Re: "Defining exposure in case-control studies: a new approach".
Explore the source record for details and available documents.
Case-Control Studies↗
Biomedical subjects
S L Zeger
Publications and source records attributed to S L Zeger.
The effects on survival of early treatment of human immunodeficiency virus infection.
BACKGROUND: Zidovudine has been shown to prolong survival in patients with the acquired immunodeficiency syndrome (AIDS) and, in persons with human immunodeficiency virus (HIV) infection but not AIDS, to delay the progression to AIDS. However, it is still uncertain whether treatment before the development of AIDS prolongs survival. METHODS: We analyzed data from a cohort of 2162 high-risk men who were already seropositive for HIV type 1 (HIV-1) and 406 men who seroconverted from October 1986 through April 1991. There were 306 deaths. The probabilities of death were compared among men at similar stages of disease who began zidovudine therapy before the diagnosis of AIDS and among those who did not. Relative risks of death were calculated for each of five initial disease states on the basis of CD4+ cell counts and clinical symptoms and signs appearing over follow-up periods of 6, 12, 18, and 24 months. Adjustments were also made for the use of prophylaxis against Pneumocystis carinii pneumonia (PCP). RESULTS: After we controlled for CD4+ cell count and symptoms, the use of zidovudine with or without PCP prophylaxis before the development of AIDS significantly reduced mortality in all follow-up periods. The relative risks of death were 0.43 (95 percent confidence interval, 0.23 to 0.78) at 6 months, 0.54 (95 percent confidence interval, 0.38 to 0.78) at 12 months, 0.59 (95 percent confidence interval, 0.44 to 0.79) at 18 months, and 0.67 (95 percent confidence interval, 0.52 to 0.86) at 24 months. After we adjusted for the effects of PCP prophylaxis, zidovudine alone significantly reduced mortality at 6, 12, and 18 months (relative risks, 0.45, 0.59, and 0.70, respectively), but not at 24 months (relative risk, 0.81). Among zidovudine users, those who also used PCP prophylaxis before the development of AIDS had significantly lower mortality at 18 and 24 months than those who did not (relative risks, 0.62 and 0.60, respectively). CONCLUSIONS: The results of this study support the hypothesis that in HIV-1 infection, early treatment with zidovudine and PCP prophylaxis improves survival in addition to slowing the progression to AIDS.
CD4-Positive T-Lymphocytes↗
The effect of state cigarette tax increases on cigarette sales, 1955 to 1988.
We evaluated the effect of state cigarette tax increases on cigarette sales in the 50 states for the years 1955 to 1988. State cigarette tax increases were associated with an average decline in cigarette consumption of three cigarette packs per capita (about 2.4%). Larger tax increases were associated with larger declines in consumption. Raising state cigarette taxes appears to be an effective public health intervention that can reduce cigarette consumption and its associated health consequences.
Bias↗
Generalized linear models with random effects; salamander mating revisited.
In recent years much effort has been devoted to extending regression methodology to non-Gaussian data, where responses are not independent. These methods for dependent responses are suitable for data from longitudinal studies or nested designs. However, use of these methods for crossed designs seems to have serious limitations due to the intensive computations involved because of the intractable nature of the joint distribution. In this paper, we cast the problem in a Bayesian framework and use a Monte Carlo method, the Gibbs sampler, to avoid current computational limitations. The flexibility of this approach is illustrated by analyzing the interesting salamander mating data reported by McCullagh and Nelder (1989, Generalized Linear Models, 2nd edition, London: Chapman and Hall).
Analysis of Variance↗
Statistical reasoning in epidemiology.
Explore the source record for details and available documents.
Biometry↗
Effect of zidovudine and Pneumocystis carinii pneumonia prophylaxis on progression of HIV-1 infection to AIDS. The Multicenter AIDS Cohort Study.
Although used widely, the effectiveness of zidovudine therapy and primary prophylaxis for Pneumocystis carinii pneumonia (PCP) in HIV-1-infected individuals, has not been assessed in a large cohort. We have done an observational study between October, 1986, and October, 1990, of a cohort of 2145 HIV-1-seropositive men and 371 who seroconverted during the study. A Markov chain transitional analysis was used to examine the effect of zidovudine and PCP prophylaxis on the probability of progression of HIV-1 infection to AIDS (after 6, 12, 18, and 24 months) after follow-up visits categorised into one of six disease states. The six starting states were based on CD4+ lymphocyte counts and the presence of HIV-related symptoms. Use of pre-AIDS zidovudine and PCP prophylaxis was associated with significant reductions in rates of progression to AIDS at 6, 12, 18, and 24 months for participants starting with less than 350 CD4+ lymphocytes/microliter. For those starting with 350 or more CD4+ lymphocytes/microliter, non-significant protective trends were seen during 12, 18, and 24 month intervals. In multivariate log-linear models virtually all the treatment effect was due to zidovudine. However, after adjusting for the effects of zidovudine, PCP prophylaxis reduced significantly the probability of progression to a first episode of PCP during 6, 12, 18, and 24 month intervals. This study suggests that early primary PCP prophylaxis is effective in preventing first episodes of PCP, and that the efficacy of zidovudine demonstrated in clinical trials can be translated to the population level.
