PubMed Health⌕ Search

SEARCH · PubMed Health

Results for “Conditional logit”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

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

179 records · Page 10Linked to original sources

Use of ordinal categorical variables in skeletal assessment of sex from the cranium.

In anthropological studies, visual indicators of sex are traditionally scored on an ordinal categorical scale. Logistic and probit regression models are commonly used statistical tools for the analysis of ordinal categorical data. These models provide unbiased estimates of the posterior probabilities of sex conditional on observed indicators, but they do so only under certain conditions. We suggest a more general method for sexing using a multivariate cumulative probit model and examine both single indicator and multivariate indicator models on a sample of 138 crania from a Late Mississippian site in middle Tennessee. The crania were scored for five common sex indicators: superciliary arch form, chin form, size of mastoid process, shape of the supraorbital margin, and nuchal cresting. Independent assessment of sex for each individual is based on pubic indicators. The traditional logistic regressions are cumbersome because of limitations imposed by missing data. The logistic regression correctly classified 66/74 males and 46/64 females, with an overall correct classification of 81%. The cumulative probit model classified 64/74 males correctly and 51/64 females correctly for an overall correct classification rate of 83%. Finally, we apply parameters estimated from the logit and probit models to find posterior probabilities of sex assignment for 296 additional crania for which pubic indicators were absent or ambiguous.

Aging↗

Age at menopause in Latin America.

OBJECTIVE: To assess the age at menopause (AM) in Latin America urban areas. DESIGN: A total of 17,150 healthy women, aged 40 to 59 years, accompanying patients to healthcare centers in 47 cities of 15 Latin American countries, were surveyed regarding their age, educational level, healthcare coverage, history of gynecological surgery, smoking habit, presence of menses, and the use of contraception or hormone therapy at menopause. The AM was calculated using logit analysis. RESULTS: The mean age of the entire sample was 49.4 +/- 5.5 years. Mean educational level was 9.9 +/- 4.5 years, and the use of hormone therapy and oral contraception was 22.1% and 7.9%, respectively. The median AM of women in all centers was 48.6 years, ranging from 43.8 years in Asuncion (Paraguay) to 53 years in Cartagena de Indias (Colombia). Logistic regression analysis determined that women aged 49 living in cities at 2,000 meters or more above sea level (OR = 2.0, 95% CI: 1.4-2.9, P < 0.001) and those with lower educational level (OR = 1.9, 95% CI: 1.3-2.8, P < 0.001) or living in countries with low gross national product (OR = 2.1, 95% CI: 1.5-2.9, P < 0.001) were more prone to an earlier onset of menopause. CONCLUSIONS: The AM varies widely in Latin America. Lower income and related poverty conditions influence the onset of menopause.

Adult↗

Commitment to work and the self-perception of disability.

Population surveys of the disabled frequently show large numbers of persons who report work activity despite a simultaneous response indicating that they are severely disabled. Using multinomial logit analysis, this article examines the characteristics of persons who express such a commitment to work. The characteristics examined are sex, race, age, family size, education, marital status, health status, and length of work experience prior to onset of disability. The logit analysis allows us to see which of these variables predict work behavior. Data from the 1972 Survey of Disabled and Nondisabled Adults and the 1974 Followup Survey are used. Data are shown in cross section (1972 and 1974) and longitudinally (patterns of change from 1972 to 1974). In 1972, the less severe the level of health problems, the higher the level of education, being male, and needing to help relatives financially were traits associated with the probability of working despite self-report of severe disability. Those who received funds from public income maintenance were not likely to have worked. Essentially, the same findings were obtained for 1974 cross sectionally and 1972-74 longitudinally. Comparison was also made between those severely disabled persons who worked full time and those who worked part time, for both 1972 and 1974. Basically, age and severity of health condition were associated with the probability of working full time as opposed to part time, suggesting that it is a work/no work decision that is more likely to discriminate among the work commitments of the severely disabled than is the type of work schedule.

Adaptation, Psychological↗

Optimal statistical design of radioimmunoassays and competitive protein-binding assays.

