Article

A simple method for estimating relative risk using logistic regression

Grupo Latinoamericano de Investigaciones Epidemiológicas, Organización Latinoamericana para el Fomento de la Investigación en Salud, Bucaramanga, Colombia.
BMC Medical Research Methodology (Impact Factor: 2.17). 02/2012; 12:14. DOI: 10.1186/1471-2288-12-14
Source: PubMed

ABSTRACT Odds ratios (OR) significantly overestimate associations between risk factors and common outcomes. The estimation of relative risks (RR) or prevalence ratios (PR) has represented a statistical challenge in multivariate analysis and, furthermore, some researchers do not have access to the available methods. Objective: To propose and evaluate a new method for estimating RR and PR by logistic regression.
A provisional database was designed in which events were duplicated but identified as non-events. After, a logistic regression was performed and effect measures were calculated, which were considered RR estimations. This method was compared with binomial regression, Cox regression with robust variance and ordinary logistic regression in analyses with three outcomes of different frequencies.
ORs estimated by ordinary logistic regression progressively overestimated RRs as the outcome frequency increased. RRs estimated by Cox regression and the method proposed in this article were similar to those estimated by binomial regression for every outcome. However, confidence intervals were wider with the proposed method.
This simple tool could be useful for calculating the effect of risk factors and the impact of health interventions in developing countries when other statistical strategies are not available.

Download full-text

Full-text

Available from: Fredi Diaz-Quijano, Dec 13, 2013
0 Followers
 · 
236 Views
  • Source
    • "The LR on the modified dataset will be mathematically equivalent to log binomial model but uses the logit link instead of log therefore it ensures that prediction be between 0 and 1. Also, this method directly provides RR instead of OR and avoids the convergence problem of log binomial model [5]. This form of analysis over represents the total cases. "
  • Source
    • "The LR on the modified dataset will be mathematically equivalent to log binomial model but uses the logit link instead of log therefore it ensures that prediction be between 0 and 1. Also, this method directly provides RR instead of OR and avoids the convergence problem of log binomial model [5]. This form of analysis over represents the total cases. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Studying the effect of exposure or intervention on a dichotomous outcome is very common in medical research. Logistic regression (LR) is often used to determine such association which provides odds ratio (OR). OR often overestimates the effect size for prevalent outcome data. In such situations, use of relative risk (RR) has been suggested. We propose modifications in Zhang and Yu and Diaz-Quijano methods. These methods were compared with stratified Mantel Haenszel method, LR, log binomial regression (LBR), Zhang and Yu method, Poisson/Cox regression, modified Poisson/Cox regression, marginal probability method, COPY method, inverse probability of treatment weighted LBR, and Diaz-Quijano method. Our proposed modified Diaz-Quijano (MDQ) method provides RR and its confidence interval similar to those estimated by modified Poisson/Cox and LBRs. The proposed modifications in Zhang and Yu method provides better estimate of RR and its standard error as compared to Zhang and Yu method in a variety of situations with prevalent outcome. The MDQ method can be used easily to estimate the RR and its confidence interval in the studies which require reporting of RRs. Regression models which directly provide the estimate of RR without convergence problems such as the MDQ method and modified Poisson/Cox regression should be preferred.
    Journal of Applied Statistics 09/2003; 31(3). DOI:10.1080/02664763.2013.840772 · 0.45 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Multimorbidity, presence of two or more health conditions, is a widespread phenomenon affecting populations' health all over the world. It becomes a serious public health concern due to its negative consequences on quality of life, mortality, and cost of healthcare services utilization. Studies exploring determinants of multimorbidity are limited, particularly those looking at vulnerable populations prospectively over time. This study aimed at identifying short and long term socioeconomic, psychosocial, and health behavioral determinants of incident multimorbidity among a cohort of the 1988 Armenian earthquake survivors. The study included a representative subsample of 725 from a larger initial cohort of the earthquake survivors. Data on this subsample were collected via four phases of this cohort study during the period 1990--2012. The final logistic regression analysis eliminated all those cases with baseline multimorbidity to investigate short and long term determinants of incident multimorbidity; this subsample included 600 participants. More than 75% of the studied sample had multimorbidity. Perceived low affordability of healthcare services, poor living standards during the post-earthquake decade, and lower education were independent predictors of incident multimorbidity developed during the period 1990--2012. Stressful life events and poor social support were among psychosocial determinants of incident multimorbidity. Participants' baseline BMI reported in 1990 was independently associated with incident multimorbidity. Most of the identified determinants of incident multimorbidity in our study population were markers of social inequities, indicating that inequities pose a serious threat to both individual and public health-related outcomes. Strategies targeting to decrease such inequities along with promotion of healthy lifestyle and strengthening of social networks may considerably reduce multimorbidity among population groups with similar socioeconomic and cultural profiles.
    International Journal for Equity in Health 08/2013; 12(1):68. DOI:10.1186/1475-9276-12-68 · 1.71 Impact Factor
Show more

Similar Publications