11.1 Supplemental Antioxidant Nutrients: Combined Vitamins and Trace Elements April 2013

Transcription

1 . Supplemental Antioxidant Nutrients: Combined Vitamins and Trace Elements April Recommendation: Based on 7 level and 7 level 2 studies, the use of supplemental combined vitamins and trace elements should be considered in critically ill patients. 23 Discussion: The committee noted that with the addition of new trials (Lindner 24, El Attar 2, González 2, Andrews 2, Manzanares 2, Valenta 2, Schneider 2 and Heyland 23), there was a moderate treatment effect but narrow confidence intervals with respect to a reduction in mortality, infections and a trend towards a reduction in mechanical ventilation similar to a recent systematic review (). The committee noted that the large REDOXS trial was negative but that the signal of benefit persisted despite its inclusion in the meta-analysis. They considered that the dose of antioxidants in the REDOXS trial may have been insufficient and there is still uncertainty about the optimal composition and dose of supplemental vitamins and trace elements. Concern was expressed about the differences in the types of antioxidant nutrients used in the studies and the heterogeneity of the trials but the high generalizability of the results from many large, multicentre trials was also noted. There were no concerns about the safety, feasibility and cost of these nutrients. The committee therefore agreed to continue with a recommendation that supplemental combined vitamins and trace elements should be considered. () Manzanares W, Dhaliwal R, Jiang X, Murch L, Heyland DK. Antioxidant micronutrients in the critically ill: a systematic review and meta-analysis. Crit Care. 22 Dec 2;(2):R 2 Recommendation: Based on 3 level and 3 level 2 studies, the use of supplemental combined vitamins and trace elements should be considered in critically ill patients. Discussion: The committee noted the strong treatment effect and narrow confidence intervals with respect to a reduction in mortality. Even with the exclusion of one small study that had poor methodological quality (Kuklinksi ), the reduction in mortality remained. The committee expressed concern about the differences in the types of antioxidant nutrients used in the studies and the heterogeneity of the trials. Despite the optimal composition and dose of supplemental vitamins and trace elements not being well established, there were no concerns about the safety, feasibility and cost of these nutrients. The committee therefore agreed to make a recommendation that supplemental combined vitamins and trace elements should be considered. These nutrients are currently being investigated and we await the results of ongoing studies to strengthen the clinical recommendations.

2 Semi Quantitative Scoring Value Definition 2 Score (,,2,3) Magnitude of the absolute risk reduction attributable to the intervention listed--a higher score indicates a Effect size larger effect size 2 Confidence interval Validity % confidence interval around the point estimate of the absolute risk reduction, or the pooled estimate (if more than one trial)--a higher score indicates a smaller confidence interval Refers to internal validity of the study (or studies) as measured by the presence of concealed randomization, blinded outcome adjudication, an intention to treat analysis, and an explicit definition of outcomes--a higher score indicates presence of more of these features in the trials appraised 3 (mortality) 2 (infections) 23 Score (,,2,3) (mortality) (infections) 3 (mortality) 3 (infections) 2 3 Homogeneity or Reproducibility Adequacy of control group Biological plausibility Similar direction of findings among trials--a higher score indicates greater similarity of direction of findings among trials 2 Extent to which the control group represented standard of care (large dissimilarities =, minor dissimilarities=2, usual care=3) 3 3 Consistent with understanding of mechanistic and previous clinical work (large inconsistencies =, minimal inconsistencies =2, very consistent =3) 2 2 Generalizability Likelihood of trial findings being replicated in other settings (low likelihood i.e. single centre =, moderate likelihood i.e. multicentre with limited patient population or practice setting =2, high likelihood i.e. multicentre, heterogenous patients, diverse practice settings = Low cost Feasible Safety Estimated cost of implementing the intervention listed--a higher score indicates a lower cost to implement the intervention in an average 2 2 Ease of implementing the intervention listed--a higher score indicates greater ease of implementing the intervention in an average 2 2 Estimated probability of avoiding any significant harm that may be associated with the intervention listed--a higher score indicates a lower probability of harm 2 2 2

