Developing a Veterans Health Administration (VHA) Serious Injury Surveillance System that Includes Adverse Event Hospitalizations

Campbell RR, Bradham DD, Sanchez-Anguiano A, et al.

Publication Details

Abstract

Objectives: The objective of this study was to examine the feasibility of applying the State and Territorial Injury Prevention Directors Association (STIPDA) consensus recommendations for using hospital discharge data in injury and adverse event surveillance to the Veterans Health Administration (VHA) population. The utility of developing an injury surveillance system that also included adverse events due to medical care was examined for its potential contributions to VHA patient safety programs and research. Methods: Selected variables from all VHA hospital inpatient discharges for 5 fiscal years (1998-2002) were extracted from the National Patient Care Dataset. The resultant dataset had more than 2.8 million records. The selected variables extracted included demographic and clinical information. Discharges for injuries and adverse events due to medications and medical complications were identified using the primary admitting diagnosis in accordance with STIPDA recommendations. The injuries and adverse events were grouped into categories using the Clinical Classification Software developed by the Agency for Healthcare Research and Quality (AHRQ). The medical care costs for these injury and adverse event hospital discharges were obtained from the VHA Decision Support System (DSS). Results: Over the study time frame, 153,153 injury and adverse event discharges occurred, with more than 1.8 million inpatient days, and $2.0 billion in direct medical care costs. In any given year, injury and adverse event discharges accounted for approximately 10 percent of total hospital medical costs and approximately 5 percent of the total discharges. Hospitalizations for adverse events associated with medical care, or medication adverse events, represented more than 50 percent of the hospitalizations. Conclusions: VHA administrative hospital discharge datasets can be used following STIPDA recommendations to monitor trends in the incidence and costs of veterans' hospitalizations for injuries as well as adverse events. The information gained from this enhanced injury surveillance system has the potential to positively affect current VHA patient safety and injury prevention initiatives.

Introduction

Patient safety has been defined by the Institute of Medicine (IOM) as “freedom from accidental injury.” 1 The IOM, in its 1999 report further noted, “[T]his definition recognizes that this is the primary safety goal from the patient's perspective.” 1 VHA former Undersecretary for Health, Kenneth W. Kizer, M.D., underscored the primacy of patient safety in health care: “The medical imperative is clear: to make health care safe, we need to redesign our systems to make errors difficult to commit and create a culture in which the existence of risk is acknowledged and injury prevention is recognized as everyone's responsibility.” 2 The VHA is the largest integrated public health care delivery system in the United States. It currently provides health care services to more than 6.8 million enrolled veterans. VHA facilities are located in all 50 states and Puerto Rico. The system currently includes 162 hospitals, hundreds of hospital-based outpatient clinics, long-term care facilities, and more than 850 community outpatient clinics.

Injury surveillance systems have been operational and producing public information on trends in injuries at the state and national level under the auspices of the Centers for Disease Control (CDC)'s National Center for Injury Prevention and Control for many years. There are no comparable comprehensive injury-surveillance systems operating and reporting publicly at the national level in the VHA system. Injury surveillance systems have traditionally not included data on adverse events due to medical care or medications. 3–7 Adverse event tracking and reporting have historically the province of risk management and, recently, of increased interest in patient safety initiatives. The patient safety movement is of relatively recent origin in the United States and the VHA's National Center for Patient Safety is a leader in this movement.

The veteran population is rapidly aging and is at risk for the same kinds of serious injuries requiring hospitalization that have been tracked with comprehensive injury-surveillance systems focused on the non-VHA population for many years. The VHA also recognizes that many injuries to patients occur as the result of medical care and is developing many comprehensive patient safety initiatives. The VHA emphasizes a culture of safety and focuses its patient safety initiatives and reporting on systems issues associated with adverse events. These reporting systems are confidential and are used internally to improve patient safety programs. There is no comparable publicly reported information on injuries and adverse events requiring hospitalizations for the VHA similar to the data readily available on the AHRQ Web site.

STIPDA is a major policy and coordinating organization for state and national injury surveillance systems in the United States. It recently published a consensus statement on the utility of using hospital discharge data for injury and adverse event surveillance. 3 The utility of applying the public health injury surveillance model to patient safety research has been reported elsewhere in the literature. 4 The goal of this research was to examine the utility of developing a national veteran injury-surveillance system based on the public health injury-surveillance approach using STIPDA case-finding recommendations. It would use all of the VHA hospital discharge data and, for the first time, track and trend patient injuries requiring hospitalizations due to all causes. This system would combine elements similar to the state's current injury surveillance systems and add patient-safety related data designed to monitor injuries resulting from medical care. 5–8 The data derived from this system would be examined for its potential contribution to injury prevention and patient safety programs in the VHA.

