THE USE OF MULTIVARIATE ANALYSIS OF COVARIANCE IN EVALUATING THE COMPARABILITY OF LABORATORY DETERMINATIONS IN COOPERATIVE STUDIES

The role of clinical chemistry has traditionally been to evaluate acutely ill or hospitalized patients. Traditional statistical methods have serious drawbacks in that they use univariate techniques. To demonstrate alternative methodology, a multivariate analysis of covariance model was developed and applied to the data from the Cooperative Study of Sickle Cell Disease.^ The purpose of developing the model for the laboratory data from the CSSCD was to evaluate the comparability of the results from the different clinics. Several variables were incorporated into the model in order to control for possible differences among the clinics that might confound any real laboratory differences.^ Differences for LDH, alkaline phosphatase and SGOT were identified which will necessitate adjustments by clinic whenever these data are used. In addition, aberrant clinic values for LDH, creatinine and BUN were also identified.^ The use of any statistical technique including multivariate analysis without thoughtful consideration may lead to spurious conclusions that may not be corrected for some time, if ever. However, the advantages of multivariate analysis far outweigh its potential problems. If its use increases as it should, the applicability to the analysis of laboratory data in prospective patient monitoring, quality control programs, and interpretation of data from cooperative studies could well have a major impact on the health and well being of a large number of individuals. ^

Comparison of compliance and immune response of drug users on the standard and accelerated hepatitis B vaccine schedules - Houston, Texas

Despite the availability of hepatitis B vaccine for over two decades, drug users and other high-risk adult populations have experienced low vaccine coverage. Poor compliance has limited efforts to reduce transmission of hepatitis B infection in this population. Evidence suggests that immunological response in drug users is impaired compared to the general population, both in terms of lower seroprotection rates and antibodies levels.^ The current study investigated the effectiveness of the multi-dose hepatitis B vaccine and compared the effect of the standard and accelerated vaccine schedules in a not-in-treatment, drug-using adult population in the city of Houston, USA.^ A population of drug-users from two communities in Houston, susceptible to hepatitis B, was sampled by outreach workers and referral methodology. Subjects were randomized either to the standard hepatitis vaccine schedule (0, 1-, 6-month) or to an accelerated schedule (0, 1-, 2-month). Antibody levels were detected through laboratory analyses at various time-points. The participants were followed for two years and seroconversion rates were calculated to determine immune response.^ A four percent difference in the overall compliance rate was observed between the standard (73%) and accelerated schedules (77%). Logistic regression analyses showed that drug users living on the streets were twice as likely to not complete all three vaccine doses (p=0.028), and current speedball use was also associated with non-completion (p=0.002). Completion of all three vaccinations in the multivariate analysis was also correlated with older age. Drug users on the accelerated schedule were 26% more likely to achieve completion, although this factor was marginally significant (p=0.085).^ Cumulative adequate protective response was gained by 65% of the HBV susceptible subgroup by 12-months and was identical for both the standard and accelerated schedules. Excess protective response (>=100 mIU/mL) occurred with greater frequency at the later period for the standard schedule (36% at 12-months compared to 14% at six months), while the greater proportion of excess protective response for the accelerated schedule occurred earlier (34% at 6 months compared to 18% at 12-months). Seroconversion at the adequate protective response level of 10 mIU/mL was reached by the accelerated schedule group at a quicker rate (62% vs. 49%), and with a higher mean titer (104.8 vs. 64.3 mIU/mL), when measured at six months. Multivariate analyses indicated a 63% increased risk of non-response for older age and confirmed the existence of an accelerating decline in immune response to vaccination manifesting after 40 years (p=0.001). Injecting more than daily was also highly associated with the risk of non-response (p=0.016).^ The substantial increase in the seroprotection rate at six months may be worth the trade-off against the faster antibody titer decrease and is recommended for enhancing compliance and seroconversion. Utilization of the accelerated schedule with the primary objective of increasing compliance and seroconversion rates during the six months after the first dose may confer early protective immunity and reduce the HBV vulnerability of drug users who continue, or have recently initiated, increased high risk drug use and sexual behaviors.^

