Ottawa Hospital Patient Safety Study: incidence and timing of adverse events in patients admitted to a Canadian teaching hospital ================================================================================================================================= * Alan J. Forster * Tim R. Asmis * Heather D. Clark * Ghiath Al Saied * Catherine C. Code * Sharon C. Caughey * Kevin Baker * James Watters * Jim Worthington * Carl van Walraven * © 2004 Canadian Medical Association or its licensors ## Abstract **Background:** Adverse events are poor patient outcomes that are due to medical care. Studies of hospital patients have demonstrated that adverse events are common, but few data describe the timing of them in relation to hospital admission. We evaluated characteristics of adverse events affecting patients admitted to a Canadian teaching hospital, paying particular attention to timing. **Methods:** We randomly selected 502 adults admitted to the Ottawa Hospital for acute care of nonpsychiatric illnesses over a 1-year period. Charts were reviewed in 2 stages. If an adverse event was judged to have occurred, a physician determined whether it occurred before or during the index hospitalization. The reviewer also rated the preventability, severity and type of each adverse event. **Results:** Of the 64 patients with an adverse event (incidence 12.7%, 95% confidence interval [CI] 10.1%–16.0%), 24 had a preventable event (4.8%, 95% CI 3.2%–7.0%), and 3 (0.6%, 95% CI 0.2%–1.7%) died because of an adverse event. Most adverse events were due to drug treatment, operative complications or nosocomial infections. Of the 64 patients, 39 (61%, 95% CI 49%–72%) experienced the adverse event before the index hospitalization. **Interpretation:** Adverse events were common in this study. However, only one-third were deemed avoidable, and most occurred before the hospitalization. Interventions to improve safety must address ambulatory care as well as hospital-based care. Patient safety has become an area of interest in Canada and in many other countries. This is partly because international research has demonstrated that adverse events, or poor outcomes caused by medical care, are common. Two US studies identified adverse events in 2.5%1 to 3.7%2 of hospitalizations, while British and Australian studies found rates of 10.8%3and 16.6%,4 respectively. One aspect of such studies that is often overlooked is the timing of the adverse event with respect to the hospitalization.5 Except for the British study,3 all previous studies included adverse events that occurred before the hospitalization under review as long as the adverse event was discovered during the hospitalization.1,2,4 Such “prehospital” events constituted between 38%2 and 49%4 of all adverse events in these studies. Yet, although the overall incidence of adverse events has been widely publicized, most of the attention on improving safety has focused on hospital care. We performed this study to determine the incidence, preventability, severity, type and timing of adverse events affecting patients in a Canadian teaching hospital. In addition, we evaluated differences in adverse events occurring before and during hospitalization. This information would confirm previous observations that many adverse events occur before admission. Importantly, we felt that a more thorough evaluation of the timing and location of adverse events could lead to more rational interventions to improve quality and safety. ## Methods The study took place in the Ottawa Hospital, a multicampus teaching hospital, and was approved by the Ottawa Hospital Research Ethics Board. We chose a random sample of 502 adult, nonpsychiatric hospitalizations from 2 acute-care facilities over a 1-year period. The sample constituted 1.3% of all hospitalizations and was stratified by facility and admitting service. As in previous studies, we used a 2-stage chart-review process (Appendix 1) to identify adverse events.1,2,3,4 In brief, in stage 1 a nurse reviewed charts to identify hospitalizations meeting at least 1 of 16 screening criteria. The criteria were identical to those used in previous studies and were selected to identify admissions in which an adverse event could have occurred. The nurse-reviewer had received a 1-hour training session on patient-safety issues, background literature and chart-review methods. In addition, the nurse and the principal investigator reviewed a training sample of 50 charts independently and discussed discrepancies in coding. In stage 2, the charts rated “screen-positive” by the nurse-reviewer were reviewed by 1 of 8 physicians (surgeons, internists, obstetricians and an emergency physician) to determine whether an adverse event had occurred. The physicians had received a 2-hour training session with the same framework as the nurse's training session. In addition, the physicians and the principal investigator reviewed a training sample of 5 charts independently and discussed discrepancies in coding. The physicians used a 6-point scale to rate their confidence that an adverse outcome was due to medical care. A score of 4 or above indicated that the outcome was an adverse event. All hospitalizations that were rated positive for an adverse event and a sample of negative-rated charts were reviewed again by another physician. Adverse events were rated for preventability, severity, type, timing and location. A preventable adverse event was one that, on the basis of implicit judgement, was felt to be due to an error in management. Severe adverse events led to permanent disability or death. The type was classified as adverse drug event, operative complication, nosocomial infection, diagnostic error (an indicated test was not ordered or a significant test result was misinterpreted) or system problem (inadequate communication, inadequate training or supervision of doctors, or inadequate functioning of hospital services). Types of adverse events