Search tips
Search criteria 


Logo of bmjThe BMJ
BMJ. 1998 June 20; 316(7148): 1853–1858.

Confidential inquiry into quality of care before admission to intensive care

Peter McQuillan, consultant in intensive care and anaesthesia,a Sally Pilkington, senior registrar in anaesthesia,a Alison Allan, registrar in anaesthesia,a Bruce Taylor, consultant in intensive care and anaesthesia,a Alasdair Short, consultant in intensive care,c Giles Morgan, consultant in anaesthesia and intensive care,d Mick Nielsen, consultant in intensive care and anaesthetics,b David Barrett, senior lecturer,e and Gary Smith, director of intensive carea


Objective: To examine the prevalence, nature, causes, and consequences of suboptimal care before admission to intensive care units, and to suggest possible solutions.

Design: Prospective confidential inquiry on the basis of structured interviews and questionnaires.

Setting: A large district general hospital and a teaching hospital.

Subjects: A cohort of 100 consecutive adult emergency admissions, 50 in each centre.

Main outcome measures: Opinions of two external assessors on quality of care especially recognition, investigation, monitoring, and management of abnormalities of airway, breathing, and circulation, and oxygen therapy and monitoring.

Results: Assessors agreed that 20 patients were well managed (group 1) and 54 patients received suboptimal care (group 2). Assessors disagreed on quality of management of 26 patients (group 3). The casemix and severity of illness, defined by the acute physiology and chronic health evaluation (APACHE II) score, were similar between centres and the three groups. In groups 1, 2, and 3 intensive care mortalities were 5 (25%), 26 (48%), and 6 (23%) respectively (P=0.04) (group 1 versus group 2, P=0.07). Hospital mortalities were 7 (35%), 30 (56%), and 8 (31%) (P=0.07) and standardised hospital mortality ratios (95% confidence intervals) were 1.23 (0.49 to 2.54), 1.4 (0.94 to 2.0), and 1.26 (0.54 to 2.48) respectively. Admission to intensive care was considered late in 37 (69%) patients in group 2. Overall, a minimum of 4.5% and a maximum of 41% of admissions were considered potentially avoidable. Suboptimal care contributed to morbidity or mortality in most instances. The main causes of suboptimal care were failure of organisation, lack of knowledge, failure to appreciate clinical urgency, lack of supervision, and failure to seek advice.

Conclusions: The management of airway, breathing, and circulation, and oxygen therapy and monitoring in severely ill patients before admission to intensive care units may frequently be suboptimal. Major consequences may include increased morbidity and mortality and requirement for intensive care. Possible solutions include improved teaching, establishment of medical emergency teams, and widespread debate on the structure and process of acute care.

Key messages

  • Suboptimal management of oxygen therapy, airway, breathing, circulation, and monitoring before admission to intensive care occurred in over half of a consecutive cohort of acute adult emergency patients. This may be associated with increased morbidity, mortality, and avoidable admissions to intensive care
  • At least 39% of acute adult emergency patients were admitted to intensive care late in the clinical course of the illness
  • Major causes of suboptimal care included failure of organisation, lack of knowledge, failure to appreciate clinical urgency, lack of supervision, and failure to seek advice
  • A medical emergency team may be useful in responding pre-emptively to the clinical signs of life threatening dysfunction of airway, breathing, and circulation, rather than relying on a cardiac arrest team
  • The structure and process of acute care and their importance require major re-evaluation and debate


Seriously ill patients may be identified by the clinical signs of life threatening dysfunction of the airway, breathing, or circulation, but these may be missed, misinterpreted, or mismanaged by clinicians of all grades. Avoidable components therefore contribute to physiological deterioration, with major consequences on morbidity, mortality, requirement for intensive care, and cost. Such deficiencies may be described as suboptimal care.16 We aimed to investigate the prevalence of suboptimal care before admission to intensive care, to examine its nature, causes, and consequences, and to suggest possible solutions.

Subjects and methods

We prospectively studied the quality of care received by 50 consecutive, adult emergency patients before their admission to intensive care units in each of two centres (Portsmouth and Southampton). The study was conducted in the winter of 1992-3 after approval by local ethics committees and all acute unit consultants.

