Skip Navigation U.S. Department of Health and Human Services
Agency for Healthcare Research Quality
Archive print banner

Feature Story

This information is for reference purposes only. It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work. Persons with disabilities having difficulty accessing this information should contact us at: Let us know the nature of the problem, the Web address of what you want, and your contact information.

Please go to for current information.

Nearly 9 percent of lab tests commonly performed for hospitalized patients are redundant

Many clinical laboratory tests for hospitalized patients appear redundant, concludes a study supported in part by the Agency for Health Care Policy and Research (HS08297). It found that at one hospital, 28 percent of 12 standard laboratory tests were repeated earlier than the recommended test-specific intervals expected to show a change in status.

For two tests (white blood cell count and chest x-ray), nearly all the initial results in the sample were abnormal, and all repeats were considered justified. For the remaining 10 tests, medical chart review revealed no clinical indication for 92 percent of repeat tests following normal initial results. Overall, 9 percent of these 10 tests appeared redundant, that is, they were repeated too early to show any clinical change in the patient.

Eliminating these redundant tests would have saved $930,000 in charges at the study hospital. Three types of tests—arterial blood gases, routine cultures, and digoxin levels—accounted for 89 percent of the projected savings. Most redundant tests might be eliminated by using computerized reminder systems, suggests David W. Bates, M.D., M.Sc., of Brigham and Women's Hospital in Boston. Dr. Bates and his colleagues retrospectively studied the charts of more than 6,000 adults discharged from a large teaching hospital during a 3-month period in 1991 to examine the proportion of commonly performed diagnostic tests that were redundant and their associated charges.

Results showed that 88 percent of these patients had at least one of 12 target tests performed. These ranged from urine, sputum, and stool cultures and serum drug levels of digoxin and theophylline to white blood cell counts, portable chest x-rays, and arterial blood gases. Tests that were most likely to be repeated were digoxin levels (62 percent) and chest x-ray (58 percent); those least likely to be repeated were some serum drug levels (gentamicin 2 percent, theophylline 4 percent) and chemistry panels (3 percent). Of early repeat tests, a mean of 42 percent followed an initial normal result.

See "What proportion of common diagnostic tests appear redundant?" by Dr. Bates, Deborah L. Boyle, Eve Rittenberg, M.A., and others, in the April 1998 American Journal of Medicine 104, pp. 361-368.

Return to Contents


Early tube feeding improves outcomes, shortens hospital stays, and lowers costs for patients undergoing bowel resection

Patients who receive total enteral nutrition (TEN)—which means they are fed via a tube inserted into the small intestine—within 12 hours after bowel resection leave the hospital 2 days earlier, have fewer infections, and incur far less overall treatment costs than similar patients treated with total parenteral nutrition (TPN)—that is, intravenous fluids and diet. Feedings into the small intestine, which regains its motility and function within 2 to 12 hours after surgery, can begin much earlier than has been conventionally practiced. Yet early feeding is widely underused due to the common clinical practice of waiting for signs of bowel function, which don't occur for 72 to 96 hours after surgery, according to Ann-Marie Hedberg, Dr.P.H., of St. Luke's Episcopal Hospital in Houston, TX.

In a study supported in part by the Agency for Health Care Policy and Research (HS08440), a feeding tube was inserted into the small intestine of 66 treatment patients at the time of surgery; feeding began at 10 cc/hour within 12 hours after bowel resection and progressed if patients were hemodynamically stable. Outcomes for these study patients were compared with the outcomes of 159 control patients who received usual care at the same Texas teaching hospital. Treatment patients had 2.7 fewer postoperative days and 2.2 fewer total hospital days, as well as 4 fewer days supported by TPN (1.8 vs. 5.8) compared with control patients.

Also, no patients in the study group developed bacterial infections, compared with 7.5 percent of patients in the control group. TPN increased by nearly 10 percent the likelihood of infection; patients in the control group, who spent more days on TPN, were nearly three times more likely to have a nosocomial (hospital-acquired) infection than patients in the treatment group. Finally, average total cost savings, including the cost for related hospital readmissions, was $2,598 per treatment patient. The researchers conclude that TPN should be reserved for bowel resection patients who are not able to tolerate postoperative enteral feedings.

For more details, see "Effectiveness of an early postoperative enteral feeding protocol," by Dr. Hedberg, David R. Lairson, Ph.D., Lu Ann Aday, Ph.D., and others, in the March 1998 Journal of Clinical Outcomes Management 5(2), pp. 21-28.

Better control of postoperative pain may reduce delirium following surgery

Up to 60 percent of patients suffer from delirium following surgery. This delirium has been associated with increased mortality, more frequent complications, and prolonged hospital stays. Better control of postoperative pain may reduce delirium and its complications after surgery, according to a study supported in part by the Agency for Health Care Policy and Research (HS06573).

