Duration of life-threatening antecedents prior to intensive care admission

Intensive Care Med (2002) 28:1629 1634 DOI 10.1007/s00134-002-1496-y ORIGINAL Ken M. Hillman Peter J. Bristow Tien Chey Kathy Daffurn Theresa Jacques Sandra L. Norman Gillian F. Bishop Grant Simmons Duration of life-threatening antecedents prior to intensive care admission Received: 19 September 2000 Accepted: 6 August 2002 Published online: 11 September 2002 Springer-Verlag 2002 Funding for the study was provided by an Australian Commonwealth Department of Health, and Family Services Research and Development Grant (HS338). K.M. Hillman ( ) P.J. Bristow K. Daffurn S.L. Norman G.F. Bishop Liverpool Hospital, Liverpool BC, NSW 1871, Australia, Australia e-mail: ken.hillman@unsw.edu.au Tel.: +61-2-98283585 Fax: +61-2-98283577 K.M. Hillman University of New South Wales, Sydney, Australia T. Chey Epidemiology Unit, South Western Sydney Area Health Service, Sydney, Australia T. Jacques St George Hospital, Sydney, Australia G. Simmons Illawarra Regional Hospital, Wollongong, Australia Abstract Objective: To document the characteristics and incidence of serious abnormalities in patients prior to admission to intensive care units. Design and setting: Prospective follow-up study of all patients admitted to intensive care in three acute-care hospitals. Patients: The study population totalled 551 patients admitted to intensive care: 90 from the general ward, 239 from operating rooms (OR) and 222 from the Emergency Department (ED). Measurements and results: Patients from the general wards had greater severity of illness (APACHE II median 21) than those from the OR (15) or ED (19). A greater percentage of patients from the general wards (47.6%) died than from OR (19.3%) and ED (31.5%). Patients from the general wards had a greater number of serious antecedents before admission to intensive care 43 (72%) than those from OR 150 (64.4%) or ED 126 (61.8%). Of the 551 patients 62 had antecedents during the period 8 48 h before admission to intensive care, and 53 had antecedents both within 8 and 48 h before their admission. The most common antecedents during the 8 h before admission were hypotension (n=199), tachycardia (n=73), tachypnoea (n=64), and sudden change in level of consciousness (n=42). Concern was expressed in the clinical notes by attending staff in 70% of patients admitted from the general wards. Conclusions: In over 60% of patients admitted to intensive care potentially life-threatening abnormalities were documented during the 8 h before their admission. This may represent a patient population who could benefit from improved resuscitation and care at an earlier stage. Keywords Critical care Severity of illness Antecedents Intensive care units Vital signs Prevention Introduction Patients have been reported who received suboptimal management during the period prior to their admission to the intensive care unit (ICU), resulting in a high mortality [1]. The problem appears to be associated particularly with general hospital wards. Patients from general wards admitted to ICU have a higher mortality than those admitted from the emergency department (ED) or operating or recovery room (OR) [2, 3, 4, 5]. Other studies have

1630 Table 1 Patient characteristics and admission source (HDU high-dependency unit) a Test of source independence, median one-way analysis test, χ 2 General Operating theatre, Emergency p a ward recovery department Number of events 90 239 222 n.s. Number of patients 84 228 219 Men (events) 39 (43.3%) 134 (56.1%) 121 (54.5%) Age (years) Mean 64.0±17.2 60.1±18.8 51.8±21.0 Median (range) 68 (17 93) 66 (15 92) 55 (15 88) <0.01 14 29 6 (6.7%) 26 (10.9%) 49 (22.1%) 30 44 7 (7.8%) 24 (10.0%) 41 (18.5%) 45 59 15 (16.7%) 43 (18.0%) 41 (18.5%) 60 74 37 (41.1%) 97 (40.6%) 55 (24.8%) 75+ 25 (27.8%) 49 (20.5%) 36 (16.2%) Days from hospital admission to ICU/HDU Mean 10.3±20.1 4.4±10.6 0.2±0.5 Median (range) 