The statistical analysis of radioimmunoassay and competitive protein binding assay data is complex. Because the response variable (percent counts) is not lineraly related to log dose, a logit transformation of the response variable usually is performed to permit linear regression analysis. This transformation induces marked heterogeneity of variance, so that iterative weighted regression programs have been used to achieve the best standard curve and the most precise dose estimates of unknowns. In this study several parameters of assay design are investigated in order to establish those designs yielding antigen concentration estimates of highest precision as well as estimates of comparable precision by either simple linear regression analysis or by the more complex weighted regression technique. Unknown estimates of highest precision are obtained when 1) the present counts of the standard doses covers a range of approximately 80 percent to 20 percent, 2) the number of standard dose levels is eight or more, 3) the number of replicates at each dose level is two or more, and 4) the percent counts of the unknowns also are within the range 80 percent ot 20 percent. Under these conditions, also, simple linear regression yields unknown estimates of comparable precision to weighted regression and therefore may be safely used.

Antigens↗

The accuracy of substituted judgments in patients with terminal diagnoses.

BACKGROUND: Patients' loved ones often make end-of-life treatment decisions, but the accuracy of their substituted judgments and the factors associated with accuracy are poorly understood. OBJECTIVE: To assess the accuracy of judgments made by surrogate decision makers; ascertain the beliefs, practices, and clinical and sociodemographic factors associated with accuracy of surrogates' decisions; assess the preferences of patients for life-sustaining treatments; and compare differences in accuracy across diagnoses. DESIGN: Cross-sectional paired interviews. SETTING: Outpatient practices of three university hospitals. PATIENTS: 250 patients with terminal diagnoses of congestive heart failure, AIDS, amyotrophic lateral sclerosis, lung cancer, and chronic obstructive pulmonary disease (50 patient-surrogate pairs in each group) and 50 general medical patients and their surrogates. MEASUREMENTS: The accuracy of surrogate predictions was measured by using scales based on 10 potential treatments in each of three hypothetical clinical scenarios. RESULTS: Preferences varied according to mode of treatment and scenario. On average, surrogates made correct predictions in 66% of instances. Accuracy was better for the permanent coma scenario than for the scenarios of severe dementia or coma with a small chance of recovery (P < 0.001). In a binary logit model, the accuracy of substituted judgments was positively associated with the patient having spoken with the surrogate about end-of-life issues (odds ratio [OR], 1.9 [95% CI, 1.6 to 2.3]), the patient having private insurance (OR, 1.4 [CI, 1.1 to 1.7]), the surrogate's level of education (OR, 1.5 [CI, 1.2 to 1.9]), and the patient's level of education (OR, 1.7 [CI, 1.4 to 2.2]). Accuracy was negatively associated with the patient's belief that he or she would live longer than 10 years (OR, 0.6 [CI, 0.5 to 0.7]), surrogate experience with life-sustaining treatment (OR, 0.4 [CI, 0.3 to 0.5]), surrogate participation in religious services (OR, 0.67 [CI, 0.50 to 0.91]), and a diagnosis of heart failure (OR, 0.6 [CI, 0.5 to 0.8]). Age, ethnicity, marital status, religion, and advance directives were not associated with accuracy. CONCLUSIONS: The accuracy of substituted judgments is associated with multiple clinically apparent patient and surrogate factors. This information can help clinicians identify conditions under which substituted judgments are likely to be accurate or inaccurate and can help target populations for education designed to improve the accuracy of surrogate decision making.

Adult↗

A discrete choice model of drug abuse treatment location.

OBJECTIVE: To identify short-term drug abuse treatment location risk factors for ten large, self-insured firms starting January 1, 1989 and ending December 31, 1991. DATA SOURCES/STUDY SETTING: Study population selected from a large database of health insurance claims for all treatment events starting January 1, 1989 and ending December 31, 1991. STUDY DESIGN: A nested binomial logit method is used to estimate firm-specific patterns of treatment location. The differences in treatment location patterns among firms are then decomposed into firm effects (holding explanatory variables constant among firms) and variable effects (holding firm-specific parameters constant). PRINCIPAL FINDINGS: Probability of inpatient drug treatment is directly related to the type of drug diagnosis. The most important factors are diagnoses of drug dependence (versus drug abuse) and/or a cocaine dependence. Firm-specific factors also make a substantive difference. Controlling for patient risk factors, firm-specific probabilities of inpatient treatment vary by as much as 87 percent. Controlling for practices of firms and their insurance carriers, differing patient risk profiles cause probabilities of inpatient treatment to vary by as much as 69 percent among firms. Use of the outpatient setting increased over the three-year period. CONCLUSIONS: There are two plausible explanations for the findings. First, people beginning treatment later in the three-year period had less severe conditions than earlier cases and therefore had less need of inpatient treatment. Second, drug abuse treatment experienced the same trend toward the increased use of outpatient care that characterized treatment for other illnesses in the 1980s and early 1990s.

