Table 1 Results of the 21 included studies

Basques et al. 2015 [8]

Retrospective observational study

GA vs. SA

n = 9.842 GA = 7.253 (73.7%) SA = 2.589 (26.3%)

Operating time; length of stay (LOS); adverse events within 30 post-operative days, rate of re-admission

30-day mortality

30- day mortality: GA vs. SA: (OR 0.98, 95% CI 0.82 to 1.20, p = 0.908); LOS: GA vs SA: (HR: 1.28, 95% CI 1.22 to 1.34, p < 0.001)

There was no difference between the groups except of length of hospital stay.

Biboulet et al. 2012 [32]

Randomised controlled study

GA vs. SA (propofol, sevoflurane)

n = 45 GA = 30 SA = 15

Blood pressure profile, heart rate profile

30-day mortality

Hypotension episodes: SA = 0 (range, 0-6) vs. propofol = 11.5 (range 1-25) vs. sevoflurane = 10 (range,1-23) (p < 0.0001); maximal decrease in arterial pressure: SA = 26% [16], propofol = 47% [8%], sevoflurane = 46% [12%] (p < 0.001)

SA in elderly patients provided better blood pressure stability than propofol and sevoflurane.

Brox et al. 2016 [23]

Retrospective observational study

GA vs. SA vs. mixed

n = 7.585 GA = 4.257 (56%) SA = 3.059 (40%) Mixed = 269 (4%)

30-day, 90-day and 365-day mortality

30-day mortality: GA = 177 (4%) vs. SA = 113 (4%) vs. mixed = 17 (6%); 90-day mortality: GA = 336 (8%) vs. SA = 224 (7%) vs. mixed = 23 (9%); 365-day mortality: GA = 661 (16%) vs. SA = 424 (14%) vs. mixed = 41 (15%)

There was no difference between the groups.

Chu et al. 2015 [25]

Retrospective observational study

GA vs. NA (spinal/epidural)

n = 104.088 GA = 52.044 NA = 52.044

In-hospital mortality

Acute stroke, transient ischemic stroke, acute myocardial infarction, acute respiratory failure, acute renal failure

In-hospital death: GA vs. NA: 1.363 (2.62%) vs. 1.107 (2.13%), p < 0.001

The GA group had a greater percentage and higher odds of adverse in-hospital events than the NA group.

Fields et al. 2010 [9]

Retrospective observational study

GA vs. SA

n = 6.133 GA = 4.318 (72.6%) SA = 1.815 (27.4%)

30-day complications, 30-day mortality

none

SA vs. GA: 30-day mortality: (6.67 vs. 5.84, p = 0.21); overall complications: (45.75% vs. 48.97%, p = 0.001)

GA had a higher risk of 30-day complications compared to SA. There was no difference related to mortality.

Heidari et al. 2011 [33]

Randomised controlled study

GA vs. NA (EA/SA)

n = 270 GA = 197 NA = 190

30-day mortality, in-hospital mortality, Length of hospital stay, postoperative complications

None

30-day mortality: GA vs. NA: 0 vs. 2, p = 0.299; In-hospital mortality: GA vs. NA: 5 vs. 0, p = 0.107; Length of hospital stay: GA vs. NA: 4.3 (1.8) vs. 3.8 (1.6), p = 0.042

The length of hospital stay was significantly longer in the GA group. The morbidity and mortality rates were similar in both groups.

Helwani et al. n [10]

Retrospective observational study

GA vs.NA (SA/EA)

n = 12.929 GA = 7.826 (60.5%) NA = 5.103 (39.5%) [SA = 4.377 (85.8%); EA = 126 (2.5%)]

30-day mortality, LOS, deep surgical site infection (dssi), cardiovascular (cvc) -, pulmonary complications (pc)

None

NA vs. GA: dssi: (OR = 0.38; 95% CI = 3% to 7%; p < 0.001); LOS: (OR = 0.73; 95% CI = 0.68 to 0.89, p < 0.001); 30-day mortality: (OR = 0.78; 95% CI = 0.43 to 1.42; p > 0.05); cvc: (OR = 0.61; 95% CI = 0.44 to 0.85; p < 0.001); pc: (OR = 0.51; 95% CI = 0.33 to 0.81; p < 0.001)

NA was associated with a reduction in dssi rates, LOS, rates of postoperative cvc and pc. There was no difference in the mortality between NA and GA.

Karademir et al. 2015 [11]

Retrospective observational study

GA vs. SA

n = 11 GA = 30 (26%) SA = 85 (74%)

1-year mortality rate

None

RA vs. GA: p = 0.63

No significant difference in the 1-year mortality between GA and SA group

Karaman et al. 2015 [12]

Retrospective observational study

GA vs. NA (SA/EA)

n = 308 GA = 105 (34.1) NA = 203 (65.9%)

Overall-mortality

None

mortality rate: GA (n = 40) vs. NA (n = 37) (p < 0.01); OR: 2.761 (95% CI:1.62 to 4.69)

The mortality rate of patients receiving GA was higher than mortality rate of patients receiving NA.

