A comparison of two methods for estimating odds ratios: Results from the National Health Survey
2008

Comparing Methods for Estimating Odds Ratios in Obesity Research

Sample size: 14176 publication Evidence: high

Author Information

Author(s): Bakhshi Enayatollah, Eshraghian Mohammad R, Mohammad Kazem, Seifi Behjat

Primary Institution: Tehran University/Medical Sciences, Iran

Hypothesis

Does the 'without dichotomizing' method provide more accurate odds ratios for obesity compared to the 'dichotomizing' method?

Conclusion

The 'without dichotomizing' method is more useful for estimating odds ratios and provides more precise results than the 'dichotomizing' method.

Supporting Evidence

  • The odds ratio estimates changed only slightly over the two methods.
  • The 'without dichotomizing' method provided shorter confidence intervals on the odds ratio parameters than the dichotomizing method.
  • Urban women had significantly higher odds of obesity than their rural counterparts.
  • Age was directly associated with obesity.
  • Education was inversely associated with obesity.
  • Non-smoker women were more obese than smokers.
  • Married women had significantly higher odds of obesity than their non-married counterparts.
  • An association was observed between economic index and obesity.

Takeaway

This study shows that when looking at weight data, it's better to keep the numbers as they are instead of splitting them into groups, because you get clearer results.

Methodology

A comparative study using data from the National Health Survey in Iran, comparing two methods for estimating odds ratios.

Potential Biases

Potential bias due to the exclusion of pregnant women and reliance on a single dataset.

Limitations

The study's cross-sectional nature limits conclusions about causality, and it did not include physical activity or income data.

Participant Demographics

14176 women aged 20–69 years, including 8957 urban and 5219 rural participants.

Statistical Information

P-Value

p<0.05

Confidence Interval

95% CI: 1.916–2.914

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1186/1471-2288-8-78

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication