Letters

Suboptimal ward care of critically ill patients

Suboptimal care should have been defined

Assessment of quality of care was flawed

Active management should prevent cardiopulmonary arrests

Inadequate staffing means problems are missed

Doctors don't review patients that nurses identify as highly dependent

Checklist may help improve referral

More intensive care beds are needed

Medical training should focus on basic skills

Course is available for surgical trainees

Medical emergency teams improve care

Authors' reply

Suboptimal care should have been defined

EDITORMcQuillan et al show that most patients receive suboptimal management of oxygen therapy, airway, breathing, circulation, and monitoring before admission to intensive care.¹ In an area of medicine renowned for objective measurement it is surprising that this study should rely on the subjective opinions of two assessors about what constituted suboptimal care. Understandably, their opinions often disagreed.

The authors accept that there are difficulties in relying on assessors' opinions, but we must not underestimate these limitations. The assessors knew the outcomes of the patients, which must have biased their opinions, particularly since suboptimal care is not defined. How suboptimal care was defined is crucial to the paper's message, and more information about the data evaluated by the assessors would have been preferable to the lengthy discussion, much of which was not directly related to the data.

Unfortunately, many of the data are self fulfilling. It is unsurprising that the suboptimally managed group scored badly on oxygen therapy and airway, breathing, and circulation and that 67% of this group were late admissions to intensive care since these were presumably the factors used to determine suboptimal management.

Nevertheless, a key message is that most of the well managed patients were admitted to intensive care units within the first day of admission, with presumably some going straight from accident and emergency. These acutely ill patients are perhaps more easily identifiable as going to need intensive care. Conversely, those patients who arrived at hospital less ill and who deteriorated while on general wards were those who received suboptimal care. There was a longer time between admission to hospital and admission to intensive care in these patients. We are not told if any of the admissions to intensive care were delayed because of lack of beds. Although there is no excuse for suboptimal care, sometimes admission to intensive care is requested because a ward with overstretched nursing staff and no high dependency beds recognises that it is unable to provide optimal care for an acutely ill patient.

David Gorard, Consultant physician.  .
Wycombe Hospital, Buckinghamshire HP11 2TT

1. McQuillan P, Pilkington S, Allan A, Taylor B, Short A, Morgan G, et al. Confidential inquiry into quality of care before admission to intensive care. BMJ 1998; 316: 1853-1858[Abstract/Free Full Text]. (20 June.)

Assessment of quality of care was flawed

EDITORMcQuillan et al made striking claims about avoidable admissions and the contribution of suboptimal care to subsequent mortality and morbidity on the basis of a study which was deeply flawed in two crucial respects.¹

Firstly, the research relied on implicit judgments of the quality of care made by two external assessors, who were presented with data abstracted from the clinical records. The authors argued that they had to use implicit assessments of the quality of care because it was too difficult to set out objective or explicit definitions of what constituted suboptimal care. If it is hard to define explicit quality standards or criteria, it will be equally hard to reach a valid and reliable implicit assessment of the quality of care. The extensive literature on implicit reviews suggests that their interrater reliability is very mixed. ^{2 3} The kappa statistics cited in this study, ranging from 0.42 to 0.53, would be regarded as at best indicating moderate reliability.⁴ The authors could have increased the reliability of the assessments by using more assessors for each case and by undertaking some training and feedback of results to assessors before the study.

Secondly, the two assessors who made the judgments about the quality of care were apparently aware of the eventual outcomes in each case. In other words, they knew about subsequent morbidity and mortality when they were making judgments about the quality of care. Implicit judgements about the quality of care are likely to be inappropriately influenced by knowledge of eventual outcome. Assessors are more likely to rate the care as suboptimal if they are told that the patient died, even though the process of care is unchanged.⁵ This means that the association between assessors' ratings of the quality of care and patients' subsequent mortality, which is made much of in the paper, may simply be an artefact of the methods used.