3.0 (0 131) 1.0 (0 91) 0.0 (0 1) <0.01 Same day 5 (5.6%) 72 (30.4%) 180 (81.1%) 1 day 18 (20.0%) 78 (32.9%) 42 (18.9%) 2 7 days 37 (41.1%) 54 (22.8%) 0 8 14 days 16 (17.8%) 15 (6.3%) 0 >14 days 14 (15.6%) 18 (7.6%) 0 Time of admission to ICU <0.05 00.00 <08.00 hours 16 (17.8%) 42 (17.6%) 59 (26.6%) 08.00 <16.00 hours 39 (43.3%) 84 (35.1%) 60 (27.0%) 16.00 <24.00 hours 35 (38.9%) 113 (47.3%) 103 (46.4%) Reason for ICU admission (primary system) <0.01 Respiratory 33 (36.7%) 7 (2.9%) 34 (15.3%) Cardiovascular 18 (20.0%) 54 (22.6%) 36 (16.2%) Neurological 13 (14.4%) 45 (18.8%) 42 (18.9%) Gastrointestinal 8 (8.9%) 84 (35.2%) 10 (4.5%) Trauma 1 (1.1%) 32 (13.4%) 54 (24.3%) Sepsis 12 (13.3%) 2 (0.8%) 14 (6.3%) Others 5 (5.6%) 15 (6.3%) 32 (14.4%) reported a surprisingly high mortality in patients discharged to general wards from ICU who were expected to survive [2, 6, 7]. The reasons for the high mortality include delayed recognition and the lack of an organized approach to at-risk patients [1]. It has been suggested that early identification of these patients, both before admission and after discharge from ICU, may decrease mortality rates [2, 8]. The primary purpose of this study was to examine the incidence of serious abnormalities prior to admission to the ICU in acute hospitals. Methods Study Design This study was a prospective assessment of all patients admitted to ICUs in three acute-care hospitals between 8 July and 31 December 1996. The hospitals had, respectively, 386, 493 beds and 532 beds. There were a total of 50,942 admissions in the three hospitals over the study period, 551 of which were to an ICU (531 patients): 90 from the general ward, 239 from the OR, and 222 from the ED. Data were not included on patients transferred to ICU from another hospital or from coronary care units or high-dependency units within the same hospital. As soon as practical after ICU admission all patients aged 14 years or over had their medical record reviewed for information including patient demographics, day of hospital admission, day of ICU admission, time of ICU admission, source of ICU admission, Acute Physiology and Chronic Health Evaluation (APACHE) II [9] diagnosis and score, Apache III [10] score, and Simplified Acute Physiology Score (SAPS) II [11] score for the first 24 h. Over 60% of the patients admitted from the general wards and OR were aged 60 years or older (Table 1). The median length of hospital stay before admission to ICU was longer for patients from the general wards (Table 1). Data were also collected on the presence of potentially lifethreatening antecedent factors in the periods 0 8 h and 8 48 h before admission to the ICU. The factors were defined as a major threat to the airway associated with neck swelling, difficulty in swallowing saliva or stridor, respiratory rate of 5 or less per minute or higher than 36 per minute, pulse rate of 40 or less per minute or higher than 140 per minute, systolic blood pressure of 90 mmhg, sudden fall in Glasgow Coma Scale of 2 or more, and

1631 Table 2 Illness severity, length of stay and hospital outcome related to admission source General ward Operating theatre, Emergency p a (n=90) recovery (n=236) department (n=212) APACHE II <0.01 Mean 22.2±9.7 16.1±9.1 19.4±10.0 Median (range) 21.0 (0 45) 15.0 (0 81) 19.0 (0 47) APACHE III <0.01 Mean 82.1±36.7 60.0±31.3 71.5±39.1 Median (range) 78.5 (0 164) 55.5 (3 148) 70.0 (0 183) SAPS II <0.01 Mean 46.7±20.1 36.0±16.4 42.4±22.9 Median (range) 44.0 (0 94) 34.0 (0 84) 42.0 (0 211) Event outcome by patient 84 228 219 Total length of stay in hospital (days) Mean 22.6 (24.1) 23.6 (26.0) 11.9 (16.6) Median (range) 14.5 (1 145) 15.5 (1 175) 6.0 (1 119) <0.01 1 day 3 (3.6%) 12 (5.3%) 42 (19.2%) 2 7 days 22 (26.2%) 37 (16.2%) 80 (36.5%) 8 14 days 17 (20.2%) 58 (25.4%) 39 (17.8%) >14 days 42 (50.0%) 121 (53.1%) 58 (26.5%) Length of stay from admission to ICU to leaving hospital (days) Mean 