Ph.D. - Biomedical Sciences (Biostatistics - Epidemiology)

Permanent URI for this collectionhttps://hdl.handle.net/10125/1930

Browse

Now showing 1 - 15 of 15

Epidemiology of HIV-associated risk factors and acquisition of HIV among high-risk women in southern Vietnam
(University of Hawaii at Manoa, 2004) Komatsu, Ryuichi
Background: HIV epidemic is rapidly increasing in Vietnam. Although injecting drug use is still the major mode of transmission countrywide, sexual transmission is becoming increasingly important, particularly in the southern provinces. Objectives: To measure HIV-associated risk factors and HIV prevalence among high-risk women in southern Vietnam. Methods: Two cross-sectional surveys were conducted in 2000 and 2002 among high-risk women in two southern provinces bordering with Cambodia. After geographical and social mapping was conducted to develop a sampling frame, 803 women in 2000 and 801 women in 2002 were interviewed for HIV associated risk factors. Serum specimens were also collected for HIV testing. Major Results: Condom use with the last commercial client remained at the same level at 85% from 2000 to 2002 (p>0.05). However, more respondents in 2002 (71.8%) worked everyday than in 2000 (40.9%) (p<0.01) and the percentage who had seven or more clients increased from 7.5% to 18.7% (p<0.01). Consistent condoms use with the non-commercial partners declined from 35.5% to 12.1% (p<0.0l). Factors associated with condom use varied based on the type of partners. Although still fairly low compared with other cities in Vietnam, injecting drug use in these southern provinces increased from 0.4% to 3.8% (p<0.01). No change was observed in HIV prevalence from 4.0% in 2000. Positive HIV serological status was firmly associated with visits to Cambodia (odds ratio=15.9) than with injecting drug use (odds ratio=5.5) in this population. Discussion: Although condom use was high, these results imply that sexual transmission of HIV is important in this population. HIV prevalence was more strongly associated with high-risk sex work in Cambodia than injecting drug use.
An epidemiological study of morbidity, mortality and use of primary health care services among mothers and children in three demographic areas of Mozambique
(University of Hawaii at Manoa, 1992) Oliver, Charles Whitney Jr.
Studies of morbidity, mortality, and use of primary health care (PHC) were conducted in Zambezia Province, Mozambique. Formerly a leading agricultural resource of the region, Zambezia was particularly traumatized by a generation of war in what remains one of the poorest countries of the developing world. Cross-sectional surveys were conducted during 1990-1991. Subjects were 3,232 mothers and children (under five years) from three areas: centros de acomodacao, camps for persons displaced by war [0], a rural town [R], and the periurban provincial capital [U]. Household interviews were complemented by medical examinations and laboratory analyses. Indices of socio-economic status (SES), knowledge and use of PHC services were calculated from composite scores. An estimated rate of cumulative child mortality (CCMR) was derived from birth histories in lieu of lack of accurate census data. Laboratory and anthropometric measures were used to determine prevalence rates and levels of intensity for P. falciparum malaria and helminthic infections, anemia, and low nutritional status. Morbidity scores were calculated from a composite of these findings. Results of the study delineate significant differences in levels of morbidity, child mortality, SES, and knowledge and use of PHC services among these groups. Aggregate and group data were further analyzed utilizing stepwise logistic regression models developed for all major variables. P falciparum malaria is holoendemic in the study areas, but was observed more often among women from [U] (OR=4.6, P < 0.001). Similar results were found for children. Malaria was associated with higher parity, and with hookworm among women and children. Hookworm infections were commonly found in all groups, although more prevalent among women from [U] and [0], and among women than children. Children from [U] had twice the prevalence of hookworm compared with the other groups. Hookworm was associated with anemia in women, and older age in children. Trichuriasis was more prevalent among women and children from [U], while strongyloidiasis more common in the rural groups. Anemia (Hb < 11.0 g dl-1 ) was commonly observed in all groups, sharing a similar, albeit higher, prevalence distribution with hookworm. Mean prevalence for children was below 8.0 g dr-1 , and below 10.0 g dl-1 for women. Among women, moderate anemia (Hb < 9.0 g dl^-1 ) was associated with hookworm, pregnancy, and