Quantifying uncertainty in cost effectiveness analyses

This study compared four alternative approaches (Taylor, Fieller, percentile bootstrap, and bias-corrected bootstrap methods) to estimating confidence intervals (CIs) around cost-effectiveness (CE) ratio. The study consisted of two components: (1) Monte Carlo simulation was conducted to identify characteristics of hypothetical cost-effectiveness data sets which might lead one CI estimation technique to outperform another. These results were matched to the characteristics of an (2) extant data set derived from the National AIDS Demonstration Research (NADR) project. The methods were used to calculate (CIs) for data set. These results were then compared. The main performance criterion in the simulation study was the percentage of times the estimated (CIs) contained the “true” CE. A secondary criterion was the average width of the confidence intervals. For the bootstrap methods, bias was estimated. ^ Simulation results for Taylor and Fieller methods indicated that the CIs estimated using the Taylor series method contained the true CE more often than did those obtained using the Fieller method, but the opposite was true when the correlation was positive and the CV of effectiveness was high for each value of CV of costs. Similarly, the CIs obtained by applying the Taylor series method to the NADR data set were wider than those obtained using the Fieller method for positive correlation values and for values for which the CV of effectiveness were not equal to 30% for each value of the CV of costs. ^ The general trend for the bootstrap methods was that the percentage of times the true CE ratio was contained in CIs was higher for the percentile method for higher values of the CV of effectiveness, given the correlation between average costs and effects and the CV of effectiveness. The results for the data set indicated that the bias corrected CIs were wider than the percentile method CIs. This result was in accordance with the prediction derived from the simulation experiment. ^ Generally, the bootstrap methods are more favorable for parameter specifications investigated in this study. However, the Taylor method is preferred for low CV of effect, and the percentile method is more favorable for higher CV of effect. ^

Project Alpha: A culturally appropriate approach to adolescent male sex education

In the United States today, adolescents face unacceptably high rates of mortality and morbidity due to the contraction of HIV/AIDS, sexually transmitted diseases and teenage pregnancy. In view of these rates, there is a need for applied preventive interventions to delay adolescent sexual behavior until adulthood. Project Alpha was a school-adopted, quasi-experimental program for adolescent male students attending Sharpstown High School in Houston, Texas. This intervention used student newsletters to provide specific role-model stories on community and student role models who have changed attitudes or improved efficacy to abstain from sexual behavior until adulthood. It was hypothesized that teenagers exposed to the intervention would show improvements in knowledge, beliefs, avoidance skills, perceived norms, intentions and self-efficacy to delay sexual behavior compared to no-treatment reference teenagers in the same school.^ In total, the Project Alpha program had a significant effect on student knowledge, beliefs (towards abstinence and having sex with multiple partners), perceived risk (HIV/STD testing), self-efficacy (could avoid sex with attractive girl who wants to have sex), perceived social norms (friends believing in sexual abstinence) and sexual intentions. However, no significant intervention effects were found in student's beliefs (that it was OK to have sex with girlfriend), perceived risk of HIV/STD, self-efficacy (to avoid sex with girlfriend) and social norms (friends believe it is OK to have sex with a girlfriend and multiple partners in the same month). ^

The association of genetic groups 1, 2, and 3 of Mycobacterium tuberculosis with pulmonary and extra-pulmonary tuberculosis

