Excess Mortality Caused by Medical Injury

Linda N. Meurer; Hongyan Yang; Clare E. Guse; Carla Russo; Karen J. Brasel; Peter M. Layde

doi:10.1370/afm.553

Published eLetters

If you would like to comment on this article, click on Submit a Response to This article, below. We welcome your input.

Submit a Response to This Article

Compose eLetter

Title *

Author Information

Contributors
First Name and Middle Initial * First or given name, e.g. 'Peter'. Last Name * Your last, or family, name, e.g. 'MacMoody'. Email Address * Your email address, e.g. higgs-boson@gmail.com Occupation * Your role and/or occupation, e.g. 'Orthopedic Surgeon'. Affiliation * Your organization or institution (if applicable), e.g. 'Royal Free Hospital'.

Statement of Competing Interests

Competing interests? *

Yes

Please describe the competing interests

CAPTCHA

This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.

What code is in the image? *

Enter the characters shown in the image.

Vertical Tabs

Jump to comment:

PATIENT SAFETY IN THE US: A LEADERSHIP ROLE FOR FAMILY MEDICINE
George E. Fryer

Published on: 20 October 2006
What�s the Real Number? Excess Mortality and Medical Injury
David W Bates

Published on: 06 October 2006
Medical Injury Approach Complements Patient Safety Indicators
Linda N Meurer

Published on: 04 October 2006
Excess Mortality Caused by Medical Injury: Put Numbers in Perspective
Chunliu Zhan

Published on: 30 September 2006

Published on: (20 October 2006)
Page navigation anchor for PATIENT SAFETY IN THE US: A LEADERSHIP ROLE FOR FAMILY MEDICINE
PATIENT SAFETY IN THE US: A LEADERSHIP ROLE FOR FAMILY MEDICINE
George E. Fryer, New York, NY
PATIENT SAFETY IN THE US: A LEADERSHIP ROLE FOR FAMILY MEDICINE
The impact of estimates of mortality due to medical error based on the results of credible, large scale studies has been seismic. A related IOM report further galvanized policy and provider groups, and led to unprecedented focus on issues of patient safety and quality of care [1]. These revelations fueled growing frustration with the world’s most...
Show More

PATIENT SAFETY IN THE US: A LEADERSHIP ROLE FOR FAMILY MEDICINE
The impact of estimates of mortality due to medical error based on the results of credible, large scale studies has been seismic. A related IOM report further galvanized policy and provider groups, and led to unprecedented focus on issues of patient safety and quality of care [1]. These revelations fueled growing frustration with the world’s most expensive health care system, already noted for poor outcomes compared with those achieved in other developed countries.
The findings of Meurer, et al, [2] provide important new insights into the nature and effects of medical error on mortality. Their calculations of comparative risk for the various areas of care can serve as a guide to prioritize responsive patient safety activities. The estimate of death attributable to medical error, though less than previously found, was substantial, and documenting error in the cases of 1 of every 7 hospital admissions is a continuation of the disturbing commentary on our fractured, confusing US system.
Medical mistakes reported by Meurer, et al, occurred in hospitals, a setting characterized in part by specialized physicians performing complex procedures on seriously ill patients. The authors make a compelling case for not giving primary care a pass. But those same data indicate that many generalist office based providers, a little more than one-third of family physicians, now seldom if ever visit their own patients who have been hospitalized [3]. Does their responsibility end precipitously at the entrance to these facilities? After all, hospitalists help staff certain of these institutions, attending mostly to matters once the domain of the patient’s primary care physician.
At a time when family medicine is being redefined, adapted to a rapidly changing world, and field tested, does the phenomenon of medical error truly reflect on this specialty? Family physicians may not have been directly implicated in the commission of most hospital errors, but they do have a potentially vital role in this country’s movement to enhance patient safety. Increasing the active involvement and reassuring visible presence of a patient’s primary care physician during his/her hospital stay might be a meaningful step toward reducing medical errors. Certainly, the patient-provider relationship, even one of long-standing, can be eroded if patients feel abandoned by their PCP (‘my doctor’), at this most difficult time for them and for the members of their families.
This discussion, like most addressing medical error, has been focused on the hospital. But earlier research suggests that among hospitalized patients about half of all medical errors reported had occurred prior to their admission. Ambulatory care settings warrant serious scrutiny and evaluation as part of any comprehensive effort to reduce errors. The American Academy of Family Physicians (AAFP) and other organizations who shared sponsorship for the Future of Family Medicine initiative [4] have given the specialty a platform from which it can lead the effort to regain public confidence in the US health care system. There is a place for family physicians at the patient’s side in any health care setting. If there when needed, the members of this specialty have a unique opportunity to advance the concept of patient safety, to help prevent the occurrence of medical errors, and to reduce their effect.
REFERENCES:
1. Institute of Medicine. Committee on Quality of Health Care in America. To Err is Human: Building a Safer Health System. Washington, DC: National Academy Press; 2000.
2. Meurer LN, Yang H, Guse CE, Russo C, Brasel KJ, Layde PM. Excess mortality caused by medical injury. Ann Fam Med. 2006;4:410-416.
3. American Academy of Family Physicians. 2005 Facts about Family Medicine. Practice Profile Survey. May 2005. http://www.aafp.org.
4. Future of Family Medicine Project Leadership Committee. The future of family medicine: a collaborative project of the family medicine community. Ann Fam Med. 2004;2:S3-S32.
Competing interests: None declared
Show Less

