Spatial analytical methods and geographic information systems: use in health research and epidemiology.
Explore the source record for details and available documents.
Cluster Analysis↗
Biomedical subjects
T E Carpenter
Publications and source records attributed to T E Carpenter.
Evaluation of temporal and spatial clustering of horses with Corynebacterium pseudotuberculosis infection.
OBJECTIVE: To determine whether horses with Corynebacterium pseudotuberculosis infections that were examined at a veterinary medical teaching hospital between July 1, 1992, and June 30, 1994 had patterns of temporal or spatial clustering. ANIMALS: 134 case and 800 control horses randomly selected from all non-case horses admitted during the study period. PROCEDURES: Admission date and geographic location were determined. Scan, Cuzick & Edwards', and Knox tests were applied to determine whether case horses had patterns of temporal or spatial clustering. RESULTS: For all windows > or = 3 days (134 case horses) and > or = 7 days (subset of 69 case horses), results of the Scan test were significant. Results of Cuzick & Edwards' test were significant for all data sets. A significant spatial cluster of case horses was observed for October, November, and December 1992. Results of the Knox test were significant for temporal intervals between 7 and 56 days and spatial intervals between 4.3 and 6.5 km. Higher Knox(x) proportions were observed for temporal intervals of 0 to 7, 8 to 14, 22 to 28, and 29 to 35 days. CONCLUSIONS: Significant spatial and temporal clustering of horses with C pseudotuberculosis infection was detected. CLINICAL RELEVANCE: Analysis of the results strongly indicates that this disease is directly or indirectly (ie, short distance and time) transmitted. In addition, data analyses indicated an incubation period of 3 to 4 weeks. The disease could be transmitted through horse-to-horse contact or from infected to susceptible horses via insects, other vectors, or contaminated soil.
Animals↗
Using a tobit regression model to analyse risk factors for foot-pad dermatitis in commercially grown broilers.
A model was developed to explain the influence of different local and regional factors on the prevalence of foot-pad dermatitis in broilers at the time of slaughter. Data from a recent two-year study of broiler foot-health in Sweden were used to construct the model. The model was based on regression analysis, using a Tobit model for censored data. This type of Tobit regression model has been used for several decades in econometrics and can also be found in the medical-scientific literature. It has, however, reportedly not been used in veterinary medicine although there appears to be several situations where it should be considered useful. Week of slaughter was significantly (p < 0.001) associated with total foot-pad score, which decreased over time. The birds' age at slaughter was also significantly (p < 0.01) associated with total foot-pad score. Five out of the 11 slaughterhouses included in the study were included as significant (p < 0.01) variables in the final model. Three feed suppliers (p < 0.05) and four regions (p < 0.01) were also included as significant variables in the model. There were also significant (p < 0.001) interactions between several of these factors (mainly, between the time variable and specific slaughterhouses, feed suppliers or regions, respectively). The information gathered in this study regarding the influence of a number of risk factors for the development of foot-pad dermatitis and their reciprocal interactions will enable us to modify the Swedish broiler foot-health programme.
Abattoirs↗
Application and evaluation of a mailed questionnaire for an epidemiologic study of Corynebacterium pseudotuberculosis infection in horses.
The objective of this study is to describe the design, application and validity of a self-administered (mailed) questionnaire to collect data on potential risk factors for Corynebacterium pseudotuberculosis infection in California horses. Horses admitted to the UC Davis Veterinary Medical Teaching Hospital (VMTH) between 1 July 1992 and 30 June 1994 served as the study base for case identification and simple random sampling of 800 control horses. A questionnaire was mailed to owners of the study horses, followed by a reminder postcard and a second copy of a questionnaire. Data were collected on owner and horse identity and demographics, horse management and use, geographic location, and general health-related issues. Return pattern over time as well as differential return proportions were described. The overall return proportion was 66% (587/890), and the completion proportion 55% (491/890). The number of returns over time followed a negative binomial distribution, with over 90% of all returns being in by the end of the fifth week after mailing, and over 99% at the end of the tenth week. Some categories within the variables age (between 2 and 3 years), breed (Thoroughbred and Standardbred horses) and gender (stallions) had significantly lower return proportions than expected (differential return; p < 0.05). The profile of these horses fits a section of the racehorse population that is served by the VMTH. Age, breed and disease status information was available from the VMTH medical records and from the questionnaire, and was used to determine the validity of the survey data. There was good agreement between the data from the two sources, and we therefore concluded that the quality of the survey information was sufficient to perform a risk-factor analysis. The mailed survey provided a rapid and cost-effective method of collecting additional information to supplement existing medical records.