Objective. To investigate the association of the three major genetic groups of Mycobacterium tuberculosis with pulmonary and extra-pulmonary tuberculosis in clustered and non-clustered TB cases in the Houston area. ^ Study design. Secondary analysis of an ambi-directional study. ^ Study population. Three hundred fifty-eight confirmed cases of tuberculosis in the Houston that occurred between October 1995 and May 1997, who had been interviewed by the Houston T13 Initiative staff at Baylor College of Medicine, and whose isolates have had their DNA fingerprint and genetic group determined. ^ Exclusions. Individuals whose mycobacterial genotype was unknown, or whose data variables were unavailable. ^ Source of data. Laboratory results, patient interviews, and medical records at clinics and hospitals of the study population. ^ Results. In clustered cases, the majority of both, pulmonary and extra-pulmonary TB cases were caused by genetic group 1. Independent factors were assessed to determine the interactions that may influence the site of infection or increase the risk for one site or another. HIV negative males were protected against extra-pulmonary TB compared to HIV negative females. Individuals ages 1–14 years were at higher risk of having extra-pulmonary TB. Group 3 organisms were found less frequently in the total population in general, especially in extra-pulmonary disease. This supports the evidence in previous studies that this group is the least virulent and genetically distinct from the other two groups. Group 1 was found more frequently among African Americans than other ethnic groups, a trend for future investigations. ^ Among the non-clustered cases, group 2 organisms were the majority of the organisms found in both sites. They were also the majority of organisms found in African Americans, Caucasians, and Hispanics causing the majority of the infections at both sites. However, group 1 organisms were the overwhelming majority found in Asian/Pacific Islander individuals, which may indicate these organisms are either endemic to that area, or that there is an ethnic biological factor involved. This may also be due to a systematic bias, since isolates from individuals from that geographic region lack adequate copies of the insertion sequence IS6110, which leads to their placement in the non-clustered population. ^ The three genetic groups of Mycobacterium tuberculosis were not found equally distributed between sites of infection in both clustered and non-clustered cases. Furthermore, these groups were not distributed in the same patterns among the clustered and non-clustered cases, but rather in distinct patterns. ^

Home care service utilization in British Columbia: Substitute or complement to acute care?

The desire to promote efficient allocation of health resources and effective patient care has focused attention on home care as an alternative to acute hospital service. in particular, clinical home care is suggested as a substitute for the final days of hospital stay. This dissertation evaluates the relationship between hospital and home care services for residents of British Columbia, Canada beginning in 1993/94 using data from the British Columbia Linked Health database. ^ Lengths of stay for patients referred to home care following hospital discharge are compared to those for patients not referred to home care. Ordinary least squares regression analysis adjusts for age, gender, admission severity, comorbidity, complications, income, and other patient, physician, and hospital characteristics. Home care clients tend to have longer stays in hospital than patients not referred to home care (β = 2.54, p = 0.0001). Longer hospital stays are evident for all home care client groups as well as both older and younger patients. Sensitivity analysis for referral time to direct care and extreme lengths of stay are consistent with these findings. Two stage regression analysis indicates that selection bias is not significant.^ Patients referred to clinical home care also have different health service utilization following discharge compared to patients not referred to home care. Home care nursing clients use more medical services to complement home care. Rehabilitation clients initially substitute home care for physiotherapy services but later are more likely to be admitted to residential care. All home care clients are more likely to be readmitted to hospital during the one year follow-up period. There is also a strong complementary association between direct care referral and homemaker support. Rehabilitation clients have a greater risk of dying during the year following discharge. ^ These results suggest that home care is currently used as a complement rather than a substitute for some acute health services. Organizational and resource issues may contribute to the longer stays by home care clients. Program planning and policies are required if home care is to provide an effective substitute for acute hospital days. ^

Congential syphilis in Harris County, Texas, 1993--1996

The specific aims of this study were to: (1) To determine the incidence of congenital syphilis (CS) for the 1993-96 birth cohorts in Houston/Harris County, based on the newly revised 1989 CS surveillance definition in Houston/Harris County, Texas; (2) To study the distribution of selected variables listed in the Centers for Disease Control and Prevention (CDC) Congenital Syphilis Case Investigation and Report Form for the reported cases of CS; (3) To ascertain the relationships of maternal demographic factors, geographic distribution, and provision and utilization of prenatal care, associated with reported congenital syphilis delivery.^ This was a descriptive study analyzing reported cases of congenital syphilis in Houston/Harris, County, TX during the years 1993-96 using the data recorded on the CDC's Congenital Syphilis Case Investigation and Report forms. The population included infants delivered during 1993-96 who were diagnosed with congenital syphilis, using the revised 1990 criteria of the CDC. This study examined the risk factors associated with the occurrence of congenital syphilis in Harris County where the prevalence of maternal syphilis infection (13.7/100,000/1995) is high. (Abstract shortened by UMI.) ^

