Research news from Agency for Healthcare Research and Quality.

Diagnostic errors that harm outpatients are typically the result of multiple individual and system breakdowns

A study of 307 closed malpractice claims shows that many missed or delayed diagnoses of outpatients lead to dire outcomes. In some cases, diagnosis of a serious condition like cancer was delayed more than a year. Over half (59 percent) of the claims studied involved diagnostic errors that harmed patients. Also, 59 percent of these errors were associated with serious harm, and 30 percent resulted in death. Cancer was the diagnosis involved in 59 percent of the errors, chiefly breast (24 percent) and colorectal (7 percent) cancer. The next most commonly missed diagnoses were infections, fracture, and heart attacks.

The most common breakdowns in the diagnostic process were failure to order an appropriate diagnostic test (55 percent), failure to create a proper followup plan (45 percent), failure to obtain an adequate history or perform an adequate physical exam (42 percent), and incorrect interpretation of diagnostic tests (37 percent) by physicians, radiologists, or pathologists. In some cases, clinicians failed to check on test results or to communicate them to patients, or they did not schedule a necessary followup appointment. In other cases, patients failed to keep an appointment to find out or follow up on abnormal test results.

Missed cancer diagnoses were more likely than other missed diagnoses to involve errors in the performance and interpretation of tests. Primary care physicians were centrally involved in most diagnostic errors. The findings reinforce the need for system interventions, such as clinical decision support systems that include alerts and reminders, to reduce these problems. The study was supported in part by the Agency for Healthcare Research and Quality (HS11886 and HS11285).

More details are in "Missed and delayed diagnoses in the ambulatory setting: A study of closed malpractice claims," by Tejal K. Gandhi, M.D., M.P.H., Allen Kachalia, M.D., J.D., Eric J. Thomas, M.D., M.P.H., and others, in the October 3, 2006, Annals of Internal Medicine 145, pp. 488-496.

Shifting from a culture of blame to a culture of safety in nursing homes could help identify and prevent medical errors

Many nursing homes continue to harbor a culture of blame. As long as staff members feel they will be blamed for medical errors they report, patient safety will remain at risk. Creating a culture of safety in nursing homes is complicated by limited resources, overwhelmed leadership, and an educationally diverse workforce. Nevertheless, by improving team communication and participation in decisions, nurse leaders can create an environment in which every team member can contribute to resident safety, asserts Jill Scott-Cawiezell, Ph.D., of the University of Missouri-Columbia. In a study supported by the Agency for Healthcare Research and Quality (HS14281), she and fellow researchers surveyed staff members of five diverse Midwestern nursing homes. The staffers were members of a team whose goal was to improve medication safety practices.

The survey asked nursing home staff how strongly they agreed or disagreed that key safety elements--communication, teamwork, and leadership--were present for medication safety practices to develop and thrive. The study authors also conducted a case study of a nursing home team trying to develop a culture of safety, which nevertheless was stuck in a culture of blame. In this particular case, staff members were assigned a point for every medication error they made. Staff members were disciplined after three points and could even be fired.

The team clearly understood the need to know about errors so that they could improve care, but no one was willing to get others in trouble. Also, the leader of the team often missed meetings due to crises or other problems. To alleviate the leadership void, the research nurse invited the nursing home administrator to the team's regular meetings. The team then became very open about safety issues in the nursing home. They also voiced their frustration with the continued blaming culture, which clearly led to underreporting of errors. The authors conclude that nursing homes must create a climate where everyone feels comfortable identifying and reporting safety concerns in order to shift to a safety culture.

See "Moving from a culture of blame to a culture of safety in the nursing home setting," by Dr. Scott-Cawiezell, Amy Vogelsmeier, M.S.N., Charlotte McKenney, B.S.N., and others, in the July 2006 Nursing Forum 41(3), pp. 133-140.

Not adjusting for pre-existing health problems may have exaggerated the number of deaths due to medical injury

In 2002, nearly 14 percent of adults hospitalized in Wisconsin suffered a care-associated medical injury which was believed to have increased their risk of dying by 48 percent compared with other patients. Researchers used the Wisconsin Medical Injuries Prevention Program (WMIPP) screening criteria to identify medical injuries (harm associated with a therapeutic or diagnostic healthcare intervention) among all 562,317 patients discharged from 134 acute care hospitals in Wisconsin in 2002. A total of 77,666 (14 percent) of discharges met WMIPP criteria for at least one medical injury. Overall, deaths occurred among 3.14 percent of those who suffered a medical injury and 2.13 percent of those who had no medical injury diagnosis upon discharge (a 48 percent difference).

