Intensive care unit safety incidents for medical versus surgical patients: A prospective multicenter study B

Journal of Critical Care (2007) 22, 177 183 Health Services Research Intensive care unit safety incidents for medical versus surgical patients: A prospective multicenter study B David J. Sinopoli MPH, MBA a,b, Dale M. Needham MD, PhD c,1, David A. Thompson DNSc, MS d, Christine G. Holzmueller BLA d, Todd Dorman MD d, Lisa H. Lubomski PhD d, Albert W. Wu MD, MPH e, Laura L. Morlock PhD e, Martin A. Makary MD, MPH f, Peter J. Pronovost MD, PhD d,e,f, * a UMDNJ-Robert Wood Johnson Medical School, Piscataway, NJ 08854 b Johns Hopkins University School of Professional Studies in Business and Education, Baltimore, MD 21218, USA c Pulmonary and Critical Care Medicine, Johns Hopkins University School of Medicine, Baltimore, MD 21287, USA d Department of Anesthesiology/Critical Care Medicine, Johns Hopkins University School of Medicine, Baltimore, MD 21287, USA e Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD, USA f Department of Surgery, Johns Hopkins University School of Medicine, Baltimore, MD, USA Keywords: Medical errors; Safety management; Sentinel surveillance; Incident reporting; Critical care; Intensive care units Abstract Purpose: The aim of this study is to determine if patient safety incidents and the system-related factors contributing to them systematically differ for medical versus surgical patients in intensive care units. Materials and Methods: We conducted a multicenter prospective study of 646 incidents involving adult medical patients and 707 incidents involving adult surgical patients that were reported to an anonymous patient safety registry over a 2-year period. We compared incident characteristics, patient harm, and associated system factors for medical versus surgical patients. Results: The proportion of safety incidents reported for medical versus surgical patients differed for only 3 of 11 categories: equipment/devices (14% vs 19%; P =.02), bline, tube, or drainq events (8% vs 13%; P =.001), and computerized physician order entry (13% vs 6%; P V.001). The type of patient harm associated with incidents also did not differ. System factors were similar for medical versus surgical patients, with training and teamwork being the most important factors in both groups. Conclusions: Medical and surgical patients in the intensive care unit experience very similar types of safety incidents with similar associated patient harm and system factors. Common initiatives to improve patient safety for medical and surgical patients should be undertaken with a specific focus on improving training and teamwork among the intensive care team. D 2007 Elsevier Inc. All rights reserved. B This study was funded by the Agency for Healthcare Research and Quality (grant no. U18HS11902). Analyses were performed at The Johns Hopkins University School of Medicine and Bloomberg School of Public Health. * Corresponding author. Department of Anesthesiology/Critical Care Medicine, Baltimore, MD 21287, USA. Tel.: +1 410 502 3231. E-mail address: ppronovo@jhmi.edu (P.J. Pronovost). 1 Supported by a Clinician-Scientist Award from the Canadian Institutes of Health Research. 0883-9441/$ see front matter D 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jcrc.2006.11.002

178 1. Introduction Patient safety incidents and medical errors occur frequently [1-5]. A prospective examination of surgical patients [6] discovered that complication rates may be 2 to 4 times higher than previously described [3], and suggested that almost half of incidents were preventable. Many reports also demonstrate that medical patients are subject to injury from preventable errors [7-9]. Although previous patient safety studies have focused on specific types of incidents (eg, adverse drug events [10-12]) or care settings (eg, intensive care units [ICUs]) [13,14], little research has directly compared incidents involving medical versus surgical patients. Given differences in admitting diagnoses, hospital treatment, and organizational culture within hospital units [15], safety incidents may systematically differ for medical versus surgical patients. A comparison of incidents between these 2 patient groups may help clinicians design safety evaluations and interventions that are most appropriate for these different patient populations. The ICU is a highly complex environment [16] that cares for both medical and surgical patients. This setting provides an important opportunity to evaluate patient safety [1,14,17]. Contemporary patient safety research focuses on the role that systems, rather than specific individuals, play in contributing to, limiting, and preventing safety incidents [17-20]. The objective of this study was to compare these contributing, limiting, and preventive system factors associated with safety incidents for medical versus surgical patients in ICUs, and to compare the characteristics and type of patient harm associated with these incidents. We hypothesize that there are significant differences between medical and surgical ICU patients in the type of safety incidents, and the associated harm and system factors. 