A comparison of Markov and generalized linear models of hospital mortality

This paper reports a comparison of three modeling strategies for the analysis of hospital mortality in a sample of general medicine inpatients in a Department of Veterans Affairs medical center. Logistic regression, a Markov chain model, and longitudinal logistic regression were evaluated on predictive performance as measured by the c-index and on accuracy of expected numbers of deaths compared to observed. The logistic regression used patient information collected at admission; the Markov model was comprised of two absorbing states for discharge and death and three transient states reflecting increasing severity of illness as measured by laboratory data collected during the hospital stay; longitudinal regression employed Generalized Estimating Equations (GEE) to model covariance structure for the repeated binary outcome. Results showed that the logistic regression predicted hospital mortality as well as the alternative methods but was limited in scope of application. The Markov chain provides insights into how day to day changes of illness severity lead to discharge or death. The longitudinal logistic regression showed that increasing illness trajectory is associated with hospital mortality. The conclusion is reached that for standard applications in modeling hospital mortality, logistic regression is adequate, but for new challenges facing health services research today, alternative methods are equally predictive, practical, and can provide new insights. ^

The effects of population density on the spread of disease

The objective of this study is to identify the relationship between population density and the initial stages of the spread of disease in a local population. This study proposes to concentrate on the question of how population density affects the distribution of the susceptible individuals in a local population and thus affects the spread of the disease, measles. Population density is measured by the average of the number of contacts with susceptible individuals by each individual in the population during a fixed-length time period. The term “contact with susceptible individuals” means sufficient contact between two people for the disease to pass from an infectious person to a susceptible person. The fixed-length time period is taken to be the average length of time an infected person is infectious without symptoms of the disease. For this study of measles, the time period will be seven days. ^ While much attention has been given to modeling the entire epidemic process of measles, attempts have not been made to study the characteristics of contact rates required to initiate an epidemic. This study explores the relationship between population density, given a specific herd immunity rate in the population, and initial rate of the spread of the disease by considering the underlying distribution of contacts with susceptibles by the individuals in the population. ^ This study does not seek to model an entire measles epidemic, but to model the above stated relationship for the local population within which the first infective person is introduced. This study describes the mathematical relationship between population density parameters and contact distribution parameters. ^ The results are displayed in graphs that show the effects of different population densities on the spread of disease. The results support the idea that the number of new infectives is strongly related to the distribution of susceptible contacts. The results also show large differences in the epidemic measures between populations with densities equal to four versus three. ^

Design and implementation of a multicriteria medical decision support system for diagnosis and treatment

Health care providers face the problem of trying to make decisions with inadequate information and also with an overload of (often contradictory) information. Physicians often choose treatment long before they know which disease is present. Indeed, uncertainty is intrinsic to the practice of medicine. Decision analysis can help physicians structure and work through a medical decision problem, and can provide reassurance that decisions are rational and consistent with the beliefs and preferences of other physicians and patients. ^ The primary purpose of this research project is to develop the theory, methods, techniques and tools necessary for designing and implementing a system to support solving medical decision problems. A case study involving “abdominal pain” serves as a prototype for implementing the system. The research, however, focuses on a generic class of problems and aims at covering theoretical as well as practical aspects of the system developed. ^ The main contributions of this research are: (1) bridging the gap between the statistical approach and the knowledge-based (expert) approach to medical decision making; (2) linking a collection of methods, techniques and tools together to allow for the design of a medical decision support system, based on a framework that involves the Analytic Network Process (ANP), the generalization of the Analytic Hierarchy Process (AHP) to dependence and feedback, for problems involving diagnosis and treatment; (3) enhancing the representation and manipulation of uncertainty in the ANP framework by incorporating group consensus weights; and (4) developing a computer program to assist in the implementation of the system. ^