14.5 (16.9) 19.3 (23.6) 11.7 (16.6) Median (range) 9.0 (0 106) 12.0 (0 173) 6.0 (0 118) <0.01 Same day 6 (7.1%) 3 (1.3%) 5 (2.3%) 1 day 11 (13.1%) 16 (7.0%) 41 (18.7%) 2 7 days 20 (23.8%) 50 (21.9%) 79 (36.1%) 8 14 days 17 (20.2%) 68 (29.8%) 37 (16.9%) >14 days 30 (35.7%) 91 (39.9%) 57 (26.0%) Discharge status from hospital (by patient: total) <0.01 b Died in hospital 40 (47.6%) 44 (19.3%) 69 (31.5%) Discharge home 29 (34.5%) 141 (61.8%) 100 (45.7%) Transfer to another hosp 5 (6.0%) 15 (6.6%) 15 (6.8%) Transfer to nursing home 2 (2.4%) 5 (2.2%) 5 (2.3%) Others 8 (9.5%) 23 (10.1%) 30 (13.7%) Total length of stay in hospital for survival of hospital stay (days) 44 184 150 Mean 29.8±27.6 25.3±24.7 15.2±18.8 Median (range) 23.5 (1 145) 16.0 (1 148) 10.0 (1 119) <0.01 a Test of source independence, median one-way analysis test, χ 2 b For percentage of deaths only repeated or prolonged seizures. The cases were also noted in which concern by attending clinical nursing or junior medical staff, relating to the patient s condition, was explicitly written in the patients clinical notes. Information was also collected on ICU and hospital discharge date and status. Data were collected by a dedicated critical care nurse at each of the three hospitals, trained in the use of a form, specifically designed to collect information on the incidence of serious abnormalities before admission to be ICU. The form had been refined in a 2-week pilot study before data collection began. The nurses were familiar with the medical record at the respective hospitals. Where the information contained in the clinical notes was unclear, it was clarified with the treating staff. Incomplete forms and those with mutually exclusive results were returned to the data collector for checking. Forms were then entered onto a database (Microsoft Access 2.0, Microsoft, Redmond, Wash., USA, 1994). A data collector then compared all forms with the database to ensure tha no typographical errors had occurred. Checking was performed to determine errors such as duplicated results and anomalous data. The original medical records were reviewed as needed and the database corrected. Data were then exported to SAS v6.12 (SAS Institute, Cary N.C., USA, 1997) for analysis. Results were tabulated as number, percentage of factor prevalence and mean or median for continuous measures. Comparison of prevalence of factor by source of ICU admission was carried out using the χ 2 test of independence and median one-

1632 Table 3 Presence of potentially life-threatening antecedent factors and admission source: non-cardiorespiratory arrest patients with abnormal antecedents within 8 h of admission to ICU ( worried concern written in clinical notes about the patient s condition by attending nursing or junior medical staff, GCS Glasgow Coma Scale) Ward Operating theatre, Emergency department p a (n=59) recovery (n=233) (n=204) n % n % n % Presence of 0 8 h antecedents ns Airway threatened 2 3.4 7 3.0 25 12.3 Respiratory rate <5 0 0 0 2 1.0 Respiratory rate >36 20 33.9 10 4.3 34 16.7 Pulse rate <40 0 7 3.0 8 3.9 Pulse rate >140 13 22.0 21 9.0 40 19.6 Systolic blood pressure <90 14 23.7 127 155.5 58 28.4 Sudden fall in GCS >2 5 8.5 13 5.6 24 11.8 Repeated/prolonged seizure 4 6.8 1 0.4 9 4.4 Worried 41 69.5 One or more antecedents ( worry ) 43 72.9 150 164.4 126 161.8 <0.01 One only 31 52.5 119 151.1 69 33.8 2 or more 12 20.3 31 13.3 57 27.9 Number of patients with abnormal 58 157 25 antecedents within 8 48 h of admission Presence of 8 48 h criteria <0.01 Airway threatened 1 1.7 0 0 Respiratory rate >36 12 20.7 2 1.3 8 32.0 Pulse rate <40 1 1.7 0 0 Pulse rate >140 6 10.3 3 1.9 4 16.0 Systolic blood pressure <90 11 19.0 22 14.0 5 20.0 Sudden fall in GCS >2 1 1.7 0 1 4.0 Repeated/prolonged seizure 3 5.2 0 1 4.0 Worried 25 43.1 One or more antecedents ( worry ) 25 43.1 23 14.7 14 56.0 <0.01 One only n (%) 16 27.6 19 12.1 9 36.0 Two or more n (%) 9 15.5 