If implicit professional judgments about the quality of care are to be used in future, the reliability and validity of those judgments should be more rigorously examined. More information about the training of assessors should be sought, better evidence of interrater reliability should be presented, and implicit reviews of the process of care should be blinded to the subsequent process and outcome to avoid bias. Because implicit and explicit review methods each have advantages and disadvantages, it may be advisable to use both and compare their results rather than to opt for one or the other.

Kieran Walshe, Senior research fellow. 
Health Services Management Centre, University of Birmingham, Birmingham B12 2RT

1. McQuillan P, Pilkington S, Allan A, Taylor B, Short A, Morgan G, et al. Confidential inquiry into quality of care before admission to intensive care. BMJ 1998; 316: 1853-1858. (20 June.)
2. Brook RH, Appel FA. Quality of care assessment: choosing a method for peer review. N Engl J Med 1973; 288: 1323-1329.
3. Koran LM. The reliability of clinical methods, data and judgements. N Engl J Med 1975; 293: 695-701[Medline].
4. Brennan P, Silman A. Statistical methods for assessing observer variability in clinical measures. BMJ 1992; 304: 1491-1494.
5. Caplan RA, Posner KL, Cheney FW. Effect of outcome on physician judgements of appropriateness of care. JAMA 1991; 265: 1957-1960[Abstract].

Active management should prevent cardiopulmonary arrests

EDITORGarrard and Young¹ asked whether McQuillan and colleagues' findings were representative of care across the United Kingdom.² We conducted a similar study in a Manchester teaching hospital aimed at identifying the incidence of preventable physiological deterioration before cardiopulmonary arrest on general medical and surgical wards.³

We analysed 47 consecutive arrests and found abnormal vital signs in 24 patients during the 24 hours before the arrest call was made. Appropriate tests were performed but results were often not acted on; senior staff were consulted before arrest in only six cases. Two patients were referred for intensive care before arrest; both were deemed unsuitable. Most importantly, cardiopulmonary resuscitation was largely unsuccessful. Nine of the 47 patients survived the arrest, and five went home alive. In patients with premonitory signs, only three survived the arrest and none left hospital.

Though we approached the subject from a different angle, our findings support and complement those of McQuillan et al. In over half our patients the arrest was preceded by a more gradual physiological decompensation and therefore opportunity existed for intervention. Ward staff need to appreciate the importance of abnormal signs and investigations and seek help promptly from experienced clinicians. Intensive care may be appropriate but is more likely to benefit patients if they are referred early. We believe that some of the cardiopulmonary arrests in our survey could have been prevented. The proposed medical emergency team would have been invaluable in assessing these patients.

We agree that a new model of treatment of critically ill ward patients is required with emphasis on early referral and treatment. However, some patients who are approaching cardiopulmonary arrest are so sick that cardiopulmonary resuscitation will not succeed and intensive care would be inappropriate. We would urge earlier, wider consideration of "do not attempt resuscitation" orders in this group. The trend should be towards proactive management, either to expedite referral for intensive care for those who need it or to allow a dignified death for those who are destined to die in any case. Too often we see a haphazard trial of cardiopulmonary resuscitation followed by hasty referral to intensive care. This is inhumane, futile, costly, and demoralising.

Jeremy Wood, Specialist registrar. 
Department of Anaesthetics, North Manchester General Hospital, Manchester M8 5RB

Andrew Smith, Consultant anaesthetist. 
Department of Anaesthesia, Royal Lancaster Infirmary, Lancaster LA1 4RP

1. Garrard C, Young D. Suboptimal care of patients before admission to intensive care. BMJ 1998; 316: 1841-1842[Free Full Text]. (20 June.)
2. McQuillan P, Pilkington S, Allan A, Taylor B, Short A, Morgan G, et al. Confidential inquiry into quality of care before admission to intensive care. BMJ 1998; 316: 1853-1858. (20 June.)
3. Smith A, Wood J. Can some in-hospital cardiopulmonary arrests be prevented? Resuscitation (in press).

Inadequate staffing means problems are missed

EDITORAs an intensive care resident, I was unsurprised by the results of McQuillan et al.¹ Their recommendations for improving management of patients before intensive care, including the medical emergency team and better training, should all be supported. A problem not dealt with is detecting the acute physiological disturbance in the first instance.