3 . Supplemental Antioxidant Nutrients: Combined Vitamins and Trace Elements April 23 Question: Does the addition of Supplemental Combined Vitamins and Trace Elements result in improved outcomes in the critically ill patient? Summary of evidence: Of the 24 studies included, there were seven level and seventeen level 2 studies reviewed that compared various antioxidants either as single nutrients (zinc, selenium) or as a combination of nutrients (selenium, copper, zinc, vit. A, C & E, N-acetylcysteine) given by various routes (IV/parenteral, enteral, combined parenteral and enteral). One study was published in 2 parts (Berger et al, Intensive Care Medicine 2;27:- and Berger et al, Nutrition Research (2):4-4) and the data listed here represent the data from the latter study (intent to treat). This study had two intervention arms i.e. selenium alone and selenium combined with zinc and α tocopherol compared to placebo and the data are presented in the meta-analysis as Berger 2a and Berger 2b respectively. Mortality: Twenty-three studies reported on mortality and when the results were aggregated, antioxidant supplementation was associated with a significant reduction in overall mortality (RR., % CI.7,.., p=.3, heterogeneity I 2 =2%; figure ). Linder (24) was excluded from the meta-analyses because the type of mortality was not specified and appeared to be days. When the studies which delivered antioxidants via parental nutrition were sub-grouped and analysed, antioxidant supplementation was associated with a significant reduction in overall mortality as well (RR., % CI.74,., p=.4, heterogeneity I 2 =%; figure ). Similarly, when the 4 studies which delivered antioxidants via enteral nutrition were sub-grouped and analysed, antioxidant supplementation was associated with a significant reduction in overall mortality (RR., % CI.4,., p=., heterogeneity I 2 =%; figure ). However, when the data from the subgroup comprised of the 3 studies which delivered antioxidants via both enteral and parental nutrition were aggregated, antioxidant supplementation had no effect on overall mortality (RR.7, % CI.2,.2, p=.3, heterogeneity I 2 =%; figure ). The test for subgroup differences was significant (p=.4). Mortality (higher vs. lower mortality in control group): Subgroup analysis showed that antioxidant supplementation was associated with a significant reduction in overall mortality among patients with higher risk of death (>% mortality in the control group) (RR.3, % CI.7,.7, p=.2, heterogeneity I 2 =3%; figure 2). There was no significant effect observed for trials of patients with a lower mortality in the control group (RR.4, % CI.7,., p=., heterogeneity I 2 =%; figure 2). The test for subgroup differences was not significant (p=.2). Infections: When the studies that reported on infectious complications were aggregated, antioxidant supplementation was associated with a significant reduction in overall infections (RR., % CI.7,., p=.4, heterogeneity I 2 =%; figure 3). When a subgroup analysis based on studies which delivered antioxidants via parental nutrition was done, antioxidant supplementation was associated with a trend towards a reduction in infectious complications (RR., % CI.72,.3, p=., heterogeneity I 2 =%; figure 3). When a subgroup analysis based on 3 studies which delivered antioxidants via enteral nutrition was done, antioxidant supplementation had no effect on infectious complications (RR., % CI., 2.4, p=.7, heterogeneity I 2 =3%; figure 3). When a third subgroup analysis based on 3 studies which delivered antioxidants via both enteral and 3