Animals↗
Risk factors associated with Corynebacterium pseudotuberculosis infection in California horses.
A case-control study was designed using equine medical records from the UC Davis Veterinary Medical Teaching Hospital (VMTH) and data derived through a mailed survey. The objective was to evaluate the associations between horse demographics, horse-management factors, and equine Corynebacterium pseudotuberculosis infection in California. Horses admitted to the VMTH between July 1 1992 and June 30 1994 served as the study base for case identification and simple random sampling of 800 controls. A questionnaire was mailed to the owners of all horses enrolled in the study to collect data on demographics, management and health-related questions. A logistic-regression model containing age, outdoor activity level, other locations in California, insect-control measures, contact with other horses, and summer pasture was developed. The final model was adjusted for the suspected confounding variables admission type, regular teaching hospital patient and breed. Horses of age between 1 and 2 yrs and between 3 and 5 yrs, and horses in contact with other horses or horses on summer pasture had significantly increased odds (p < 0.05) of being diagnosed with C. pseudotuberculosis infection. The results support the hypotheses that the disease predominantly affects young-adult horses of all breeds and both sexes, and that management factors play an important role in occurrence of the disease. Since the existing serological test system is not reliable and destruction of infected animals is not feasible, the most-logical approach for disease prevention is the early identification and isolation of clinical cases and the implementation of management changes like improvement of stable hygiene and insect control and change of pasture practices.
Animals↗
Assessment of spatial and temporal clustering of ampicillin- and tetracycline-resistant strains of Pasteurella multocida and P haemolytica isolated from cattle in California.
OBJECTIVE: To determine whether ampicillin- and tetracycline-resistant strains of Pasteurella multocida and P haemolytica isolated from California cattle with pneumonia were spatially and temporally clustered and to compare overall estimates of percentages of these isolates resistant to these antimicrobials with estimates obtained on the basis of regional and temporal information. DESIGN: Epidemiologic study. SAMPLE POPULATION: Records of P multocida and P haemolytica isolates obtained from lung or tracheal wash samples collected from California cattle with pneumonia between July 1, 1991 and July 31, 1996. Only isolates obtained from samples submitted by dairies and calf ranches were used. PROCEDURE: Spatial clustering of ampicillin- and tetracycline-resistant isolates was assessed by use of nearest-neighbor and Cuzick and Edwards' analyses. Linear clustering along a north-south line was assessed by use of runs and maximum length of runs tests. Temporal clustering was assessed by use of scan tests. Spatial-temporal clustering was assessed by use of Barton's method. Regional estimates of percentages of P multocida and P haemolytica resistant to ampicillin or tetracycline were calculated. RESULTS: There was significant spatial clustering of resistant isolates and significant linear clustering along a north-south line. Significant differences in regional estimates of percentages of antimicrobial-resistant isolates were found. CLINICAL IMPLICATIONS: Results support the hypothesis that antimicrobial-resistant organisms can be clustered at the local level and reinforce the need to establish regional estimates of percentages of bacterial isolates that will be susceptible to commonly used antimicrobials.
Ampicillin Resistance↗
Prevalence and control of foot-pad dermatitis in broilers in Sweden.
1. A cross-sectional observational study was carried out to investigate spatial, temporal and management aspects of foot-pad dermatitis in Swedish broilers. The efficacy of a control programme was evaluated. 2. Flock information on producer, breed, foodstuff manufacturer, region, abattoir, date of slaughter, age at slaughter, planned and actual stocking density was recorded. A total of 6988 flocks, representing approximately 110 million broilers was examined. A total of 175 broiler producers from 15 geographical regions is represented. 3. The total foot-pad score per flock ranged from 0 to 200, with a mean of 34.7 and a standard deviation of 40.9. 4. Simple linear regression analysis showed that the mean total score on a weekly basis decreased (P < 0.001) over time, mainly because of a decrease in the prevalence of severe foot-pad lesions. 5. There was an association (P < 0.05) between slaughterhouse, foodstuff supplier and geographic region respectively and the mean total foot pad score of the flocks slaughtered.
Abattoirs↗
Risk analysis of quarantine station performance: a case study of the importation of equine infectious anemia virus-infected horses into California.
