Quality assessment and study characteristics
We screened and evaluated 4720 studies, assessing 194 for eligibility. The selected 58 studies, including 38,787 patients, were enrolled in our analysis (Fig. 1) (1, 4–7, 9–11, 17–65). There were 4718 burn patients with AKI and 1764 burn patients with RRT. A total of 93.53% of the burn patients with RRT were AKI patients. Figure 1 shows a flow chart of the identification and selection of the studies.
The main features of these studies are shown in Table 1. Risk of bias is summarized in Additional file 1: Table S1 (cohort or case-control studies) and Fig. S1 (RCT studies). Among the 58 included studies, 23 were from North America (1, 4, 9, 17–19, 27, 29, 31, 33, 34, 36, 37, 42–44, 48–50, 54, 59, 61, 62), 18 were from Asia (5, 7, 10, 11, 22, 24–26, 28, 39, 40, 46, 47, 51, 52, 58, 64, 66), 15 were from European countries (6, 20, 21, 23, 30, 35, 38, 45, 53, 55–57, 60, 65, 67), one was from South America (32), one was Africa (41). A total of 12 studies could not be used for analysis of the prevalence of RRT, seven of which only reported on RRT patients (11, 18, 21, 37, 40, 43, 45), four of which were RCT studies (9, 10, 51, 52), and the remaining one of which were historical controls (44). Most of the 58 studies (41/60) were retrospective cohort studies, but 12 were prospective cohort studies (7, 35, 39, 41, 43, 46, 47, 50, 57, 58, 64, 65), and 5 RCT studies (9, 10, 51, 52, 56).
Table 1
Summary of the baseline characteristics of the studies included in the meta-analysis.
Study
|
Nationality
|
Study type
|
Sample size
|
AKI definition
|
Admitted time
|
AKI numbers
|
RRT numbers
|
RRT mortality (%)
|
Akers 2012
|
America
|
Retrospective cohort
|
171
|
≥ 0.5 mg/dL SCr increase, any time during therapy
|
2006–2009
|
38
|
33
|
19(57.58)
|
Béchir 2010
|
Switzerland
|
Prospective cohort
|
30
|
Dialysis
|
1997.8-1999.7
|
5
|
5
|
NA
|
Béchir 2013
|
Switzerland
|
Randomized controlled trial
|
45
|
Dialysis
|
2009.11-2013.1
|
12
|
12
|
NA
|
Boucher 2016
|
America
|
Prospective cohort
|
10
|
AKIN criteria
|
NA
|
9
|
10
|
3(30)
|
Chrysopoulo 1999
|
America
|
Retrospective cohort
|
1404
|
Oliguria for at least 36 hours (urine output < 350 mL/d), a blood urea nitrogen-creatinine ratio of less than 20, Scr > 2 mg/dL, and the requirement for dialysis after injury
|
1981–1998
|
76
|
67
|
61(91.04)
|
Chun 2018
|
Korea
|
Prospective cohort
|
76
|
AKIN criteria
|
2014.2-2015.9
|
32
|
20
|
19(95)
|
Chung 2008
|
America
|
Retrospective cohort
|
102
|
RIFLE classification
|
2003–2007
|
34
|
18
|
10(55.56)
|
Chung 2017
|
America
|
Randomized controlled trial
|
37
|
Oliguria (< 20 ml/hour) for > 24 hours or an increase Scr > 2 mg/dl in males or > 1.5 mg/dl in females over a period of < 4days