Risk, outcomes, and cost of emergency center visits in cancer patients

Cancer is a chronic disease that often necessitates recurrent hospitalizations, a costly pattern of medical care utilization. In chronically ill patients, most readmissions are for treatment of the same condition that caused the preceding hospitalization. There is concern that rather than reducing costs, earlier discharge may shift costs from the initial hospitalization to emergency center visits. ^ This is the first descriptive study to measure the incidence of emergency center visits (ECVs) after hospitalization at The University of M. D. Anderson Cancer Center (UTMDACC), to identify the risk factors for and outcomes of these ECVs, and to compare 30-day all-cause mortality and costs for episodes of care with and without ECVs. ^ We identified all hospitalizations at UTMDACC with admission dates from September 1, 1993 through August 31, 1997 which met inclusion criteria. Data were electronically obtained primarily from UTMDACC's institutional database. Demographic factors, clinical factors, duration of the index hospitalization, method of payment for care, and year of hospitalization study were variables determined for each hospitalization. ^ The overall incidence of ECVs was 18%. Forty-five percent of ECVs resulted in hospital readmission (8% of all hospitalizations). In 1% of ECVs the patient died in the emergency center, and for the remaining 54% of ECVs the patient was discharged home. Risk factors for ECVs were marital status, type of index hospitalization, cancer type, and duration of the index hospitalization. The overall 30-day all-cause mortality rate was 8.6% for hospitalizations with an ECV and 5.3% for those without an ECV. In all subgroups, the 30-day all-cause mortality rate was higher for groups with ECVs than for those without ECVs. The most important factor increasing cost was having an ECV. In all patient subgroups, the cost per episode of care with an ECV was at least 1.9 times the cost per episode without an ECV. ^ The higher costs and poorer outcomes of episodes of care with ECVs and hospital readmissions suggest that interventions to avoid these ECVs or mitigate their costs are needed. Further research is needed to improve understanding of the methodological issues involved in relation to health care issues for cancer patients. ^

Sequential updating via dynamic hierarchical models: A Bayesian approach to longitudinal data analysis

Most statistical analysis, theory and practice, is concerned with static models; models with a proposed set of parameters whose values are fixed across observational units. Static models implicitly assume that the quantified relationships remain the same across the design space of the data. While this is reasonable under many circumstances this can be a dangerous assumption when dealing with sequentially ordered data. The mere passage of time always brings fresh considerations and the interrelationships among parameters, or subsets of parameters, may need to be continually revised. ^ When data are gathered sequentially dynamic interim monitoring may be useful as new subject-specific parameters are introduced with each new observational unit. Sequential imputation via dynamic hierarchical models is an efficient strategy for handling missing data and analyzing longitudinal studies. Dynamic conditional independence models offers a flexible framework that exploits the Bayesian updating scheme for capturing the evolution of both the population and individual effects over time. While static models often describe aggregate information well they often do not reflect conflicts in the information at the individual level. Dynamic models prove advantageous over static models in capturing both individual and aggregate trends. Computations for such models can be carried out via the Gibbs sampler. An application using a small sample repeated measures normally distributed growth curve data is presented. ^

Detection of enterotoxigenic Escherichia coli in foods by polymerase chain reaction

Enterotoxigenic Escherichia coli (ETEC) causes significant morbidity and mortality in infants of developing countries and is the most common cause of diarrhea in travelers to these areas. Enterotoxigenic Escherichia coli infections are commonly caused by ingestion of fecally contaminated food. A timely method for the detection of ETEC in foods would be important in the prevention of this disease. A multiplex polymerase chain reaction (PCR) assay which has been successful in detecting the heat-labile and heat-stable toxins of ETEC in stool was examined to determine its utility in foods. This PCR assay, preceded by a glass matrix and chaotropic DNA extraction, was effective in detecting high numbers of ETEC in a variety of foods. Ninety percent of 121 spiked food samples yielded positive results. Samples of salsa from Guadalajara, Mexico and Houston, Texas were collected and underwent DNA extraction and PCR. All samples yielded negative results for both the heat-labile and heat-stable toxins. Samples were also subjected to oligonucleotide probe analysis and resulted in 5 samples positive for ETEC. Upon dilution testing, it was found that positive PCR results only occurred when 12,000 to 1,000,000 bacteria were present in 200 mg of food. Although the DNA extraction and PCR method has been shown to be both sensitive and specific in stool, similar results were not obtained in food samples. ^

