George Maldonado

Determinants of primary care specialty choice: a non-statistical meta-analysis of the literature

This paper analyzes and synthesizes the literature on primary care specialty choice from 1987 through 1993. To improve the validity and usefulness of the conclusions drawn from the literature, the authors developed a model of medical student specialty choice to guide the synthesis, and used only high-quality research (a final total of 73 articles). They found that students predominantly enter medical school with a preference for primary care careers, but that this preference diminishes over time (particularly over the clinical clerkship years). Student characteristics associated with primary care career choice are: being female, older, and married; having a broad undergraduate background; having non-physician parents; having relatively low income expectations; being interested in diverse patients and health problems; and having less interest in prestige, high technology, and surgery. Other traits, such as value orientation, personality, or life situation, yet to be reliably measured, may actually be responsible for some of these associations. Two curricular experiences are associated with increases in the numbers of students choosing primary care: required family practice clerkships and longitudinal primary care experiences. Overall, the number of required weeks in family practice shows the strongest association. Students are influenced by the cultures of the institutions in which they train, and an important factor in this influence is the relative representation of academically credible, full-time primary care faculty within each institutions governance and everyday operation. In turn, the institutional culture and faculty composition are largely determined by each schools mission and funding sources–explaining, perhaps, the strong and consistent association frequently found between public schools and a greater output of primary care physicians. Factors that do not influence primary care specialty choice include early exposure to family practice faculty or to family practitioners in their own clinics, having a high family medicine faculty-to-student ratio, and student debt level, unless exceptionally high. Also, students view a lack of understanding of the specialties as a major impediment to their career decisions, and it appears they acquire distorted images of the primary care specialties as they learn within major academic settings. Strikingly few schools produce a majority of primary care graduates who enter family practice, general internal medicine, or general practice residencies or who actually practice as generalists. Even specially designed tracks seldom produce more than 60% primary care graduates. Twelve recommendations for strategies to increase the proportion of primary care physicians are provided.

Can landscape ecology untangle the complexity of antibiotic resistance

Bacterial resistance to antibiotics continues to pose a serious threat to human and animal health. Given the considerable spatial and temporal heterogeneity in the distribution of resistance and the factors that affect its evolution, dissemination and persistence, we argue that antibiotic resistance must be viewed as an ecological problem. A fundamental difficulty in assessing the causal relationship between antibiotic use and resistance is the confounding influence of geography: the co-localization of resistant bacterial species with antibiotic use does not necessarily imply causation and could represent the presence of environmental conditions and factors that have independently contributed to the occurrence of resistance. Here, we show how landscape ecology, which links the biotic and abiotic factors of an ecosystem, might help to untangle the complexity of antibiotic resistance and improve the interpretation of ecological studies.

How far from non-differential does exposure or disease misclassification have to be to bias measures of association away from the null?

A well-known heuristic in epidemiology is that non-differential exposure or disease misclassification biases the expected values of an estimator toward the null value. This heuristic works correctly only when additional conditions are met, such as independence of classification errors. We present examples to show that, even when the additional conditions are met, if the misclassification is only approximately non-differential, then bias is not guaranteed to be toward the null. In light of such examples, we advise that evaluation of misclassification should not be based on the assumption of exact non-differentiality unless the latter can be deduced logically from the facts of the situation.

Good practices for quantitative bias analysis

Quantitative bias analysis serves several objectives in epidemiological research. First, it provides a quantitative estimate of the direction, magnitude and uncertainty arising from systematic errors. Second, the acts of identifying sources of systematic error, writing down models to quantify them, assigning values to the bias parameters and interpreting the results combat the human tendency towards overconfidence in research results, syntheses and critiques and the inferences that rest upon them. Finally, by suggesting aspects that dominate uncertainty in a particular research result or topic area, bias analysis can guide efficient allocation of sparse research resources. The fundamental methods of bias analyses have been known for decades, and there have been calls for more widespread use for nearly as long. There was a time when some believed that bias analyses were rarely undertaken because the methods were not widely known and because automated computing tools were not readily available to implement the methods. These shortcomings have been largely resolved. We must, therefore, contemplate other barriers to implementation. One possibility is that practitioners avoid the analyses because they lack confidence in the practice of bias analysis. The purpose of this paper is therefore to describe what we view as good practices for applying quantitative bias analysis to epidemiological data, directed towards those familiar with the methods. We focus on answering questions often posed to those of us who advocate incorporation of bias analysis methods into teaching and research. These include the following. When is bias analysis practical and productive? How does one select the biases that ought to be addressed? How does one select a method to model biases? How does one assign values to the parameters of a bias model? How does one present and interpret a bias analysis?. We hope that our guide to good practices for conducting and presenting bias analyses will encourage more widespread use of bias analysis to estimate the potential magnitude and direction of biases, as well as the uncertainty in estimates potentially influenced by the biases.

A systematic approach to conducting a non-statistical meta-analysis of research literature

Literature analyses and syntheses are becoming increasingly important as a means of periodically bringing coherence to a research area, contributing new knowledge revealed by integrating single studies, and quickly informing scientists of the state of the field. As a result, there is a need for approaches that can provide replicable, reliable, and trustworthy results. Over the last decade many researchers have begun using the statistical meta-analysis approach to integrate studies. However, the single studies conducted in many areas are not of the type amenable to statistical meta-analysis but are more appropriate for non-statistical analysis and synthesis. The present paper describes (1) a rigorous approach to conducting a non-statistical meta-analysis of research literature and (2) an example of how this approach was applied to the literature of determinants of primary care specialty choice published between 1987 and 1993. This approach includes model development, literature retrieval, literature coding, rating references for quality, annotating high-quality references, and synthesizing only the subset of the literature found of sufficient quality to be considered. Also, the basic results of each included study are reported in the synthesis so that readers have before them all the “data points” used in the synthesis. Thus, readers can draw their own interpretations without having to re-collect the data, just as they would be able to do in any single study that presents original data as well as conclusions and discussion.