2. Methods This research was conducted using prospective patient safety reports from a voluntary, anonymous, Web-based ICU Safety Reporting System (ICUSRS). This study focused on incidents reported to the ICUSRS from July 1, 2002, to June 30, 2004. During this time, 20 adult ICUs (3 medical, 10 surgical, 7 medical-surgical) located primarily in the northeastern, midatlantic, and southern United States reported patient safety incident data to the ICUSRS. Of the 20 participating ICUs, 19 were at teaching hospitals, with a mean (range) number of ICU beds of 17 (9-24). For this research, incidents were defined as unsafe conditions and events in ICUs that could, or did, lead to patient harm [21]. 2.1. Background on the ICUSRS The ICUSRS has served as a model for other systems and has been actively used to assess patient safety in the ICU, as described elsewhere [16,22-25]. Reports at participating hospitals were submitted by ICU staff who received training on appropriate use of this Web-based reporting system. 2.2. Data collection The data provided in incident reports included characteristics of the patient, ICU and incident, and the associated patient harm. bsurgicalq patients were identified by ICU staff, and defined as any patient undergoing surgery immediately before or during their ICU admission. The type of incident reported (eg, airway, medication) was coded centrally by an ICUSRS project team member according to a written taxonomy. Each report may be associated with more than one type of incident and patient harm. A trained ICUSRS staff member along with an ICU clinician checked for appropriateness of the coding for type of incident. Initially, the principal investigator (an ICU physician and patient safety expert) reviewed 100% of the coded cases. There was 99.6% interrater reliability for the first 200 cases. As a result, reliability was checked for only 20% of the subsequent cases. Further methodological information can be found elsewhere [22]. In each report, ICU staff also indicated relevant system factors according to an established methodology [26]. These reports evaluated the role of 7 possible system factors: patient, provider, team, training, task, organization, and management [19]. These factors can play a role in leading to an incident (bcontributing factorq), minimizing the negative impact of an incident after it occurs (blimiting factorq), and/ or preventing an incident (bpreventive factorq). Each factor has a standard definition to facilitate consistent reporting, as reported in detail elsewhere [23,24]. Because these factors are not mutually exclusive, incident reports frequently include multiple system factors. 2.3. Statistical analysis D.J. Sinopoli et al. v 2 tests were performed to compare the incident and patient characteristics, harm, and system factors for safety incidents involving medical versus surgical patients. Significant differences observed in the univariable analyses of patient type and system factors were further analyzed using multivariable logistic regression models adjusting for significant univariable differences observed in patient characteristics, ICU type, and incident characteristics. Collinearity of the patient, incident, and system factor variables were evaluated using variance inflation factors [27]. Statistical interactions were also evaluated within the multivariable logistic regression model. Separate logistic regression analyses were performed for each category of system factor (ie, contributing, limiting, and preventive). Results were considered statistically significant at P V.05. STATA version 8.0 [27] was used for all analyses. Institutional Review Board approval for the overall study was granted by the Johns Hopkins University School of Medicine, and each participating site obtained approval from its respective Institutional Review Board.