The study found that after controlling for known preoperative risk factors for delirium—for example, age, alcohol abuse, and type of surgery—higher pain scores at rest were associated with a 20 percent greater risk of postoperative delirium over the first 3 postoperative days. The method of postoperative analgesia, type of opioid used, and cumulative opioid dose were not associated with an increased risk of delirium.

The peak incidence of delirium was on postoperative day 2. Increased pain at rest was the only type of pain associated with an increased risk of developing delirium. Although pain with movement or maximal pain may represent more of an acute physiologic stress, patients experience pain at rest for more hours of the day. Therefore, pain at rest is more likely to affect their sleep-wake cycle and hormonal levels.

Delirium is often accompanied by sleep-wake cycle disruption, explain the researchers, who are from Brigham and Women's Hospital and Beth Israel Deaconess Medical Center. They used a visual analog scale to assess pain at rest, pain with movement, and maximal pain over the previous 24 hours when they interviewed patients older than 50 years admitted to a hospital for major elective noncardiac surgery during 1992 and 1993. They assessed postoperative delirium among these patients by using the confusion assessment method during the interviews and/or by using data from the medical record and the hospital's nursing intensity index.

See "The impact of postoperative pain on the development of postoperative delirium," by Eileen P. Lynch, M.D., Marissa A. Lazor, M.D., Janice E. Gellis, M.D., and others, in Anesthesia and Analgesia 86, pp. 781-785, 1998.

Researchers examine variations in use of intensive care units

Different hospital characteristics and geographic location continue to drive variations in use of intensive care units (ICUs) as well as use of ICU technology. This is exemplified in a study of variations in ICU use for patients with human immunodeficiency virus (HIV)-related Pneumocystis carinii pneumonia (PCP), which was supported in part by the Agency for Health Care Policy and Research (HS06494). A second study by AHCPR's Center for Organization and Delivery Studies demonstrates that hospital factors rather than payer differences affect use of ICUs in two States. The last two studies, supported in part by AHCPR (National Research Service Award fellowship HS00106), advise clinicians on how to find and critically appraise prognostic studies on specific medical conditions, as well as quality of care studies that compare outcomes among ICUs.

Curtis, J.R., Bennett, C.L., Horner, R.D., and others (1998). "Variations in intensive care unit utilization for patients with human immunodeficiency virus-related Pneumocystis carinii pneumonia: Importance of hospital characteristics and geographic location." Critical Care Medicine 26(4), pp. 668-675.

PCP remains the most common cause of respiratory failure and the most common reason for ICU admission in patients with HIV infection. Whether patients with PCP are admitted to intensive care and receive mechanical ventilation depends not only on patient characteristics but also on hospital characteristics and geographic location, concludes this study. It demonstrates that hospitals vary in their threshold for admitting patients with HIV-related PCP to the ICU and for forgoing ICU care for patients who die. Among 2,174 patients with PCP, 18 percent received care in an ICU. White patients, patients with Medicaid insurance, and patients with a prior AIDS-defining illness were the least likely to receive care in an ICU.

On the other hand, patients in county- or State-owned hospitals and patients in hospitals with more PCP experience also were less likely to be cared for in an ICU. These differences in ICU use persisted even after researchers controlled for patients' severity of illness and other characteristics. Significant geographic variation in ICU use also persisted after controlling for both patient and hospital characteristics.

The authors conclude that hospital and geographic variations in ICU use may make it difficult to generalize ICU outcomes across different hospitals. Their findings are based on a retrospective review of the medical records of 2,174 patients with HIV-related PCP in a stratified random sample of patients hospitalized for this problem from 1987 to 1990 at 96 hospitals in five cities.

Friedman, B.S., Steiner, C., and Scott, J. (1998). "Rationing of an expensive technology in the U.S.: Hospital intensive care units in two states, 1992." In D. Chinitz, and J. Cohen (Editors), Governments and Health Systems. New York: John Wiley & Sons, pp. 483-496.

Some studies have challenged the effectiveness of ICU care, a day of which is estimated to cost three to five times as much as a day of care on a regular hospital unit. Given the great cost of high-technology ICU care, it is not surprising that many payers give physicians and hospitals incentives to provide less ICU care than patients and their doctors would prefer. However, this study shows that differences in insurance coverage had little impact on ICU admission. On the other hand, hospital-specific factors were very important. High expected ICU occupancy, low ICU bed capacity, and high costs each implied lower than even odds of ICU admission.

In this book chapter, Bernard S. Friedman, Ph.D., and Claudia Steiner, M.D., of AHCPR, and Jane Scott, Ph.D., of the University of Maryland School of Medicine, present an analysis of hospital differences based on 1992 hospital discharge data for Massachusetts and Florida drawn from AHCPR's Hospital Cost and Utilization Project-3 database. The analysis revealed that, after controlling for patients' clinical conditions and severity of illness, variation in ICU use for nonemergency and nonsurgical admissions across hospitals was more important than variation across patients' insurance coverage.