Acquired Immunodeficiency Syndrome↗
On estimating efficacy from clinical trials.
We define 'biologic efficacy' as the effect of treatment for all persons who receive the therapeutic agent to which they were assigned. It measures the biologic action of treatment among compliant persons. In a randomized trial with one treatment and one placebo control, one can theoretically estimate efficacy by comparing persons who complete the treatment regimen with controls who similarly complete the control regimen. In practice, however, we make this comparison with reservation because a control protocol often presents a different challenge for compliance than does the treatment, so that the compliant subgroups are not comparable. Standard practice employs intent-to-treat comparisons in which one compares those randomized to treatment and control without reference to whether they actually received the treatment. Intent-to-treat comparisons estimate the 'programmatic effectiveness' of a treatment rather than its biologic efficacy. This paper introduces and derives the statistical properties of an alternative estimator of biologic efficacy that avoids the potential selection bias inherent in a comparison of compliant subgroups. The method applies to randomized trials with a dichotomous outcome measure, whether or not a placebo is given to the control group. The idea is to compare the compliers in the treatment group to an inferred control subgroup chosen to eliminate selection bias. The methodology was motivated by and is illustrated in the analysis of a randomized community trial of the impact of vitamin A supplementation on children's mortality.
Bias↗
An application of longitudinal methods to the analysis of menstrual diary data.
Despite the considerable morbidity associated with menstrual dysfunction and mounting evidence that women's endogenous endocrine environment influences their long term health, epidemiologic investigation of menstruation is limited. A major obstacle has been the difficulty in analyzing menstrual diary data. This paper describes the variability in menstrual cycle length in college women using a longitudinal perspective. We first characterize the distribution of cycle length and show that it can be approximated by a mixture of a nearly symmetric distribution centered at 28 days and a stochastically larger component which produces a long right tail. After assessing the degree of heterogeneity in cycling patterns, we propose an analytical approach that examines cycle lengths within the symmetric portion of the distribution and cycle lengths within the tail of the distribution separately using random effects models.
Adolescent↗
Zidovudine use in AIDS-free HIV-1-seropositive homosexual men in the Multicenter AIDS Cohort Study (MACS), 1987-1989.
Zidovudine use data were examined in the Multicenter AIDS Cohort Study to determine (i) if the proportion of pre-AIDS participants (i.e., CD4+ cells less than 200/mm3 or AIDS-related complex) taking zidovudine is high enough to explain a slower than expected rise in AIDS incidence in U.S. homosexual men since mid-1987; (ii) which factors are associated with starting zidovudine and clinical trials of zidovudine; and (iii) if pre-AIDS patients, as a group, are being undertreated. Data on zidovudine use, clinical trial participation, and sociodemographic, clinical, and hematologic variables were collected every 6 months from 1,195 AIDS-free HIV-1-seropositive homosexual men from April 1987 to September 1989. Overall prevalence of zidovudine use rose from 3.6% in mid-1987 (visit 7) to 23% in mid-1989 (visit 11). Of those with less than 200 CD4+ lymphocytes/mm3, the prevalence of zidovudine use rose from 23% (24% if those taking zidovudine or placebo as part of a clinical trial are included) at visit 7 to 58% (69%) at visit 11. Of those with ARC, 20% (23%) were using zidovudine at visit 7 and 55% (65%) at visit 11. Although numbers were small, the advanced ARC participants (CD4+ cells less than 200/mm3 and two or more symptoms) reported the highest treatment rates (50, 78, 80, 60, and 74% at visits 7-11, respectively). By September 1989, 42% (31%) of those with CD4+ lymphocyte levels less than 200/mm3 were still not receiving zidovudine, suggesting that many high-risk, pre-AIDS individuals are being undertreated. To explore this finding further, we examined a range of sociodemographic, hematologic, and clinical variables to determine which factors best predicted initiation of zidovudine therapy outside of clinical trials. In multivariate analyses, CD4+ lymphocyte number was the most consistent predictor of initiation of therapy over all four study visits. For each 100 cells/mm3 deficit, the odds ratios were 2.3 (95% C.I. of 1.7-3.1) at visit 7 and 1.7% (95% C.I. of 1.4-2.0) at visit 11. Symptom status and education level were also associated with starting zidovudine, but not at all visits. The relatively low predictive power of the clinical variables raises and the possibility that nonclinical factors not measured in the MACS (drug cost, third-party insurance restrictions, and individual preferences) may play an important role in predicting zidovudine use. Finally, comparisons were made between seropositive participants starting clinical trials of zidovudine and the rest of the study population. No important differences were found in demographic or major clinical variables between clinical trial participants and zidovudine nonusers in this study.
