Background
Methods
Search strategy and study identification
Study eligibility criteria
Data abstraction
-
(a) the extent to which randomization was concealed, (b) blinding, (c) analysis based on the intention-to-treat (ITT) principle, (d) comparability of groups at baseline, (e) extent of follow-up, (f) description of treatment protocol, (g) co-interventions, (h) definition of clinical outcomes.
Statistical analysis
Subgroup analyses
Results
Study identification and selection
Author | Year | Population | Setting | Total patientsa
| EN group | PN group | Reference |
---|---|---|---|---|---|---|---|
Rapp et al. | 1983 | Head-injured patients | Single-center | 38 | 18 | 20 | [37] |
Adams et al. | 1986 | Critically ill trauma | Single-center | 46 | 23 | 23 | [38] |
Young et al. | 1987 | Brain-injured patients | Single-center | 51 | 28 | 23 | [39] |
Peterson et al. | 1988 | Critically ill patients with abdominal trauma | Single-center | 59 | 29 | 30 | [40] |
Cerra et al. | 1988 | Critically ill patients | Single-center | 70 | 33 | 37 | [41] |
Moore et al. | 1989 | Abdominal trauma | Single-center | 75 | 39 | 36 | [42] |
Kudsk et al. | 1992 | Abdominal trauma | Single-center | 98 | 52 | 46 | [43] |
Dunham et al. | 1994 | Blunt trauma | Single-center | 28b
| 12 | 16 | [44] |
Borzotta et al. | 1994 | Closed head injury | Single-center | 59 | 36 | 23 | [45] |
Hadfield et al. | 1995 | Mixed ICU medical-surgical | Single-center | 24 | 13 | 11 | [46] |
Kalfarentzos et al. | 1997 | Severe acute pancreatitis | Single-center | 38 | 18 | 20 | [47] |
Woodcock et al. | 2001 | ICU patients requiring nutrition support | Single-center | 38 | 17 | 21 | [27] |
Casas et al. | 2007 | Severe acute pancreatitis | Single-center | 22 | 11 | 11 | [48] |
Chen et al. | 2011 | Medical ICU | Single-center | 98b
| 49 | 49 | [49] |
Justo Meirelles et al. | 2011 | Traumatic brain injury | Single-center | 22 | 12 | 10 | [21] |
Wang et al. | 2013 | Surgical ICU (severe acute pancreatitis) | Single-center | 121b
| 61 | 60 | [22] |
Sun et al. | 2013 | Surgical ICU (severe acute pancreatitis) | Single-center | 60 | 30 | 30 | [50] |
Harvey et al. | 2014 | Mixed medical-surgical | Multi-center | 2400 | 1200 | 1200 | [16] |
Study | Methods (score) | Mortality, N (%)a
| Infections, N (%)b
| LOS, days, mean ± SD (N) | Mechanical ventilation, days, mean ± SD (N) | Caloric intakec
| |||||
---|---|---|---|---|---|---|---|---|---|---|---|
EN | PN | EN | PN | EN | PN | EN | PN | EN | PN | ||
1. Rapp et al. 1983 [37] | C.Random: not sure | 9/18 (50) | 3/20 (15) | NR | Hospital 49.4d
| Hospital 52.6d
| 10.3d
| 10.4d
| 685
P = 0.001 | 1750 | |
ITT: no | |||||||||||
Blinding: no (4) | |||||||||||
2. Adams et al. 1986 [38] | C.Random: not sure | 1/23 (4) | 3/23 (13) | 15/23 (65) | 17/23 (74) | ICU 13 ± 11 (19) | ICU 10 ± 10 (17) | 12 ± 11 (17) | 10 ± 10 (13) | 2088 NSf
| 2572 |
