Table 2.
Results of reanalyses
| Study, treatment comparison; condition | Primary outcome measures | Effect size (type) | Effect size: published paper | P value | Effect size: reanalysis | P value |
|---|---|---|---|---|---|---|
| Luedtke 201527*: transcranial direct current stimulation (TDCs) versus sham TDCs; chronic low back pain | Visual analogue scale, time 1 | MD (99% CI) | 1 (−8.69 to 6.30) | 0.68 | 1.42 (−6.06 to 8.90) | 0.62 |
| Visual analogue scale, time 2 | MD (99% CI) | 3 (−10.32 to 6.73) | 0.58 | 5.19 (−3.62 to 14.00) | 0.13 | |
| Oswestry disability index, time 1 | MD (99% CI) | 1 (−1.73 to 1.98) | 0.86 | −0.12 (−1.99 to 1.75) | 0.87 | |
| Oswestry disability index, time 2 | MD (99% CI) | 0.01 (−2.45 to 2.62) | 0.92 | −0.49 (−3.15 to 2.18) | 0.64 | |
| Cohen 201520: epidural steroid injections versus gabapentin; lumbosacral radicular pain | Average leg pain, time 1 | MD (95% CI) | 0.4 (−0.3 to 1.2) | 0.25 | 0.4 (−0.3 to 1.2) | 0.26 |
| Average leg pain, time 2 | MD (95% CI) | 0.3 (−0.5 to 1.2) | 0.43 | 0.3 (−0.5 to 1.2) | 0.45 | |
| Hyttel-Sorensen 201524: near infrared spectroscopy versus usual care; preterm infants | Burden of hypoxia and hyperoxia | RC% (95% CI) | −58 (−35 to −74) | <0.001 | −58 (−36 to −72) | <0.001 |
| Bousema 201618†: hotspot targeted interventions versus usual interventions; malaria | nPCR parasite prevalence, time 1, zone 1 | MD (95% CI) | 10.2 (−1.3 to 21.7) | 0.075; 0.024 | 10.2 (−2.5 to 22.9) | 0.095; 0.047 |
| nPCR parasite prevalence, time 2, zone 1 | MD (95% CI) | 4.2 (−5.1 to 13.8) | 0.326; 0.265 | 4.3 (−5.3 to 13.9) | 0.328; 0.157 | |
| nPCR parasite prevalence, time 1, zone 2 | MD (95% CI) | 3.6 (−2.6 to 9.7) | 0.219; 0.216 | 3.6 (−3.3 to 0.1) | 0.239; 0.339 | |
| nPCR parasite prevalence, time 2, zone 2 | MD (95% CI) | 1.0 (−7.0 to 9.1) | 0.775; 0.713 | 1.0 (−7.6 to 9.7) | 0.778; 0.858 | |
| nPCR parasite prevalence, time 1, zone 3 | MD (95% CI) | 3.8 (−2.4 to 10.0) | 0.199; 0.187 | 4.0 (−2.1 to 10.2) | 0.170; 0.253 | |
| nPCR parasite prevalence, time 2, zone 3 | MD (95% CI) | 1.0 (−8.3 to 10.4) | 0.804; 0.809 | 1.0 (−8.4 to 10.5) | 0.805; 0.814 | |
| Gysin-Maillart 201621‡: brief therapy versus usual care; suicide attempt | Suicide attempt | HR (95% CI) | 0.17 (0.07 to 0.46) | <0.001 | 0.17 (0.06 to 0.49) | <0.001 |
| Lombard 201626: low intensity intervention versus general health intervention; obesity | Change in weight (unadjusted) | MD (95% CI) | −0.92 (−1.67 to −0.16) | . | −0.92 (−1.66 to −0.18) | . |
| Change in weight (adjusted) | MD (95% CI) | −0.87 (−1.62 to −0.13) | . | −0.89 (−1.63 to −0.15) | . | |
| Polyak 201630: cotrimoxazole prophylaxis cessation versus continuation; HIV | Malaria, pneumonia, diarrhea, or mortality (ITT) | IRR (95% CI) | 2.27 (1.52 to 3.38) | <0.001 | 2.27 (1.51 to 3.39) | <0.001 |
| Malaria, pneumonia, diarrhea, or mortality (PP) | IRR (95% CI) | 2.27 (1.52 to 3.39) | <0.001 | 2.27 (1.52 to 3.40) | <0.001 | |
| Robinson 201531: blood stage plus liver stage drugs or blood stage drugs only; malaria | Plasmodium vivax infection by qPCR, time 1 | HR (95% CI) | 0.18 (0.14 to 0.25) | <0.001 | 0.17 (0.13 to 0.24) | <0.001 |
| P vivax infection by qPCR, time 2 | HR (95% CI) | 0.17 (0.12 to 0.24) | <0.001 | 0.16 (0.11 to 0.22) | <0.001 | |