4 parental nutrition was done, antioxidant supplementation was associated with a trend towards a reduction in infectious complications (RR., % CI.77,., p=., heterogeneity I 2 =%; figure 3). The test for subgroup differences was not significant (p=.72). Infections (higher vs. lower mortality in control group): Subgroup analysis showed that antioxidant supplementation was associated with a significant reduction in infectious complications among patients with higher risk of death (>% mortality in the control group) (RR., % CI.77,., p=., heterogeneity I 2 =%; figure 4). There was no significant effect observed for patients in trials with a lower mortality in the control group (RR.7, % CI.,., p=.2, heterogeneity I 2 =%; figure 4). The Maderazo study was not included in the analysis since it does not report on mortality. The test for subgroup differences was not significant (p=.). length of stay: When the studies that reported length of stay as a mean ± standard deviation were aggregated, antioxidant supplementation had no effect on length of stay (WMD.3, % CI -.,., p=.33, heterogeneity I 2 =%; figure ). The result was the same for each of the 3 subgroups: six studies which delivered antioxidants via parental nutrition (WMD., % CI -2.47, 2.2, p=., heterogeneity I 2 =2%; figure ), one study which delivered antioxidants via enteral nutrition (WMD 3.3, % CI -.,., p=.; figure ), and three studies which delivered antioxidants via both enteral and parental nutrition (WMD.3, % CI -.7,.7, p=., heterogeneity I 2 =%; figure ). The test for subgroup differences was not significant (p=.7). length of stay: When the studies that reported hospital length of stay as a mean ± standard deviation were aggregated, antioxidant supplementation had no effect on hospital length of stay (WMD -., % CI -4.7, 2.4, p=.3, heterogeneity I 2 =%; figure ). The result was the same for 2 of the subgroups: two studies which delivered antioxidants via parental nutrition (WMD -.3, % CI -2., 3., p=., heterogeneity I 2 =%; figure ), and one study which delivered antioxidants via enteral nutrition (WMD -2., % CI -24.,.2, p=.; figure ). However, in the subgroup of 3 studies in which antioxidants were delivered via both enteral and parental nutrition, antioxidant supplementation was associated with a trend towards a reduction in hospital length of stay (WMD -.4, % CI -., 4., p=.2, heterogeneity I 2 =3%; figure ). The test for subgroup differences was not significant (p=.). Duration of mechanical ventilation: When the studies that reported duration of ventilation as a mean ± standard deviation were aggregated, antioxidant supplementation was associated with a trend towards a reduction in duration of ventilation (WMD -.7, % CI -3.7,.3, p=., heterogeneity I 2 =74%; figure 7). Subgroup analysis showed that antioxidant supplementation had no effect on duration of ventilation in the subgroup of studies in which antioxidants were delivered via parental nutrition (WMD -2.22, % CI -.7,.2, p=.2, heterogeneity I 2 =7%; figure 7), nor in the subgroup consisting of study in which antioxidants were delivered via both enteral and parental nutrition (WMD.4, % CI -., 2.7, p=.73; figure 7). However, in the subgroup of the 2 studies where antioxidants were delivered via enteral nutrition, antioxidant supplementation was associated with a significant reduction in duration of ventilation (WMD -2., % CI -4., -.4, p=., heterogeneity I 2 =3%; figure 7). There was a trend towards a difference between the subgroups (p=.). 4

5 Conclusions: ) Antioxidant nutrients are associated with a significant reduction in overall mortality in critically ill patients. 2) Antioxidant nutrients are associated with a significant reduction in overall infectious complications in critically ill patients. 3) Antioxidant nutrients have no effect on length of stay in critically ill patients. 4) Antioxidant nutrients have no effect on hospital length of stay in critically ill patients. ) Antioxidant nutrients are associated with a trend towards a reduction in duration of ventilation in critically ill patients. Level study: if all of the following are fulfilled: concealed randomization, blinded outcome adjudication and an intention to treat analysis. Level 2 study: If any one of the above characteristics are unfulfilled.

6 Table. Randomized Studies Evaluating Supplemental Combined Vitamins And Trace Elements in Critically Ill Patients Study Population Methods Score Intervention Studies in which antioxidants were delivered via PN ) Kuklinski Patients with acute pancreatic necrosis N=7 C. Random: not sure ITT: no Blinding: no (4) PN + selenium supplementation ( µg /d) vs. PN without selenium supplementation 2) Young Severely head injured patients, ventilated N= C. Random: yes Blinding: double (7) 2 mg elemental zinc via PN, then progressing to oral zinc from - days vs. 2. mg elemental zinc, then progressing to oral placebo 3) Zimmerman 7 Patients with SIRS, APACHE > and multiorgan failure score > N=4 C. Random: no Blinding: no () µg Na-Selenite as a bolus IV then µg Na-Selenite/24 hrs as a continuous infusion over 2 days vs. standard 4) Berger Burns > 3 % TBSA N=2 C. Random: yes Blinding: double blind (2) IV Copper (4.4 µmol), selenium ( µg), zinc (4 µmol) + standard trace elements vs. standard trace elements (Copper 2 µmol, selenium 32 µg, zinc µmol) from day -, all received early EN ) Angstwurm Patients with systematic inflammatory response syndrome from s N=42 C. Random: not sure Blinding: no () PN with high dose selenium (3 µg x 3 days, 2 µg x 3 days and µg x 3 days and 3 µg thereafter) vs. low dose selenium (3 µg/day for duration of study) ) Berger 2 Trauma patients, surgical N=32 C. Random: yes ITT: no Blinding: double blind () IV Selenium supplementation ( µg/day ) vs. placebo (Selenium group randomized further to two groups: µg Selenium alone vs. µg Selenium + mg α tocopherol + 3 mg zinc) given slowly for st days after injury (All groups received EN) 7) Lindner 24 Patients with acute pancreatitis admitted to the N=7 C. Random: not sure ITT: no Blinding: single () IV sodium selenite dose of 2 µg on day, µg on days 2-, and 3 µg from day until discharge vs placebo (isotonic.% IV NaCl solution).

