Two methods for point and interval estimation of relative risk for log-linear exposure-response relations in meta-analyses of published ordinal categorical exposure-response data have been proposed. The authors compared the results of a meta-analysis of published data using each of the 2 methods with the results that would be obtained if the primary data were available and investigated the circumstances under which the approximations required for valid use of each meta-analytic method break down. They then extended the methods to handle nonlinear exposure-response relations. In the present article, methods are illustrated using studies of the relation between alcohol consumption and colorectal and lung cancer risks from the ongoing Pooling Project of Prospective Studies of Diet and Cancer. In these examples, the differences between the results of a meta-analysis of summarized published data and the pooled analysis of the individual original data were small. However, incorrectly assuming no correlation between relative risk estimates for exposure categories from the same study gave biased confidence intervals for the trend and biased P values for the tests for nonlinearity and between-study heterogeneity when there was strong confounding by other model covariates. The authors illustrate the use of 2 publicly available user-friendly programs (Stata and SAS) to implement meta-analysis for dose-response data.
1 February 2006 in The Stata Journal: Promoting communications on statistics and Stata
This paper presents a command, glst, for trend estimation across different exposure levels for either single or multiple summarized case–control, incidence-rate, and cumulative incidence data. This approach is based on constructing an approximate covariance estimate for the log relative risks and estimating a corrected linear trend using generalized least squares. For trend analysis of multiple studies, glst can estimate fixed- and random-effects metaregression models.
24 October 2014 in International Journal of Behavioral Nutrition and Physical Activity
Walking and cycling have shown beneficial effects on population risk of all-cause mortality (ACM). This paper aims to review the evidence and quantify these effects, adjusted for other physical activity (PA). We conducted a systematic review to identify relevant studies. Searches were conducted in November 2013 using the following health databases of publications: Embase (OvidSP); Medline (OvidSP); Web of Knowledge; CINAHL; SCOPUS; SPORTDiscus. We also searched reference lists of relevant texts and reviews. Eligible studies were prospective cohort design and reporting walking or cycling exposure and mortality as an outcome. Only cohorts of individuals healthy at baseline were considered eligible. Extracted data included study population and location, sample size, population characteristics (age and sex), follow-up in years, walking or cycling exposure, mortality outcome, and adjustment for other co-variables. We used random-effects meta-analyses to investigate the beneficial effects of regular walking and cycling. Walking (18 results from 14 studies) and cycling (8 results from 7 studies) were shown to reduce the risk of all-cause mortality, adjusted for other PA. For a standardised dose of 11.25 MET.hours per week (or 675 MET.minutes per week), the reduction in risk for ACM was 11% (95% CI = 4 to 17%) for walking and 10% (95% CI = 6 to 13%) for cycling. The estimates for walking are based on 280,000 participants and 2.6 million person-years and for cycling they are based on 187,000 individuals and 2.1 million person-years. The shape of the dose-response relationship was modelled through meta-analysis of pooled relative risks within three exposure intervals. The dose-response analysis showed that walking or cycling had the greatest effect on risk for ACM in the first (lowest) exposure interval. The analysis shows that walking and cycling have population-level health benefits even after adjustment for other PA. Public health approaches would have the biggest impact if they are able to increase walking and cycling levels in the groups that have the lowest levels of these activities. The review protocol was registered with PROSPERO (International database of prospectively registered systematic reviews in health and social care) PROSPERO 2013: CRD42013004266.
Background: The relationship between dietary sodium intake and blood pressure (BP) has been tested in clinical trials and nonexperimental human studies, indicating a direct association. The exact shape of the dose–response relationship has been difficult to assess in clinical trials because of the lack of random-effects dose–response statistical models that can include 2-arm comparisons. Methods: After performing a comprehensive literature search for experimental studies that investigated the BP effects of changes in dietary sodium intake, we conducted a dose–response meta-analysis using the new 1-stage cubic spline mixed-effects model. We included trials with at least 4 weeks of follow-up; 24-hour urinary sodium excretion measurements; sodium manipulation through dietary change or supplementation, or both; and measurements of systolic and diastolic BP at the beginning and end of treatment. Results: We identified 85 eligible trials with sodium intake ranging from 0.4 to 7.6 g/d and follow-up from 4 weeks to 36 months. The trials were conducted in participants with hypertension (n=65), without hypertension (n=11), or a combination (n=9). Overall, the pooled data were compatible with an approximately linear relationship between achieved sodium intake and mean systolic as well as diastolic BP, with no indication of a flattening of the curve at either the lowest or highest levels of sodium exposure. Results were similar for participants with or without hypertension, but the former group showed a steeper decrease in BP after sodium reduction. Intervention duration (≥12 weeks versus 4 to 11 weeks), type of study design (parallel or crossover), use of antihypertensive medication, and participants’ sex had little influence on the BP effects of sodium reduction. Additional analyses based on the BP effect of difference in sodium exposure between study arms at the end of the trial confirmed the results on the basis of achieved sodium intake. Conclusions: In this dose–response analysis of sodium reduction in clinical trials, we identified an approximately linear relationship between sodium intake and reduction in both systolic and diastolic BP across the entire range of dietary sodium exposure. Although this occurred independently of baseline BP, the effect of sodium reduction on level of BP was more pronounced in participants with a higher BP level.
Introduction: Although previous studies found that cigarette smoking is associated with risk of rheumatoid arthritis (RA), the dose-response relationship remains unclear. This meta-analysis quantitatively summarizes accumulated evidence regarding the association of lifelong exposure to cigarette smoking assessed as pack-years with the risk of RA. Methods: Relevant studies were identified by a search of MEDLINE and EMBASE from 1966 to October 2013, with no restrictions. Reference lists from retrieved articles were also reviewed. Studies that reported relative risks (RR) or odds ratio (OR) estimates with 95% confidence intervals (CIs) for the association between pack-years of cigarette smoking and rheumatoid arthritis were included in a dose-response random-effects meta-regression analysis. Results: We included 3 prospective cohorts and 7 case-control studies in the meta-analysis. They included a total of 4,552 RA cases. There was no indication of heterogeneity (Pheterogeneity = 0.32) and publication bias did not affect the results. Compared to never smokers, the risk of developing RA increased by 26% (RR = 1.26, 95% CI 1.14 to 1.39) among those who smoked 1 to 10 pack-years and doubled among those with more than 20 pack-years (RR for 21 to 30 pack years = 1.94, 95% CI 1.65 to 2.27). The risk of RA was not increasing further for higher exposure levels (RR for >40 pack-years = 2.07, 95% CI 1.15 to 3.73). The risk of RA was statistically significantly higher among rheumatoid factor (RF)-positive RA cases (RR = 2.47, 95% CI 2.02 to 3.02) compared to RF-negative (RR = 1.58, 95% CI 1.15 to 2.18) when comparing the highest versus lowest category of pack-years for the individual studies. Conclusions: Lifelong cigarette smoking was positively associated with the risk of RA even among smokers with a low lifelong exposure. The risk of RA did not further increase with an exposure higher than 20 pack-years.
1 March 2011 in The Stata Journal: Promoting communications on statistics and Stata
The use of flexible models for the relationship between a quantitative covariate and the response variable can be limited by the difficulty in interpreting the regression coefficients. In this article, we present a new postestimation command, xblc, that facilitates tabular and graphical presentation of these relationships. Cubic splines are given special emphasis. We illustrate the command through several worked examples using data from a large study of Swedish men on the relation between physical activity and the occurrence of lower urinary tract symptoms.