The surveillance system would be based on STIPDA's recent recommendations on the use of hospital discharge data for injury and adverse event surveillance. Medication-related hospitalizations and hospitalizations for complications due to medical care are unique new additions in this injury surveillance study. Past studies of patient safety—such as the Harvard medical practice studies in New York, Colorado, and Utah—have primarily focused on studying the delivery of care associated with hospitalizations. These studies were based on a sample of medical record reviews and attempted to identify errors and preventability, but did not extensively examine comprehensive discharge datasets for all admissions coded for adverse events due to medical care. 9–10 Recently researchers have used hospital discharge datasets to publish studies of injury hospitalizations or of hospitalizations for adverse events related to medical care. 11–16 Currently there is a great deal of interest in developing national patient-safety data systems. 16–20 This interest parallels national efforts in injury surveillance and patient safety. 1, 18, 19

The STIPDA report noted that hospital discharge data have several attributes that make them ideal candidates for injury surveillance research. 3 First, injury hospitalizations are an indication of the seriousness of the injury and are prime targets for injury prevention efforts. Secondly, they noted that hospital discharge datasets are now being collected in 42 states and represent one of the most widely available sources of standardized statewide health care data. 3 Population-based surveillance is the preferred level of analysis for injury and adverse event surveillance. 1, 3, 19, 21 Finally, because of the geographic information available in hospital discharge data, they provide an opportunity for prevention programs to be based at the county and city level. STIPDA also noted “hospital discharge data may be more useful than vital records for surveillance in less-populated areas for particular types of injuries that seldom result in death but do result in hospitalizations.” 3

Methods

Study design

The study was a national retrospective analysis of the number and types of hospitalizations for injuries and adverse events that occurred in the VHA system during fiscal years 1998-2002. Injuries and adverse events are defined by International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) diagnosis codes that have been aggregated into mutually exclusive categories using the AHRQ-developed Clinical Classification Software (CCS.) 22 The CCS categories that define injuries and adverse events are listed in Table 1.

Table 1. Trends in VHA injury and adverse event hospitalizations: by discharge frequency.

Table 1

Trends in VHA injury and adverse event hospitalizations: by discharge frequency.

In collaboration with other interested organizations, STIPDA published recommendations for the appropriate use of hospital discharge data for injury surveillance purposes. In particular, its recommendation that the primary admitting diagnosis be used to identify injury hospitalizations was adopted as the injury case-finding method for this study. 3 The VHA hospital discharge datasets have diagnosis coding that allow the primary admitting diagnosis to be readily identified. 23

Sources of data

Selected variables from all of the VHA hospital inpatient discharge datasets for 5 fiscal years (1998-2002) were extracted from the centralized VHA data repository at the Austin Automation Center (AAC, Medical Inpatient SAS datasets.) 23 This produced a discharge dataset with more than 2.8 million discharge records from the Patient Treatment File (PTF.) The selected variables extracted included demographic information on the patient; all ICD-9-CM coded clinical diagnoses associated with the inpatient stay; information on procedures and surgeries associated with the hospitalization; and discharge status. The direct medical care costs associated with these 5 fiscal years of discharges were obtained from the DSS data at the AAC. The AHRQ's single level Clinical Classification Software was used in the feasibility study to group the primary discharge diagnoses into clinical classes. The hospital discharges coded with the 20 CCS injury classes (225–244) were identified and incorporated into the database. 22 These CCS injury classes, comprising mutually exclusive groupings of ICD-9-CM codes in the 800–999 range, included various types fractures; sprains; wounds; complications due to procedures or use of medical devices; and medication and other poisoning-related discharges. 22

The clinical information included the primary admitting diagnosis, up to nine additional secondary diagnoses, information concerning the length of stay, and the Diagnosis-Related Group (DRG) assigned to the hospitalization. Demographic variables included age, gender, race, the percentage of service-connected disability for the veteran, and information concerning the Veterans Integrated Service Network (VISN) and the facility where the care was provided. The DSS cost data of VHA provided average cost data by CCS injury categories for the analyses. Enrollment and workload data were obtained from the VHA Planning System Support Group (PSSG) enrollment data and from VHA national and VISN health care utilization reports. 24