4 2.6 5 20.0 Presence of antecedents in both 0 8 and 8 48 h periods 58 58 157 157 25 25 Airway threatened 1 1.7 0 0 0 0 Respiratory rate >36 6 10.3 1 0.6 3 12.0 Pulse rate >140 2 3.4 3 1.9 3 12.0 Systolic BP <90 8 13.8 19 12.1 2 8.0 Sudden fall in GCS >2 1 1.7 0 0 1 4.0 Repeated/prolonged seizure 3 5.2 0 0 1 4.0 Worried 19 32.8 One or more antecedents ( worry ) 21 36.2 21 13.3 11 45.8 a Test of source independence, median one-way analysis test, χ 2 way analysis (χ 2 approximation) for median scores. Event was the primary unit of analysis. This study was part of a larger investigation on medical emergency processes in the three hospitals. The study proceeded after review by the three hospitals ethics committees and subsequent ratification by the relevant university body. Each body waived the need for informed consent. Results The time of ICU admission differed significantly between the three areas, with most patients from the general wards and OR admitted during the day or evening, whereas patients from the ED were admitted mostly after 4.00 p.m. The primary reason for ICU admission also differed significantly: general ward patients were mainly due to cardiorespiratory disorder, those from the OR to gastrointestinal, cardiovascular or neurological system disorders, and those from the ED to trauma, neurological, cardiovascular or respiratory disorders. Patients from the general wards had significantly higher severity of illness as measured by APACHE II and III and SAPS II (Table 2). The hospital mortality rate among patients admitted from the general wards (48%) was significantly higher than that from the OR (19%) or ED (31%). In patients who survived to hospital discharge the median hospital stay was significantly longer in those admitted from the general wards (median 23.5 days) than in those from the OR (median 16 days) or ED (median 10 days). Of the 496 ICU patients 55 were admitted after cardiorespiratory arrest, and 319 (64%) had life-threatening

1633 abnormalities before admission. Of 341 patients 62 (18%) had abnormalities in the 8- to 48-h period before admission to the ICU, and 53 patients (16%) had serious abnormalities during both time periods. Almost threequarters of patients from the general wards had at least one potentially life-threatening antecedent factor during the 8 h before admission to the ICU (Table 3). A large proportion of patients from OR (64%) and ED (62%) also had at least one of these factors present. Patients from the ED had a higher proportion with two or more antecedent factors (28%) than those from general wards (20%) or OR (13%; Table 3). Concern was frequently recorded in the clinical notes of patients admitted to the ICU from the general wards by attending nursing and junior medical staff (Table 3). Discussion Well over one-half of patients admitted to the ICU had potentially serious vital sign abnormalities within 8 h and many for up to 48 h before admission. Because the abnormality was only noted when documented in the clinical notes, the real incidence may have been even higher. A high incidence of patients with shock on admission to ICU has been noted in other studies [12, 13]. Many patients in our study were hypotensive, tachycardic or tachypnoic. Long periods of ischaemia and hypoxia before definitive treatment in the ICU is one of the postulated reasons as to why therapeutic measures in the ICU do not improve survival [13, 14, 15]. The management of patients prior to and after their admission from the ICU may have a marked effect on eventual outcome, without necessarily reflecting on the care delivered within the ICU itself. The importance of this lead-time bias in interpreting inter-hospital comparisons of ICUs was recognized soon after severity scoring systems were developed for ICUs [16, 17]. The presence of untreated and severe abnormalities in vital signs found in this study for many hours before admission to the ICU raises