- Email: [email protected]
Traumatic Spinal Cord Injury Mortality, 1981–1998
AEP Vol. 17, No. 9 September 2007 : 723–751
ABSTRACTS (ACE)
(81%) had more than two risk factors. Compared to younger HCV patients, older patients were more likely to have exposure to blood products (32% vs. 17%; p!0.001) but less likely to have injection drug use (59% vs. 74%; p!0.001). Proportion of newly diagnosed patients associated with blood transfusion before 1990 significantly declined (p!0.05) from 16.7% in the period 1998 to 2000, to 13.5% in the period 2004 to 2006. CONCLUSION: The above figures are useful for evaluating the epidemiological changes of HCV infection and for anticipating the future economical cost of the hepatitis C treatment. The recognition of the potential presence of multiple risk factors may have important implications determining approaches to HCV surveillance, particularly the use of hierarchical algorithms in the study of risk factors. doi: 10.1016/j.annepidem.2007.07.055
P52 EFFECT OF ANTIBIOTICS TAKEN FOR OTHER ILLNESSES ON INCIDENCE OF HELICOBACTER PYLORI INFECTION IN CHILDREN CS Broussard, KJ Goodman, CV Phillips, RS Day, CC Aragaki, MA Smith, LA Fischbach, University of Texas Health Science Center, School of Public Health, Houston, TX PURPOSE: This analysis estimated rate ratios for the effect of antibiotic exposure on the incidence of a first detected H. pylori infection in children. We hypothesized that antibiotics taken for common infections decrease the incidence of H. pylori infections that persist long enough to be detected. METHODS: The Pasitos Cohort Study (1998–2005) investigated predictors of H. pylori infection in children from El Paso, Texas, and Juarez, Mexico. We screened children for infection at target intervals of 6 months from 6 to 84 months of age, using the 13C-urea breath test corrected for body-size-dependent variation in CO2 production. Reported exposure was defined in two ways: 1) any systemic antibiotic use in the immediately preceding at-risk interval (interval-specific exposure) and 2) average number of courses across all at-risk intervals (average exposure). Cox proportional hazards regression models included country of residence, mother’s education, and prenatal care adequacy. We defined infection onset time as the midpoint between last negative and first positive test. For model checking, we also estimated rate ratios with an interval-censored survival model using a Weibull distribution. RESULTS: Of 589 children with complete data, 247 had a first detected infection during 15,762 person-months at risk; >1 antibiotic courses were reported at 48% of 1,920 visits. Rate ratios (95% CI) comparing 0 to O0 courses were 1.0 (0.79–1.3) for intervalspecific and 0.45 (0.34–0.61) for average exposure. Across interval-specific exposure levels, estimates showed weak effects; rate ratios comparing low and high average exposure to none were 0.38 (0.27–0.55) and 0.50 (0.37–0.69), respectively. Intervalcensored models produced similar results. CONCLUSION: Incidental antibiotic exposure was associated with a decreased rate of detected acquisition of childhood H. pylori infection when measured as the average across at-risk person-time. Reasons for inconsistent effects across exposure definitions and levels need further study. doi: 10.1016/j.annepidem.2007.07.056
741
P53 LUNG CANCER AND HISTORY OF PULMONARY TUBERCULOSIS AMONG THE U.S. ELDERLY Y Yu, R Pfeiffer, EA Engels, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Bethesda, MD PURPOSE: The complex processes of inflammatory response induced by pulmonary tuberculosis may increase the risk for lung cancer. We conducted a registry-based case-control study to test this hypothesis. METHODS: The Surveillance, Epidemiology and End Results (SEER)-Medicare linked database was used to examine the association of lung cancer and history of tuberculosis among persons aged 65 and older. We identified lung cancer cases from SEER registries and frequency-matched population-based controls sampled from Medicare beneficiaries. Medicare claims data were used to identify inpatient and outpatient records of tuberculosis. We