Killer Care: How Medical Error Became America's Third Largest Cause of Death, And What Can Be Done About It
(The following is excerpted from Killer Care: How Medical Error Became America's Third Largest Cause of Death, And What Can Be Done About It by James B. Lieber. Reprinted here with permission by the publisher OR Books © 2015.)
By the end of the twentieth century, the health-care industry was avid about getting its arms around the problem of error. The news about accidents among the rich and poor, famous and obscure was alarming the public and undermining its confidence in the profession. Malpractice claims were on the rise; so were monetary payouts. Insurers, hospitals, and physicians pushed back with “tort reform” designed to award and divert litigation. Since lawsuits often arose from treatment, “quality of care” initiatives grew as a method of preventing cases. All of these issues focused researchers on two problems. First, how should medical errors be defined? Second, how many of them were there? Clearly, they were killing and maiming people, causing lawsuits, and mounting costs. So size mattered critically.
In 1978, a pathologist and lawyer, Don Harper Mills at the University of Southern California School of Medicine, published an early analysis of the frequency of accidents caused by health-care in the Western Medical Journal. These were termed “potentially compensable events” or PCE’s (meaning that a jury could award malpractice money damages). Supported by the California Medical Association and the California Hospital Association, Mills and his team analyzed 20,684 patient charts from twenty-three California hospitals involving discharges in 1974. They found that 4.65 percent of the patients experienced PCE’s of varying severity. This was a fairly shocking rate of about one in twenty admissions. Of the PCE’s, 80 percent were temporary disabilities (lasting less than thirty days), 3.8 percent were lasting functional disabilities, and 9.7 percent were fatal. The rate of PCE’s markedly increased in patients age 65 and older.
In the 1980s, the Harvard Medical Practice Study—formed by representatives from Harvard’s Medical School, School of Public Health, Law School, Kennedy School of Government, Department of Biostatistics, and the Brigham and Women’s Hospital—assembled to address these problems. One of its leaders was Dr. Lucian Leape, a respected pediatric surgeon and medical educator who had been Martin Makary’s public health professor.
Among Leape and Makary’s principal interests were deficient and impaired physicians. Leape was famous for speaking at national conferences to hundreds of physicians and asking them to “raise your hand if you know of a physician you work with who should not be practicing because he or she is dangerous?” Every hand had gone up, and still goes up today when Makary asks the same question. However, three decades ago, Leape and the Harvard Practice Study were not looking to cast out incompetent physicians or buttress tort reform. They were simply trying to define and gauge the size of the problem of errors.
The study involved fifty-one non-psychiatric hospitals in New York State. From a total of 2,671,863 patients discharged in 1984, the team randomly picked 30,121 patient charts and analyzed them for “adverse events.” Reported in 1991 in the New England Journal of Medicine, the study defined an adverse event as “an injury that was caused by medical management [rather than the underlying disease] and that prolonged the hospitalization, produced a disability at the time of discharge, or both.”
The vetting of patient records was rigorous, labor-intensive, and occurred in two stages. The first round involved nurses who marked certain records “positive” for adverse events. Each suspect chart was reviewed by a pair of trained doctors, usually internists or surgeons, to assure that the adverse events occurred during or resulted from hospital treatment, and were not caused prior to admission. Plus, the physicians examined the records for negligence. An audit for quality control then was conducted of 1 percent of all charts by professional medical records analysts.
The findings were stark: an estimated 3.7 percent of patients had endured adverse events, and 1 percent experienced negligence from treatment below the standard of care. Most adverse events were the result of negligence. An example would be an unexpected bad reaction to a drug that was improperly prescribed. In order to reach reasonable conclusions, the study tracked records after discharge. For instance, it was able to find negligence in the case of a man who was admitted with rectal bleeding, received an examination, tested negative, and developed colon cancer twenty-two months later. Patients 65 and older had twice the rates of adverse events and negligence as patients under age 45.
