The Number of Events per Confounder for Valid Estimation of Risk Difference Using Modified Least-Squares Regression

Hagiwara, Yasuhiro; Fukuda, Musashi; Matsuyama, Yutaka

doi:10.1093/aje/kwy158

Cited by 5 publications

(4 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several candidate methods are available for enhancing the performance of standard errors and the resulting confidence intervals. In modified Poisson regression, sandwich standard errors with small sample correction may be explored, as has been considered for linear regression [ 32 ] and modified least-squares regression for risk difference estimation [ 33 , 34 ]. In regression standardization, the bootstrap method may be preferred for small sample sizes if the computational time is not critical [ 13 , 35 ].…”

Section: Discussionmentioning

confidence: 99%

A simulation study of regression approaches for estimating risk ratios in the presence of multiple confounders

Fuyama

Hagiwara

Matsuyama

2021

Emerg Themes Epidemiol

Self Cite

View full text Add to dashboard Cite

Background Risk ratio is a popular effect measure in epidemiological research. Although previous research has suggested that logistic regression may provide biased odds ratio estimates when the number of events is small and there are multiple confounders, the performance of risk ratio estimation has yet to be examined in the presence of multiple confounders. Methods We conducted a simulation study to evaluate the statistical performance of three regression approaches for estimating risk ratios: (1) risk ratio interpretation of logistic regression coefficients, (2) modified Poisson regression, and (3) regression standardization using logistic regression. We simulated 270 scenarios with systematically varied sample size, the number of binary confounders, exposure proportion, risk ratio, and outcome proportion. Performance evaluation was based on convergence proportion, bias, standard error estimation, and confidence interval coverage. Results With a sample size of 2500 and an outcome proportion of 1%, both logistic regression and modified Poisson regression at times failed to converge, and the three approaches were comparably biased. As the outcome proportion or sample size increased, modified Poisson regression and regression standardization yielded unbiased risk ratio estimates with appropriate confidence intervals irrespective of the number of confounders. The risk ratio interpretation of logistic regression coefficients, by contrast, became substantially biased as the outcome proportion increased. Conclusions Regression approaches for estimating risk ratios should be cautiously used when the number of events is small. With an adequate number of events, risk ratios are validly estimated by modified Poisson regression and regression standardization, irrespective of the number of confounders.

show abstract

Section: Discussionmentioning

confidence: 99%

A simulation study of regression approaches for estimating risk ratios in the presence of multiple confounders

Fuyama

Hagiwara

Matsuyama

2021

Emerg Themes Epidemiol

Self Cite

View full text Add to dashboard Cite

show abstract

“…For continuous outcomes, the mean difference was estimated using linear models, adjusting study and baseline serum zinc concentration as fixed-effect covariates. For binomial outcomes, the risk difference was estimated with linear probability models by the modified least-squares method, adjusting study as a fixed-effect covariate [19,20]. Heterogeneity of treatment effects between trials was tested with the interaction term between treatment and study as a fixed effect.…”

Section: Discussionmentioning

confidence: 99%

Efficacy and Safety of Polaprezinc (Zinc Compound) on Zinc Deficiency: A Systematic Review and Dose–Response Meta-Analysis of Randomized Clinical Trials Using Individual Patient Data

