Recent research has shown an association between smoking in pregnancy and low birthweight. Many authors have concluded that the relationship is causal but some have argued that it is the smoker rather than the smoke which is responsible. This thesis examines the relationship between the smoking habit in pregnancy and birthweight corrected for gestational age using data from the St. George's Hospital Birthweight Study. Adjustment is made for confounding factors so that the effect of smoking can be estimated. The statistical problem of adjusting birthweight for gestational age when very early births are included is discussed and a solution presented in the form of a birthweight ratio. The relationship is examined between birthweight ratio and many socioeconomic and psychological factors and shows that few are associated with reduced birthweight. Those associations which are observed can be explained by smoking. Alcohol and caffeine are only related to birthweight in smokers. When the smoking habit is analysed in terms of quantity and constituents, a threshold is observed whereby women smoking a low number of low yield cigarettes have mean birthweight similar to non-smokers. For women smoking higher numbers of cigarettes but a low yield brand mean birthweight is reduced by the same amount (6% or more) as women smoking high yield brands. The effect on birthweight of alcohol and caffeine in smokers only is adjusted for smoking by using this threshold. This shows that smoking, alcohol and caffeine are all associated with reduced birthweight. For alcohol and caffeine consumption this relationship is strongest in early pregnancy and weakest near delivery. The association between smoking and birthweight is not explained by any of the wide range of confounding factors examined. This provides evidence that the relationship is a causal one.

My three papers evaluate the effectiveness and cost-effectiveness of clinical and policy strategies to improve women’s health, focusing on human papillomavirus (HPV) vaccination in the U.S. and maternal health care in a developing country context. Paper 1 presents a claims-based econometric analysis of the U.S.’s Patient Protection and Affordable Care Act provision requiring the elimination of cost-sharing for recommended preventive care. I evaluate the effect of this value-based insurance design intervention on HPV immunization rates among girls and young women enrolled in private insurance plans. My regression approach uses variation in the intensity and timing of the intervention across plans to distinguish policy effects from background trends. I find that the policy was associated with modest increases in age-specific vaccination rates. Increases in vaccination per dollar reduction in cost-sharing were notably larger among beneficiaries in socioeconomically disadvantaged areas. Nevertheless, vaccination rates under free preventive care were well below federal targets, highlighting the need for additional interventions to increase HPV vaccine coverage. In Paper 2, I undertake a comparative effectiveness analysis of HPV vaccination by dose level within a U.S. cohort of adolescent girls and young women. Rates of screening-detected cervical abnormalities in claims are compared among recipients of zero, one, two, or three doses, using a marginal structural model approach to adjust for a broader set of potential confounders than would be possible with conventional regression methods. Findings from these analyses complement prior evidence from immunogenicity trials, and although protective effects appear greatest with three doses, support the value of HPV vaccination even when incomplete. Vaccine effect estimates are largest with respect to high-grade lesions that are precursors to cervical cancer. Using primary data from a randomized experiment, Paper 3 examines the cost-effectiveness of pay-for-performance interventions among obstetric care providers in rural Karnataka, India. I construct a decision analytic model to quantify incremental costs and life years under alternative policy scenarios, combining obstetric complication outcomes and program expenditures from the trial with published evidence on complication-related mortality and medical costs. Results suggest that an incentive program based on input quality is not cost-effective in its current form, but could become economically attractive if program activities can be adjusted to reduce costs while maintaining similar health effects. Performance data collection costs were substantial in this resource-limited setting and represent a key barrier to cost-effectiveness. / Health Policy

No description available.

No description available.

Postmenopausal women rely on hormone replacement therapy to manage menopausal symptoms. Few studies have assessed the consequences of estrogen progestin replacement therapy (EPRT), a formulation prescribed to postmenopausal women with an intact uterus. The aim of this thesis was to conduct a systematic review and meta-analysis assessing the association between breast cancer risk and EPRT use among postmenopausal women. Electronic databases, handsearching and advice from experts were used to locate studies. A fixed effects model was used to calculate the combined estimate and its standard error. Statistical tests for association and homogeneity were performed. The association between breast cancer risk and EPRT exposure was significant for both case control (OR = 1.38; 95% CI 1.23--1.56; chi2 = 48.30, p < 0.0001) and cohort studies (RR = 1.42; 95% CI 1.29--1.57; chi 2 = 27.95, p < 0.0001). Results indicate that postmenopausal women on EPRT have a higher breast cancer risk, particularly with long term use and testosterone derived formulations.

