These data demonstrate two major findings. First, non-cardiac morbidity is more common than cardiac morbidity, yet is associated strongly with higher preoperative cardiac risk as defined by the mRCRI. Secondly, mRCRI and POSSUM appear to be inadequate predictors of postoperative morbidity in individual patients for routine clinical practice.
Large epidemiological studies have shown that perioperative morbidity is associated with dramatic differences in post-discharge life expectancy across different operations and health systems.21
In the past, retrospective chart review has been frequently used to assess the type and frequency of complications. This approach has many methodological limitations including that the frequency and accuracy of reported complications might be inadequate for accurately assessing risk.22
POMS is the only published prospective method for describing short-term morbidity after major surgery and is a reliable and validated survey with high inter-rater reliability.10
POMS domains identify morbidity of a type and severity that could delay discharge from hospital and are recorded from indicators defining clinically important consequences rather than traditional diagnostic categories. The importance of prospectively gathering morbidity data in a defined, rigorous manner is borne out by the failure of administrative databases to satisfactorily describe postoperative complications.23,24
Our data demonstrate that in a surgical population undergoing orthopaedic operations performed by limited numbers of consultant surgical and anaesthetic teams, the RCRI is associated clearly with prospectively acquired, standardized measures of morbidity. We have also shown that morbidity is clearly associated with longer hospital stay, consistent with previous validation studies of POMS.10
In this high-volume orthopaedic population, the intrinsic limitation of patients' exercise capacity,25
plus the cost and expertise required by certain preoperative tests,11
may necessitate an alternative objective assessment of perioperative risk. A clear association was observed between pre-existing, chronic medical morbidity and the recognition by clinicians that postoperative care in a critical care environment was appropriate for many patients in the higher mRCRI groups. However, the association between higher mRCRI score, development of morbidity, and length of hospital stay persisted, regardless of immediate postoperative environment. Several other evidence-based factors that influence postoperative morbidity were similar between mRCRI groups. The high standard of anaesthetic management of these patients across mRCRI groups was demonstrated by normal postoperative temperatures and the attainment of haematocrit levels above those associated with increased risk of cardiac ischaemia.15,16
Our outcome data (hospital length of stay and in-hospital mortality) stand favourably with national4
comparisons. This includes data from recent large studies from the USA, where although the mean length of stay appears shorter, more than 48% of the patients are transferred to skilled nursing, intermediate care facilities.5
Furthermore, up to 9% US orthopaedic patients are readmitted to hospital within 3 months. After major hip or knee surgery in the US Medicare system, 10% patients were readmitted within 30 days after discharge, representing 1.5% of all surgical re-admissions.26
Failure to rehabilitate, pneumonia, postoperative infections, and gastrointestinal dysfunction (including bleeding) were among the commonest problems associated with re-admission. Thus, we observed similar patterns of morbidity reported in other hospitals and health-care systems27–30
who do not use POMS.
This is an observational, two-centre study with inherent limitations, notwithstanding that we conducted an adequately powered study that is the largest prospectively defined and collected set of morbidity data in a very specific, tightly defined surgical subpopulation. We cannot assume that all higher risk patients were appropriately treated medically before operation. Since no intervention was conducted, these data can only provide associative conclusions. Minor risk factors have not been shown to be independently associated with increased perioperative cardiac risk. Thus, the assumption that the minor risk factors are weighted appropriately requires further exploration. However, it is well recognized that abnormalities in the baseline ECG of randomly selected, asymptomatic populations are associated with subsequent all-cause mortality, including cardiovascular and coronary heart disease,31,32
independently of other risk factors.33
These data provide three significant clinical implications. First, the occurrence of early morbidity on POD 3 was associated strongly with length of stay within both high- and low-risk mRCRI groups. Nevertheless, these data demonstrate that higher mRCRI is associated with more overall complications and longer hospital stay even in the absence of early postoperative morbidity. Secondly, cardiovascular morbidity—specifically, cardiac ischaemia—was a relatively infrequently recorded morbidity domain. Individual mRCRI factors, even intermediate factors such as previous myocardial infarction or heart failure, were not associated with morbidity and length of stay. Rather, the composite mRCRI score, many domains of which are not directly related to cardiovascular morbidity, conferred greater risk of postoperative morbidity. This finding is consistent with the lack of specific cardiac morbidity and an overall constellation of multiple, chronic morbidities, perhaps reflecting general deconditioning and likely lack of cardiorespiratory reserve. Thirdly, the modest predictive value of mRCRI was similar to the performance of POSSUM for predicting morbidity, suggesting that other objective measures require exploration. By definition, POSSUM cannot perform as a true preoperative predictor of morbidity since the POSSUM score can only be calculated after surgery. Objective measures of cardiopulmonary reserve including cardiopulmonary exercise testing have been reported in the orthopaedic population but are often deemed to be impractical due to the limitations imposed by musculoskeletal pain, deformity, or both. In one small study, only ~60% of the patients with end-stage arthritis of the hip and knee were able to undertake conventional bicycle ergometry.25
Thus, stratifying risk on the basis of plasma biomarkers, such as brain natriuretic peptide,34
may confer additional predictive value.
In conclusion, greater preoperative cardiac risk is associated with an increased risk of non-cardiac morbidity and delayed discharge from hospital in elective orthopaedic surgery. Further work is required to ascertain whether perioperative interventions can be based upon the mRCRI. With increases in the prevalence of arthritis, obesity, and an ageing population, the need for measuring orthopaedic postoperative outcome/morbidity based upon appropriate risk stratification will be of great importance in driving quality health care.