The rate of negligence for adverse events was 28 percent, which varied depending on the location in the hospital. Fourteen percent of adverse events occurred in the operating room, 41 percent in the patient’s room, and 70 percent in the emergency room. The most frequent adverse cases in these hospitals were medication reactions (19 percent), followed by wound infections (14 percent), and surgical complications (13 percent). In many ways, the study was ominous. The 1 percent negligence rate from “sub-standard care” was chilling, but ironically it also was hopeful. Patient management errors accounted for 58 percent of adverse events—half of which were caused by negligence. That meant that reducing and preventing a huge block of harm did not require gains in medical research or knowl-edge, but merely implementing organizational improvements, resulting from what accident experts including Lucian Leape for called “human factors analysis.”
Scientifically rigorous and showcased as a two part series in the New England Journal of Medicine, the profession’s premier journal, and reported by the New York Times, the Harvard Practice Study received a brief flurry of attention, according to Leape, and then fell from sight, making a limited impact. Perhaps that was because it lacked statistical extrapolation to estimate the size of the national problem. Polled physicians during the period continued to see errors as not a catastrophic public-health crisis, believing that these mistakes killed less than 5,000 patients per year.
Regional and individual hospital studies persisted that showed the widespread nature of the error problem and also that medicine as an industry was an outlier. One of the last preserves of workplace individualism, it was slow to adopt team approaches, quality assurance, or to prioritize safety.
Although somewhat frustrated by the lack of attention gained by the Harvard study, Leape continued to drive the reform process. In 1997, he testified before the the U.S. House of Representatives Veterans Affairs Committee’s Subcommittee on Health: “Why is it,” he asked, “that when you enter a hospital your chances of dying from an accident are one in two hundred, but when you climb on an airplane, your chances of dying are one in two million?” Later, Leape organized and participated in a conference with aviation scientists and government officials, including Senator Bill Frist, on “Reducing Medical Error: Can You Be As Safe in a Hospital As You Are in a Jet?” With other medical policy heavyweights, Leape served on the Committee on the Quality of Health Care in America of the Institute of Medicine (IOM), a part of the National Academy of Sciences that was chartered by Congress in 1863 to advise the government on scientific matters. In 1970, it created the IOM, its third institute, following the National Research Council (1916) and the National Academy of Engineering (1964), in order “to secure the services of eminent members of appropriate professions in the examination of policy matters pertaining to the health of the public.”
In 1999, the IOM Committee on Quality of Health Care authored To Err is Human: Building a Safer Health System, because “errors are responsible for an immense burden of patient injury, suffering, and death.” The nearly 300 page report, which synthesized the relevant studies to date, and brought to bear the talents of hundreds of people from medicine, business, and government, is the most important policy document that the field has generated since the Carnegie Foundation in 1910 produced Medical Education in the United States under the leadership of Abraham Flexner. The Flexner Report, as it became known, produced much needed reforms and the standardization of medical training, and also resulted in the closure of many sketchy medical schools. To Err is Human already has had an equivalent impact, becoming nothing less than the magna carta of the medical errors/patient safety movement.
While now superseded in certain respects, To Err is Human credibly sized the problem in a way that took hold with the health care policymakers and the public. The IOM extended the data from New York (in the Harvard Medical Practice study) and a study from a recent large-scale analysis of Utah and Colorado to all U.S. hospital admissions in 1992, which totaled 33.6 million patients.
The Utah and Colorado survey produced adverse events in 2.9 percent of hospitalizations of which 6.6 percent were fatal. In New York the figures were 3.7 percent and 13.6 percent respectively. In both studies, medical errors that could have been prevented caused more than half of the deaths. The extrapolation to the wider society indicated that error fatalities ranged from 44,000 (at the Colorado and Utah rate) to 98,000 (at the New York rate). The IOM ranked medical errors as America’s eighth leading cause of death ahead of motor vehicle accidents (43,458), breast cancer (42,297), AIDS (16,516), and workplace accidents (about 6,000).
