals (CI) were calculated for overall mortality, pregnancy-related mortality, and pregnancy-unrelated mortality associated with delivery method. RESULTS: Thirty-two women (12.1 per 100,000 singleton live births) died within 6 months of delivery of their first child. Eleven of 32 deaths were pregnancy related (4.1 per 100,000 singleton live births, 95% CI 1.6, 6.5), and 21 of the 32 deaths were not pregnancy related (7.9 per 100,000 singleton live births, 95% CI 4.5, 11.3). The pregnancy-related mortality rate was higher among women delivered by cesarean (10.3/100,000) than among women delivered vaginally (2.4/100,000). In logistic regression analyses, women who had cesarean delivery were not at significantly higher risk of death overall after adjustment for maternal age (OR 1.7, 95% CI 0.3, 3.6), pregnancy-related death after adjustment for maternal age and severe preeclampsia (OR 2.2, 95% CI 0.6, 7.9), or pregnancy-unrelated death after adjustment for maternal age and marital status (OR 0.9, 95% CI 0.3, 2.7), relative to women who had vaginal delivery. CONCLUSION: Cesarean delivery might be a marker for serious preexisting morbidities associated with increased mortality risk rather than a risk factor for death in and of itself. Data from additional sources such as medical records and autopsy reports are necessary to disentangle preexisting mortality risk from risk associated solely with delivery method.

Who gets protection orders for intimate partner violence?

BACKGROUND: It is unknown how victims of intimate partner violence (IPV) who seek civil protection orders differ from IPV victims who do not. METHODS: To compare characteristics of women with and without protection orders, 448 women with police or court contact for an IPV incident in Seattle, Washington, were interviewed. Data collected included demographic characteristics of the subject and her abuser, abuse history, and the subject's mental and physical health. RESULTS: IPV victims who obtained protection orders were more likely than victims without protection orders to be employed full-time, be pregnant, be married, aged over 24, and less likely to be involved with perpetrator at index incident. The perpetrators for both groups were similar, and the majority had a current or previous alcohol/drug problem and a previous criminal history. Both groups of victims had been psychologically and physically abused during the previous year and nearly all had symptoms of depression. However, at the index incident, women who sought protection orders were less likely to be physically assaulted or injured, but more likely to have family members or friends physically assaulted. CONCLUSIONS: Financial independence and abuse of family or friends are important factors associated with the decision to seek a protection order in IPV.

Recommendations for the design of epidemiologic studies of endometriosis.

This paper proposes a standard definition of endometriotic disease for epidemiologic studies and suggests subject-selection strategies to increase the validity of clinic- or population-based studies of the disease. Although endometriosis can be defined simply as the presence of ectopic endometrial tissue, emerging evidence indicates that to be pathologic, such tissue must persist and progress. The proposed disease definition incorporates the concepts of persistence and progression, and its use may increase the likelihood of observing true associations in etiologic studies. Potential threats to validity of substantial magnitude exist in both clinic- and population-based epidemiologic studies of endometriosis. In clinic-based studies, control subjects (infertility clinic patients, women delivering infants, or women undergoing tubal ligation) often are not representative of the population from which the cases arose, and bias can be considerable for behavioral and hormone-related exposures. In population-based studies, substantial case underascertainment may exist, and diagnosed cases may be a biased sample of all potential cases in the population. Although neither the ideal design nor the ideal case and control groups are likely to be achievable in epidemiologic studies of endometriosis, the subject-selection strategies suggested may improve the validity of studies that are obliged to depart from the ideal.

Rates and relative risk of hospital admission among women in violent intimate partner relationships.

OBJECTIVES: This study assessed the history of hospitalization among women involved in violent intimate relationships. METHODS: In this 1-year retrospective cohort study, female residents of King County, Washington, who were aged 18 to 44 years and who had filed for a protection order were compared with nonabused women in the same age group. Outcome measures included overall and diagnosis-specific hospital admission rates and relative risk of hospitalization associated with abuse. RESULTS: Women known to be exposed to a violent intimate relationship were significantly more likely to be hospitalized with any diagnosis (age-specific relative risks [RRs] ranging from 1.2 to 2.1), psychiatric diagnoses (RR = 3.6, 95% confidence interval [CI] = 2.8, 4.6), injury and poisoning diagnoses (RR = 1.8, 95% CI = 1.2, 2.8), digestive system diseases (RR = 1.9, 95% CI = 1.3, 2.9), and diagnoses of assault (RR = 4.9, 95% CI = 1.1, 22.1) or attempted suicide (RR = 3.7, 95% CI = 1.6, 9.2) in the year before filing a protection order. CONCLUSIONS: This study showed an increased relative risk of both overall and diagnosis-specific hospitalizations among abused women. Intimate partner violence has a significant impact on women's health and use of health care.

