Mortality and utilisation of critical care resources amongst high-risk surgical patients in a large NHS trust*


  • *

    This study was presented at the European Society of Intensive Care Medicine Annual Scientific Congress, Barcelona, 2006.

R. M. Pearse


Previous reports describe a population of non-cardiac surgical patients at high risk of complications and death. Outcomes are sub-optimal for such patients, perhaps in part related to inadequate provision or ineffective utilisation of critical care resources. In this study, data describing 26 051 in-patient non-cardiac surgical procedures performed in a large NHS Trust between April 2002 and March 2005 were extracted from local databases. Of these procedures, 2 414 (9.3%) were high risk with an overall mortality rate of 12.2% and a prolonged hospital stay (high-risk population median (IQR) 16 (9–30) days vs standard risk 3 (2–6) days). Mortality rates for specific procedures were consistent with UK averages. However, only 852 (35.3%) high-risk patients were admitted to a critical care unit at any stage after surgery. Of 294 high-risk patients who died, only 144 (49.0%) were admitted to a critical care unit at any time and only 75 (25.6%) of these deaths occurred within a critical care area. Mortality rates were high amongst patients discharged and readmitted to critical care (37.7%) and amongst those admitted to critical care following initial postoperative care on a standard ward (29.9%). These data suggest that the outcome of high-risk general surgical patients could be improved by adequate provision and more effective utilisation of critical care resources.

In a recent British study of 4.1 million selected non-cardiac surgical procedures, we were able to identify a high-risk surgical population which accounted for only 12.5% of in-patient surgical procedures but more than 80% of postoperative deaths [1]. This study confirmed the suspicion that low overall postoperative mortality rates conceal the existence of a large sub-population of patients at much greater risk of postoperative complications and death. Another important finding of this analysis was that fewer than 15% of high-risk surgical patients were admitted to a critical care unit at any time following surgery. The small number of patients who did receive this level of care were discharged after a median of only 24 h and subsequently lingered for many days on standard surgical wards (median (IQR) stay 16 (10–30) days). However, in this study more detailed analysis of critical care utilisation was not possible because not every critical care unit in the participating centres contributed data to the Intensive Care National Audit & Research Centre (ICNARC) database.

Recent clinical trials have shown that substantial improvements in outcome may be achieved for high-risk surgical patients through the use of protocolised cardiorespiratory therapies delivered in a critical care setting [2–5]. However, continued debate over both the size of the high-risk surgical population and the potential benefit of admitting such patients to critical care represent important obstacles to the introduction of such therapeutic approaches into routine practice. We therefore decided to examine in more detail the role of critical care in the management of high-risk surgical patients in the UK. We used methods similar to those employed in the previous national audit, but conducted the present study in a single institution, so that individual patient data could be cross-referenced using additional local databases and where necessary hospital notes. This analysis also provided an opportunity to confirm or refute the existence of a definable population of high-risk surgical patients within UK hospitals. The aim of this study was to identify patients undergoing selected high-risk non-cardiac surgical procedures within a large NHS Trust and describe critical care resource provision and utilisation for this population in relation to outcome.


The study was performed in a large NHS Trust incorporating two teaching hospitals which between them provide all major secondary and tertiary surgical services. This service evaluation was conducted with the approval of the local audit committee. One hospital has 114 surgical beds with a 13-bed intensive care unit which provides postoperative care for cardiac and other surgical patients as well as medical intensive care for oncology and HIV-positive patients. The other hospital has 214 surgical beds with a 16-bed intensive care unit which provides care for medical, non-cardiac surgical and neurosurgical patients and a six-bed high dependency unit providing care for non-cardiac surgical patients including a number of patients admitted following traumatic injury. Using an analogous design to the previous national study [1], data were extracted from two databases. The local Hospital Episodes Statistics (HES) database [6], is maintained by clerical coding staff and includes data on all clinical activity within the Trust. Validation is performed locally by Trust information managers. The intensive care audit database is maintained locally by dedicated intensive care audit staff and contributes data to the ICNARC case mix programme. These data are subject to local and central internal error checks [7].