low knowledge of PHC. It was associated with higher CCMR (≥ 200), age (< 2 years), lower nutritional status, and recent febrile episode among children. Prevalence of severe anemia (Hb < 7.0 g dl^-1 ) was very high (even for Sub-Saharan Africa), and was associated with malaria among children from [U]. Malaria, helminthic infection, and anemia were found to be associated between mothers and children suggesting common etiologies. Prevalence of wasting (weight-for-height Z scores < -2) did not differ among children from all groups. Wasting was found to be associated with anemia, age (< 2 years), and in the [U] group with low maternal Body Mass Index (<18.0) (OR=3.0S, P < 0.001). Growth stunting (height-for-age Z scores < -2) was more prevalent among rural children (OR=3.77, p < 0.001), associated with anemia, age ( > 2 years), and low SES. In [U] children underweight (weight-for-age Z scores < -2) or stunted were found associated with low birth weight (<2.5 kg), indicating failure in catch up growth during early childhood. The peak age for underweight and wasting among the urban children was a year later than the rural groups. Women and children from [U] had higher morbidity scores (OR = 2.52, P < 0.001) and (OR = 1.71, P < 0.001) respectively. A high morbidity score was associated with pregnancy, neonatal death, and low knowledge of PHC among women, while related to growth stunting, and age (> 2 years) in children. Women with higher CCMR ( ~ 200) were associated with low SES, low knowledge and use of PHC. Also, it was more likely that their child had severe anemia. The CCMR was lower in [U] than the rural groups, however interpretation remains limited since it could reflect a wide range of risk factors beyond the scope of this study, e.g. from exposure to the effects of war among older women in the rural groups to exposure to improved antenatal facilities among younger women in the urban group. Another unusual finding was that women from [U] had lower SES more often than group [R] (OR=1.97, P < 0.001). Lower SES scores were found associated with maternal anemia, a neonatal death, age (> 24 years), and low knowledge of PHC. Knowledge of PHC was low in all groups, although lower in [D] (OR=2.34, P < 0.001). A low score was found associated with anemic and underweight children, and low use of PHC services. Similarly, [D] had more women with lower use of PHC services. Low use of PHC was found associated with higher CCMR, growth stunting, and low knowledge of PHC. The finding that morbidity and SES levels were worse in [U] than [R] may be explained by the supposition that much of the urban area has become marginalized as a result of war-induced migration into the remains of an abandoned, formerly Portuguese urban enclave. Thus as a result of the capital's inability to cope with this massive influx, the situation in many periurban bairros more resembled that found in the accommodation camps than the traditional rural town. Results of logistic regression analyses revealed that individual morbidity indicators, e.g. moderate anemia, to be more informative than composite scores of morbidity or PHC. The development of sensitive measures of mortality in Mozambique are precluded until an accurate census can be undertaken, once peace is finally established throughout the country. The complex web of associations elucidated from morbidity indicators in comparison with PHC scores suggest that the inclusion of inexpensive morbidity studies in development of maternal and child care programs may be of greater utility than PHC interviews alone. The numerous correlations between maternal and child morbidity indicators suggest common etiologies, and may prove useful in community diagnosis if prospective studies prove confirmatory. Innovative and cost-effective triage protocol is needed to be employed at every opportunity for PHC intervention, e.g. when women attending antenatal clinics are found with anemia, hookworm, or malaria, her youngest child could be treated presumptively as well. The observation of higher prevalences of morbidity (both individual and composite indicators) and lower levels of SES among the periurban group merits further study of these phenomena in similar areas of the developing world. The cumulative impact of a generation of war and socioeconomic devastation continues to adversely affect the delivery of PHC services in rural Mozambique. However, the consequences of poverty and massive internal migration have exacerbated the precarious balance of health in periurban areas, and new strategies are needed to prevent large scale pandemics (e.g. AIDS and drug-resistant malaria) in the rapidly emerging periurban communities of the developing world in the future.
