Skip Navigation U.S. Department of Health and Human Services www.hhs.gov
Agency for Healthcare Research Quality www.ahrq.gov
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 It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work.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: https://info.ahrq.gov. Let us know the nature of the problem, the Web address of what you want, and your contact information.

Please go to www.ahrq.gov for current information.

New model projects 11 percent fewer deaths in 20 years if U.S. smokers stopped smoking

If all adult Americans who smoked in the mid-1970s had quit, there would now be about 11 percent fewer overall deaths in the United States, concludes a study supported by the Agency for Health Care Policy and Research (HS07002). Men would account for 58 percent of deaths avoided by year 20, women for 41 percent. Men have historically smoked more than women, so that their exposure before they stop smoking is greater and has a greater impact on their health, explains Louise B. Russell, Ph.D., of Rutgers University's Institute for Health.

Dr. Russell and her colleagues developed a simulation model based on data from a longitudinal survey of a representative sample of over 11,000 U.S. adults 25 and older, which included risk factors for all-cause mortality, hospital admissions, and nursing home admissions that occurred after baseline (1971 to 1975) through 1987. They estimated the probability of survival for each adult in the sample each year after baseline as a function of baseline risk factors. Risk factors included, among others, diabetes, smoking, high blood pressure, and elevated serum cholesterol. The researchers then projected the total number of deaths each year based on these risk factors.

To test the model's validity, the researchers compared actual deaths with projected deaths for the first 12 years of followup, the longest period of followup available for all participants since some did not enter the sample until 1975. The researchers found that projected deaths closely matched observed deaths by age and sex. They kept baseline risk factors at their observed values, but changed smokers to former smokers to compute annual projections for 5 and 20 years after baseline. Five years after baseline, cumulative deaths were 15 percent lower than projected if no one smoked, with a decline of 11 percent in total deaths by the 20th year.

Details are in "Modeling all-cause mortality: Projections of the impact of smoking cessation based on the NHEFS," by Dr. Russell, Jeffrey L. Carson, M.D., William C. Taylor, M.D., and others, in the April 1998 American Journal of Public Health 88, pp. 630-636.

Editor's Note: Select to access a summary in the March/April 1998 issue of a related article by Dr. Russell and colleagues on the relationship between smoking cessation and the likelihood of hospitalization ("Biomedical risk factors for hospital admission in older adults," by Jane E. Miller, Dr. Russell, Diane M. Davis, and others, Medical Care 36(3), pp. 411-421, 1998).

Return to Contents

Outcomes/Effectiveness Research

Outcomes are improving for older patients who undergo carotid endarterectomy

Carotid endarterectomy (surgical removal of plaque from the carotid artery) is performed to prevent stroke in increasingly older patients. Complications and deaths associated with the procedure are declining among older patients, according to a study by David C. Hsia, J.D., M.D., M.P.H., of the Agency for Health Care Policy and Research. Dr. Hsia and his colleagues from the Office of the Inspector General, Department of Health and Human Services, analyzed Medicare bills to calculate carotid endarterectomy frequency, rate, and 30-day perioperative mortality according to patient demography and hospital characteristics.

Medicare patients account for 75 percent of all carotid endarterectomies performed. After initially peaking at 61,273 procedures in 1985, the frequency of carotid endarterectomy among Medicare beneficiaries declined to 46,571 in 1989 in response to reports questioning the procedure's effectiveness in stroke prevention. Procedure rates then rose to 108,275 in 1996, when additional reports affirmed its usefulness.

Perioperative mortality among Medicare patients declined from 3 percent in 1985 to 1.6 percent in 1996, paralleling but not matching the lower mortality rates of clinical trials. In patients over age 75 years, the share of surgery steadily increased from 34.5 percent in 1985 to 43.7 percent in 1996. However, patients aged 85 years and older suffered twice the average perioperative mortality. Men and white patients had significantly higher surgery rates than women and black patients throughout the study period.

In 1985, hospitals performing more than 50 procedures per year made up 12 percent of the 2,747 institutions but had nearly 45 percent of the patients. By 1996, 28 percent of the 2,607 hospitals did more than 50 carotid endarterectomies annually and performed the procedure on 70 percent of the patients.