We examined the risk of importing and mistakenly releasing equine infectious anemia virus (EIAV)-infected horses into California. A computer simulation model was constructed to evaluate current and alternative quarantine station procedures; 150,000 iterations were performed to simulate 15 different scenarios of 10,000 horses imported into the state over a 14-year period. Simulation results showed that under current conditions of low EIAV prevalence in exporting countries, increasing the quarantine period would not decrease the number of EIAV-infected horses mistakenly released from quarantine. In a worst case scenario of high EIAV prevalence in exporting countries, the model predicted 10 EIAV-infected horses would be imported, of these 1 or none would escape detection and would be released mistakenly if quarantine duration were 3 or 14 days, respectively. This model may be applied to other quarantine station situations for evaluating the importation risk for EIAV and other diseases.
Animals↗
Temporal aspects of foot-pad dermatitis in Swedish broilers.
This study aimed at analysing the temporal aspects of foot-pad dermatitis in Swedish broilers. The information on disease prevalence and severity was based on a 2-year foot-health surveillance programme where information on producer, breed, feed manufacturer, region, abattoir, date of slaughter and several other variables was recorded. The seasonal effects were evaluated using classical multiplicative decomposition time series analysis. This study shows that there has been a significantly consistent decrease in the prevalence of severe foot-pad lesions during the first 2 years of the programme. Looking at the development over time we found a trend-cycle component of 10.4% and a seasonal component of 48.7% resulting in a total adjusted R-square value of 58.5% for the total foot-pad score. This means that almost 60% of the variation in flock foot-pad score can be explained by the variable 'time', and that this variation was mainly related to seasonal effects but also to a general decreasing trend over the study period. However, substantial differences in temporal patterns among slaughterhouses, feed suppliers and regions were found. The time series analysis approach was found to be useful for this type of investigation when evaluating the effects of an intervention programme, and it can also be applied for projecting the future development of disease status in a stable population.
Abattoirs↗
Spatial aspects of foot-pad dermatitis in Swedish broilers.
This study aimed at analysing spatial and spatio-temporal aspects of foot-pad dermatitis in Swedish broilers. The information on disease prevalence and severity was based on a two-year foot-health surveillance programme where information on producer, breed, feed manufacturer, region, abattoir, date of slaughter and several other variables was recorded. The level of clustering in space was analysed on 2-digit zipcode level using Moran's I test which measures similarity of location. The level of clustering in space was also analysed using the Ipop test, which takes the population at risk into consideration. The examination of time-space interaction was carried out using the Barton method and the Knox method. We found a significant (p < 0.001) clustering of regions with respect to foot-pad dermatitis score using Moran's I test, and a significant (p < 0.0001) clustering in space also when related to the number of flocks delivered from each region. The flocks with very high prevalence of foot-pad dermatitis were significantly (p < 0.05) clustered in both time and space, i.e. the flocks with high prevalence of lesions came from the same geographic area during the same time periods. This information will permit us to focus the control efforts within the foot-health surveillance programme on specific regions in specific time periods, thus making the programme more effective.
Abattoirs↗
Simulation analysis of the effect of herd immunity and age structure on infection of a cattle herd with bluetongue viruses in Queensland, Australia.
A state-transition model based on Leslie matrix formulation was used to investigate the effects of herd immunity and age structure on the infection of a simulated cattle herd with bluetongue viruses under Australian climatic conditions. Increasing duration of immunity decreased the prevalence of infection. A duration of immunity of 33 months was consistent with prevalence estimates made from previous serological studies of bluetongue virus. Herd prevalence displayed slowly dampening cyclical variation over time (most pronounced when a short duration of immunity was simulated). Increasing calving and mortality risk rates in the simulated herd increased prevalence, whereas increasing age at first calving decreased prevalence. Manipulation of calving rates had the greatest effect on the predicted prevalence of infection in the herd. Simulation of a number of herd-management scenarios suggested that management systems in which cattle are bred early and where high calving rates are achieved are likely to contribute to high levels of infection with bluetongue viruses. Results confirm the importance of management factors in influencing the prevalence of infectious diseases in animal populations.
Aging↗
Effect of different sampling techniques on odds ratio estimates using hospital-based cases and controls.