Prevalence and risk factors for human immunodefiency virus and hepatitis C virus co-infection among drug users in innercity neighborhoods in Houston, Texas

Background. There are 200,000 HIV/HCV co-infected people in the US and IDUs are at highest risk of exposure. Between 52-92% of HIV infected IDUs are chronically infected with HCV. African Americans and Hispanics bear the largest burden of co-infections. Furthermore HIV/HCV co-infection is associated with high morbidity and mortality if not treated. The present study investigates the demographic, sexual and drug related risk factors for HIV/HCV co-infection among predominantly African American injecting and non-injecting drug users living in two innercity neighborhoods in Houston, Texas. ^ Methods. This secondary analysis used data collected between February 2004 and June 2005 from 1,889 drug users. Three case-comparison analyses were conducted to investigate the risk factors for HIV/HCV co-infection. HIV mono-infection, HCV mono-infection and non-infection were compared to HIV/HCV co-infection to build multivariate logistic regression models. Race/ethnicity and age were forced into each model regardless of significance in the univariate analysis. ^ Results. The overall prevalence of HIV/HCV co-infection was 3.9% while 39.8% of HIV infected drug users were co-infected with HCV and 10.7% of HCV infected drug users were co-infected with HIV. Among HIV infected IDUs the prevalence of HCV was 71.7% and among HIV infected NIDUs the prevalence of HCV was 24%. In the multivariate analysis, HIV/HCV co-infection was associated with injecting drug use when compared to HIV mono-infection, with MSM when compared to HCV mono-infection and with injecting drug use as well as MSM when compared to non-infection. ^ Conclusion. HIV/HCV co-infection was associated with a combination of sexual and risky injecting practices. More data on the prevalence and risk factors for co-infection among minority populations is urgently needed to support the development of targeted interventions and treatment options. Additionally there should be a focus on promoting safer sex and injecting practices among drug users as well as the expansion of routine testing for HIV and HCV infections in this high risk population.^

Retrospective analysis of Methicillin-Resistant Staphylococcus Aureus (MRSA) skin and Soft-Tissue Infections (SSTI) in a university hospital emergency department (ED)

Study Objective: Identify the most frequent risk factors of Community Acquired-MRSA (CA-MRSA) Skin and Soft-tissue Infections (SSTIs) using a case series of patients and characterize them by age, race/ethnicity, gender, abscess location, druguse and intravenous drug-user (IVDU), underlying medical conditions, homelessness, treatment resistance, sepsis, those whose last healthcare visit was within the last 12 months, and describe the susceptibility pattern from this central Texas population that have come into the University Medical Center Brackenridge (UMCB) Emergency Department (ED). ^ Methods: This study was a retrospective case-series medical record review involving a convenience sample of patients in 2007 from an urban public hospital's ED in Texas that had a SSTI that tested positive for MRSA. All positive MRSA cultures underwent susceptibility testing to determine antibiotic resistance. The demographic and clinical variables that were independently associated with MRSA were determined by univariate and multivariate analysis using logistic regression to calculate odds ratios (OR), 95% confidence intervals, and significance (p≤ 0.05). ^ Results: In 2007, there were 857 positive MRSA cultures. The demographics were: males 60% and females 40%, with the average age of 36.2 (std. dev. =13) the study population consisted of non-Hispanic white (42%), Hispanics (38%), and non-Hispanic black (18.8%). Possible risk factors addressed included using recreational drugs (not including IVDU) (27%) homelessness (13%), diabetes status (12.6%) or having an infectious disease, and IVDU (10%). The most frequent abscess location was the leg (26.6%), followed by the arm and torso (both 13.7%). Eighty-three percent of patients had one prominent susceptibility pattern that had a susceptibility rate for the following antibiotics: trimethoprim/sulfamethoxazole (TMP-SMX) and vancomycin had 100%, gentamicin 99%, clindamycin 96%, tetracycline 96%, and erythromycin 56%. ^ Conclusion: The ED is becoming an important area for disease transmission between the sterile hospital environment and the outside environment. As always, it is important to further research in the ED in an effort to better understand MRSA transmission and antibiotic resistance, as well as to keep surveillance for the introduction of new opportunistic pathogens into the population. ^