On the basis of methodology used for confidential inquiries1,2,6 detailed questionnaires were completed by us during structured interviews with (a) the admitting clinical team and (b) the intensive care team. The interviews concentrated on events between hospital admission and admission to intensive care. The questionnaires comprised tick and data entry boxes and a page for summarising history, clinical findings, assessment, thought processes, resuscitation, treatment, and response to treatment. Emphasis was placed on the recognition, investigation, monitoring, and management of abnormalities of airway, breathing, circulation, and oxygen therapy and monitoring. Interviews took place as soon as possible after a patient’s admission to intensive care, which ranged from minutes to days. Severity of illness was recorded using the acute physiology and chronic health evaluation (APACHE II7) scoring system, using most extreme values in the first 24 hours in intensive care. Data on duration of stay in intensive care and intensive care and hospital outcomes were also collected. Casemix adjusted expected mortality was calculated from the APACHE data7 and compared with actual mortality to produce a standardised mortality ratio.

Data from the questionnaires were made anonymous and sent to two extraregional intensivist assessors (GM, primary specialty anaesthesia and AS, primary specialty nephrology). Clinical notes were not included.

The assessors specifically considered quality of medical care and appropriateness and timeliness of admission to intensive care. Care considered suboptimal was defined and the causes outlined. A 10 cm linear analogue scale was used to score the adequacy of management of (a) oxygen therapy, (b) airway, (c) breathing, (d) circulation, and (e) monitoring.

The database was analysed with Microsoft Excel (Microsoft, Seattle, WA) and Minitab (Minitab, PA). Non-parametric data were compared using χ2 and Kruskal-Wallis tests. A sample size of 100 was arbitrarily chosen and no estimate of clinical effect or power analysis was undertaken.


Of the 100 patients admitted to intensive care, 51 were general medical, 28 general surgical, eight orthopaedic, three obstetrics and gynaecology, three urology, two neurosurgery, two ophthalmology, one ENT, one haematology, and one thoracic surgery. No significant differences were found between the two centres in age, sex distribution, incidence of inappropriate admission, late admission, or suboptimal care, and the casemix was broadly similar. Severity of illness was greater in Portsmouth than in Southampton (median APACHE scores 21.6 and 16 respectively) (P=0.03).

The assessors agreed that 20 patients (group 1) were well managed and that 54 patients (group 2) received suboptimal care. They disagreed on the quality of care before admission to intensive care in 26 patients (group 3) (table (table1).1). For internal validation, assessors were separately asked to classify patients according to quality of care: 10.5% received excellent care (AS 4%, GM 17%), 21.5% received good quality care (AS 20%, GM 23%), 17.5% received adequate care (AS 25%, GM 10%), and 50.5% received inadequate care (AS 51%, GM 50%). In each quality of care group the casemix was broadly similar, with no significant differences in APACHE II scores between groups (table (table1).1).

Table 1
Patient demographic data, assessor opinions on appropriateness of admission and linear analogue scores, and mortality

Agreement between assessors was moderate8: κ values for questions on late admission, appropriateness, and suboptimal care were 0.50, 0.50, and 0.42 respectively. The weighted κ value for categorisation into excellent, good, adequate, or inadequate care classes was 0.53. In groups 1, 2, and 3 intensive care mortalities were 5 (25%), 26 (48%), and 6 (23%) respectively (P=0.04). By partitioning the 3 × 2 table (mortality in table table1)1) into two separate 2 × 2 tables,9 a comparison of groups 1 and 2 gave a P value of 0.07. Hospital mortalities were 7 (35%), 30 (56%), and 8 (31%) (P=0.07) (table (table1),1), and standardised mortality ratios were 1.23 (95% confidence interval 0.49 to 2.53), 1.4 (0.94 to 2.00), and 1.26 (0.54 to 2.48) respectively. More patients received suboptimal care (agreed by both assessors) before admission to intensive care in the intensive care non-survivors group (26/37, 70%) than in the survivors group (28/63, 44%) (P=0.04).

Admission to intensive care was considered avoidable definitely in 4.5% of patients, probably in 4%, and possibly in 32.5%, and in 7.5%, 7.5%, and 41.5% respectively of group 2 (table (table2).2). A minimum of 4.5% (definitely avoidable) and a maximum of 41% of adult emergency patients (sum of definitely, probably, and possibly avoidable) might have avoided admission to intensive care with better management of airway, breathing, and circulation.