questions that ICU outcome status may not be testing the standard of treatment in the ICU as much as the system proximal to the ICU. Other studies have demonstrated serious deficiencies in the care of patients prior to admission to ICU [1, 2, 3, 4, 5, 6]. The reasons that have been postulated for suboptimal care prior to ICU admission include lack of knowledge by attending clinical staff, failure to appreciate clinical urgency, lack of appropriate supervision and failure to seek advice [1, 5, 8]. While patients from all sites prior to ICU admission had a high incidence of serious abnormalities in vital signs, the environments from which the patients are admitted are quite different. For example, in the OR, the patient is usually closely supervised by a clinician with training in anaesthesia. Similarly, ED provides monitoring and supervision by staff trained in all aspects of advanced resuscitation. The high number of patients with one or more serious abnormalities from the ED and OR were presumably being closely monitored and appropriately managed, despite being temporarily unstable. In contrast patients from the general wards with serious abnormalities in vital signs were in an environment with less monitoring and inconsistent supervision by staff, who may or may not have had training in advanced resuscitation. Other studies have demonstrated a poorer outcome for patients from general wards [1, 2, 3, 5]. Patients from the general wards in this study had a greater severity of illness on admission to the ICU (Table 1). There was a higher mortality and longer hospital stay in patients admitted to ICU from the general wards than in those from OR or ED. A higher rate of serious abnormalities occurred in patients from the general wards, and this may have resulted in poor outcomes. The presence of these abnormalities in ward patients may identify a high-risk group of patients who could benefit from systems to deliver higher acuity care. Specific concern was expressed in the patient notes by attending nursing and junior medical staff in 70% of all patients within 8 h of admission to the ICU from the general wards and in 43% of patients during the 8- to 48-h period before admission. This is consistent with previous studies which have shown attending staff not to have appropriate skills and knowledge about seriously ill patients in the general ward environment [1]. It also emphasizes the need for a more appropriate system-wide response to these patients. Hypotension was the commonest antecedent factor, especially in patients admitted from the OR. Tachypnoea and tachycardia were also common, regardless of admission source. Tachypnoea is associated with hypoxia and has previously been identified as being associated with patients who are at risk of serious deterioration [4, 18]. Repeated or sustained seizures was the only specific disease category used and has for many reasons been associated with sudden death [19]. While the other abnormalities in vital signs such as level of consciousness, bradycardia and bradypnoea were less common, they remain serious and potentially life-threatening. Editorials urging more proactive management have been prompted by studies demonstrating inadequate care before admission to the ICU [8, 20, 21]. One of the options for early identification and resuscitation of patients in an acute-care hospital is the institution of a medical emergency team which replaces the conventional cardiac arrest team [22, 23, 24]. The definition of serious physiological abnormalities used in this study coincide with the medical emergency team criteria for early identification of patients. Staff trained in all aspects of advanced resuscitation respond immediately to in-hospital patients with these criteria. The high incidence of potentially treatable antecedents to ICU admission may be a basis for a more systematic approach to seriously ill patients across the whole hospital.

