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

Research Briefs

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 It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work.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.

Auerbach, A.D., Landefeld, S., Shojania, K.G. (2007, September) "The tension between needing to improve care and knowing how to do it." AHRQ grant HS11416. New England Journal of Medicine 357(6), pp. 608-13.

The authors present seven arguments for and against the rapid dissemination of quality improvement interventions. The arguments for proceeding quickly include: many thousands of patients are injured or killed annually by medical errors; any harms from quality improvement are likely to be less than those from preserving the status quo; and the nature of quality improvement exempts it from the usual strategies of assessment.

Several arguments caution against proceeding quickly and for taking the time for thorough evaluation. For example, in the treatment of disease, where the need to improve is equally urgent, we demand rigorous evidence that a therapy works before recommending it widely. Knowing the harms and opportunity costs of quality improvement is important to understanding the net benefit to patients, which may be small.

Finally, given the complexity of quality and safety problems and their causes, rigorous study designs are needed to properly evaluate them. The authors suggest resisting the temptation to circumvent traditional models of evidence in the case of quality improvement. They assert that in pursuing quality and safety improvement, we should not blind ourselves to harms, squander scarce resources, or be deluded about the effectiveness of our efforts.

Cohen-Mansfield, J. and Lipson, S. (2006, Summer). "To hospitalize or not to hospitalize? That is the question: An analysis of decision making in the nursing home." (AHRQ grant HS09833). Behavioral Medicine 32(2), pp. 64-70.

A survey of 6 physicians and a nurse practitioner at 1 nursing home of 52 residents asked about the medical event leading to consideration of hospitalization, the decisionmaking process, factors involved in making treatment decision, and the role of advance directives. Hospitalized residents had fewer treatments considered (1.8 vs. 3.5) and fewer treatments chosen (1.3 vs. 1.8) than those who were not hospitalized. The physicians rated general treatment practice of the condition in question as important or very important in 68 percent of hospitalized residents, but in only 31.6 percent of nonhospitalized residents. Cost and quality of life were rated as more important for nonhospitalized residents (48 and 76 percent, respectively) than for hospitalized residents (17.4 and 43.5 percent, respectively). Advance directives for "no hospitalization" were associated with fewer hospitalizations.

Ditto, P.H., Jacobson, J.A., Smucker, W.D., and others (2006, August). "Context changes choices: A prospective study of the effects of hospitalization on life-sustaining treatment preferences." (AHRQ grant HS08180). Medical Decision Making 26, pp. 313-322.

Researchers analyzed responses of 401 elderly individuals, who participated in 3 annual interviews about advance directives. A subsample of 88 individuals, who were hospitalized for more than 48 hours during the course of the study, participated in an additional "recovery" interview conducted about 2 weeks after their release from the hospital. At each interview, people indicated their desire to receive four life-sustaining medical treatments: cardiopulmonary resuscitation (CPR) for cardiac arrest; artificial nutrition and hydration (ANH) for inability to eat or drink; gall bladder surgery for a life-threatening gall bladder infection; and antibiotics for life-threatening pneumonia.

The treatments were for four serious illness scenarios: Alzheimer's disease; terminal cancer with pain; coma with no chance of recovery; and stroke with a slight chance of recovery. Among patients who had been hospitalized, CPR refusal decisions remained highly stable across all three interviews. However, 56 percent of all decisions to receive CPR at the prehospitalization interview changed to decisions to refuse treatment at the recovery interview, and 56 percent of those initially unstable decisions changed back to wanting CPR by the time of the posthospitalization interview several months later. Thus, physicians and family members should be sensitive to the possibility that decisions to receive life-sustaining treatment stated by healthy individuals may be particularly susceptible to changing circumstances, advise the researchers.

Fleishman, J.A. and Zuvekas, S. H. (2007, July). "Global self-rated mental health: Associations with other mental health measures and with role functioning." Medical Care 45(7), pp. 602-609.

The purpose of this research was to compare results from the global self-reported mental health (SRMH) item in the Medical Expenditure Panel Survey (MEPS) with other measures of emotional well-being or psychological distress. Two of these measures, the mental health component summary (SF-MCS) and the mental health subscale (SF-MH), were drawn from the SF-12 (a survey measuring eight domains of physical and mental health).