Kim et al. 2013 [13]

Retrospective observational study

GA vs. SA vs. EA

n = 506 GA = 246 (48.62%) SA = 249 (49.21%) EA = 11 (2.17%)

30-day mortality, pulmonary complications (pc); cardiac complications (cc); Delirium

None

30-day mortality: GA = 7 (2.8%); SA = 4 (1.6%); EA = 0 (0.0%); p = 0.57; pc: GA = 91 (37.1%), SA = 74 (29.8%), EA = 3 (27.3%), p = 0.20; cc: GA = 4 (1.6%), SA = 2 (0.8%), EA = 0 (0.0%) p = 0.52; Delirium: GA = 31 (12.7%), SA = 27 (10.9%), EA = 1 (9.1%), p = 0.8

Methods of anaesthesia did not influence mortality and postoperative complications.

Le-Wendling et al. 2012 [14]

Retrospective observational study

GA vs. RA (single injection spinal, continuous spinal, continuous epidural) with or without continuous nerve block

n = 308 GA = 235 (76.30%) RA = 73 (23.70%)

In-hospital mortality, hospitalization costs (hc), Length of stay (LOS)

Re-hospitalization

hc: RA vs. GA ($ 16.789 + 631 vs. $ 16.815 + 643, p = 0.9557); LOS: 6.4 vs. 6.6 days, p = 0.64; in-hospital mortality: 2 (2.74) vs. 9 (3.83)

There was no difference in postoperative morbidity, rates of re-hospitalization, in-patient mortality or hc in patients receiving RA or GA.

Neuman et al. 2012 [27]

Retrospective observational study

GA vs. NA

n = 18.158 GA = 12.904 NA = 5.254

In-hospital mortality

Pulmonary and cardiovascular complications

In-hospital mortality: GA vs. NA: 325 (2.5%) vs. 110 (2.1%), p = 0.090

The mortality rate was similar between the two groups.

Neuman et al. 2014 [15]

Retrospective observational study

GA vs. NA (spinal/epidural)

n = 56.729 GA = 40.825 (72%) NA = 15.904 (28%)

30-day mortality

Length of stay (LOS)

30-day mortality: NA = 5.3%, GA = 5.4% (difference 0.1%; 95 CI −0.5 to 0.3; p = 0.55); LOS: RA = 6 days (95% CI: 6 to 6.1) vs. GA = 6.3 days (95% CI: 6.2 vs. 6.3), difference LOS: −0.2 days (95% CI: −0.3 `to 0.2; p < 0.001)

30-day mortality did not differ significantly between GA and NA. NA was associated with a shorter LOS.

Parker et al. 2015 [31]

Randomised controlled study

GA vs. SA

n = 322 GA = 164 (50.93%) SA = 158 (49.07%)

Mortality after 30, 90, 120 and 365 days

Surgical outcome, general complications, hospital stay (LOS)

30-day mortality: GA vs SA (4.9% vs. 3.2%; p = 0.57); 90 days: (7.3% vs 7.6%; p = 1.00); 120 days: (7.3% vs 7.6%; p = 0.55); 365 days: (11.7% vs 20.2%; p = 0.05); LOS in days (standard deviation): GA = 15.9 (13.7); RA = 16.2 (14.6); p = 0.75

No differences between GA and SA.

Patorno et al. 2014 [16]

Retrospective observational study

GA vs NA (spinal/epidural) vs. GA + NA

n = 73.284 GA = 61.554 (84.0%) NA = 6.939 (9.47%) GA + NA = 4.791 (6.53%)

In-hospital mortality

none

In-hospital mortality: GA vs. NA: (risk ratio 0.93, 95% CI 0.78 to 1.11)

Mortality risk did not differ significantly between GA and NA.

Rashid et al. 2013 [17]

Retrospective observational study

GA vs. NA (epidural/spinal)

n = 194 GA = 107 (55.15%) NA = 87 (44.85%)

Operating time, length of stay (LOS), blood loss, mortality

none

Operative time: GA = 1.54 ± 0.6, NA = 1.24 ± 0.39, p < 0.01; LOS: GA = 9.35 ± 9.0, NA = 8.63 ± 3.6, p = 0.484; blood loss: GA = 928 ± 360, NA = 912 ± 400, p = 0.758; mortality: GA = 4, NA = 5

There were no differences between LOS, blood loss and mortality. The only significant difference was in the operating time.