Association between method of delivery and maternal rehospitalization.

CONTEXT: Despite nearly 4 million deliveries in the United States each year, minimal information exists on unintended health consequences following childbirth, particularly in relation to delivery method. OBJECTIVE: To assess the risk for maternal rehospitalization associated with cesarean or assisted vaginal delivery compared with spontaneous vaginal delivery. DESIGN: Retrospective cohort study of data from the Washington State Birth Events Record Database for 1987 through November 1, 1996. SETTING AND PARTICIPANTS: All primiparous women without selected chronic medical conditions who delivered live singleton infants in nonfederal short-stay hospitals in Washington State (N =256,795). MAIN OUTCOME MEASURES: Relative risks (RRs) of rehospitalization within 60 days of cesarean or assisted vaginal vs spontaneous vaginal deliveries. RESULTS: A total of 3149 women (1.2%) were rehospitalized within 60 days of delivery. In logistic regression analyses adjusting for maternal age, rehospitalization was found to be more likely among women with cesarean delivery (RR, 1.8; 95% confidence interval [CI], 1.6-1.9) or assisted vaginal delivery (RR, 1.3; 95% CI, 1.2-1.4) than among women with spontaneous vaginal delivery. Cesarean delivery was associated with significantly increased risks of rehospitalization for uterine infection, obstetrical surgical wound complications, and cardiopulmonary and thromboembolic conditions. Among women with assisted vaginal delivery, significant increased risks were seen for rehospitalization with postpartum hemorrhage, obstetrical surgical wound complications, and pelvic injury. CONCLUSIONS: Women with cesarean and assisted vaginal deliveries were at increased risk for rehospitalization, particularly with infectious morbidities. Effective strategies for preventing and controlling peripartum infection should be an obstetrical priority.

Maternal depressive symptoms and child behavior problems in a nationally representative normal birthweight sample.

OBJECTIVE: To evaluate the association between maternal depressive symptoms and child behavior problems in a nationally representative sample of U.S. mothers of normal birthweight babies. METHODS: We analyzed data from the 1988 National Maternal and Infant Health Survey (NMIHS) and a 1991 follow-up survey. Depressive symptoms were measured at both surveys using the CES-D, and child behavior problems were assessed by maternal self-report at follow-up. RESULTS: Approximately 28% of the 5303 mothers reported depressive symptoms at a mean of 17 months after delivery, as did 20% at 36 months. In multivariate analyses, women with depressive symptoms at either or both surveys were significantly more likely than women without depressive symptoms to report that their children had frequent temper tantrums or difficulty getting along with other children, and were difficult to manage, unhappy, or fearful. Compared to women without depressive symptoms, the risks of reporting three out of the five child behavior problems for women with depressive symptoms were OR = 1.6 (CI = 1.1-2.1), 1988 only; OR = 2.3 (CI = 1.6-3.3), 1991 only; and OR = 3.6 (2.6-5.0), both 1988 and 1991. CONCLUSIONS: Study findings indicate that a substantial proportion of mothers of young children in the United States experience depressive symptoms and that their children are at significantly increased risk of maternally reported behavior problems. Our results suggest that efforts to identify and treat depression in new mothers should be increased and that mothers whose children are found to have behavior problems should be assessed for depression.

Maternal factors and breast cancer risk among young women.

The results from previous studies have provided evidence to support the hypothesised association between intrauterine oestrogen exposure and subsequent risk of breast cancer. Information has not been available to study this relationship for several perinatal factors thought to be related to pregnancy oestrogen levels. Data collected from the mothers of women in two population-based case-control studies of breast cancer in women under the age of 45 years (510 case mothers, 436 control mothers) who were diagnosed between 1983 and 1992 in three western Washington counties were used to investigate further the relationship between intrauterine oestrogen exposure and risk of breast cancer. A pregnancy weight gain of 25-34 pounds was associated with breast cancer risk (odds ratio [OR] = 1.5; 95% confidence interval [CI] 1.1, 2.0); however, women whose mothers gained 35 pounds or more were not at increased risk. Use of antiemetic medication in women with any nausea and vomiting (OR = 2.9; 95% CI 1.1, 8.1) and use of diethylstilboestrol (DES) (OR = 2.3; 95% CI 0.8, 6.4) appeared to be positively associated with breast cancer risk. The results from this study provide limited support for the hypothesis that in utero oestrogen exposure may be related to subsequent breast cancer risk among young women.

Maternal physical activity in pregnancy and infant size for gestational age.