|
2012–2016
|
37
|
37
|
23(62.16)
|
Chung 2018
|
America
|
Retrospective cohort
|
4086
|
KDIGO criteria
|
2012–2016
|
160
|
170
|
85(50)
|
Clark 2019
|
America
|
Retrospective cohort
|
1040
|
KDIGO criteria
|
2008–2015
|
601
|
58
|
36(62.07)
|
Coca 2007
|
America
|
Retrospective cohort
|
304
|
RIFLE classification
|
1998–2003
|
81
|
11
|
8(72.73)
|
Damkat-Thomas 2011
|
America
|
Retrospective cohort
|
41
|
RIFLE classification
|
2006–2010
|
17
|
5
|
2(40)
|
Davies 1979
|
England
|
Retrospective cohort
|
1064
|
NA
|
1958–1979
|
28
|
25
|
22(88)
|
Demsey 2019
|
Canada
|
Retrospective cohort
|
151
|
AKIN criteria
|
2010–2016
|
64
|
18
|
7(38.89)
|
Dépret 2018
|
France
|
Retrospective cohort
|
87
|
KDIGO criteria
|
2012.1-2015.1
|
55
|
21
|
NA
|
Gille 2014
|
Germany
|
Retrospective cohort
|
18
|
NA
|
2004–2009
|
18
|
18
|
2(11.11)
|
Haberal 1993
|
Turkey
|
Retrospective cohort
|
915
|
NA
|
1979–1989
|
19
|
19
|
15(78.95)
|
Hladik 2001
|
Czech
|
Retrospective cohort
|
40
|
NA
|
1996–2000
|
10
|
40
|
28(70)
|
Holm 1999
|
Germany
|
Retrospective cohort
|
328
|
Scr > 2.0 mg/dl (with rising tendency) combined with a blood urea nitrogen level > 200 mg/dl or in patients with anuria or oliguria (urine volume < 400 ml/24 h) with anasarca and/or hyperkalemia
|
1994–1998
|
48
|
48
|
41(85.42)
|
Hong 2013
|
Korea
|
Prospective cohort
|
45
|
RIFLE classification
|
2011–2012
|
11
|
5
|
4(80)
|
Hu 2012
|
China
|
Retrospective cohort
|
396
|
RIFLE classification
|
2006–2010
|
151
|
25
|
NA
|
Hundeshagen 2017
|
America
|
Retrospective cohort
|
246(adults)
|
KDIGO criteria
|
2004–2016
|
26
|
3
|
NA
|
Kim 2003
|
Korea
|
Retrospective cohort
|
147
|
Scr ≥ 2 mg/dL
|
2000.1-2000.12
|
28
|
3
|
3(100)
|
Knowlin 2018
|
America
|
Retrospective cohort
|
7539
|
Using ICD-9 codes
|
2002–2012
|
194
|
1
|
NA
|
Kumar 2016
|
America
|
Retrospective cohort
|
254
|
AKIN criteria
|
2011–2013
|
190
|
10
|
NA
|
Kuo 2016
|
China
|
Retrospective cohort
|
145
|
KDIGO criteria
|
2004–2006
|
59
|
9
|
7(77.78)
|
Kuo 2018
|
China
|
Retrospective cohort
|
301
|
AKIN criteria
|
2006–2011
|
34
|
28
|
NA
|
Kym 2015
|
Korea
|
Prospective cohort
|
85
|
RIFLE classification
|
2012–2013
|
48
|
22
|
NA
|
Leblanc 1997
|
Canada
|
Retrospective cohort
|
970
|
NA
|
1987–1994
|
16
|
16
|
13(81.25)
|
Liu 2016
|
China
|
Randomized controlled trial
|
41
|
NA
|
2013.1-2015.7
|
NA
|
20
|
7(35)
|
Liu 1986
|
China
|
Retrospective cohort
|
6050
|
NA
|
1958–1983
|
53
|
15
|
8(53.33)
|
Lopes 2007
|
Portugal
|
Retrospective cohort
|
126
|
Doubling of baseline Scr
|
2004–2006
|
32
|
11
|
NA
|
Mariano 2010
|
Italy
|
Retrospective cohort
|
548
|
NA
|
2000–2007
|
98
|
70
|
50(71.43)
|
Mason 2016
|
Canada
|
Retrospective cohort
|
330
|
Scr > 1.5 mg/dL
|
2004–2011