Competing Interests: None declared.
Published on: (6 October 2006)
Page navigation anchor for What�s the Real Number? Excess Mortality and Medical Injury
What�s the Real Number? Excess Mortality and Medical Injury
David W Bates, Boston, MA
Other Contributors:
The number of deaths caused by iatrogenic injury in the U.S. has been highly controversial [1]. To estimate this figure, Meurer et al used claims data to determine how many deaths might be related to a medical injury diagnosis in Wisconsin, and then extrapolated their figures to the U.S. as a whole [2].
The fundamental flaw in this approach—which cannot be overcome without primary data collection—is that coding...
Show More

The number of deaths caused by iatrogenic injury in the U.S. has been highly controversial [1]. To estimate this figure, Meurer et al used claims data to determine how many deaths might be related to a medical injury diagnosis in Wisconsin, and then extrapolated their figures to the U.S. as a whole [2].
The fundamental flaw in this approach—which cannot be overcome without primary data collection—is that coding is so poor for many of the key diagnoses that it results in large underestimates of the proportion of deaths caused by iatrogenic events. In studies we have done involving primary data collection, almost none of the preventable deaths would have been detected by their approach (Bates personal commentary).
Another point emphasized by Meurer et al is that comorbidity may confound the relationship between having an adverse event and mortality. The issue, though, is that comorbidity and severity are probably effect modifiers as well as confounders, so simply adjusting for the confounder does not solve the problem, as the authors imply.
The net result is that Meurer et al’s estimate of the number of deaths is hardly the final word on this issue. A better estimate of the proportion of deaths would require a large-scale study with primary data collection, evaluating the care received by large numbers of patients in a representative population of patients who died in the U.S. This approach would have many methodological problems of its own, especially hindsight bias. Furthermore, errors that result in deaths are not always readily apparent (like delayed diagnosis) and many errors are missed. While doing such a study would provide useful insights, it would be very expensive and none is currently planned.
We believe that rather than waiting for such a study, it would be better to acknowledge that major problems with safety resulting in unnecessary deaths exist in healthcare today, and that organizations should systematically begin to implement changes such as those in the National Quality Forum’s safe practices list3 and the 100,000 Lives campaign [4,5]. Additional research should also be done to identify new safety strategies. At the same time, approaches like Meurer et al’s can be used to obtain rough estimates of performance, especially serially, but the absolute results must be interpreted with considerable humility. If we are to have confidence in the absolute numbers, coding would need to be greatly improved and standardized for many of the specific conditions involved.
References
(1) Hayward RA, Hofer TP. Estimating hospital deaths due to medical errors: preventability is in the eye of the reviewer. JAMA 2001; 286(4):415-420.
(2) Meurer LN, Yang H, Guse CE, Russo C, Brasel KJ, Layde PM. Excess mortality caused by medical injury. Annals of Family Medicine 2006; 4(5):410-416.
(3) National Quality Forum. Safe Practices for Better Healthcare. Washington, DC: National Quality Forum, 2003.
(4) Berwick DM, Calkins DR, McCannon CJ, Hackbarth AD. The 100 000 Lives Campaign. JAMA 2006; 295(3):324-327.
(5) IHI announces that hospitals participating in 100,000 lives campaign have saved an estimated 122,300 lives. 14 June 2006. Institute for Healthcare Improvement. 5 Oct 2006 <http://www.ihi.org/NR/rdonlyres/1C51BADE-0F7B-4932-A8C3- 0FEFB654D747/0/UPDATED100kLivesCampaignJune14milestonepressrelease.pdf>.
Competing interests: None declared
Show Less