Adult↗

Cadmium inhalation exposure estimates: their significance with respect to kidney and liver cadmium burden.

Cadmium exposure histories based on employment records, area monitoring techniques, and personal sampling data were obtained for 82 industrially exposed workers. From these data, a time-weighted cumulative exposure index (TWE) was calculated for each worker by multiplying the length of time (ti) in a given work area by the estimated exposure conditions for that area and year (Ei) and then summing these values for the total exposure history TWE = sigma iEiti The cadmium body burden for each worker was measured directly by the in vivo prompt-gamma neutron activation technique. The cadmium content of the left kidney and the liver were measured. For the actively employed workers, a significant correlation (r = 0.70, p less than 0.001) was observed between the exposure index (TWE) and the liver cadmium burden. Furthermore, whenever the worker's liver burden exceeded 40 ppm and the exposure index exceeded 400-500 micrograms/m3 X yr, there was evidence of renal abnormalities (usually elevated urinary beta 2-microglobulin). The highest correlation (r = 0.83, p less than 0.001) was obtained between the kidney cadmium burden data and the exposure index for the active workers without evidence of kidney dysfunction. The percentage of workers with renal abnormalities was found to increase as the exposure index increased. When this relationship was examined using linear logistic regression analysis, the following model was indicated: logit p = 1.24 In TWE--8.34 where p is the individual worker's probability of being classified as having renal dysfunction.

Adult↗

[Comparative study of the effect of 3 M-cholinergic blockaders on memory and motor activity in rats].

Experiments on rats were made to study and compare pharmacologically three central antimuscarinic drugs, namely scopolamine, N-methyl-4-piperidinol benzylate and tropine phenylcyclopentylglycolate. A study was made of the effects of the compounds on rat motor activity by the Animex system and on short-term memory by passive avoidance behavior. Logit analysis was employed to define the ED50 of the compounds that inhibit the arecoline-induced tremor. Definite differences were discovered in the effects of the antimuscarinic drugs on CNS functions. For example, M-methyl-4-piperidinol benzylate applied in a wide dosage range induced motor hyperactivity without affecting the short-term memory, whereas tropine phenylcyclopentylglycolate provoked motor hyperactivity and memory disorders at equal doses. As regards the doses producing both the effects, scopolamine occupied the intermediate position. No correlation was found between central antimuscarinic activity of the compounds under study and their influences on motor hyperactivity and formation of passive avoidance behavior.

Animals↗

Return and onward migrations in Canada, 1976-1981: an explanation based on personal and ecological variables.

"This paper applies a three-level nested logit model to the micro data of the 1981 Canadian census to explain the 1976-81 interprovincial migration choices of the non-natives (those whose province of residence was different from province of birth), aged 20 to 44, by personal factors and provincial attributes. Important personal factors include mother tongue, level of education, family type, and age. Influential provincial attributes include economic variables (income level, employment growth, and unemployment), distance, and cultural similarity. The main finding is that not only onward migrants but also return migrants were sensitive to the interprovincial variation in economic opportunities." (SUMMARY IN FRE)

Americas↗

The relationship between severity of obstructive sleep apnoea/hypopnoea syndrome (OSAHS) and lateral cephalometric radiograph values: a clinical diagnostic tool.

BACKGROUND: This prospective cross-sectional cohort study examined the relationship between radiographic anatomy and the severity of OSAHS. The severity of OSAHS can be measured subjectively in terms of the Epworth scale and objectively in terms of the apnoea/hypopnoea Index (AHI). METHODS: 121 lateral cephalometric radiographs were traced under uniform conditions and a series of 56 landmarks identified, from which 48 angular and linear measurements were made. Significant changes occurred when comparison of these measurements with the severity of OSAHS were made. RESULTS: Body mass index (BMI), the maxillary-mandibular planes angle (MMPA) and the linear measurement between points 7 and 12, (the pharyngeal dimension measured from the tip of the soft palate to the corresponding horizontal point on the posterior pharynx), increased significantly with increasing severity of OSAHS as measured by the Epworth score. Overjet, lower lip length, and the distance from the hyoid bone to a point B on the mandible all increased significantly with increasing severity of OSAHS, as measured by the AHI. The hyoid bone was found to rotate counter clockwise as the severity of OSAHS increased, as a result the distance between the most anterior superior point on the hyoid bone and the maxillary plane was seen to decrease as severity of OSAHS increased in terms of AHI. CONCLUSIONS: Some radiographic anatomical features show significant change as the severity of OSAHS increases and these features could be used in the identification of patients who have severe OSAHS. The Logit equation derived from the findings of this study may also be a useful clinical tool in predicting the likelihood of a subject suffering from severe OSAHS.