AIDS-Related Complex↗
Risk factors for regret after tubal sterilization: 5 years of follow-up in a prospective study.
The Collaborative Review of Sterilization is a prospective, multicenter study that interviewed 7,590 women before they underwent tubal sterilization and then conducted yearly follow-up interviews that included questions on sterilization regret. These women contributed 26,641 observations (for up to 5 years after the procedure, 1978 to 1988) to an analysis of the presterilization characteristics most consistently associated with poststerilization regret. Young age at the time of sterilization was the strongest predictor of regret, regardless of parity or marital status; among women 20 to 24 years of age at sterilization, an average of 4.3% reported regret over the follow-up period. The rate of regret was significantly lower for women 30 to 34 years of age (2.4%).
Adult↗
Possible beneficial effects of neutralizing antibodies and antibody-dependent, cell-mediated cytotoxicity in human immunodeficiency virus infection.
We studied the relationship between early human immunodeficiency virus type 1 (HIV-1) specific immune responses and pathogenesis of infection in participants enrolled in the multicenter AIDS cohort study (MACS). Sera collected at 6-month intervals for 2 years (visit 1-5) from 39 persons who seroconverted by enzyme-linked immunosorbent assay (ELISA) 6 months (visit 2) after enrollment were examined for isotype-specific Western blot reactivity, neutralizing antibodies (NA) against two divergent strains of HIV-1 (HIV-1IIIB and HIV-1RF), and for antibodies capable of participating in antibody-dependent, cell-mediated cytotoxicity (ADCC). These results were compared with changes in CD4+ cell number and episodes of lymphadenopathy. Twenty-five subjects had antibodies of at least one isotype reactive to at least one HIV-1 protein by Western blot at visit 1, before they became ELISA positive. NA reactive with HIV-1IIIB were detected before those reactive with HIV-1RF. NA were first observed in 11 sera at visit 2, in 22 sera at visit 3, and in 3 sera at visit 4; sera from three patients remained nonneutralizing through visit 5. In most cases, NA were detected after a decline in CD4+ cell numbers. The data are consistent with the interpretation that NA develop after about 16 to 18 months of declining CD4+ cell numbers, following which the rate of decline in CD4+ cell numbers slows. In contrast, HIV-1 envelope antigen-specific ADCC responses were first observed in 11 subjects at visit 1 when all 39 were NA and ELISA negative, in 12 subjects at visit 2, in 13 subjects at visit 3, and 1 subject at visit 4. Early ADCC responses were associated with high mean % CD4+ cell numbers and absence of lymphadenopathy throughout the 2-year observation period. Not all subjects who developed ADCC developed NA. In some subjects, ADCC and NA were detectable for the first time at the same visit, for others ADCC was detectable prior to NA, and for a few NA was detectable prior to ADCC. These findings suggest that ADCC and neutralization are mediated by different antibody populations, that they may partially inhibit the progress of HIV-1 infection, and that the late appearance of NA may relate to the failure of immunity to effect recovery from this infection.
Acquired Immunodeficiency Syndrome↗
Statistical methods for monitoring the AIDS epidemic.
This article describes statistical methods for monitoring the epidemic of the acquired immunodeficiency syndrome (AIDS). A log-linear model is proposed to estimate AIDS incidence and its growth rate while taking account of delays in case reporting. An empirical Bayes approach for estimating the epidemic growth rate in low prevalence subgroups is introduced. These methods are illustrated with an analysis of AIDS incidence trends for seven risk groups in each of six geographic regions using the Centers for Disease Control AIDS case registry data as of September 1987. The analysis finds that AIDS incidence is currently doubling about once every two years and that the relative composition of new cases is shifting away from the older epidemics such as in north-eastern homosexual communities.
Acquired Immunodeficiency Syndrome↗
Hemodilution causes size-dependent constriction of pial arterioles in the cat.
