Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “sampling design”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,459 records · Page 81Linked to original sources
PubMed1991

Changes in the incidence of alcohol-impaired driving in the United States, 1973-1986.

Studies of motor vehicle fatality data have indicated that alcohol involvement in fatal crashes has declined substantially in the United States since 1980. To determine the actual incidence of alcohol-impaired drivers on U.S. roads, a national roadside survey using portable breath-testing devices was carried out in 32 localities in the spring of 1986. The same sampling design and survey procedures used in a 1973 national roadside survey were followed as much as possible. The 1986 survey found 3.1% of the late-night weekend drivers to have a blood alcohol concentration (BAC) of 0.10% or more, compared to 4.9% of drivers in 1973. Similarly, 8.3% of the 1986 drivers were at or above 0.05% BAC, compared to 13.5% in 1973. The data indicate that the incidence of alcohol-impaired driving on weekend nights has fallen by one-third or more in the United States since 1973 and that the decline affected most population subgroups.

Accidents, Traffic↗
PubMed2002

Effects of minimum drinking age laws: review and analyses of the literature from 1960 to 2000.

OBJECTIVE: The goal of this article is to review critically the extant minimum legal drinking age (MLDA) research literature and summarize the current state of knowledge regarding the effectiveness of this policy. METHOD: Comprehensive searches of four databases were conducted to identify empirical studies of the MLDA published from 1960 to 1999. Three variables were coded for each study regarding methodological quality: (1) sampling design, (2) study design and (3) presence or absence of comparison group. RESULTS: We identified 241 empirical analyses of the MLDA. Fifty-six percent of the analyses met our criteria for high methodological quality. Of the 33 higher quality studies of MLDA and alcohol consumption, 11 (33%) found an inverse relationship; only 1 found the opposite. Similarly, of the 79 higher quality analyses of MLDA and traffic crashes, 46 (58%) found a higher MLDA related to decreased traffic crashes; none found the opposite. Eight of the 23 analyses of other problems found a higher MLDA associated with reduced problems; none found the opposite. Only 6 of the 64 college-specific studies (9%) were of high quality; none found a significant relationship between the MLDA and outcome measures. CONCLUSIONS: The preponderance of evidence indicates there is an inverse relationship between the MLDA and two outcome measures: alcohol consumption and traffic crashes. The quality of the studies of specific populations such as college students is poor, preventing any conclusions that the effects of MLDA might differ for such special populations.

Age Factors↗
PubMed1998

A population-based study of access to immunization among urban Virginia children served by public, private, and military health care systems.

BACKGROUND: Pediatric immunization rates have increased in the United States since 1990. Nevertheless, national survey data indicate that up to one third of 2-year-old children in some states and urban areas lack at least one recommended dose of diphtheria-tetanus-pertussis (DTP)-, polio-, or measles-containing vaccines. Immunization has become a key measure of preventive pediatric health care in the United States. To achieve and maintain the national immunization goal that 90% of children receive all recommended immunizations by 2 years of age, the role of the health care system in immunization delivery must be examined. Urban eastern Virginia has a diverse population that obtains immunization services from public, private, and military providers and insurers. At the time of this survey, immunization services in Virginia were available free to all children through public health clinics and to military families when using a military facility. OBJECTIVE: To examine access to pediatric immunization services and health system factors associated with underimmunization in a representative sample of children at 12 and 24 months of age. METHODS: We conducted a household survey in urban eastern Virginia from April through September 1993. A total of 12 770 households in Norfolk and Newport News, VA, were selected for inclusion in the study using probability-proportionate-to-size cluster sampling. Use of probability-proportionate-to-size sampling ensured that children within each city had equal probability of being included in the survey. Selected households were visited by trained interviewers to determine their eligibility, defined as having at least one child 12 to 30 months of age residing in the household. In eligible households, parents were asked to participate in a standardized, 15-minute interview. Survey respondents were asked about household demographics, and for each eligible child, the immunization history, health insurance, the name and location of all immunization providers, the usual immunization provider, and any problems the parent had experienced accessing immunization services with that child. Up-to-date (UTD) immunization status was defined as having all recommended doses of DTP, polio, and measles-mumps-rubella at 12 months (three DTP and two polio immunizations) and 24 months (four DTP, three polio, and one measles-mumps-rubella immunizations). The child's immunization history was assessed from parent and provider records only. Data analysis accounted for the survey's cluster sampling design (ie, within-cluster correlation). Because the immunization rates of the two cities did not differ significantly, unweighted analyses were used for ease of computation. Significance was determined for contingency tables by Wald's chi2 test. RESULTS: A total of 749 children (91% of eligible households) participated in the survey. Study children were born between October, 1990, and July, 1992. Immunization records were obtained for 705 children (94%). Eighty-seven percent of respondents were mothers, 44% were African-American, 40% of children were military dependents, and 40% were enrolled in the Women, Infants and Children (WIC) program. Sixty-five percent of children were UTD at 12 months and 53% at 24 months. Parents reported that their children's usual immunization providers were private doctors (34%); public health, hospital clinics, or community health centers (32%); and military clinics or a military contract provider (34%). At least one problem accessing immunization services was reported by 35% of respondents, ranging from 29% among those who used a private doctor as their child's usual immunization provider to 46% among those using a military contract provider. Overall, the most commonly reported problem was clinic waiting time (12%), with reports of waiting time as a problem occurring most often among those using the military contract provider (22%) and public health clinics (17%). (ABSTRACT TRUNCATED)

