PubMed Health⌕ Search

Biomedical subjects

Sander Greenland

Publications and source records attributed to Sander Greenland.

At least 37 records · Page 2Linked to original sources

Socioeconomic status and childhood leukaemia: a review.

BACKGROUND: A long-held view links higher socioeconomic status (SES) to higher rates of childhood leukaemia. Some recent studies exhibit associations in the opposite direction. METHODS: We reviewed journal literature through August 2002 for associations between childhood leukaemia and socioeconomic measures. We determined the direction of each association and its P-value. We described the results with regard to study design, calendar period, geographic locale, and level of the socioeconomic measures (individual or ecological). For measures with sufficient number of results, we computed summary P-values across studies. RESULTS: Case-control studies conducted in North America since 1980 have involved subject interviews or self-administered questionnaires and have consistently reported inverse (negative) associations of childhood leukaemia with individual-level measures of family income, mother's education, and father's education. In contrast, associations have been consistently positive with father's occupational class in record-based case-control studies and with average occupational class in ecological studies. CONCLUSIONS: Connections of SES measures to childhood leukaemia are likely to vary with place and time. Validation studies are needed to estimate SES-related selection and participation in case-control studies. Because different socioeconomic measures (such as income and education) and individual-level and ecological-level measures may represent different risk factors, we advise researchers to report these measures separately rather than in summary indices of social class.

Adolescent↗

A method to automate probabilistic sensitivity analyses of misclassified binary variables.

BACKGROUND: Misclassification bias is present in most studies, yet uncertainty about its magnitude or direction is rarely quantified. METHODS: The authors present a method for probabilistic sensitivity analysis to quantify likely effects of misclassification of a dichotomous outcome, exposure or covariate. This method involves reconstructing the data that would have been observed had the misclassified variable been correctly classified, given the sensitivity and specificity of classification. The accompanying SAS macro implements the method and allows users to specify ranges of sensitivity and specificity of misclassification parameters to yield simulation intervals that incorporate both systematic and random error. RESULTS: The authors illustrate the method and the accompanying SAS macro code by applying it to a study of the relation between occupational resin exposure and lung-cancer deaths. The authors compare the results using this method with the conventional result, which accounts for random error only, and with the original sensitivity analysis results. CONCLUSION: By accounting for plausible degrees of misclassification, investigators can present study results in a way that incorporates uncertainty about the bias due to misclassification, and so avoid misleadingly precise-looking results.

Bias↗

Improved estimation of controlled direct effects in the presence of unmeasured confounding of intermediate variables.

Adjusting for a causal intermediate is a common analytic strategy for estimating an average causal direct effect (ACDE). The ACDE is the component of the total exposure effect that is not relayed through the specified intermediate. Even if the total effect is unconfounded, the usual ACDE estimate may be biased when an unmeasured variable affects the intermediate and outcome variables. Using linear programming optimization to compute non-parametric bounds, we develop new ACDE estimators for binary measured variables in this causal structure, and use root mean square confounding bias (RMSB) to compare their performance with the usual stratified estimator in simulated distributions of target populations comprised of the 64 possible potential response types as well as distributions of target populations restricted to subsets of 18 or 12 potential response types defined by monotonicity or no-interactions assumptions of unit-level causal effects. We also consider target population distributions conditioned on fixed outcome risk among the unexposed, or fixed true ACDE in one stratum of the intermediate. Results show that a midpoint estimator constructed from the optimization bounds has consistently lower RMSB than the usual stratified estimator both unconditionally and conditioned on any risk in the unexposed. When conditioning on true ACDE, this midpoint estimator performs more poorly only when conditioned on an extreme true ACDE in one stratum of the intermediate, yet outperforms the stratified estimator in the other stratum when interaction is permitted. An alternate 'limit-modified crude' estimator can never perform less favourably than the stratified estimator, and often has lower RMSB.

Bias↗

Revisiting mortality predictability of serum albumin in the dialysis population: time dependency, longitudinal changes and population-attributable fraction.