ICU safety incidents for medical versus surgical patients 179 3. Results A total of 1582 incidents involving adult patients were reported to the ICUSRS. We excluded 195 (12.3%) reports for which patient surgical status was unavailable, and 34 (2.1%) potentially duplicate reports. There were no significant differences in the excluded versus included group of incidents with respect to patient or incident characteristics, patient harm, type of incident, or associated system factors. The final data set included 646 (48%) and 707 (52%) events involving adult medical and surgical patients, respectively. 3.1. Patient, ICU, and incident characteristics Of the reported incidents, 57% involved men and 48% involved patients older than 65 years. These patient characteristics did not differ between medical and surgical reports ( P =.98 and P =.21, respectively) (Table 1). The admitting diagnoses were more frequently circulatory or respiratory, and less frequently digestive or genitourinary for incident reports involving medical versus surgical patients ( P b.001) (Table 1). As expected, there was a significant difference in the proportion of incident reports, by ICU type, for medical versus surgical patients ( P b.001). For example, a larger proportion of incidents for surgical versus medical patients was reported from surgical ICUs (62% vs 12%) (Table 1). However, mixed medical-surgical ICUs were an important source of incident reports for both medical (43%) and surgical (30%) patients. No significant differences were noted in the characteristics of safety incidents for medical and surgical patients (Table 1). Overall, nurses reported more than 70% of incidents. Approximately 80% of incidents were considered preventable and more than 40% caused harm. The timing of incidents in both groups was similar. Incidents were reported to patients family or friends in 18% of cases. 3.2. Incident type and harm The types of safety incidents reported for both patient groups were similar (Fig. 1). Issues related to bcommunication,q bclinical management,q and bicu managementq were reported in more than 50% of incidents for both medical and surgical patients. There was a significant difference in 3 of the 11 types of incidents. Incidents involving bequipment/devicesq and a bline, tube, or drainq were less common in medical versus surgical patients (14% vs 19% [ P =.02] and 8% vs 13% [ P =.001], respectively), whereas incidents related to computerized physician order entry were more common in medical patients (13% vs 6%; P b.001). Comparison of the types of patient harm associated with reported incidents revealed no significant differences (Fig. 2). Physical injury and anticipated or actual prolonged length of stay occurred in approximately 15% and 10% of all reports, respectively. Death, at the time the incident was reported, occurred in 2% or less of cases. Table 1 Characteristics of reported ICU safety incidents involving medical versus surgical patients Characteristic Number (%) of incidents a P a 3.3. System factors Medical (n = 646) Surgical (n = 707) Patient characteristics Male 366 (57) 399 (57).98 Age N65 y 276 (46) 321 (49).21 Diagnosis at admission b b.001 Circulatory/Respiratory 209 (58) 164 (47) Digestive/Genitourinary 41 (11) 94 (27) Infectious 36 (10) 21 (6) Endocrine/Metabolic/ 36 (10) 16 (5) Hematologic Neurological/Mental 20 (6) 10 (3) Injury/Poisoning 10 (3) 18 (5) Neoplasm 2 (1) 11 (3) Skin/Musculoskeletal 4 (1) 15 (4) ICU type b.001 Medical 285 (45) 58 (8) Surgical 74 (12) 431 (62) Medical-Surgical 274 (43) 208 (30) Incident characteristics Timing.08 Weekday (7 am-7 pm) 288 (47) 360 (53) Weeknight (7 pm-7 am) 203 (33) 213 (31) Weekend 124 (20) 112 (16) Reporter.22 Nurse 479 (75) 489 (72) Doctor 87 (14) 98 (14) Other c 68 (11) 93 (14) Preventable 491 (81) 513 (80).62 Harmed patient d 282 (44) 297 (42).57 Informed Family/Friends 118 (18) 126 (18).85 a The total number of incidents reported for each characteristic may vary slightly due to missing data. P values were calculated using the v 2 statistic. b Diagnosis data only reported in ICUSRS from July 2002 to October 2003 for 358 medical and 349 surgical patients. c Includes nurse practitioners, respiratory therapists, pharmacists, and nurses aides. d Harm includes any death, physical injury, physiologic change, discomfort, psychological distress, dissatisfaction of relatives and anticipated or actual prolonged length of stay. See Fig. 2 for more detailed breakdown. The system factors associated with incidents in medical versus surgical patients were evaluated in 3 separate analyses based on their independent roles as contributing, limiting, or preventive factors. First, as contributing factors, 3 of 7 system factors significantly differed between medical and surgical patients: patient, provider, and team factors contributed more frequently to incidents reported in medical versus surgical patients (each P V.02) (Fig. 3). However, in a multivariable logistic regression analysis adjusting for differences in patient diagnosis, ICU type, and incident

180 D.J. Sinopoli et al. Fig. 1 Type of safety incident. Each safety incident reported can have more than one type. btherapeuticq incidents relate to medication and other therapeutic interventions; bairwayq incidents relate to the patient s anatomic airway or endotracheal tube; bcpoeq incidents relate to computerized physician order entry systems. An asterisk (*) indicates a significant difference (all P V.02) by v 2 test. timing (Table 1), only patient system factors remained significantly different (odds ratio, 1.50; 95% confidence interval, 1.20-1.88) between medical and surgical patients (results not shown). Second, a comparison of the system factors in their potential role for limiting the negative impact of an incident revealed no significant differences between medical versus surgical patients ( P z.21 for all comparisons, results not shown). Third, comparing the system factors in their potential role for preventing safety incidents revealed 2 factors that differed between medical and surgical patients. Incidents reported in medical versus surgical patients more frequently involved a provider (57% and 47%; P b.001) or team (39% and 34%; P =.04) preventive system factor. However, in a multivariable logistic regression analysis (as described above), these factors were not significantly different between medical and surgical ICU patients (results not shown). In all multivariable logistic regression models, there was no significant collinearity or interaction terms. In all 3 roles (contributing, limiting, and preventive), training and team system factors were the most common in both medical and surgical patients, being reported in 30% to 50% of all incidents. 3.4. Sensitivity analyses Data regarding the diagnosis of patients at admission were available for only 707 of the safety incidents, largely comprising the first 15 months of data collection (Table 1). As a post hoc sensitivity analysis, separate multivariable regression analyses were performed using a model that excluded this patient diagnosis variable. This sensitivity analysis did not result in any material difference in findings from the primary analysis (results not shown). Fig. 2 Type of patient harm. Multiple types of patient harm can be associated with each reported safety incident. Comparison of medical versus surgical patients for each type of harm was not significant ( P N.05) by v 2 test. LOS indicates length of stay.