Competing Interests: None declared.
Published on: (4 October 2006)
Page navigation anchor for Medical Injury Approach Complements Patient Safety Indicators
Medical Injury Approach Complements Patient Safety Indicators
Linda N Meurer, Milwaukee, WI, USA
Other Contributors:
We welcome the opportunity to respond to comments from Dr. Zhan at the Agency for Healthcare Research and Quality. Dr. Zhan describes a number of differences between the Wisconsin Medical Injury Prevention Program Medical Injury Screening Criteria and the patient safety indicators of the AHRQ and the Complications Screening Program. We do not believe, however, that these differences reflect a deficiency of either the WM...
Show More

We welcome the opportunity to respond to comments from Dr. Zhan at the Agency for Healthcare Research and Quality. Dr. Zhan describes a number of differences between the Wisconsin Medical Injury Prevention Program Medical Injury Screening Criteria and the patient safety indicators of the AHRQ and the Complications Screening Program. We do not believe, however, that these differences reflect a deficiency of either the WMIPP criteria or the CSP or AHRQ/PSI. Rather, they reflect different purposes and different theoretical underpinnings of the different criteria.
The CSP and AHRQ/PSI are intended to identify substandard care during a given hospital stay. Therefore they exclude patient safety problems that occurred prior to the index hospitalization, such as on an outpatient basis or in a previous hospitalization. They identify a specific set of patient safety problems determined likely to be due to inappropriate care, and therefore potentially preventable. This relatively narrow definition misses some of the adverse events identified through detailed chart review studies such as the Harvard Medical Practice Study. The PSI area indicators further capture events leading up to hospitalization, but only for 7 specific event types.1
The WMIPP criteria are not focused on identifying episodes of substandard care during a given hospitalization. Rather they are intended to give an overall estimate of the magnitude and pattern of medical injuries occurring in the hospital as well as those medical injuries occurring on an outpatient basis which require subsequent hospitalization. It is very difficult to reproducibly determine whether a medical injury occurring to a patient is due to negligence. For that reason, and because of the importance of identifying all risks associated with health care interventions the WMIPP includes all medical injuries without reference to negligence. In doing so, we also recognize that injuries that seem inevitable according to current standards may well be prevented as medicine advances. A surveillance tool such as the WMIPP criteria can help to identify sources of harm and potential targets for future investigation.
There are advantages to each of these approaches. The CSP and AHRQ/PSI use sophisticated methods to target important threats to patient safety, and can be a useful tool for hospitals attempting to improve quality of care, to guide patient safety interventions and monitor their impact. They do not, however, give a comprehensive view of all patient safety problems. They may underestimate events that occur on an outpatient basis and those that require a prolonged period to become manifest.
The WMIPP criteria provide a broader picture of the patterns of medical injury from a population perspective. They help identify risks of health care interventions which merit attention, even if they are not clearly due to negligence or inappropriate care. They can also help identify “up stream” sources of medical injuries that may be profitably addressed. Examples of such medical injuries that ultimately end up frequently being treated in hospitals include urinary tract infections incurring in nursing homes and bleeding disorders and other complications associated with coumadin use. In a recent demonstration project that we undertook using the WMIPP criteria, hospitals identified a substantial number of each of those types of medical injuries and participated in efforts to reduce their occurrence.2
As Dr. Zhan described, the WMIPP criteria do have limitations inherent in administrative datasets, and the injuries we identified represent a different subset of patient safety issues than those found by the PSI’s. For drug injuries, these appeared to over-represent low acuity events. More detailed “drilling down” is needed to make meaningful conclusions regarding specific injury categories.
In summary, WMIPP are not an effort to “reinvent the wheel” of CSP or AHRQ/PSI. We believe the major differences of the WMIPP criteria and the CSP and AHRQ/PSI reflect their different underpinnings and purposes and are actually a strength, not a weakness. The evolution of patient safety surveillance methods using routinely collected data is still in its infancy. The rapid development of electronic medical records systems holds potential for developing far more sophisticated screening tools than either the WMIPP criteria or the CSP or AHRQ/PSI, which are based on limited hospital discharge data. Ultimately the success of any patient safety surveillance tool will depend on how effective it is at helping to identify patient safety problems requiring intervention, and monitoring the impact of those interventions.
1. Patient Safety Indicators Overview. AHRQ Quality Indicators. February 2006. http://www.qualityindicators.ahrq.gov/psi_overview.htm; Accessed October 4, 2006. Agency for Healthcare Research and Quality, Rockville, MD.
2. Meurer JR, Meurer LN, Grube J, Brasel KJ, McLaughlin C, Hargarten SW, Layde PM for the Wisconsin Medical Injury Reporting System (WMIRS) Research Group: Coupling Performance Feedback with Evidence-based Educational Resources: A Model to Promote Medical Injury Prevention. In: Advances in Patient Safety: From Research to Implementation (Volume4- Programs, Tools and Products), Education and Training. AHRQ-DOD 2005.
Competing interests: None declared
Show Less