For more information, see "Epidemiology of carotid endarterectomy among Medicare beneficiaries: 1985-1996 update," by Dr. Hsia, Linda M. Moscoe, B.A., and W. Mark Krushat, M.P.H., Sc.D., in Stroke 29, pp. 346-350, 1998. Reprints (AHCPR Publication No. 98-R039) are available from the AHCPR Publications Clearinghouse.

Maternal leptin concentration during pregnancy is not an accurate predictor of infant birthweight

High blood levels of the protein product leptin have been linked to high body mass index (BMI) and obesity. Yet a woman's leptin blood levels during pregnancy do not correlate with her infant's birthweight, according to the Patient Outcomes Research Team (PORT) on the Prevention of Low Birthweight in Minority and High-Risk Women. The PORT is led by Robert L. Goldenberg, M.D., and supported by the Agency for Health Care Policy and Research (contract 290-92-0055). In a recent PORT study, more obese women had higher leptin concentrations during pregnancy, but those leptin blood levels—regardless of maternal BMI—did not correlate with infant birthweight.

Researchers at the University of Alabama at Birmingham studied 135 pregnant women who delivered full-term infants (66 with and 69 without growth-retarded infants). They evaluated the relationship between serum leptin in these pregnant women and weight gain during pregnancy, their prepregnancy BMI, and the birthweight of their infants. Cord serum leptin levels were also measured. Results showed that mean maternal serum leptin concentrations adjusted for BMI were highest at about 22 to 27 weeks' gestation (29.8 ng/mL) and declined thereafter until term (25.2 ng/mL).

There was no significant correlation between maternal serum leptin concentrations and birthweight in any BMI group or during any period of pregnancy in women with and without growth-retarded infants. However, mean cord serum leptin correlated significantly with birthweight, with higher levels of serum leptin correlating with higher birthweight. Although leptin—which has a molecular weight of 16,000—probably does not cross the placenta, there was a significant correlation in leptin concentrations between maternal serum and cord serum. Mean cord serum leptin concentration (10.8 ng/mL) ranged from 20 percent to 67 percent of maternal serum leptin at 37 weeks' gestation, depending on maternal BMI. The researchers conclude that it is likely that the correlation between maternal and fetal leptin concentrations is mediated by the increase in fat mass of the maternal-fetal unit.

Details are in "Serum leptin concentrations during pregnancy and their relationship to fetal growth," by Tsunenobu Tamura, M.D., Dr. Goldenberg, Kelley E. Johnston, and Suzzane P. Cliver, in the March 1998 Obstetrics & Gynecology 91(3), pp. 389-395.

Return to Contents

Clinical Decisionmaking

Specific factors identify asthmatics most at risk for relapse after ER treatment

About one-fourth of adult asthmatics relapse within 10 days of visiting the emergency department (ED) for asthma episodes. By 2 months, nearly half (45 percent) suffer a relapse, according to a recent article reporting a study funded in part by the Agency for Health Care Policy and Research (HS07103 and HS07969). The study, led by Robert J. Zalenski, M.D., M.A., and Michael T. McDermott, M.D., shows that these asthma patients typically had made three or more visits to the ED in the prior 6 months, had difficulty performing work or other activities due to their asthma in the prior month, and/or had discontinued hospital-based treatment for the ED asthma episode within 24 hours without having achieved at least 50 percent of their peak expiratory flow rate (ability to breathe out).

These three risk factors for relapse may help to identify patients with poorly controlled asthma who probably need more intensive and comprehensive management, explains the article's lead author Madeline McCarren, Ph.D., of the University of Illinois at Chicago. The researchers examined the association between a patient's medical history, presenting symptoms, and treatment-related characteristics with asthma relapse (in this case, relapse means an unscheduled return to the ED or doctor's office for worsening asthma symptoms) within 8 weeks of an ED visit. They studied 284 mostly poor, black, urban, male adult asthmatics who visited the Cook County Hospital ED for a moderate to severe asthma episode from 1992 to 1995.

Patients with none of the three risk factors had a 14 percent probability of relapse within 8 weeks compared with 38 percent for patients with one risk factor, 63 percent for those with two risk factors, and 100 percent for patients with three risk factors. Patients who completed the postdischarge oral anti-inflammatory course of the steroid prednisone were half as likely to relapse as patients who did not take all of their medicine. The patients at highest risk displayed an accelerated relapse rate after 3 to 4 weeks, shortly after they would have finished steroid therapy. This suggests that longer courses of prednisone might reduce relapse rates, notes Dr. McCarren.