ICU safety incidents for medical versus surgical patients 181 Fig. 3 Factors contributing to safety incidents. Multiple factors can contribute to each safety incident reported. An asterisk (*) indicates a significant difference (all P V.02) by v 2 test; however, in a multivariable logistic regression analysis adjusting for differences in patient diagnosis, ICU type, and incident timing (see Table 1), only patient-related system factors remained significantly different (odds ratio, 1.50; 95% confidence interval, 1.20-1.88) between medical and surgical patients. As a second post hoc sensitivity analysis, we assessed if the incident characteristics, associated harm, and system factors between medical and surgical patients differed for the 3 ICUs, each reporting more than 10% of all the ICUSRS incidents versus the remaining 17 ICUs. Comparison of results from the separate multivariable regression analyses for each group of ICUs revealed no material differences (results not shown). 4. Discussion This multicenter prospective study comparing 1353 patient safety incidents in critically ill medical versus surgical patients demonstrated great similarity in patient and incident characteristics, harm, and associated system factors between these 2 distinct patient groups. Our findings were contrary to the hypothesis that such factors would differ for medical versus surgical ICU patients. Because little prior research had been done to address this specific question, we based our hypothesis on knowing that the diagnosis, treatment, and medical complexity often differs between medical and surgical ICU patients, and that these differences have been associated with patient safety and medical errors [1,5,9,23,24,28,29]. The few differences between medical and surgical patients that were observed in this study were not unexpected. For example, the higher frequency of incidents related to a line, tube, or drain in surgical patients reflects the greater frequency with which these procedures are performed in this patient group. Similarly, the increased frequency of incidents related to computerized order entry in medical patients is likely due to a higher number of medication orders in this group. These differences in the nature of medical and surgical patients were also reflected in the finding that patient-related system factors, in their role as contributing to safety incidents, were the only system factors significantly different between these 2 groups. Consistent with prior literature, our study also demonstrated that most of the safety incidents were preventable [5,6,28] and may be serious, resulting in physical injury and an increased length of stay [9,30]. Furthermore, safety incidents very commonly involve communication issues among the health care team [14,28,31,32]. Relatively few incidents involving either medical or surgical patients were reported to family or friends, consistent with previous studies [33,34] and perhaps due to fear of malpractice suits, damage to reputation, or avoidance of awkward or uncomfortable situations [35]. Finally, we demonstrated that training and team system factors were critical in contributing to, limiting, and preventing safety incidents [13,32,36,37]. 4.1. Implications for patient safety The similarity between medical and surgical patients in the type of incident, associated harm, and system factors has important implications. In the United States, critically ill medical and surgical patients are often cared for by different physicians, and sometimes in different ICUs that frequently have different clinical protocols and practice patterns [38]. These differences may reflect a belief that critically ill medical and surgical patients are fundamentally different. However, our study of patients in both mixed and separate medical and surgical ICUs demonstrates that, from a safety perspective, there are few important differences between these patient groups. The similarity in medical issues (eg, respiratory failure, sepsis, etc) faced by critically ill patients may be more important than differences in their admitting diagnoses and hospital care. Consequently, across different types of critically ill patients and ICUs, patient safety evaluations and interventions can be similar what is beneficial for one patient group or ICU will likely be

182 beneficial for other patients and ICUs. Furthermore, our study highlights that efforts to improve patient safety in the ICU should focus on training- and team-related system factors that are important across different types of patients and ICUs. 4.2. Limitations This study has potential limitations. First, the ICUSRS is a voluntary reporting system, and safety incidents are likely incompletely reported or underreported, although we cannot directly estimate the magnitude of this limitation [3,9,19,21,31]. Consequently, even if data on the number of patient admissions or ICU days were available within ICUSRS, these results and those from any other selfreported safety database should not be used to estimate the incidence of medical errors. However, the consistency of this study and other ICUSRS reports [23,24] with prior research using