Seroepidemiological study of human antibodies to the major merozoite surface coat precursor protein of Plasmodium falciparum (gp195) from a hyperendemic area of the Philippines
(University of Hawaii at Manoa, 1990) Kramer, Kenton Jay
The seroepidemiology of the major merozoite surface coat antigen (gp195) was investigated in a human population living on the island of Palawan, Philippines. A strain of Plasmodium falciparum (Pf857), isolated from Palawan, was used as a source of gp195 antigens. Malaria surveys were conducted in the Napsan region of Palawan during June and September 1986. In total, one thousand four-hundred and fifty-eight individuals were screened for anti-gp195 antibodies using an enzyme immunoassay (EIA). The results of the June and September surveys were comparable. Approximately 90% of the people tested had anti-gp195 antibodies. The lowest seroprevalence (approximately 78%) was found in the 0-4 year old age group. By age group 10-19 years the seroprevalence of anti-gp195 antibodies had plateaued near 90%. Anti-gp195 antibody titers were determined from 59 randomly selected individuals using an EIA. The anti-gp195 antibody titers ranged from zero to 1/102,400. Anti-gp195 antibody titers increased with age (r = 0.355, p < 0.01). The lowest antibody titers were found in the 0-4 year old age group. The 5-9 year old age group had anti-gp195 antibody titers comparable with the older age groups. similar results were also obtained using two yeast recombinant polypeptides based on the gp195 amino acid sequence. This suggests that the recombinant polypeptides share some B cell epitopes with the native protein. Immunoblotting experiments, using gp195 polypeptides, revealed that all the serum samples tested had antibodies to the gp195 precursor (195-kDa). However, only individuals with anti-gp195 EIA titers greater than 1/400 had antibodies against the 42-kDa processing fragment. This implies that this fragment is either poorly immunogenic or partially hidden from the immune system. This finding is important and may explain Why immunity to P. falciparum is acquired only after repeated exposures. Future work should concentrate on the role of the 42-kDa processing fragment in immunity to P. falciparum. No significant differences were found in the anti-gp195 EIA titers or immunoblot patterns against gp195 between individuals with a positive P. falciparum blood film and those with a negative slide. This implies that if antibodies are important in immunity, the relevant antibodies are probably recognizing a limited number of specific gp195 epitopes.
The impact of mammography utilization on breast cancer incidence in Hawaii
(University of Hawaii at Manoa, 1996) Maskarinec, Gertraud
This research is an ecologic study investigating the association between mammography utilization and breast cancer incidence in Hawaii with the hypothesis that geographic areas with high mammography rates have higher breast cancer incidence rates than geographic areas with low mammography rates. Insurance claims for mammograms received during 1992/93 were combined with breast cancer incidence data from the state-wide Hawaii Tumor Registry, the 1990 Census ZIP File, and death records. Insurance claims data were obtained from four private and three public health plans and covered approximately 85% of the state's female population age 40 years old and older. Age-specific breast cancer incidence rates for each of the 79 ZIP code areas were regressed on mammography rates and selected demographic variables using both multiple linear and multiple logistic regression. An estimated 42% of women 40 years old and older had received at least one mammogram during the study period 1992 to 1993, with the highest rate (45%) in women 50 to 64 years old. Overall, 23% of the variation in age-specific breast cancer incidence was associated with mammography utilization, another 23% with age, and 4% with educational attainment. The relation between mammography use and breast cancer incidence was strongest for women 50 to 64 years old and weakest for women 40 to 49 years old. Mammography utilization predicted 17% of the variance in localized disease, but very little of the variance of in situ or advanced disease. Neither mammography utilization nor demographic variables were found to be good predictors of geographic variations in breast cancer mortality in Hawaii. In summary, geographic areas in Hawaii with high mammography utilization experienced higher breast cancer incidence rates than areas with low mammography utilization. The size of this effect was consistent with the increase in breast cancer rates observed since the mid-eighties, supporting the hypothesis that the increase in breast cancer incidence is attributable to screening and early detection. However, the long term increase in breast cancer incidence requires alternate explanations.
