Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Effect Modifier, Epidemiologic”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

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

At least 19 recordsLinked to original sources
PubMed2000

The use of randomized control trials in complementary therapies: exploring the issues.

The current popularity of complementary therapies presents an interesting challenge to nurses and midwives. If they are to deliver such therapies themselves, or support patients in choosing appropriate therapies they will need to consider the professional and legal issues, in particular those regarding safety. Evidence for the effectiveness for complementary therapies is also a requirement in order that their integration into nursing practice can be justified. Purchasers are currently hampered by the lack of credible evidence for effectiveness and until that evidence is provided, access to such therapies through the National Health Service (NHS) will remain limited. The form that evidence should take has led to a lively debate about possible methodological approaches. There appears to be a clash between the medical profession and those working in the field of complementary therapy research, with the medical establishment advocating randomized control trials (RCTs). This contrasts with the view held by some advocates of complementary therapies that the RCT approach is reductionist and not applicable to such approaches. The pivot of the debate on the methodological approaches for evaluating complementary therapies is the contrast of two apparently different and diverse world-views, and the assertion that methods developed in one world-view are not transferable to the other. There is also some confusion within the field of complementary therapy over the applicability of RCTs to therapies such as acupuncture, and the mistaken assumption that trials which include a control group, are also required to be double-blind. This paper is based on the need for good quality evidence of effectiveness in complementary therapy. It will set out the concerns associated with the use of RCTs within complementary therapy, together with the benefits and limitations of this approach. The paper will go on to review research options and propose some suggestions for future methodological approaches.

Complementary Therapies↗
PubMed1996

Perinatal factors and risk of breast cancer.

A high level of endogenous estrogen in utero has been hypothesized to be a possible risk factor for breast cancer. We used information from two population-based case-control studies to investigate the relation between perinatal factors and risk of invasive breast cancer among women age 21-45 years (746 cases, 960 controls) and women age 50-64 years (401 cases, 439 controls). Breast cancer cases were ascertained through a population-based cancer registry, and controls were selected by random digit dialing. After adjustment for age, menopausal status, and maternal smoking, the birthweight-breast cancer association in women age 21-45 years followed a J-shaped curve, with women whose birthweight was less than 2,500 gm [odds ratio (OR) = 1.3; 95% confidence interval (CI) = 0.9-2.0] and 4,000 gm or more (OR = 1.7; 95% CI = 1.1-2.5) at increased risk. Women age 50-64 years who were 4,000 gm or more at birth appeared to be at slightly reduced risk of breast cancer (OR = 0.6; 95% CI = 0.3-1.1). With the exception of maternal smoking, there was little effect of other perinatal factors on breast cancer risk in either group. These results support the hypothesized association between intrauterine estrogen exposure and subsequent risk of breast cancer.

Adult↗
PubMed1996

Assessing the relative sensitivity to change of rheumatoid arthritis activity measures: is the type of treatment an important third variable?

Observational studies and meta-analyses of controlled clinical trials have been used to identify which measures of rheumatoid arthritis activity are most sensitive to change. These analyses often pool studies of different drugs, although it is not known if arthritis activity measures are differentially responsive to different drugs. In meta-analyses, estimates of the relative sensitivity to change of different measures may also be confounded by differences in drug efficacy, if studies of different drugs contribute different measures to the meta-analysis. To determine if the type of treatment acts as an important effect modifier or confounder in studies of the relative sensitivity to change of arthritis activity measures, we computed effect sizes for four measures (weighted tender joint count, grip strength, duration of morning stiffness, and erythrocyte sedimentation rate) used in each of 16 trials of five different disease-modifying antirheumatic drugs (methotrexate, sulfasalazine, cyclosporin A, intramuscular gold, and D-penicillamine) in rheumatoid arthritis. In a complete factorial analysis of variance, effect sizes differed significantly among drugs (p = 0.0006), but differed only marginally among measures (p = 0.08). No interaction was detectable between drugs and measures. These results suggested that effect modification by drugs was not present, but that pooled estimates of the sensitivity to change of different measures may be confounded in meta-analyses, if trials of more efficacious drugs contribute different measures than trials of less efficacious drugs. In a similar analysis of 26 trials of nine nonsteroidal anti-inflammatory drugs, we found significant differences in effect sizes among measures (p < 0.0001), but no differences among drugs (p = 0.96), and no interaction between drugs and measures. This study suggests that pooled analyses of the relative sensitivity to change of arthritis activity measures based on trials of different disease-modifying drugs may be confounded by drug effects, but confounding by drug effects is unlikely if these meta-analyses are based on trials of different nonsteroidal anti-inflammatory drugs. Although the power of these analyses to detect small interaction effects was limited, effect modification by drugs was not observed, indicating that the measures we examined were not strongly differentially responsive to different drugs.