7 ) Angstwurm 27 Multicentre mixed s N=24 C.Random: not sure ITT: no Blinding: double () µg Selenium IV within hr followed by µg Selenium for 4 days vs. NaCl (.%) (all patients received EN or PN) ) Berger 27 Burns > 2 % TBSA N=2 C.Random: not sure Blinding: no () IV ml of Copper ( µmol) + Selenium (37 µgm + zinc (74 µmol) vs. NaCl (.%) from admission for - days. Both groups were on EN. ) Forceville 27 Septic shock patients from 7 s N= C.Random: not sure ITT: no Blinding: double () 4µg Selenium IV on day followed by µg Selenium for days vs. NaCl (.%) (all patients received EN or PN) ) Mishra 27 Septic patients N=4 C.Random: not sure Blinding: double () 474 µg Selenium IV x 3 days followed by 3 µg x 3 days, µg x 3 days and 3. µg thereafter vs. 3. µg Selenium (all patients received EN or PN). 2) El-Attar 2 COPD patients N= C.Random: yes Blinding: yes (2) IV selenium as sodium selenite µg/day, zinc 2 mg/day and manganese.4 mg/day vs. none. TE were administered during the period on mechanical ventilation 3) González 2 Medical/surgical pts N= C.Random: yes Blinding: double (7) day sodium selenite µg, day 2 sodium selenite µg and thereafter 2 µg during seven additional days vs selenite µg/d 4) Andrews 2 Mixed N=2 C. Random: yes Blinding: double (3) µg selenium supplemented PN (2.g nitrogen, 2kcal) vs. standard PN (2.g nitrogen, 2kcal) initiated after admission (actual median 2. days) for 7 days (actual duration, mean 4. days). ) Manzanares 2 Septic or trauma patients N=3 C. Random: not sure ITT: no (except mortality) Blinding: single () IV Selenium supplementation loading dose 2 µg (2 hours) on day followed by µg/day for days vs. NaCl as placebo 7

8 ) Valenta 2 Patients with sepsis or SIRS N= C. Random: not sure Blinding: no () IV Selenium supplementation loading dose µg on day followed by µg/day for -4 days + <7µg/day of Na-selenite added to PN. vs. NaCl + <7µg/day of Na-selenite added to PN. Studies in which antioxidants were delivered via EN 7) Maderazo Blunt Trauma N=4 C. Random: yes Blinding: double (7) 2 mg Ascorbic acid, then mg + mg α tocopherol in ml of DW vs. ml of DW (Experimental group divided into 2 groups, 2 mg ascorbic acid vs. mg α tocopherol).given as 2 hr infusions from Day -7. (All groups received enteral nutrition or po intake) ) Preiser 2 Mixed N= C. Random: not sure ITT: no Blinding: single (7) Antioxidant rich formula via EN (33 µg / ml vit. A, 3 mg/ ml Vit C & 4. mg/ ml Vit E) vs. isonitrogenous, isocaloric standard formula (7 µg / ml vit. A, mg/ ml Vit C and. mg/ ml Vit E) from Day - 7 ) Nathens 22 General Surgical/Trauma N=77 C.Random: not sure ITT: no Blinding: no (7) α tocopherol IU q h via naso or orogastric tube and ascorbic acid mg q h via IV vs. standard care 2) Crimi 24 Mixed N=224 C.Random: not sure ITT: no Blinding: no (7) Vit C ( mg), Vit E (4 IU) within 72 hrs for days vs. isotonic saline (all groups received EN) 2) Schneider 2 patients with sepsis or SIRS N= C.Random: not sure Blinding: single blind () Fresenius Kabi Intestamin (3µg selenium, zinc 2mg, vitamin C mg, Vitamin E mg) vs. Fresubin original plus 2mL water delivered via duodenal tube and initiated within first 4h of admission. Both groups received Fresenius Kabi original fiber and supplemental PN if <% adequacy Studies in which antioxidants were delivered simultaneously via PN and EN 22) Porter Surgical Penetrating trauma patients with injury severity score 2 N= C. Random: yes Blinding: no () µg selenium IV q hrs + 4 IU Vit E, mg Vit. C q hrs and g of N-acetylcysteine (NAC) q hrs via nasogastric or oral route, from Day -7 vs. none