|
48
|
37
|
NA
|
Muñoz 2017
|
Spain
|
Retrospective cohort
|
840
|
KDIGO criteria
|
1992–2012
|
466
|
34
|
NA
|
Mustonen 2008
|
Finland
|
Retrospective cohort
|
1380
|
Scr > 120umol/L (1.4 mg/dL); for chronic renal insufficiency patients, 2-fold rise in Scr or Scr rose > 100 µmol/l during 1 day
|
1988–2001
|
93
|
32
|
20(62.5)
|
Peng 2005
|
China
|
Randomized controlled trial
|
20
|
NA
|
2001.6-2001.10
|
NA
|
10
|
1(10)
|
Planas 1982
|
America
|
Retrospective cohort
|
29
|
Scr level above initial values to a level equal to or greater than 1.5 mg/dL
|
1980–1982
|
11
|
3
|
2(66.67)
|
Pronina 2015
|
Canada
|
Retrospective cohort
|
1405
|
AKIN or RIFLE criteria
|
2006–2014
|
53
|
21
|
7(33.33)
|
Queiroz 2016
|
Brazil
|
Retrospective cohort
|
293
|
An elevation in baseline serum
creatinine greater than or equal to 50% from baseline
|
2010–2012
|
77
|
52
|
NA
|
Rakkolainen 2018
|
Finland
|
Retrospective cohort
|
187
|
Scr ≥ 120umol/L (1.4 mg/dl)
|
2006–2015
|
51
|
21
|
9(42.86)
|
Ren 2015
|
China
|
Prospective cohort
|
95
|
KDIGO criteria
|
2013.4-2013.9
|
11
|
5
|
4(80)
|
Sabry 2009
|
Egypt
|
Prospective cohort
|
40
|
Scr > 2 mg/dL and blood urea nitrogen > 25 mg/dL
|
2007.5-2007.12
|
9
|
4
|
2(50)
|
Saffle 1993
|
America
|
Retrospective cohort
|
529
|
Scr > 132.6umol/L (1.5 mg/dL)
|
1987–1991
|
143
|
5
|
5(100)
|
Sánchez-Sánchez 2016
|
Spain
|
Prospective cohort
|
165
|
RIFLE classification
|
2008.10-2011.12
|
32
|
15
|
14(93.33)
|
Schneider 2012
|
America
|
Retrospective cohort
|
220
|
RIFLE classification
|
2006–2008
|
103
|
25
|
NA
|
Sen 2015
|
America
|
Prospective cohort
|
30
|
RIFLE classification
|
NA
|
14
|
3
|
NA
|
Soltani 2009
|
America
|
Retrospective cohort
|
3356
|
NA
|
1994–2004
|
38
|
33
|
23(69.7)
|
Steinvall 2008
|
Sweden
|
Prospective cohort
|
127
|
RIFLE classification
|
1997–2005
|
31
|
4
|
3(75)
|
Stewart 2013
|
America
|
Retrospective cohort
|
1967
|
AKIN criteria
|
2003–2008
|
640
|
70
|
49(70)
|
Tang 2018
|
China
|
Retrospective cohort
|
157
|
AKIN criteria
|
2014.8
|
89
|
82
|
NA
|
Tremblay 2000
|
Canada
|
Retrospective cohort
|
12
|
NA
|
1995–1998
|
12
|
12
|
6(50)
|
Witkowski 2016
|
Poland
|
Retrospective cohort
|
225
|
Decrease in GFR of less than 60 ml/min at admission, decrease in GFR of more than 75% compared to baseline or decrease in the daily diuresis of less than 500 ml for at least 24 h
|
2012–2013
|
135
|
9
|
9(100)
|
Yang 2014
|
Korea
|
Prospective cohort
|
90
|
RIFLE classification
|
2011–2012
|
55
|
22
|
17(77.27)
|
Yim 2015
|
Korea
|
Prospective cohort
|
97
|
AKIN criteria
|
2012–2013
|
40
|
23
|
NA
|
Yoon 2017-Burns
|
Korea
|
Retrospective cohort
|
84
|
RIFLE classification
|
2007–2010
|
84
|
84
|
71(84.5)
|
Yoon 2017-PLOS ONE
|
Korea
|
Retrospective cohort
