David G. Kleinbaum

Emory University | EU · Department of Epidemiology

Objective To compare risk of surgical site infection (SSI) following cesarean delivery between women covered by Medicaid and private health insurance. Study design Retrospective cohort. Study population Cesarean deliveries covered by Medicaid or private insurance and reported to the National Healthcare Safety Network (NHSN) and state inpatient di...

Background: Initiation of complementary feeding is often delayed in Bangladesh and likely contributes to the high burden of infant undernutrition in the country. Methods: Pregnant women at 28–32 weeks’ gestation were recruited for a cohort-based evaluation of a community-based nutrition education program. To identify predictors of the timing of int...

Episodes of adenolymphangitis (ADL) are a recurrent clinical aspect of lymphatic filariasis (LF) and a risk factor for progression of lymphedema. Inter-digital entry lesions, often found on the web spaces between the toes of those suffering from lymphedema, have been shown to contribute to the occurrence of ADL episodes. Use of antifungal cream on...

Episodes of adenolymphangitis (ADL) are a recurrent clinical aspect of lymphatic filariasis (LF) and a risk factor for progression of lymphedema. Inter-digital entry lesions, often found on the web spaces between the toes of those suffering from lymphedema, have been shown to contribute to the occurrence of ADL episodes. Use of antifungal cream on...

Background: Childhood undernutrition is a major public health problem in Bangladesh. Evaluating child nutrition programs is a priority. Objective: The objective of this study was to evaluate a community-based nutrition education program (implemented from 2011 to 2013) aimed at improving infant and young child feeding (IYCF) practices and growth in...

The association between suboptimal infant feeding practices and growth faltering is well-established. However, most of this evidence comes from cross-sectional studies. To prospectively assess the association between suboptimal infant feeding practices and growth faltering, we interviewed pregnant women at 28–32 weeks’ gestation and followed-up the...

Dataset. (ZIP)

Among dialysis facilities participating in a bloodstream infection (BSI) prevention collaborative, access-related BSI incidence rate improvements observed immediately following implementation of a bundle of BSI prevention interventions were sustained for up to 4 years. Overall, BSI incidence remained unchanged from baseline in the current analysis....

Objective To determine the association between household food security and infant complementary feeding practices in rural Bangladesh. Design Prospective, cohort study using structured home interviews during pregnancy and 3 and 9 months after delivery. We used two indicators of household food security at 3-months’ follow-up: maternal Food Composit...

Episodes of acute adenolymphangitis (ADL) are often the first clinical sign of lymphatic filariasis (LF). They are often accompanied by swelling of the affected limb, inflammation, fever, and general malaise and lead to the progression of lymphedema. Although ADL episodes have been studied for a century or more, questions still remain as to their e...

Inappropriate complementary feeding (CF) practices are common in Bangladesh. To identify predictors of the timing of introduction of solid/semi‐solid/soft foods (CF initiation) in rural Bangladesh, we prospectively interviewed 2073 pregnant women at 28 weeks' pregnancy and at postnatal months 3 and 9. We assessed CF knowledge and attitudes at the 3...

The use of contaminated surface water continues to be a pressing issue in areas of the world where people lack improved drinking water sources. In northern coastal Ecuador, many communities rely on untreated surface water as their primary source of drinking water. We undertook a study to explore how microscale river hydrodynamics affect microbial w...

The use of contaminated surface water continues to be a pressing issue in areas of the world where people lack improved drinking water sources. In northern coastal Ecuador, many communities rely on untreated surface water as their primary source of drinking water. We undertook a study to explore how microscale river hydrodynamics affect microbial w...

Background: Men who have sex with men (MSM) are the most affected risk group in the United States' human immunodeficiency virus (HIV) epidemic. Sexual concurrency, the overlapping of partnerships in time, accelerates HIV transmission in populations and has been documented at high levels among MSM. However, concurrency is challenging to measure emp...

Lymphedema management programs have been shown to decrease episodes of adenolymphangitis (ADLA), but the impact on lymphedema progression and of program compliance have not been thoroughly explored. Our objectives were to determine the rate of ADLA episodes and lymphedema progression over time for patients enrolled in a community-based lymphedema m...