9 23) Berger 2 Mixed N=2 C.Random: not sure Blinding: no () IV Selenium supplementation loading dose 4 µg/day + zinc ( mg) + Vit C 27 mg + Vit B 3 mg + Vit E enteral mg + Vit E 2. mg IV for 2 days followed by half the dose of all vs. standard vitamins. (All groups received EN or PN) 24) Heyland 23 Multicentre mixed s N=2 DW: dextrose % in water TBSA: total body surface area C.Random: yes Blinding: double (2) µg selenium via PN + 3 µg selenium, 2 mg zinc, mg beta carotene, mg vitamin E, mg vitamin C via EN vs. placebo via PN and EN Table. Randomized Studies Evaluating Combined Vitamins And Trace Elements in Critically Ill Patients (continued) Study Mortality Experimental Control Studies in which antioxidants were delivered via PN Infections Experimental Control LOS Experimental Control Ventilator Days Experimental Control ) Kuklinski / () / () NR NR NR NR NR NR 2) Young 4/33 (2) /3 (2) NR NR NR NR NR NR 3) Zimmerman 7 3/2 () /2 (4) NR NR NR NR NR NR 4) Berger / () / (). ±. (-4) per patient 3. ±. (2-) per patient 3 ± 2 () 4 ± 27 () 3 ± 3 () ± 3 () ± () 2 ± () ) Angstwurm 7/2 (33) /2 (2) NR NR NR NR (3-23) (-43)

10 ) Berger 2 (a) Se alone 2/ (22) (b) Se+AT+Zn / () / () (a) Se alone / () (b) Se+AT+Zn 3/ (27) /2 (42) (a) Se alone. ± 4. () 2 ± 7 (). ±. () 4 ± 3 () (a) Se alone.2 ± 3. () (b) Se+AT+Zn 4. ± 3. () 4.2 ±.2 () (b) Se+AT+Zn. ± 4.4 () ± 4 () 7) Linder 24 Not specified /32 (.) Not specified 3/3 (.) NA NA 24 (-44) 2 (-4) NA NA ) Angstwurm 27 2-day 4/ (4) 2-day /22 () HAP / () HAP /22 (). ± () 2.7 ± (22) NR NR ) Berger 27 / () / () 2. ±. per pt 3. ±.3 per pt 3 ± 27 () 47 ± 37 () 7. ± () 2. ± () ) Forceville 27 2-day 4/3 (4) -month /3 () -year % 2-day 3/2 (4) -month 2/2 () -year 7% Superinfection /3 (3) Superinfection 2/2 (7) 2 (7-4) 2 (7-) (-3) 33 (-) (7-34) 4 (-23) ) Mishra 27 / (44) / () 2-day / (44) /22 () /22 () 2-day /22 (). ±. per patient. ±. per patient 2.3 ±.2 () 2. ± 2. () NR NR 2) El-Attar 2 2/4 () /4 (3) VAP /3 (4) VAP 7/34 (2) NR NR.4 ± 7.3 (4) 7. ± 7. (4) 3) González 2 /34 () /34 (24) NR NR 2(2-4) 7(4-2) (7-2) 3 (-4)

