Education Section

Epidemiology and prevention of unexpected in-hospital deaths

M. Buist S. Bernard J. Anderson 
Dandenong Hospital, PO Box 478, Dandenong, Victoria 3175, Australia

Correspondence to: M. Buist, Dandenong Hospital, PO Box 478, Dandenong, Victoria 3175, Australia

Introduction Incidence Risk factors

Prevention Conclusion References

Keywords: Epidemiology, prevention, in-hospital deaths 
Surg J R Coll Surg Edinb Irel., 1 October 2003, 265-268

Introduction: The medical literature has only documented the importance of adverse events in terms of patient outcomes and cost to the health service over the last decade. This review focuses on the incidence, risk factors and prevention on unexpected in-hospital deaths. Discussion: The incidence of unexpected in-hospital deaths in the two major retrospective studies was 0.5% and 0.8% of all hospital admissions. A prospective study of surgical patients documented an incidence of nearly 7.1%. The risk factors that have been associated with hospital death include: increasing age; non-elective operative procedure; negligence; human error and unstable bedside observations in the hours prior to death. Although a number of intreventions have been suggested to prevent unexpected in-hospital deaths, very few studies have been performed to determine effectiveness. In a single institution historical control study, the introduction of a Medical Emergency Team was associated with 50% reduction in the casemix adjusted incidence of in-hospital cardiac arrest calls. Conclusion: There are a number of studies to suggest that in-hospital deaths are both predictable and preventable. More work is required to determine effective strategies to manage this problem

INTRODUCTION
The medical literature has only documented the importance of adverse events in terms of patient outcomes and cost to health services over the last decade. This review focuses on the most serious adverse event that can occur to hospital in-patients, namely unexpected death. Many of the latter are unexpected and probably preventable through identification of risk factors.

INCIDENCE
The problem of defining ‘unexpected’ has hampered the study of in-hospital unexpected deaths. The best data comes from the Harvard Medical Practice Study (HMPS) and the Quality in Australian Health Care Study (QAHCS).^1-3 In both studies, trained nursing and specialist medical observers undertook large retrospective case note reviews of acute care hospital in-patient medical records to determine if adverse events had occurred and to document the associated outcomes. Although the methodologies in both studies were similar, they had slight differences in definition and identification of adverse events, which the authors of the QAHCS estimated could increase the incidence of adverse events by 20% to 25%.

In fact, there was a fourfold difference in the incidence of overall adverse events between the two studies; 3.7% for HMPS and 16.6% for QAHCS. However, the incidence of death as a result of the adverse event was broadly similar; 0.5% in HMPS, and 0.8% in the QAHCS. The incidence of unexpected in-hospital deaths could be less than these results, as they include deaths due to adverse events that occurred prior to and after the index admission. As documented in Table 1, these results are consistent with other smaller, retrospective studies.

The difficulty with all these studies is that they are retrospective and that the methodology used to determine incidence and preventability is subjective. Both the HMPS and the QAHCS included unexpected hospital deaths as one of the 18 adverse events they identified. Death, as a distinct outcome from the other 17 adverse events included in the medical record screen, would be relatively easy to identify. However, the medical reviewers used a subjective six-point scale to rank the degree of causation linking medical errors to adverse events that resulted in death.

In the HMPS, there was 89% agreement between the two trained medical specialist observers on the presence of an adverse event (kappa 0.61), and 93% agreement on causation due to negligence but with a lower significance (kappa 0.24). In the QAHCS the agreement between the two medical reviewers for the presence of an adverse event was 80% (kappa 0.55), and for event causation the agreement was 71% (kappa 0.42). Thus, from these two studies we can conclude that the rated incidence of death associated with an adverse event was accurate, but that the rated degree to which the death was unexpected, preventable or caused by the health system is more problematic. This is hardly surprising, as despite a death being unexpected, it may well be inevitable due to the patients, underlying co-morbidities.

The only prospective study to examine the incidence of adverse events and their outcomes was performed in a Melbourne teaching hospital, over a four-month period in 1999.7 This study, examined 1125 consecutive surgical admissions, both elective and emergency, that had a hospital stay of greater than 48 hours. The adverse events that were documented were: acute myocardial infarction; pulmonary embolism; acute pulmonary oedema; unscheduled tracheostomy; respiratory failure; cardiac arrest, cerebrovascular accident; severe sepsis, acute renal failure; unplanned intensive care unit (ICU) admission and death. There were 414 adverse events that affected 190 patients. Eighty patients (7.1%) died or 1 in 14 admissions. This study did not estimate the degree to which negligence, causation or preventability may have contributed to the adverse event. As such, the incidence of unexpected in-hospital deaths could not be determined.

