advertisement
Meta-analysis of RCT's
January 2003, Rome, Italy
Luca Rossetti, Stefano Miglior and Nicola Orzalesi
At a recent expert meeting in Rome Luca Rossetti presented the following conclusions of a meta-analysis of seven studies on the effect of glaucoma treatment: Epstein et al. 1989, Kass et al. 1989, Schulzer et al. 1991, CNTG 1998, Heijl et al. 2000, OHTS 2002, EMGT 2002.
Meta-Analytical Conclusions:
- Pooled estimate of IOP reduction = - 4.5 mmHg (18.3%)
- Highly significant protective effect of treatment on VF and optic disc (OR = 0.51, 0.40-0.65)
- M-A allows for "giving more weight" to bigger RCTs: now we know that treatment works!• Important improvement of quality of RCTs
- Questions: need to treat many patients to have a small clinical effect? See Coleman,'Your Special Attention for'.
- 13% of treated got worse despite treatment
Note: Using the same longitudinal dataset, Katz et al. (Arch Ophthalmol 1999) have shown that different rates of VF progression can be obtained depending on the methods used for analysis.
Comment on Epstein et al.:
- "Positive results" trial (? Only in the alternative analysis, when intention-to-treat analysis considered, the difference did not reach statistical significance, p = 0.07)
- 23/107 (21.5%) dropouts
- No sample size calculation mentioned
- VF outcome: most manual, and only later changed to automatic
Comment on Kass et al.:
- "Positive results" trial (?)
- 19/62 (30.6%) patients excluded from the study
- No intention-to-treat analysis mentioned
- One-tailed test selected
- Study design (intra-patient eye randomization)
Comment on Schulzer et al.:
- "Negative" results trial
- 31/141 (22%) patients excluded from the study
- Intention-to-treat analysis mentioned
- No sample size calculation reported (the goal was to enroll 300 patients?)
- High risk population ? (31% with family history of POAG, 29% with a c/d > 0.5, 20% with a baseline IOP of 30 mmHg or more)
Comment on CNTG:
- "Positive" results trial when patients with cataract were removed from the study (!)
- "Negative" results trial when intention-to-treat analysis was applied
- 5/145 (3.4%) patients withdrawn
- Study power calculation reported
Comment on Heijl et al.:
- "Negative" results trial
- 43/90 (47.8%) patients withdrawn
- Intention-to-treat analysis mentioned
- No sample size calculation reported
Comment on OHTS:
- "Positive" results trial
- 310/1,636 (18.9%) patients died, inactive, nonadherence to randomization. 93/1,636 (5.7%) developed VF or optic disc changes due to other causes
- Intention-to-treat analysis reported
- Sample size calculation described
Comment on EMGT:
- "Positive" results trial
- 28/255 (10.9%) patients died or were lost to f-up
- Intention-to-treat analysis reported
- Sample size calculation described
- 52 % of patients had baseline IOP < 21 mmHg