11 4) Andrews 2 4/2 (33) -month 7/2 (43) 4/2 (33) -month 4/2 (4) Confirmed 4/2 (4) Confirmed 2/2 (4) 3.2 (IQR 7., 23.7) 2. (IQR 4.7, 2.4). (IQR.3, 2.4) 3.2 (IQR.-7.) NR NR ) Manzanares 2 3/ (2) / (33) / (3) 7/ (44) VAP 3/ (2) VAP 7/ (44) 4 ± () 3 ± () ± () ± 4 () ) Valenta 2 2-day /7 (2) 2-day 24/7 (32) NR NR NR NR NR NR Studies in which antioxidants were delivered via EN 7) Maderazo NR NR 3/2 (4) / (2) NR NR NR NR ) Preiser 2 3/2 () /2 (4) 3/7 () /7 (3) 3/2 () /7 () (3-2) (3-) NR NR ) Nathens 22 3/3 () /3(2) 2-day 4/3 () /24 (3) /24(3) 2-day 7/24 (2) 3/3 (2) 44/24 ().3 (mean) 4. (mean).4 (mean). (mean) 3.7 (mean) 4. (mean) 2) Crimi 24 2-day 4/2 (44) 2-day 7/2 () NR NR 2. (mean) 27. (mean).2 ± 2.3 (2). ±. (2) 2) Schneider 2 /2 (2) /2 (2) From day 3/2 () From day /24 (3) 2. ± 2 (2) 44.4 ± 3. (2) 2. ±. (2) 47.2 ± 4. (2) 3. ±.2 (2) 27.2 ±. () Studies in which antioxidants were delivered simultaneously via PN and EN

12 22) Porter / / / () / () 22 ± 2.2 () 3.3 ± 23.4 () 3. ± 2. () 4 ± 3 () NR NR 23) Berger 2 /2 () 4/2 (4) 3-month 4/2 (4) / () / () 3-month / () 3/2 (3) 34/ (3). ±.4 (2) 23 ± 2 (2).4 ±.7 () 2 ± 2 () Vent-free days 2. ±.7 Vent-free days 2. ±.2 24) Heyland 23 2/7 (3) 4-day 4/7 (2) 2-day /7 (3) 3-month 23 (3) -month 2 (4) / (33) 4-day 32/ (22) 2-day 73/ (2) 3-month 222 (3) -month 23(4) All /7 (27) VAP 7/7 (2) All / (3) VAP / () 4.2 ± 22.7 (7) 3.2 ±.2 (7) 3. ± 23. () 2. ± 44. (). ± 2.4 (7). ±.7 () 2

13 Figure. Overall Mortality (with sub-analyses according to routes of administration) Study or Subgroup.. AOX via PN Kuklinski Young Zimmerman Berger Angstwurm Berger 2b Berger 2a Angstwurm 27 Mishra Forceville Berger 27 El-Attar González Andrews Valenta Manzanares Subtotal (% CI) AOX Control Risk Ratio Risk Ratio Events Total Events Total Weight M-H, Random, % CI Year M-H, Random, % CI %.%.3%.2% 3.%.2%.4% 2.7%.% 4.%.3%.3%.% 4.%.% 2.%.7% 2 22 Heterogeneity: Tau² =.; Chi² = 2., df = (P =.); I² = % Test for overall effect: Z = 2. (P =.4).7 [.,.].47 [.,.3].3 [.2,.2] 3. [.4,.].4 [.3,.32].33 [.2, 7.3] 2.44 [.2, 22.].7 [.,.]. [.,.43]. [.,.7]. [.7, 2.] 2. [., 2.].7 [.2,.3]. [.7,.2].7 [.4,.32].7 [.3,.]. [.74,.] AOX via EN Preiser Nathens Crimi Schneider Subtotal (% CI) %.% 4.%.7% 2.4% Heterogeneity: Tau² =.; Chi² = 2.3, df = 3 (P =.); I² = % Test for overall effect: Z = 3.3 (P =.)..3 AOX via PN & EN Porter Berger 2 Heyland Subtotal (% CI) % 2.3% 23.% Heterogeneity: Tau² =.; Chi² =.7, df = (P =.4); I² = % Test for overall effect: Z =. (P =.3).3 [.4, 2.2].4 [.,.].4 [.,.2]. [.37, 2.74]. [.4,.] Not estimable.4 [., 3.2]. [.,.24].7 [.2,.2] Total (% CI) 3 7.%. [.7,.] 3 7 Heterogeneity: Tau² =.2; Chi² = 2.22, df = 2 (P =.2); I² = 2% Test for overall effect: Z = 2. (P =.3) Test for subgroup differences: Chi² =.2, df = 2 (P =.4), I² = 2.2% Favours AOX Favours control 3