To identify predictors of infant complementary feeding (CF) practices in rural Bangladesh, we studied pregnant women and their infants followed‐up at age 3 and 9 mo. Information on sociodemographics, maternal diet, and infant diet and feeding practices were collected by interview. Maternal food composition score (FCS) at infant age 3 mo, calculated...

Background: Receipt of nephrology care prior to end-stage renal disease (ESRD) is a strong predictor of decreased mortality and morbidity, and neighborhood poverty may influence access to care. Our objective was to examine whether neighborhood poverty is associated with lack of pre-ESRD care at dialysis facilities. Methods: In a multi-level ecol...

In epidemiologic studies, we use a measure of disease frequency to determine how often the disease or other health outcome of interest occurs in various subgroups of interest. We describe two basic types of measures of disease frequency in this chapter, namely, measures of incidence and measures of prevalence. The choice typically depends on the st...

We hope that you will find the ActivEpi resources a valuable reference. This page on the ActivEpi CD ROM is designed to make it easy for you to find and use the reference materials in ActivEpi. In addition to the icons on this page, you may want to take advantage of some special web services found by clicking the WEB icon in the control bar.

Both stratum-specific analyses and overall assessment require a point estimate, an interval estimate, and a test of hypothesis. In this lesson, we focus on overall assessment, which is the most conceptually and mathematically complicated of the four steps. For overall assessment, the point estimate is an adjusted estimate that is typically in the f...

Selection bias concerns systematic error that may arise from the manner in which subjects are selected into one’s study. In his lesson, we describe examples of selection bias, provide a quantitative framework for assessing selection bias, show how selection bias can occur in different types of epidemiologic study designs, and discuss how to adjust...

In the previous lesson on Measures of Effect, we focused exclusively on ratio measures of effect. In this lesson, we consider difference measures of effect and other related measures that allow the investigator to consider the potential public health impact of the results obtained from an epidemiologic study.

The primary objective of most epidemiologic research is to obtain a valid estimate of an effect measure of interest. In this activity, we illustrate three general types of validity problems, distinguish validity from precision, introduce the term bias, and discuss how to adjust for bias.

Matching is an option for control that is available at the study design stage. We previously introduced matching on page 13-2 in Lesson 13. We suggest that you review that activity before proceeding further with this lesson. The primary goal of matching is to gain precision in estimating the measure of effect of interest. There are other advantages...

Information bias is a systematic error in a study that arises because of incorrect information obtained on one or more variables measured in the study. The focus here is on the consequences of having inaccurate information about exposure and disease variables that are dichotomous, that is, when there is misclassification of exposure and disease tha...

Confounding is a form of bias that concerns how a measure of association may change in value depending on whether variables other than the exposure variable are controlled in the analysis.

Epidemiology is the study of health and illness in human or other (veterinary) populations. In this course, we consider real-world health and illness problems, and we show how epidemiologic concepts and methods allow us to study, understand and solve such problems. And, most important, we apply each new concept or method as we develop it to help yo...

Suppose we are studying whether there is a link between exposure to a toxic chemical and the development of lung cancer in a chemical industry. To answer this question properly, we would want to isolate the effect of the chemical from the possible influence of other variables, particularly age and smoking status, two known risk factors for lung can...

A key stage of epidemiologic research is the study design. This is defined to be the process of planning an empirical investigation to assess a conceptual hypothesis about the relationship between one or more exposures and a health outcome. The purpose of the study design is to transform the conceptual hypothesis into an operational hypothesis that...

In epidemiologic studies, we compare disease frequencies of two or more groups using a measure of effect. We will describe several types of measures of effect in this chapter. The choice of measure typically depends on the study design being used.

Background: Men who have sex with men (MSM) represent the largest HIV risk group in the United States. Sexual concurrency catalyzes HIV transmission in populations by increasing the indirect exposure of one's sex partners to one another. Although individual-level (egocentric) designs have demonstrated a high prevalence of concurrency among MSM res...

This lesson considers how the assessment of confounding gets somewhat more complicated when controlling for more than one risk factor. In particular, when several risk factors are being controlled, we may find that considering all risk factors simultaneously may not lead to the same conclusion as when considering risk factors separately. We have pr...