Child, Preschool↗
PubMed2001

Differential impact of recent Medicaid expansions by race and ethnicity.

OBJECTIVE: Between 1989 and 1995, expansions in Medicaid eligibility provided publicly financed health insurance to an additional 7 million poor and near-poor children. It is not known whether these expansions affected children's insurance coverage, use of health care services, or health status differently, depending on their race/ethnicity. The objective of this study was to examine, by race/ethnicity, the impact of the recent Medicaid expansions on levels of uninsured individuals, health care service utilization, and health status of the targeted groups of children. METHODS: Using a stratified set of longitudinal data from the National Health Interview Surveys of 1989 and 1995, we compared changes in measures of health insurance coverage, health services utilization, and health status for poor white, black, and Hispanic 1- to 12-year-old children. To control for underlying trends over time, we subtracted 1989 to 1995 changes in these outcomes among nonpoor children from changes among the poor children for each race/ethnicity group. Measures of coverage included uninsured rates and Medicaid rates. Utilization measures included annual probability of visiting a doctor, annual number of doctor visits, and annual probability of hospitalization. Health status measures included self-reported health status and number of restricted-activity days in the 2 weeks before the interview. Differences in means were analyzed with the use of Student's t tests accounting for the clustering sample design of the National Health Interview Surveys. RESULTS: Among poor children between 1989 and 1995, uninsured rates declined by 4 percentage points for whites, 11 percentage points for blacks, and 19 percentage points for Hispanics. Medicaid rates for these groups increased by 16 percentage points, 22 percentage points, and 23 percentage points, respectively. With respect to utilization, the annual probability of seeing a physician increased 7 percentage points among poor blacks and Hispanics but only 1 percentage point among poor whites (not significant) for children in good, fair, or poor health. Among those in excellent or very good health, the respective increases were 1 percentage point for poor whites (not significant), 7 percentage points for poor blacks, and 3 percentage points for poor Hispanics (not significant). Significant increases in numbers of doctor visits per year were recorded only for poor Hispanics who were in excellent or very good health, whereas significant decreases in hospitalizations were recorded for Hispanics who were in good fair or poor health. Measures of health status remained unchanged for poor children over time. The recorded decreases in uninsured rates and increases in Medicaid coverage remained robust to adjustments for underlying trends for all 3 race/ethnicity groups. With respect to adjusted measures of utilization and health status, the only significant differences found were among poor blacks who were in good, fair, or poor health and who registered increases in the likelihood of hospitalization and in poor Hispanics who were in excellent or very good health and who registered decreases in the numbers of restricted-activity days. CONCLUSIONS: Recent expansions in the Medicaid program from 1989 to 1995 produced greater reductions in uninsured rates among poor minority children than among poor white children. Regardless of race/ethnicity, poor children did not seem to experience significant changes during the period of the expansions in either their level of health service utilization or their health status.Medicaid, health insurance, health status, health services.

Black or African American↗
PubMed2006

Full breastfeeding duration and associated decrease in respiratory tract infection in US children.