| Harris 201523: nurse delivered walking intervention versus usual care; older adults | Daily step count | MD (95% CI) | 1037 (513 to 1560) | <0.001 | 1037 (513 to 1560) | <0.001 |
| Adrion 201616: low dose betahistine versus placebo; Meniere’s disease | Vertigo attack | RaR (95% CI) | 1.036 (0.942 to 1.140) | 0.759 | 1.033 (0.942 to 1.133) | 0.777 |
| Vertigo attack | RaR (95% CI) | 1.012 (0.919 to 1.114) | 0.777 | 1.011 (0.921 to 1.109) | 0.777 | |
| Selak 201432: fixed dose combination treatment versus usual care; patients at high risk of cardiovascular disease in primary care | Use of antiplatelet, statin, ≥2 blood pressure lowering drugs | RR (95% CI) | 1.75 (1.52 to 2.03) | <0.001 | 1.75 (1.52 to 2.03) | <0.001 |
| Change in systolic blood pressure | MD (95% CI) | −2.2 (−5.6 to 1.2) | 0.21 | −2.2 (−5.6 to 1.2) | 0.21 | |
| Change in diastolic blood pressure | MD (95% CI) | −1.2 (−3.2 to 0.8) | 0.22 | −1.2 (−3.2 to 0.8) | 0.22 | |
| Change in low density lipoprotein cholesterol level | MD (95% CI) | −0.05 (−0.17 to 0.08) | 0.46 | −0.05 (−0.17 to 0.08) | 0.46 | |
| Chesterton 201319: transcutaneous electrical nerve stimulation versus usual care; tennis elbow | Change in elbow pain intensity | MD (95% CI) | −0.33 (−0.96 to 0.31) | 0.31 | −0.33 (−1.01 to 0.34) | 0.33 |
| Atukunda 201417: sublingual misoprostol versus intramuscular oxytocin; prevention of postpartum haemorrhage | Blood loss >500 mL | RR (95% CI) | 1.64 (1.32 to 2.05) | <0.001 | 1.65 (1.32 to 2.05) | <0.001 |
| Paul 201529: trimethoprim-sulfamethoxazole versus vancomycin; severe infections caused by meticillin resistant Staphylococcus aureus | Treatment failure (ITT) | RR (95% CI) | 1.38 (0.96 to 1.99) | 0.08 | 1.38 (0.96 to 1.99) | 0.08 |
| Treatment failure (PP) | RR (95% CI) | 1.24 (0.82 to 1.89) | 0.36 | 1.24 (0.82 to 1.89) | 0.36 | |
| All cause mortality (ITT) | RR (95% CI) | 1.27 (0.65 to 2.45) | 0.57 | 1.27 (0.65 to 2.45) | 0.57 | |
| All cause mortality (PP) | RR (95% CI) | 1.05 (0.47 to 2.32) | 1 | 1.05 (0.47 to 2.32) | 1 | |
| Omerod 201528: ciclosporin versus prednisolone; pyoderma gangrenosum | Speed of healing over six weeks (cm2/day) | MD (95% CI) | 0.003 (−0.20 to 0.21) | 0.97 | 0.003 (−0.20 to 0.21) | 0.97 |
| Hanson 201522: home based counselling strategy versus usual care; neonatal care | Neonatal mortality | OR (95%) CI | 1.0 (0.9 to 1.2) | 0.339 | 1.0 (0.9 to 1.2) | 0.339 |
nPCR=nested polymerase chain reaction; qPCR=quantitative real time PCR; MD=mean differences; RC%=relative change in percentage; HR=hazard Ratio; RaR=rate ratio; RR=relative risk; OR=odds ratio; ITT=intention to treat; PP=per protocol.
*
Study judged as not fully reproduced but reaching same conclusion because of differences in numerical estimates: for one of these analyses current authors found a difference in number of patients analyzed (one patient more in reanalysis).
†
P values of adjusted analyses were computed to complement non-adjusted analyses.
‡
Study judged as not fully reproduced but reaching same conclusion because initial analysis did not take into account repeated nature of data. This was corrected in reanalysis without major differences in estimates.