Table 2
Opinions of two independent extraregional assessors (AS and GM) on avoidability of admission to intensive care and how far suboptimal care contributed to morbidity or mortality

The assessors agreed that 39% of the patients were admitted late in the clinical course, and disagreed on a further 25% of the patients (table (table1).1). All patients in group 1 were admitted to intensive care at a time considered appropriate by the assessors. Of group 2 patients, 37 (69%) were admitted late (P<0.0001), the source time (period between hospital admission and admission to intensive care) being longest in this group (table (table1).1). Length of stay in intensive care was not significantly different between groups whether non-survivors were included (P=0.38) or excluded (P=0.55) (table (table1).1).

Of the patients in group 2, the assessors concluded that suboptimal care contributed to morbidity or mortality definitely in 32.5%, probably in 21%, and possibly in 32.5% (table (table2).2). Scores for management of oxygen therapy, airway, breathing, and circulation, and monitoring (table (table1)1) were all lowest in group 2 patients (P<0.0001).

Suboptimal care in group 2 patients occurred in similar proportions in the general medical (24/51, 47%) and general surgical (18/28, 64%) subpopulations (data not shown). Using APACHE II point deciles to divide the 100 patients into groups on the basis of severity of illness, the assessors agreed that suboptimal care occurred in 41-64% of patients in each decile (table (table3).3).

Table 3
Stratification by APACHE II* deciles

The principle causes of suboptimal care were failure of organisation, lack of knowledge, failure to appreciate clinical urgency, lack of experience, lack of supervision, and failure to seek advice (table (table4).4).

Table 4
Causes of suboptimal care, and agreement between assessors


Hospital mortality for critically ill patients is a complex function of age, surgical or medical status, elective or emergency status, comorbidities, physiological reserve, nature and severity of acute illness, and quality of care.7 Apart from long term sociopolitical strategies to improve the health of the nation,10 only quality of care and perhaps severity of illness (by earlier recognition of critical illness, intervention, and referral to intensive care) may potentially be influenced by clinicians.

The assessment of quality (and the quality of measuring it) is difficult and examination of the process may be more sensitive than measures of outcome.11,12 All have limitations. This study, modelled on an earlier inquiry,1 is subject to similar limitations of power and outcome bias. Like other studies1,2 we relied on the opinions of assessors on what constituted suboptimal care and its causes (table (table4),4), as objective definitions for all conceivable scenarios were impractical. Other limitations included assessor agreement (the existence of group 3 reflecting reality, where even experts frequently disagree), small patient numbers, and wide confidence intervals. These factors may explain why the mortality of group 3 was similar to group 1 rather than occurring between groups 1 and 2.

The statistically insignificant effect (at the 5% level) of gross variations in quality of care before admission to intensive care on mortality and standardised mortality ratios generated from the APACHE scoring system, may have several possible explanations: inadequate sample size (including the existence of group 3), APACHE is insufficiently sensitive for detecting the effect, quality of care before admission to intensive care has no impact on mortality, or greater mortality prevented by intensive care and post-intensive care factors.

As an index of quality of care in intensive care the reliability of APACHE and standardised mortality ratios have been criticised.13 APACHE purports to predict hospital outcome for populations of patients, therefore reflecting the totality of quality of hospital care. This study suggests that the quality of care before admission to intensive care may influence outcome. Casemix adjustment for the adult emergency patient at high risk may prove inadequate using APACHE as elective and paediatric patients have low mortalities. These factors may explain the apparently high standardised mortality ratio of 1.23 in group 1, a figure identical to the standardised mortality ratio in the southwest Thames audit group (17 intensive care units, including Portsmouth14). The validity of standardised mortality ratios may also be compromised by lead time bias,15 where resuscitation instigated in intensive care may ameliorate physiology (and APACHE score) before transfer to intensive care,16,17 and the limited applicability of data, casemix, and practice from the United States to Britain.

