About this example
An early meta-analysis of small trials suggested that intravenous magnesium halved mortality after myocardial infarction. The ISIS-4 trial, with over 58,000 patients, then found no benefit. Egger, Davey Smith and Altman (2001) used these 16 trials to show how a funnel plot reveals the discrepancy: the small trials sit far to the left of the large one.
Here the fixed-effect summary is drawn with pseudo 95% confidence limits and Egger’s regression line. The line runs from the precise ISIS-4 estimate towards the imprecise small trials, and Egger’s test is highly significant. Bertschat 1989 has a zero cell and gets the usual 0.5 continuity correction, as in metafor.
The data
| Study | Events T | Total T | Events C | Total C |
|---|---|---|---|---|
| Morton 1984 | 1 | 40 | 2 | 36 |
| Rasmussen 1986 | 9 | 135 | 23 | 135 |
| Smith 1986 | 2 | 200 | 7 | 200 |
| Abraham 1987 | 1 | 48 | 1 | 46 |
| Feldstedt 1988 | 10 | 150 | 8 | 148 |
| Shechter 1989 | 1 | 59 | 9 | 56 |
| Ceremuzynski 1989 | 1 | 25 | 3 | 23 |
| Bertschat 1989 | 0 | 22 | 1 | 21 |
| Singh 1990 | 6 | 76 | 11 | 75 |
| Pereira 1990 | 1 | 27 | 7 | 27 |
| Shechter 1991 | 2 | 89 | 12 | 80 |
| Golf 1991 | 5 | 23 | 13 | 33 |
| Thogersen 1991 | 4 | 130 | 8 | 122 |
| LIMIT-2 1992 | 90 | 1159 | 118 | 1157 |
| Shechter 1995 | 4 | 107 | 17 | 108 |
…and 1 more row. Open the example to see all of them.
Open the example, replace the data with your own, and export the figure as SVG, PDF, PNG or TIFF.