analyzed the association between case-control status and tuberculosis using logistic regression models adjusted for sex, age group, race, SEER registry, and year of diagnosis/selection. We also assessed the association with larynx and bladder cancers as a comparison, because these two types of cancers are not known to be related to tuberculosis. RESULTS: We identified 107,389 lung cancer cases and the same number of controls. Odds of having tuberculosis within the past 15 years were 4.2 times higher among cases than controls (95% CI 3.5–5.1). The association was stronger when history of tuberculosis was defined in terms of inpatient claims (OR Z 4.7) compared to outpatient claims (OR Z 3.8). The association was strongest in the year preceding lung cancer diagnosis (OR Z 10.4), but was still significant in the 1–10 year period prior (OR Z 2.5, p!0.0001). We also found increased odds of having tuberculosis among larynx cancer cases (OR Z 3.6, p!0.0001), but not among bladder cancer cases (OR Z 1.2, pZ0.3). CONCLUSION: History of tuberculosis seems to be associated with increased lung cancer risk. The stronger association in the year prior to cancer diagnosis suggests ascertainment bias. We were not able to adjust for smoking, which may be a confounder (based on increased odds of tuberculosis for larynx cancer). The possible etiologic association between tuberculosis and lung cancer needs to be clarified. doi: 10.1016/j.annepidem.2007.07.057 INJURY P54 TRAUMATIC SPINAL CORD INJURY MORTALITY, 1981–1998 L Lineberry1, A Selassie1, E Hill1, J Nicholas1, A Varma2, D Lackland1, S Patel2, 1Department of Biostatistics, Bioinformatics, and Epidemiology, Medical University of South Carolina, 2Department of Neurosurgery, Medical University of South Carolina PURPOSE: Our objectives were to 1) assess the long-term trend of, and 2) identify risk factors for traumatic spinal cord injury (TSCI) mortality. METHODS: This analysis utilized hospital discharge data and multiple cause of data from the TSCI surveillance system in South Carolina from the years 1981–1998. ICD-9 codes 806 and 952 were
742
ABSTRACTS (ACE)
AEP Vol. 17, No. 9 September 2007 : 723–751
used to identify TSCI cases. A total of 4,353 persons sustained a TSCI, with 1,726 resulting in death. Poisson regression was used to examine trend, and multiple logistic regression was used to identify risk factors for TSCI mortality. RESULTS: The rate of TSCI mortality was 27.4 per million population between 1981–1998. A 3% annual decrease in the TSCI mortality rate was found from 1981 through 1998 (p!0.0001). Specifically, TSCI mortality rates declined the most per year in motor vehicle crashes, males, and whites. Rates in South Carolina were higher than the United States, specifically for males and blacks. Adjusted for covariates, individuals of older ages, black race, with a cervical TSCI, and with a more severe injury were associated with higher odds of in-hospital mortality. Females had lower odds of in-hospital mortality than males (ORZ0.6, 95%CI: 0.4–0.9). CONCLUSION: Although rate is decreasing, TSCI mortality, especially MVC-related, is still a significant problem, with South Carolina having higher rates of TSCI mortality than the U.S. The association between gender and in-hospital mortality needs further exploration. doi: 10.1016/j.annepidem.2007.07.058
P55 MOTOR VEHICLE CRASHES AND INJURY AMONG HIGH SCHOOL AND COLLEGE AGED DRIVERS. MIAMI-DADE COUNTY, FL 2005 MX Bustamante, G Zhang, E O’Connell, D Rodriguez, R BorrotoPonce, Miami-Dade County Health Department, Miami, FL PURPOSE: To describe motor vehicle crashes and injury occurring among high school and college aged drivers in order to guide prevention efforts targeted toward reducing motor vehicle related injuries among the young population. METHODS: The 2005 crash data used in this study was obtained from the Florida Department of Highway Safety and Motor Vehicles. Data was analyzed using SAS by age, injury severity, alcohol/ drug involvement, restraint use, day of week, and time of day. RESULTS: There were 16,311 drivers between the ages of 15–24 involved in motor vehicle crashes. Of