Prediction of cardiovascular disease risk factors in Mexican-American children from parental risk factors, and social and demographic characteristics in the HHANES

Objectives. Cardiovascular disease (CVD) including CVD secondary to diabetes type II, a significant health problem among Mexican American populations, originates in early childhood. This study seeks to determine risk factors available to the health practitioner that can identify the child at potential risk of developing CVD, thereby enabling early intervention. ^ Design. This is a secondary analysis of cross-sectional data of matched Mexican American parents and children selected from the HHANES, 1982–1984. ^ Methods. Parents at high risk for CVD were identified based on medical history, and clinical and physical findings. Factor analysis was performed on children's skinfold thicknesses, height, weight, and systolic and diastolic blood pressures, in order to produce a limited number of uncorrelated child CVD risk factors. Multiple regression analyses were then performed to determine other CVD markers associated with these Factors, independently for mothers and fathers. ^ Results. Factor analysis of children's measurements revealed three uncorrelated latent variables summarizing the children's CVD risk: Factor1: ‘Fatness’, Factor2: ‘Size and Maturity’, and Factor3: ‘Blood Pressure’, together accounting for the bulk of variation in children's measurements (86–89%). Univariate analyses showed that children from high CVD risk families did not differ from children of low risk families in occurrence of high blood pressure, overweight, biological maturity, acculturation score, or social and economic indicators. However, multiple regression using the factor scores (from factor analysis) as dependent variables, revealed that higher CVD risk in parents, was significantly associated with increased fatness and increased blood pressure in the children. Father's CVD risk status was associated with higher levels of body fat in his children and higher levels of blood pressure in sons. Mother's CVD risk status was associated with higher blood pressure levels in children, and occurrence of obesity in the mother associated with higher fatness levels in her children. ^ Conclusion. Occurrence of cardiovascular disease and its risk factors in parents of Mexican American children, may be used to identify children at potentially higher risk for developing CV disease in the future. Obesity in mothers appears to be an important marker for the development of higher levels of body fatness in children. ^

Behavioral and psychosocial factors associated with sexual risk in a population of crack cocaine users

The purpose of this dissertation was to examine the relationship between key psychosocial and behavioral components of the Transtheoretical Model and the Theory of Reasoned Action for sexual risk reduction in a population of crack cocaine smokers and sex workers, not in drug treatment. ^ The first study examined the results of an analysis of the association between two principal constructs in the Transtheoretical Model, the processes of change and the stages of change for condom use, in a high risk population. In the analysis of variance for all respondents, the overall F-test revealed that people in different stages have different levels of experiential process use, F(3,317) = 17.79, p = 0.0001 and different levels of behavioral process use, F(3,317) = 28.59, p = .0001. For the experiential processes, there was a significant difference between the precontemplation/contemplation stage, and both the action, and maintenance, stages.^ The second study explored the relationship between the Theory of Reasoned Action “beliefs” and the stages-of-change in the same population. In the analysis of variance for all participants, the results indicate that people in different stages did value the positive beliefs differently, F(3,502) = 15.38, p = .0001 but did not value the negative beliefs differently, F(3,502) = 2.08, p = .10. ^ The third study explored differences in stage-of-change by gender, partner type drug use, and HIV status. Three discriminant functions emerged, with a combined χ2(12) = 139.57, p = <.0001. The loading matrix of correlations between predictors and discriminant functions demonstrate that the strongest predictor for distinguishing between the precontemplation/contemplation stage and the preparation, action, and maintenance stages (first function) is partner type (.962). The loadings on the second discriminant function suggest that once partner type has been accounted for, ever having HIV/AIDS (.935) was the best predictor for distinguishing between the first three stages and the maintenance stage. ^ These studies demonstrate that behavioral change theories can contribute important insight to researchers and program planners attempting to alter HIV risk behavior in high-risk populations. ^