BACKGROUND: Hypoalbuminaemia is a marker of malnutrition-inflammation complex syndrome (MICS) and a strong predictor of cardiovascular (CV) death in maintenance haemodialysis (MHD) patients. However, serum albumin may change over time. Hence, its time-varying associations with outcome may be different. METHODS: Associations between 3-month averaged serum albumin levels, measured in a single laboratory using bromocresol green, and CV mortality were studied longitudinally in a 2-year cohort of 58,058 MHD patients. Mortality predictability of fixed baseline and trimonthly-varying serum albumin concentrations were compared. RESULTS: Hazard ratios (HRs) of CV death strictly increased across decrements of baseline serum albumin, whereas the HR for time-varying serum albumin decrements below 3.8 g/dl did not differ. A drop in serum albumin in the first 6 months was associated with increasing all-cause and CV death risks in the subsequent 18 months, while a rise in serum albumin was a predictor of better survival independent of baseline serum albumin. The multivariate adjusted population-attributable fraction of death due to baseline serum albumin <3.8 g/dl was 19%. CONCLUSIONS: Time-varying hypoalbuminaemia predicts all-cause and CV death differently from fixed measures of serum albumin in MHD patients. An increase in serum albumin over time is associated with better survival independent of baseline serum albumin or other MICS surrogates. If this association is causal, an intervention that could increase serum albumin >3.8 g/dl might reduce the number of MHD deaths in the USA by approximately 10,000 annually. Nutritional interventions examining benefits of increasing serum albumin in MHD patients are urgently needed.

Aged↗

Epidemiologic measures and policy formulation: lessons from potential outcomes.

This paper provides a critique of the common practice in the health-policy literature of focusing on hypothetical outcome removal at the expense of intervention analysis. The paper begins with an introduction to measures of causal effects within the potential-outcomes framework, focusing on underlying conceptual models, definitions and drawbacks of special relevance to policy formulation based on epidemiologic data. It is argued that, for policy purposes, one should analyze intervention effects within a multivariate-outcome framework to capture the impact of major sources of morbidity and mortality. This framework can clarify what is captured and missed by summary measures of population health, and shows that the concept of summary measure can and should be extended to multidimensional indices.

Journal Article↗

Proper interpretation of non-differential misclassification effects: expectations vs observations.

BACKGROUND: Many investigators write as if non-differential exposure misclassification inevitably leads to a reduction in the strength of an estimated exposure-disease association. Unfortunately, non-differentiality alone is insufficient to guarantee bias towards the null. Furthermore, because bias refers to the average estimate across study repetitions rather than the result of a single study, bias towards the null is insufficient to guarantee that an observed estimate will be an underestimate. Thus, as noted before, exposure misclassification can spuriously increase the observed strength of an association even when the misclassification process is non-differential and the bias it produced is towards the null. METHODS: We present additional results on this topic, including a simulation study of how often an observed relative risk is an overestimate of the true relative risk when the bias is towards the null. RESULTS: The frequency of overestimation depends on many factors: the value of the true relative risk, exposure prevalence, baseline (unexposed) risk, misclassification rates, and other factors that influence bias and random error. CONCLUSIONS: Non-differentiality of exposure misclassification does not justify claims that the observed estimate must be an underestimate; further conditions must hold to get bias towards the null, and even when they do hold the observed estimate may by chance be an overestimate.

Bias↗

Reverse epidemiology of hypertension and cardiovascular death in the hemodialysis population: the 58th annual fall conference and scientific sessions.