I was on call for anaesthesia last weekend. On Sunday the preregistration house officer for surgery had 102 ward patients to look after. The medical preregistration house officer was caring for 114 patients, with the help of one half of a senior house officer. Even with the best acute medical emergency training these doctors cannot be proactive in the care of this number of patients, most of whom they have never met before. Under such pressure these doctors can only react to problems identified to them. We now seem to rely on the ward nurses to call the "physiology police," but with more than eight patients per trained nurse on the medical and surgical wards, detection of something physiologically abnormal is not reliable. I am sure this hospital is not unique in this situation.

To have any chance of improving the quality of acute medical care on general wards there must be either fewer patients or more medical and nursing staff. Treatment cannot start until the patient's acute problem is identified.

Paul J Youngs, Specialist registrar in anaesthesia. 
Royal Devon and Exeter Hospital, Exeter EX2 5DW

1. McQuillan P, Pilkington S, Allan A, Taylor B, Short A, Morgan G, et al. Confidential inquiry into quality of care before admission to intensive care. BMJ 1998; 316: 1853-1858. (20 June.)

Doctors don't review patients that nurses identify as highly dependent

EDITORMcQuillan and colleagues report suboptimal care before admission in 54% of patients admitted to intensive care units in two hospitals.¹ In 1993 we performed an audit at our hospital to assess the number of patients in selected general wards who would be more appropriately cared for in intensive care or a high dependency unit. The survey was performed daily over two weeks and included general medical, general surgical, and cardiology wards. The night sister initially identified the most dependent patients. During the study 56 assessments were made of 39 patients. We recorded the grades of medical staff attending the patients and the frequency with which the patients were seen. Severity of nursing workload was assessed with the therapeutic intervention score.

Requirement for more nursing was given as the reason for referral for assessment in 34 (87%) patients. Twenty eight were thought to require more monitoring, and 18 were thought to require more intensive treatment or organ support. The nursing staff directly looking after each patient were then asked to judge whether admission to intensive care or high dependency units was required. In 20 of 56 cases (36%) nurses thought admission was necessary. There were significant overlaps between the therapeutic intervention scores of ward patients judged to require high dependency or intensive care (13-36), ward patients judged not to require such care (11-32), and patients in intensive care units at the time of the audit (24-70).

Of the patients identified as requiring more intensive care by the nursing staff, only 11 (55%) were reviewed daily by a consultant and only four (20%) were reviewed at least four to six hourly by a registrar or consultant.

These data suggest that nursing staff on general medical and surgical wards identify a significant number of patients whom they feel warrant admission to a high dependency or intensive care unit. Worryingly, most of such patients identified during this audit were not reviewed regularly by experienced medical staff.

Tim Ringrose, Specialist registrar. 
Christopher Garrard, Director. 
Intensive Care Unit, John Radcliffe Hospital, Oxford OX3 9DU

1. McQuillan P, Pilkington S, Allan A, Taylor B, Short A, Morgan G, et al. Confidential inquiry into quality of care before admission to intensive care. BMJ 1998; 316: 1853-1858. (20 June.)

Checklist may help improve referral

EDITORMcQuillan et al¹ highlighted an important question facing hospitals todaynamely, how can patients receive a tailored continuum of care in the face of the effects of Calman training and the pressure to reduce ward nursing numbers and grades? Their recommendations, although exhaustive, are not all achievable within an acceptable time frame. Individual hospitals must initially find a solution that is locally achievable within present resources.

After a critical incident involving a patient admitted through the accident and emergency unit to a medical ward and belatedly referred to intensive care we compiled a list of conditions for which senior medical and intensive care advice must be sought (box). Unlike most guidelines these do not dictate a clinical pathway but serve as a trigger for more senior involvement in the management of patients at an earlier stage. A second major difference was the involvement of intensive care staff for patients that may not necessarily require ventilation but need correction of their physiological parameters. Since the implementation of these guidelines referrals to intensive care have been earlier and appropriate. We plan to augment this list with physiological variables² and distribute it to the acute medical and surgical wards.

