Length of hospital stay
On the basis of the literature, 30 studies reporting post-operative length of hospital stay were included although 25 give useable data for statistical analysis12,13,15,16,18–23,25,27–29,31,36,38–43,45,46,48. A total of 25 studies with 3,958 patients and 8 pharmacological treatments are reported. The extracted data are detailed in Supplemental Table 8.1. The following treatments were studied and included in the network analysis: NSAIDs, gastrografin, colloid infusion, oral carbohydrates, prokinetics, probiotics, erythromycin and opioid antagonists. Figure 6 reports the network map for the 8 pharmacological treatment classes analysed. The main results are reported in Figure 6: network map, relative effect Bayesian plot, rankogram and the surface under the cumulative ranking curve (SUCRA).
Of the 8 treatments studied, prokinetics (Mean difference (MD) (hours) – 1.9; credible interval – 3.8, - 0.040 (Figure 6.B); SUCRA 0.34 (Figure 6.C)) showed a significantly faster onset of first stools compared to the control treatments. Gastrografin was the best treatment for the duration of flatus recovery with a probability of P=0.41. The ranking (rank1+rank2+rank3) in descending order of the top three treatments from the best to the third was as follows: gastrografin had a 73% (0.41+0.20+0.11) probability of being among the top three therapies followed by colloid infusion at 50% (0.21+0.17+0.11) and then prokinetics with 47% (0.05+0.17+0.25). The bias studies are summarised in Supplemental Figure 8.6. The overall bias was rated low risk in 92% of studies and of some concern in 8% of studies. The highest ratio of some concern was for deviations from intended interventions (24%).