Competing Interests: None declared.
Published on: (30 September 2006)
Page navigation anchor for Excess Mortality Caused by Medical Injury: Put Numbers in Perspective
Excess Mortality Caused by Medical Injury: Put Numbers in Perspective
Chunliu Zhan, Rockville, Maryland, USA
The greatest accomplishment since the release of the Institute of Medicine report, “To Err Is Human”, (1) in 1999 is that now we all agree that errors and iatrogenic injuries in healthcare are real and should be reduced. However, many disagreements remain regarding definitions, taxonomy, measures, and ways to improve patient safety. Meurer et al’s article reflects both the achievement and the many challenges in the field...
Show More

The greatest accomplishment since the release of the Institute of Medicine report, “To Err Is Human”, (1) in 1999 is that now we all agree that errors and iatrogenic injuries in healthcare are real and should be reduced. However, many disagreements remain regarding definitions, taxonomy, measures, and ways to improve patient safety. Meurer et al’s article reflects both the achievement and the many challenges in the field of patient safety.
Meurer et al screened hospital discharge abstracts for 688 specific ICD-9-CM codes and flagged a discharge as having suffered “medical injury” if one or more of the targeted codes were present on the abstract. This method took advantage of the large volume of data collected for billing or administrative purpose. But the convenience is not without a price – administrative data have many flaws, and addressing and bypassing the flaws is a critical component in any research effort using this data source. (2) Many consensuses have emerged from previous efforts, especially from the development of the Computer Screening Program (CSP) by Iozzoni and colleagues (3) and, more recently, the Patient Safety Indicators (PSIs) by the Agency for Healthcare Research and Quality (AHRQ). (4) A brief review of these consensuses may put Meurer et al’s findings into proper context.
1. Concept and Definition. Many terms have been used to refer to “untoward harm” in health care, including adverse event, complication, iatrogenic injury, medical injury, patient harms, etc. These terms are wrapped under a more general term, patient safety, which also includes terms such as medical errors, near- misses, mishaps, and so on. A key underlining characteristic of these concepts is that the events (errors or harms or patient safety events) described by these terms are preventable when appropriate care is provided, and occurrence of such events is indicative of patient safety lapses. Both the CSP and the AHRQ PSIs attempted to identify preventable patient safety events. Meurer et al defined “medical injury” more broadly, under an “injury control model” that included injuries occurring “in the course of appropriate care”. Their choice may stem from the difficulty in separating preventable medical injuries from complications of care, but can be misleading.
2. ICD-9-CM codes identifying patient safety events. Meurer et al used substantially more codes than CSP and PSIs in flagging medical injuries. For example, they included ICD-9-CM codes for adverse effects of drugs, which are expected and may not be preventable for some drugs. These codes have little relevance to patient safety.
3. Principal versus secondary diagnosis. If a targeted code appears as the principal diagnosis in the discharge abstract, the flagged event is considered as the cause of an admission, indicating medical injuries or errors occurred outside of the admitting hospital. A selected code that appears on the secondary diagnosis, on the other hand, is indicative of safety lapse in the admitting hospital. It is critical to differentiate them. The AHRQ PSIs include a set of indicators to identify in-hospital events, as well as a set of indicators that capure both in-hospital events and events leading to admissions, which are called “area indicators” since the indicators count all the patient safety events appearing in the hospital service area, regardless of whether they were originated in the hospital. It is not clear whether Meurer’s study made such a distinction.