With the increasing subspecialisation of general medicine the management of medical emergencies has been sidelined. This has occurred at a time when the specialty of accident and emergency medicine is beginning to come of age. All undiagnosed emergency patients should be admitted to hospital through accident and emergency departments so that an accurate assessment and appropriate transfer can be made.

The recent disquiet at unfavourable clinical outcomes makes it increasingly untenable to rely on cardiac arrest teams and intensive care units to salvage ward patients near to death. Time to put systems in place to ensure the matching of health care to the continuum of illness is one thing we do not have.

C McAllister, Lead clinician, intensive care unit. 
S J McGovern, Consultant in accident and emergency medicine. 
Craigavon Area Hospital Group Trust, Portadown BT63 5QQ

1. McQuillan P, Pilkington S, Allan A, Taylor B, Short A, Morgan G, et al. Confidential inquiry into quality of care before admission to intensive care. BMJ 1998; 316: 1853-1858. (20 June.)
2. Ridley SA. Intermediate care. Possibilities, requirements and solutions. Anaesthesia 1998; 53: 654-664[Medline].

More intensive care beds are needed

EDITORMcQuillan and colleagues show that patients often receive suboptimal care before admission to intensive care.¹ We would like to highlight another factor adversely affecting the care of patients. This is the frequency with which hospitals cannot admit patients to their own intensive care unit because of a shortage of staffed and available beds.

In the Northwest region each day an average of three patients are transferred to another intensive care unit. This can rise to nine a day during peak periods. All intensive care units in the Northwest region are contacted four times daily by the Intensive Care Bed Information Service to ascertain bed availability. When only 10 of the 183 adult general intensive care beds remain available an amber alert is declared by the NHS Executive Northwest Regional Office, and this information is faxed to all trusts. When only five beds remain a red alert is declared. During June amber alert conditions were met 17 times and there were six occasions when a red alert could have been issued. The true situation is worse since paediatric and specialist services are not included and there are no alerts at night or weekends.

We audit transfers against published standards ^{2 3} and over the past two years have clearly shown that transfers are increasingly caused by a lack of staffed intensive care beds in the host hospital. Transfers for this reason have increased by 300% in Greater Manchester and 200% in the rest of the Northwest.

Despite close liaison with local and regional managers, the health authorities appear unable to address the fundamental issue of insufficient investment in intensive care and high dependency units in the northwest of England. Political direction is aimed at reducing waiting times for elective surgery.

Unless McQuillan and colleagues' strategy to improve the care of the acutely ill patient succeeds, in the absence of sufficient high dependency and intensive care beds, it seems inevitable that patients will continue to be transferred unnecessarily.

Peter W Duncan, Chairman, Association of North Western Intensive Care Units. 
Royal Preston Hospital, Preston PR2 4HT

Peter Nightingale, Secretary, Association of North Western Intensive Care Units. 
Withington Hospital, Manchester M20 2LR

Ian Macartney, Clinical adviser, Intensive Care Bed Information Service. 
North Manchester General Hospital, Manchester M8 6RB

Johanna Ryan, Regional intensive care audit coordinator. 
Bolton General Hospital, Bolton BL4 0JR

Maire P Shelly, Local adviser in intensive care medicine. 
Withington Hospital, Manchester M20 2LR

1. McQuillan P, Pilkington S, Allan A, Taylor B, Short A, Morgan G, et al. Confidential inquiry into quality of care before admission to intensive care. BMJ 1998; 316: 1853-1858. (20 June.)
2. Intensive Care Society. Guidelines for transport of the critically ill adult. London: ICS, 1997:6.
3. Royal College of Anaesthetists. Basic specialist training guide. London: RCA , 1991.

Medical training should focus on basic skills

EDITORTwo recent articles have suggested that care is suboptimal on NHS general medical and surgical wards. ^{1 2} McQuillan et al showed that over 50% of admissions to intensive care may have been avoidable with improved care in the preceding hours and days. Smith and Power² reviewed a recent Audit Commission report that showed problems with provision of pain control after surgery. Both suggest that changes in organisation and service provision are required.