Analyses

The primary admitting diagnosis ICD-9-CM codes were aggregated into the 20 injury categories using the AHRQ's CCS injury coding taxonomy. The CCS is an ICD-9-CM grouper software program that was used to aggregate the diagnosis codes into mutually exclusive diagnosis categories. It was used because it is a widely available, standardized taxonomy used by the VHA and the civilian sector. Also, data based on CCS injury categories is readily available on the AHRQ Web site with national and state-specific data on trends and costs associated with the injury and adverse event categories. 25 The primary focus of the analysis was on identifying the number, relative proportions, time trends, and costs associated with injury and adverse event hospitalizations. Particular emphasis was placed on analyses of the adverse event hospitalizations to emphasize their importance in VHA patient safety-related programs and initiatives.

Descriptive analyses produced frequency distributions for hospitalizations by CCS injury and adverse event classes. Injury hospitalizations, medication-related hospitalizations (CCS categories 241 and 242), and adverse events (CCS categories 237 and 238) were analyzed. Five-year trends in the number of injury discharges were computed nationally and by VISNs over the 5-year time period. Injury hospitalization costs were computed using national average cost data from the VHA DSS. The national average inpatient discharge costs by fiscal year for each CCS category of injury were used to calculate the direct medical care costs associated with the hospitalizations. Costs were computed and trended for each CCS injury class nationally, and by VISNs, over the 5 years. The proportion of total injuries associated with each CCS category was computed. Analyses were conducted using Microsoft Access, Microsoft Excel, and Statistical Analytical System (SAS) software Version 8.2.

We received human subject protection permissions to conduct this study from the VHA Research and Development Committee and the associated University Institutional Review Board.

Results

Overall trends

There were 153,153 hospitalizations involving 99,550 unique patients with an injury or adverse event over the 5-year time period. Approximately 53 percent of the study admissions were coded as adverse events related to medical care, 39 percent were injury coded, and 8 percent were medication coded. By major CCS categories, there were 82,041 hospitalizations associated with adverse events related to medical care (CCS categories 237 and 238), 59,172 injury hospitalizations, and 11,940 medication-related hospitalizations (CCS categories 241 and 242). There were over 1.8 million bed days of care (BDOC) and over $2.0 billion in direct medical care costs associated with these hospitalizations.

CCS categories

The overall trends in specific CCS category discharges and total BDOC were fairly constant over the 5 years for the VHA system. The estimates of the inpatient total costs of direct medical care per year, however, increased from approximately $383 million to more than $450 million. Analyses conducted at the 22 VISN level found that there was considerable variation in the number, proportions, and direct costs of medical care in the individual CCS injury classes. National trends in the number of discharges associated with the CCS categories are listed in Table 1.

Inpatient mortality

There were 3,916 patients (3.9 percent of the 99,550 unique patients) who were coded for inpatient deaths associated with these hospitalizations over the 5-year study period. Approximately 2.6 percent of the total discharges (153,153 total discharges) over the 5 years were coded with inpatient deaths. For individual CCS categories, the highest percentages of the total discharges coded with an inpatient death over the 5-year period were associated with intracranial injuries (10.6 percent), hip fractures (6 percent), and spinal cord injuries (5.6 percent).

The adverse events associated with medical care (CCS 237 and 238) jointly accounted for 4.7 percent of the inpatient deaths overall for the 5-year time frame. However, when one examines the trends in the percentage of inpatient deaths associated with these two adverse event CCS categories by individual fiscal year (FY), the trends show that the percentage grew from approximately 45 percent in FY1998 (335 of 745) to nearly 60 percent in FY2002 (482 of 808). Nearly 50 percent (1934 of 3916) of the total inpatient deaths over 5 years in this study were associated with CCS categories 237 and 238. The trends in number of inpatient admissions by CCS categories coded for mortality are displayed in Table 2.

Table 2. Trends in inpatient mortality associated with injuries and adverse event hospitalizations.

Table 2

Trends in inpatient mortality associated with injuries and adverse event hospitalizations.