Admission rates for adult patients to the ICU were about 80 percent higher in Florida than in Massachusetts. Teaching hospitals were less likely to admit patients to an ICU, but they provided somewhat more service to those admitted. Investor-owned and government hospitals were prone to make greater use of ICUs. HMO/managed care patients had somewhat lower use of services once admitted to the ICU, but hospital characteristics seemed to outweigh the effects of insurance coverage.

Randolph, A.G., Guyatt, G.H., and Richardson, W.S. (1998). "Prognosis in the intensive care unit: Finding accurate and useful estimates for counseling patients." Critical Care Medicine 26(4), pp. 767-772.

How can an ICU doctor have confidence in the prognosis he or she gives to parents of a 17-year-old head trauma victim who has been in a vegetative state for 3 months? One obvious approach is to analyze research studies that evaluate the prognosis of similar patients. But to make this worthwhile, doctors must know how to find and evaluate such articles, note the authors of this study. They advise clinicians how to identify and evaluate articles that contain valid prognostic information that will be useful in counseling ICU patients and their families.

The authors point out three questions to ask in article evaluation. One, are the results valid; that is, was there a representative sample of patients, were the patients homogeneous with respect to prognostic risk, was followup sufficiently complete, and were objective and unbiased outcome criteria used? Two, what are the results; that is, how likely are the outcomes over time, and how precise are the estimates of probability? And three, are the results useful for counseling patients and families? Were the study patients and their management similar to the doctor's own patient, and was the followup sufficiently long?

Randolph, A.G., Guyatt, G.H., and Carlet, J. (1998). "Understanding articles comparing outcomes among intensive care units to rate quality of care." Critical Care Medicine 26(4), pp. 773-781.

ICU patient prognosis is examined in a different way in this study. Clearly ICU patients who are much sicker and therefore have a poorer prognosis are more likely to have poor outcomes than other ICU patients. Therefore, patient prognosis must be adjusted for severity of illness in order to properly gauge better-than- and worse-than-expected outcomes among ICUs. To the extent that the spectrum of patients in ICUs differ in their prognosis, and these differences remain unaccounted for, attributing differences in outcomes to variable quality of care will be Variations in use of intensive care units misleading, contend these researchers.

The researchers note that valid comparisons of the outcomes among ICUs are made when the outcome measures used are accurate and comprehensive as well as clinically important; the ICUs being compared serve similar patients; the sampling of patients is sufficient and unbiased; appropriate risk adjustment is undertaken by applying a valid model to reliably collected data; and the comparisons focus on care delivered in the ICU. Before changes in ICU policy are made based on these outcome differences, it is important to clarify which factors might have resulted in extreme outcomes and whether these results are applicable in the ICU population that will see the impact of the changes.

Return to Contents

Managed Care

Researchers identify factors that encourage health plan choice among Medicaid managed care enrollees

States moving to mandatory managed care in their Medicaid programs have tried to present enrollees with more than one health plan option. The States that are most successful in getting enrollees to choose a plan are those that combine the Medicaid eligibility application and plan selection process, devise a comprehensive educational strategy, and give recipients personalized attention, according to a study by researchers at the Agency for Health Care Policy and Research.

Presenting Medicaid enrollees with a choice has the potential to increase their satisfaction and continuity of enrollment and provide them with the means to identify the plans and providers best equipped to meet their particular needs, explain Irene Fraser, Ph.D., Director of AHCPR's Center for Organization and Delivery Studies. Dr. Fraser and her colleagues Cindy Brach, M.P.P., also of AHCPR, and Elizabeth Chait, M.Ed., of Brookline, MA, conducted structured interviews in 1995 and 1997 with policymakers and operational staff in seven States that had State- or county-wide mandatory managed care Medicaid programs: Arizona, Maryland, Massachusetts, Michigan, Minnesota, Oregon, and Wisconsin.

Results showed that the proportion of enrollees selecting their own health plans varied considerably from State to State, ranging from 47 to 99 percent in 1994 and from 60 to 100 percent in 1996 to 1997. The same two States—Oregon and Minnesota—exhibited much higher levels of choice in both time periods. The percent of new enrollees in Oregon's managed care selection plan increased from 70 percent in 1994 to 100 percent in 1997, after procedures were instituted that required plan selection as part of the application. Selection rates in Minnesota counties have ranged from 87 to almost 100 percent. In both States, educational sessions and plan selection opportunities are closely tied to the application sites. This contrasts with most States, where application for Medicaid coverage occurs first and information on how to make a choice among health plans is provided only after eligibility is determined. Both States also had a broad-based strategy, which included educational materials comparing plans and one-on-one counseling. Efforts such as these obviously are critical in improving the quality of choice, note the authors.