Shih et al. 2010 [18]

Retrospective observational study

GA vs. SA

n = 335 GA = 167 (49.85%) SA = 168 (50.15%)

perioperative morbidity, duration of surgery, length of stay (LOS), blood loss

none

GA vs. SA: duration: 165 min. vs. 150 min.; p < 0.001, LOS: 9 days vs. 8 days, p = 0.04); overall mortality: (5/167 [3%] vs. 2/168 [1.2%]; p = 0.25); overall morbidity: (21/167 [12.6] vs. 9/168 [5.4%]; p = 0.02)

GA increased the risk of postoperative morbidity in octogenarian patients after hip fracture repair. Patients with pre-existing respiratory diseases were especially vulnerable.

Seitz et al. 2014 [19]

Retrospective observational study

GA (inhalational, intravenous, GA combined with epidural or local anaesthesia) vs. SA

n = 20.973 GA = 8.818 (42.1%) SA = 12.155 (57.9%)

30-day mortality, 30-day postoperative medical complication, ICU till 7 days after surgery, length of stay (LOS)

none

GA vs. SA: 30-day mortality: GA = 691 (11.3%) vs. SA = 665 (10.8%), p = 0.44; ICU: GA = 371 (6%) vs.SA = 259 (4.2%), P < 0.001, 30-day postoperative medical complication: GA = 1.165 (19%), SA = 1.169 (19%) p = 0.92; LOS in days (± standard deviation): GA = 16.1 (20.2), SA = 16.0 (23.6), p = 0.72

GA and SA were associated with similar rates of most postoperative events.

Sevtap et al. 2013 [20]

Retrospective observational study

GA vs. SA vs. EA

n = 185 GA = 67 (36.21%) SA = 67 (36.21%) EA = 51 (27.58%)

7-day mortality, 30-day-mortality

Blood loss, blood transfusion, length of stay (LOS)

7-day mortality: GA = 3 (4.4%), SA = 2 (2.9%), EA = 1 (1.9%), p = 0.738; 30-day mortality: GA = 4 (1.4%), SA = 6 (5.9%), EA = 4 (5.8%), p = 0.805; LOS: GA = 13.6 ± 8.9, SA = 12.5 ± 5.2, EA = 15.7 ± 9.4, p = 0.228

There was no difference in the 7-day and 30-day mortality between the anaesthesia techniques. Further there were no differences in the other factors.

Tung et al. 2016 [24]

Retrospective observational study

GA vs. RA (epidural/spinal)

n = 17.189 GA = 6.063 (35.1%) NA = 11.153 (64.9%)

30-day all-cause mortality, 30-day all cause readmission, 30-day specific cause readmission

30-day mortality: GA = 104 (1.7%), NA = 189 (1.7%), p = 0.891, (OR 0.89, 95% CI [0.67 to 1.18], p = 0.409) 30-day readmission all-cause: GA = 771 (12.8%), NA = 1332 (12%), (OR 0.83, 95% CI 0.75 to0.93, p = 0.001), Surgical site infection readmission: (OR 0.69, 95% CI 0.49 to 0.97, p = 0.031)

There was no difference in the 30-day mortality between the two groups. NA is associated with a decreased 30-day all-cause readmission and surgical site infection readmission compared to GA

White et al. 2014 [21]

Retrospective observational study

GA vs. SA

n = 65.535 GA = 35.373 (53.97%) SA = 23.665 (36.11%)

30-day mortality

none

30-day mortality: GA = 1.066 (7.0%) vs. SA = 1.345 (7.3%); p = 0.053

No differences between GA and SA.

White et al. 2016 [26]

Retrospective observational study

GA vs. SA (with or without peripheral nerve block)

n = 11.085

GA = 985

GA with block = 4.364

GA + SA = 458

SA = 1.506

SA with block = 3.234

Unknown = 538

30-day mortality

Length of stay (IQR [range])

Intraoperative blood pressure

None

30-day mortality:

GA vs. SA: 53 (5.4) vs. 87 (5.8)

LOS: GA (n = 883): 13.2 (8.0-23.4 [1.3-165.8]) vs. SA (n = 1.319): 13.2 (8.0-22.8 [0.2-287.9])

There was no significant difference in the 30-day mortality and the length of stay between the groups.

Whiting et al. 2015 [22]

Retrospective observational study

GA vs. NA

n = 7.764 GA = 5.840 (75.2%) NA = 1.924 (24.8%) (SA = 1.813 (23.4%); nerve block = 111 (1.4%))

Minor complications, major complications, total complications within 30-day postoperative; 30-day mortality

none

SA vs GA: minor complications (OR: 1.43; CI 95%: 1.15 to 1.77; p = 0.001), major complications (OR: 1.01; CI 95%: 0.81 to 1.24; p = 0.950), total complications (OR: 1.24; CI 95%: 1.05 to 1.48; p = 0.014), 30-day mortality (OR: 1.20; CI 95%: 0.92 to 1.56; p = 0.169)

NA was associated with significantly greater odds of minor and total perioperative complications compared with GA.