Maintenance hemodialysis patients in the United States have a high prevalence (approximately 80%) of systolic hypertension and a high mortality (approximately 20% per year). Some reports indicate a paradoxical association between hypertension and mortality in hemodialysis patients (ie, a normal to low blood pressure is associated with poor outcome), whereas high pressure confers survival advantages, a phenomenon referred to as "reverse epidemiology." We hypothesized that malnutrition-inflammation complex syndrome may be a cause of this paradoxical association. We studied a 15-month cohort of 40 933 hemodialysis patients in the United States whose predialysis and postdialysis blood pressure values were recorded routinely during each hemodialysis treatment. Patients were 59.8+/-15.3 years old; 54% were women and 46% diabetics. Cox proportional hazard models were used for blood pressure categories (systolic <110, > or =190 mm Hg; diastolic <50, > or =110; and increments of 10 mm Hg in between). Unadjusted, case-mix and dialysis dose-adjusted, and additional malnutrition-inflammation-adjusted hazard ratios of all-cause and cardiovascular death showed progressively increasing all-cause and cardiovascular death risk for decreasing blood pressure values. The lowest mortality was associated with predialysis systolic pressure of 160 to 189 mm Hg, whereas normal to low predialysis pressure values were associated with significantly increased mortality. Adjustment for the malnutrition-inflammation mitigated only a small portion of paradoxical associations between the low blood pressure and mortality. Predialysis systolic hypertension remained a significant predictor of highest all-cause and cardiovascular survival rate. Although these associations may not be causal, they call into question whether treatment goals for the general population can be applied to dialysis patients or other similar populations.

Aged↗

Epidemiologic review of marijuana use and cancer risk.

Marijuana is the most commonly used illegal drug in the United States and is considered by young adults to be the illicit drug with the least risk. On the other hand, marijuana smoke contains several of the same carcinogens and co-carcinogens as the tar from tobacco, raising concerns that smoking of marijuana may be a risk factor for tobacco-related cancers. We reviewed two cohort studies and 14 case-control studies with assessment of the association of marijuana use and cancer risk. In the cohort studies, increased risks of lung or colorectal cancer due to marijuana smoking were not observed, but increased risks of prostate and cervical cancers among non-tobacco smokers, as well as adult-onset glioma among tobacco and non-tobacco smokers, were observed. The 14 case-control studies included four studies on head and neck cancers, two studies on lung cancer, two studies on non-Hodgkin's lymphoma, one study on anal cancer, one study on penile cancer, and four studies on childhood cancers with assessment of parental exposures. Zhang and colleagues reported that marijuana use may increase risk of head and neck cancers in a hospital-based case-control study in the United States, with dose-response relations for both frequency and duration of use. However, Rosenblatt and co-workers reported no association between oral cancer and marijuana use in a population-based case-control study. An eightfold increase in risk among marijuana users was observed in a lung cancer study in Tunisia. However, there was no assessment of the dose response, and marijuana may have been mixed with tobacco. Parental marijuana use during gestation was associated with increased risks of childhood leukemia, astrocytoma, and rhabdomyosarcoma, but dose-response relations were not assessed. In summary, sufficient studies are not available to adequately evaluate marijuana impact on cancer risk. Several limitations of previous studies include possible underreporting where marijuana use is illegal, small sample sizes, and too few heavy marijuana users in the study sample. Recommendations for future studies are to (1) focus on tobacco-related cancer sites; (2) obtain detailed marijuana exposure assessment, including frequency, duration, and amount of personal use as well as mode of use (smoked in a cigarette, pipe, or bong; taken orally); (3) adjust for tobacco smoking and conduct analyses on nonusers of tobacco; and (4) conduct larger studies, meta-analyses, or pooled analyses to maximize statistical precision and investigate sources of differences in results. Despite the challenges, elucidation of the association between marijuana use and cancer risk is important in weighing the benefits and risks of medical marijuana use and to clarify the impact of marijuana use on public health.

Animals↗

Association of morbid obesity and weight change over time with cardiovascular survival in hemodialysis population.