Potential biases introduced by the use of hospital admission records have rarely been discussed in the veterinary literature. Veterinary Medical Teaching Hospital (VMTH) patient records kept at the University of California, Davis (UCD) School of Veterinary Medicine provide a unique opportunity to perform in-depth analyses on the effect of different control selection (sampling) techniques on odds ratio (OR) estimates for disease risk factors in a retrospective case-control study. Horses with Corynebacterium pseudotuberculosis abscesses (134) and the (secondary) study base population (source for controls) were identified, and a 'gold standard' OR for each category of the factors admission type, age, breed and sex was derived. Example data were used to calculate sampling ratios (SRs), defined as the ratio between any sample proportion (of an arbitrary risk factor) and the study base proportion for this risk factor. Sampling ratios different from 1.0 introduced biases into the observed OR estimates, when compared with the 'gold standard' OR. Three randomized samples (simple random, stratified random, systematic sampling), one matched (on date of admission) and three different diagnosis samples ('colic', 'cuts and lacerations', 'fractures') were selected from the study base, and the SRs for all categories of the four factors were derived. The matched and two different disease samples ('colic' and 'fractures') had especially wide ranges of observed SRs (and large errors in the OR estimates), whereas simple random and systematic sampling had comparably narrow ranges (less biased OR estimates). For the three randomized sampling techniques under study, repeated sampling was used to derive SR distributions. The SRs were approximately normally distributed. Analysis of variance and covariance showed that simple random and systematic sampling provided SR distributions with means closest to 1.0 (expected value) and small standard deviations. The OR estimates obtained from records selected by these two sampling techniques therefore were least biased. The findings demonstrate the importance of selecting appropriate sampling techniques in addition to properly defining the study (base) population. Sampling design introduces uncertainty into the OR estimates. The direction of the bias, however, depends on the OR between factor and disease in the source population (the 'gold standard'), and on the direction and magnitude of the SR. When combining the results from single and repeated sampling we conclude that sampling design is most influential on the range of the observed SRs (single samples), on the absolute deviation of the SR from 1.0 (expressed as SR delta Mean) and on the SR standard deviation (SD) (repeated sampling).
Animals↗
Economics of feeding pasteurized colostrum and pasteurized waste milk to dairy calves.
OBJECTIVE: To estimate the marginal contribution of pasteurization of waste milk and colostrum to gross margin per calf at weaning and to estimate the minimum number of cattle on a dairy farm for pasteurization to be profitable. DESIGN: Randomized, controlled, clinical trial. ANIMALS: 300 Holstein calves. PROCEDURE: The performance of calves fed pasteurized colostrum and waste milk was compared with the performance of calves fed nonpasteurized colostrum and waste milk. Costs, revenues, and gross margins for the 2 groups were compared. RESULTS: Calves fed pasteurized colostrum and waste milk were worth an extra $8.13 in gross margin/calf, compared with calves fed nonpasteurized colostrum and waste milk. The minimum number of cattle for which feeding pasteurized colostrum and waste milk was calculated to be economically feasible was 315 calves/d (1,260-cow dairy farm). CLINICAL IMPLICATIONS: An economic benefit was associated with feeding pasteurized colostrum and waste milk. Additional benefits that may accrue include higher mean weight gain and lower mortality rate of calves as well as calves that have fewer days in which they are affected with diarrhea and pneumonia.
Animal Feed↗
Analysis of spatial and temporal clustering of horses with Salmonella krefeld in an intensive care unit of a veterinary hospital.
OBJECTIVE: To determine whether clustering existed in the spatial or temporal distribution of horses that shed Salmonella krefeld in their feces during hospitalization. DESIGN: Retrospective analysis of medical records. ANIMALS: 219 horses housed in the intensive care unit of a veterinary medical teaching hospital from October 1991 through May 1992. PROCEDURE: Bacteriologic culturing of fecal samples was used to identify horses shedding S krefeld. For affected horses, the scan statistic was used to analyze temporal clustering, and Knox's method was used to analyze temporal-spatial clustering. RESULTS: 20 horses were identified as shedding S krefeld in their feces. Significant temporal clustering of affected horses was observed for periods of 5, 6, 7, and 8 days. Temporal-spatial analysis did not detect a significant distribution for any combination of time and distance among affected horses. CLINICAL IMPLICATIONS: Detection of temporal clustering and concurrent random temporal-spatial distribution of affected horses suggested that affected horses were grouped in time, but means of transmission was not related to proximity between horses.
Animals↗
Financial evaluation of vaccination and testing alternatives for control of parvovirus-induced reproductive failure in swine.
