Revised cardiac risk index and postoperative morbidity after elective orthopaedic surgery: a prospective cohort study

Abstract

Background: The revised cardiac risk index (RCRI) is associated strongly with increased cardiac ischaemic risk and perioperative death. Associations with non-cardiac morbidity in non-cardiac surgery have not been explored. In the elective orthopaedic surgical population, morbidity is common but preoperative predictors are unclear. We hypothesized that RCRI would identify individuals at increased risk of non-cardiac morbidity in this surgically homogenous population.

Methods: Five hundred and sixty patients undergoing elective primary (>90%) and revision hip and knee procedures were studied. A modified RCRI (mRCRI) score was calculated, weighting intermediate and low risk factors. The primary endpoint was the development of morbidity, collected prospectively using the Postoperative Morbidity Survey, on postoperative day (POD) 5.

Results: Morbidity on POD 5 was more frequent in patients with mRCRI ≥ 3 {relative risk 1.7, [95% confidence interval (CI): 1.4-2.1]; P<0.001}. Time to hospital discharge was delayed in patients with mRCRI score ≥ 3 (log-rank test, P=0.0002). Pulmonary (P<0.001), infectious (P=0.001), cardiovascular (P=0.0003), renal (P<0.0001), wound (P=0.02), and neurological (P=0.002) morbidities were more common in patients with mRCRI score ≥ 3. Pre/postoperative haematocrit, anaesthetic/analgesic technique, and postoperative temperature were similar across mRCRI groups. There were significant associations with hospital stay, as measured by the area under the receiver-operating characteristic curves for mRCRI 0.64 (95% CI: 0.58-0.70) and POSSUM 0.70 (95% CI: 0.63-0.75).

Conclusions: mRCRI score ≥ 3 is associated with increased postoperative non-cardiac morbidity and prolonged hospital stay after elective orthopaedic procedures. mRCRI can contribute to objective risk stratification of postoperative morbidity.

Figures

Fig 1

Preoperative mRCRI score and POMS-defined morbidity at any time point on PODs 3, 5, 8, and 15. Day 3 POMS was not collected in the Edinburgh cohort. The percentage of patients in each mRCRI group sustaining morbidity is shown. P-values refer to differences between mRCRI groups (Fisher's exact test).

Fig 2

Individual morbidity domains recorded at any pre-defined POMS data collection time point according to the mRCRI group. Day 3 POMS was not collected in the Edinburgh cohort. Note that variable prevalence between different morbidity domains. P-values refer to differences between mRCRI groups (Fisher's exact test).

Fig 3

Time for patients to become morbidity-free after operation according to the preoperative mRCRI category. (a) mRCRI groups ≥3 sustained longer periods of postoperative morbidity (P=0.01). (b) Time to become morbidity-free after operation according to the preoperative mRCRI score ≥3 vs <3.

Fig 4

Length of hospital stay according to the mRCRI group. (a) Median (25th–75th percentiles) length of stay for each preoperative mRCRI score. Asterisks denote differences between mRCRI groups (P=0.03). (b) Time-to-discharge analysis showing relationship between each preoperative mRCRI score and length of hospital stay (until 30 days after operation). (c) Length of stay was persistently higher in morbidity-free patients with preoperative mRCRI ≥3. For clarity, discharge rates for mRCRI grouped into scores ≥3 or <3 are shown as for (b), for 30 days after operation (P=0.01).

Fig 5

Time-to-discharge analyses demonstrating the impact of morbidity and mRCRI group on length of hospital stay. The presence of any POMS-defined morbidity on POD 3 (a: log-rank test, P=0.005) and POD 5 (b: log-rank test, P=0.01) was associated with prolonged hospital stay, within each mRCRI group. Length of stay was persistently higher in morbidity-free patients with mRCRI ≥3 on both PODs 3 and 5.