1634 References 1. McQuillan P, Pilkington S, Alan A, Taylor B, Short A, Morgan G, Nielsen, Barrett D, Smith G (1998) Confidential inquiry into quality of care before admission to intensive care. BMJ 316:1853 1858 2. Goldhill DR, Sumner A (1998) Outcome of intensive care patients in a group of British intensive care unit. Crit Care Med 26:1337 1345 3. Escarce JJ, Kelley MA (1990) Admission source to the medical intensive care unit predicts hospital death independent of APACHE II score. JAMA 264:2389 2394 4. Sax FL, Charlson ME (1987) Medical patients at high risk for catastrophic deterioration. Crit Care Med 15:510 515 5. Lundberg JS, Perl TM, Wiblin T, Costigan MD, Dawson J, Nettleman MD, Wenzel RP (1998) Septic shock: an analysis of outcomes for patients with onset on hospital wards versus intensive care units. Crit Care Med 26:1020 1024 6. Wallis CB, Davies HTO, Shearer AJ (1997) Why do patients die on general wards after discharge from intensive care units? Anaesthesia 52:9 14 7. Lawrence A, Havill JH (1999) An audit of deaths occurring in hospital after discharge from the Intensive Care Unit. Anaesth Intensive Care 27:185 189 8. Bion J (1995) Rationing intensive care. Preventing critical illness is better, and cheaper, than cure. BMJ 310:682 683 9. Knaus WA, Draper EA, Wagner DP, Zimmerman JE (1985) APACHE II: a severity of disease classification system. Crit Care Med 13:818 829 10. Knaus WA, Wagner DP, Draper EA, Zimmerman JE, Bergner M, Bastos PG, Sino CA, Murphy DJ, Lotring T, Damiano A (1991) The APACHE III prognostic system risk prediction of hospital mortality for critically ill hospitalised adults. Chest 100:1619 1636 11. Le Gall J-R, Lemeshow S, Saulnier F (1993) A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA 270:2957 2963 12. Gutierrez G, Palizas F, Doglio G, Wainsztein N, Gallesio A, Pacino J, Dibin A, Schiavi E, Jorge M, Pusajo J (1992) Gastric intramucosal ph as a therapeutic index of tissue oxygenation in critically ill patients. Lancet 339:195 199 13. Hayes MA, Timmins AC, Yau EH, Palazzo M, Hinds CJ, Watson D (1994) Elevation of systemic oxygen delivery in the treatment of critically ill patients. N Engl J Med 330:1717 1722 14. Kirschenbaum L (1998) Antibodies to TNF-α: too little, too late? Crit Care Med 26:1625 1626 15. Chernow B (ed) (1996) Back to the drawing board. Crit Care Med 24:1097 1098 16. Bion JF, Edlin SA, Ramsay G, McCabe S, Ledingham IM (1985) Validation of a prognostic score in critically ill patients undergoing transportation. BMJ 291:432 434 17. Dragsted L, Jorgensen J, Jensen N-H, Bonsing E, Jacobsen E, Knaus WA, Qvist J (1989) Interhospital comparisons of patient outcome from intensive care: importance of lead-time basis. Crit Care Med 17:418 422 18. Tobin MJ (1992) Breathing pattern analysis. Intensive Care Med 18:193 201 19. Nashef L, Brown S (1996) Epilepsy and sudden death. Lancet 348:1324 1325 20. Garrad C, Young D (eds) (1998) Suboptimal care of patients before admission to intensive care. BMJ 316:1841 1842 21. Hillman KM (1996) Reducing preventable deaths and containing costs: the expanding role of intensive care medicine. Med J Aust 164:308 309 22. Lee A, Bishop G, Hillman KM, Daffurn K (1995) The medical emergency team. Anaesth Intensive Care 23:183 186 23. Hourihan F, Bishop G, Hillman KM, Daffurn K, Lee A (1995) The medical emergency team: a new strategy to identify and intervene in high risk patients. Clin Intensive Care 6:269 272 24. Hillman KM, Bishop G, Lee A, Daffurn K, Bauman A, Crispin C, Ince L, Bristow P, Hourihan F (1996) Identifying the general ward patient at high risk of cardiac arrest. Clin Intensive Care 7:242 243