The other two measures were the K6 scale (a 6-item scale of psychological distress) and the Patient Health Questionnaire-2 (PHQ-2), a 2-item screener for depression. The authors found that responses to the SRMH were moderately associated with other measures of mental health. However, the other measures assessing emotional well-being and psychological distress were more strongly correlated with each other than with the SRMH.

With regard to role functioning, SRMH scores were correlated with physical and emotional role functioning, but not as strongly as were the other mental health measures. The authors recommend further research, possibly including cognitive interviewing techniques, into the factors affecting responses to the SRMH. The authors also recommend that the self-reported mental health (SRMH) item not be used as the sole indicator of emotional well-being or psychological distress.

Gorawara-Bhat, R., Cook, M. A., and Sachs, G. A. (2007). "Nonverbal communication in doctor-elderly patient transactions (NDEPT): Development of a tool." (AHRQ grant HS14088). Patient Education and Counseling 66, pp. 223-34.

The authors developed a new tool, the NDEPT, for observing and coding nonverbal communications in doctor-elderly patient interactions in exam rooms. For purposes of the study, the authors divided nonverbal communication into two major dimensions: the physical setting of exam rooms enclosing interaction and the body language of the physician unfolding within the exam room.

The NDEPT was developed based on 50 videotapes of routine exam room visits between physicians and their elderly patients. The researchers measured static, dynamic, and kinesic attributes. Static attributes include furniture and equipment. Dynamic attributes include the interaction distance, vertical height distance, angle of interaction, and physical barriers between doctor and patient. Kinesic attributes of the physician include stance, eye contact, facial expression, gesture, and touch. One conclusion was that the spatial configuration of the exam room either facilitated or impeded the manifestation of the physician's kinesic attributes. The authors recommend testing and validation with larger samples.

Johnson, M.D., Meredith, L.S., Hickey, S.C., and Wells, K.B. (2006, September-October). "Influence of patient preference and primary care clinician proclivity for watchful waiting on receipt of depression treatment." (AHRQ grant HS08349). General Hospital Psychiatry 28, pp. 379-386.

Primary care patients with depression who prefer watchful waiting to treatment are less likely to receive treatment for their depression, especially those with subsyndromal depression (depressive symptoms only), according to a new study. Brief established questionnaires can help providers distinguish depressive disorders from symptoms only and patients should be involved in decisions about the types of treatments available, such as medication and/or therapy.

Researchers found that patient knowledge about depression and its treatment was also significantly associated with the likelihood of receiving treatment. The clinicians' general tendency toward watchful waiting was not significantly associated with the likelihood that their individual patients received treatment for depression.

Katzan, I., Dawson, N.V., Thomas, C.L., and others (2007, May). "The cost of pneumonia after acute stroke." (AHRQ grant HS09969). Neurology 68, pp. 1938-1943.

This study found that hospitalization costs for stroke patients who develop pneumonia ($21,043) are nearly $15,000 more than for patients who require treatment for stroke alone ($6,206). Patients who develop pneumonia after stroke are also 70 percent more likely to need extended care once they are discharged from the hospital, further adding to costs.

When study data were extrapolated to the 553,000 similar patients admitted to U.S. hospitals each year, the cost for pneumonia complications in this group approached $460 million. Researchers studied 11,286 Medicare beneficiaries in 29 hospitals in the Cleveland area who were admitted after suffering strokes. Nearly 6 percent of patients developed pneumonia in the hospital. Patients who were admitted from nursing homes or who were more severely ill were more likely to develop pneumonia. The rate of pneumonia infections was similar for teaching and nonteaching hospitals.

Lee, J.Y., Rozier, R. G., Lee, S. D., and others (2007, Spring). "Development of a word recognition instrument to test health literacy in dentistry: The REALD-30—a brief communication." (AHRQ grant 13233). Journal of Public Health Dentistry 67(2), pp. 94-98.

This study explored dental health literacy by developing and testing a screening instrument for dental word recognition, the Rapid Estimate of Adult Literacy in Dentistry (REALD-30). The REALD-30 contains 30 words covering etiology, anatomy, prevention, and treatment that are commonly used in brochures and written materials provided to dental patients.