alternative study designs supports the validity of our system for identifying important system factors for improving patient safety. In addition, there were no differences in the relationships evaluated in this study, between ICUs contributing relatively large numbers of incidents and ICUs contributing fewer incidents to the ICUSRS, suggesting consistency in findings across study sites regardless of the volume of incidents reported. Furthermore, more severe or notable incidents were more likely to have been reported, which may be the most important incidents for system factors analysis, thus minimizing the impact of this potential limitation on our study. Second, the reporting of system factors by ICU staff may be subjective and vary by ICU site or type [39], introducing a potential bias when compared to studies that directly observed incidents. However, training was provided to staff at participating ICUs to help assure consistency and quality. Furthermore, the observed consistency in safety reports across the types of ICUs and patients in this study provides reassurance regarding the limited impact of this potential bias. Third, only 20 ICUs across the United States were involved in this study, and 19 of these were based at teaching hospitals. Thus, the study results may not be generalizable to all ICUs. However, to date, this is the largest such ICU safety reporting project conducted in the United States and demonstrates the feasibility of this approach for larger-scale implementation. Fourth, our study may not have had sufficient power to detect all potential differences between medical and surgical patients, resulting in a false-negative finding. However, our analyses were performed using all available data, reflecting over 1300 incidents occurring in adult patients over the study period, and is the largest study of its kind. These factors may help minimize this potential limitation. In conclusion, medical and surgical patients in the ICU experience similar types of safety incidents, with similar associated patient harm and underlying system factors. Common initiatives to improve patient safety should be undertaken for both medical and surgical patients, with a specific focus on improving training and teamwork among the health care team. Acknowledgments We thank Maureen Fahey and Fern Dickman for their assistance with database issues. References D.J. Sinopoli et al. [1] Bracco D, Favre JB, Bissonnette B, et al. Human errors in a multidisciplinary intensive care unit: a 1-year prospective study. Intensive Care Med 2001;27:137-45. [2] Brennan TA, Leape LL, Laird NM, et al. Incidence of adverse events and negligence in hospitalized patients: results of the Harvard Medical Practice Study, 1991. Qual Saf Health Care 2004;13:145-51. [3] Kohn LT, Corrigan JM, Donaldson MS, editors. To Err is Human: Building a Safer Health System. Washington (DC)7 National Academy Press; 1999. [4] Reason J. Human error: models and management. BMJ 2000;320: 768-70. [5] Wilson RM, Runciman WB, Gibberd RW, et al. The quality in Australian health care study. Med J Aust 1995;163:458-71. [6] Healey MA, Shackford SR, Osler TM, et al. Complications in surgical patients. Arch Surg 2002;137:611-8. [7] Bates DW, Cullen DJ, Laird N, et al. Incidence of adverse drug events and potential adverse drug events. Implications for prevention. ADE prevention study group. JAMA 1995;274:29-34. [8] Leape LL, Bates DW, Cullen DJ, et al. Systems analysis of adverse drug events. ADE prevention study group. JAMA 1995;274:35-43. [9] Weingart SN, Wilson RM, Gibberd RW, et al. Epidemiology of medical error. BMJ 2000;320:774-7. [10] Cullen DJ, Sweitzer BJ, Bates DW, et al. Preventable adverse drug events in hospitalized patients: a comparative study of intensive care and general care units. Crit Care Med 1997;25:1289-97. [11] Osmon S, Harris CB, Dunagan WC, et al. Reporting of medical errors: an intensive care unit experience. Crit Care Med 2004;32:727-33. [12] Pronovost P, Weast B, Schwarz M, et al. Medication reconciliation: a practical tool to reduce the risk of medication errors. J Crit Care 2003;18:201-5. [13] Baggs JG, Schmitt MH, Mushlin AI, et al. Association between nursephysician collaboration and patient outcomes in three intensive care units. Crit Care Med 1999;27:1991-8. [14] Donchin Y, Gopher D, Olin M, et al. A look into the nature and causes of human errors in the intensive-care unit. Crit Care Med 1995;23:294-300. [15] Pronovost P, Sexton B. Assessing safety culture: guidelines and recommendations. Qual Saf Health Care 2005;14:231-3. [16] Wu AW, Pronovost P, Morlock L. ICU incident reporting systems. J Crit Care 2002;17:86-94. [17] Pronovost P, Wu AW, Dorman T, et al. Building safety into ICU care. J Crit Care 2002;17:78-85. [18] Leape LL. Error in medicine. JAMA 1994;272:1851-7. [19] Vincent C, Taylor-Adams S, Stanhope N. Framework for analysing risk and safety in clinical medicine. BMJ 1998;316:1154-7. [20] McNutt R, Abrams R, Hasler S. Why blame systems for unsafe care? Lancet 2004;363:913-4. [21] Pronovost PJ, Nolan T, Zeger S, et al. How can clinicians measure safety and quality in acute care? Lancet 2004;363:1061-7.