This lesson discusses methods for carrying out statistical inference procedures for epidemiologic data given in a simple two-way table. We call such procedures simple analyses because we are restricting the discussion here to dichotomous disease and exposure variables only and we are ignoring the typical analysis situation that considers the contro...

The field of epidemiology was initially concerned with providing a methodological basis for the study and control of population epidemics. Now, however, epidemiology has a much broader scope, including the study of both acute and chronic diseases, the quality of health care, and mental health problems. As the focus of epidemiologic inquiry has broa...

This Companion Textbook supplements the ActivEpi CD-ROM, sold separately. The ActivEpi CD-ROM provides a multimedia presentation of concepts, commonly taught in an introductory epidemiology course. ActivEpi mixes a full array of media to motivate, explain, visualize and apply epidemiological concepts. Virtually all of the material on the ActivEpi C...

Background The relationship between resource utilization and postoperative length of stay (PLOS) following congenital heart disease surgery is unknown. Methods We performed a retrospective cohort study using data from the Pediatric Health Information Systems database. We included subjects 1month to 1year of age with a PLOS of 1month following elect...

This chapter considers outcome events that may occur more than once over the follow-up time for a given subject. Such events are called “recurrent events.” Modeling this type of data can be carried out using a Cox PH model with the data layout constructed so that each subject has a line of data corresponding to each recurrent event. A variation of...

In the United States, considerable geographic variation in the rates of culture-confirmed Campylobacter infection has been consistently observed among sites participating in the Foodborne Diseases Active Surveillance Network (FoodNet). We used data from the FoodNet Population Surveys and a FoodNet case-control study of sporadic infection to examine...

In this chapter, we consider issues involved when designing studies with time-to-event outcomes. This includes determining the expected number of events to occur over the study period, the total number of subjects to be recruited into the study, the power of the study, the time period over which to accrue the study subjects, the time period over wh...

We begin with a brief review of the purposes of survival analysis, basic notation and terminology, and the basic data layout for the computer.

We begin with a brief review of the characteristics of the Cox proportional hazards (PH) model. We then give an overview of three methods for checking the PH assumption: graphical, goodness-of-fit (GOF), and time-dependent variable approaches.

The Cox model is the most widely used survival model in the health sciences, but it is not the only model available. In this chapter we present a class of survival models, called parametric models, in which the distribution of the outcome (i.e., the time to event) is specified in terms of unknown parameters. Many parametric models are acceleration...

Purpose: Many centers have implemented efforts to decrease length of stay (LOS) following surgery for congenital heart disease in order to lessen hospital-related morbidity and potentially to reduce resource use. We sought to determine the relationship between daily resource utilization and LOS following surgery for congenital heart disease. Meth...

In this chapter, the logistic model is extended to handle outcome variables that have dichotomous correlated responses. The analytic approach presented for modeling this type of data is the generalized estimating equations (GEE) model, which takes into account the correlated nature of the responses. If such correlations are ignored in the modeling...

This chapter continues the previous chapter (Chap. 6) that gives general guidelines for a strategy for determining a best model using a logistic regression procedure. The focus of this chapter is the interaction and confounding assessment stages of the model building strategy.

We begin this chapter by giving the rationale for having a strategy to determine a “best” model. Focus is on a logistic model containing a single dichotomous exposure variable that adjusts for potential confounding and potential interaction effects of covariates considered for control. A strategy is recommended, which has three stages: (1) variable...

This introduction to logistic regression describes the reasons for the popularity of the logistic model, the model form, how the model may be applied, and several of its key features, particularly how an odds ratio can be derived and computed for this model.

In this chapter, the standard logistic model is extended to handle outcome variables that have more than two ordered categories. When the categories of the outcome variable have a natural order, ordinal logistic regression may be appropriate.

We begin our discussion of statistical inference by describing the computer information required for making inferences about the logistic model. We then introduce examples of three logistic models that we use to describe hypothesis testing and confidence interval estimation procedures. We consider models with no interaction terms first, and then we...

In this chapter, we describe and illustrate methods for assessing the extent that a fitted binary logistic model can be used to distinguish the observed cases (Y = 1) from the observed noncases (Y = 0).

Regression diagnostics are techniques for the detection and assessment of potential problems resulting from a fitted regression model that might either support, compromise, or negate the assumptions made about the regression model and/or the conclusions drawn from the analysis of one’s data.