|
216
|
AKIN criteria
|
2009–2015
|
190
|
216
|
176(81.48)
|
You 2018
|
China
|
Randomized controlled trial
|
82
|
KDIGO stage 3
|
2014–2017
|
9
|
41
|
11(26.83)
|
NA, not available; Scr, serum creatine. |
Prevalence and mortality of RRT under different AKI diagnostic criteria
We analysed 46 literatures that reported the prevalence of RRT in burn patients (Table 2) (1, 4–7, 17, 19, 22–36, 38, 39, 41, 42, 46–50, 53–62, 64–68). The prevalence rates of RRT were 5.14% (95%CI 4.54%-5.74%) in all burn patients and 35.8% (95%CI 29.54%-42.07%) in AKI patients. The prevalence of RRT among burn patients in the ICU was 10.92% (95%CI 8.71%-13.14%)(4–7, 19, 25, 30, 32, 34, 39, 42, 46, 47, 50, 54, 56, 57, 59, 60, 62, 65, 67). A total of 25 studies with RIFLE, AKIN and KDIGO as AKI diagnostic criteria were analysed (4, 6, 7, 19, 24–26, 31, 35, 36, 38, 39, 46, 47, 49, 50, 54, 58–62, 64–66). The prevalence of RRT in these burn patients was 29.58% (95%CI 23.65%-35.52%).
Moreover, we analysed the results of 41 studies that reported RRT mortality in burn patients (Table 3). The mortality of all burn patients with RRT was 65.52% (95%CI 58.41%-72.64%) (4, 5, 9–11, 17–23, 25, 27, 28, 30, 31, 33–46, 48, 51, 52, 54, 55, 58, 59, 64, 65, 67). The mortality of patients with RRT in ICU was 62.7% (95%CI 53.7%-71.7%) (4, 5, 9, 10, 18, 19, 25, 30, 34, 39, 40, 42–46, 54, 59, 65, 67). Of the 41 studies, excluding those only conducted in RRT patients, 20 gave not only the mortality rate of all burn patients but also the mortality rate of burn patients with RRT (4, 5, 19, 23, 25, 30, 33–36, 38, 39, 42, 46, 48, 54, 58, 59, 65, 67). Based on the results of the above 20 articles, it was found that the mortality rate of RRT patients was 30.33% (95%CI 22.06%-38.59%) of the total. The results of 20 studies with RIFLE, AKIN, and KDIGO as AKI diagnostic criteria showed that the mortality of RRT in burn patients was 67.16% (95%CI 57.40%-76.93%) (4, 10, 11, 18, 19, 25, 31, 35, 36, 38–40, 43, 44, 46, 54, 58, 59, 64, 65). Three studies reported deaths in all burn patients undergoing RRT (5, 33, 38, 42). According to different mortality categories, the mortalities of 14 days, 28 days and 60 days ranged from 30–50%, while those of ICU and hospital were 56.98% and 68.89%, and overall mortality further increased to 75.24% (Additional file 1: Table S2).
According to the three diagnostic criteria of RIFLE, AKIN, and KDIGO, the prevalence of RRT was KDIGO<RIFLE<AKIN, and that of mortality was KDIGO<AKIN<RIFLE. The prevalence of RRT was 14.79% (95%CI 9.02%-20.56%) and that of mortality was 55.29% (95%CI 39.46%-71.12%) in the six literatures with KDIGO classification as the diagnostic standard, which was lower than other AKI diagnostic standards.