RISK FACTORS
In the absence of studies that have explicitly examined unexpected inhospital deaths, we are left to extract information on risk factors from the large adverse event studies. In addition, there are several studies that have looked at high risk groups for in-hospital deaths and that have retrospectively examined the physical observations of those patients prior to the unexpected inhospital deaths.

Both the HMPS and the prospective study from Melbourne identified increased age as a risk factor for death from adverse events.^1,2,7 The HMPS documented that persons over 65 years of age had double the risk of an adverse event, than persons aged between 16 and 44 years of age. The prospective study from Melbourne documented a mortality rate of 20%, in patients aged more than 75 years who underwent with unscheduled surgery.

The HMPS estimated that 51.3% of the deaths from adverse events were caused by negligence. In a re-examination of the 2351 adverse events from the QAHCS, 81.8% of events were associated with human error.⁸ The human errors included failure in technical performance, failure to act on available information, failure to investigate or consult, failure to attend, and lack of care. Furthermore, the QAHCS found that ‘delay’ was associated with 20% of adverse events and that 86% to 90% of these events were assessed to be preventable.⁸

Several studies have adopted a different approach to the identification of risk factors for unexpected in-hospital deaths. Rather than looking at adverse events and factors associated with them they have looked at either the patients who had an unexpected in-hospital death or those who had a high risk event (in hospital cardiac arrest or unplanned ICU admission), and examined the observation charts prior to the index event. Another Australian study of 50,942 acute care admissions to three hospitals, performed over a six-month study period in 1996, documented the antecedents of 778 deaths.⁹ Of these, only 66 were classified as unexpected in that they did not have a ‘do not resuscitate’ (DNR) order and nor were their deaths preceded by a cardiac arrest or ICU admission. In the eight hours prior to the deaths of these patients 50% had documented severe abnormalities in the observation charts or concerns noted in the nursing or medical record. Furthermore, 33% of these patients had abnormal observations, or concerns noted up to 48 hours prior to their death. The most common abnormal observations were hypotension (systolic blood pressure < 90mmHg) and tachypnoea (respiratory rate > 36 per minute).

Several studies have examined cardiac arrest calls or unplanned ICU admissions from within hospitals on the assumption that these events were ‘unexpected’ in that there were no DNR orders in place or that whatever process was going on could be reversed with intensive care interventions. Despite the fact that survival to hospital discharge from in-hospital cardiac arrests is only in the range of 15 to 30%, there is good evidence that the majority of these arrests are not unexpected.^10-17 In common with the findings from the Australian three hospital study, retrospective analysis of simple bedside observations prior to inpatient cardiac arrest call or referral for ICU admission have demonstrated prolonged periods of documented clinical instability in a significant number of patients.9 A retrospective case note review at the Jackson Memorial Medical Centre, Florida, over a four-month period in 1987, documented 64 consecutive inhospital cardiac arrests in the general ward areas.¹⁸ Of these, 54 (84%), had documented observations of clinical deterioration or new complaints within eight hours of the arrest. In a similar study performed at the Cook County Hospital, Illinois, 150 cardiac arrests were observed in the medical wards over a 20-month period from 1990 to 1991.19 In 99 (64%) of these cases, a nurse or physician documented deterioration in the patient’s condition within six hours of the cardiac arrest. The hospital mortality rate of the 150 cardiac arrests was 91%. More recently, in a 28 week period at the Manchester Royal Infirmary, reported in 1999, 47 cardiac arrest calls in the general ward areas were analysed.²⁰ Twenty-four (51%) had premonitory signs prior to the cardiac arrest call. Similarly, in our study in a tertiary care hospital in metropolitan Melbourne over the calendar year of 1997, there was a median period of documented clinical instability of 6.5 hours (range 0-432 hours) prior to either a cardiac arrest call or ICU referral amongst 122 in-hospital patients.²¹

Finally, at least in Australia, several studies indicate that the medical undergraduate syllabus does not provide graduates with the basic skills to manage acute life threatening emergencies.^22-24 These studies have identified deficiencies not only in intellectual and procedural abilities, but also in communication. This is a worrying state, as in the first instance it often falls to the most junior members of the medical team to assess and manage these acute clinical situations.

In summary, increasing age, unplanned surgery, negligence, human error and treatment delays have all been associated with unexpected in-hospital death.