In this chapter, we present examples of GEE models applied to three datasets containing correlated responses. The examples demonstrate how to obtain odds ratios, construct confidence intervals, and perform statistical tests on the regression coefficients. The examples also illustrate the effect of selecting different correlation structures for a GE...

In this chapter, several important special cases of the logistic model involving a single (0, 1) exposure variable are considered with their corresponding odds ratio expressions. In particular, focus is on defining the independent variables that go into the model and on computing the odds ratio for each special case. Models that account for the pot...

In this chapter, we consider five issues on modeling Strategy, which were not covered in the previous two chapters on this topic: 1. Modeling strategy when there are two or more exposure variables 2. Screening variables when modeling 3. Collinearity diagnostics 4. Influential observations 5. Multiple testing Each of these issues represent important...

In this chapter, the discussion of methods to analyze outcome variables that have dichotomous correlated responses is expanded to include approaches other than GEE. Three other analytic approaches are discussed. These include the alternating logistic regressions algorithm, conditional logistic regression, and the generalized linear mixed model appr...

In this chapter, the standard logistic model is extended to handle outcome variables that have more than two categories. Polytomous logistic regression is used when the categories of the outcome variable are nominal, that is, they do not have any natural order. When the categories of the outcome variable do have a natural order, ordinal logistic re...

In the United States, the proportion of foodborne illness outbreaks associated with consumption of contaminated domestic and imported fresh fruits and vegetables (produce) has increased over the past several decades. To address this public health concern, the goal of this work was to identify and quantify factors associated with microbial contamina...

Logistic regression is a popular statistical tool used to analyze epidemiologic data. It is typically used when the outcome or disease/event under study is dichotomous. Logistic regression involves postulating a model to explain the disease process (dependent variable) using exposure and control variables (independent variables), fitting the model...

To investigate community health workers' (CHW) adherence over time to guidelines for treating ill children and to assess the effect of refresher training on adherence. Analysis of 7151 ill-child consultations performed by 114 CHWs in their communities from March 1997-May 2002. Adherence was assessed with a score (percentage of recommended treatment...

Evaluation of a community health worker (CHW) programme in Siaya district, Kenya, showed CHWs commonly made errors in managing childhood illness. We assessed the effect of multiple interventions on CHW healthcare practices. A sample of 192 ill-child consultations performed by 114 CHWs in a hospital outpatient department between February and March 2...

Acinetobacter infections have increased and gained attention because of the organism's prolonged environmental survival and propensity to develop antimicrobial drug resistance. The effect of multidrug-resistant (MDR) Acinetobacter infection on clinical outcomes has not been reported. A retrospective, matched cohort investigation was performed at 2...

A Pocket Guide to Epidemiology is a stand-alone introductory text on the basic principles and concepts of epidemiology. The primary audience for this text is the public health student or professional, clinician, health journalist, and anyone else at any age or life experience that is interested in learning what epidemiology is all about in a conven...

A key stage of epidemiologic research is the study design. This is defined to be the process of planning an empirical investigation to assess a conceptual hypothesis about the relationship between one or more exposures and a health outcome. The purpose of the study design is to transform the conceptual hypothesis into an operational hypothesis that...

The primary objective of most epidemiologic research is to obtain a valid estimate of an effect measure of interest. In this chapter we illustrate three general types of validity problems, distinguish validity from precision, introduce the term bias, and discuss how to adjust for bias.

In the previous chapter on Measures of Effect, we focused exclusively on ratio measures of effect In this chapter, we consider difference measures of effect and other related measures that allow the investigator to consider the potential public health impact of the results obtained from an epidemiologic study.

Selection bias concerns systematic error that may arise from the manner in which subjects are selected into one’s study. In his chapter we describe examples of selection bias, provide a quantitative framework for assessing selection bias, show how selection bias can occur in different types of epidemiologic study designs, and discuss how to adjust...

Epidemiology is the study of health and illness in human populations. For example, a randomized clinical trial conducted by Epidemiologists at the Harvard School of Public Health showed that taking aspirin reduces heart attack risk by 20 to 30 percent. Public health studies in the 1950’s demonstrated that smoking cigarettes causes lung cancer. Envi...

