Hamling J, Lee P, Weitkunat R, Ambuhl MFacilitating meta-analyses by deriving relative effect and precision estimates for alternative comparisons from a set of estimates presented by exposure level or disease category. Stat Med 27: 954-970

P.N. Lee Statistics and Computing Ltd., 17 Cedar Road, Sutton, Surrey, U.K.
Statistics in Medicine (Impact Factor: 1.83). 03/2008; 27(7):954-70. DOI: 10.1002/sim.3013
Source: PubMed


Epidemiological studies relating a particular exposure to a specified disease may present their results in a variety of ways. Often, results are presented as estimated odds ratios (or relative risks) and confidence intervals (CIs) for a number of categories of exposure, for example, by duration or level of exposure, compared with a single reference category, often the unexposed. For systematic literature review, and particularly meta-analysis, estimates for an alternative comparison of the categories, such as any exposure versus none, may be required. Obtaining these alternative comparisons is not straightforward, as the initial set of estimates is correlated. This paper describes a method for estimating these alternative comparisons based on the ideas originally put forward by Greenland and Longnecker, and provides implementations of the method, developed using Microsoft Excel and SAS. Examples of the method based on studies of smoking and cancer are given. The method also deals with results given by categories of disease (such as histological types of a cancer). The method allows the use of a more consistent comparison when summarizing published evidence, thus potentially improving the reliability of a meta-analysis.

Download full-text


Available from: Peter N Lee, Jan 25, 2016
  • Source
    • "Additionally, the significant association between passive maternalsmoking and increased risk of preterm birth was consistently observed in studies with more than 100 cases of preterm birth and in studies adjusted for maternal age, socioeconomic status and/or education, body mass index, and parity (Table 5).[30]. Additionally, we excluded 2 studies[33,37]in which risk estimates were recalculated by the effective-count method proposed by Hamling et al[38]; this result was robust (SOR = 1.21, 95% CI = 1.07– 1.37, I 2 = 42.6%). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Previous studies investigating the relationship between passive maternal smoking and preterm birth reveal inconsistent results. We conducted the current meta-analysis of observational studies to evaluate the relationship between passive maternal smoking and preterm birth. We identified relevant studies by searching PubMed, EMBASE, and ISI Web of Science databases. We used random-effects models to estimate summary odds ratios (SORs) and 95% confidence intervals (CIs) for aforementioned association. For the analysis, we included 24 studies that involved a total of 5607 women who experienced preterm birth. Overall, the SORs of preterm birth for women who were ever exposed to passive smoking versus women who had never been exposed to passive smoking at any place and at home were 1.20 (95%CI = 1.07-1.34,I2 = 36.1%) and 1.16 (95%CI = 1.04-1.30,I2 = 4.4%), respectively. When we conducted a stratified analysis according to study design, the risk estimate was slightly weaker in cohort studies (SOR = 1.10, 95%CI = 1.00-1.21,n = 16) than in cross-sectional studies (SOR = 1.47, 95%CI = 1.23-1.74,n = 5). Additionally, the associations between passive maternal smoking and preterm birth were statistically significant for studies conducted in Asia (SOR = 1.26, 95%CI = 1.05-1.52), for studies including more than 100 cases of preterm birth (SOR = 1.22, 95%CI = 1.05-1.41), and for studies adjusted for maternal age (SOR = 1.27,95%CI = 1.09-1.47), socioeconomic status and/or education (SOR = 1.28, 95%CI = 1.10-1.49), body mass index (SOR = 1.33, 95%CI = 1.04-1.71), and parity (SOR = 1.27, 95%CI = 1.13-1.43). Our findings demonstrate that passive maternal smoking is associated with an increased risk of preterm birth. Future prospective cohort studies are warranted to provide more detailed results stratified by passive maternal smoking during different trimesters of pregnancy and by different types and causes of preterm birth.
    Full-text · Article · Jan 2016 · PLoS ONE
  • Source
    • "This means that the off-diagonal values of the covariance matrix V(ε i ) = S i are different from zero. Two different methods have been proposed to approximate the correlation between the non-referent log(RR)s (Greenland and Longnecker, 1992; Hamling et al., 2008). For each study, the vector of regression parameters β i and its variance–covariance matrix V(β i ) can be efficiently estimated through the generalized least squares (GLS) estimator (Greenland and Longnecker, 1992; Orsini et al., 2012). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Goodness of fit evaluation should be a natural step in assessing and reporting dose-response meta-analyses from aggregated data of binary outcomes. However, little attention has been given to this topic in the epidemiological literature, and goodness of fit is rarely, if ever, assessed in practice. We briefly review the two-stage and one-stage methods used to carry out dose-response meta-analyses. We then illustrate and discuss three tools specifically aimed at testing, quantifying, and graphically evaluating the goodness of fit of dose-response meta-analyses. These tools are the deviance, the coefficient of determination, and the decorrelated residuals-versus-exposure plot. Data from two published meta-analyses are used to show how these three tools can improve the practice of quantitative synthesis of aggregated dose-response data. In fact, evaluating the degree of agreement between model predictions and empirical data can help the identification of dose-response patterns, the investigation of sources of heterogeneity, and the assessment of whether the pooled dose-response relation adequately summarizes the published results. © 2015 The Authors. Research Synthesis Methods published by John Wiley & Sons, Ltd.
    Full-text · Article · Dec 2015 · Research Synthesis Methods
  • Source
    • "Evidence for non-linearity was determined by the significance level of the second meta-regression coefficient. When no evidence for non-linearity was found, we conducted linear two-stage meta-regressions (Hamling et al., 2008; Orsini and Greenland, 2006). To determine potential threshold effects, we conducted categorical meta-analyses on the relationship between alcohol consumption (cut-points defined by 20 g/day intervals). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background: Pancreatitis is a highly prevalent medical condition associated with a spectrum of endocrine and exocrine pancreatic insufficiencies. While high alcohol consumption is an established risk factor for pancreatitis, its relationship with specific types of pancreatitis and a potential threshold have not been systematically examined. Methods: We conducted a systematic literature search for studies on the association between alcohol consumption and pancreatitis based on PRISMA guidelines. Non-linear and linear random-effect dose-response meta-analyses using restricted cubic spline meta-regressions and categorical meta-analyses in relation to abstainers were conducted. Findings: Seven studies with 157,026 participants and 3618 cases of pancreatitis were included into analyses. The dose-response relationship between average volume of alcohol consumption and risk of pancreatitis was monotonic with no evidence of non-linearity for chronic pancreatitis (CP) for both sexes (p = 0.091) and acute pancreatitis (AP) in men (p = 0.396); it was non-linear for AP in women (p = 0.008). Compared to abstention, there was a significant decrease in risk (RR = 0.76, 95%CI: 0.60-0.97) of AP in women below the threshold of 40. g/day. No such association was found in men (RR = 1.1, 95%CI: 0.69-1.74). The RR for CP at 100. g/day was 6.29 (95%CI: 3.04-13.02). Interpretation: The dose-response relationships between alcohol consumption and risk of pancreatitis were monotonic for CP and AP in men, and non-linear for AP in women. Alcohol consumption below 40. g/day was associated with reduced risk of AP in women. Alcohol consumption beyond this level was increasingly detrimental for any type of pancreatitis. Funding: The work was financially supported by a grant from the National Institute on Alcohol Abuse and Alcoholism (R21AA023521) to the last author.
    Full-text · Article · Nov 2015 · EBioMedicine
Show more