Details are in "Prediction of relapse within eight weeks after an acute asthma exacerbation in adults," by Drs. McCarren, McDermott, Zalenski, and others, in the Journal of Clinical Epidemiology 51(2), pp. 107-118, 1998.

Current strategy for routine thyroid function tests should be reconsidered

It is currently recommended that the free thyroxine (FT4) test be used prior to the more costly thyrotropin (TSH) test for thyroid function testing. This strategy should be reconsidered, according to a study supported in part by the Agency for Health Care Policy and Research (HS07381).

Researchers at the Hawaii MEDTEP Minority Research Center examined the effectiveness and cost of alternative test sequences: TSH test followed by the FT4 test, if the TSH test result is abnormal; the FT4 test first, followed by the TSH test; and performing both tests together. They applied these strategies to consecutive patients referred for a thyroid function test to a health maintenance organization (HMO), a multispecialty fee-for-service group, a military hospital, and a commercial laboratory.

The TSH and FT4 tests had similar sensitivities for detecting clinical hyperthyroidism (overactive thyroid) and hypothyroidism (underactive thyroid). However, the TSH test detected subclinical function and monitored thyroxine treatment better. The FT4 test detected central hypothyroidism and monitored rapidly changing thyroid function better. Direct costs for the two tests were the same, but charges for the TSH test were higher. The average direct cost per patient, starting with the FT4 test, was $4.61; starting with the TSH test, $5.90; and with both tests together, $6.50.

The researchers favor the TSH-FT4 test sequence for nearly all routine screening, diagnostic testing, and treatment monitoring because it optimizes diagnostic accuracy, reduces laboratory work, and simplifies ordering by clinicians. The small additional cost of starting with the TSH instead of FT4 assay is outweighed by the added value of detecting the subclinical function states and by more effective thyroxine therapy monitoring, explain the researchers. However, they urge caution for patients with rapidly changing thyroid function and those suspected of having or known to have central hypothyroidism, conditions best detected by the FT4 assay.

For more information, see "Alternative sequences of thyrotropin and free thyroxine assays for routine thyroid function testing," by Robert A. Nordyke, M.D. (deceased), Thomas S. Reppun, M.D., Lynn D. Madanay, M.D., and others, in the February 9, 1998 Archives of Internal Medicine 158, pp. 266-272.

Computer system may someday provide quick trauma advice to emergency medicine physicians

A patient with a gunshot wound to the abdomen arrives at the emergency department (ED) in shock and losing blood rapidly. A relatively inexperienced resident surgeon decides to do a CT (computerized tomography) scan of the abdomen to locate the source of the bleeding and then take the patient to the operating room. The attending physician intervenes and recommends an abdominal x-ray, an effective procedure that is faster than a CT scan. The attending physician also suggests that the resident x-ray the chest to verify that the bullet did not travel upward and cause chest-cavity injuries.

Emergency physicians may eventually turn to TraumaTIQ to find similar advice displayed on a computer screen as they work on a patient. TraumaTIQ is the critiquing interface for TraumaAID, an on-line decision support system for managing trauma patients. TraumaTIQ was developed by Abigail S. Gertner, Ph.D., of the University of Pittsburgh, and Bonnie L. Webber, Ph.D., of the University of Pennsylvania. This work was done while Dr. Gertner was a graduate student at the University of Pennsylvania. Their work was supported in part by the Agency for Health Care Policy and Research (HS06740).

With TraumaTIQ, an ED nurse chronologically records all the findings, tests, and treatments pertaining to the patient on an electronic version of a standard trauma flow sheet. Each entry triggers TraumaTIQ's critiquing process. With each new input, the system interprets the physician's orders in terms of underlying goals, evaluates the inferred plan by comparing it to TraumAID's recommended plan, and then generates a critique that calls attention to potential problems. TraumaTIQ runs in conjunction with TraumAID on both Unix/X Windows® and Macintosh platforms. TraumaTIQ is now being field tested to see if it can be integrated into the fast-moving, highly structured ED environment.