This Ph.D. project sought to determine the expression, regulation, and function of the KIT-SCF receptor-ligand system in human epithelia] ovarian cancer. The expression of c- KIT and SCF in normal ovaries, in cultured ovarian surface epithelium (OSE), and in epithelia] ovarian tumors was analyzed. Normal OSE expressed SCF, but not c- KIT ; however, epithelia] invaginations and inclusion cysts often expressed KIT protein. Of 15 benign ovarian tumors and tumors of low malignant potential, 87% expressed c- KIT , and 92% of these co-expressed SCF, suggesting the possibility of autocrine growth regulation. Of 35 malignant ovarian cancers, 71% expressed c- KIT (92% co-expressed SCF), with a trend for decreased c- KIT expression in advanced stage disease. Of 34 patients with malignant tumors for whom follow-up information was available (median follow-up time of 24 months), 9 had tumors that did not express c- KIT , 8 (89%) of whom have died and the remaining 1 has recurrent disease. Of the 25 patients with tumors expressing c- KIT , 56% are still alive, eight of whom have no evidence of disease. Importantly, statistical analysis indicated that patients whose tumors did not express c- KIT had a significantly shorter (p < 0.05) disease-free survival time than patients who had KIT-expressing tumors. Studies were carried out to identify intraovarian growth regulatory factors which may regulate c- KIT and SCF expression in ovarian cancer cells, and to determine whether activated KIT can affect the proliferation and survival of these cells. HEY cells, which co-expressed KIT and SCF, were treated with transforming growth factor (TGF)-α, TGF-β, and dibutyryl cyclic AMP (dbcAMP) and their cellular proliferation and expression of c- KIT and SCF were examined. A series of transfection studies were carried out to determine if enforced c- kit expression inhuman ovarian carcinoma cells could regulate cellular proliferation. Transient transfection of c- kit into HEY cells resulted in decreased proliferation. Similarly, stable transfection of c- kit into A2780-cp cells, which do not express endogenous c- KIT , also resulted in a decreased proliferative rate. In contrast to the ovarian cancer cells, increased proliferation was documented for NIH 3T3 fibroblast cells transiently transfected with c- kit . Together, these results suggest that the positive prognostic value of c- KIT expression in ovarian tumors is related to its negative growth regulatory function in ovarian cancer cells. (Abstract shortened by UMI.)

Abstract Not Available.

Background. Decision support interventions (DSIs) are interventions used by patients and their practitioners to help make difficult shared healthcare decisions. The efficacy of these interventions is well established. However, there are no formal comparisons of the efficacy of different delivery methods. Interactive computerized delivery methods have the advantage of allowing patients to control the flow of information and to receive feedback on their comprehension. Objective. To compare the efficacy of an interactive computerized DSI for women considering long term hormone replacement therapy (HRT), to that of a validated audio-booklet-based version of the same intervention. Study design. Fifty-one peri- and post-menopausal women aged 40--70 who were literate in English, and who showed no evidence of cognitive impairment, were randomized to use either the computerized or the standard audio-booklet version of the DSI. The patients were interviewed with a pre- and post-intervention questionnaire. Interventions. (1) Standard audio-booklet-based decision aid: A 40-minute audio tape guided participants through an illustrated booklet describing the risks and benefits of post-menopausal HRT. (2) Interactive computer based decision aid: The computerized version of the DSI presents identical information as synchronized text/audio/animation. (Abstract shortened by UMI.)

Le but de l'étude consiste à évaluer auprès de femmes francophones l'efficacité d'un support d'aide à la décision sur la qualité de la prise de décision en regard de l'hormonothérapie substitutive. Une expérimentation avec un devis avant-après avec groupe unique a été menée auprès de 40 femmes âgées de 47 à 64 ans. Les variables à l'étude ont été mesurées à l'aide des questionnaires et des échelles de mesures portant sur: (1) les connaissances; (2) les attentes; (3) les valeurs; et (4) le niveau de conflit décisionnel en regard des bénéfices et des risques de l'hormonothérapie substitutive. Les résultats de cette étude révèlent une amélioration significative ( p < 0,05) des connaissances et une amélioration des attentes réalistes de la femme par rapport aux risques et aux bénéfices de l'hormonothérapie, après l'utilisation du support d'aide à la décision. Quant aux valeurs, les résultats révèlent une meilleure cohérence des valeurs de la femme par rapport à l'hormonothérapie après l'intervention. Enfin, les résultats démontrent, en général, une diminution significative du niveau de conflit décisionnel. Les femmes se sentent plus informées, supportées et ont la perception qu'elles ont pris une décision efficace. Les résultats de cette étude suggèrent que l'utilisation d'un support d'aide à la décision chez les femmes ménopausées contribue à l'amélioration de la qualité de la prise de décision par rapport à l'hormonothérapie substitutive. Cette étude doit être considérée comme un projet pilote, lequel devra être suivi d'une étude auprès d'un plus grand échantillon afin d'examiner la généralisation possible des résultats à l'ensemble de la population féminine francophone.

Objectives. (1) Primary, to identify factors influencing the decision to take long-term preventive hormone replacement therapy (LTP-HRT) assessed 2 months after the counseling interview with the physician. (2) Secondary, to identify factors influencing the decision to take LTP-HRT assessed after reviewing a decision support intervention and after a counseling interview with their physician and to identify the factors that are associated with women being uncertain regarding the decision to take LTP-HRT. Method. Data, collected from women who were participating in a RCT to evaluate the effectiveness of a decision support intervention, was used to develop multivariate statistical models at three specific time points. The time points were after reviewing the decision support intervention, after a counseling interview with their physician, and 2 months after the counseling interview. The physician was the unit of randomization in the original study and thus multivariate statistical modeling adjusting for correlated observations using Generalized Estimating Equations was utilized. (Abstract shortened by UMI.)