Data were extracted for all adult (age ≥ 18 years) surgical admissions to hospital and to critical care between 1 April 2002 and 31 March 2005. These dates were selected to provide the maximum amount of data whilst avoiding any bias arising from changes to the data collection process. All data relating to length of stay was calculated by subtracting the discharge date from the admission date. Surgical specialities with highly developed treatment pathways (often including critical care) such as cardiothoracic surgery and neurosurgery were excluded. Surgical specialities analysed included general and vascular surgery, orthopaedics, gynaecology, ear nose and throat, plastic surgery, urology, maxillofacial and oral surgery. Admissions involving endoscopy, day case surgery, cardio-thoracic surgery, neurosurgery, organ transplantation, obstetrics or the surgical management of burns were excluded. There are 6 920 surgical procedure codes in the Office of Population, Censuses and Surveys (OPCS) classification [8]. Surgical admissions to hospital were identified in the HES database by the presence of one of 4 910 codes that satisfied the inclusion criteria. Where more than one surgical procedure was performed during the same hospital admission, only the first procedure was included in the analysis. Several alternative OPCS codes may exist for any given procedure [8]. In order to reduce bias arising from discrepancies in the coding process, procedures were categorised into elective (includes elective and scheduled cases according to ICNARC definition) and non-elective (includes urgent and emergency cases according to ICNARC definition). Healthcare Resource Groups (HRG) are based on clinical similarity and resource homogeneity [8]. The codes, examples of which are presented in the report of our previous national study [1], may specify the presence of a complicating medical condition, complexity of surgery or a particular age group. HRGs were then ranked according to mortality rates. High risk surgical procedures were prospectively defined as those procedures included in a HRG with a mortality rate of 5% or more. The remaining procedures were classified as standard risk.

Surgical admissions to critical care units were identified in the ICNARC database by source of admission, and were only included if the primary reason for admission was not an excluded surgical procedure. Critical care admissions were prospectively divided into admissions directly from the operating theatre and admissions to critical care following a period of postoperative care on a standard ward. Data describing critical care resource use was verified through a manual check of individual patient details using the ICNARC database.

Data are presented as median (IQR). Categorical data were tested with Fisher’s exact test and continuous data with the unpaired t-test with Welch’s correction. Analysis was performed using GraphPad Prism version 4.0 (GraphPad Software, San Diego, CA, USA). Significance was set at p < 0.05.


Hospital admissions

During the 36 months of the study, there were 26 051 hospital admissions involving one of the selected non-cardiac surgical procedures, with 390 (1.5%) deaths. The median age was 46 (32–63) years and 13 169 (50.6%) patients were male. There were 16 294 elective surgical admissions with 65 (0.4%) deaths and 9 757 non-elective surgical admissions with 325 (3.33%) deaths in hospital. The duration of hospital stay was shorter for elective admissions when compared to non-elective admission (3 (2–7) days vs. 4 (2–12) days; p < 0.0001).

Out of 427 HRGs, 69 were associated with a mortality rate of 5% or greater. From these, 2 414 (approximately 800 per annum) individual high risk surgical procedures were identified with a total of 294 (12.2%) deaths (Figure 1). The high-risk surgical population accounted for 75.4% of deaths in hospital but only 9.3% of admissions. Complex or major surgery, advanced age, the presence of a complicating medical condition or a combination of these was specified by 40 (58%) of the 69 high risk HRG codes compared to 119 (33%) of 358 standard-risk HRG codes. Although less than 10% of patients were classified as high risk, this population utilised 46 138 in-patient bed days (23% of total in-patient bed days).

Figure 1.

 Standard and high risk populations of patients undergoing selected non-cardiac surgical procedures. Data presented as median [IQR] or absolute values (%). LOS; length of hospital stay. *p < 0.0001 between groups.

Critical care admissions

There were 1 627 surgical admissions to critical care which met the inclusion criteria with 297 (18.3%) deaths. Of these patients, 1 470 (90.4%) were admitted directly to critical care from the operating theatre. Two hundred and fourteen patients were discharged prematurely from critical care because of bed shortages (defined according to ICNARC criteria) with 32 (15.0%) deaths. Of the 297 patients who were admitted to critical care (either immediately postoperatively or subsequently) and later died, 188 (63.3%) did so after initial discharge from critical care. Sixty-one patients who were discharged from critical care were subsequently readmitted, of whom 23 (37.7%) died. One hundred and fifty-seven patients were admitted to critical care following an initial period of postoperative care on a standard ward, of whom 47 (29.9%) patients died. The median duration of standard ward care for these patients was 2 (1–3) days for elective patients and 2 (1–7) days for emergency patients. A total of 775 surgical patients classified by our criteria as standard risk were admitted to a critical care facility postoperatively (Figure 2). The overall mortality rate for these patients was 4.5%. Within this was a subgroup who were admitted to critical care after a period on a standard ward of whom 29 (21.2%) died; the duration of standard ward care for this subgroup was 2 (1–5) days.

Figure 2.

 Sub-groups of the standard risk population of surgical patients. Data presented as median (IQR) or absolute values (%).