SURVIVAL PREDICTION FOR BRAIN TUMOR PATIENTS USING GENE EXPRESSION DATA

Brain tumor is one of the most aggressive types of cancer in humans, with an estimated median survival time of 12 months and only 4% of the patients surviving more than 5 years after disease diagnosis. Until recently, brain tumor prognosis has been based only on clinical information such as tumor grade and patient age, but there are reports indicating that molecular profiling of gliomas can reveal subgroups of patients with distinct survival rates. We hypothesize that coupling molecular profiling of brain tumors with clinical information might improve predictions of patient survival time and, consequently, better guide future treatment decisions. In order to evaluate this hypothesis, the general goal of this research is to build models for survival prediction of glioma patients using DNA molecular profiles (U133 Affymetrix gene expression microarrays) along with clinical information. First, a predictive Random Forest model is built for binary outcomes (i.e. short vs. long-term survival) and a small subset of genes whose expression values can be used to predict survival time is selected. Following, a new statistical methodology is developed for predicting time-to-death outcomes using Bayesian ensemble trees. Due to a large heterogeneity observed within prognostic classes obtained by the Random Forest model, prediction can be improved by relating time-to-death with gene expression profile directly. We propose a Bayesian ensemble model for survival prediction which is appropriate for high-dimensional data such as gene expression data. Our approach is based on the ensemble "sum-of-trees" model which is flexible to incorporate additive and interaction effects between genes. We specify a fully Bayesian hierarchical approach and illustrate our methodology for the CPH, Weibull, and AFT survival models. We overcome the lack of conjugacy using a latent variable formulation to model the covariate effects which decreases computation time for model fitting. Also, our proposed models provides a model-free way to select important predictive prognostic markers based on controlling false discovery rates. We compare the performance of our methods with baseline reference survival methods and apply our methodology to an unpublished data set of brain tumor survival times and gene expression data, selecting genes potentially related to the development of the disease under study. A closing discussion compares results obtained by Random Forest and Bayesian ensemble methods under the biological/clinical perspectives and highlights the statistical advantages and disadvantages of the new methodology in the context of DNA microarray data analysis.

An analysis of the shortage of registered nurses (RNs) in acute care hospitals in El Paso, Texas

The purpose of this analysis of the shortage of Registered Nurses (RNs) in acute care hospitals in El Paso, Texas, was to evaluate twenty-two specific organizational and/or patient care unit (nursing unit) characteristics that effect the retention and turnover of professional nurses. Vacancy Rates were used to measure the level of the shortage in each hospital and nursing unit in the study. Vacancy Rates are a function of both RN retention and RN turnover. Seventy-three patient care units in five acute care hospitals were included in the study population.^ Fredrick Herzberg's motivational - hygiene theory was used to explain the types of characteristics or factors that can effect worker dissatisfaction. Dissatisfiers (hygiene factors) are those work place characteristics that influence workers to leave the job. The twenty-two potentially dissatisfying work place characteristics were either organizational or patient care unit specific in nature. The focus of the study was to evaluate high vacancy rates caused by both low retention of RNs and high turnover rates. Retention and turnover are a function of workers (RNs) not staying in their jobs, therefore hygiene factors were appropriate characteristics to study.^ Various multivariate analysis techniques were used to assess both the individual and combined effects of the hygiene factors on Vacancy Rates, Retention and Turnover. Results suggest that certain organizational and patient care unit characteristics are associated with and have a statistically significant effect on vacancy rates, and the retention and turnover of RNs. The type of Hospital was of particular interest in this regards. For-Profit facilities were less effected by most of the study variables than the Not-for-Profits. ^

