Can we do a meta-analysis of data of 3 regions. A particular disease was treated with the same treatment but implemented thoroughly in 2 regions and not so thoroughly in 1 region. How to proceed with analysis? What is the best way to analyse these data. The data are of a disease (cases and deaths) as follows:

          Region 1      Region 2         Region 3           
        Cases   Dead    Cases   Dead     Cases  Dead    Total cases     Total dead cases
2006    2320    528     1484    108       73    3       3877            639
2007    3024    645     1592    75        32    1       4648            721
2008    3012    537     1920    53         3    0       4935            590
2009    3073    556     1477    40       246    8       4796            604
2010    3540    494     1460    26       138    1       5138            521
Total  14969    2760    7933    302      492    13     23394            3075

Updated These are unpublished data.

  • 1
    $\begingroup$ What is the aim of the analysis? What question(s) are you trying to answer? $\endgroup$ – onestop Feb 10 '11 at 16:59
  • $\begingroup$ @onestop: Thank you for your time. The same treatment was given in all 3 regions but with different levels of dedication to the suffering patient due to various administrative reasons. The aim is to show that there is a significant increase in the number of deaths in 1 region due to poor implementation of the treatment guidelines. $\endgroup$ – DrWho Feb 11 '11 at 5:04
  • $\begingroup$ @DrWho: I'm not sure if meta-analysis is the correct approach to analyze this dataset. It seems more like classic longitudinal design. I think you should reconsider your methodology. $\endgroup$ – Andrej Feb 11 '11 at 7:56
  • $\begingroup$ @Andrej: can you kindly say a few more words about the design? $\endgroup$ – DrWho Feb 11 '11 at 14:32
  • $\begingroup$ @DrWho, i think it would be more helpful if you said a few more words about the design! Are the numbers under 'cases' incident cases or prevalent cases? Are they always different people in each year? If someone became a case in 2006 and died in 2007, how would their data be entered in the above table? $\endgroup$ – onestop Feb 12 '11 at 9:52

In light of the clarifications in the comments to the question, I'd suggest using logistic regression with indicator variables for each region and a linear trend over year, and also consider adding an interaction between year and region to allow the linear trend to differ between regions.

If you want something simpler, you could apply Pearson's chi-squared test to the data for each year separately. It looks like there is very strong evidence that region 1 has higher death rates than either of the other two regions in all years except 2008, when region 3 had too few cases to be able to estimate the proportion of deaths with any precision.

Statistics can say nothing about the reason that region 1 differs from the others, of course.

  • $\begingroup$ Thank you for the detailed and very useful advice. If I use only Region 1 and Region 2, and apply Pearson's chi-squared test to the data for each year separately, would it be better to do a Meta-analysis also for these 2 regions? There is heterogeneity in the sample data and a Random model looked more appropriate and gave a significance at a p value of 0.000. Egger’s linear regression method intercept (B0) 10.34631, 95% confidence interval (1.05905, 19.63357), with t=3.54535, df=3. The 1-tailed p-value is 0.01911, and the 2-tailed p-value is 0.03822 (significant funnel asymmetry). $\endgroup$ – DrWho Feb 13 '11 at 13:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.