Body Mass Index↗

Influence of body weight, age, and weight gain on fertility and prolificacy in four breeds of ewe lambs.

Breeding ewes to lamb at 1 yr of age can improve profitability for some production systems. The first objective of this study was to evaluate the effect of age and weight at breeding and total postweaning weight gain on reproductive performance of ewe lambs. The second objective was to compare the effects of weight and age variables in four major sheep breeds (Columbia, Polypay, Rambouillet, and Targhee). Weights, ages, and the binary traits of fertility (pregnant or nonpregnant) and prolificacy (one lamb born vs. two or more) were collected on 2,055 ewe lambs at the U.S. Sheep Experiment Station, Dubois, ID, from 1984 through 1988. The effects of age and weight at breeding and total weight gain from weaning to breeding on fertility and prolificacy were analyzed with a logit model in a maximum likelihood analyses. Differences (P < 0.001) among breeds for fertility were identified, with a 93% fertility rate for Polypay ewe lambs compared with lower fertility rates in Columbia, Targhee, and Rambouillet ewe lambs (50, 60, and 75%, respectively). The percentage of multiple births (prolificacy rate) also was higher (P < 0.001) in the Polypay (47%) than in Columbia, Targhee, and Rambouillet breeds (1, 13, and 14%, respectively). Averaged across breeds, weight at breeding had a positive effect on fertility and prolificacy (P < 0.001), whereas total weight gain from weaning to breeding had a positive effect only on fertility (P < 0.027). In separate analyses for each breed, increasing age (P < 0.001) and weight at breeding (P < 0.001) increased the probability of pregnancy in Rambouillet ewe lambs. The probability of pregnancy for Targhee ewe lambs increased (P < 0.005) with weight at breeding. Increasing weight at breeding increased (P < 0.004) the probability of multiple births in all breeds. Increasing total postweaning weight gain increased (P < 0.007) the probabilities of multiple births in Rambouillet and Targhee ewe lambs. In conclusion, Polypay ewe lambs were superior in fertility and prolificacy to Columbia, Rambouillet, and Targhee under Western range conditions. Improved reproductive performance of Columbia, Rambouillet, and Targhee ewe lambs may be achieved by increasing age and weight at breeding and postweaning gain.

Age Factors↗

An assessment of non-randomized medical treatment of long-term schizophrenia relapse using bivariate binary-response transition models.

The analyses of observational longitudinal studies involving concurrent changes in treatment and medical conditions present difficulties because of the multitude of directions of potential relationships: past medication influences current symptoms; past symptoms influence current medication; and current medication is associated with current symptoms. In the context of a long-term study of non-randomized pharmacological treatment of schizophrenic relapse, we present an analysis of bivariate discrete-time transitional data with binary responses in an attempt to understand the transitional and concurrent relationships between schizophrenia relapse and medication use. A naive analysis does not show any association between previous medication and current relapse. However, we provide evidence suggesting that current treatment may impact current relapse for those who have previously taken medication, but not for those who haven't taken medication in the past. When univariate models are specified to assess these associations, the bivariate nature of the problem requires a choice of which response, relapse or medication, should be the dependent variable. In this case, the choice of relapse or medication as a dependent variable does matter. Hence, our results derive from models where both relapse and medication are treated as dependent variables. Specifically, we specify a bivariate log odds ratio for current relapse and current medication use and a separate univariate logit component for each of these outcomes. Each of these components contains transitional associations with previous relapse and medication. Such models represent extensions of univariate transitional association models (e.g. Diggle et al. (1994)) and correspond to bivariate transitional models (e.g. Zeger and Liang (1991)). We incorporate changes in transitional associations into the full-data parametric model for final inference, and investigate if these temporal changes are due to learning effects or the impact of drop-out. We also perform residual analyses and sensitivity analyses in the context of missing data patterns.

Journal Article↗

A comparison of unmet needs for dental and medical care among persons with HIV infection receiving care in the United States.