these, over a 1/3 were injured, 14% were unrestrained, and 1.5% were alcohol and/or drug related. Alcohol and/or drug use among college aged drivers 20– 24 years of age was higher than that in other age groups. The majority of crashes (38%) occurred between the hours of 12 PM and 5 PM. Friday and Saturday accounted for 30% of all crashes in this age group. High school and college aged drivers were less likely to use safety restraints compared to all other age groups. Those without restraints were 5 times more likely to suffer severe injuries. Drivers who were using alcohol and/or drugs were twice as likely to be unrestrained and three times more likely to be severely injured. Among those using alcohol and/or drugs 27% were not restrained. Over half of drivers using alcohol and/or drugs crashed between the hours of 12 AM and 5 AM. CONCLUSION: Interventions targeting 15–24 year olds should focus on increasing safety belt use in this age group. Furthermore, college based interventions should also aim to reduce impaired driving due to alcohol and/or drugs. Since the majority of crashes occurred between the hours of 12 PM and 5 PM, intervention campaigns should target students as they leave the school premises for lunch or after class. In addition, localities should make use of readily available data such as the Florida Department of Highway
Safety and Motor Vehicles crash database in order to enhance injury surveillance activities and program planning and evaluation. doi: 10.1016/j.annepidem.2007.07.059 METHODS P56 LOW BIRTH WEIGHT RATES IN TWO COUNTIES OF SOUTH FLORIDA: SMALL-AREA VARIATION AND TEMPORAL TRENDS N Kellier, T Niyonsenga, Stempel School of Public Health Florida International University, Miami, FL PURPOSE: Low birth weight (LBW) is a primary cause of infant mortality and morbidity. The main hypothesis is that infant LBW rates would decrease annually by 9% to meet Health People 2010 goal of 5% by 2010. Moreover, observed LBW rates would be comparable over geographical areas in the counties and among race/ ethnic groups. This study focused on LBW rates over 8 years at the small area level, the zip code, within each county. The purpose is to estimate the trends, quantify the magnitude of the small area variation and identify geographic area with high rates. METHODS: This study utilized data from Florida Department of Health extracted from birth records for residents of Miami-Dade and Broward counties born in calendar years 1998 to 2005. For each zip code area within the counties, the number of LBW cases and the total number of births were provided yearly. The Poisson approximation to the Binomial distribution was used for statistical analyses and inference. Spatial statistical methods included the spatial scan statistic to detect clusters. Maps were used to highlight geographical variation and clusters of areas with high LBW rates. RESULTS: Descriptive results showed that for Miami-Dade County, LBW rates varied over time from 74.1 G 26.9 to 100.1 G 108.2 cases per 1000 births while in Broward County, they varied from 78.1 G 21.9 to 87.5 G 29.6 cases per 1000 births. There was no decreasing trend over both counties. For both counties, LBW rates exhibited non-negligible amount of small-area variation as captured by the coefficients of variation (CV). CV values ranged from 34.5% to 124.7% for Miami-Dade, and from 21.0% to 33.8% for Broward, with less variation within Broward County areas. CONCLUSION: In this multi-ethnic population of South Florida, LBW rates vary significantly at the small-area level and over time. There is no decrease in rates over time as expected. The rationale of this study is to focus further analysis on factors such as poverty, maternal social support, and smoking levels at the zip code level to explain the observed variation and the area-specific temporal trends. doi: 10.1016/j.annepidem.2007.07.060
P57 PREVALENCE AND CORRELATES OF SEXUAL RISK BEHAVIORS BY GENDER AMONG MULTI-ETHNIC ADOLESCENTS T Niyonsenga, WM Hlaing, Stempel School of Public Health Florida International University, Miami, FL PURPOSE: Risky sex and its related behaviors (initiation of sex, alcohol, and drugs, school performance, delinquency) among