Cause-specific failure probability with covariate effects
(University of Hawaii at Manoa, 1996) Choi, Kwisook
When infectious diseases were the main killers, elimination of their effects on mortality rates was possible. However, in modern society in which chronic diseases are major causes of death, elimination of disease effect is no longer relevant in estimating mortality rates. The cause-specific failure probability (CSFP) under the unrealistic assumption of elimination of other causes (net probability) is always larger than that under the practical situation where all other causes act simultaneously (crude probability), given competing risks. The proportional hazards model is fit to estimate the CSFP when covariate effects are considered. Fitting the model is performed simply by treating the patients who fail due to other causes as censored observations. When the coefficient of a covariate is positively related to the hazards of other causes, the estimate of net probability is increased over that of crude probability according to the increasing value of the covariate. The more the association of a covariate is related to other causes, the larger the difference is. However, the interpretation is complicated because the probability is related to one or more of the cause-specific hazards, and these hazards are also influenced by covariates. If the proportionality assumption of Cox's model is violated for an explanatory variable, stratification for the variable is desirable and CSFP is estimated in each stratum. If the proportional hazards model fits selected causes, strata, or time zones, a combination of non-parametric and semiparametric hazard and survival functions can be used to estimate the CSFP. An illustrative example is given for prostate cancer patients in Hawaii.
A study of determinants of malaria in Kelantan, Malaysia
(University of Hawaii at Manoa, 1995) Bjorge, Steven Scott
The candidate conducted operational research, in collaboration with the Vector-Borne Diseases Control Program (VBDCP), Kelantan, Malaysia, to analyze determinants of malaria incidence at different geographic scales: the state (inter-annual and intra-annual rainfall, and other seasonal events); the district (age- and sex-specific risk groups in relation to blood-film collection methods: Passive Case Detection [PCD], Active Case Detection [ACD], Investigation Surveys [INV]); and the village/individual domicile (environmental and behavioral variables). Routine operational data from the VBDCP were used as far as possible. Statewide statistics showed clear association of malaria with intra-annual rainfall variation but in a mixed correlation: high or low rainfall reduce malaria incidence, while moderate rainfall increases incidence. Inter-annual EI Nino Southern Oscillation events, which cause drought in southeast Asia, were associated with reduced malaria incidence. Rubber price and production, and fruit seasons could not be conclusively associated with incidence. All malaria blood-films examined during 1991 in Gua Musang district were entered to a database permitting age- and sex-stratified analysis of the populations sampled by PCD, ACD, and INV. Relative to the census population, the blood-film collection over-samples children and young adult males, but these risk groups still have higher incidence. PCD was most efficient in detecting malaria cases, but ACD and INV were most effective in detecting falciparum gametocytemia. A case-control study of behavioral and environmental determinants of malaria (largely taken from the routine case investigation form) in Jerek village proved inconclusive due to small sample size, resulting from low malaria transmission.
An exploratory comparison of vertebral fracture prevalence and risk factors among native Japanese, Japanese-American, and Caucasian women
(University of Hawaii at Manoa, 1994) Huang, Chün
In this cross-sectional population-based study, the prevalence of vertebral fractures in elderly women was compared among native Japanese in Hiroshima, Japanese-Americans in Hawaii, and North American Caucasians in Minnesota. Compared with Japanese-American women, the age-adjusted odds ratios for native Japanese women were significantly and consistently greater than 1.0 (range from 1.6 to 2.6, depending on fracture definition), while the age-adjusted odds ratios for Caucasian women living in Minnesota were closer to 1.0 (range from 0.5 to 1.5, depending on fracture definition). These data indicated that the age-adjusted overall prevalence of vertebral fracture among Japanese-American women was quite different from the prevalence in Japan, but more similar to the prevalence in the U.S., suggesting non-genetic factors may have some impact on vertebral fractures. Spine bone mineral density (BMD), a major predictor of vertebral fracture prevalence in this study, was found to be lower among native Japanese than among Japanese-American women even after adjusting for age. This difference was mainly due to the differences in body size and menstrual history between the two populations. On the average, native Japanese women were shorter and lighter, and tended to have a later menarche, an earlier menopause, and a shorter period between menarche and menopause. In linear regression analysis, age, weight and menstrual history (age at menopause or years between menarche and menopause) were found to be significantly associated with BMD, and thus might contribute to fracture risk indirectly through their effects on BMD. However, this study also shows that age and menstrual history provided complementary information about fracture risk and explained additional difference in fracture prevalence after adjusting for BMD, suggesting that BMD is a major but not a sole risk factor for vertebral fractures. Age-related and menopause-related mechanisms may also play an important role in spine fracture independent of BMD. The observed differences in vertebral fracture prevalence and related risk factors between native Japanese and Japanese-American women may be evidence for environmental effects, such as changes in nutrition and lifestyle.