OBJECTIVE: To identify the preferable testing and vaccination strategy for control of porcine parvovirus (PPV) during a 6-month period. DESIGN: Decision-tree analysis and computer simulations. SAMPLE POPULATION: Computer modeling of 300-sow farrow-to-finish herd. PROCEDURE: Serologic testing of 30 females to estimate herd PPV prevalence versus not testing any females was the initial decision alternative. On the basis of serologic test results, herds were classified into 1 of 3 PPV-risk categories: low (> or = 80% seropositive females), moderate (40 to < 80% seropositive females), or high (< 40% seropositive females). Vaccinating all females, only gilts, or not vaccinating was the second decision alternative. RESULTS: For initial model assumptions (test sensitivity and specificity = 0.95; test cost = $5/female; vaccination cost = $0.30/dose; vaccination efficacy = 0.95; and foregone gross margin = $10.85/pig), vaccination of all females (with or without serologic testing) was preferable, but the financially preferable option was to omit serologic testing. Most profitable vaccination option varied with foregone gross margin, vaccination cost, and efficacy. For herds in which all sows were known to be immune, vaccinating only gilts was financially preferable, and serologic testing was not warranted. Variation is expected monetary losses was less in vaccination options than with nonvaccination. CLINICAL IMPLICATIONS: For most herds in the United States, serologic screening for PPV prior to selection of a vaccination program is unlikely to be cost-effective, because vaccination is inexpensive ($0.30/dose) and effective (95%). At current profit margins ($10.85/pig), vaccination of all females has the least-risk and is the preferred option.
Animals↗
Spatial analysis of seroconversion of sentinel cattle to bluetongue viruses in Queensland.
OBJECTIVE: To assess quantitatively the spatial distribution of seroconversion of Queensland cattle to bluetongue viruses. DESIGN: A sentinel herd study. SAMPLE POPULATION: Sixty-nine sentinel herds at 30 locations. PROCEDURE: Spatial clustering of seroconversion to bluetongue viruses was investigated during the period from 1990 to 1994. RESULTS: Seroconversion to only two bluetongue virus serotypes, 1 and 21, was observed. The 14 herds, in which seroconversion to bluetongue virus serotype 1 was detected, were located only along the eastern coastal and subcoastal region of Queensland, and were significantly (P < 0.05) clustered. Locations at which seroconversion to serotype 21 was detected, were not significantly clustered. The results generally agree with field observations, except for the failure to detect seroconversion to bluetongue viruses in north-western Queensland. CONCLUSION: Bluetongue infection of cattle in north-western Queensland may be temporally sporadic. The dominance of serotype 1 in the Queensland cattle population may be the result of differential transmission by potential vector species. Long-term surveillance programs are important for defining disease status of animal populations.
Animals↗
Infection of cattle with bluetongue viruses in Queensland, Australia: results of a sentinel herd study, 1990-1992.
Between 1990 and 1992, 47 sentinel herds of 10-20 cattle each were established throughout Queensland, Australia to monitor bluetongue virus infection. Sixteen herds at 12 locations seroconverted to bluetongue viruses during the study. Herd incidence rates ranged from 0.0 to 3.45 seroconversions per cattle-year at risk. The mean incidence rate was 0.29 seroconversions per cattle-year at risk (95% confidence interval 0.23-0.36), and the median incidence rate was 0.32 seroconversions per cattle-year at risk (95% CI 0.0-0.54). Only seroconversions to bluetongue virus serotypes 1 and/or 21 were detected in the study; no evidence was found of infection with other serotyped previously isolated in Australia (3, 9, 15, 16, 20, 23). Seroconversion to serotype 1 occurred more frequently than seroconversion to serotype 21. Seroconversion generally occurred between April and July, being most common in May, but varied considerably within each of the 3 years of the study. This study confirmed the presence of only serotypes 1 and 21 in Queensland and demonstrated the sporadic and restricted nature of bluetongue virus infection in the Queensland cattle herd.
Animals↗
Infection of cattle in Queensland with bluetongue viruses: II. Distribution of antibodies.
A survey of the distribution of serum antibodies to bluetongue viruses in Queensland cattle herds was conducted in 1989. A total of 410 herds were selected in which sera from 20 or more cattle had been tested with the agar gel immunodiffusion (AGID) test. Estimates of prevalence were standardised for age of cattle and adjusted using results from serum neutralisation (SN) tests. Spatial statistical methods were used to investigate the distribution of these selected herds with respect to the prevalence of serum antibodies to bluetongue viruses. The mean herd prevalence was 5.2% (95% confidence interval 4.5 to 5.9) and the median herd prevalence was 3.5% (95% CI 2.8 to 4.2). Prevalence was highest in far north Queensland (17.3%) and lowest in south-west Queensland (0.3%). Significant (P < 0.01) clustering of selected herds with respect to prevalence of serum antibodies was found. Herds with prevalence of serum antibodies to bluetongue viruses above the overall mean herd prevalence were significantly (P < 0.05) clustered in the north-west, far north and northern and southern coastal regions of Queensland. Clustering within the north-west and southern coastal regions of Queensland was more pronounced than the clustering detected in the northern coastal and far north regions.
Animals↗