OBJECTIVE: The American Academy of Pediatrics recommends exclusive breastfeeding for an infant's first 6 months of life. When compared with exclusive breastfeeding for 4 months, greater protection against gastrointestinal infection, but not respiratory tract infection, has been demonstrated for the 6-month duration. The objective of this study was to ascertain if full breastfeeding of > or = 6 months compared with 4 to < 6 months in the United States provides greater protection against respiratory tract infection. METHODS: Secondary analysis of data from the National Health and Nutrition Examination Survey III, a nationally representative cross-sectional home survey conducted from 1988 to 1994, was performed. Data from 2277 children aged 6 to < 24 months, who were divided into 5 groups according to breastfeeding status, were compared. Children who required neonatal intensive care were excluded. SUDAAN software was used to account for the complex sampling design. Logistic regression adjusted for confounding factors. Outcome measures included adjusted odds of acquiring pneumonia, > or = 3 episodes of cold/influenza, > or = 3 episodes of otitis media (OM), or wheezing in the past year or acquiring first OM at < 12 months of age. RESULTS: In unadjusted analyses, infants who were fully breastfed for 4 to < 6 months (n = 223) were at greater risk for pneumonia than those who were fully breastfed for > or = 6 months (n = 136) (6.5% vs 1.6%). There were not statistically significant differences in > or = 3 episodes of cold/influenza (45% vs 41%), wheezing (23% vs 24%), > or = 3 episodes of OM (27% vs 20%), or first OM at < 12 months of age (49% vs 47%). Adjusting for demographic variables, childcare, and smoke exposure revealed statistically significant increased risk for both pneumonia (odds ratio [OR]: 4.27; 95% confidence interval [CI]: 1.27-14.35) and > or = 3 episodes of OM (OR: 1.95; 95% CI: 1.06-3.59) in those who were fully breastfed for 4 to < 6 months compared with > or = 6 months. CONCLUSIONS: This nationally representative study documents increased risk of respiratory tract infection including pneumonia and recurrent OM in children who were fully breastfed for 4 vs 6 months. These findings support current recommendations that infants receive only breast milk for the first 6 months of life.

Birth Weight↗
PubMed2006

Pediatric fireworks-related injuries in the United States: 1990-2003.

OBJECTIVE: Our goal was to describe the epidemiology of pediatric fireworks-related injuries among children aged 19 years and younger by using a nationally representative sample. DESIGN: We performed a retrospective analysis of data from the US Consumer Product Safety Commission's National Electronic Injury Surveillance System from 1990-2003. RESULTS: An estimated 85800 pediatric fireworks-related injuries were treated in US emergency departments during the 14-year study period. Injured children had a mean age of 10.8 years, and 77.9% were male. Fireworks users accounted for 49.5% of the injuries, whereas 22.2% of the injuries were to bystanders; however, user status could not be determined in 28.3% of cases. The overall fireworks-related injury rate decreased significantly during the study period, but subgroup analysis did not indicate consistent declines among all ages and types of fireworks. Injuries were most commonly caused by firecrackers (29.6%), sparklers/novelty devices (20.5%), and aerial devices (17.6%). The most commonly injured body sites were the eyeball (20.8%), face (20.0%), and hands (19.8%), and the most common injury type was burns (60.3%). Approximately 91.6% of all children with fireworks-related injuries were treated and released from hospital emergency departments, 5.3% were admitted, and 2.3% were transferred to another institution. Bystanders accounted for 13.3% of admitted cases and 20.6% of transferred cases. CONCLUSIONS: Consumer fireworks cause serious preventable injuries among pediatric fireworks users and bystanders in the United States. Parents should be advised to take their children to safer public fireworks displays rather than allowing consumer fireworks to be used by or near their children. A national restriction of consumer fireworks, in accordance with the policy recommendations of the American Academy of Pediatrics, should be implemented to reduce the burden of fireworks-related injuries among children.

Adolescent↗
PubMed2004

[Drug utilization in adults: prevalence and individuals determinants].

OBJECTIVE: To study epidemiological patterns of drug utilization and its individual determinants and to classify drugs used into pharmacological groups. METHODS: In a population-based cross-sectional study, 3,182 subjects aged 20 years or more were selected from an urban area in Southern Brazil using a multi-stage sampling design. Data were collected through home interviews using a structured questionnaire. Drug use in the previous 15 days was recorded. Crude analysis was performed using the Chi-square test for heterogeneity and trend and t-test for means' comparison. The adjusted analysis was carried out using Poisson regression. RESULTS: The overall prevalence of drug utilization was 65.9%. The highest prevalences of drug use after adjusting for confounding were seen among elderly women in the upper economic class who had ill health self-perception. Analgesics, anti-inflammatory and anti-hypertensive drugs were the most frequently used drugs. CONCLUSIONS: The prevalence of drug use was found to be higher than in other national and international studies. Studies focusing on individual determinants of drug utilization can identify those groups with higher use. This knowledge might encourage specific strategies to decrease drug use and lead to restrictive policies concerning drug prescription and sale.