A nested case-control study of dietary and serum antioxidant exposures and the risk of developing Parkinson's disease
(University of Hawaii at Manoa, 1994) Grandinetti, Andrew
Free radical oxidation may be involved in the etiology of idiopathic Parkinson's disease (IPD).A nested case-control study of the association of IPO with serum and dietary antioxidants was conducted in the Honolulu Heart Program (ID-IP) cohort of 8,006Japanese men. Eighty-four definite IPO cases occurring in the HHP cohort were verified by autopsy, record review, or direct neurological examination, and age-matched with 336 controls. Dietary and serum antioxidant data for this study were obtained from the first HHP cohort examination (Exam I) in 1965-1968. Conditional logistic regression was used to estimate odds ratios (ORs) for each variable, and as a test of trend for ordinal data and continuous data collapsed into quartiles. The adjusted odds ratio (aOR) relating IPO occurrence to vitamin E intake greater than the Recommended Dietary Allowance (RDA)was not significant (aOR=O.64, 95 percent a 0.35-1.17). The test for trend was also not significant. Intakes of several foods with high vitamin E content identified in a 24 hour dietary recall in 196>1968 were inversely associated with IPD occurrence, but only legume consumption was significantly associated (aOR =0.27, 95 percent CI 0.09-0.78). IPD occurrence was, however, associated with vitamin C intake above the RDA (OR =1.74,95 percent a 1.07-2.85). The association was attenuated by adjustment for smoking and coffee drinking (aOR =1.51, 95 percent a 0.90-2.52), and the test for trend was also not significant after adjustment for smoking and coffee drinking. IPD was also inversely associated with consumption of a western diet (aOR =0.6095 percent a 0.35-1.01) when compared with a Japanese diet, but no significant trend was detected after adjustment for smoking and coffee drinking. IPD was also associated with serum. uric acid (VA) concentration (aOR = 0.82, 95 percent a 0.69-0.99). The odds ratio for VA above the median (OR=O.60, 95 percent a 0.37-0.97) was only slightly attenuated after adjustment for smoking history and coffee consumption (aOR = 0.66, 95 percent a 0.41-1.08). IPD may be associated with prior dietary factors, but the relationship may be complex and difficult to measure retrospectively using epidemiologic techniques.
Maternal education and infant mortality in Thailand: comparison between the proportional hazards models with multiplicative and additive risk functions
(University of Hawaii at Manoa, 1995) Boosaba Sanguanprasit
This study has two objectives: (1) to compare the fit of the proportional hazards models with the multiplicative risk function to the one with the additive risk function by applying both functional forms to infant mortality data, and (2) to investigate the association between maternal education and other socioeconomic, demographic, environmental and health care factors in relation to infant mortality in Thailand. The data, used in this study, came from the 1987 Thailand Demographic and Health Survey (1987 TDHS). Results showed that the proportional hazards model with the multiplicative risk function fitted the infant mortality data in Thailand better than the additive one. The investigation of the associations between maternal education and its associations with other factors mentioned above in relation to infant mortality suggested that maternal education had a significant negative effect numerically on infant mortality even when all other variables under study were controlled. The education was strongly associated with family economic condition, place of residence, toilet facilities, and birth order greater than one with short preceding birth intervals. There was suggestion of interaction effects between maternal education, on the one hand, and place of residence, birth order greater than one with short preceding birth interval, and health care utilization, on the other, on infant mortality. The effects of these factors were significantly reduced if maternal education was increased from lower than secondary to at least secondary level. Suggestions are offered to reduce infant mortality in Thailand as follows: 1. The compulsory education should be strengthened and controlled to ensure that all graduates from school could read with comprehension. 2. Different governmental departments, such as education, public health, economic, community development, should take part in planning for reducing infant mortality. 3. Safe drinking water supply should be launched and carried out by local authorities; special attention should be paid for those living in urban areas, where safe water supplies were not accessible. 4. Breastfeeding program should be initiated to encourage working mothers to breastfeed their children during working hours.