ITT: yes | Hospital 30 ± 21 (19) | Hospital 31 ± 29 (17) | |||||||||
Blinding: no (8) | |||||||||||
3. Young et al. 1987 [39] | C.Random: not sure | 10/28 (36) | 10/23 (43) | 5/28 (18) | 4/23 (17) | NR | NR | 1671
P = 0.02 | 2299 | ||
ITT: no | |||||||||||
Blinding: no (6) | |||||||||||
4. Peterson et al. 1988 [40] | C.Random: not sure | NR | 2/21 (10) | 8/25 (32) | ICU 3.7 ± 0.8 (21) | ICU 4.6 ± 1.0 (25) | NR | Kcal on day 5 | |||
2204
P = 0.04 | 2548 | ||||||||||
ITT: no | Hospital 13.2 ± 1.6 (21) | Hospital 14.6 ± 1.9 (24) | |||||||||
Blinding: no (5) | |||||||||||
5. Cerra et al. 1988 [41] | C.Random: not sure | ICU 7/31 (22) | ICU 8/35 (23) | NR | NR | NR | Non-protein kcal | ||||
1684 NSf
| 2000 | ||||||||||
ITT: no | |||||||||||
Blinding: no (2) | |||||||||||
6. Moore et al. 1989 [42] | C.Random: yes | NR | 5/29 (17) | 11/30 (37) | NR | NR | Non-protein kcal on day 5 | ||||
ITT: no | 1847
P = 0.01 | 2261 | |||||||||
Blinding: no (10) | |||||||||||
7. Kudsk et al. 1992 [43] | C.Random: not sure | ICU 1/51 (2) | ICU 1/45 (2) | 9/51 (16) | 18/45 (40) | Hospital 20.5 ± 19.9 (51) | Hospital 19.6 ± 18.8 (45) | 2.8 ± 4.9 (51) | 3.2 ± 6.7 (45) | Kcal/kg/d | |
ITT: no | 15.7
P < 0.05 | 19.1 | |||||||||
Blinding: single (10) | |||||||||||
8. Dunham et al. 1994 [44] | C.Random: not sure | 1/12 (7) | 1/15 (8) | NR | NR | NR | NSf
| ||||
ITT: no | |||||||||||
Blinding: no (8) | |||||||||||
9. Borzotta et al. 1994 [45] | C.Random: not sure | 5/28 (18) | 1/21 (5) | 51/28 (28) | 39/21 (21) | Hospitale 39 ± 23.1 | Hospitale 36.9 ± 14 | NR | 2097 NSf
| 1961 | |
ITT: no | |||||||||||
Blinding: no (6) | |||||||||||
10. Hadfield et al. 1995 [46] | C.Random: not sure | ICU 2/13 (15) | ICU 6/11 (55) | NR | NR | NR | NSf
| ||||
ITT: no | |||||||||||
Blinding: no (7) | |||||||||||
11. Kalfarentzos et al. 1997 [47] | C.Random: not sure | ICU 1/18 (6) | ICU 2/20 (10) | 5/18 (28) | 0/20 (50) | ICU 11 (5–21)d
| ICU 12 (5–24)d
| 15 (6–16)d
| 11 (7–31)d
| Non-protein kcal/kg/d | |
ITT: no | Hospital 40 (25–83)d
| Hospital 39 (22–73)d
| 24.1 NSf
| 24.5 | |||||||
Blinding: single (9) | |||||||||||
12. Woodcock et al. 2001 [27] | C.Random: yes | 9/17 (53) | 5/21 (24) | 6/16 (38) | 11/21 (52) | 33.2 ± 43 (16) | 27.3 ± 18.7 (18) | NR | % caloric target (30 kcal/kg/d) achieved | ||
ITT: yes | 54.1
P < 0.001 | 96.7 | |||||||||
Blinding: single (12) | |||||||||||
13. Casas et al. 2007 [48] | C.Random: no/unsure | Hospital 0/11 (0) | Hospital 2/11 (18) | 1/11 (9) | 3/11 (27) | Hospital 30.2 (average) | Hospital 30.7 (average) | NR | Kcal/kg/d | ||
ITT: Yes | 20.1 NSf
| 20.8 | |||||||||
Blinding: no (8) | |||||||||||
14. Chen et al. 2011 [49] | C.Random: yes | 20-day 11/49 (22) | 20-day 10/49 (20) | 5/49 (10) | 18/49 (37) | ICU 9.09 ± 2.75 | ICU 9.60 ± 3.06 | 7.95 ± 2.11 | 8.23 ± 2.42 | NR | |
ITT: yes | |||||||||||
Blinding: no (7) | Hospital 23.32 ± 5.6 | Hospital 22.24 ± 3.27 | |||||||||