Injury from dairy cattle activities.

Animals have been implicated as an important source of injury for farm household members. Little is known, however, about the specific activities associated with the animal/livestock operations that place a person at increased or decreased risk for injuries. The primary aim of this case‐control study was to identify which dairy cattle operation activities (that is, milking, feeding, cleaning barns, trimming and treating feet, dehorning, assisting with difficult calvings, and doing treatments) were associated with an increased or decreased risk of injury. We found milking to have the greatest increase in risk for injury. The ratios for increasing hours per week spent at milking (0, 1–10, 11–20, 21–30, 31–63) were 1.0, 2.3, 5.5, 10.9, and 20.6, respectively. We also found an increased rate ratio associated with trimming or treating hooves (rate ratio = 4.2).

Exposure-measurement error is frequently ignored when interpreting epidemiologic study results

IntroductionOne important source of error in study results is error in measuring exposures. When interpreting study results, one should consider the impact that exposure-measurement error (EME) might have had on study results.MethodsTo assess how often this consideration is made and the form it takes, journal articles were randomly sampled from original articles appearing in the American Journal of Epidemiology and Epidemiology in 2001, and the International Journal of Epidemiology between December 2000 and October 2001.ResultsTwenty-two (39%) of the 57 articles surveyed mentioned nothing about EME. Of the 35 articles that mentioned something about EME, 16 articles described qualitatively the effect EME could have had on study results. Only one study quantified the impact of EME on study results; the investigators used a sensitivity analysis. Few authors discussed the measurement error in their study in any detail.ConclusionsOverall, the potential impact of EME on error in epidemiologic study results appears to be ignored frequently in practice.

An updated study of mortality among North American synthetic rubber industry workers

Aim: This study evaluated the mortality experience of workers from the styrene-butadiene industry. Methods: The authors added seven years of follow up to a previous investigation of mortality among 17 924 men employed in the North American synthetic rubber industry. Analyses used the standardised mortality ratios (SMRs) to compare styrene-butadiene rubber workers’ cause specific mortality (1943–98) with those of the United States and the Ontario general populations. Results: Overall, the observed/expected numbers of deaths were 6237/7242 for all causes (SMR = 86, 95% CI 84 to 88) and 1608/1741 for all cancers combined (SMR = 92, 95% CI 88 to 97), 71/61 for leukaemia, 53/53 for non-Hodgkin’s lymphoma, and 26/27 for multiple myeloma. The 16% leukaemia increase was concentrated in hourly paid subjects with 20–29 years since hire and 10 or more years of employment in the industry (19/7.4, SMR = 258, 95% CI 156 to 403) and in subjects employed in polymerisation (18/8.8, SMR = 204, 95% CI 121 to 322), maintenance labour (15/7.4, SMR = 326, 95% CI 178 to 456), and laboratory operations (14/4.3, SMR = 326, 95% CI 178–546). Conclusion: The study found that some subgroups of synthetic rubber workers had an excess of mortality from leukaemia that was not limited to a particular form of leukaemia. Uncertainty remains about the specific agent(s) that might be responsible for the observed excesses and about the role of unidentified confounding factors. The study did not find any clear relation between employment in the industry and other forms of lymphohaematopoietic cancer. Some subgroups of subjects had more than expected deaths from colorectal and prostate cancers. These increases did not appear to be related to occupational exposure in the industry.

Interpreting Model Coefficients When the True Model Form Is Unknown

In this paper, we critically examine mathematical modeling. We outline the major assumptions required by modeling methods used in epidemiology and discuss in detail one fundamental assumption that is usually violated in epidemiologic studies: the assumption that the structural model form is correctly specified. We apply concepts from the econometrics literature to examine how epidemiologic inference may be affected when the structural model form is incorrectly specified. Because the structural model is almost always misspecified in practice, tests and confidence intervals for model coefficients do not refer to “true population parameters“ in the ordinary sense. Rather, these statistics concern parameters that depend on features of study design, as well as the effects under study. In cohort studies analyzed with multiplicative rate models, model parameters are interpretable as approximations to log standardized rate ratios; unfortunately, such interpretations are not as accurate for other models and designs. We therefore conclude that model coefficients can serve as reasonable effect summaries in some, but not all, situations. (Epidemiology 1993;4:310–318)

Chemical exposures in the synthetic rubber industry and lymphohematopoietic cancer mortality.

Objective: This study evaluated the association between exposure to several chemicals and mortality from lymphohematopoietic cancer (LHC) among 16,579 synthetic rubber industry workers who were followed up from 1943 to 1998. Methods: Poisson regression analyses examined LHC rates in relation to butadiene, styrene, and DMDTC exposure. Models provided maximum likelihood estimates of the relative rate for the contrast between categories of one agent, adjusting for other agents and for additional potential confounders. Results: Cumulative exposure to 1,3-butadiene was associated positively with all leukemia (relative rates of 1.0, 1.4, 1.2, 2.9, and 3.7, respectively, for exposures of 0, >0 to <33.7, 33.7 to <184.7, 184.7 to <425.0, and 425.0+ ppm-years), chronic myelogenous leukemia and to a lesser extent with chronic lymphocytic leukemia. Adjusting for styrene and DMDTC attenuated these associations. After controlling for butadiene, neither styrene nor DMDTC displayed a consistent exposure–response trend with all leukemia, chronic myelogenous leukemia, or chronic lymphocytic leukemia. Conclusions: This study found a positive association between butadiene and leukemia that was not explained by exposure to other agents examined.