However, after adjustment for patients' coexisting illnesses, severity of illness, and other factors (baseline mortality risk), the excess risk of dying associated with medical injury disappeared. The only types of medical injuries that were still associated with increased odds of dying were related to procedure-related complications or to a device, implant, or graft. Both medical injury and in-hospital mortality risks were higher among older patients who had multiple coexisting medical problems at admission.

These findings suggest that previous unadjusted risks of dying attributable to medical injury, which did not account for patients' baseline mortality risk, may have exaggerated the number of deaths due to medical injury. In this study, researchers adjusted for baseline risk of death using an index of coexisting disease, age, sex, diagnosis, hospital characteristics, and clustering within hospital. Injuries related to procedures or to a device, implant, or graft were still associated with increased adjusted odds mortality of 39 percent and 16 percent, respectively. The study was supported in part by the Agency for Healthcare Research and Quality (HS11893). See "Excess mortality caused by medical injury," by Linda N. Meurer, M.D., M.P.H., Hongyan Yang, M.S. Clare E. Guse, M.S., and others, in the September 2006 Annals of Family Medicine 4(5), pp. 410-416.

Eleven medications account for one-third of medication errors that harm hospitalized children

One-third of reported medication errors that harm hospitalized children involve 11 medications that have been in use for a considerable time. These errors are commonly due to wrong dosing and missed doses, according to a study using a national voluntary medication error reporting system, MEDMARX[R]. Researchers at the University of North Carolina Center for Education and Research on Therapeutics examined all pediatric medication error records submitted to the MEDMARX[R] program by subscribing hospitals and related health systems from January 1, 1999, to December 31, 2003.

They identified 816 harmful outcomes involving 242 medications during the 5-year period. About 4.2 percent of all pediatric medication errors were harmful and 11 medications from 3 drug classes were responsible for one-third of harmful medication errors. Opioid analgesics (morphine and fentanyl) were involved in 11.5 percent of errors, followed by antimicrobial agents (vancomycin, ceftriaxone, and gentamicin, 7.5 percent), antidiabetic agents (insulin, 4.5 percent), fluids and electrolytes (potassium chloride and total parenteral nutrition, 4.4 percent), bronchodilators (albuterol), inotropic agents (dopamine), and anticoagulants (heparin).

Over half of opioid analgesics and nearly one-fourth of antidiabetics in this study were given at the wrong dose. Dosage errors were often due to confusion between drug weight volumes and drug dosages, misprogramming of infusion pumps to deliver drugs per minute rather than per hour, and inappropriate recording of pounds instead of kilograms. Omission errors often involved a specific change in care or in the environment of the patient, such as transfer between units, between shift changes, or following a procedure. The study was supported by the Agency for Healthcare Research and Quality (HS10397).

See "Harmful medication errors in children: A 5-year analysis of data from the USP's MEDMARX[R] program," by Rodney W. Hicks, M.P.A., M.S.N., A.R.N.P., Shawn C. Becker, M.S.N., R.N., and Diane D. Cousins, R.Ph., in the August 2006 Journal of Pediatric Nursing 21(4), pp. 290-298.

Computerized and age-specific drug alerts can reduce both inappropriate prescribing of drugs and unnecessary drug alerts

Elderly patients are commonly prescribed medications that are potentially harmful to them, such as tertiary tricyclic amine antidepressants, long-acting benzodiazepines, or propoxyphene. To prevent this, computerized order entry (CPOE) systems include drug-specific alerts that warn clinicians about medications that are potentially inappropriate for older people each time the drug is ordered, regardless of the patient's age. Such alerts often annoy physicians when they are prescribing these drugs for younger patients. However, researchers recently modified a CPOE system so that age-specific drug alerts only occurred when clinicians prescribed target drugs to elderly patients. The system then suggested an alternative medication. This approach limited the number of unnecessary alerts faced by prescribers, while still maintaining the effectiveness of the drug-specific alerts.

The study was led by researchers at the HMO Research Network Center for Education and Research in Therapeutics at Harvard Pilgrim Health Care, which is supported by the Agency for Healthcare Research and Quality (HS11843). The research team randomly assigned seven practices to receive age-specific prescribing alerts plus an academic detailing intervention (interactive educational program on medications that can potentially harm the elderly). Eight practices received age-specific alerts alone.