Critical care resource utilisation by the high-risk surgical population

Patients were classified as high risk on the basis that they underwent a procedure with an overall mortality rate of 5% or greater. Only 852 of these high-risk patients were admitted to a critical care unit at any stage during their hospital admission (35.3% of the overall high-risk population) (Figure 3). Of the 294 high-risk patients who died, the median time from hospital admission to the first surgical intervention was 2 (1–3) days for elective patients and 3 (1–11) days for non-elective patients. The median time between hospital admission and death was 16 (6–32) days for elective patients and 22 (7–43) days for non-elective patients. For those admitted to critical care, the median time between hospital admission and death was 28 (12–48) days compared to 15 (3–34) days for those not admitted to critical care. Only 144 (49.0%) high-risk patients who died were admitted to a critical care area at any stage during their admission and only 75 (25.6%) of these deaths occurred in a critical care area.

Figure 3.

 Sub-groups of the high-risk population of surgical patients. Data presented as median (IQR) or absolute values (%). LOS; length of hospital stay.


The principal finding of this study is that, despite having an overall mortality of 12% and accounting for over 70% of post-operative deaths, only one-third of high-risk surgical patients were admitted to critical care at any stage following surgery. Our findings suggest that the failure to admit such patients to critical care immediately after surgery contributes to the poor outcomes for the high risk group as a whole. Premature discharge from, and the need re-admit patients to critical care were both factors associated with a particularly high mortality rate. Similarly, mortality was high amongst patients admitted to intensive care following an initial period of postoperative care on a standard ward (30%). The median duration between surgery and critical care admission for this group was only 2 days. It seems likely that outcomes would be improved for these patients if they were admitted to critical care immediately after surgery to receive protocolised cardio-respiratory care. This approach seems to be very effective following cardiac surgery and provides the opportunity for use of additional interventions which may improve survival still further [2–5]. Finally only 49% of the total population of high-risk surgical patients who died were admitted to a critical care facility at any stage and only 26% died in a critical care area. These observations indicate a failure to provide an appropriate level of care even to the highest risk patients.

Whilst the number of high-risk patients admitted to a critical care unit in our institution (35%) compares favourably with our national study (15%) [1], the findings of both studies clearly indicate that this resource is underprovided and/or under-utilised. This conclusion is supported by the observation that postoperative critical care admission is routine for cardiac surgical patients, who have a high incidence of comorbid disease and yet undergo major surgery with an overall mortality rate in the UK of only 3.5% [9]. Although in many instances limited resources prevent admission of high-risk surgical patients to critical care, it is unclear whether the lack of critical care facilities is compounded by clinicians’ underestimation of risk or doubts over the potential benefits of this facility. Whilst the mortality rate for standard risk patients admitted to critical care was significant (4.5%), suggesting clinicians are able to identify patients at greater risk, it is interesting to note that most such patients underwent elective surgery whilst our data suggest that it is the emergency surgical population that is in greatest need of critical care admission. This raises the possibility that involvement of a more senior clinician may be a key factor in the decision to admit a patient to critical care following surgery.

The findings of this single centre study are consistent with published data describing the high-risk surgical population [1, 10–22]. The size and nature of the high-risk population in our institution is similar to that identified in the national UK study and although it might appear that a greater proportion of high-risk surgical patients are admitted to a critical care area in our Trust, this observation may also be explained by an underestimate of critical care admissions in the national study [1]. In keeping with the current study, reports from the National Confidential Enquiry into Perioperative Deaths (NCEPOD) also describe advanced age, comorbid disease, major and urgent surgery as the key characteristics of the high-risk patient [12, 13]. The high mortality rates and prolonged hospital stays described for high-risk patients in this study are also consistent with published data from Europe and North America [10, 23–25].

The use of healthcare databases to test clinical hypotheses is associated with a number of important limitations. In particular, HES data may underestimate mortality rates because the coding process is not designed to capture detailed mortality data [12, 13, 26]. In the current analysis cross-referencing data describing individual patients between the two databases has allowed us to minimise this source of error. Ideally, prospective studies should be performed to evaluate this issue in more detail. Such a study would allow identification of the reasons underlying decisions regarding the allocation of critical care resources to individual patients.

This study confirms that the high-risk surgical population accounts for the majority of postoperative complications and deaths. Although readily identified, only around one third of such patients are admitted to a critical care unit at any stage after surgery. Where critical care admission does occur it is often for short periods, whilst the duration of hospital stay is prolonged. The allocation of additional critical care facilities to this population is likely to be a cost-effective means of improving both short- and long-term outcomes.


The authors would like to thank Gail Marshall and Jovin Synott for their assistance with data extraction. There was no external funding source for this research. The authors have no conflicts of interest.