Details are in "TraumaTIQ: Online decision support for trauma management," by Drs. Gertner and Webber, in the January 1998 issue of IEEE Intelligent Systems, pp. 32-39.

Videos of prostate cancer treatment alternatives benefit both physicians and patients

When physicians show men with prostate cancer a standardized video presentation on treatment alternatives for their condition prior to the doctor's discussion of these alternatives, patients appear to benefit. Patients gain a better understanding of the risks and benefits of various treatment approaches. Also, the doctor can then focus on risk-benefit tradeoffs of treatments rather than merely describing them when talking to the patient. In addition, these videos can be incorporated into busy medical office practices, concludes a study supported by the Agency for Health Care Policy and Research (HS08397).

During 1994 and 1995, the Prostate Patient Outcomes Research Team, in a substudy led by Peter Albertsen, M.D., M.S., at the University of Connecticut, showed 111 men with newly diagnosed localized prostate cancer a videotape about the risks and benefits of four treatment options: radical surgery, external beam radiation, hormonal therapy, and watchful waiting. The researchers assessed the usefulness of the video by having each patient complete a questionnaire before and after viewing the video and again following a discussion with his treating physician.

Upon entering the physicians' offices, 56 percent of men felt they had only a fair or poor knowledge of prostate cancer or its treatment options. After viewing the video, only 6 percent felt this way; 80 percent felt they had a good or very good understanding (compared with 38 percent before), and 14 percent felt they had an excellent understanding (versus 6 percent before). Followup discussion with their personal physicians further increased patients' knowledge. The final survey showed 40 percent of patients felt they had an excellent understanding and 55 percent a good or very good understanding of prostate cancer and its treatment options. Also, 75 to 84 percent of patients felt they participated "a lot" in the treatment decision, and 93 percent were satisfied with their decision.

See "Assessment of the feasibility and impact of shared decision making in prostate cancer," by Erol Onel, M.D., Cristina Hammond, B.A., John H. Wasson, M.D., and others, in Urology 51(1), pp. 63-66, 1998.

Return to Contents

Health Care Delivery

Physicians can identify women with bothersome incontinence by asking them a few simple questions

Urinary incontinence strikes many older women, interfering with their ability to enjoy daily activities, social interactions, and travel. Often, these women are too embarrassed to ask their doctors about their problem. Primary care physicians should take the lead by asking women if they are bothered by incontinence and, if so, about their voiding and leaking patterns, suggests a study supported by the Agency for Health Care Policy and Research (HS08716).

A few questions can substitute for two lengthy standardized questionnaires that are typically used to identify the impact of urinary incontinence on quality of life, according to the Wake Forest University researchers who performed the study. They conducted a telephone survey with 384 incontinent women from 41 rural and urban primary care practices in North Carolina.

Women were identified as incontinent if they had experienced at least one episode of incontinence per week in the previous 3 months. The survey gathered demographic data—such as age, race, marital status, and education—and included several questions about the subject's symptoms of urinary incontinence. Each incontinent participant also completed the seven-item Incontinence Impact Questionnaire and the six-item Urogenital Distress Inventory. The women's responses during the telephone survey were then correlated with their answers on the standardized instruments.

The researchers found that a positive response to the simple question, "Do you consider this accidental loss of urine a problem that interferes with your day-to-day activities or bothers you in other ways?" was a good predictor of incontinence severe enough to affect the woman's life quality (as indicated on the questionnaires). Simply asking women this question and asking them about the amount and frequency of urine loss and daily voids would enable physicians to identify and treat affected women without the use of time-consuming questionnaires, conclude the researchers.

For more information, see "Relationship between patient reports of urinary incontinence symptoms and quality of life measures," by Deirdre Robinson, M.D., Katherine F. Pearce, M.D., John S. Preisser, Ph.D., and others, in the February 1998 Obstetrics & Gynecology 91(2), pp. 224-228.

Social conversation during physician visits has a positive effect on patient satisfaction

Doctors may sometimes cut short social conversation during office visits with sicker patients in order to focus more on the medical aspects of the patients' condition. Although social conversation may appear to divert valuable time from strictly medical concerns, some sick patients may interpret reluctance to engage in social conversation as a lack of regard for them, leaving them less satisfied with a doctor's visit. These are the findings of a recent study supported by the Agency for Health Care Policy and Research (HS06999).