To Err is Human dealt with many other critical topics including why errors occur, reporting systems, establishing standards (best practices), and creating safety regimes. It pointed to Not for sale or successes in other industries, suggested ways to cope with accidents without creating malpractice claims, generally simplified the definition of adverse event (an injury resulting from a medical intervention), and somewhat complicated the definition of error: “failure of a planned action to be completed as intended or use of a wrong plan to achieve an aim.” It divided errors into active and latent. An active error “occurs at the level of the front-line operator and whose effects are felt almost immediately.” Latent errors exist “in the design, organization, training, or maintenance that lead to operator errors and whose effects typically lie dormant in the system for lengthy periods of time,” including such glitches as using similar vials of drugs for adults and children, inadequate numbers of nurses to prevent pressure ulcers, and insufficient or untrained housekeeping staff to keep down infections.
In To Err is Human, the Institute of Medicine vigorously maintained that errors should become a great public focus of health care, and that they should be mandatorily reported. The IOM took on specific problems like wrong site surgery and medication errors. It insisted upon a change in the culture of medicine. The era of the autonomous doctor hero had ended. No single human being should be expected to remember and act upon the hundreds if not thousands of pieces of information intrinsic to the history and diseases of the patient, as well as the numerous drug facts and compliance points for procedures necessary to treat an individual. Error was human; it would occur and cause harm with regularity unless the standardization and enforcement of best practices was rehearsed and applied by teams trained to prevent it or spot it early, eradicate it, or blunt it.
But it was those numbers—44,000 to 98,000 preventable medical error deaths per year—that stuck in the public mind. Death consecrates a problem in America and there was a lot of it, even though the figures did not include infections, patients who died from not actually receiving the treatments that had been recommended, and those killed by diagnostic errors.
Predictably there was pushback from the profession. In its “Controversies” section, the Journal of the American Medical Association carried criticism from a team at the University of Indiana headed by Clement J. McDonald, M.D., a geriatrician. They believed that the numbers, especially the 98,000 fatalities derived from New York hospitals in the 1980s, were exaggerated and alarmist. They particularly decried that the overall number of alleged hospital deaths more than doubled those from car crashes. Drivers “survive their ride if collisions are avoided,” but patients “have high disease burdens and high death risks even before they enter the hospital.” Thus, severely ill patients probably would have died soon anyway. The IOM was making unfair “headline claims.” Plus, “[t]he message in the IOM report is hot and shrill. It shouts about death and disability in U.S. hospitals: Preventable adverse events are a cause of death.”
Lucian Leape vehemently responded in the JAMA with an article entitled “Institute of Medicine Medical Error Figures Are Not Exaggerated.” The IOM report, he noted, “has galvanized a national movement to inspire patient safety. It’s about time.” The notion of “excessive mortality” due to the fact that some people maybe died soon anyway struck him, as it would have struck their families, as cold, but also wrong. High-severity patients numbered less than thirty in the study. Leape pointed to “screening criteria” that kept most of those likely to die out of the study, including patients who had do-not-resuscitate directives, were terminally ill, had had acute myocardial infarctions, pneumonia, or had undergone certain types of major surgery. Many ICU patients also did not meet the criteria. After Leape flogged the “excessive mortality” charge as wholly wanting, he argued that the critics missed the main point of the study:
Knowing that some of the patients “would have died anyway” is important for physicians because it lessens the burden of guilt. Physicians feel responsible deaths due to errors, which is appropriate and key to physicians’ professionalism. But we also feel shame and guilt, which is inappropriate and misguided, since errors are rarely due to carelessness. Failure to understand this is to miss the main message of the IOM report, which is not the mortality figures, but the admonition encapsulated in its title, To Err is Human.
The transforming insight for medicine from human factors research is that errors are rarely due to personal failings, inadequacies, and carelessness. Rather, they result from defects in the design and conditions of medical work that lead careful, competent, caring physicians and nurses to make mistakes that are no different from the simple mistakes people make every day, but which can have devastating consequences for patients. Errors result from faulty systems not from faulty people, so it is the systems that must be fixed. Errors are excusable; ignoring them is not.