ABSTRACTS (ACE)
(81%) had more than two risk factors. Compared to younger HCV patients, older patients were more likely to have exposure to blood products (32% vs. 17%; p!0.001) but less likely to have injection drug use (59% vs. 74%; p!0.001). Proportion of newly diagnosed patients associated with blood transfusion before 1990 significantly declined (p!0.05) from 16.7% in the period 1998 to 2000, to 13.5% in the period 2004 to 2006. CONCLUSION: The above figures are useful for evaluating the epidemiological changes of HCV infection and for anticipating the future economical cost of the hepatitis C treatment. The recognition of the potential presence of multiple risk factors may have important implications determining approaches to HCV surveillance, particularly the use of hierarchical algorithms in the study of risk factors. doi: 10.1016/j.annepidem.2007.07.055
P52 EFFECT OF ANTIBIOTICS TAKEN FOR OTHER ILLNESSES ON INCIDENCE OF HELICOBACTER PYLORI INFECTION IN CHILDREN CS Broussard, KJ Goodman, CV Phillips, RS Day, CC Aragaki, MA Smith, LA Fischbach, University of Texas Health Science Center, School of Public Health, Houston, TX PURPOSE: This analysis estimated rate ratios for the effect of antibiotic exposure on the incidence of a first detected H. pylori infection in children. We hypothesized that antibiotics taken for common infections decrease the incidence of H. pylori infections that persist long enough to be detected. METHODS: The Pasitos Cohort Study (1998–2005) investigated predictors of H. pylori infection in children from El Paso, Texas, and Juarez, Mexico. We screened children for infection at target intervals of 6 months from 6 to 84 months of age, using the 13C-urea breath test corrected for body-size-dependent variation in CO2 production. Reported exposure was defined in two ways: 1) any systemic antibiotic use in the immediately preceding at-risk interval (interval-specific exposure) and 2) average number of courses across all at-risk intervals (average exposure). Cox proportional hazards regression models included country of residence, mother’s education, and prenatal care adequacy. We defined infection onset time as the midpoint between last negative and first positive test. For model checking, we also estimated rate ratios with an interval-censored survival model using a Weibull distribution. RESULTS: Of 589 children with complete data, 247 had a first detected infection during 15,762 person-months at risk; >1 antibiotic courses were reported at 48% of 1,920 visits. Rate ratios (95% CI) comparing 0 to O0 courses were 1.0 (0.79–1.3) for intervalspecific and 0.45 (0.34–0.61) for average exposure. Across interval-specific exposure levels, estimates showed weak effects; rate ratios comparing low and high average exposure to none were 0.38 (0.27–0.55) and 0.50 (0.37–0.69), respectively. Intervalcensored models produced similar results. CONCLUSION: Incidental antibiotic exposure was associated with a decreased rate of detected acquisition of childhood H. pylori infection when measured as the average across at-risk person-time. Reasons for inconsistent effects across exposure definitions and levels need further study. doi: 10.1016/j.annepidem.2007.07.056
741
P53 LUNG CANCER AND HISTORY OF PULMONARY TUBERCULOSIS AMONG THE U.S. ELDERLY Y Yu, R Pfeiffer, EA Engels, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Bethesda, MD PURPOSE: The complex processes of inflammatory response induced by pulmonary tuberculosis may increase the risk for lung cancer. We conducted a registry-based case-control study to test this hypothesis. METHODS: The Surveillance, Epidemiology and End Results (SEER)-Medicare linked database was used to examine the association of lung cancer and history of tuberculosis among persons aged 65 and older. We identified lung cancer cases from SEER registries and frequency-matched population-based controls sampled from Medicare beneficiaries. Medicare claims data were used to identify inpatient and outpatient records of tuberculosis. We analyzed the association between case-control status and