There was no significant correlation (r=-0.224, P = 0.159) between the year of publication and the mortality of burn patients with RRT (Additional file 1: Fig. S2). According to the year of publication, the patients were divided into four subgroups (Additional file 1: Fig. S3) from 2010–2020, 2000–2009, 1990–1999, 1989 and before. The mortality of the 2010–2020 group was 60.42%±25.35%, that of the 2000–2009 group was 61.55%±23.29%, that of the 1990–1999 group was 87.33%±8.44%, and that of the 1989 and before group was 63.52%±25.05%. There was no significant difference between groups (P = 0.139). After 2010, three studies still reported that the mortality of RRT patients was more than 90% (38, 58, 65).
Table 2
The prevalence of RRT in burn patients with different diagnostic criteria.
Diagnosis
|
N. of Trials
|
Patients
|
I2 (%)
|
P
|
Prevalence (%)
|
95%CI
|
All burn patients
|
46
|
34076
|
97
|
<0.01
|
5.14
|
4.54–5.74
|
RRT of AKI patients
|
39
|
3929
|
98
|
<0.01
|
35.8
|
29.54–42.07
|
ICU
|
22
|
5628
|
90%
|
<0.01
|
10.92
|
8.71–13.14
|
RIFLE classification
|
11
|
596
|
78
|
<0.01
|
28.85
|
21-36.69
|
AKIN classification
|
9
|
1180
|
98
|
<0.01
|
43.96
|
28-59.91
|
KDIGO classification
|
6
|
1218
|
83
|
<0.01
|
14.79
|
9.02–20.56
|
Summary of RIFLE, AKIN, KDIGO
|
25
|
3044
|
96
|
<0.01
|
29.58
|
23.65–35.52
|
Table 3
The mortality of RRT in burn patients with different diagnostic criteria.
Diagnosis
|
N. of Trials
|
Patients
|
I2 (%)
|
P
|
RRT mortality (%)
|
95%CI
|
Summary of all literatures
|
41
|
1342
|
90
|
<0.01
|
65.52
|
58.41–72.64
|
ICU
|
20
|
797
|
85%
|
<0.01
|
62.7%
|
53.7–71.7
|
RIFLE classification
|
9
|
185
|
77
|
<0.01
|
70.08
|
56.4-83.75
|
AKIN classification
|
7
|
370
|
90
|
<0.01
|
66.73
|
52.01–81.45
|
KDIGO classification
|
5
|
283
|
81
|
<0.01
|
55.29
|
39.46–71.12
|
Summary of RIFLE, AKIN, KDIGO
|
20
|
811
|
90
|
<0.01
|
67.16
|
57.40-76.93
|
The effect of RRT on the mortality of burn patients
Compared with the absolute values of serum creatinine or urine volume as the grading standard defined by AKI (such as SOFA or other definition), RIFLE, AKIN, and KDIGO grading can more accurately reflect the actual incidence of AKI. Therefore, we used cohort studies with RIFLE, AKIN, and KDIGO as diagnostic criteria for subgroup analysis. The standard of ARF was defined as RIFLE-Failure level and above, AKIN-3 or KDIGO Level 3. The results showed that the risks of death in burn patients with AKI and ARF were 5.19 times (95%CI 3.87–6.97) and 4.77 times (95%CI 3.01–7.57) higher than those in non-AKI patients, while the risk of death in RRT patients was increased to 5.78 (95%CI 3.43–9.75), but there was no significant difference between groups (P = 0.87, Fig. 2). A meta-analysis of the cohort study did not show that RRT could reduce the risk of death in burn patients with AKI.
Three RCT studies compared the effects of standard care (non-RRT) and CRRT on patients' prognosis (10, 51, 52). Two RCT studies did not find that CRRT had a positive effect on the mortality of burn patients (51, 52). You 2018 (10) found that early HVHF treatment may reduce 90-day mortality of patients with severe burns (≥ 80% TBSA) (P = 0.049). A meta-analysis of these three RCT studies also did not show that CRRT could reduce the risk of death in burn patients with AKI compared with the standard care (non-RRT) (RR = 0.64, 95%CI 0.40–1.03, P = 0.06; Fig. 3).