Complications of medical care

The CCS categories coded for complications associated with a device, implant, or graft (CCS 237) and complications of surgical procedures or medical care (CCS 238) were analyzed using the CCS multilevel categories to provide more information about the types of cases aggregated into these two categories. Complications associated with devices, implants, or grafts accounted for 53 percent (44,436 of 82,041) of the hospitalizations. Complications due to surgical procedures or medical care accounted for 47 percent (37,605 of 82,041) of the admissions. These 82,041 discharges were disaggregated and analyzed using the CCS multilevel software. There were 79,906 cases analyzed using the multilevel coding taxonomy. Nearly 25 percent of the admissions over the 5 years were coded for “malfunctions” of a device, implant, and graft. Discharges coded for “postoperative infections” and “other complications of surgical and medical procedures” constituted 18 percent and 15 percent of the admissions respectively. The results of this analysis are in Table 3.

Table 3. Frequencies of CCS multilevel codes for medical adverse events over time (CCS 237–238).

Table 3

Frequencies of CCS multilevel codes for medical adverse events over time (CCS 237–238).

Drug adverse event admissions

There were 11,940 admissions coded as poisonings due to psychotropic drugs or due to other medications and drugs during the study period. Approximately 27 percent of the total admissions associated with medication adverse events were associated with psychotropic drugs. Poisonings due to benzodiazepine tranquilizers consistently ranked as the primary ICD-9-CM coding for drug poisoning admissions over the five-year study period. Admissions with ICD-9-CM codes as due to the “adverse effects of medical or biological substances” and “poisonings due to antidepressants” were also problematic drug adverse events. The CCS classes associated with these psychotropic drugs and medications do not include illicit drugs such as heroin or cannabis.

Frequency of admission by unique patients

The 153,153 injury and adverse event coded admissions were associated with 99,550 unique patients. More than 90 percent of the unique patients who were admitted over the study period had one or two injury-coded admissions. Only a very few number of patients had more than 10 admissions over the 5-year time frame. Table 4 summarizes trends in adverse event discharges (CCS classes 237–238, 241–242), total direct costs of medical care, BDOC, average lengths of stay (ALOS), and the number of unique patients.

Table 4. Trends in adverse event hospitalizations: discharges, costs, BDOC, ALOS, unique patients.

Table 4

Trends in adverse event hospitalizations: discharges, costs, BDOC, ALOS, unique patients.

Discussion

The trends in the number of injury and adverse event coded hospitalizations in the VHA and their costs have remained fairly constant over the 5-year study period. This is similar to observed trends in these 20 CCS categories available online from the AHRQ Web site. In any given year, the injury and adverse event-related VHA discharges accounted for approximately 5 percent of the total discharges, but consumed nearly 10 percent of the acute hospital expenditures. Previous studies have reported similar cost profiles associated with injury and adverse event hospitalizations. 6–14 The opportunities for tracking, trending, and committing resources these injuries and adverse events through injury prevention and patient safety programs is evident from these kinds of injury surveillance analyses.

National initiatives in research on patient safety using hospital discharge data are currently underway, as previously noted. Some important observations must be made about the use of hospital discharge data for injury surveillance and adverse event reporting. First, one cannot use hospital discharge data alone for epidemiological studies of the incidence and prevalence of injury conditions. An injury hospitalization represents one possible tier on the “injury pyramid” as a setting of care. Measuring the incidence of injuries at this level will not provide a complete picture of the incidence, prevalence, mortality, morbidity, and costs associated with treating any particular category of injury across the many settings of care where injuries are treated. Second, hospital discharge data is primarily collected for billing purposes and not to conduct injury studies, which is why there is no specific information available on the timing, or the preventability, of the injury in discharge data. 3 One cannot generally tell with certainty from the diagnosis coding in discharge datasets whether the injury or adverse event was present on admission or developed while in the hospital. This is especially problematic unless one chooses to use the STIPDA criteria that focus on the admitting diagnosis. However, by using only the admitting diagnosis to identify cases included in the surveillance system, one runs the risk of underestimating the true number of injury cases. This is especially true in the case of adverse events that occurred during hospitalization and are coded in the secondary diagnosis fields. Diagnosis coding issues to identify injury cases of interest in medical discharge records are well-established issues in the injury surveillance field. Previous STIPDA consensus statements have focused on specification of the particular codes used to define important types of injuries. 8

This study chose to adopt the STIPDA case-finding definition to combine injuries and adverse events in one dataset. This approach was chosen because the VHA is a comprehensive, nationally managed care program with an enrolled population. There are clinical and financial incentives inherent in the VHA system that would support these kinds of comprehensive public health-oriented injury and adverse event studies. There are comprehensive electronic medical records and related health care datasets that allow for the detailed study of injuries and adverse events on a national basis.