Epidemiology of malaria in Punjab, Pakistan: a case study in a rural community near Lahore
(University of Hawaii at Manoa, 1985) Suleman, Mohammad
Assessment of the comparability of frequency and quantitative dietary intake measurements for epidemiologic studies of diet-disease associations
(University of Hawaii at Manoa, 1982) Chu, Susan Ying
Recent epidemiologic studies of chronic diseases have necessitated the development of dietary methods that can be used with large population samples, at relatively low cost, while still yielding valid results. One of the simplest approaches for measuring dietary intakes is to ask only about the frequency with which foods are eaten over a specified time period. The advantages of the frequency dietary methods are obvious. They can be administered cheaply and quickly to large numbers of people, and the numerous problems associated with the collection of quantitative dietary data are avoided. A number of studies reported in the literature have relied on frequency dietary data, occasionally translated into quantitative values by statistical conversions, to investigate possible associations between diet and disease. However, it is not known whether frequency dietary data can be used in place of the more laboriously collected quantitative dietary data without affecting the validity of the results. The few attempts to validate frequency dietary methods against traditional quantitative dietary methods (e.g., Burke dietary histories, food records) have been limited by the size of the sample and cursory data analyses. The main objective of the present study was to determine whether frequency dietary intake data can substitute for quantitative intake data in epidemiologic studies without affecting the diet-disease associations measured. This was done by determining the extent of agreement between individual frequency and quantitative dietary intake assessments collected from 342 male subjects participating in a case-control study concerning diet and prostate cancer risk. The dietary information was obtained using a recall interview method designed to assess the subject's past, usual dietary habits. The questionnaire consisted of a list of 113 foods, selected to cover the major sources of dietary fat, vitamin A, zinc, and cadmium. The, frequency and quantitative amount of intake were recorded separately for each food item eaten. Quantitative amounts were estimated using photographs of small, medium, and large portions of each item. Since the precision of data required varies according to the objectives of a particular investigation, the extent of agreement between frequency and quantitative measurements was examined for each dietary factor (44 food items, 20 food groups, and 8 nutrients were evaluated), at both the group and individual level, and with regard to the specific study objective. If the study objective involves the comparison of the intakes of individuals or groups, e.g., case-control, cohort, or aggregate correlational studies, then relative agreement is sufficient to determine the interchangeability of frequency and quantitative dietary intake measurements. In other words, the correspondence between the two measurements does not have to be exact, but it must be consistent among different subjects (or groups) to allow the substitution of frequency for quantitative dietary data without changing the measured diet-disease association. If, however, information on the actual amount consumed is required, e.g., in metabolic balance studies, then absolute agreement (intakes based on the two measurements correspond exactly in magnitude) is necessary to demonstrate interchangeability between frequency and quantitative intake data. At the aggregate level, frequency and quantitative dietary measurements were found not comparable for estimating the exact amount (absolute agreement) of mean dietary intakes of food items, food groups, or nutrients. There was a moderate degree of relative agreement between the mean intakes of food items (the two approaches ranked five ethnic groups on mean intakes equally (τ= 1.0) for 43% of the food items tested), but not for food groups (τ = 1.0 for 15% of the food groups tested). For the nutrients, relative agreement between mean intakes based on the two measurements was even lower. This suggests that frequency and quantitative dietary measurements can be used interchangeably to rank groups by mean intakes of certain food items (e.g., in ecological analyses), but not of food groups or nutrients. At the individual level, there was essentially no absolute agreement between the two types of dietary measurements for any dietary ,factor. There was a moderate degree of relative agreement between individual frequency and quantitative intake measurements, but because of the somewhat limited nature of the quantitative measurements collected, the findings were interpreted conservatively. In general, the extent of agreement was insufficient for one to be confident that frequency intake data can replace quantitative data without affecting the statistical findings. This suggests caution in interpreting the findings of epidemiologic studies that have used frequency of intake data to estimate quantitative dietary intakes, especially when individual intakes are compared. Frequency dietary data, however, may be useful in exploratory stages of investigating diet-disease associations, especially when the dietary hypotheses concern specific food items.