ICU safety incidents for medical versus surgical patients 183 [22] Holzmueller CG, Pronovost PJ, Dickman F, et al. Creating the webbased Intensive Care Unit Safety Reporting System. J Am Med Inform Assoc 2005;12:130-9. [23] Needham DM, Thompson DA, Holzmueller CG, et al. A system factors analysis of airway events from the Intensive Care Unit Safety Reporting System (ICUSRS). Crit Care Med 2004;32: 2227-33. [24] Needham DM, Sinopoli DJ, Thompson DA, et al. A system factors analysis of bline, tube, and drainq incidents in the intensive care unit. Crit Care Med 2005;33:1701-7. [25] Wu AWM, Holzmueller CGB, Lubomski LHP, et al. Development of the ICU Safety Reporting System. J Patient Saf 2005;1:23-32. [26] Vincent C. Understanding and responding to adverse events. N Engl J Med 2003;348:1051-6. [27] STATA statistical software: release 8.0. TX7 College Station; 2002. [28] Frey B, Kehrer B, Losa M, et al. Comprehensive critical incident monitoring in a neonatal-pediatric intensive care unit: experience with the system approach. Intensive Care Med 2000;26:69-74. [29] Metnitz PG, Reiter A, Jordan B, et al. More interventions do not necessarily improve outcome in critically ill patients. Intensive Care Med 2004;30:1586-93. [30] Classen DC, Pestotnik SL, Evans RS, et al. Adverse drug events in hospitalized patients. Excess length of stay, extra costs, and attributable mortality. JAMA 1997;277:301-6. [31] Buckley TA, Short TG, Rowbottom YM, et al. Critical incident reporting in the intensive care unit. Anaesthesia 1997;52:403-9. [32] Pronovost PJ, Wu AW, Sexton JB. Acute decompensation after removing a central line: practical approaches to increasing safety in the intensive care unit. Ann Intern Med 2004;140:1025-33. [33] Blendon RJ, DesRoches CM, Brodie M, et al. Views of practicing physicians and the public on medical errors. N Engl J Med 2002;347:1933-40. [34] Wu AW, Folkman S, Mcphee SJ, et al. Do house officers learn from their mistakes. JAMA 1991;265:2089-94. [35] Gallagher TH, Waterman AD, Ebers AG, et al. Patients and physicians attitudes regarding the disclosure of medical errors. JAMA 2000;289:1001-7. [36] Nap RE, Silva Alvaro M, Fidler V, et al. Collaborative practice and clinical outcomes in the ICU. Crit Care 2003;4:221. [37] Pollack MM, Patel KM, Ruttimann E. Pediatric critical care training programs have a positive effect on pediatric intensive care mortality. Crit Care Med 1997;25:1637-42. [38] Diringer MN, Edwards DF. Admission to a neurologic/neurosurgical intensive care unit is associated with reduced mortality rate after intracerebral hemorrhage. Crit Care Med 2001;29:635-40. [39] Stambouly JJ, McLaughlin LL, Mandel FS, et al. Complications of care in a pediatric intensive care unit: a prospective study. Intensive Care Med 1996;22:1098-104.