Age-specific alerts resulted in continued effectiveness of the drug-specific alerts over a 1-year period. Group academic detailing did not enhance the effect of the alerts; however, the age-specific alerts led to fewer false-positive alerts for clinicians. During the drug-specific intervention (January to June 2002), each physician received an average of 18 alerts, 14 (82 percent) of which were false-positive (i.e., prescribed for non-elderly patients). During the age-specific intervention (January to June 2004), each physician received an average of four drug alerts, all of which were for elderly patients.

More details are in "Computerized prescribing alerts and group academic detailing to reduce the use of potentially inappropriate medications in older people," by Steven R. Simon, M.D., M.P.H., David H. Smith, R.Ph., Ph.D., Adrianne C. Feldstein, M.D., M.S., and others, in the June 2006 Journal of the American Geriatric Society 54, pp. 963-968.

Primary care patients with pain and psychosocial problems benefit from nurse telephone calls

Primary care patients who suffer from pain are often plagued by psychosocial problems, including emotional problems, substance abuse, and domestic violence. These patients can gain substantial and sustained reduced pain and improvement in psychosocial problems, as well as physical function, when a nurse educator talks to them by telephone about problem-solving strategies and basic pain management skills, concludes a new study. Operating from a central location, nurses were able to achieve positive outcomes with an average of three telephone calls to each patient. This approach could benefit many primary care patients, given that more than one-third of Americans aged 19 to 69 years report levels of pain similar to the study patients, and about 40 percent also report psychosocial problems, notes Tim A. Ahles, Ph.D., of Dartmouth-Hitchcock Medical Center.

Dr. Ahles and fellow researchers randomized patients from 14 rural primary care practices (47 physicians) to usual care or intervention groups. Overall, 644 patients reported pain and psychosocial problems and 693 patients reported pain problems without psychosocial problems. Patients in the intervention group received information tailored to their problems and concerns (INFO). Their physicians received feedback about their patients' problems and concerns (FEED) and a nurse-educator (NE) telephoned patients to teach problem-solving strategies and basic pain management skills.

At the 6-month assessment, more patients in the INFOFEED + NE group than in the usual care group achieved meaningful improvement in bodily pain (53 vs. 40 percent), physical role (41 vs. 26 percent), social functioning (55 vs. 37 percent), and vitality (46 vs. 28 percent). However, there was no advantage in emotional role (32 vs. 22 percent) or physical function (36 vs. 27 percent). At 12 months, clinically meaningful differences were sustained for several of these measures. However, patients who received INFOFEED alone without the nurse calls experienced minimal improvements over the usual care group, which were not sustained at the 12 month assessment. The study was supported in part by the Agency for Healthcare Research and Quality (HS10265).

See "A controlled trial of methods for managing pain in primary care patients with or without co-occurring psychosocial problems," by Dr. Ahles, John H. Wasson, M.D., Janette L. Seville, Ph.D., and others, in the July 2006 Annals of Family Medicine 4(4), pp. 341-350.

State children's health insurance and premium-subsidy programs do not always provide a bridge to private health insurance

When families' incomes increase and they are no longer eligible for State Children's Health Insurance Programs (SCHIPs) and premium-subsidy programs, many of them are still not able to afford private health insurance premiums without help. Over 85 percent of parents in a study of low-income Oregon families--whose children were disenrolled from the SCHIP Oregon Health Plan (OHP) or Oregon's premium-subsidy program, the Family Health Insurance Assistance Program (FHIAP)--said they would have kept their children in these programs, if possible.

One solution to ensuring children's health insurance coverage would be to raise the income eligibility ceiling for these programs, suggest the study authors. Since their survey was conducted, Oregon has implemented a modest increase in the income limit for both programs, from 170 to 185 percent of the Federal poverty level. In this study, half of children disenrolled from Oregon's SCHIP failed to requalify, because their families made too much money to meet the income eligibility requirements. Many of the remaining children did not reapply, because their parents thought they were no longer eligible. The reasons were similar for children leaving FHIAP.

These programs did not provide a bridge to nonsubsidized private health insurance for these children. Only one-third of OHP children and one-half of FHIAP children (whose parents were better educated and possibly had more access to job-related insurance) had insurance coverage after leaving these programs. Care access for these children was driven largely by health insurance coverage. Insured children were more likely to have a usual source of care and to have seen a physician when they needed one. The study was supported by the Agency for Healthcare Research and Quality (HS10463).

See "What happens to children who lose public health insurance coverage?" by Janet B. Mitchell, Susan G. Haber, and Sonja Hoover, in the October 2006 Medical Care Research and Review 63(5), pp. 623-635.