The relevance of the suboptimal care received by 54% of this cohort before admission to intensive care is not negated by the lack of statistical significance for effect on outcome. The assessors believe suboptimal care had a substantial impact on individual morbidity, mortality, and requirement for intensive care resources (avoidable admissions). Furthermore, in a more recent study in which 32 of 87 (37%) patients admitted to intensive care from the ward suffered suboptimal care, McGloin et al blinded their assessors to outcome, had a no disagree group, and found a highly significant increase in mortality in the group receiving suboptimal care.18 Although we believe our local situation has improved, McGloin et al’s study suggests this may not be so universally. Clinically significant effects also occur if appropriate referrals to intensive care are delayed, refused, or transferred elsewhere, elective surgery is disrupted, and if direct or medicolegal costs are raised. Ethical dilemmas arise as to what, how, and by whom information on deficiencies of care should be imparted to patients and their families.

Suboptimal care occurred in 41-64% of patients in each APACHE decile—that is, at all levels of severity of illness. It is therefore probable that a similar pattern occurs in patients not referred to intensive care, a contention supported by others although not specifically addressed by this study.1,2,6

Although length of stay in intensive care was not statistically different, the assessors believed between 4.5% and 41% of admissions were potentially avoidable. Thus better care before admission to intensive care may reduce intensive care bed days.

Failings of clinicians of all grades over a wide range of tasks and knowledge have been shown.1924 This study suggests a fundamental problem of failure to appreciate that airway, breathing, and circulation are the prerequisites of life and that their dysfunction are the common denominators of death. The assessors’ conclusions on the causes of suboptimal care (table (table4)4) suggest multifactorial organisational (structure) and clinical (process) problems. The box shows possible solutions.

Suggestions to improve quality of care before admission to intensive care


Recognition by referring teams that a problem of quality of care exists

Increased emphasis on care of critically ill patient by royal colleges in teaching and exams

Organisation and structure

Trainee intensive care posts open to all specialties in larger centres

Recognition of intensive care as a specialty

House officer posts in intensive care or anaesthesia for 3 months

Appointment of acute care (general) physicians to deal with acute medicine

Emphasise that prime roles of consultants are patient care (supervision of juniors) and teaching

Expansion of continuing medical education to a maintenance of standards programme allowing consultants to spend time accompanying and observing other clinicians’ practice, including other specialties and subspecialties

Alter consultant sessions to recognise need to be involved in acute care and supervison and teaching of trainees

Multidisciplinary rotations of senior house officers: intensive care, accident and emergency, and major specialties

Expansion of accident and emergency departments to include acute admissions/high dependency unit facilities

Expansion of intensive care and high dependency unit facilities (close to intensive care). Rotate high dependency unit nurses into intensive care periodically

Clinical process

Improved recognition of serious illness, physiological derangements of airway, breathing, and circulation—make clinicians extra vigilant to physiological abnormalities (“physiology police”)

Change the acute care ethos: when patients or volume of work are difficult call in a senior staff member

Increased hands on involvement of consultant in acute care

Acute care trainees require accreditation in appropriate advanced life support course (or should be sent on such course)

Extensive initial preparation programme for new doctors. Junior doctors should not be given responsibility unprepared and unsupported

Improved teaching of all grades of staff including medical students. House officers and senior house officers to spend a week in intensive care unit during medical and surgical posts

Replace cardiac arrest teams with medical emergency teams with defined calling criteria (see table 5)

Guidelines and audit

Cross specialty audit and morbidity, mortality, and critical incident meetings

Greater development of guidelines and best practice for patient management

Audit adherence to standards and guidelines

Consider peer review sessions to examine delivery of care

Recognise everyone makes mistakes and it is usually more educational to examine our errors than our successes. The emphasis should be on education and not vilification

The national confidential inquiry into perioperative deaths recommends that surgical trainees should be actively trained, should readily seek senior advice, and should not operate unsupervised at night.1 Trauma reports recommend senior experienced input.25 Few equivalent studies or recommendations to set standards for medical patients exist. In our cohort the majority of patients were treated predominantly by trainees, often unsupervised.26 Some consultants acknowledge that their skills in acute general medicine could be improved.27