PURPOSE: To estimate the effect of several types of maternal physical activity in pregnancy on size for gestational age and length of gestation. METHODS: Telephone interviews, birth certificates, and medical records provided data on physical activity and other factors for a random sample of 291 Colorado residents. Backward polychotomous logistic regression modeling yielded estimates of the odds ratios for size for gestational age (appropriate versus small or large) and length of gestation (term versus pre-term or post-term) in relation to second and third trimester maternal physical activity. RESULTS: Performance of any moderate or vigorous physical activity for two hours per week or more in any month was associated with a decreased risk of large infant size for gestational age (LGA; odds ratio = 0.3, 95% confidence interval = 0.2, 0.7), but had no significant effect on risk of small infant size for gestational age (SGA; odds ratio (OR) = 0.8, 95% confidence interval (CI) = 0.3, 2.3). Length of gestation was not affected by prenatal physical activity. CONCLUSIONS: These results suggest that prenatal physical activity may decrease risk of LGA, as might be expected given its salutary effects on glucose tolerance.

Levels of antibodies to transferrin and alpha 2-HS glycoprotein in women with and without endometriosis.

PROBLEM: To establish an enzyme-linked immunosorbent assay (ELISA) for determining the levels of antibodies to transferrin and alpha 2-HS glycoprotein in the serum of women with and without endometriosis. METHOD OF STUDY: Serum samples were obtained from 105 normal women, who were randomly selected for a population-based epidemiologic study, and 123 patients with active endometriosis. An ELISA using transferrin and alpha 2-HS glycoprotein as antigens was established. RESULTS: The levels of antibodies to transferrin and alpha 2-HS glycoprotein in the serum of patients with endometriosis were approximately 21 times higher than those in the serum of control subjects without endometriosis. Only 2% of control subjects had false positive levels of these antibodies, and 5% of patients with endometriosis had false negative levels of these antibodies (specificity, 98.1 and 98.1, respectively, for anti-transferrin and anti-alpha 2-HS glycoprotein; sensitivity, 95 and 96.7, respectively, for anti-transferrin and anti-alpha 2-HS glycoprotein). CONCLUSIONS: Patients with endometriosis have significantly higher levels of antibodies to transferrin and alpha 2-HS glycoprotein than control subjects. Testing women for antibodies to transferrin and alpha 2-HS glycoprotein will provide a specific noninvasive diagnosis of endometriosis.

Unintended pregnancy among female soldiers presenting for prenatal care at Madigan Army Medical Center.

More than half of all pregnancies in this country are unintended at the time of conception. Although women make up 13% of the U.S. Army, there are no epidemiological studies of unintended pregnancy in U.S. Army soldiers. This descriptive study was conducted to determine the prevalence of and factors associated with unintended pregnancy in female soldiers presenting for prenatal care at Madigan Army Medical Center. All soldiers presenting for prenatal care from mid-February 1996 through mid-February 1997 were asked to fill out an anonymous, 12-page written questionnaire. All but one soldier (99.7%) agreed to participate (N = 347). Overall, 55% of soldiers presenting for prenatal care reported that their pregnancies were unintended at the time of conception. The majority of officers (60%) and noncommissioned officers (65%) reported that their pregnancies were intended. In contrast, only 39% of the junior enlisted soldiers reported that their pregnancies were intended at the time of conception. The majority of women (62%) who reported that their pregnancies were unintended were not using any form of birth control during the month they conceived. The most common reason for not using birth control was the stopping of contraception because of side effects. The most common contraceptive method used by those women who reported birth control use during the month they conceived was the male condom. The majority of pregnancies among junior enlisted soldiers presenting for prenatal care at Madigan Army Medical Center are unintended. Prevention programs should target this subpopulation of soldiers and address knowledge, attitudes, and beliefs related to the use of contraception among those at risk for unintended pregnancy.

PIP: To determine the prevalence of and factors associated with unintended pregnancy in women in the US Army, a questionnaire was administered to all female soldiers presenting for prenatal care to the Madigan Army Medical Center (Fort Lewis, Washington) in a 12-month period in 1996-97. 347 of the 348 such women agreed to complete the anonymous questionnaire. Mean age was 24.2 years; 48% were White, 33% were Black, and 10% were Hispanic. Overall, 189 women (55%) reported their pregnancy was unintended at the time of conception. This rate was 61% among junior enlisted soldiers, 40% among officers, and 35% among noncommissioned officers. 117 women with unintended pregnancies (62%) were not using any form of contraception at the time of conception, primarily because of side effects experienced with previous use of oral contraceptives or Depo-Provera. Unintended pregnancy was significantly associated with lower rank, younger age, single marital status, lower income, lower educational level, and involvement in a relationship of less than 1 year's duration. Since soldiers who became pregnant during the study period and elected to have an abortion were not included in this sample, the actual prevalence of unintended pregnancy among woman soldiers is considerably higher than 55%.

Analysis, assessment, and presentation of risk-adjusted statewide obstetrical care data: the StORQS II study in Washington State. Statewide Obstetrics Review and Quality System.