Regional Citrate Anticoagulation
Four articles reported the studies of regional citrate anticoagulation (RCA) for CRRT in burn patients, including 3 retrospective cohort studies (21, 55, 69) and 1 RCT study (70). The cohort studies found that the filter survival time (CRRT) of the citrate group was significantly longer than that of the heparin group (average 28.5–28.7 h vs 19-19.4 h) (55, 69). When the dialysis mode was SLED, the filter survival time of the citrate group was still better than that of the heparin group (average 8–13 h vs 6.5-7 h) (55, 70). Moreover, the citric acid group also had some advantages in the recovery time of urine volume and the length of stay in ICU (69). After 24 h of treatment, the coagulation index (PT and APTT) and the inflammation index (PCT and CRP) of the citric acid group were significantly lower than those of the heparin group (69). However, there was no significant difference in the risk of death between the two groups (P = 0.51, Additional file 1: Fig. S4). Only a small-sample RCT study was conducted to compare the effects of local heparin anticoagulation and local citrate anticoagulation for SLED-HF in severe burn patients (TBSA greater than 50%) with sepsis (70). The results also showed that the citric acid group had significant advantages in single treatment time, treatment time reaching rate (12 h), coagulation index (PT and APTT), and inflammation index (PCT and CRP) (70). All these literatures suggest that the incidence of bleeding events in the heparin group was significantly higher than that in the citric acid group (55, 69, 70). The incidence of hypocalcaemia in the citric acid group was 7.27% (4/55). The incidence rates of metabolic acidosis and metabolic alkalosis in the citric acid group were slightly higher than those in the heparin group, but the differences were not statistically significant (69).
RRT-related adverse reactions
Nine articles reported the incidence of RRT-related adverse reactions. The total incidence was 28.77% (63/229) (9, 10, 21, 23, 27, 28, 30, 37, 51), including thrombocytopenia 0.44% (1/229), bleeding 10.92% (25/229), thromboembolism 1.75% (4/229), secondary infection 9.61% (22/229), electrolyte disorder 2.62% (6/229), and imbalance syndrome 0.44% (1/229). Among them, only Chung 2017 reported 6 patients with electrolyte disorder. Other literatures may not mention the occurrence of electrolyte disorder due to certain concerns, which may underestimate the prevalence of electrolyte disorder. A total of 16 patients with peritoneal dialysis were reported in 9 articles; most of these patients were from Liu 1986 (28). Among the 16 patients, 1 had "unbalanced syndrome" and improved after stopping dialysis; 4 had abdominal infection, 3 survived, and 1 changed to haemodialysis and ultimately died. Considering the immature technology at that time, the incidence of PD adverse reactions may be overestimated.
Renal outcome
A recent study found an odds of dialysis of 2.40 in burn patients who developed AKI compared with the general Finnish population (71). Eleven studies followed up the long-term renal outcomes of burn patients who survived RRT (9, 18, 19, 21, 27, 30, 34, 36, 37, 50, 54). A total of 184 patients survived after renal replacement therapy; 64.13% of them (118/184) were dialysis-independent after discharge, 25% (46/184) needed temporary required dialysis, and 10.87% (20/184) needed long-term dialysis (more than 6 months after discharge). Thalji 2017 found that one year after burn, the proportion of chronic dialysis in non-AKI patients was 0.33% (56/16985), significantly lower than that in patients with AKI, which was 4.58% (26/568) (72). The proportion of severe CKD (defined as stage 3–5 CKD) in non-AKI patients (0.71%) was also lower than those in AKI patients (5.81%) (72). Gille reported that 3 of the 16 surviving burn patients undergoing CRRT had CKD progression (GFR < 45 ml/min.1.73 m2, CKD 3b) (18.75%) (21). Two patients had slightly impaired GFR (< 90 ml/min.1.73 m2, CKD 2) before the burn trauma. One patient had normal GFR (21).