Incidence, etiology and risk factors of neonatal seizures Harris County, Texas, 1992--1994

This study was conducted to determine the incidence and etiology of neonatal seizures, and evaluate risk factors for this condition in Harris County, Texas, between 1992 and 1994. Potential cases were ascertained from four sources: discharge diagnoses at local hospitals, birth certificates, death certificates, and a clinical study of neonatal seizures conducted concurrent with this study at a large tertiary care center in Houston, Texas. The neonatal period was defined as the first 28 days of life for term infants, and up to 44 weeks gestation for preterm infants.^ There were 207 cases of neonatal seizures ascertained among 116,048 live births, yielding and incidence of 1.8 per 1000. Half of the seizures occurred by the third day of life, 70% within the first week, and 93% within the first 28 days of life. Among 48 preterm infants with seizures 15 had their initial seizure after the 28th day of life. About 25% of all seizures occurred after discharge from the hospital of birth.^ Idiopathic seizures occurred most frequently (0.5/1000 births), followed by seizures attributed to perinatal hypoxia/ischemia (0.4/1000 births), intracranial hemorrhage (0.2/1000 births), infection of the central nervous system (0.2/1000 births), and metabolic abnormalities (0.1/1000 births).^ Risk factors were evaluated based on birth certificate information, using univariate and multivariate analysis (logistic regression). Factors considered included birth weight, gender, ethnicity, place of birth, mother's age, method of delivery, parity, multiple birth and, among term infants, small birth weight for gestational age (SGA). Among preterm infants, very low birth weight (VLBW, $<$1500 grams) was the strongest risk factor, followed by birth in private/university hospitals with a Level III nursery compared with hospitals with a Level II nursery (RR = 2.9), and male sex (RR = 1.8). The effect of very low birth weight varied according to ethnicity. Compared to preterm infants weighing 2000-2999 grams, non-white VLBW infants were 12.0 times as likely to have seizures; whereas white VLBW infants were 2.5 times as likely. Among term infants, significant risk factors included SGA (RR = 1.8), birth in Level III nursery private/university hospitals versus hospitals with Level II nursery (RR = 2.0), and birth by cesarean section (RR = 2.2). ^

Non-sexual household transmission of HCV infection: A cohort study

Objective. This study was designed to determine the prevalence and incidence of HCV infection among non-sexual household contacts of HCV-infected women and to describe the association between HCV infection and potential household risk factors in order to examine whether non-sexual household contact is a route of transmission for HCV infection. ^ Methods. A baseline prevalence survey included 409 non-sexual household contacts of 241 HCV-infected index women in the Houston area from 1994 to 1997. A total of 470 non-sexual household contacts with no evidence of HCV infection at baseline investigation were re-assessed approximately three years after baseline enrollment. Information on potential risk factors was collected through face to face interviews and blood samples were tested for anti-HCV with ELISA-2 and Matrix/RIBA-2. The relationships between HCV infection and potential risk factors were examined by using univariate and multivariate logistic regression analyses. ^ Results. The overall prevalence of anti-HCV positivity among 409 non-sexual household contacts was 4.4%. The highest prevalence of anti-HCV was found in parents (19.5%), followed by siblings (8.1%) and other relatives (5.6%); the children had the lowest prevalence of anti-HCV (1.2%). The univariate analysis showed that IDU, blood transfusion, tattoos, sexual contact with injecting drug users, more than 3 sexual partners in a lifetime, history of a STD, incarceration, previous hepatitis, and contact with hepatitis patients were significantly associated with HCV infection, however, sharing razors, nail clippers, toothbrushes, gum, food or beds with HCV-infected women, and history of dialysis, health care job, body piercing, and homosexual activities were not. Multivariate analysis found that IDU (OR = 221.7 with 95% CI of 22.8 to 2155.7) and history of a STD (OR = 11.7 with 95% CI of 1.2 to 113.1) were the only variables significantly associated with HCV infection. No such associations remained for other risk factors. The three-year cumulative incidence of anti-HCV among 352 non-sexual household contacts of HCV-infected women was zero. ^ Conclusion. This study has provided no evidence that non-sexual household contact is a likely route of transmission for HCV infection. The risk of sharing razors, nail clippers, toothbrushes, gum, food and/or beds with HCV-infected women is not evident and has not been shown to be the likely mode for HCV spread among family members. This study does suggest that IDU is the likely route of transmission for most HCV infection. Association also has been shown independently with a history of STD. The prevalence of anti-HCV among non-sexual household contacts was low. Exposure to common parenteral risk factors and sexual transmission between sexual partners may account for HCV spread among household members of HCV-infected persons. ^