Actually, deaths in the study almost certainly were underestimated. The screeners had relied on retrospective chart reviews. But chart quality depends on accuracy and honesty by doctors, an undefined portion of whom do not record their errors, perhaps to avoid peer scrutiny and malpractice claims. Plus these were all patients in hospitals, not individuals subjected to errors in doctors’ offices, clinics, or stand-alone surgical centers. So, the person whose melanoma was missed in the office or whose colonoscopy was misinterpreted in the clinic was not included. The IOM had not focused on misdiagnoses in or out of the hospital (where most of them occurred). Leape also raised the pro foundly troubling matter that: “some errors are not even known to clinicians caring for the patient. Autopsy studies for example, have found misdiagnoses in 20 percent to 40 percent of cases. On balance, the reliance on information extracted from records most likely led to a substantial underestimate of the prevalence of injury.”
In 2004, Healthgrades, a for-profit hospital rating agency, published a patient safety study based on computer models that appeared in JAMA and applied them to the Medicare data from all 50 states and the District of Columbia, during a three-year record (from 2000–2002). Medicare patients, exclusive of obstetrical cases, made up about 45 percent of admissions. The study showed 195,000 deaths per year from error, or as Healthgrades Vice President Samantha Collier, M.D., put it: “the equivalent of 390 jumbo jets full of people are dying each year due to likely preventable, in hospital medical errors, making this one of the leading killers in the U.S.”
During the brief study period, patient safety incidents cost the Medicare system an additional $8.54 billion. Three issues—failure to rescue, decubitus ulcers (pressure ulcers or bedsores), and sepsis following surgery—produced about 60 percent of the errors involving Medicare patients (one of four of whom would die). The 195,000 annual deaths would have placed error sixth on the Centers for Disease Control and Prevention list of killers, ahead of diabetes, pneumonia, Alzheimer’s, and renal disease.
If the old were at high risk, so were the young. Like the Quaid twins, they were particularly vulnerable to medication errors. In 2007, a study in the Journal of Pediatrics showed that 11 percent of children in hospitals experienced medication errors. Often there is no standard dose for infants and small children. Amounts must be diluted and calibrated by patient size and weight. Especially errors that were potentially harmful. Not for sale or with newborns whose livers, kidneys, and immune systems still are developing, toxicity is more of a danger than for adults. Plus, children are in far greater danger even when not hospitalized. In 2006, the American Academy of Pediatrics reported that 26 percent of children treated as outpatients experienced medication errors that were potentially harmful.
The surging popularity of electronic medical records in the twenty-first century produced new error detection methods unavailable to the Harvard Practice Study or the IOM when they analyzed New York and Colorado medical records. Mainly of these were “trigger” tools. Triggers in charts noted interventions, shifts, or changes in a patient’s care signaling a response to a possible error. For example, giving Naloxone (Narcan), an opiate receptor blocker to a patient like Josie King likely revealed an overdose of morphine or methadone.
Most trigger systems such as ones used by the federal Agency for Health Care Quality and some states like Utah and Missouri focused on discharge summaries. A “Global Trigger Tool” was developed by the Institute of Healthcare Improvement in Cambridge, Massachusetts from the work of David Classen, M.D., a professor of medicine at the University of Utah on the staff of LDS Hospital in Salt Lake City. Classen also served on the Patient Safety and Healthcare Information Technology Committee of the IOM.
The tool was global and “longitudinal,” in the sense that it drew from all charts and records, included far more triggers, and was followed by physician and nurse auditing as in the IOM studies. Predictably in head to head comparisons with federal and state trigger systems, the global tool proved more “sensitive,” meaning that it picked up additional errors and more “specific” in the sense that it caught greater numbers of false positives.
In addition to using Narcan, some of the global triggers included any “code or arrest” (meaning the heart or lungs stopped), a sudden drop in hematocrit (centrifuged red blood cells) of greater than 25 percent (anemia suggesting kidney failure), a fall, a “bounce back” readmission within thirty days, a transfer to a higher level of care, a return to surgery, an intubation following anesthesia care, a new pneumonia, a sudden stop of a medication, the use of Vitamin K, a coagulant following anti-clotting medication, third- and fourth-degree lacerations (severe tearing during childbirth), and a return to the emergency room within forty-eight hours.
The global trigger tool comprised nine general fields to search for adverse events. These included anesthesia, cardiac/myocardial infarction, device failures, falls, medication, pressure injuries (bedsores), procedures, pulmonary embolism/deep vein thrombosis (clots), and surgery. These basically followed the New York/Colorado frameworks. Again, diagnostic lapses were not included.