Health hazard appraisal for Hawaii
(University of Hawaii at Manoa, 1981) Tseng, Wen-hsiung Robert
Age-parity-duration-specific measures of fertility: application to China
(University of Hawaii at Manoa, 1992) Ng, Siu-Man
Maximizing the use of blocking in record linkage: theory and simulation
(University of Hawaii at Manoa, 1991) Khan, Mahmudul Huq
Two data files can be compared only if they have common variables. Pairs are formed by the combination of two records, one from each file. The actual status of a pair of records is either a match or a nonmatch, but is usually unknown. Record linkage methodology is used to make a decision on the actual status of each observed pair. Decision-errors are committed whenever the actual status and the decision are different. The new decision-rule is based on minimizing the sum of these errors of misclassification. Discrete discriminant analysis has been used to discriminate between match and nonmatch pairs. It is assumed that there are chance factors associated with any pair of records belonging to the sets of match and nonmatch pairs. A high value of the chance factor implies that the reliability of the pair belonging to the corresponding set is low. This is the basic premise in this research. Two discriminant scores are computed from the conditional probabilities of the observed values among match and nonmatch pairs, and the true proportion of match and nonmatch pairs. A pair is decided as a link or a nonlink, depending on which score is larger. The discriminant scores are estimated by using the chance probabilities and applying Bayes' theorem on conditional probability. In order to reduce the number of unnecessary comparisons, the files are often blocked by one or more variables. A new method for choosing the best blocking scheme has been proposed in this study. Some of the actual match pairs may be missed because they fall in different blocks. The idea is to select the scheme for which the failure to pick up the match pairs is the least. In comparison to some of the commonly used linkage procedures, the new method is statistically sound, yet it only requires the frequency counts of the variables in the files. Simulated experiments using the new method have been very successful. without any data error the results were 100% correct. Even with 10% data errors 80% of the true matches were correctly decided.
Pregnancy intendedness in Hawaiʻi: findings from the Hawaiʻi pregnancy risk assessment monitoring system, 1999
(University of Hawaii at Manoa, 2002-12) Sato, Alvin H.; Baruffi, Gigliola; Biomedical Sciences
The purpose of this study was to investigate the relationship between unintended pregnancies and demographic factors in Hawaiʻi, and to investigate the relationship between unintended pregnancies and violence in Hawaiʻi. This study used an observational, cross-sectional probability sample drawn women from June 1999 through December 1999. Women who were selected received Hawaiʻi Pregnancy Risk Assessment Monitoring System (PRAMS) questionnaires generated from the Hawaiʻi State Department of Health. The main variables from the PRAMS questionnaire investigated in the study were maternal pregnancy intendedness, physical abuse 12 months before pregnancy and physical violence during pregnancy. Controlling variables investigated were maternal age, maternal education, marital status, geographical location, maternal ethnicity, parity status, Quest status, source of prenatal care, and contraceptive use. This dissertation supports the proposed ecological conceptual framework for understanding unintended pregnancies and physical violence during and around the time of pregnancy. Using this model, it could be hypothesized that women with unintended pregnancies tend to live in less stable environments and therefore unintended pregnancies and physical violence could be part of the same social problems. In general, bivariate analysis showed unintended pregnancies were significantly associated with SES factors. However, controlling for all variables in the model, only maternal age, parity status, marital status and contraceptive use were significantly associated with unintended pregnancies. Crude odds ratio showed a significant association between maternal pregnancy intendedness and physical abuse twelve months prior to pregnancy. However, when controlling for all variables in the study model proved to be not significant. A better understanding of unintended pregnancies and physical abuse around the time of pregnancies would help professionals in Hawaiʻi to screen for at risk individuals.

Browse

Recent Submissions