et al. 2020

View full text Add to dashboard Cite

Zinc intake is recommended for zinc deficiency. In clinical practice, polaprezinc has been used as a zinc replacement therapy for zinc deficiency. However, the efficacy of polaprezinc has not been established. To confirm the efficacy on zinc deficiency of polaprezinc and provide additional information on an appropriate regimen, we conducted a systematic review using individual patient data (IPD). We searched PubMed, the Japanese database Ichushi, and the database owned by the marketing authorization holder of polaprezinc. Randomized placebo-controlled trials that reported the serum zinc concentration were eligible. The mean difference of the change from baseline in serum zinc concentration was estimated using a fixed-effects model. The linear dose–response relationship and the subgroup effects were also assessed. Out of 54 unique randomized clinical trials (RCTs), four studies met the eligibility criteria, and we could access IPD for all of them. All three doses of polaprezinc (75 mg, 150 mg, and 300 mg) and the placebo group were examined. The dose-combined overall polaprezinc increased the change from baseline by a mean of 9.08 µg/dL (95% confidence interval: 5.46, 12.70; heterogeneity: I 2 = 0.61%) compared to the placebo. A significant dose–response relationship was confirmed (p < 0.001). Baseline serum zinc concentration was considered an effect modifier in polaprezinc 300 mg. All doses of polaprezinc were tolerable, but a dose–response relationship with adverse events (AEs) was observed in gastrointestinal disorders. The dose of 300 mg may be useful among patients with baseline serum zinc concentration of less than 70 µg/dL, and 150 mg for 70 µg/dL or more.

show abstract

“…Data were compared using one‐way analysis of variance, the Chi‐squared test, and Fisher's exact test, as appropriate. Differences in risk and robust 95% confidence intervals (CIs) of contamination according to sites and topical disinfectants were estimated using uni‐ and multivariate analyses with modified least‐squares regression and a robust standard error estimator 17,18 . The same patients were considered to be a random effect in the above model.…”

Section: Methodsmentioning

confidence: 99%

“…Differences in risk and robust 95% confidence intervals (CIs) of contamination according to sites and topical disinfectants were estimated using uni‐ and multivariate analyses with modified least‐squares regression and a robust standard error estimator. 17 , 18 The same patients were considered to be a random effect in the above model. Age, sex, and doctors' experience were adjusted as confounders in multivariate analyses.…”

Section: Methodsmentioning

confidence: 99%

Sites of blood collection and topical disinfectants associated with contaminated cultures: An ambidirectional cohort study

Ota,

Nishioka,

Hamada

et al. 2023

J of Gen and Family Med

View full text Add to dashboard Cite

BackgroundWe aimed to determine whether puncture sites for blood sampling and topical disinfectants are associated with rates of contaminated blood cultures in the emergency department (ED) of a single institution.MethodsThis single‐center, ambidirectional cohort study of 548 consecutive patients ≥20 years of age was performed in the ED of a university hospital in Japan over a 13‐month period. Pairs of blood samples were collected for aerobic and anaerobic cultures from patients in the ED. Physicians selected puncture sites and topical disinfectants according to their personal preference.ResultsPotential contamination was identified in 110 of the 548 patients (20.1%). One hundred fourteen (20.8%) patients showed true‐positive results for bacteremia, and 324 (59.1%) patients showed true‐negative results. Multivariate analysis revealed more frequent contamination when puncture sites were disinfected with povidone‐iodine (PVI) than with alcohol/chlorhexidine (ACHX) (adjusted risk difference, 19.1%; 95% confidence interval [CI]), 15.7–22.6; p < 0.001). In terms of blood collection sites, femoral and central venous (CV) catheter with PVI disinfection showed more frequent contamination than venous sites with ACHX (adjusted risk differences: 26.6%, 95% CI 21.3–31.9, p < 0.001 and 41.1%, 95% CI 22.2–59.9, p < 0.001, respectively).ConclusionRates of contaminated blood cultures were significantly higher when blood was collected from the CV catheter or femoral sites with PVI as the topical disinfectant.

show abstract

The Number of Events per Confounder for Valid Estimation of Risk Difference Using Modified Least-Squares Regression

Cited by 5 publications

References 40 publications

A simulation study of regression approaches for estimating risk ratios in the presence of multiple confounders

A simulation study of regression approaches for estimating risk ratios in the presence of multiple confounders

Efficacy and Safety of Polaprezinc (Zinc Compound) on Zinc Deficiency: A Systematic Review and Dose–Response Meta-Analysis of Randomized Clinical Trials Using Individual Patient Data

Sites of blood collection and topical disinfectants associated with contaminated cultures: An ambidirectional cohort study

Contact Info

Product

Resources

About