By congressional mandate, the Office of Inspector General of the Department of Health and Human Services applied global triggers to a study of hospitalized Medicare patients in 2008. The results were stark. Unexpected adverse events impacted about one in seven patients, and caused 15,000 deaths per month. About 44 percent of the events were likely or clearly preventable. The most common flaws involved medication, patient care (such as giving too much fluid intravenously as with Andy Warhol), and infections.
The additional costs to Medicare from errors amounted to $324 million per month, mostly from lengthened hospital stays. AHRQ Director Carolyn Clancy found that adverse events were occurring at an “alarming rate” and vowed to improve it.
A decade after the IOM report, the medical community was eager to see if the heightened consciousness about patient safety and quality had reduced errors. Two landmark studies suggested tragically that it had not. One, published in the New England Journal of Medicine in November 2010, looked at records from 2002–2007 of North Carolina hospitals with a high commitment to error prevention. Applying IHI Global Trigger Tool software, it found no significant reduction, which was sobering, especially because the IOM had called for a 50 percent reduction over five years. The authors led by Christopher Landrigan, M.D., of Brigham and Women’s Hospital and Harvard Medical School were frustrated, but still hopeful on a limited basis: “Although the absence of large-scale improvement is a cause for concern, it is not evidence that current efforts to improve safety are futile. On the contrary, data have shown that focused efforts to reduce discrete harms, such as nosocomial [hospital-based] infections and surgical complications can significantly improve patient safety.”
Analyzing over 2,300 admissions, the North Carolina study found that one out of four patients suffered harm. Over 63 percent of the injuries were preventable. The authors used a non-theoretical, practical definition of medical error: harm that was preventable. In almost 2.5 percent of cases—up from 1 percent in the 1999 IOM study—these preventable faults contributed to death. Over 3 percent of patients were permanently damaged, and 8 percent endured a life-threatening injury like hemorrhaging on the operating table. Were there more errors in safety-conscious hospitals since To Err is Human in 1999? Probably not, since the fatal numbers in the IOM studies preceded global trigger tools and by almost all accounts now seemed deflated.
The following year, researchers shook the profession with an article in Health Affairs entitled “‘Global Trigger Tool’ Shows that Adverse Events in Hospitals May be Ten Times Greater than Previously Measured.” Dr. David Classen, who did the seminal research for global triggers, served as lead author of the study, which looked at three mid-size to large (ranging from 550 to 1,000 beds) teaching hospitals associated with medical schools in the West and Northwest that participated on the condition of anonymity. The study defined harm as “unintended physical injury resulting from or contributed to by medical care that requires additional monitoring, treatment or hospitalization, or that results in death.” As radical as it was complex, the definition did not separate out harms based on mistakes or errors. As the Classen team wrote: “Because of prior work with Trigger Tools and the belief that ultimately all adverse events may be preventable, we did not attempt to evaluate the preventability or ameliorability (whether harm could have been reduced) of these adverse events.”
Each of these participating anonymous hospitals had a program for patient safety, error detection, and reporting. Each was a “tertiary” care facility, meaning that it received referrals of complicated cases (which in themselves can be more likely to lead to adverse events) from other providers. Harms were broken down into seven types: medication-related, procedure-related (excluding infection), nosocomial (hospital) infection, pulmonary VTE (venous thromboembolism), pressure ulcers, device failure, patient falls, and other. As in most studies, medication errors were most frequent (here, 45.7 percent) and diagnostic errors were not specifically considered. Deaths amounted to 4 percent of adverse events. Harm occurred in 33 percent of cases.
When different detection methods were applied, global triggers found over 90 percent of events, the government’s Patient Safety Indicators (based on discharge summaries) found 8.5 percent, and voluntary reporting disclosed only 2 percent (afraid of censure and malpractice, doctors and nurses seldom willingly self-accuse). Classen, et al. warned: “reliance on voluntary reporting and the Patient Safety Indicators could produce misleading conclusions about the current safety of care in the U.S. health-care system and misdirect efforts to improve patient safety.”