were not mutually exclusive; for example, a patient with a postoperative wound infection was classified as having both a nosocomial infection and an operative injury. As in previous studies,1,2,4 we included adverse events that occurred before or during the index hospitalization as long as the event was first discovered during that hospitalization. We did not include adverse events that developed after the index hospitalization.1,2 We determined whether “prehospital” adverse events occurred during ambulatory care, during an emergency department visit or during a previous hospitalization. “Ambulatory care” consisted of treatment in physician offices, in the patient's home or in a nursing home. If an event occurred in the emergency department and the patient was directly admitted, then the event location was coded as “in-hospital.” However, if the patient was sent home from the emergency department and an adverse event required subsequent admission to hospital, the adverse event was classified as prehospital. The unit of analysis was the hospitalization. We calculated event rates per 100 hospitalizations. We used the Wilson Score method to calculate 95% confidence intervals (CIs). We analyzed 3 simple variables as potential risk factors for the occurrence of adverse events (patient age [in quartiles], admission status [“elective” if the chart contained a completed “request-for-admission” form] and admitting service [medicine, surgery or obstetrics and gynecology]). The χ2 test was used to compare event risk between strata and the χ2 test for trend to compare event risk between ranked strata. We determined independent predictors of in-hospital adverse events using a multivariate logistic regression model. Variables were entered into the model if they were significant in univariate testing at *p* < 0.2. ## Results We reviewed all the charts for the 502 randomly selected hospitalizations. At least 1 of the screening criteria was present in 312 (62.2%) of the charts, and adverse events (Appendix 2) were identified in 64 charts (incidence 12.7%, 95% CI 10.1%–16.0%). Of the 502 patients, 24 had a preventable adverse event (4.8%, 95% CI 3.2%–7.0%), 3 (0.6%, 95% CI 0.2%– 1.7%) died because of an adverse event, 9 (1.8%, 95% CI 0.4%–3.3%) had permanent disabilities, and 52 (10.4%, 95% CI 8.0%–13.3%) had temporary disabilities; none of the deaths was considered preventable, whereas 5 of the permanent disabilities were. Adverse events were most commonly adverse drug events (50%), surgical complications (31%) or nosocomial infections (19%) (Table 1). View this table: [Table1](http://www.cmaj.ca/content/170/8/1235/T1) Table 1. Only 25 of the patients experienced their adverse events after admission to hospital (5.0%, 95% CI 3.1%–6.9%). The remaining 39 (7.8%, 95% CI 5.7%–10.4%) experienced them before the index hospitalization. Similar proportions of prehospital and in-hospital adverse events were preventable (32% and 40%, respectively; *p* = 0.5) or severe (18% and 20%, respectively; *p* = 0.8). Except for operative events, the types of prehospital adverse events were similar to the types of in-hospital adverse events (Table 1). Of the 39 prehospital adverse events, 20 (31% of all adverse events) occurred during ambulatory care, 16 (25% of all adverse events) during a previous hospitalization and 3 (5% of all adverse events) during a previous emergency department visit. Of the 20 prehospital adverse events that occurred during ambulatory care, 18 (90%) were adverse drug events, and 9 (45%) were preventable. Of the 16 adverse events that occurred during previous hospitalizations, 10 (62%) were operative complications, 3 (19%) were adverse drug events, 3 (19%) were nosocomial infections, and 4 (25%) were preventable. All 3 adverse events that occurred during previous emergency department visits were due to diagnostic errors and were preventable. The difference in the proportion of adverse events that were judged to be preventable in each location was significant (*p* < 0.05). Table 2 presents the incidence of different types of adverse events by patient age group, admitting service and admission status. Overall risk of adverse events was significantly associated with patient age (*p* < 0.001 by the χ2 test for trend), admitting service (*p* = 0.001) and admission status (*p* = 0.04). When only in-hospital adverse events were considered, however, the risk remained significantly associated with age (*p* = 0.004) but was no longer associated with admission status and was less strongly associated with admitting service (*p* = 0.03). After controlling for patient age with a multivariate logistic model, we found that admitting service was not significantly associated with risk of in-hospital adverse events, although the total number of adverse events was small. Risk of adverse events was the same in both facilities. View this table: [Table2](http://www.cmaj.ca/content/170/8/1235/T2) Table 2. ## Interpretation We found that 12.7% of hospitalizations in a Canadian teaching hospital were associated with an adverse event and that 38% of all of the adverse events were preventable. We also determined that 61% of the events occurred before the patient was admitted to the facility. The types and preventability of prehospital adverse events were similar to those of in-hospital events. However, prehospital events that occurred in the ambulatory setting were almost always adverse drug events and were often preventable, whereas those related to previous hospitalizations were more often surgical complications and less often preventable. Although we used similar methods, comparisons between our study and previous ones need to be made cautiously owing to subtle differences in definitions and reviewer behaviour and international differences in charting practices.5,6 However, prehospital adverse events were less common in the US studies than in our study and the Australian one. This could