The critical role and regulation of transcription factor FoxM1 in gastric cancer development and progression

The mammalian Forkhead Box (Fox) transcription factor (FoxM1) is implicated in tumorgenesis. However, the role and regulation of FoxM1 in gastric cancer remain unknown.^ I examined FoxM1 expression in 86 cases of primary gastric cancer and 57 normal gastric tissue specimens. I found weak expression of FoxM1 protein in normal gastric mucosa, whereas I observed strong staining for FoxM1 in tumor-cell nuclei in various gastric tumors and lymph node metastases. The aberrant FoxM1 expression is associated with VEGF expression and increased angiogenesis in human gastric cancer. A Cox proportional hazards model revealed that FoxM1 expression was an independent prognostic factor in multivariate analysis. Furthermore, overexpression of FoxM1 by gene transfer significantly promoted the growth and metastasis of gastric cancer cells in orthotopic mouse models, whereas knockdown of FoxM1 expression by small interfering RNA did the opposite. Next, I observed that alteration of tumor growth and metastasis by elevated FoxM1 expression was directly correlated with alteration of VEGF expression and angiogenesis. In addition, promotion of gastric tumorigenesis by FoxM1 directly and significantly correlated with transactivation of vascular endothelial growth factor (VEGF) expression and elevation of angiogenesis. ^ To further investigate the underlying mechanisms that result in FoxM1 overexpression in gastric cancer, I investigated FoxM1 and Krüppel-like factor 4 (KLF4) expressions in primary gastric cancer and normal gastric tissue specimens. Concomitance of increased expression of FoxM1 protein and decreased expression of KLF4 protein was evident in human gastric cancer. Enforced KLF4 expression suppressed FoxM1 protein expression. Moreover, a region within the proximal FoxM1 promoter was identified to have KLF4-binding sites. Finally, I found an increased FoxM1 expression in gastric mucosa of villin-Cre -directed tissue specific Klf4-null mice.^ In summary, I offered both clinical and mechanistic evidence that dysregulated expression of FoxM1 play an important role in gastric cancer development and progression, while KLF4 mediates negative regulation of FoxM1 expression and its loss significantly contributes to FoxM1 dysregulation. ^

The prevalence of bulimia nervosa and bulimic behaviors among university students in Texas

This cross-sectional study examined by questionnaire the prevalence of bulimia nervosa and bulimic behaviors in a sample of 1175 undergraduate students enrolled in two state-supported universities in Texas. In one university, the student population was predominantly white; in the other, it was predominantly black. Fifty-nine percent of the respondents were female and 41% were male. Information regarding age, sex, ethnicity, college major, college year, marital status, housing arrangements, religion, socioeconomic status, height, weight, dieting behaviors, and family history of alcoholism, drug abuse, and depression was collected. Bulimia status was assessed using the Revised Bulimia Test (BULIT-R), which is based on the DSM-III-R criteria for bulimia nervosa. Only 1.3% of the females and 0.4% of the males were classified as having bulimia nervosa. The prevalence of bulimic behaviors was considerably higher; 6.4% of the females and 3.6% of the males were classified as having bulimic behaviors. Univariate analysis showed the following factors to be significantly associated with bulimic behaviors: female gender, single marital status, high BMI, a family history of alcoholism, drug abuse, or depression, and certain dieting behaviors. In the present study, ethnicity did not prove to be a significant factor associated with bulimia nervosa or bulimic behaviors. Multivariate analysis showed that, in comparison to normal/underweight individuals, the odds of having bulimic behaviors for severely overweight subjects were 2.23 (95% CI: 1.43, 3.50). Students who were dieting at the time of the study were 3.22 times (95% CI: 2.05, 5.06) as likely to have bulimic behaviors as were students who had never dieted. This study concludes there is a need to distinguish between bulimia nervosa and bulimic behaviors when estimating prevalence of a population. ^