Adult↗
PubMed2006

[Domestic violence against Mexican teenagers].

OBJECTIVE: To assess students' perception of domestic violence. METHODS: The study was carried out in a random sample representative (n=5,982) of high-school students aged between 13 and 19 years in Chiapas and San Luis Potosi, Mexico, in 2003. The students answered a self-administered questionnaire on sexual education, sexuality, pregnancy and risk behavior, and domestic violence. Considering the effects of the sample design, a logistic regression analysis was carried out in order to verify the association between socioeconomic and demographic characteristics and risk of domestic violence. RESULTS: The multivariate statistical analysis confirmed that female students are more vulnerable to domestic violence in both states studied (Chiapas: OR=1.87, p=0.01; San Luis: OR=1.53, p=0.00). Moreover, alcohol (Chiapas: OR=2.52, p=0.00; San Luis: OR=1.93, p=0.00) or drug use (Chiapas: OR=2.27, p=0.02; San Luis: OR=1.82, p=0.01) were the two most significant variables for students exposure to domestic violence. The association between alcohol consumption and student's age increases the risk of violence. The youngest students had 51% (Chiapas) and 73% (San Luis) more risk of experience violence by a relative than older students. CONCLUSIONS: Teenagers remain one of the most vulnerable population groups to domestic violence due to their physical, affective, economic and social dependence on other adults. The study was one of the first attempts to measure domestic violence against students.

Adolescent↗
PubMedDate not supplied

[Early onset depression: prevalence, course, and treatment seeking delay].

OBJECTIVE: To estimate the prevalence of early onset depression in the Mexican population and compare its natural history, comorbidity and treatment latency between early onset and adult onset depression. MATERIAL AND METHODS: The National Survey of Psychiatric Epidemiology (ENEP) is representative of the Mexican urban population aged 18 to 65. The diagnostic instrument is the Composite International Diagnostic Interview (CIDI). Statistical analysis was performed with descriptive and logistic regression analyses, considering the multistage, stratified and weighted sample design of the survey. RESULTS: Two percent of the Mexican population has suffered from depression as children or adolescents, with an average 7 episodes during their lifetime (compared with 3 episodes for adult onset); the first episode lasted a mean of 31 months (compared with 16 for adult onset) and generally did not receive any treatment. CONCLUSIONS: The longer duration of the first episode and the greater number of lifetime episodes for subjects with early onset is attributed to the lack of early detection and of timely treatment of depression in Mexican youth. The English version of this paper is available at: http://www.insp.mx/salud/index.html.

Adolescent↗
PubMedDate not supplied

[Prevalence of violent events and post-traumatic stress disorder in the Mexican population].

OBJECTIVE: To report the rate of exposure to different violent events, their demographic correlates, the prevalence of Post-Traumatic Stress Disorder (PTSD), and the impact on quality of life. MATERIAL AND METHODS: The National Survey of Psychiatric Epidemiology is representative of the Mexican urban population aged 18 to 65. The survey was undertaken in 2001 and 2002 using the Composite International Diagnostic Interview (CIDI- 15) computerized version. The statistical analyses take into account the multistage, stratified, and weighted sample design. Kaplan-Meier and logistic regressions were performed. RESULTS: Sixty-eight percent of the population has been exposed to at least one stressful life event. Exposure varies by sex (rape, harassment, and sexual abuse are more frequent in women; accidents and being a victim of burglary among men) and by age (more frequent in children, adolescents, young adult women, and the elderly). By sex, 2.3% of women and 0.49% of men present PTSD. Rape, harassment, kidnapping, and sexual abuse are the events most associated with PTSD. CONCLUSIONS: The results suggest the need to increase treatment coverage to attend the consequences of violence, taking into consideration the important gender and age variations.

Adolescent↗
PubMed2003

[District allocation and utilization of health care services in Pau da Lima, Salvador, Bahia, Brazil].