OBJECTIVE: To develop an effective, concise presentation of hospital-specific birth event and delivery-related complication rates, including significant deviations from expected rates calculated using risk-adjusted peer hospital data, for distribution to all Washington State hospitals with delivery services. DATA SOURCES: Complete data for calendar year 1993, including inpatient discharge records for mothers and newborns, birth certificates, and infant death records, for 74 hospitals from Washington state source files. STUDY DESIGN: Institutions were classified into four peer groups based on presence of neonatal intensive care units, number of births, and rural/urban location. Twenty-three clinical indicators of procedure (e.g., cesarean section) and complication rates were analyzed and presented. METHODS: For each indicator, observed and expected rates (adjusted within peer group for categorized baseline risk factors) were calculated and presented by institution. Effective graphic and numeric techniques for presenting significant deviations from expected rates were developed. Results were calculated in terms of numbers of events as well as rates. Approaches applicable to institutions with small numbers of deliveries were selected. PRINCIPAL FINDINGS: Exact confidence intervals (C.I.s) for event rates were superior to binomial or Poisson approximations for small hospitals. For calculating expected rates, indirect adjustment was used due to small numbers within risk factor categories. For all indicators, observed and expected rates along with 95 percent C.I.s for the true rate were presented graphically by institution for each peer group. Transforming C.I.s into "statistically acceptable ranges" allowed hospital personnel to assess their performance in terms of actual numbers of events as well as rates. CONCLUSIONS: Readily available statistical methods and straightforward descriptive approaches allow accurate presentation of outcomes for both large and small institutions.

Young maternal age and depressive symptoms: results from the 1988 National Maternal and Infant Health Survey.

OBJECTIVES: The goal of this study was to provide population-based estimates of the prevalence of depressive symptoms among primiparous US adolescent mothers. METHODS: Data from the live-birth component of the 1988 National Maternal and Infant Health Survey were analyzed. RESULTS: The prevalence of depressive symptoms varied by age and race, from 14% among White adult mothers to 48% among Black mothers 15 to 17 years old. After control for income and marital status, the increased prevalence of depressive symptoms associated with adolescent motherhood was greatly diminished (for 15- to 17-year-old Black women and 18- to 19-year-old White women) or eliminated (for 18- to 19-year-old Black women and 15- to 17-year-old White women). CONCLUSIONS: Adolescent mothers experience high rates of depressive symptoms relative to adult mothers, and mental health and other interventions that alleviate the exacerbating influence of poverty and unmarried status are warranted.

Validity and reliability of subject and mother reporting of perinatal factors.

The authors used data from a population-based case-control study of breast cancer in women aged < or = 44 years (cases, n = 975; controls, n = 866) conducted between 1994 and 1996 in three counties of western Washington state to assess the validity and reliability of reported perinatal factors. For a sample of participants, exposure information from self-administered questionnaires was validated with information from birth certificates (cases, n = 378; controls, n = 283). Detailed information regarding perinatal characteristics of their daughters was also collected from subjects' mothers (case mothers, n = 510; control mothers, n = 436) to assess the reliability of subjects' reporting of these events. Although reporting of birth weight by subjects (cases, r = 0.83; controls, r = 0.80) and their mothers (case mothers, r = 0.89; control mothers, r = 0.84) was highly correlated with the birth certificates, there was differential measurement error by subjects; cases reported birth weight accurately on average, but controls tended to underestimate their birth weight. Agreement between the subject and mother report was excellent for birth weight (cases, r = 0.85; controls, r = 0.87) and good for other perinatal factors, but birth order and maternal diethylstilbestrol use were underreported among cases and reported accurately among controls. Differential measurement error of birth weight by case-control status resulted in biased odds ratios for breast cancer risk.

Correlates and effect of non-response in a postpartum survey of obstetrical care quality.

This study investigated the differences between unprompted respondents, prompted respondents, and non-respondents to a postpartum postal survey, and determined the likely impact of non-response on the accuracy of calculations of patient assessments of obstetrical care quality. Birth certificate and hospital discharge data were obtained for 1664 live births at three hospitals in Washington State between 8/91-10/91 and linked with 1268 completed postpartum maternal postal surveys. Non-white race, public insurance payer, unmarried status, and smoking in pregnancy were independent risk factors for non-participation. Among participants, non-white race, unmarried status, and having an infant who was low birthweight, preterm, or discharged late were independent risk factors for prompted response. The inclusion of prompted respondents did not substantially alter the calculated proportion of women rating obstetrical care quality as low, and these figures were similar to proportions estimated for the entire intended cohort using a modification of Drane's method. A one-time mailing of an obstetrical care quality survey can provide information similar to that obtained with more extensive follow-up even though substantial differences may exist between unprompted and prompted respondents, and with adjustment for factors related to non-participation and timing of response, it may be possible to obtain accurate estimation of outcome prevalences for the entire intended cohort.