Judith A. Hall, Ph.D., of Northeastern University, and her colleagues examined the relationship between the health status of patients, the physician's patterns of communication, and patients' satisfaction with the care received. They defined "social conversation" as casual chit-chat, which usually precedes getting down to the business of the medical visit. The researchers studied two groups of patients to find out why sicker patients seem to be less satisfied with their medical care than healthier patients.

For the first group, the researchers measured health status and satisfaction with care of 114 patients with initial visits to Boston rheumatologists; the second group included 649 patients who made continuing care visits to Baltimore area internists and primary care physicians. In both studies, physician/patient communication was taped during the visits, and rating instruments were used after the visits to survey patients about their satisfaction.

The first study showed that patient health and physician communication were not significantly related to patient satisfaction with care, but that poor health directly produced dissatisfaction. However, the study sample may not have been large enough to show the effects of differences in physician communication styles. The second study also showed that illness directly produced dissatisfaction, but it also showed that physicians' use of social conversation seemed to be a mediating link between health and patient satisfaction. Patients in better health received more social conversation, which in turn influenced their higher rating of the physician's psychosocial responsiveness and their satisfaction. These findings may be of particular concern to managed care organizations trying to conserve resources by limiting the length of physician visits—the effort may result in reduced patient satisfaction, cautions Dr. Hall.

See "Why are sicker patients less satisfied with their medical care? Tests of two explanatory models," by Dr. Hall, Michael A. Milburn, Ph.D., Debra L. Roter, Dr.P.H., and Lawren H. Daltroy, Dr.P.H., in Health Psychology 17(1), pp. 70-75, 1998.

Computer reminders can increase physician-patient discussions about advance directives

Simple computer-generated reminders to discuss advance directives given to primary care doctors can increase the rate of these discussions as well as completion of advance directives among elderly outpatients with serious illnesses, finds a study supported by the Agency for Health Care Policy and Research (HS07632). Advance directives name a surrogate decisionmaker or provide written instructions about end-of-life treatment decisions should the patient be incapable of making the decisions at that time.

The study shows that physicians who did not receive reminders discussed advance directives with 4 percent of study patients compared with 24 percent for physicians who received reminders during scheduled outpatient visits. What's more, unreminded doctors completed advance directive forms with only 4 percent of study patients compared with 15 percent for physicians who received reminders. Overall, 45 percent of patients whose doctors discussed advance directives with them completed at least one type of advance directive, according to William M. Tierney, M.D., the study's principal investigator.

Previous studies have shown that both physicians and patients believe the other should initiate advance directive discussions. In this study, computer reminders broke the impasse and motivated physicians to take the first step. Researchers from the Indiana University School of Medicine, the Regenstrief Institute for Health Care, the Indianapolis Veterans Affairs Medical Center, and St. Louis University randomized 147 primary care physicians at an outpatient general medicine practice to receive computer-generated reminders to discuss one or both of two types of advance directives (written instructions about end-of-life care preferences or a proxy directive) or to receive no reminder.

Details are in "Effectiveness of computer-generated reminders for increasing discussions about advance directives and completion of advance directive forms," by Paul R. Dexter, M.D., Frederic D. Wolinsky, Ph.D., Gregory P. Gramelspacher, M.D., and others, in the January 1998 Annals of Internal Medicine 128, pp. 102-110.

Select for information on the final report from this project ("Advance directives, proxies, and electronic medical records," HS07632).

Return to Contents

Managed Care

Researchers examine impact of managed care on physician behavior, quality of care, and coverage of medical technology

Today, more than half of all Americans are enrolled in managed care organizations (MCOs). The rapid growth in managed care has been accompanied by a slowing in the growth of national health care spending, which some attribute in part to managed care incentives to physicians to provide less care to patients.

Concern about financial arrangements and contractual agreements between MCOs and physicians has prompted considerable State regulatory activity. Concern also exists about the potential negative impact that cost control by MCOs may have on quality and on patients' access to expensive medical technology. Fred J. Hellinger, Ph.D., of the Agency for Health Care Policy and Research's Center for Organization and Delivery Studies, examines the first two issues in articles summarized here. A third article by AHCPR researcher Claudia A. Steiner, M.D., M.P.H., and Anne Elixhauser, Ph.D., formerly with AHCPR and now with MEDTAP International, examines how MCOs make coverage decisions about medical technology.