15. Justo Meirelles et al. 2011 [21] | C.Random: no | Not specified 1/12 (8.3) | Not specified 1/10 (10) | Total infectious complications 2/12 (16.7) | Total infectious complications 4/10 (40) | ICU 14 (5–26) | ICU 14 (6–24) | NR | Cumulative kcal over 5d | ||
ITT: no | Pneumonia 2/12 (16.7) | Pneumonia 2/10 (20) | 5985
P = 0.34 | 6586 | |||||||
Blinding: no (5) | Sepsis 0 | Sepsis 2/10 (20) | |||||||||
16. Wang et al. 2013 [22] | C.Random: no | Hospital 3/61 (5) | Hospital 7/60 (12) | Pancreatic sepsis 13/61 (21) | Pancreatic sepsis 24/60 (40) | NR | NR | NR | |||
ITT: no | |||||||||||
Blinding: double (7) | MODS 15/61 (24.6) | MODS 22/60 (36.7) | |||||||||
17. Sun et al. 2013 [50] | C.Random: no | Hospital 2/30 (7) | Hospital 1/30 (3) | Pancreatic 3/30 (10) | Pancreatic 10/30 (33) | ICU 9 (5–14) | ICU 12 (8–21) | NR | NR | ||
ITT: no | MODS 5/30 (17) | MODS 13/30 (43) | |||||||||
Blinding: no (6) | SIRS 12/30 (40) | SIRS 22/30 (73) | |||||||||
18. Harvey et al. 2014 [16] | C.Random: yes | ICU 352/1197 (29.4) | ICU 317/1190 (26.6) | Total infectious complications 194/1197 (16.2)g
| Total infectious complications 194/1191 (16.3)g
| ICU 11.3 ± 12.5 (1197) | ICU 12 ± 13.5 (1190) | 8.2 ± 9.3 (1197) | 8.7 ± 11.5 (1189) | Cumulative kcal/kg/d over 5d | |
ITT: yes | Hospital 450/1186 (37.9) | Hospital 431/1185 (36.4) | Pneumonia 143/1197 (11.9) | Pneumonia 135/1191 (11.3) | Hospital 26.8 ± 33.2 (1186) | Hospital 27.5 ± 33.9 (1185) | 74 NSf
| 89 | |||
30-day 409/1195 (34.2) | 30-day 393/1188 (33.1) | Bloodstream infections 21/1197 (1.8) | Bloodstream infections 27/1191 (2.9) | ||||||||
Blinding: no (8) | 90-day 464/1188 (39.1) | 90-day 442/1184 (37.3) | Surgical infections 12/1197 (1.0) | Surgical infections 10/1191 (0.8) |
Effect of EN versus PN on mortality
Effect of EN versus PN on infectious complications
Effect of EN versus PN on ICU and hospital length of stay
Effect of EN versus PN on mechanical ventilation
Effect of trial quality and publication date on outcomes and risk of publication bias
Discussion
Comparison to other meta-analyses
Effect of dissimilar caloric intake
Effect of trial quality and publication bias
Strength and limitations
Conclusions
Key messages
-
This updated meta-analysis on effects of EN versus PN on clinical outcomes included 18 RCTs with 3347 randomized critically ill patients
-
There was no significant difference in mortality between patients fed via EN or PN
-
Compared to PN, the use of EN was associated with a significant reduction of infectious complications and ICU LOS
-
This positive treatment effect of EN compared to PN may be attributed to differences in caloric intake and significant publication bias among aggregated trials
-
EN still should be considered first-line nutritional therapy over PN in critically ill patients with a functioning gastrointestinal tract