be due to differences in the availability of primary health care: since greater proportions of the population in Canada and Australia are eligible for primary health care, it is possible that greater proportions receive medical treatment in the community. Alternatively, treatments may be monitored and followed more closely in the United States, which would make it less likely that complications would lead to hospitalization. The higher proportion of prehospital adverse events in Canada could also be due to the fact that the study hospital is a referral centre. Our study has both strengths and limitations. We used standard methods, evaluated more than 1 institution and enrolled appropriate specialists to conduct the chart reviews. However, the 2 facilities that we evaluated were teaching hospitals and in the same city. Current methods for detecting adverse events are hampered by being retrospective and based on chart review. This bias probably results in a conservative rate estimate. Determinations of adverse events are based on implicit criteria and are only moderately reliable.7 We tried to address this limitation by presenting all of our adverse events, so that readers could make their own judgements about validity. In summary, adverse events are relatively common. Most are the consequence of therapies that are provided correctly but have inherent risks. However, many adverse events are potentially preventable. Therefore, increased efforts must be made to reduce their incidence. The higher rate of prehospital adverse events needs to be confirmed. A larger, multicentre Canadian study is underway and may help shed light on this intriguing finding. Regardless, it is clear that quality-improvement efforts must address ambulatory care as well hospital care. ## Appendix 1 View this table: [Table3](http://www.cmaj.ca/content/170/8/1235/T3) Appendix 1. ## Appendix 2 View this table: [Table4](http://www.cmaj.ca/content/170/8/1235/T4) Appendix 2. ## Appendix 2 View this table: [Table5](http://www.cmaj.ca/content/170/8/1235/T5) Appendix 2. Continued. ## Footnotes * This article has been peer reviewed. *Contributors:* Alan Forster conceived and designed the study and, with Carl van Walraven, interpreted the data. Alan Forster analyzed the data and drafted the article. All of the authors acquired data, revised the draft critically for important intellectual content and approved the version submitted for publication. *Acknowledgements:* This research was supported by a research grant from the Ottawa Hospital and the Ottawa Internists Research Group. Carl van Walraven is an Ontario Ministry of Health Career Scientist. *Competing interests:* None declared. ***Correspondence to:**** Dr. Alan J. Forster, Ottawa Hospital, Civic Campus, C406–1053 Carling Ave., Ottawa ON K1Y 4E9; fax 613 761-5351;* aforster{at}ohri.ca ## References 1. 1. Thomas EJ, Studdert DM, Burstin HR, Orav EJ, Zeena T, Williams EJ, et al. Incidence and types of adverse events and negligent care in Utah and Colorado. Med Care 2000;38:261-71. [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.1097/00005650-200003000-00003&link_type=DOI) [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=10718351&link_type=MED&atom=%2Fcmaj%2F170%2F8%2F1235.atom) [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000085644100003&link_type=ISI) 2. 2. Brennan TA, Leape LL, Laird NM, Hebert L, Localio AR, Lawthers AG, et al. Incidence of adverse events and negligence in hospitalized patients. Results of the Harvard Medical Practice Study I. N Engl J Med 1991;324:370-6. [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.1056/NEJM199102073240604&link_type=DOI) [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=1987460&link_type=MED&atom=%2Fcmaj%2F170%2F8%2F1235.atom) [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=A1991EW29600004&link_type=ISI) 3. 3. Vincent C, Neale G, Woloshynowych M. Adverse events in British hospitals: preliminary retrospective record review. BMJ 2001;322:517-9. [Abstract/FREE Full Text](http://www.cmaj.ca/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiYm1qIjtzOjU6InJlc2lkIjtzOjEyOiIzMjIvNzI4NS81MTciO3M6NDoiYXRvbSI7czoyMToiL2NtYWovMTcwLzgvMTIzNS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 4. 4. Wilson RM, Runciman WB, Gibberd RW, Harrison BT, Newby L, Hamilton JD. The Quality in Australian Health Care Study. Med J Aust 1995;163: 458-71. [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=7476634&link_type=MED&atom=%2Fcmaj%2F170%2F8%2F1235.atom) [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=A1995TM31000004&link_type=ISI) 5. 5. Thomas EJ, Studdert DM, Runciman WB, Webb RK, Sexton EJ, Wilson RM, et al. A comparison of iatrogenic injury studies in Australia and the USA. I: Context, methods, casemix, population, patient and hospital characteristics. Int J Qual Health Care 2000;12:371-8. [Abstract/FREE Full Text](http://www.cmaj.ca/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NjoiaW50cWhjIjtzOjU6InJlc2lkIjtzOjg6IjEyLzUvMzcxIjtzOjQ6ImF0b20iO3M6MjE6Ii9jbWFqLzE3MC84LzEyMzUuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 6. 6. Runciman WB, Webb RK, Helps SC, Thomas EJ, Sexton EJ, Studdert DM et al. A comparison of iatrogenic injury studies in Australia and the USA. II: Reviewer behaviour and quality of care. Int J Qual Health Care 2000;12:379-88. [Abstract/FREE Full Text](http://www.cmaj.ca/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NjoiaW50cWhjIjtzOjU6InJlc2lkIjtzOjg6IjEyLzUvMzc5IjtzOjQ6ImF0b20iO3M6MjE6Ii9jbWFqLzE3MC84LzEyMzUuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 7. 7. Thomas EJ, Lipsitz SR, Studdert DM, Brennan TA. The reliability of medical record review for estimating adverse event rates. Ann Intern Med 2002; 136: 812-6. [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=12044129&link_type=MED&atom=%2Fcmaj%2F170%2F8%2F1235.atom) [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000175964500004&link_type=ISI)