tuberculosis using logistic regression models adjusted for sex, age group, race, SEER registry, and year of diagnosis/selection. We also assessed the association with larynx and bladder cancers as a comparison, because these two types of cancers are not known to be related to tuberculosis. RESULTS: We identified 107,389 lung cancer cases and the same number of controls. Odds of having tuberculosis within the past 15 years were 4.2 times higher among cases than controls (95% CI 3.5–5.1). The association was stronger when history of tuberculosis was defined in terms of inpatient claims (OR Z 4.7) compared to outpatient claims (OR Z 3.8). The association was strongest in the year preceding lung cancer diagnosis (OR Z 10.4), but was still significant in the 1–10 year period prior (OR Z 2.5, p!0.0001). We also found increased odds of having tuberculosis among larynx cancer cases (OR Z 3.6, p!0.0001), but not among bladder cancer cases (OR Z 1.2, pZ0.3). CONCLUSION: History of tuberculosis seems to be associated with increased lung cancer risk. The stronger association in the year prior to cancer diagnosis suggests ascertainment bias. We were not able to adjust for smoking, which may be a confounder (based on increased odds of tuberculosis for larynx cancer). The possible etiologic association between tuberculosis and lung cancer needs to be clarified. doi: 10.1016/j.annepidem.2007.07.057 INJURY P54 TRAUMATIC SPINAL CORD INJURY MORTALITY, 1981–1998 L Lineberry1, A Selassie1, E Hill1, J Nicholas1, A Varma2, D Lackland1, S Patel2, 1Department of Biostatistics, Bioinformatics, and Epidemiology, Medical University of South Carolina, 2Department of Neurosurgery, Medical University of South Carolina PURPOSE: Our objectives were to 1) assess the long-term trend of, and 2) identify risk factors for traumatic spinal cord injury (TSCI) mortality. METHODS: This analysis utilized hospital discharge data and multiple cause of data from the TSCI surveillance system in South Carolina from the years 1981–1998. ICD-9 codes 806 and 952 were
742
ABSTRACTS (ACE)
AEP Vol. 17, No. 9 September 2007 : 723–751
used to identify TSCI cases. A total of 4,353 persons sustained a TSCI, with 1,726 resulting in death. Poisson regression was used to examine trend, and multiple logistic regression was used to identify risk factors for TSCI mortality. RESULTS: The rate of TSCI mortality was 27.4 per million population between 1981–1998. A 3% annual decrease in the TSCI mortality rate was found from 1981 through 1998 (p!0.0001). Specifically, TSCI mortality rates declined the most per year in motor vehicle crashes, males, and whites. Rates in South Carolina were higher than the United States, specifically for males and blacks. Adjusted for covariates, individuals of older ages, black race, with a cervical TSCI, and with a more severe injury were associated with higher odds of in-hospital mortality. Females had lower odds of in-hospital mortality than males (ORZ0.6, 95%CI: 0.4–0.9). CONCLUSION: Although rate is decreasing, TSCI mortality, especially MVC-related, is still a significant problem, with South Carolina having higher rates of TSCI mortality than the U.S. The association between gender and in-hospital mortality needs further exploration. doi: 10.1016/j.annepidem.2007.07.058
P55 MOTOR VEHICLE CRASHES AND INJURY AMONG HIGH SCHOOL AND COLLEGE AGED DRIVERS. MIAMI-DADE COUNTY, FL 2005 MX Bustamante, G Zhang, E O’Connell, D Rodriguez, R BorrotoPonce, Miami-Dade County Health Department, Miami, FL PURPOSE: To describe motor vehicle crashes and injury occurring among high school and college aged drivers in order to guide prevention efforts targeted toward reducing motor vehicle related injuries among the young population. METHODS: The 2005 crash data used in this study was obtained from the Florida Department of Highway Safety and Motor Vehicles. Data was analyzed using SAS by age, injury severity, alcohol/ drug involvement, restraint use, day of week, and time of day. RESULTS: There were 16,311 drivers between the ages of 15–24 involved in motor vehicle crashes. Of these, over a 1/3 were injured, 14% were unrestrained, and 