14 Figure 2: Mortality (with sub-analyses according to high (>%) or low mortality in the control group) Study or Subgroup.2. High mortality Kuklinski Young Zimmerman Angstwurm Crimi Forceville Angstwurm 27 Mishra González Andrews Manzanares Valenta Schneider Heyland Subtotal (% CI) AOX Control Risk Ratio Risk Ratio Events Total Events Total Weight M-H, Random, % CI Year M-H, Random, % CI %.%.3% 3.% 4.% 4.% 2.7%.%.% 4.% 2.%.%.7% 2.3% 2.3% 47 2 Heterogeneity: Tau² =.3; Chi² = 2.24, df = 3 (P =.7); I² = 3% Test for overall effect: Z = 2.3 (P =.2).7 [.,.].47 [.,.3].3 [.2,.2].4 [.3,.32].4 [.,.2]. [.,.7].7 [.,.]. [.,.43].7 [.2,.3]. [.7,.2].7 [.3,.].7 [.4,.32]. [.37, 2.74]. [.,.24].3 [.7,.7] Low mortality Berger Porter Preiser Berger 2b Berger 2a Nathens Berger 27 Berger 2 El-Attar Subtotal (% CI) % 2.4%.2%.4%.%.3% 2.%.3% 7.7% Heterogeneity: Tau² =.; Chi² = 3.4, df = 7 (P =.7); I² = % Test for overall effect: Z =.4 (P =.) Total (% CI) 3 7.% 3 7 Heterogeneity: Tau² =.2; Chi² = 2.22, df = 2 (P =.2); I² = 2% Test for overall effect: Z = 2. (P =.3) Test for subgroup differences: Chi² =., df = (P =.2), I² = 3.% 3. [.4,.] Not estimable.3 [.4, 2.2].33 [.2, 7.3] 2.44 [.2, 22.].4 [.,.]. [.7, 2.].4 [., 3.2] 2. [., 2.].4 [.7,.]. [.7,.] Favours AOX Favours control 4

15 Figure 3. Infections (with sub-analyses according to routes of administration)

16 Figure 4. Infections (with sub-analyses according to high (>%) or low mortality in the control group) Study or Subgroup.4. High mortality Angstwurm 27 Andrews Manzanares Heyland Subtotal (% CI) AOX Control Risk Ratio Risk Ratio Events Total Events Total Weight M-H, Random, % CI Year M-H, Random, % CI % 33.3%.% 4.3% 7.4% 2 3 Heterogeneity: Tau² =.; Chi² =., df = 3 (P =.7); I² = % Test for overall effect: Z =.4 (P =.). [.4, 2.43]. [.7,.4].4 [.4,.4]. [.7,.]. [.77,.] Low mortality Porter Preiser Berger 2a Berger 2b Nathens Berger 2 El-Attar Subtotal (% CI) %.3%.%.% 7.%.%.2% 23.% Heterogeneity: Tau² =.; Chi² = 4., df = (P =.); I² = % Test for overall effect: Z =.4 (P =.2).3 [.33,.7] 2. [.2, 22.3].33 [., 3.24]. [.2, 2.2]. [.3,.2].2 [.7,.4].7 [.24,.2].7 [.,.] Mortality not reported Maderazo 3 Subtotal (% CI) Heterogeneity: Not applicable Test for overall effect: Not applicable 2.%.7 [.72, 3.] Not estimable Total (% CI) 47 4.% Heterogeneity: Tau² =.; Chi² =.73, df = (P =.4); I² = % Test for overall effect: Z = 2.2 (P =.2) Test for subgroup differences: Chi² =., df = (P =.), I² = %. [.7,.] Favours AOX Favours control