This cross-sectional study is aimed at the identification of patterns in the utilization of health care services in Pau da Lima, a neighborhood in the city of Salvador, Bahia. In 1992, a household survey was carried out with 384 families selected through a random cluster sampling design. In each household, the family head or a surrogate informant was asked to answer a questionnaire about their use of any type of health care in the last month. From the total of 1,887 individuals, 236 reported at least one visit to health facilities during the referent period This means an estimated prevalence of health services utilization of 12.5%. With regard to the type of services, 25.9% of all visits were made to public sector facilities. Attendants were mainly women (73.7%) from 15 to 29 years of age. The main reason reported for the choice of the health facility was availability (63.7%), and the major reason for seeking health care was disease-related conditions (75.7%), rather than preventive procedures. These findings may result from the still unfinished process of distritalização(district allocation of services), which is evident in the lack of health services' infrastructure or coverage, as well as in persistent organizational problems. These issues may lead the population living in the respective district to seek health care outside the assigned catchment area. The authors discuss these findings in the context of the expansion of the private sector in the health area and the financial collapse of the public budget in Brazil.

English Abstract↗
PubMed2003

[Arterial hypertension in Ilha do Governador, Rio de Janeiro, Brazil: I. Methodology].

The aim of this paper (first of a series of two) is to present the main methodological aspects used in a cross-sectional study whose proposals were to establish the prevalence of arterial hypertension in adults in Ilha do Governador (a district of the city of Rio de Janeiro, Brazil), and to examine its relationship to other risk factors, including those for cardiovascular diseases, so as to establish its control status. Sampling design (of clusters in strata, according to mean household income) is described in detail, in addition to measurement methods, and training of observers. Indicators of performance in field work and quality control are also presented as results of the application of the metodology of study.

English Abstract↗
PubMed2003

[Non-fatal work-related accidents in adolescents].

This community-based survey estimates the annual incidence of non-fatal work-related injuries in adolescents and describes their characteristics. Using a random area sampling design, 2,513 households were selected from the city of Salvador, capital of Bahia State, Northeast Brazil. The study population included 361 young people 10 to 20 years of age who answered individual questionnaires on socio-demographic, occupational, and health data. Annual incidence of non-fatal work-related injuries was 6.4%. Among females, most injuries were mild and were related to household chores; some 38.5% required medical care, and 36.4% caused permanent non-disabling lesions. Among males, work-related injuries took place mainly in public byways, and the majority (60.0%) were treated in emergency-care facilities. These results show that work-related injuries need to be considered in adolescent protection policies.

Accidents, Occupational↗
PubMed2003

[Prevalence of overweight and obesity and associated factors among adolescents in the Northeast and Southeast regions of Brazil, 1996 to 1997].

The purpose of this study was to assess the prevalence and factors associated with overweight and obesity (OW/O) among Brazilian adolescents, based on the Living Standards Survey conducted in 1996 and 1997 (IBGE, 1998a). The sample included 1,027 and 854 adolescents aged 15 to 19 years from the Northeast and Southeast regions, respectively. Body Mass Index (w/h2) was calculated. Statistical analysis considered expansion factors and sample design. Prevalence of OW/O was 8.45% in the Northeast and 11.53% in the Southeast. In the Northeast there was a higher risk of OW/O for girls (PR = 3.00), with the same situation in both urban (PR = 3.21) and rural areas (PR = 2.27). In the Southeast, the risk of OW/O was lower among girls (PR = 0.58). The same situation occurred in girls from urban areas (PR = 0.51), but in rural areas there was an increased risk among girls (PR = 1.86). Per capita monthly income was associated with the risk of OW/O among boys in both regions. Programs to prevent OW/O should incorporate nutritional reeducation and encourage physical exercise, besides considering socioeconomic, cultural, and regional differences and the location of the target population.

Adolescent↗
PubMed2006

[Arterial hypertension in the elderly: prevalence, associated factors, and control practices in Campinas, São Paulo, Brazil].

The objective of this study was to assess the prevalence of reported hypertension among the elderly in Campinas, São Paulo, Brazil, identifying related factors, the use of healthcare services, and knowledge and practices related to treatment options. This was a cross-sectional population-based study, with stratified clustered multiple-stage sampling. Data analysis for the 426 individuals (60 years of age and older) considered the sampling design and outlining effect. Hypertension prevalence was 51.8% (46.4% in men and 55.9% in women) and was higher among the elderly with less education (55.9%), immigrants from other States (60.2%), and the overweight or obese (57.2%). The results indicate that healthcare services are ensuring access to medical care (71.6% have regular physician appointments) and medication (86.7% take their routine medication), without distinction as to socioeconomic status. However, social inequalities persist in knowledge and use of other measures to control blood pressure, like appropriate diet and exercise, which are also underused by more privileged socioeconomic strata.