The common denominator in these (and many other) issues is not a lack of care but an inability of nursing and medical staff to give effective treatment. One aspect of this is insufficient resources. Effective monitoring, treatment, and review of acutely ill and postoperative patients takes considerable time. This time is not available within the current funding. Many are already fighting to improve this situation.

The second and perhaps more fundamental aspect is that of training. McQuillan and others have noted poor application of fundamental principles of airway, breathing, and circulation; pain control; physiology; etc. Care of emergencies and basic acute care, postoperative care, and pain control are bread and butter for senior and preregistration house officers. Yet we are increasingly seeing how inadequately medical training prepares us for this.

Rather than increasing consultant input and specialist teams, surely it would be more effective to train our medical students in these skills early on. Current training prepares well for exams but leaves students ill prepared for meeting the needs of patients. I had minimal practical training in spotting the signs of a patient in physiological decline. Thus junior doctors may discuss the intricacies of the surgery on the consultant ward round while the patient travels further into renal failure. This is not a failure of care by them (although would be seen as such by the public and the court) but of their training.

I had to wait six years after qualifying to have the opportunity to be taught how to recognise a sick child and to give the treatment needed while waiting for further help. Most medical students can quote all the causes of polyarteritis nodosa (which they may never see) but few of electromechanical dissociation (which they will see often). This list is almost endless.

Although pain, intensive care, and anaesthetic specialists will always be required to intervene with ward patients, they should need to be called only when basic measures are already well under way. Most aspects of basic monitoring; maintaining airways, breathing, and circulation; fluid management; and pain control should be well within the ability of properly trained students by the time of qualification.

Carl Pritchard, Senior house officer in anaesthesia. 
Southampton University Hospitals NHS Trust, Southampton SO16 6YD

1. McQuillan P, Pilkington S, Allan A, Taylor B, Short A, Morgan G, et al. Confidential inquiry into quality of care before admission to intensive care. BMJ 1998; 316: 1853-1857. (20 June.)
2. Smith G, Power I. Audit and bridging the analgesic gap. Anaesthesia 1998; 53: 521[Medline].

Course is available for surgical trainees

EDITORMcQuillan et al and the accompanying editorial document current deficiencies in critical care. ^{1 2} Efforts to improve surgical critical carethat is, management of emergencies and unexpected complications and perioperative care of patients having major surgeryhave been under way for some time. Four years ago, the Royal College of Surgeons of England commissioned a working party of intensivists, anaesthetists, and surgeons to develop a consensus programme to improve training in surgical critical care for junior doctors. A practical three day course on the care of the critically ill surgical patient (CCrISP) has been developed which deals specifically with many of the deficiencies identified in the articles.

The aim of the course is to try to prevent surgical patients deterioratingoften to the point where they require intensive careby identifying and correcting problems early. The course emphasises the use of a system of assessment to avoid simple errors which account for many avoidable adverse episodes. The system begins with the correction of airway, breathing, and circulation but moves rapidly on to the identification and treatment of the underlying cause. Candidates learn this system, discuss it in a range of realistic clinical scenarios, and then practise it on simulated patients. Candidates read a course manual beforehand and through lectures and practical sessions cover theory and practice necessary for surgical critical care in the ward or high dependency unit. Topics include monitoring techniques, nutrition, sepsis, renal failure, communication, and pain management in addition to detailed control of airway, breathing, and circulation. Calling for help and seeking timely senior input is emphasised throughout.

The college has run the course successfully for two years, and it has now been established at Hope Hospital in Manchester and in Leeds. Feedback from candidates three months after their course shows that 85% were influenced considerably in their approach to critically ill patients and that 90% used the advocated system of assessment frequently. Six other centres are establishing the course in their region. Many postgraduate deans have indicated their support, and the college has advised that the course is highly recommended for all basic surgical trainees. Trainees from other disciplines may benefit from similar courses.