PREVENTION
Despite the general lack of data concerning the incidence and risk factors for unexpected in-hospital deaths, there are a wide range of interventions proposed to prevent them. These include medical emergency teams (MET’s), outreach teams, ‘the ICU without walls’, the ‘hospitalist’ and various education systems for both under- and post-graduate medical trainees. However, apart from two non-blinded, non-randomised studies that have examined the efficacy of METs, there are no studies that examine the usefulness of any of these interventions.^25,26

The largest study was performed in three New South Wales hospitals in Australia in 1996. One hospital had a MET in operation since 1991 and the other two hospitals served as controls.²⁵ Over a 6 month period the study examined 50,942 consecutive adult acute (non-psychiatric) in-patient hospital admissions. The outcomes of the study were in-hospital cardiac arrest incidence, all hospital deaths, hospital deaths without DNR orders, total ICU and high dependency unit (HDU) admissions, and unanticipated ICU/HDU admissions. There was no difference in the incidence of in-hospital cardiac arrests and all hospital deaths, between the hospital with the MET and the two control hospitals. However, both of the control hospitals had a significantly greater incidence of unanticipated ICU/HDU admissions than the hospital with the MET (odds ratio of 1.48 and 1.58 versus 1.0). Also, one of the control hospitals had a greater incidence of death with no DNR orders than the MET hospital (odds ratio of 1.77 versus 1.0).

Our group performed the other study that examined the efficacy of a MET system in a single hospital using a historical control over the two previous years.²⁶ We documented a reduction in the incidence of in-hospital cardiac arrests from 3.8 per 1000 hospital admissions in 1996, to 2.0 per 1000 hospital admissions in 1999 after the MET system had been operational for three years. After adjustment for the hospital case-mix over the two time periods, the MET intervention was associated with a 50% reduction in the incidence of in-hospital cardiac arrests (odds ratio 0.50, 95% confidence interval 0.35 to 0.73). The setting up of a MET was associated with a reduction in the overall hospital mortality rate of 2.0 per 1000 hospital admissions over the study time period. The mortality rate from those patients who did have a cardiac arrest was also reduced from 56/73 (77%) to 26/47 (55%) after the MET intervention (p< 0.001). The conclusions that can be drawn from this study are limited by the use of a historical control in a single institution and the fact that we explicitly used performing the research to drive the outcomes of management. This included the progressive introduction of education, audit, and staff debriefing sessions to facilitate the implementation of the MET system throughout the hospital. This study did not examine the differential effects of the MET intervention, or the associated support activities that may have also influenced the results. As such, the applicability of these results to other institutions is questionable. Despite this, similar results have been replicated, in terms of in-hospital cardiac arrest incidence and overall hospital mortality reduction, at another Melbourne teaching hospital using a similar study design.²⁷

From the limited results available we can conclude that a system like the MET can decrease the incidence of unexpected cardiac arrest calls, which in itself is a desirable outcome, but whether MET can influence unexpected hospital mortality is less clear. This is primarily because a MET intervention may alter perception of the likelihood, or ‘unexpectedness’, of death. In our study, in 13 instances the MET intervention resulted in the implementation of DNR orders, and 11 of these subsequently died.²⁶ This suggests that at least in some instances the DNR status of the patient is not clear, or is inappropriate. In these circumstances, without a MET system, when patients have a cardiac arrest the attending medical team would assess the event as unexpected. The MET system allows for the appropriate classification of these patients either into for resuscitation or DNR, before the mortality status at discharge is determined. Thus, the MET may just simply transfer hospital mortality from unexpected, to expected, by appropriate allocation of DNR status.

CONCLUSION
Despite the limitations of the data, there is a body of evidence that a significant number of in-hospital deaths are unexpected. There is also some data to suggest that these deaths may be preventable by the detection of certain risk factors prior to the index event. However, further intervention studies are required to truly determine the preventability of these deaths.