The researchers also identified three methods of assigning recipients who did not choose plans: random assignment systems, assigning enrollees based on current needs and previous source of care, and assignments that provide incentives to contractors. Design decisions related to State contracting processes also can affect choice and continuity of care, the authors found. For example, the combination of exclusive contracting processes and frequent recompetitions can overrule prior choices and force recipients to switch plans often.

In conclusion, the authors point out that States face conflicting goals, wanting to keep costs low while rewarding quality and improving outcomes-they are trying to be creative in balancing these competing priorities, and they are moving away from using just financially driven assignment protocols.

See "Choice in managed Medicaid programs: State variation in enrollment strategies, choice levels, and assignment methods," by Dr. Fraser, Ms. Chait, and Ms. Brach, in the September/October 1998 Health Affairs 17(5), pp. 165-173.

Return to Contents

Primary Care

Clinical factors rather than concern about lawsuits usually cause physicians to delay childhood immunizations

Families whose children are injured by vaccines can use the National Vaccine Injury Compensation Program (NVICP) for compensation or they can pursue a malpractice case against a physician in a civil suit. Less than 30 percent of family physicians and pediatricians believe that the NVICP (and similar State programs) protect them against vaccine-related lawsuits, and only 12 to 13 percent believe that providing vaccines to children significantly increases their chances of a malpractice suit. Nevertheless, it is not fear of lawsuits, but questionable clinical factors that prompt most of these physicians to inappropriately delay vaccinations, concludes a study supported by the Agency for Health Care Policy and Research (HS07286).

Researchers from the University of North Carolina at Chapel Hill analyzed responses to a survey of 1,165 pediatricians and 1,849 family physicians on their perceptions of the legal and financial risks of providing childhood immunizations. The survey showed that only two-thirds of family physicians would provide a diphtheria, tetanus, and pertussis (DTP) vaccine to a child with a family history of seizures, and only slightly more than half would give the same vaccine to a child who experienced a fever following a previous dose of the vaccine.

Pediatricians were more likely to believe that the DTP vaccine could be administered safely to children with a family history of seizures, minor respiratory tract illnesses, or a previous local reaction to a DTP injection. With few exceptions, liability issues were not significantly associated with vaccine practices. These data suggest that educational efforts focused on liability issues alone will have little effect on inappropriate delaying of pediatric immunizations.

For more information, see "Vaccine-associated liability risk and provider immunization practices," by Gary L. Freed, M.D., M.P.H., Teresa Kauf, Ph.D., Victoria A. Freeman, R.N., Dr.P.H., and others, in the March 1998 Archives of Pediatric and Adolescent Medicine 152, pp. 285-289.

CQI teams and academic detailing are generally ineffective in increasing compliance with guidelines

During the past two decades, much effort has been invested in the development and dissemination of clinical practice guidelines. We now know, however, that the availability of clinical guidelines in isolation has generally failed to promote voluntary change in practice patterns.

Even techniques used successfully by pharmaceutical representatives—one-on-one discussion of a product and its uses (academic detailing, AD)—or the use of continuous quality improvement (CQI) teams have failed to improve guideline compliance or clinical outcomes for two very common chronic conditions-hypertension and depression. These are the findings of the first multisite randomized controlled trial of guideline implementation using AD techniques and CQI teams, which was supported by the Agency for Health Care Policy and Research (HS07652).

Researchers led by Harold I. Goldberg, M.D., of the University of Washington, randomized 15 small group practices of providers and patients into three groups: academic detailing alone, AD plus CQI teams, and usual care (UC), to examine the impact of these approaches on implementation of depression and hypertension practice guidelines. For hypertension, neither the AD nor the CQI team intervention, either alone or in combination, was associated with improvement in prescribing patterns or hypertension control rates. Across all clinics, no intervention effects were found in either depression recognition, improved prescribing, or reduced symptomatology, with the exception of a small decrease (4.7 percent) in first-generation tricyclic use associated with academic detailing versus usual care. This change was consistent with the guideline recommendation to avoid antidepressants that frequently cause side effects.

The researchers point out that both AD techniques and CQI teams are socially complex interventions whose implementation is sensitive to local differences in the organizations' cultures and personnel and the diseases involved. They conclude that it is unlikely at this point that widespread dissemination using either approach would markedly improve compliance with chronic disease care guidelines.

See "A randomized controlled trial of CQI teams and academic detailing: Can they alter compliance with guidelines?" by Dr. Goldberg, Edward H. Wagner, M.D.,M.P.H., Stephan D. Fihn, M.D., M.P.H., and others, in the March 1998 Joint Commission Journal on Quality Improvement 24(4), pp. 130-142.

Return to Contents
Proceed to Next Section

The information on this page is archived and provided for reference purposes only.


AHRQ Advancing Excellence in Health Care