Hellinger, F.J. (1998, April). "Regulating the financial incentives facing physicians in managed care plans." American Journal of Managed Care 4, pp. 663-674.

Dr. Hellinger's review of the literature shows that during the past 18 months, 9 States have passed legislation requiring that managed care plans disclose financial incentives they offer participating physicians. Eight States have passed legislation restricting the methods used by health plans to compensate physicians. Thirty-four States have passed bills that restrict techniques that managed care plans can use to reduce the use of health services.

Thirty-two States have passed laws banning "gag rules." These are clauses sometimes found in contracts between health plans and physicians that prohibit physicians from discussing with patients treatment options not covered under the plan.

These State regulations, as well as some Federal regulations, are based on the assumption that placing physicians at too much financial risk can result in the withholding, denial, or delay of medically necessary services. However, information about how physicians react to financial incentives is difficult to ascertain because different types of financial incentives affect the cost and quality of care in different ways.

Dr. Hellinger calls for research into how treatment regimens are affected by both the form and the strength of financial incentives facing physicians. He points out that current decisions about the use and regulation of financial incentives are being based on anecdotal experience and political factors.

Reprints (AHCPR Publication No. 98-R064) are available from the AHCPR Publications Clearinghouse.

Hellinger, F.J. (1998, April). "The effect of managed care on quality: A review of recent evidence." Archives of Internal Medicine 158, pp. 833-841.

Although there is not evidence to indicate that managed care has decreased the overall effectiveness of health care, it may adversely affect the health of some vulnerable subpopulations. In particular, evidence indicates that sick enrollees who are either poor or elderly fare worse in health maintenance organizations (HMOs) than in traditional plans. Also, enrollees in managed care plans are less satisfied with their care and have more problems in gaining access to specialized services. These conclusions are based on a review of recent studies examining the relationship between managed care and quality of care—that is, the effectiveness of care, satisfaction with care, and access to care.

In almost all studies of satisfaction with care and in all studies of access to care, people in HMOs (including Medicare beneficiaries) were less satisfied and had greater difficulties accessing care than those in traditional health plans. HMO beneficiaries who were sick or poor or both were the most dissatisfied and experienced the most problems in gaining access to specialized services.

Few data exist on the impact on care of newer types of managed care plans that invest heavily in information systems and rely on financial incentives to alter practice patterns. Nor have studies addressed how certain types of financial incentives (for instance, capitation, withholds, and bonuses) affect quality of care. Dr. Hellinger concludes that in the absence of research, it is likely that managed care plans will continue to use financial incentives to control costs without a full understanding of their impact on quality of care.

Reprints (AHCPR Publication No. 98-R056) are available from the AHCPR Publications Clearinghouse.

Steiner, C.A., and Elixhauser, A. (1998, January). "Managed care, technology assessment and coverage of medical technology." Today's Internist, pp. 22-27.

While clinical considerations are primary, cost-effectiveness is ranked among the top five most important factors in MCO decisions about medical technology coverage, according to Drs. Steiner and Elixhauser. Also, cost considerations appear to vary by whether or not specialty physicians are placed at financial risk.

Medical directors in MCOs where physicians who provide specialty care are primarily salaried or reimbursed on a fee-for-service (FFS) basis (that is, they do not bear financial risk) are more likely to list increased cost as a reason to deny coverage than their counterparts in MCOs that pay specialty physicians on a capitated basis. When specialty physicians are capitated, the MCO may be more likely to seek their input in making coverage decisions, since they have a shared interest in controlling costs. In contrast, in FFS-based or salaried arrangements, centralized coverage decisions are a primary means to achieve cost control.

These findings are based on a summary of the results of three surveys in which pharmaceutical directors, medical directors, and other decisionmakers in MCOs responded to questions about coverage of medical technologies. The surveys also revealed that, for the most part, MCOs favor covering a particular technology when findings indicate that the technology is medically acceptable, reasonable, and necessary; FDA-approved; cost effective compared with its alternative; efficacious; and has a lower complication rate than alternative techniques.

Reprints (AHCPR Publication No. 98-R038) are available from the AHCPR Publications Clearinghouse.

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