The debate about what kinds of cases to study in patient safety research is well known. Whether one chooses to study all injuries coded as related to medical care, or only a subset of preventable injuries that are clearly related to medical error is a judgment call. 4, 26 If one chooses to study only admitting diagnosis for injuries and adverse events there will be an underreporting of cases. Many adverse events occur during the inpatient episode and may be coded in the secondary diagnosis fields, such as postoperative wound infections. Previous pilot studies by the authors have shown that by adopting the STIPDA case-finding definition and not including a search of the secondary diagnoses, one would identify approximately 50 percent of the injury and adverse event hospitalizations in any given year. By way of example, in FY2002 in the VHA, there were 30,681 hospital discharges with an admitting diagnosis for an injury or adverse event. If one included the secondary diagnosis fields, the total coded admissions would be 65,848. Similar results can be observed in the online AHRQ Healthcare Cost and Utilization Project (HCUP) HCUPNet data, where one can select the primary diagnosis or all listed diagnoses. 25 Future research studies will incorporate the secondary diagnoses to ensure a more comprehensive study of injuries and adverse events involving hospitalizations and will also include veteran injuries and adverse events that are paid for in non-VHA facilities by Medicare.

There are other limitations inherent in using hospital discharge data in injury surveillance systems that include adverse events. Important risk factors are not routinely entered in discharge datasets. Risk factor identification facilitates injury-prevention program development and should be an important component of injury surveillance systems. 21 Injury coding is incomplete in hospital discharge datasets. 21 Recent reports demonstrate that ICD-9-CM mechanism of injury coding (E-coding) is present in about 60 percent of the discharge records on a national level. 8, 15 Pilot studies conducted by the authors have found that approximately 50 percent of the VHA injury hospitalizations in the study dataset had an E-code. Increased national attention to reporting the ICD-9-CM mechanism of injury code and the place of occurrence codes is important for injury prevention and adverse event tracking initiatives. 1, 21

This analysis may have limitations in its generalizability because it involved a predominantly male veteran population. Approximately 5 percent of the injury hospitalizations in the VHA involved females. However, it provides a unique contribution to the injury surveillance and adverse event literature because no published national time trend data or analysis on veteran injury or hospitalizations for adverse events exists. The utility of adapting the public health injury surveillance model using the STIPDA consensus-driven case definitions for an “injury hospitalization” applied to VHA hospital discharge data has been demonstrated. The use of the AHRQ Clinical Classification Software to present the data in 20 injury and adverse event categories produced the first national study of its kind using VHA discharge datasets. Because the case-finding definitions have been “standardized” using the CCS categories, one can compare the VHA experience with the public and private sector using readily available data online and published studies from AHRQ.

The VHA has well-established patient safety initiatives and adverse event tracking and reporting systems under the guidance of its National Center for Patient Safety. 27 These programs include well-established and effective programs in surgery and medication safety. 28, 29 The current analysis of major types of drugs that lead to hospitalizations for poisonings in this study will augment the current medication safety programs being undertaken by the VA's new Center for Medication Safety. 29 The study results also provide an opportunity for comparisons of injury and adverse event hospitalizations with the non-VHA sector because of the availability of similar data on the HCUPNet Web site. 25 Analysis of the non-VHA hospital sector's data, available from the AHRQ Healthcare Utilization Project (HCUP), can provide information on the types of complications that occur in the use of medical devices in rehabilitation, and complications due to medical care. 29

This current analysis and injury surveillance dataset does not replace important VHA patient safety programs that are centered on inpatient reporting systems, rather it supplements those initiatives. It can produce data about the development of injuries and complications outside of the hospital setting. These injuries or adverse events can occur in other types of health care settings, at home, or elsewhere in the community. Additionally, an important component in any injury surveillance or patient safety program is information on baseline measurements to gauge the effects of intervention programs over time. This dataset provides additional information about injuries and adverse events that develop or occur outside of the inpatient setting and may be used to supplement the current patient safety reporting systems in the VHA.