1.5% were alcohol and/or drug related. Alcohol and/or drug use among college aged drivers 20– 24 years of age was higher than that in other age groups. The majority of crashes (38%) occurred between the hours of 12 PM and 5 PM. Friday and Saturday accounted for 30% of all crashes in this age group. High school and college aged drivers were less likely to use safety restraints compared to all other age groups. Those without restraints were 5 times more likely to suffer severe injuries. Drivers who were using alcohol and/or drugs were twice as likely to be unrestrained and three times more likely to be severely injured. Among those using alcohol and/or drugs 27% were not restrained. Over half of drivers using alcohol and/or drugs crashed between the hours of 12 AM and 5 AM. CONCLUSION: Interventions targeting 15–24 year olds should focus on increasing safety belt use in this age group. Furthermore, college based interventions should also aim to reduce impaired driving due to alcohol and/or drugs. Since the majority of crashes occurred between the hours of 12 PM and 5 PM, intervention campaigns should target students as they leave the school premises for lunch or after class. In addition, localities should make use of readily available data such as the Florida Department of Highway
Safety and Motor Vehicles crash database in order to enhance injury surveillance activities and program planning and evaluation. doi: 10.1016/j.annepidem.2007.07.059 METHODS P56 LOW BIRTH WEIGHT RATES IN TWO COUNTIES OF SOUTH FLORIDA: SMALL-AREA VARIATION AND TEMPORAL TRENDS N Kellier, T Niyonsenga, Stempel School of Public Health Florida International University, Miami, FL PURPOSE: Low birth weight (LBW) is a primary cause of infant mortality and morbidity. The main hypothesis is that infant LBW rates would decrease annually by 9% to meet Health People 2010 goal of 5% by 2010. Moreover, observed LBW rates would be comparable over geographical areas in the counties and among race/ ethnic groups. This study focused on LBW rates over 8 years at the small area level, the zip code, within each county. The purpose is to estimate the trends, quantify the magnitude of the small area variation and identify geographic area with high rates. METHODS: This study utilized data from Florida Department of Health extracted from birth records for residents of Miami-Dade and Broward counties born in calendar years 1998 to 2005. For each zip code area within the counties, the number of LBW cases and the total number of births were provided yearly. The Poisson approximation to the Binomial distribution was used for statistical analyses and inference. Spatial statistical methods included the spatial scan statistic to detect clusters. Maps were used to highlight geographical variation and clusters of areas with high LBW rates. RESULTS: Descriptive results showed that for Miami-Dade County, LBW rates varied over time from 74.1 G 26.9 to 100.1 G 108.2 cases per 1000 births while in Broward County, they varied from 78.1 G 21.9 to 87.5 G 29.6 cases per 1000 births. There was no decreasing trend over both counties. For both counties, LBW rates exhibited non-negligible amount of small-area variation as captured by the coefficients of variation (CV). CV values ranged from 34.5% to 124.7% for Miami-Dade, and from 21.0% to 33.8% for Broward, with less variation within Broward County areas. CONCLUSION: In this multi-ethnic population of South Florida, LBW rates vary significantly at the small-area level and over time. There is no decrease in rates over time as expected. The rationale of this study is to focus further analysis on factors such as poverty, maternal social support, and smoking levels at the zip code level to explain the observed variation and the area-specific temporal trends. doi: 10.1016/j.annepidem.2007.07.060
P57 PREVALENCE AND CORRELATES OF SEXUAL RISK BEHAVIORS BY GENDER AMONG MULTI-ETHNIC ADOLESCENTS T Niyonsenga, WM Hlaing, Stempel School of Public Health Florida International University, Miami, FL PURPOSE: Risky sex and its related behaviors (initiation of sex, alcohol, and drugs, school performance, delinquency) among