BACKGROUND: In maintenance hemodialysis (MHD) outpatients, a reverse epidemiology is described, ie, baseline obesity appears paradoxically associated with improved survival. However, the association between changes in weight over time and prospective mortality is not known. METHODS: Using time-dependent Cox models and adjusting for changes in laboratory values over time, the relation of quarterly-varying 3-month averaged body mass index (BMI) to all-cause and cardiovascular mortality was examined in a 2-year cohort of 54,535 MHD patients from virtually all DaVita dialysis clinics in the United States. RESULTS: Patients, aged 61.7 +/- 15.5 (SD) years, included 54% men and 45% with diabetes. Time-dependent unadjusted and multivariate-adjusted models, based on quarterly-averaged BMI controlled for case-mix and available time-varying laboratory surrogates of nutritional status, were calculated in 11 categories of BMI. Obesity, including morbid obesity, was associated with better survival and reduced cardiovascular death, even after accounting for changes in BMI and laboratory values over time. Survival advantages of obesity were maintained for dichotomized BMI cutoff values of 25, 30, and 35 kg/m2 across almost all strata of age, race, sex, dialysis dose, protein intake, and serum albumin level. Examining the regression slope of change in weight over time, progressively worsening weight loss was associated with poor survival, whereas weight gain showed a tendency toward decreased cardiovascular death. CONCLUSION: Weight gain and both baseline and time-varying obesity may be associated with reduced cardiovascular mortality in MHD patients independent of laboratory surrogates of nutritional status and their changes over time. Morbidly obese patients have the lowest mortality. Clinical trials need to verify these observational findings.

Adult↗

Causation and causal inference in epidemiology.

Concepts of cause and causal inference are largely self-taught from early learning experiences. A model of causation that describes causes in terms of sufficient causes and their component causes illuminates important principles such as multi-causality, the dependence of the strength of component causes on the prevalence of complementary component causes, and interaction between component causes. Philosophers agree that causal propositions cannot be proved, and find flaws or practical limitations in all philosophies of causal inference. Hence, the role of logic, belief, and observation in evaluating causal propositions is not settled. Causal inference in epidemiology is better viewed as an exercise in measurement of an effect rather than as a criterion-guided process for deciding whether an effect is present or not.

Causality↗

Interval estimation by simulation as an alternative to and extension of confidence intervals.

There are numerous techniques for constructing confidence intervals, most of which are unavailable in standard software. Modern computing power allows one to replace these techniques with relatively simple, general simulation methods. These methods extend easily to incorporate sources of uncertainty beyond random error. The simulation concepts are explained in an example of estimating a population attributable fraction, a problem for which analytical formulas can be quite unwieldy. First, simulation of conventional intervals is illustrated and compared to bootstrapping. The simulation is then extended to include sampling of bias parameters from prior distributions. It is argued that the use of almost any neutral or survey-based prior that allows non-zero values for bias parameters will produce an interval estimate less misleading than a conventional confidence interval. Along with simplicity and generality, the ease with which simulation can incorporate these priors is a key advantage over conventional methods.

Computer Simulation↗

Monte Carlo sensitivity analysis and Bayesian analysis of smoking as an unmeasured confounder in a study of silica and lung cancer.

Conventional confidence intervals reflect uncertainty due to random error but omit uncertainty due to biases, such as confounding, selection bias, and measurement error. Such uncertainty can be quantified, especially if the investigator has some idea of the amount of such bias. A traditional sensitivity analysis produces one or more point estimates for the exposure effect hypothetically adjusted for bias, but it does not provide a range of effect measures given the likely range of bias. Here the authors used Monte Carlo sensitivity analysis and Bayesian bias analysis to provide such a range, using data from a US silica-lung cancer study in which results were potentially confounded by smoking. After positing a distribution for the smoking habits of workers and referents, a distribution of rate ratios for the effect of smoking on lung cancer, and a model for the bias parameter, the authors derived a distribution for the silica-lung cancer rate ratios hypothetically adjusted for smoking. The original standardized mortality ratio for the silica-lung cancer relation was 1.60 (95% confidence interval: 1.31, 1.93). Monte Carlo sensitivity analysis, adjusting for possible confounding by smoking, led to an adjusted standardized mortality ratio of 1.43 (95% Monte Carlo limits: 1.15, 1.78). Bayesian results were similar (95% posterior limits: 1.13, 1.84). The authors believe that these types of analyses, which make explicit and quantify sources of uncertainty, should be more widely adopted by epidemiologists.