4. Complication versus comorbidity. If a targeted code appears in a secondary diagnosis, a further distinction needs to be made. Some secondary diagnoses, for example, bed sores, that present at admission are considered comorbid diseases. Bed sores could be developed during hospitalization, and are thus considered complications or adverse events because adequate care could have prevented them from developing. The current billing form, UB94, does not include data elements to identify which diagnoses are present at admission, presenting one of the biggest challenges in flagging patient safety events. Muerer et al’s study apparently could not bypass this data flaw. The UB04, the new uniform billing form to be launched soon, includes such data elements and will substantially increase the value of the claims data in patient safety research. Sometimes it is also possible to separate preventable medical injuries from comorbid conditions based on other information available from the claims data. For example, it is unlikely that a patient develops bed sores within 5 days of hospital stay. Blood stream infections may not be preventable if a patient enters the hospital with infections and immunosuppression. The CSP and the PSIs developed algorithms to exclude such events from being counted as patient safety events. By using all discharges as denominator to calculate rates, Meurer et al apparently did not apply such an exclusion method.
5. Individual types versus “medical injury” (summary of events). The excess mortality risk for having nausea is apparently different from having an allergic reaction to a drug. Lumping medical injuries of vastly different natures together to estimate excess mortality may be more concealing than telling.
6. Validity of the ICD-9-CM codes in identifying patient safety events. As Meurer et al discussed, there remain questions that (1) ICD-9-CM codes were coded correctly and (2) to what extent ICD-9-CM codes identify true events indicative of patient safety (or lack of).
7. Statistical significance versus clinical significance. The large volume of administrative data presents another perplexing problem: small but statistically significant differences. Such phenomenon may also complicate Meurer et al’s findings.
Meurer et al took an “injury control model” that is broader than the patient safety concept, and their approach to define “medical injuries” failed to take full advantage of data and latest development in this area. While recognizing the value of the new findings, we should be cautious in interpreting them. For example, it would be a disservice to all if we equal adverse drug events to Meurer et al’s “medical injuries” associated with drugs and biologics and conclude that such events do not lead to excess mortality risk.
Administrative data are gold mines, but digging them the wrong way could produce misleading information. AHRQ PSIs represent the best effort to target patient safety events for further investigation, and these indicators are constantly refined as we know more about the data and as the data sources and coding systems improve. Researchers should take advantage of such advancements, rather than re-inventing the wheel.
Reference: 1. Institute of Medicine. To Err is Human, Building a Safer Health System. Washington, DC: National Academies Press; 1999. 2. Zhan C, Miller M. Administrative data-based patient safety research: a critical review. Quality and Safety of Health Care. 2003;12(2 supl.): ii58 -ii63. 3. Iezzoni LI, Davis RB, Palmer RH, et al. Does the Complications Screening Program flag cases with process of care problems? Using explicit criteria to judge processes. Int J Qual Health Care. Apr 1999;11(2):107- 118. 4. Agency for Healthcare Research and Quality. Patient Safety Indicators. Available at http://www.qualityindicators.ahrq.gov/psi_download.htm. Accessed September 26, 2006. Rockville, MD: Agency for Healthcare Research and Quality; 2003.
Competing interests: None declared
Show Less

Competing Interests: None declared.