In order to test its internal reliability and its relationship to two dental outcomes (perceived dental health status and oral health-related quality of life), it was administered to a convenience sample of 202 English-speaking patients. Several other tests on adult literacy in medicine, functional health literacy, and an oral health profile were given at the same time. The REALD-30 had good internal reliability; however, results on validity were mixed.

Test results correlated with the two literacy tests. The REALD-30 results were associated with oral health-related quality of life, but were not associated with perceived dental health status. The authors recommend research going beyond word recognition to include a future focus on developing a comprehensive dental health literacy instrument that includes comprehension, numeracy, and verbal components.

Localio, A. R., Margolis, D.J., and Berlin, J.A. (2007). "Relative risks and confidence intervals were easily computed indirectly from multivariable logistic regression." (AHRQ grants HS10399 and HS11481). Journal of Clinical Epidemiology 60, pp. 874-882.

The authors performed simulations on two hypothetical groups of patients at a single center to assess alternative statistical methods for estimating relative risks and their confidence intervals when outcomes are common. Confidence intervals are essential to support estimates of relative risk of treatment or exposure from multivariable regression models.

The authors' simulations demonstrate why a method of substitution commonly used in leading journals fails when baseline risk and relative risk are not small—the very situations for which it was designed. The confidence intervals are too narrow and therefore precision of estimates is overstated. By contrast, other regression models seem to work better. For example, confidence intervals based on either bootstrapping resampling or the delta method demonstrate appropriate coverage when outcomes are common. In general, the bootstrap method seems to perform somewhat better. The authors conclude that estimates of risks and their ratios should be based on a regression model that best fits the data.

MacMullen, N.J., Tymkow, C., and Shen, J.J. (2006, July/ August). "Adverse maternal outcomes in women with asthma: Differences by race." (AHRQ grant HS13506). American Journal of Maternal Child Nursing 31(4), pp. 263-268.

This study found that minority pregnant women with asthma have significantly higher rates of preterm labor, gestational diabetes, and infection of the amniotic cavity than pregnant white women. Black women were the youngest (24 years old) and had the highest incidence of preterm labor (5.5 percent) and pregnancy-induced hypertension (5 percent). Asian/Pacific Islander women had the greatest occurrence of gestational diabetes (7.2 percent) and were over three times more likely than white women to have infection of the amniotic cavity (5.7 vs. 1.8 percent). Black and Hispanic women also had more infections of the amniotic cavity than white women (3.1 and 2.7 vs. 1.8 percent, respectively). Otherwise, pregnancy outcomes of Hispanic women were similar to those of white women, except that postdate pregnancy was less likely to be over 42 weeks.

The findings were based on examination of 11 adverse maternal outcomes across 4 ethnic groups of 13,900 pregnant women with asthma, who gave birth between 1998 and 1999. Their data came from AHRQ's Healthcare Cost and Utilization Project National Inpatient Sample of hospitalized patients.

Main, D. S., Henderson, W. G., Pratte, K., and others. (2007, June). "Relationship of processes and structures of care in general surgery to postoperative outcomes: A descriptive analysis." (AHRQ grant HS0119). Journal of the American College of Surgery 204 (6), pp. 1157-65.

The National Surgical Quality Improvement Program (NSQIP) offers participating hospitals the opportunity to learn more about the quality of their surgical care through collection and feedback of institution-level information on risk-adjusted morbidity and mortality. In order to systematically collect quantitative data on structures and processes of surgical care (not previously a focus of the NSQIP), the authors developed the Structures and Processes of Surgical Care Survey and administered it at 123 Veterans Administration (VA) hospitals. The final response rate was 73 percent, with 90 VA hospitals self-reporting full data.

Fourteen of the 35 structure and process characteristics were correlated with the hospital's observed-to-expected ratio (O/E) for morbidity; however, only four characteristics were correlated with the hospital's O/E ratio for mortality. For example, risk-adjusted morbidity was higher in centers with lower monthly surgical volume per full-time equivalents (FTE) in surgery, anesthesia, and nursing. Risk-adjusted mortality was higher in centers with a smaller percentage of patients whose anesthesia provider was the same during pre-, intra-, and post-operative care. The authors believe that their results support the feasibility and utility of measuring structure and process measures using a self-report survey.