Adult↗

Model-based estimation of relative risks and other epidemiologic measures in studies of common outcomes and in case-control studies.

Some recent articles have discussed biased methods for estimating risk ratios from adjusted odds ratios when the outcome is common, and the problem of setting confidence limits for risk ratios. These articles have overlooked the extensive literature on valid estimation of risks, risk ratios, and risk differences from logistic and other models, including methods that remain valid when the outcome is common, and methods for risk and rate estimation from case-control studies. The present article describes how most of these methods can be subsumed under a general formulation that also encompasses traditional standardization methods and methods for projecting the impact of partially successful interventions. Approximate variance formulas for the resulting estimates allow interval estimation; these intervals can be closely approximated by rapid simulation procedures that require only standard software functions.

Case-Control Studies↗

Associations of race/ethnicity with HIV prevalence and HIV-related behaviors among young men who have sex with men in 7 urban centers in the United States.

Using data from a multisite venue-based survey of male subjects aged 15 to 22 years, we examined racial/ethnic differences in demographics, partner type, partner type-specific condom use, drug use, and HIV prevalence in 3316 US black, multiethnic black, Latino, and white men who have sex with men (MSM). We further estimated associations of these factors with HIV infection and their influence on racial/ethnic disparities in HIV prevalence. HIV prevalences were 16% for both black and multiethnic black participants, 6.9% for Latinos, and 3.3% for whites. Paradoxically, potentially risky sex and drug-using behaviors were generally reported most frequently by whites and least frequently by blacks. In a multiple logistic regression analysis, positive associations with HIV included older age, being out of school or work, sex while on crack cocaine, and anal sex with another male regardless of reported condom use level. Differences in these factors did not explain the racial/ethnic disparities in HIV prevalence, with both groups of blacks experiencing more than 9 times and Latinos experiencing approximately twice the fully adjusted odds of infection compared with whites. Understanding racial/ethnic disparities in HIV risk requires information beyond the traditional risk behavior and partnership type distinctions. Prevention programs should address risks in steady partnerships, target young men before sexual initiation with male partners, and tailor interventions to men of color and of lower socioeconomic status.

Adolescent↗

The value of risk-factor ("black-box") epidemiology.

Risk-factor epidemiology has been denigrated by some as an empty search for associations, unguided by underlying theory. It has been defended for occasionally identifying useful (if poorly understood) potential interventions. We further defend risk-factor epidemiology as a valuable source of seemingly unrelated facts that await coherent explanation by novel theories and that provide empiric tests of theories. We illustrate these points with a theory that invokes lipid peroxidation as an explanation of an apparently incoherent accumulation of facts about renal-cell carcinoma. The example illustrates the value of viewing epidemiologic, laboratory, and clinical observations as a body of facts demanding explanation by proposed causal theories, whether or not those observations were collected with any hypothesis in mind.

Carcinoma, Renal Cell↗

Bounding analysis as an inadequately specified methodology.

The bounding analysis methodology described by Ha-Duong et al. (this issue) is logically incomplete and invites serious misuse and misinterpretation, as their own example and interpretation illustrate. A key issue is the extent to which these problems are inherent in their methodology, and resolvable by a logically complete assessment (such as Monte Carlo or Bayesian risk assessment), as opposed to being general problems in any risk-assessment methodology. I here attempt to apportion the problems between those inherent in the proposed bounding analysis and those that are more general, such as reliance on questionable expert elicitations. I conclude that the specific methodology of Ha-Duong et al. suffers from logical gaps in the definition and construction of inputs, and hence should not be used in the form proposed. Furthermore, the labor required to do a sound bounding analysis is great enough so that one may as well skip that analysis and carry out a more logically complete probabilistic analysis, one that will better inform the consumer of the appropriate level uncertainty. If analysts insist on carrying out a bounding analysis in place of more thorough assessments, extensive analyses of sensitivity to inputs and assumptions will be essential to display uncertainties, arguably more essential than it would be in full probabilistic analyses.

Comment↗