Iain D Anderson, Hillsborough tutor in critical care. 
Brian J Rowlands, Chairman of critical care working party. 
Raven Department of Education, Royal College of Surgeons of England, London WC2A 3PN

1. McQuillan P, Pilkington S, Allan A, Taylor B, Short A, Morgan G, et al. Confidential inquiry into quality of care before admission to intensive care. BMJ 1998; 316: 1853-1858. (20 June.)
2. Garrard C, Young D. Suboptimal care of patients before admission to intensive care. BMJ 1998; 316: 1841-1842. (20 June.)

Medical emergency teams improve care

EDITORMcQuillan and colleagues suggest using medical emergency teams to help overcome deficiencies in acute care.¹ Our unit pioneered this concept and a medical emergency team system has operated since 1990.² The identification of patients early on in their physiological deterioration is intuitively sensible; the potential benefits are outlined by McQuillan and colleagues. Once such patients are identified, however, there must be provision to monitor them more closely and use treatments that cannot be safely provided on a normal ward. In short, medical emergency team systems must be coupled with adequate high dependency unit facilities.

Our hospital has a total of 532 beds, with eight ventilated intensive care beds and 12 high dependency beds. Of 493 responses by medical emergency teams in 1997 (only 10% for cardiac arrests), 92 (19%) resulted in patients being admitted to intensive care or high dependency units. Thus, the teams not only identify deteriorating patients but take intensive care expertise to the wards.

The parlous state of intensive care bed provision in the United Kingdom is well known.³ High dependency provision is at best patchy or, if available, caters solely for single specialties such as neurosurgery. If Britain is to address seriously the issues raised by McQuillan and colleagues, creation and expansion of high dependency facilities will be required.

Adverse effects of the medical emergency system include deskilling of ward medical staff. This can be ameliorated by having trainees rotate through intensive care. Deskilling of ward nursing staff does occur, and this risks an increase in the number of calls to medical emergency teams and greater need for high dependency unit facilities as staff become uncomfortable and unwilling to manage sick patients on the ward. Resistance from primary specialty consultants to the transfer of patients to high dependency units is also a concern that needs addressing.

The cost effectiveness of this approach is difficult to quantify. Savings may come from reduced admission to intensive care and length of stay. Irrespective of this, however, we believe that the system improves quality of care for our sickest patients. McQuillan and colleagues show that this is desperately needed.

Michael Mercer, Senior registrar. 
Stephen J Fletcher, Senior registrar. 
Gillian F Bishop, Staff specialist. 
Intensive Care Unit, Liverpool Hospital, PO Box 103, Liverpool, Sydney, NSW 2170, Australia

1. McQuillan P, Pilkington S, Allan A, Taylor B, Short A, Morgan G, et al. Confidential inquiry into quality of care before admission to intensive care. BMJ 1998; 316: 1853-1858. (20 June.)
2. Lee A, Bishop G, Hillman KM, Daffurn K. The medical emergency team. Anaesth Intens Care 1995; 23: 183-186[Medline].
3. Ryan DW. Providing intensive care. BMJ 1996; 312: 654[Free Full Text].

Authors' reply

EDITORGorard and Walshe criticise the method and analysis of data in our study. We based our methods closely on confidential inquiries such as those into perioperative and maternal mortality in which outcome is evident to assessors and definitions of quality of care are not predetermined. As medicolegal cases attest, disagreement between experts is common. More assessors and greater training may not improve interrater reliability.

Our study was conceived to develop a tool to assess quality of care before admission to intensive care. Pilot studies are rarely perfect first time. McGloin et al confirm our findings (blinding assessors to outcome, not allowing interobserver disagreement); 37% of their patients received suboptimal care with a significantly increased mortality.¹ Despite imperfections, these studies are compelling and concur with the experience of most British intensivists and other clinicians. As intensive care is required for about 1% of patients, about 0.5% of people admitted to hospital may receive suboptimal care.

Wood and Smith's findings confirm previous studies showing that 60-80% of patients who have cardiorespiratory arrests show premonitory signs. Amalgamation of data on 33 612 patients from three large UK databases² (Intensive Care National Audit and Research Centre, Critical Care Audit, personal communication) shows that cardiopulmonary resuscitation occurs within 24 hours of admission to intensive care in 3.5% of patients referred from theatre or recovery (mortality 49.5%), 14.3% of accident and emergency referrals (mortality 65.1%), and 15.9% of ward referrals (mortality 73.3%). Thus ward patients may be exposed to a high risk of avoidable cardiorespiratory arrest which carries a particularly grave prognosis.