Age Distribution↗
PubMed2006

Prevalence and correlates of drug use disorders in Mexico.

OBJECTIVE: To describe the prevalence of drug use disorders, the correlates of drug use, and the utilization of specialized treatment services for drug users among the Mexican urban population 18-65 years old. METHODS: The data were collected in 2001 and 2002 in the Mexican National Comorbidity Survey. The sample design was stratified probabilistically for six geographical areas of the country in a multistage process for census count areas, city blocks, groups of households, and individuals. The data were weighted, taking into account the probability of selection and the response rate. The information was collected using a computerized version of the World Mental Health Survey edition of the Composite International Diagnostic Interview. The weighted response rate for individuals was 76.6%. RESULTS: Overall, 2.3% of the population reported any illicit use of drugs in the preceding 12 months; marijuana and cocaine were the substances most often used. Low levels of education were significantly associated with use, abuse, and dependence. Use of any drug was significantly more common among those who were in the youngest age group (18-29 years), were male, or were living in the Northwest region of the country. Overall, 1.4% had a lifetime history of drug abuse or dependence, with this being much more common for men (2.9%) than for women (0.2%). The 12-month prevalence of drug abuse or dependence was 0.4% overall (0.9% for men, and 0.0% for women). The rate of treatment during the preceding 12 months for those with the 12-month criteria for abuse or dependence was 17.1%; 14.8% were seen in specialized treatment centers; 2.8% reported having attended self-help groups. CONCLUSIONS: A noticeable number of Mexicans have a drug use disorder, but demand for treatment is limited, in part due to stigma. Our results indicate that there is an urgent need to organize the specialized services for persons with a substance abuse disorder according to the prevalence of dependence on different substances and the variation in prevalence in the different regions of the country.

Adolescent↗
PubMed2006

[Dental caries in Brazil: decline, polarization, inequality and social exclusion].

OBJECTIVE: To examine the caries situation in Brazilian schoolchildren between 1980 and 2003 and to determine the distribution of caries and access to treatment in this population group. METHODS: We employed secondary data concerning the number of decayed, missing or filled teeth (DMFT). The studies whose data were used differed in terms of study type, study design, sampling methods, and diagnostic criteria, but yielded national estimates that are considered valid for the DMFT index. Therefore, a trend analysis based on these studies was thought to be feasible. RESULTS: Analysis of DMFT values revealed a high frequency of dental caries in 1980 and a moderate frequency in the 1990s. In 2003, the DMFT index was still within the moderate range. Between 1980 and 2003, a 61.7% decrease in DMFT frequency was observed. The percentage of children with no DMFT increased from 3.7% in 1986 to 31.1% in 2003. On the other hand, in the segment of the study population least affected, the percentage of children who received care increased from 26.3 in 1986 to 34.7 in 2003, yet in the group with DMFT this percentage fell from 50.2 in 1986 to 39.3 in 2003. In the segment with DMFT >/=6, the percentage of those who received care remained stable (28%). During the study period, 60% of the dental caries were found in 20% of the study population. CONCLUSION: An important decline in DMFT was observed between 1980 and 2003, perhaps as a result of increased access to fluoridated water and toothpaste and of changes in the goals of public oral health programs. Despite the improvement, caries is unevenly distributed in the pediatric population; a small proportion of individuals carries most of the disease burden. In addition, the proportion of teeth with caries that went untreated did not change between 1980 and 2003. Reducing socioeconomic disparities and adopting public health measures that target and reach the most vulnerable groups remain a challenge for policy makers in Brazil.

Brazil↗
PubMed2006

Model selection for geostatistical models.

We consider the problem of model selection for geospatial data. Spatial correlation is often ignored in the selection of explanatory variables, and this can influence model selection results. For example, the importance of particular explanatory variables may not be apparent when spatial correlation is ignored. To address this problem, we consider the Akaike Information Criterion (AIC) as applied to a geostatistical model. We offer a heuristic derivation of the AIC in this context and provide simulation results that show that using AIC for a geostatistical model is superior to the often-used traditional approach of ignoring spatial correlation in the selection of explanatory variables. These ideas are further demonstrated via a model for lizard abundance. We also apply the principle of minimum description length (MDL) to variable selection for the geostatistical model. The effect of sampling design on the selection of explanatory covariates is also explored. R software to implement the geostatistical model selection methods described in this paper is available in the Supplement.

Animals↗