Effect of cancer chemotherapy on the frequency of minisatellite repeat number changes in human sperm

The objective of this study was to determine whether cancer chemotherapy induces detectable mutations in DNA of the human germline and whether minisatellite repeat number changes can be used as a sensitive indicator of genetic damage in human sperm caused by mutagens. We compared the mutation frequencies in sperm of the same cancer patients pre- and post-, pre- and during, or during and post-treatment. Small pool polymerase chain reaction (SP-PCR) (DNA equivalent to approximately 100 sperm) and Southern blotting techniques were used to detect mutations and quantify the frequency of repeat number changes at the minisatellite MS205 locus. One pre- and one post-treatment semen sample was obtained from each Hodgkin's disease patient treated with either: (1) a regimen without alkylating agents, Novantrone, Oncovin, Vinblastine, and Prednisone (NOVP), 4 patients; (2) a regimen containing alkylating agents, Cytoxan, Vinblastine, Procarbazine, and Prednisone (CVPP)/Adriamycin, Bleomycin, DTIC, CCNU, and Prednisone (ABDIC), 2 patients; and (3) a regimen containing alkylating agents, Mechlorethamine, Oncovin, Procarbazine, and Prednisone (MOPP), 1 patient. One pre- and one during treatment semen sample from each of two Hodgkin's disease patients treated with Adriamycin, Bleomycin, Vinblastine, and Dacarbazine (ABVD) were obtained. One during and one post-treatment semen sample from a Hodgkin's disease patient treated with NOVP were also obtained. At least 7900 sperm in each sample were screened for the repeat number changes at the MS205 locus by multi-aliquots of SP-PCR. The mutation frequencies of pre- and post-treatment for the four patients treated with NOVP were 0.22 and 0.18%; 0.24 and 0.16%; 0.35 and 0.28%; and 0.19 and 0.18%. With CVPP/ABDIC, they were 0.22 and 0.23%; and 0.94 and 0.98% for the two patients and with MOPP they were 0.79 and 1.14%. The mutation frequencies of pre- and during treatment with ABVD were 0.09 and 0.07%; and 0.34 and 0.27% for the two patients. The mutation frequencies of during and post-treatment with NOVP for one patient were 0.31 and 0.25%. A statistically significant increase in mutation frequency was only found in the patient treated with MOPP. According to the time of samples collected after or during treatment and the above results, we conclude that there is no effect of NOVP and CVPP/ABDIC regimens on the mutation frequency in spermatogonia. The spermatocytes are not highly sensitive to chemotherapy agents compared to spermatogonia at the minisatellite MS205 locus. MOPP treatment may increase the mutation frequency at the MS205 locus in spermatogonia. ^

Evaluation of the accuracy of multilevel analysis parameter estimates for intra-cluster correlation in correlated binary data

Many studies in biostatistics deal with binary data. Some of these studies involve correlated observations, which can complicate the analysis of the resulting data. Studies of this kind typically arise when a high degree of commonality exists between test subjects. If there exists a natural hierarchy in the data, multilevel analysis is an appropriate tool for the analysis. Two examples are the measurements on identical twins, or the study of symmetrical organs or appendages such as in the case of ophthalmic studies. Although this type of matching appears ideal for the purposes of comparison, analysis of the resulting data while ignoring the effect of intra-cluster correlation has been shown to produce biased results.^ This paper will explore the use of multilevel modeling of simulated binary data with predetermined levels of correlation. Data will be generated using the Beta-Binomial method with varying degrees of correlation between the lower level observations. The data will be analyzed using the multilevel software package MlwiN (Woodhouse, et al, 1995). Comparisons between the specified intra-cluster correlation of these data and the estimated correlations, using multilevel analysis, will be used to examine the accuracy of this technique in analyzing this type of data. ^