REFERENCES
1. Brennan TA, Leape LL, Laird N, et al. Incidence of adverse events and negligence in hospitalised patients: results of the Harvard Medical Practice Study I. N Engl J Med 1991; 324: 370-376.
2. Leape LL, Brennan TA, Laird N, et al. The nature of adverse events in hospitalised patients: results of the Harvard Medical Practice Study II. N Engl J Med. 1991; 324: 377-384.
3. Wilson RM, Runciman WB, Gibberd RW, et al. The Quality in Australian Health Care Study. Med J Aust 1995; 163: 458-471.
4. California Medical Association. Report of the Medical Insurance Feasibility study. San Francisco: California Medical Association, 1977.
5. Thomas EJ, Studdert DM, Burstin HR, Orav EJ, Zeena T, Williams EJ et al. Incidence and types of adverse events and negligence care in Utah and Colorado in 1992. Med Care (in press).
6. Vincent C, Neale G, Woloshynowych M. Adverse events in British hospitals: preliminary record review. BMJ 2001; 322 (7285): 517-9.
7. Bellomo R, Goldsmith D, Russell R, Uchino S. Postoperative serious adverse events in a teaching hospital: a prospective study.
8. Wilson RM, Harrison BT, Gibberd RW, Hamilton JD. An analysis of the causes of adverse events from the Quality in Australian Health Care Study. Med J Aust 1999; 170: 411-415.
9. Hillman KM, Bristow PJ, Chey T, et al. Antecedents to hospital deaths. Intern Med J 2001; 31 (6): 343-8.
10. Berger R, Kelley M. Survival after in-hospital cardiopulmonary arrest of noncritically Ill patients: A prospective study. Chest 1994; 106: 872-9.
11. McGrath RB. In-house cardiopulmonary resuscitation- after a quarter of a century. Ann Emerg Med 1987; 16: 1365-8.
12. Ebell MH, Becker LA, Barry HC, Hagen M. Survival after in-hospital cardiopulmonary resuscitation. A meta-analysis. J Gen Intern Med 1998; 13 (12): 805-16.
13. Bedell SE, Delbanco TL, Cook EF, Epstein FH. Survival after cardiopulmonary resuscitation in the hospital. NEJM 1983; 309: 569-76.
14. Debard ML. Cardiopulmonary resuscitation: analysis of six years’ experience and review of the literature. Ann Emerg Med 1981; 10 (8): 408-15.
15. George AL, Folk BP, Crecelius Pl, Campbell WB. Pre-arrest morbidity and other correlates of survival after in-hospital cardiopulmonary arrest. The American Journal of Medicine 1989; 87: 28-34.
16. Bedell SE, Deitz DC, Leeman D, Delbanco TL. Incidence and characteristics of preventable iatrogenic cardiac arrests. JAMA 1991; 265: 2815-2820.
17. Tortolani AJ, Risucci DA, Rosati RJ, Dixon R. In-hospital cardiopulmonary resuscitation: patient and resuscitation factors associated with survival. Resuscitation 1990; 20: 115-128.
18. Schein RM, Hazdat N, Pena M, et al. Clinical antecedents to in-hospital-cardiopulmonary arrest. Chest 1990; 00 1388-1392.
19. Franklin C, Mathew J. Developing strategies to prevent in-hospital cardiac arrest: Analyzing responses of physicians and nurses in the hours before the event. Crit Care Med 1994; 22:244-247.
20. Smith AF, Wood J. Can some in-hospital cardio-respiratory arrests be prevented? A prospective survey. Resuscitation 1998; 37 (3): 133-7.
21. Buist MD, Jarmolowski E, Burton PR, Bernard SA, Waxman BP, Anderson J. Recognising clinical instability in hospital patients before cardiac arrest or unplanned admission to Intensive Care. A pilot study in a tertiary-care hospital. Med J Aust 1999; 171: 22-25.
22. Harrison GA, Hillman KM, Fulde GWO, Jacques TC. The need for undergraduate education in critical care. Anaesth Int Care 1999; 27 :53-8.
23. Taylor D. Undergraduate procedural skills training in Victoria: Is it adequate? Med J Aust 1997; 171 : 22-25.
24. Buist M, Jarmolovski E, Burton P, McGrath B, Waxman B, Meek R. Can Interns manage clinical instability in hospital patients? A survey of recent graduates. Focus on health professional education 2001; 3 (3):20-8.
25. Bristow PJ, Hillman KM, Chey T, et al. Rates of in-hospital arrests, deaths and intensive care admissions: the effect of a medical emergency team. Med J Aust 2000; 173: 236-240.
26. Buist MD, Moore GE, Bernard SA, Waxman BP, Anderson JN, Nguyen TV. Effects of a medical emergency team on reduction of incidence of and mortality from unexpected cardiac arrests in hospital: preliminary study. BMJ 2002; 324: 1-6.
27. Bellomo R, Goldsmith D, Uchino S et al. The effect of responding to signs of physiological instability on in-hospital cardiac arrests and mortality. Personal communication.

Copyright: 9 August 2003