OBJECTIVE: Oral health conditions associated with HIV disease are frequently more severe than those of the general population, making access to both dental and medical care important. Using the domains specified in the Behavioral Model of Health Services Use, this paper examines the correlates of unmet needs for dental and medical care in a nationally representative sample of patients with HIV. METHODS: This investigation is a cross-sectional study using baseline data from the HIV Cost and Services Utilization Study (HCSUS), the first nationally representative study of persons in care for HIV. Using probability-based techniques, 4,042 people were randomly selected in January 1996, and 2,864 (71%) completed a structured interview that included questions on unmet needs for dental and medical care. Regression analysis was used to identify variables associated with having unmet needs for dental care only, medical care only, and both medical and dental care. RESULTS: Of the estimated 230,900 people in treatment for HIV in the United States, approximately 58,000 had unmet medical or dental needs based on self-reported data. Unmet dental needs were more than twice as prevalent as unmet medical needs (32,900 vs 14,300), and 11,600 people were estimated to have both types of unmet needs. Multinomial logit regression showed that persons with low income had increased odds of reporting unmet needs for both dental and medical care. The uninsured and those insured by Medicaid without dental benefits had more than three times the odds of having unmet needs for both types of care than did the privately insured. CONCLUSIONS: To serve both the dental and medical needs of diverse populations affected by HIV disease, greater coordination of services is needed. In addition, state insurance programs for people with HIV should consider the feasibility of expanding their benefit structure to include dental care benefits.

Adult↗

Predicting audiometric status from distortion product otoacoustic emissions using multivariate analyses.

OBJECTIVES: 1) To determine whether multivariate statistical approaches improve the classification of normal and impaired ears based on distortion product otoacoustic emission (DPOAE) measurements, in comparison with the results obtained with more traditional single-variable applications of clinical decision theory. 2) To determine how well the multivariate predictors, derived from analysis on a training group, generalized to a validation group. 3) To provide a way to apply the multivariate approaches clinically. DESIGN: Areas under the relative operating characteristic (ROC) curve and cumulative distributions derived from DPOAE, DPOAE/Noise, discriminant function (DF) scores and logit function (LF) scores were used to compare univariate and multivariate predictors of audiometric status. DPOAE and Noise amplitudes for 8 f2 frequencies were input to a discriminant analysis and to a logistic regression. These analyses generated a DF and LF, respectively, composed of a linear combination of selected variables. The DF and LF scores were the input variables to the decision theory analyses. For comparison purposes, DPOAE test performance was also evaluated using only one variable (DPOAE or DPOAE/Noise when f2 = audiometric frequency). Analyses were based on data from over 1200 ears of 806 subjects, ranging in age from 1.3 to 96 yr, with thresholds ranging from -5 to >120 dB HL. For statistical purposes, normal hearing was defined as thresholds of 20 dB HL or better. For the multivariate analyses, the database was randomly divided into two groups of equal size. One group served as the "training" set, which was used to generate the DFs and LFs. The other group served as a "validation" set to determine the robustness of the DF and LF solutions. RESULTS: For all test frequencies, multivariate analyses yielded greater areas under the ROC curve than univariate analyses, and greater specificities at fixed sensitivities. Within the multivariate techniques, discriminant analysis and logistic regression yielded similar results and both yielded robust solutions that generalized well to the validation groups. The improvement in test performance with multivariate analyses was greatest for conditions in which the single predictor variable resulted in the poorest performance. CONCLUSIONS: A more accurate determination of auditory status at a specific frequency can be obtained by combining multiple predictor variables. Although the DF and LF multivariate approaches resulted in the greatest separation between normal and impaired distributions, overlap still exists, which suggests that there would be value in continued efforts to improve DPOAE test performance.

Acoustic Stimulation↗

The radioimmunoassay of follicle-stimulating hormone (FSH)-suppressing protein (FSP): stimulation of bovine granulosa cell FSP secretion by FSH.