Conclusions

The feasibility of using VHA hospital discharge data to identify and track trends in injuries that include iatrogenic injuries has been demonstrated. The merging of injury surveillance data with patient safety data may produce a more comprehensive picture of the incidence and costs associated with hospitalizations for serious injuries of all types. Hospital discharge data is widely available at the state level and in many national non-VHA datasets based on well-established sampling frames. 30, 31 Several countries base important parts of their national patient safety reporting systems on analyzing injury hospitalizations and patients' experiences (e.g., the UK, Australia, New Zealand, and Canada.) 32–35 The utility and feasibility of using the AHRQ-developed CCS to group ICD-9-CM codes into mutually exclusive injury and poisoning groupings provides a consistent framework to present and compare the data across health care systems. Additionally, the availability of national and state estimates using the HCUPNet online data system for injuries, and the new Inpatient Quality Indicators, allows for comparisons of patient safety-related measures to be made at the state, regional, and national levels. 25 The STIPDA recommendation to use the primary admitting diagnosis as the primary means to identify injury and adverse event-related admissions in hospital discharge data represents the culmination of many years of discussion and consensus-building on case finding definitions. It provides another perspective on the burden to society of hospitalizations for serious injuries and adverse events.

This project built upon the authors' earlier studies on VHA injury surveillance and emerging initiatives in patient safety reporting systems to produce this VHA study dataset and analyses. The opportunities to build upon the basic structure of the current injury surveillance dataset with links to data from other settings of care, pharmacy data, and cost data is an opportunity to further both injury surveillance and patient safety-related research.

One of the unique features of the VHA administrative datasets for health services research is the use of a patient identifier that allows many administrative datasets to be linked. The use of this “scrambled social security number” in the VHA datasets greatly facilitates the identification of true incidence of injuries by unique patients. 23 It also allows one to develop and track episodes of care associated with the treatment for an injury across inpatient, outpatient, and other health care settings. Future VHA research will expand upon current patient safety research initiatives and use the surveillance data to track the incidence of fall-related injuries and medication adverse events; develop cost and utilization models to track the treatment of injuries in episodes of care; and identify risk factors for injuries to aid in focused injury prevention and patient safety programs. The VHA is a leader in patient safety research and intervention programs that can provide important models and research that may be generalizable to other health care sectors.

Acknowledgments

The research reported/outlined here was supported by the Department of Veteran Affairs, Veterans Health Administration, Health Science Research Service. Robert R.Campbell, Ph.D., was the principal investigator in Tampa, FL. The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs.