With the exception of infarct related ventricular fibrillation, the outcome after cardiorespiratory arrest is fairly poor and most of these patients (60-80%) show premonitory signs.2833 This supports changing emphasis from the traditional cardiac arrest team to a medical emergency team,3436 aiming at early recognition of sick patients and prevention of cardiorespiratory arrest. Such a team, including an intensive care specialist registrar or consultant and medical specialist registrar or acute care physician, would attend all potentially life threatening abnormalities of airway, breathing, or circulation. Table Table55 shows possible referral criteria. Pre-emptive action by the medical emergency team—that is, early recognition, referral, and decision making for sick patients, should enhance acute intensive medical care that is proactive. Adaptation to the rise in emergency work37 has been advocated,3840 and the recent Royal College of Physicians report is timely and welcome.41

Table 5
Suggested minimum calling criteria for medical emergency team (modified from Lee et al34)

All human practice (including that of doctors before, during, and after intensive care) has elements of excellence, adequacy, and deficiency, and the public deserves to know that quality of care is regularly examined. The study of error is not to apportion blame but to ask why and institute appropriate changes in organisation and clinical care.


In this study suboptimal care of severely ill patients before admission to intensive care was common and influenced morbidity, mortality, and requirement for intensive care. Remedial measures may substantially reduce emergency admissions to intensive care, and mortality. Although Osler noted many years ago that: “Patients do not die of their disease, they die of the physiologic abnormalities of their disease,”42 the concept of doctors as “physiology police” may have been lost. Training should emphasise that airway, breathing, and circulation are the prerequisites of life and their dysfunction are the common denominators of death. The greatest impact on the outcome from intensive care units may arise from improvements in input to intensive care particularly in the quality of acute care.


We thank Carol Orchard, Alan Grimes, Dr Sue Hill, and Dr John Lunn for their help in the design and completion of the study, and the clinicians in Portsmouth and Southampton for allowing the study of acute care.


Editorial by Garrard and Young


Funding: This study was supported by the audit departments of Portsmouth Hospitals NHS Trust and Southampton University Hospitals NHS Trust. We thank Lilly Industries for financial support in preparing and printing the questionnaires.

Conflict of interest: None.


