Article type
Year
Abstract
Background: A meta-analysis was performed on the data from clinical trials comparing endoscopic and conventional cholecystectomy (EC and CC) by the length of hospital stay and the complication rate. The correlation between the quality of the trials and the validity of trial results was investigated.
Methods: Medline and our own archives with 60000 entries were searched for the terms 'endoscopic', 'open', 'conventional', 'laparoscopic', and 'cholecystectomy'. Reference lists were handsearched and experts were interviewed to identify all trials comparing EC and CC. The trials were classified by prospectively defined selection criteria according to design (randomised, concomitant, historic controls, etc.). The quality of each trial was assessed by 2 independent researchers. For the randomised trials (RTs) the Jadad-score (with a maximum of 5 points) as well as the Evans-Pollock-score for surgical studies (with a maximum of 100 points) were determined. If the two raters agreed, their judgement was used; if not a consensus decision was made. In a random-effects model the pooled odds ratio (OR) for the postoperative complication rate and the mean difference in length of hospital stay was calculated, and a sensitivity analysis was performed.
Results: The search with the keywords yielded 20,637 articles. Applying the selection criteria, 111 trials were suitable for meta-analysis, looking for double publication, etc. leaded to a final data pool of 69 trials. Of these 19 were randomised, 10 gave complication rates and 7 days in hospital. For the RTs the Jadad-score ranged from 0 - 3, with 8 trials ( 2 points, the Evans-Pollock-score ranged from 36 - 84, with 9 trials ( 63 points. Of the non-RTs 36 had a concomitant control group and 14 a historic one. There was only one double-blind study. OR =1 indicates equal risk, OR < 1 shows an advantage for the endoscopic method in the postoperative complication rate. The ORs with the 95% confidence intervals (CIs) are given in the table. The non randomised studies showed significant lower risk for endoscopic cholecystectomy with 95% CI excluding 1. RTs all inclu study design OR 95% CI non-RTs with historic controls 0.35 0.2 - 0.6 non-RTs with concomitant contr. 0.41 0.3 - 0.55 RTs 0.77 0.44 - 1.34 RTs with Evans Pollock ( 62 0.76 0.53 - 1.7 RTs with Jadad ( 2 1.01 0.54 - 1.9 ded 1 in the 95% CI. The same phenomenon was seen with hospital stay. The better the qualtity of the study, the smaller was the mean difference in hospital stay (historic controls: 3.55 days (2.96-4.14) and RTs with Evans-Pollock ( 63: 1.4 days (0.79-2.00)). But all CIs excluded 0 (no difference). The only double-blind trial showed no significant difference in complication rate as well as hospital stay.
Conclusions: The pooled difference of the two methods was the smaller the higher the quality of the studies was. If we assume that complication rate influences length of hospital stay, then the difference in hospital days seems artificial, because the high quality studies showed similar complication rates.
Methods: Medline and our own archives with 60000 entries were searched for the terms 'endoscopic', 'open', 'conventional', 'laparoscopic', and 'cholecystectomy'. Reference lists were handsearched and experts were interviewed to identify all trials comparing EC and CC. The trials were classified by prospectively defined selection criteria according to design (randomised, concomitant, historic controls, etc.). The quality of each trial was assessed by 2 independent researchers. For the randomised trials (RTs) the Jadad-score (with a maximum of 5 points) as well as the Evans-Pollock-score for surgical studies (with a maximum of 100 points) were determined. If the two raters agreed, their judgement was used; if not a consensus decision was made. In a random-effects model the pooled odds ratio (OR) for the postoperative complication rate and the mean difference in length of hospital stay was calculated, and a sensitivity analysis was performed.
Results: The search with the keywords yielded 20,637 articles. Applying the selection criteria, 111 trials were suitable for meta-analysis, looking for double publication, etc. leaded to a final data pool of 69 trials. Of these 19 were randomised, 10 gave complication rates and 7 days in hospital. For the RTs the Jadad-score ranged from 0 - 3, with 8 trials ( 2 points, the Evans-Pollock-score ranged from 36 - 84, with 9 trials ( 63 points. Of the non-RTs 36 had a concomitant control group and 14 a historic one. There was only one double-blind study. OR =1 indicates equal risk, OR < 1 shows an advantage for the endoscopic method in the postoperative complication rate. The ORs with the 95% confidence intervals (CIs) are given in the table. The non randomised studies showed significant lower risk for endoscopic cholecystectomy with 95% CI excluding 1. RTs all inclu study design OR 95% CI non-RTs with historic controls 0.35 0.2 - 0.6 non-RTs with concomitant contr. 0.41 0.3 - 0.55 RTs 0.77 0.44 - 1.34 RTs with Evans Pollock ( 62 0.76 0.53 - 1.7 RTs with Jadad ( 2 1.01 0.54 - 1.9 ded 1 in the 95% CI. The same phenomenon was seen with hospital stay. The better the qualtity of the study, the smaller was the mean difference in hospital stay (historic controls: 3.55 days (2.96-4.14) and RTs with Evans-Pollock ( 63: 1.4 days (0.79-2.00)). But all CIs excluded 0 (no difference). The only double-blind trial showed no significant difference in complication rate as well as hospital stay.
Conclusions: The pooled difference of the two methods was the smaller the higher the quality of the studies was. If we assume that complication rate influences length of hospital stay, then the difference in hospital days seems artificial, because the high quality studies showed similar complication rates.