In epidemiologic studies, we compare disease frequencies of two or more groups using a measure of effect. We will describe several types of measures of effect in this chapter. The choice of measure typically depends on the study design being used.

Confounding is a form of bias that concerns how a measure of effect may change in value depending on whether variables other than the exposure variable are controlled in the analysis. Interaction/effect modification, which is different from confounding, compares estimated effects after other variables are controlled.

This is an analysis option for the control of extraneous variables that involves the following steps: 1. Categorize all variables. 2. Form combinations of categories (i.e., strata). 3. Perform stratum-specific analyses. 4. Perform overall E-D assessment if appropriate. Both stratum-specific analyses and overall assessment require a point estimate,...

This chapter discusses methods for carrying out statistical inference procedures for epidemiologic data given in a simple two-way table. We call such procedures simple analyses because we are restricting the discussion here to dichotomous disease and exposure variables only and we are ignoring the typical analysis situation that considers the contr...

The field of epidemiology was initially concerned with providing a methodological basis for the study and control of population epidemics. Now, however, epidemiology has a much broader scope, including the study of both acute and chronic diseases, the quality of health care, and mental health problems. As the focus of epidemiologic inquiry has broa...

Matching is an option for control that is available at the study design stage. We previously introduced matching in Chapter 13. We suggest that you review that chapter before proceeding further with this chapter. The primary goal of matching is to gain precision in estimating the measure of effect of interest. There are other advantages to matching...

This chapter considers how the assessment of confounding gets somewhat more complicated when controlling for more than one risk factor. In particular, when several risk factors are being controlled, we may find that considering all risk factors simultaneously may not lead to the same conclusion as when considering risk factors separately. We have p...

In epidemiologic studies, we use a measure of disease frequency to determine how often the disease or other health outcome of interest occurs in various subgroups of interest. We describe two basic types of measures of disease frequency in this chapter, namely, measures of incidence and measures of prevalence. The choice typically depends on the st...

In previous chapters, we have discussed and illustrated several important concepts concerning the control of additional (extraneous) variables when assessing a relationship between an exposure variable and a health-outcome variable. In this chapter, we briefly review these concepts and then provide an overview of several options for the process of...

Information bias is a systematic error in a study that arises because of incorrect information obtained on one or more variables measured in the study. The focus here is on the consequences of having inaccurate information about exposure and disease variables that are dichotomous, that is, when there is misclassification of exposure and disease tha...

To determine whether results from an evaluation that involved observation of community health workers while they performed patient consultations in a hospital reflected normal everyday practices. Comparison of two samples of ill-child consultations: (i) consultations performed during an evaluation in which we observed community health workers in a...

Most previous studies of the association between psychosocial stress and musculoskeletal illness among computer users have been cross-sectional and have yielded inconsistent results. The association between a measure of psychosocial stress, "job strain", and incident neck-shoulder and arm-hand musculoskeletal symptoms was investigated among recentl...

Depression and obesity have become major health problems with increasing prevalence. Given the limited effectiveness of treatment for weight problems, the identification of novel, potentially modifiable risk factors may provide insights on new preventive approaches to obesity. The purpose of this study was to test the hypothesis that depressive sym...

We consider three sources of bias affecting observational studies and ways to address them: selection bias, information bias, and confounding. We discuss these sources of bias with respect to cohort and case–control designs. Even if a study is internally valid and not subject to these sources of bias, a judgment must be made whether the results gen...

Selection biases can arise in cohort studies if the analysis cohort is not representative of the cohort of interest, either as a result of biased sampling, selective losses to follow-up, or nonresponse. A particular type of biased sampling can arise when one studies a cohort with prevalent rather than incident disease. Misclassification of disease...

Psychologic factors are increasingly recognized to influence the onset and course of asthma. Previous cross-sectional community-based studies have provided evidence for a relatively specific association between asthma and panic. To examine concurrent and longitudinal associations between asthma and panic in young adults. Prospective community-based...

This greatly expanded second edition of Survival Analysis- A Self-learning Text provides a highly readable description of state-of-the-art methods of analysis of survival/event-history data. This text is suitable for researchers and statisticians working in the medical and other life sciences as well as statisticians in academia who teach introduct...