Behavioral facets of cancer chemoprevention trials

Few studies have explored factors related to participation in cancer chemoprevention trials. The purpose of this dissertation was to conduct investigations in this emerging field by studying aspects of participation at three phases of cancer chemoprevention trials: at enrollment, during a placebo run-in period, and post-trial. In all three studies, subjects had a history of cancer and were at high risk of recurrence or second primary tumors.^ The first study explored correlates of enrollment in a head and neck cancer chemoprevention trial by comparing participants and eligible nonparticipants. Of 148 subjects who met the trial's preliminary eligibility criteria, 40% enrolled. In multivariate analysis, enrollment was positively associated with being male (OR 2.36) and being employed (OR 2.73). The most commonly cited reason for declining participation among nonparticipants was transportation.^ The second study examined outcomes of an eight-week placebo run-in period in a head and neck cancer chemoprevention trial. Of 391 subjects, 91.3% were randomized after the run-in. Adherence to drug capsules ranged from 0% to 120.3% (mean $\pm$ SD, 95.8% $\pm$ 15.1). In multivariate analysis, the main variable predicting run-in outcome was race; white subjects were 3.45 times more likely to be randomized than non-white subjects. Subjects with Karnofsky scores of 100 were 2.13 times more likely to be randomized than were subjects with lower scores.^ The third study used post-trial questionnaires to assess subjects' (n = 64) perceptions of participation in a cancer chemoprevention trial. The most highly rated trial benefit was the perception of potential colon cancer prevention, and the most troublesome barrier was erroneous billing for study visits. Perceived benefits were positively associated with interest in participating in future trials of the same (p = 0.05) and longer (p = 0.02) duration, and difficulty with trial pills and procedures was inversely related to interest in future placebo-controlled trials (p = 0.01).^ These are among the first behavioral studies to be completed in the rapidly growing field of cancer chemoprevention. Much work has yet to be done, however, to advance our understanding of the complex issues relating to chemoprevention trial participation. ^

Factors influencing participation in individuals with disability: A secondary data analysis

Individuals with disabilities face numerous barriers to participation due to biological and physical characteristics of the disability as well as social and environmental factors. Participation can be impacted on all levels from societal, to activities of daily living, exercise, education, and interpersonal relationships. This study evaluated the impact of pain, mood, depression, quality of life and fatigue on participation for individuals with mobility impairments. This cross sectional study derives from self-report data collected from a wheelchair using sample. Bivariate correlational and multivariate analysis were employed to examine the relationship between pain, quality of life, positive and negative mood, fatigue, and depression with participation while controlling for relevant socio-demographic variables (sex, age, time with disability, race, and education). Results from the 122 respondents with mobility impairments demonstrated that after controlling for socio-demographic characteristics in the full model, 20% of the variance in participation scores were accounted for by pain, quality of life, positive and negative mood, and depression. Notably, quality of life emerged as being the single variable that was significantly related to participation in the full model. Contrary to other studies, pain did not appear to significantly impact participation outcomes for wheelchair users in this sample. Participation is an emerging area of interest among rehabilitation and disability researchers, and results of this study provide compelling evidence that several psychosocial factors are related to participation. This area of inquiry warrants further study, as many of the psychosocial variables identified in this study (mood, depression, quality of life) may be amenable to intervention, which may also positively influence participation.^