17 Figure. LOS Study or Subgroup.. AOX via PN Berger Berger 2b Berger 2a Angstwurm 27 Berger 27 Mishra Manzanares Subtotal (% CI) AOX Control Mean Difference Mean Difference Mean SD Total Mean SD Total Weight IV, Random, % CI Year IV, Random, % CI Heterogeneity: Tau² = 2.3; Chi² = 7.4, df = (P =.2); I² = 2% Test for overall effect: Z =. (P =.).% 3.% 3.%.%.%.7% 2.% 32.4% -. [-.7,.7] -2. [-.2, 2.] -. [-., 4.] 2.4 [-.2, 4.2] -2. [-3.4,.4]. [-2., 3.]. [-.3, 7.3]. [-2.47, 2.2] AOX via EN Schneider Subtotal (% CI) 2. 2 Heterogeneity: Not applicable Test for overall effect: Z =. (P =.) %.% 3.3 [-.,.] 3.3 [-.,.] 2..3 AOX via PN & EN Porter Berger 2 Heyland Subtotal (% CI) Heterogeneity: Tau² =.; Chi² =.2, df = 2 (P =.44); I² = % Test for overall effect: Z =.2 (P =.).2% 4.% 7.%.% -3. [-3.,.].4 [-.4,.4].4 [-2.7, 2.7].3 [-.7,.7] 2 22 Total (% CI) 47 3.% Heterogeneity: Tau² =.; Chi² =.4, df = (P =.4); I² = % Test for overall effect: Z =. (P =.33) Test for subgroup differences: Chi² =.2, df = 2 (P =.7), I² = %.3 [-.,.] - - Favours AOX Favours control 7

18 Figure. LOS Study or Subgroup.. AOX via PN Berger Berger 2b Berger 2a Subtotal (% CI) AOX Control Mean Difference Mean Difference Mean SD Total Mean SD Total Weight IV, Random, % CI Year IV, Random, % CI Heterogeneity: Tau² =.; Chi² =.3, df = 2 (P =.3); I² = % Test for overall effect: Z =. (P =.) 2.%.%.4% 3.% -2. [-37.4, 3.4] -4. [-4., 32.]. [-37.3, 73.3] -.3 [-2., 3.] AOX via EN Schneider Subtotal (% CI) Heterogeneity: Not applicable Test for overall effect: Z =.2 (P =.) % 2.% -2. [-24.,.2] -2. [-24.,.2] 2..3 AOX via PN & EN Porter Berger 2 Heyland Subtotal (% CI) Heterogeneity: Tau² =.7; Chi² = 3.22, df = 2 (P =.2); I² = 3% Test for overall effect: Z =. (P =.2) 2.2% 44.% 47.% 3.7% Total (% CI) 77 7.% Heterogeneity: Tau² =.; Chi² = 4.42, df = (P =.2); I² = % Test for overall effect: Z =.4 (P =.3) Test for subgroup differences: Chi² =.2, df = 2 (P =.), I² = % -7.7 [-42., 7.] -3. [-.4, 2.4].7 [-3.4, 7.4] -.4 [-., 4.] -. [-4.7, 2.4] Favours AOX Favours control

19 Figure 7. Duration of mechanical ventilation Study or Subgroup.7. AOX via PN Berger Berger 2b Berger 2a Berger 27 El-Attar Manzanares Subtotal (% CI) AOX Control Mean Difference Mean Difference Mean SD Total Mean SD Total Weight IV, Random, % CI Year IV, Random, % CI % 2.2%.%.% 3.4%.4%.% Heterogeneity: Tau² = 7.; Chi² = 22.7, df = (P =.4); I² = 7% Test for overall effect: Z =.3 (P =.2) -3. [-.34,.34] -. [-3.4, 3.4] 2. [-.3,.3] -. [-.4,.4] -.4 [-.7, -.3]. [-3.,.] [-.7,.2] AOX via EN Crimi Schneider Subtotal (% CI) Heterogeneity: Tau² =.; Chi² =.3, df = (P =.3); I² = 3% Test for overall effect: Z = 3.27 (P =.).4% 2.% 22.2% -2.7 [-3.24, -2.] 3.3 [-.2, 4.] -2. [-4., -.4] AOX via PN & EN Heyland. Subtotal (% CI) Heterogeneity: Not applicable Test for overall effect: Z =.34 (P =.73)..7.%.%.4 [-., 2.7].4 [-., 2.7] 22 Total (% CI) 4 3.% Heterogeneity: Tau² =.3; Chi² = 3.22, df = (P =.2); I² = 74% Test for overall effect: Z =.3 (P =.) Test for subgroup differences: Chi² = 4.2, df = 2 (P =.), I² =.7% -.7 [-3.7,.3] - - Favours AOX Favours control