Cerebral blood flow (CBF) rises as hematocrit (Hct) falls. We previously attributed this rise in CBF to two independent factors of equal importance, decreased arterial O2 content and decreased blood viscosity. We hypothesized that decreased arterial O2 content would dilate cerebral arterioles and that the magnitude of the vasodilation would depend on the magnitude of the passive fall in vascular resistance attributable to decreased viscosity. The present study was designed to test the hypothesis that anemia is accompanied by cerebral vasodilation. Using a closed cranial window, we measured the diameters of 42 pial arterioles (35-305 microns) in 7 cats as serial isovolemic hemodilution lowered Hct by 44% from 31 +/- 4 to 17 +/- 3%. Hemodilution increased CBF (microsphere technique) but did not change mean arterial blood pressure or arterial blood gases. Anticipated vasodilation did not occur; instead, pial arterioles constricted as Hct fell. Maximum vasoconstriction was observed when Hct reached 65-70% of the initial value. Vasoconstriction lessened as Hct was lowered further, but arteriolar diameters at the lowest Hcts remained less than base-line levels. Constriction was greater in small (less than 100 microns) than in large (greater than or equal to 100 microns) arterioles. The initial constriction of pial arterioles may represent myogenic vasoconstriction in response to flow-induced vasodilation of more proximal portions of the cerebrovascular bed and/or to washout of an endogenous vasodilator. Arteriolar relaxation with more profound hemodilution may reflect superimposed metabolic vasodilation.
Animals↗
Village and household clustering of xerophthalmia and trachoma.
The clustering of xerophthalmia in villages and in households was assessed among preschool children surveyed in the Lower Shire Valley, Malawi, and in Aceh province, Indonesia. Trachoma clustering was similarly assessed among the same children in Malawi. Trachoma clustered much more than xerophthalmia among villages and among households. The impact of xerophthalmia clustering on sample size considerations for future surveys or interventions was similar in Malawi and Indonesia. Village clustering of xerophthalmia would necessitate a twofold increase in sample size. Household clustering in the absence of village clustering would have almost no impact on sample size. Village clustering of trachoma would necessitate a ninefold increase in sample size. Household clustering would increase sample size requirements by 26%.
Child, Preschool↗
Twenty-four-hour ambulatory blood pressure in normotensive adolescent children of hypertensive and normotensive parents.
Twenty-four-hour ambulatory blood pressure measurements were obtained during usual daily activities from two groups of normotensive adolescents: Cases had at least one hypertensive parent, and controls had both parents normotensive. Automatic measurements were programmed at 7.5-minute intervals. For each subject, within each of three time periods, the data were summarized by descriptors of the frequency distribution and the frequency spectrum. The time periods were sleep, school hours, and other waking hours. Regression analyses examined the relation between the blood pressure monitoring descriptors and case-control status adjusted for non-blood-pressure variables found to relate to case-control status. In stepwise logistic regression analysis with case-control status as the dependent variable, it was found that the mean level of diastolic pressure during school hours was significant, whereas the standard resting measurement was not. In linear regression analyses with blood pressure monitoring descriptors and standard resting measurements as dependent variables, several monitoring descriptors were found to be significantly related to the interaction between case-control status and certain variables that can affect blood pressure, such as weight and the presumed stress of school; no such relations were found for the standard resting measurements. These findings demonstrate the superiority of the ambulatory monitoring variables over the standard resting measurements in studying case-control differences.
Adolescent↗
Markov regression models for time series: a quasi-likelihood approach.
This paper discusses a quasi-likelihood (QL) approach to regression analysis with time series data. We consider a class of Markov models, referred to by Cox (1981, Scandinavian Journal of Statistics 8, 93-115) as "observation-driven" models in which the conditional means and variances given the past are explicit functions of past outcomes. The class includes autoregressive and Markov chain models for continuous and categorical observations as well as models for counts (e.g., Poisson) and continuous outcomes with constant coefficient of variation (e.g., gamma). We focus on Poisson and gamma data for illustration. Analogous to QL for independent observations, large-sample properties of the regression coefficients depend only on correct specification of the first conditional moment.
Action Potentials↗
Models for longitudinal data: a generalized estimating equation approach.
This article discusses extensions of generalized linear models for the analysis of longitudinal data. Two approaches are considered: subject-specific (SS) models in which heterogeneity in regression parameters is explicitly modelled; and population-averaged (PA) models in which the aggregate response for the population is the focus. We use a generalized estimating equation approach to fit both classes of models for discrete and continuous outcomes. When the subject-specific parameters are assumed to follow a Gaussian distribution, simple relationships between the PA and SS parameters are available. The methods are illustrated with an analysis of data on mother's smoking and children's respiratory disease.
Child↗
On the use of concordant pairs in matched case-control studies.
A new estimator of the common odds ratio in one-to-one matched case-control studies is proposed. The connection between this estimator and the James-Stein estimating procedure is highlighted through the argument of estimating functions. Comparisons are made between this estimator, the conditional maximum likelihood estimator, and the estimator ignoring the matching in terms of finite sample bias, mean squared error, coverage probability, and length of confidence interval. In many situations, the new estimator is found to be more efficient than the conditional maximum likelihood estimator without being as biased as the estimator that ignores matching. The extension to multiple risk factors is also outlined.
Biometry↗