Anti-Inflammatory Agents, Non-Steroidal↗
PubMed1999

Age at exposure modifies the effects of low-level ionizing radiation on cancer mortality in an occupational cohort.

In a cohort of 4,563 nuclear workers followed retrospectively from 1950 to 1994, we found that age at exposure modified the effects of external radiation dose on cancer mortality. Analyses involved application of conditional logistic regression to risk sets of age- and calendar time-matched cancer deaths, with covariates treated as time dependent and with cumulative radiation doses divided according to the age intervals in which exposure occurred. After adjustment for confounding factors, we found that workers exposed to external radiation after the age of 50 years experienced exposure-related elevations in mortality from cancer at any site [rate ratio (RR) = 1.98; 95% confidence interval (CI) = 0.63-6.26], radiosensitive solid cancer (RR = 3.29; 95% CI = 1.10-9.89), and lung cancer (RR = 3.89; 95% CI = 1.23-12.3) substantially greater (1.6- to 3.5-fold greater) than were seen in coworkers exposed at all earlier ages. In contrast, all of the radiation doses contributing to mortality from cancers of the blood and lymph system were received before age 50 (for age <50, RR = 2.73 and 95% CI = 1.46-5.10; for age > or =50, RR = 0.24 and 95% CI = 0.00-687). Our results for cancer of any site are consistent with the results of previous studies examining the effects of exposure age in nuclear workers. Thus, effects of low-level radiation doses may depend on exposure age, and furthermore, patterns of effect modification by age may differ by type of cancer.

Age Factors↗
PubMed1998

Restriction as a method for reducing bias in the estimation of direct effects.

The direct effect of a treatment on some outcome is that part of the treatment's effect not referred through a specified covariate intermediate on the pathway between treatment and outcome. Such direct effects are often of primary interest in a data analysis. Unfortunately, standard methods of analysis (for example, stratification or modelling) do not, in general, produce consistent estimates of direct effects whether or not the covariate is 'controlled'. Robins and co-authors have proposed two methods for estimation of direct effects applicable when reliable information is available on the covariate. We propose a third approach for reducing bias: data restriction. By restricting the analysis to strata of the data in which the effect of treatment on the covariate is small, we can (under certain assumptions) reduce bias in estimating treatment's direct effect. We discuss these points with reference to difference and ratio measures of treatment effect. The approach will sometimes be applicable even with an unmeasured or poorly measured covariate. We illustrate these points with data from an observational study of the effect of hormone replacement therapy on breast cancer.

Bias↗
PubMed2003

Robustness of inference on measured covariates to misspecification of genetic random effects in family studies.

Family studies to identify disease-related genes frequently collect only families with multiple cases. It is often desirable to determine if risk factors that are known to influence disease risk in the general population also play a role in the study families. If so, these factors should be incorporated into the genetic analysis to control for confounding. Pfeiffer et al. [2001 Biometrika 88: 933-948] proposed a variance components or random effects model to account for common familial effects and for different genetic correlations among family members. After adjusting for ascertainment, they found maximum likelihood estimates of the measured exposure effects. Although it is appealing that this model accounts for genetic correlations as well as for the ascertainment of families, in order to perform an analysis one needs to specify the distribution of random genetic effects. The current work investigates the robustness of the proposed model with respect to various misspecifications of genetic random effects in simulations. When the true underlying genetic mechanism is polygenic with a small dominant component, or Mendelian with low allele frequency and penetrance, the effects of misspecification on the estimation of fixed effects in the model are negligible. The model is applied to data from a family study on nasopharyngeal carcinoma in Taiwan.

Analysis of Variance↗
PubMed1990

Questioning the status quo: sickness absence research so far claims more than it should!

This article is in the nature of a cautionary tale for those working alongside any group of medics who are required to collect epidemiological data, or plan based upon data collected by medical experts without having access to that data. The field of work is occupational health and the specialised area is the relationship between socio-economic variables and absence from work due to sickness. The over-riding criticism of the sickness absence literature to date is that the use of analytical statistics is poor or in most cases almost non-existent and that there is very little understanding of the epidemiological concepts of 'confounding' and 'effect modification'. An example is given of one particular statistical procedure, multi-variate hierarchical log-linear modelling, which if used by future researchers could prevent the recurrence of some of the previous problems.

Absenteeism↗
PubMed1997

Delayed entry into prenatal care: effect of physical violence.