Youngs and Gorard suggest that staffing shortages contribute to suboptimal ward care. However, greater ward presence requires more staff or reorganisation of work patterns. Consultant expansion has increased subspecialisation and diluted on call rotas. There is little evidence of increased consultant involvement in the care of acute patients or in teaching the necessary skills to trainees, despite rising numbers of emergency admissions and the effects of the Calman recommendations. Ringrose and Garrard found that few sick patients were reviewed daily by consultants. Contracting, competition, and waiting list initiatives have overemphasised elective work, leaving conflicting pressures between elective and emergency duties. Improvements in quality of acute care, an integral part of the government's white paper The New NHS, may not be possible without reducing elective workload.

The Royal College of Physicians³ seems to share McAllister and McGovern's concern that subspecialisation has sidelined acute general medicine. Mechanisms to ensure alerting of the intensive care team or an appropriate acute care physician are important developments but should not be an alternative to the responsible consultant being part of the receiving team. Too often the first consultant input occurs on the "post-take" round. In many hospitals consultants do the elective work and trainees deal with the emergency workload, often with little or no supervision.⁴

Our assessors identified that delays in admission to intensive care were caused by late referral and not bed availability. We believe that, even when all beds are occupied, the intensive care unit has a responsibility to ensure that other critically ill patients receive appropriate and timely care. Essential intensive care interventions can be initiated on the ward. Once stabilised, the patient may be transferred to another intensive care unit. This ensures that appropriateness of intensive care takes precedence over local bed availability.

It is time to challenge the traditional view of the intensive care unit as an isolated area of technological medicine and to develop the role of the intensive care team into a critical care service central to hospital acute medical care.⁵ Pritchard's call for improved training echoes our belief that training in critical care management should begin at undergraduate level and involve critical care doctors as teachers. This can be consolidated by postgraduate courses such as described by Anderson and Rowlands. Improved early intervention, using systems such as the medical emergency teams and "calling criteria" outlined by McAllister and McGovern, would then dovetail with high dependency and intensive care units to provide a seamless acute care service.

Peter McQuillan, Consultant in intensive care and anaesthesia. 
Sally Pilkington, Senior registrar in anaesthetics. 
Alison Allan, Registrar in anaesthetics. 
Bruce Taylor, Consultant in intensive care and anaesthetics. 
Gary Smith, Consultant in intensive care. 
Mick Nielson, Consultant in intensive care and anaesthetics. 
Intensive Care Unit, Southampton General Hospital, Southampton SO16 6YD

Alasdair Short, Consultant in intensive care and anaesthetics. 
Intensive Care Unit, Broomfield Hospital, Chelmsford CM1 7ET

Giles Morgan, Consultant in intensive care and anaesthetics. 
Intensive Care Unit, Royal Cornwall Hospital, Treliske, Truro TR1 3L

Charles Collins, Consultant in intensive care. 
Royal Devon and Exeter Hospital, Exeter EX2 5DW

1. McGloin H, Adam S, Singer M. The quality of pre-ICU care influences outcome of patients admitted from the ward. Clin Intens Care 1997; 8: 104.
2. Goldhill DR, Sumner A. Outcome of intensive care patients in a group of British intensive care units. Crit Care Med 1998; 26: 1337-1345[Medline].
3. Royal College of Physicians of London. Future patterns of care by general and specialist physicians. Meeting the need of adult patients in the UK. London: RCP , 1996.
4. Allan A, McQuillan PJ, Taylor BL, Nielson MS, Collins CH, Short ALK, et al. Who sees the critically ill patient before ICU admission? Clin Intens Care 1994; 5: 152.
5. McQuillan PJ. The central role of the critical care services in the structure and process of acute medicine. Intensive Care Society Newsletter Summer 1997.