References

1.
Kohn LT, Corrigan JM, Donaldson MS, editors. To err is human: building a safer health system. A report of the Committee on Quality of Healthcare in America. Institute of Medicine. Washington, DC: Academy Press; 2000.
2.
Leape LL, Woods DD, Hatlie MJ, et al. Promoting patient safety by preventing medical error. JAMA. 1998 Oct 28;280(16)1,444–7. [PubMed: 9801008]
3.
Injury Surveillance Workgroup. Consensus recommendations for using hospital discharge data for injury surveillance. Marietta, GA: State and Territorial Injury Prevention Directors Association; 2003.
4.
Layde PM, Cortes LM, Teret SP, et al. Patient safety efforts should focus on medical injuries. JAMA. 2002 Apr 17;287(15):1,993–7. [PubMed: 11960544]
5.
Centers for Disease Control and Prevention. Recommended framework for presenting injury mortality data. MMWR. 1977 Aug 29;46(RR14):1–30. [PubMed: 9301976]
6.
National Center for Health Statistics. Health, United States, 1996-97 and Injury Chartbook. Hyattsville, MD; 1997.
7.
Fingerhut LA, Warren M. Injury Chartbook. In: Health, United States, 1996-97. Hyattsville, MD; National center for Health Statistics, 1997.
8.
Thomas C, Butler J, Davies M, et al. State injury indicators report, 2nd edition—1999 data. Atlanta: Centers for Disease Control and Prevention, National Center for Injury Prevention and Control; 2004.
9.
Brennan T A, Leape L L, Laird N M. et al. Incidence of adverse events and negligence in hospitalized patients: results of the Harvard Medical Practice Study I. NEJM. 1991 Feb 7;324(6):370–6. [PubMed: 1987460]
10.
Thomas E J, Studdert D M, Burstin H R. et al. Incidence and types of adverse events and negligent care in Utah and Colorado. Medical Care. 2000 Mar;38(3):261–71. [PubMed: 10718351]
11.
Utah Health Data Committee. Adverse events related to medical care, Utah: 1995-99. Salt Lake City: Utah Department of Health; 2001.
12.
Romano P S, Geppert J J, Davies S. et al. A national profile of patient safety in U.S. hospitals. Health Affairs. 2003 Mar–Apr;22(2):154–66. [PubMed: 12674418]
13.
Florida Agency for Health Care Administration. Adverse drug effects. health outcomes series. Florida State Center for Health Statistics. Tallahassee, FL; 2000.
14.
Florida Agency for Health Care Administration. Septicemia Hospitalizations. Health Outcomes Series. Florida State Center for Health Statistics. Tallahassee, FL: 2000.
15.
Shinoda-Tagawa T, Clark D E. Trends in hospitalization after injury: older women are displacing young men. Injury Prevention. 2003;9:214–219. [PMC free article: PMC1730984] [PubMed: 12966008]
16.
Aspen P, Corrigan JM, Wolcott J, Erikson SM, editors. patient safety: achieving a new standard of care. Committee on Data Standards for Patient Safety, Institute of Medicine. National Academy Press: 2004.
17.
Agency for Healthcare Research and Quality. National healthcare quality report. Department of Health and Human Services. Rockville, MD: 2003. (Prepublication copy: http://www​.ahrq.gov/.)
18.
Leather S, McCarthy D. The quality of health care in the United States: a chartbook. Commonwealth Fund: New York, NY; 2002.
19.
Agency for Healthcare Research and Quality. Fact sheet: Patient Safety Reporting Systems and Research in Health and Human Services. http://www​.ahrq.gov/qual​/taskforce/hhsrepor.htm. (Last accessed 2004 Dec.)
20.
Williams S. Patient safety improvement using reporting systems. Grant Number U18 HS11885. Utah State Department of Health: 2001. Abstract available at http:www​.gold.ahrq.gov/.
21.
Institute of Medicine. Reducing the burden of injury: Advancing prevention and treatment. Washington, DC: National Academy Press; 1999.
22.
Clinical Classifications Software (ICD-9-CM) Summary and Download. Agency for Health Care Policy and Research. Rockville, MD:2003 July. http://www.hcup-us.ahrq.gov/toolssoftware/ ccs/ccs.jsp.
23.
VIReC Research User Guide: FY2002 VHA Medical SAS® Inpatient Datasets. Edward J. Hines, Jr. VA Hospital. Hines, IL: Veterans Affairs Information Resource Center; 2003 April. http://www​.virec.research​.med.va.gov/DataSourcesName​/Medical-SAS-Datasets​/SASDocumentation.htm.
24.
VA Health Care. Systemwide workload, FY 1997-2002. VHA Office of Policy and Planning. http://www.va.gov/vetdata/ ProgramStatics.
25.
HCUPnet. Healthcare cost and utilization project. Rockville, MD: Agency for Healthcare Research and Quality; 2003 Dec. http://www​.ahrq.gov/hcupnet/.
26.
McNutt RA, Abrams R, Arons DC. Patient Safety Committee. Patient safety efforts should focus on errors. JAMA. 2002 Apr 17;287(15):1,997–2,001. [PubMed: 11960545]
27.
VA National Center for Patient Safety. http://www​.patientsafety.gov/.
28.
National Surgery Quality Improvement Program (NSQIP). http://www.nsqip.org/ main/summary.asp.
29.
VHA Pharmacy Benefits Management (PBM). http://www.vapbm.org/PBM/ menu.asp.
30.
Fact sheet: databases and related resources from HCUP. AHRQ Publication No. 02-P030. Agency for Healthcare Research and Quality; Rockville, MD: 2002 Sept. http://www​.ahrq.gov/data/hcup/datahcup​.htm.
31.
Centers for Disease Control and Prevention. National Center for Health Statistics overview of datasets available. http://www​.cdc.gov/nchs/express.htm.
32.
Department of Health (UK). An organisation with a memory: report of an expert group on learning from adverse events in the NHS. London: 2000.
33.
Runciman WB, Moller J. Iatrogenic injury in Australia. Australian Patient Safety Foundation; 2001. http://www​.apsf.net.au/journals.html.
34.
Davis P, Lay-Yee R, Briant R. et al. Adverse events in New Zealand public hospitals I: occurrence and impact. N Z Med J. 2002 Dec 13;115(1167):U268. [PubMed: 12552260]
35.
Baker GR, Norton P. Patient safety and healthcare error in the Canadian healthcare system. http://www​.hc-sc.gc.ca​/english/care/report/index.html.