1. National confidential enquiry into perioperative deaths 1991/2. London: National Confidential Enquiry into Perioperative Deaths; 1993.
2. Report on confidential enquiries into maternal deaths in the United Kingdom 1988-90. London: HMSO; 1994.
3. Dillner L. Babies’ deaths linked to suboptimal care. BMJ. 1995;310:757. [PubMed]
4. Sharples PM, Storey A, Aynsley-Green A, Eyre JA. Avoidable factors contributing to the death of children with head injuries. BMJ. 1990;300:87–91. [PMC free article] [PubMed]
5. Jeffery RV, Jones JJ. Avoidable factors contributing to the death of head injury patients in general hospitals in the Mersey region. Lancet. 1981;ii:459. [PubMed]
6. Payne JN, Milner PC, Saul C, Bowns IR, Hannay DR, Ramsay LE. Local confidential enquiry into avoidable factors in deaths from stroke and hypertensive disease. BMJ. 1993;307:1027–1030. [PMC free article] [PubMed]
7. Knaus WA, Draper EA, Wagner DP, Zimmerman JS. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13:818–829. [PubMed]
8. Altman DG. Practical statistics for medical research. London: Chapman and Hall; 1991.
9. Siegal S, Castellan NJ. Statistics for the behavioural sciences. McGraw-Hill: New York; 1988.
10. Department of Health, United Kingdom. The health of the nation. HMSO: London; 1992.
11. Kassirer JP. The quality of care and the quality of measuring it. N Engl J Med. 1993;29:1263–1264. [PubMed]
12. Davies HTO, Crombie IK. Assessing the quality of care. BMJ. 1995;311:766. [PMC free article] [PubMed]
13. Boyd O, Grounds RM. Physiological scoring systems and audit. Lancet. 1995;341:1573–1574. [PubMed]
14. Pappachan VJ, Millar BW, Bennett ED, Smith GB. Outcome comparisons after case mix adjustment using APACHE III (A3) system in 17 UK ICUs. Clin Intensive Care. 1997;8:97.
15. Tunnell RD, Smith GB. The effect of lead-time bias on the severity of illness scoring and mortality prediction in critically ill patients. Clin Intensive Care. 1996;7:55.
16. Goldhill DR, Withington PS. Mortality prediction by APACHE II. Anaesthesia. 1996;51:719–723. [PubMed]
17. Palazzo M, Soni N, Hinds C. Physiological scoring systems and audit. Lancet. 1993;342:307. [PubMed]
18. McGloin H, Adam S, Singer M. The quality of pre-ICU care influences outcome of patients admitted from the ward. Clin Intensive Care. 1997;8:104.
19. Thwaites BC, Shankar S, Niblett D, Saunders J. Can consultants resuscitate? J Roy Coll Phys. 1992;26:265–267. [PubMed]
20. Tham KY, Evans RJ, Rubython EJ, Kinnaird TD. Management of ventricular fibrillation by doctors in cardiac arrest teams. BMJ. 1994;309:1408–1409. [PMC free article] [PubMed]
21. Rolfe S, Harper NJN. Ability of hospital doctors to calculate drug doses. BMJ. 1995;310:1173–1174. [PMC free article] [PubMed]
22. Stoneham M, Saville GM, Wilson IH. Knowledge about pulse oximetry among medical and nursing staff. Lancet. 1994;344:1339–1342. [PubMed]
23. Montgomery H, Hunter S, Morris S, Naunton-Morgan R, Marshall RM. Interpretation of electocardiograms by doctors. BMJ. 1994;309:1551–1552. [PMC free article] [PubMed]
24. Gould TH, Upton PM, Collins P. A survey of the intended management of acute postoperative pain by newly qualified doctors in the South West region of England. Anaesthesia. 1994;49:807–810. [PubMed]
25. Report of the working party on the management of patients with major injuries. Royal College of Surgeons of England, 1985.
26. Allan A, McQuillan PJ, Taylor BL, Nielsen MS, Collins CH, Short AIK, et al. Who sees the critically ill patient before ICU admission? Clin Intensive Care. 1994;5:152.
27. Mather HM, Elkeles RS. Attitudes of consultant physicians to the Calman proposals: a questionnaire survey. BMJ. 1995;311:1060–1062. [PMC free article] [PubMed]
28. Sax FL, Charlson ME. Medical patients at high risk for catastrophic deterioration. Crit Care Med. 1987;15:510–515. [PubMed]
29. Franklin C, Mathew J. Developing strategies to prevent inhospital cardiac arrest: analyzing responses of physicians and nurses in the hours before the event. Crit Care Med. 1994;22:244–247. [PubMed]
30. Schein RMH, Hazday N, Pena M, Ruben BH, Sprung CL. Clinical antecedents to in-hospital cardiopulmonary arrest. Chest. 1990;98:1388–1392. [PubMed]
31. Bedell SE. Incidence and characteristics of preventable iatrogenic cardiac arrest. JAMA. 1991;265:2815–2820. [PubMed]
32. Dubois RW. Preventable deaths; who, how often and why? Ann Int Med. 1988;109:582–589. [PubMed]
33. George Jr AL. Prearrest morbidity and other correlates of survival after in hospital cardiac arrest. Am J Med. 1989;87:28–34. [PubMed]
34. Lee A, Bishop G, Hillman KM, Daffurn K. The medical emergency team. Anaesth Intensive Care. 1995;23:183–186. [PubMed]
35. Daffurn K, Lee A, Hillman KM, Bishop GF, Bauman A. Do nurses know when to summon emergency assistance? Intensive Crit Care Nurs. 1994;10:115–120. [PubMed]
36. Hournihan F, Bishop G, Hillman KM, Daffurn K, Lee A. The medical emergency team: a new strategy to identify and intervene in high-risk surgical patients. Clin Intensive Care. 1995;6:269–272.
37. Capewell S. The continuing rise in emergency admissions. BMJ. 1996;312:991–992. [PMC free article] [PubMed]
38. Moss F, McNicol M. Alternative models of organisation are needed. BMJ. 1995;310:925–928. [PMC free article] [PubMed]
39. Smith J. Consultants of the future. BMJ. 1995;310:953–954. [PMC free article] [PubMed]
40. Delamothe A. Wanted: guidelines that doctors will follow. BMJ. 1993;307:218. [PMC free article] [PubMed]
41. Royal College of Physicians of London. Future patterns of care by general and specialist physicians. London: RCP, Sept; 1996.
42. Pope A. The penguin dictionary of quotations. In: Cohen JM and Cohen MJ, eds. Hertfordshire: Omega, 1960.

Articles from The BMJ are provided here courtesy of BMJ Publishing Group