OBJECTIVE: To assess whether women who experienced physical violence by their partner during the 12 months before delivery were more likely to delay entry into prenatal care than were women who had not experienced physical violence. METHODS: We analyzed data from the Pregnancy Risk Assessment Monitoring System. The sample included 27,836 women who delivered live infants during 1993-1994 in nine states and were surveyed 2-6 months after delivery. We calculated risk ratios and 95% confidence intervals (CIs) to measure the association between physical violence within the 12 months before delivery and entry into prenatal care. RESULTS: The prevalence of delayed entry into prenatal care (entering after the first trimester) was 18.1% and that of reported physical violence was 4.7%. Overall, women who experienced physical violence were 1.8 times more likely (95% CI 1.5, 2.1) to have delayed entry into prenatal care than women who had not experienced such violence. When stratifying by selected maternal characteristics, this association was found only for groups of women who were 25 years of age or older or were of higher socioeconomic status. CONCLUSION: Older women and women of higher socioeconomic status who reported physical violence were more likely to delay entry into prenatal care than younger or less affluent women.

Adult↗
PubMed1998

Social class and substance use disorders: the value of social class as distinct from socioeconomic status.

The relationship between social class and substance use disorders (SUDs) is explored and compared to the relationship between SES and SUDs. Social class and SES are two different conceptualizations of socioeconomic inequality (SEI) which emanate from two different theoretical orientations in sociology. SES is commonly used in epidemiological research and is usually measured in terms of education, income or occupational prestige. Social class is less known and less used. Here, following the work of Wright et al. (Wright, E. O., Hachen, D. and Costello, C. et al. (1982) The American class structure. American Sociological Review 47, 709-726) it is measured in terms of four types of control people have in their work place: ownership, control over budget decisions, control over other workers, and control over one's own work. Data are derived from an epidemiological survey, conducted in Israel, using a two stage sampling procedure for the identification of cases. In the first stage 4914 respondents were screened with the Psychiatric Epidemiological Research Interview (PERI). In the second stage (n = 2741), those who screened positive (and a sample of the negatives) were diagnosed by psychiatrists using a structured interview that yielded diagnoses according to the Research Diagnostic Criteria (RDC). The results indicate that those who are advantaged in terms of ownership, i.e. self-employed, have higher rates of SUDs compared to employees. Furthermore, it appears that most disorders have an onset subsequent to entry into the current job, indicating that ownership plays a causal role in the onset of SUDs rather than the other way around. These results are contrasted with those of a previous report from the same study by Dohrenwend et al. (Dohrenwend, B. P., Levav, I. and Shrout, P. E. et al. (1992) Socioeconomic status and psychiatric disorders: the causation selection issue. Science 255, 946-952) which showed just the opposite association between SES and SUDs, i.e. those who are advantaged in terms of SES have lower rates of SUDs. As an explanation of these apparently conflicting results, the possibility is entertained that social class and SES represent independent causal pathways to the onset of SUDs with social class mainly related to primary SUDs and SES mainly to secondary SUDs.

Adult↗
PubMed1997

Advancing the standards of clinical research: the urgent need for new methods and better data.

Some of the weaknesses of the randomized controlled trial (RCT) are pointed out, among them unavoidable biases, the possibility of coming to erroneous conclusions by the 'luck of the draw' and the fact that most such experiments lack a doctor/patient relationship and are of little assistance in clinical decision making. An exaggerated propensity on the part of the medical profession to regard the RCT as the only 'scientific' method for research into therapeutics is noted. The potential advantages of post-marketing surveillance of new interventions in a physician-centred manner are outlined. Such a data base could ultimately be expanded into a data base that records all activity in a given practice. The PACE strategy, employed by haematologists in the north of England, and the sentinel practice primary care research networks used by family doctors contain elements of the system proposed here. However, each possesses disadvantages that limit its applicability to special cases. Random choice of practices for any particular observational study of a new intervention could lend it some of the features of an experiment.

Bias↗
PubMed2000

Statistical methods in epidemiology. IV. Confounding and the matched pairs odds ratio.

PURPOSE: This paper introduces readers to the problem of confounding in epidemiology, how it differs from effect modification and how to deal with it statistically. There are several options for dealing with confounding. These include techniques based on matching, on stratification and on regression. This paper reviews the first method, matching and the matched pairs odds ratio. METHOD: For 1:1 pair matching (1 case and 1 matched control), the matched pairs odds ratio is introduced as the ratio of discordant pairs. A method for calculating confidence intervals based on the normal approximation is described. RESULTS: Some properties of the matched pair design are illustrated by taking examples from the authors' own teaching experiences. CONCLUSION: Matching remains a difficult design option in epidemiology. Its 'best' use is for special types of studies such as for those on twin pairs.

Confidence Intervals↗
PubMed2004

Methods of calculating deaths attributable to obesity.

