Sample Size Guidelines for Logistic Regression from Observational Studies with Large Population: Emphasis on the Accuracy Between Statistics and Parameters Based on Real Life Clinical Data

Abstract

BackgroundDifferent study designs and population size may require different sample size for logistic regression. This study aims to propose sample size guidelines for logistic regression based on observational studies with large population.MethodsWe estimated the minimum sample size required based on evaluation from real clinical data to evaluate the accuracy between statistics derived and the actual parameters. Nagelkerke r-squared and coefficients derived were compared with their respective parameters.ResultsWith a minimum sample size of 500, results showed that the differences between the sample estimates and the population was sufficiently small. Based on an audit from a medium size of population, the differences were within ± 0.5 for coefficients and ± 0.02 for Nagelkerke r-squared. Meanwhile for large population, the differences are within ± 1.0 for coefficients and ± 0.02 for Nagelkerke r-squared.ConclusionsFor observational studies with large population size that involve logistic regression in the analysis, taking a minimum sample size of 500 is necessary to derive the statistics that represent the parameters. The other recommended rules of thumb are EPV of 50 and formula; n = 100 + 50i where i refers to number of independent variables in the final model.

Cultural Adaptation, Validation, and Primary Application of a Questionnaire to Assess Intentions to Eat Low-Glycemic Index Foods among Rural Chinese Women

“…The present study data were drawn from the baseline survey used in the project mentioned above. According to the rule of thumb, 15–20 subjects are required for per predictor in a logistic regression model [ 21 ]. We assumed 10–15 predictors in this study, and considering a 20% non-response or drop-off rate, the range of the final sample size was determined to be 180 to 360.…”

Associations between the Chronotypes and Eating Habits of Hong Kong School-Aged Children

“…Questionnaires that were returned with any incomplete data relating to the CCTQ or that contained more than two missing items concerning eating habits were excluded. We targeted sample size to be 500, which is the recommended minimum sample size for observational studies using logistic regression (based on the Events Per Variable criterion (EPV) of 50) [22]. Considering 20% of missing data, we aimed to collect data from 600 children and their parents.…”

Is higher psychotropic medication burden associated with involuntary treatment under the Mental Health Act? A four-year Australian cohort study

“…The cohort was identified from an electronic discharge report; the researchers checked the clinical record of each patient to ensure psychotropic medication(s) were prescribed at the point of discharge, moving back in time (from the 31st January of each study year) until a sample of 200 consecutive discharges was achieved. The sample size (n = 800) was deemed large enough to provide sufficient statistical power [31], as well as an accurate account of prescribing practices within the selected facility.…”

“…Additional file 1: Table S1. Maximum daily doses (MDD) of psychotropic medications as per clinical guidelines [31,32]. Authors' contributions AW was involved in study design, ethics approval, the data collection protocol, student supervision and training, data analysis and drafted the manuscript.…”