A RIA for bovine (b) FSH-suppressing protein (FSP) was developed using an antiserum raised in a rabbit to purified 39-kDa bFSP, iodinated 35-kDa FSP as tracer, and purified 35-kDa bFSP as standard. Purified 35-kDa FSP was iodinated using the iodogen procedure, and the iodinated FSP was purified by dye affinity chromatography. After a logit log-dose transformation of the dose-response curves, parallel displacement lines were observed between 31-, 35-, and 39-kDa FSP, bovine follicular fluid, bovine granulosa cell culture medium, and medium from bovine granulosa cells stimulated with bFSH. The specificity of the assay was investigated by comparing the immunoassay levels of FSP with in vitro bioassay levels based on the ability of FSP/inhibin to suppress FSH in rat anterior pituitary cell cultures in fractions obtained throughout the purification procedure of FSP from bovine follicular fluid. This demonstrated that 1) the FSP immunoactivity was associated with in vitro bioactivity in all fractions of the purification procedure; 2) a number of inhibin-related and other proteins showed low (less than 0.5%) or nondetectable cross-reactivity in the RIA; and 3) the in vitro biological to immunological ratios for 31-, 35-, and 39-kDa FSP were similar, indicating that the RIA detects all forms of purified bFSP. The secretion of FSP by bovine granulosa cells in culture was investigated in the presence and absence of bFSH and bLH, respectively. FSP production was proportional to granulosa cell number and decreased from highest levels at 24 h to lowest levels at 96 h of culture. The addition of either bFSH or 8-bromo-cAMP to the culture medium stimulated FSP production by a factor of 2-3 at 48 and 72 h of culture, while the addition of bLH had no effect on FSP production. Theca interna tissue cultured under the same conditions did not produce FSP. In contrast to FSP, stimulation of bovine granulosa cells with bFSH or bLH had no effect on inhibin production during the 96 h of culture, while the addition of bFSH and bLH caused a stimulation of progesterone production at 48 and 72 h of culture. It is concluded that 1) the RIA described here is specific for all mol wt forms of bFSP; 2) FSP was secreted by bovine granulosa cells and not thecal cells in vitro; and 3) FSP secretion by bovine granulosa cells in vitro is regulated by bFSH and not bLH.

Animals↗

Social environmental factors and preteen health-related behaviors.

PURPOSE: To examine associations among risk and protective factors with negative (substance use, delinquent behavior, sedentary recreation, empty calorie consumption, suicidal behavior) and positive behaviors (prosocial recreation, productive behavior, exercise, nutrition behavior, academic achievement, seatbelt use). METHODS: Data were from sixth-grade public school students (n = 133,629) in 2001. Factor analysis produced five risks, five protectors, seven negative and six positive behaviors. Associations were tested among single and cumulative risks and protectors with behaviors (linear, logit regression) and combinations of high and low risks and protectors with behaviors (analysis of variance, Chi-square). RESULTS: Individual and cumulative risks were related to higher and protectors were related to lower negative behaviors. Protectors were associated with higher positive behaviors, with some exceptions. Risks and their sum were associated with lower academic achievement and seatbelt use, but were linked to higher, rather than lower productive behavior. Being bullied or victimized was correlated with higher levels of exercise, good nutrition, and prosocial recreation. The high risk/low protection combination had the highest negative behaviors and low risk/high protection the lowest, but for positive behaviors, high protectors with either high or low risks showed higher positive behaviors. CONCLUSIONS: These findings provide new information about how (a) risks and protectors are associated with negative behaviors besides substance use and delinquency, (b) cumulative protectors, as well as risks, are related to negative and positive behaviors, and (c) combinations of high and low risks and protectors are related to behaviors. The unusual findings for positive behaviors merit further exploration.

Activities of Daily Living↗

A comparison of survival models for assessing risk of racehorse fatality.

Survival analysis was used to assess risk factors for fatal injuries on UK race courses. This allowed assessment of variation due to temporal horse-level effects, including previous racing intensity and historical distribution of race types, as well as race-level factors. Comparisons were made between measuring survival time as number of days and as number of races to injury from the first race. Two related models were presented for time as number of races to injury: a Cox regression model fitted using partial likelihood, with the Efron approximation to handling ties, and a discrete-time logit model fitted using maximum likelihood. The latter approach had the advantages of being computationally more efficient and enabling the testing of different functional forms for the dependence of hazard on time. Retrospective data were available from all race starts on the 59 courses in Britain from 1990 to the end of 1999, as analysed by . The analysis was conducted on the data for the 47,424 horses that had started racing in the UK: 538,895 starts with 1,228 fatal injuries. Horses starting racing abroad were excluded, but some included horses would have raced abroad at some stage during their racing career. The results for the selected models were broadly consistent with each other and with previously published studies. Steeplechase and hurdle races had a higher risk of fatal injury than flat races (relative hazards 1.5 and 1.7, respectively). Risk increased with the firmness of surface, age and race distance (reaching a plateau at 20 furlongs) and decreased with previous racing intensity (reaching a plateau after seven races run in the last 12 months). Horses running their first race of a new type were also found to be at higher risk (relative hazard 1.5). The main difference between the models for time as number of days and number of races concerned the role of age: age at race was identified as the more important factor in the latter model, whereas, age at first race was more significant in the former model.

Age Factors↗