Previously reported estimates of deaths attributable to obesity in the United States have been based on a method that only partially adjusts for confounding and does not allow for effect modification. In this study, the authors investigated the possible magnitude and direction of bias in estimating deaths attributable to obesity when such a method is used. Hypothetical examples are based on 1991 US population data and published relative risks. Incomplete adjustment for confounding of the obesity-mortality relation by age and sex led to a 17% overestimation of deaths due to obesity. Additional bias resulted from slight differences between the derivation cohort and the target population. For example, a difference of three percentage points in the proportion of people 80 years of age or older led to a 42% overestimation of deaths due to obesity. In addition, these estimates appear to be sensitive to minor differences in relative risks between a derivation cohort and the target population. A difference of 0.20 in relative risks almost doubled the number of deaths (97% overestimation). Estimates of deaths attributable to obesity can be biased if confounding and effect modification are not properly taken into account or if the relative risks are not estimated accurately.

Adolescent↗
PubMed2006

Variable selection for propensity score models.

Despite the growing popularity of propensity score (PS) methods in epidemiology, relatively little has been written in the epidemiologic literature about the problem of variable selection for PS models. The authors present the results of two simulation studies designed to help epidemiologists gain insight into the variable selection problem in a PS analysis. The simulation studies illustrate how the choice of variables that are included in a PS model can affect the bias, variance, and mean squared error of an estimated exposure effect. The results suggest that variables that are unrelated to the exposure but related to the outcome should always be included in a PS model. The inclusion of these variables will decrease the variance of an estimated exposure effect without increasing bias. In contrast, including variables that are related to the exposure but not to the outcome will increase the variance of the estimated exposure effect without decreasing bias. In very small studies, the inclusion of variables that are strongly related to the exposure but only weakly related to the outcome can be detrimental to an estimate in a mean squared error sense. The addition of these variables removes only a small amount of bias but can increase the variance of the estimated exposure effect. These simulation studies and other analytical results suggest that standard model-building tools designed to create good predictive models of the exposure will not always lead to optimal PS models, particularly in small studies.

Confounding Factors, Epidemiologic↗
PubMed2006

Dietary flavonoid intake and breast cancer risk among women on Long Island.

Flavonoids are found in a variety of foods and have anticarcinogenic properties in experimental models. Few epidemiologic studies have examined whether flavonoid intake is associated with breast cancer in humans. In this study, the authors investigated whether dietary flavonoid intake was associated with reduced risk of breast cancer in a population-based sample of US women. They conducted a case-control study among women who resided in Nassau and Suffolk counties on Long Island, New York. Cases and controls were interviewed about known and suspected risk factors and asked to complete a food frequency questionnaire regarding their average intake in the prior 12 months. A total of 1,434 breast cancer cases and 1,440 controls provided adequate responses. A decrease in breast cancer risk was associated with flavonoid intake; the decrease was most pronounced among postmenopausal women for flavonols (odds ratio (OR) = 0.54, 95% confidence interval (CI): 0.40, 0.73), flavones (OR = 0.61, 95% CI: 0.45, 0.83), flavan-3-ols (OR = 0.74, 95% CI: 0.55, 0.99), and lignans (OR = 0.69, 95% CI: 0.51, 0.94). The authors conclude that intake of flavonols, flavones, flavan-3-ols, and lignans is associated with reduced risk of incident postmenopausal breast cancer among Long Island women. These results suggest that US women can consume sufficient levels of flavonoids to benefit from their potential chemopreventive effects.

Adult↗
PubMed1989

Analytical implications of epidemiological concepts of interaction.

In contrast to definitions based on statistical or biological concepts, Rothman has adopted an unambiguous epidemiological definition of interaction in which two factors are not 'independent' if they are component causes in the same sufficient cause. This leads to the adoption of additivity of incidence rates as the state of 'no interaction'. However, there are other considerations which generally favour the use of multiplicative models. This implies an apparent dilemma as to how an analysis can be conducted which combines the advantages of ratio measures of effect with the assessment of independence in terms of a departure from additivity. These apparently contradictory goals can be reconciled through the analysis of separate and joint effects. This approach is discussed with reference to studies of asbestos exposure, cigarette smoking and lung cancer.

Causality↗
PubMed2001

Ecologic versus individual-level sources of bias in ecologic estimates of contextual health effects.

A number of authors have attempted to defend ecologic (aggregate) studies by claiming that the goal of those studies is estimation of ecologic (contextual or group-level) effects rather than individual-level effects. Critics of these attempts point out that ecologic effect estimates are inevitably used as estimates of individual effects, despite disclaimers. A more subtle problem is that ecologic variation in the distribution of individual effects can bias ecologic estimates of contextual effects. The conditions leading to this bias are plausible and perhaps even common in studies of ecosocial factors and health outcomes because social context is not randomized across typical analysis units (administrative regions). By definition, ecologic data contain only marginal observations on the joint distribution of individually defined confounders and outcomes, and so identify neither contextual nor individual-level effects. While ecologic studies can still be useful given appropriate caveats, their problems are better addressed by multilevel study designs, which obtain and use individual as well as group-level data. Nonetheless, such studies often share certain special problems with ecologic studies, including problems due to inappropriate aggregation and problems due to temporal changes in covariate distributions.

Bias↗