Mercer, S.L., DeVinney, B.J., Fine, L.J., Green, L.W., and Dougherty, D. (2007). "Study designs for effectiveness and translation research. Identifying tradeoffs." American Journal of Preventive Medicine 33(2), pp. 139-154.

This article reports on a conference of research experts and federal and foundation funders, which considered the applicability of different study designs to examining the effectiveness and translation of complex, multilevel health interventions. They considered the strengths and limitations of nine different types of randomized and nonrandomized study designs, as well as natural experiments.

The choice of study design is shaped by the specific research question; the level of understanding and certainty about the underlying theory, mechanisms, and efficacy of an intervention; the possibility of randomizing individuals or groups; and the availability of natural experiments. It is also influenced by the level of available resources; the extent of generalization required; and the views of intended users of the research and study subjects. Symposium participants supported the need for replication of studies and the use of alternative study designs, since no one study can demonstrate causality. They also highlighted the importance of reviewing studies across disciplines and traditions that contribute to causal inference within health, including epidemiology, economics, and medical anthropology, among others.

Reprints (AHRQ Publication No. 07-R071) are available from the AHRQ Publications Clearinghouse.

Muller, K.E., Edwards, L. J., Simpson, S.L., and others. (2007, March). "Statistical tests with accurate size and power for balanced linear mixed models." (AHRQ Grant HS13353). Statistics in Medicine 26, pp. 3639-60.

In the analysis of Gaussian data from medical imaging research and other fields, linear mixed models have been widely used due to their convenience. However, the standard mixed model tests, when used in small samples, frequently have greatly inflated test size. Mixed models are not required for many applications with correlated outcomes in medical imaging. If special cases are stated as a general linear multivariate model, they can then be analyzed with either the univariate (UNIREP) or multivariate (MULTIREP) approach to repeated measures. Since both of these test types always control test size and have good power approximation, mixed model tests should never be used when either a UNIREP or MULTIREP tests applies. The researchers ran simulations to demonstrate that new power approximations for all four UNIREP tests eliminate most inaccuracy in existing methods.

Pelton, S.I., Huot, H., Finkelstein, J.A., and others (2007, June). "Emergence of 19A as virulent and multidrug resistant pneumococcus in Massachusetts following universal immunization of infants with pneumococcal conjugate vaccine." (AHRQ grant HS10247). Pediatric Infectious Disease Journal 26(6), pp. 468-472.

Although the pneumococcal conjugate vaccine (PCV7) has reduced infection with Streptococcus pneumoniae serotypes targeted by the vaccine, it has increased infection with some nonvaccine serotypes (NVT) among children in Massachusetts. Three years after the vaccine's introduction, children under 2 years in that State suffered a significant increase in pneumonia due to a multidrug-resistant (MDR) strain of the NVT 19A.

The MDR NVT 19A has emerged as the most frequent cause of IPD in Massachusetts. Serotype 19A was identified from nearly 30 percent of all invasive pneumococcal disease episodes among children younger than 5 years between 2001 and 2005, but 44 percent of cases in 2005. From 2002 to 2004, the majority of isolates of serotype 19A demonstrated intermediate susceptibility to penicillin and full susceptibility to the antibiotics ceftriaxone and amoxicillin. However, by 2005, serotype 16A had become resistant to these and other antibiotics. Clearly, in Massachusetts, the S. pneumoniae strains colonizing healthy children have undergone substantial shifts subsequent to introduction of the PCV7.

Prosser, L. A. and Wittenberg, E. (2007, May-June). "Do risk attitudes differ across domains and respondent types?" (AHRQ grant HS14010). Medical Decision Making, 27(3), pp. 281-287.

Patient attitudes toward risk can affect their treatment choices. Current practice assumes risk neutrality; however, little is known about whether patients exhibit different attitudes toward risk than nonpatients or whether people have different risk preferences for different goods such as money or health. Prospect theory suggests that the patient could be predicted to make riskier or less risky choices than a community member, because the patient faces possible outcomes from a different point of reference.

The authors evaluated the risk attitudes of patients with multiple sclerosis (MS) and a community sample over a health gamble and two money gambles. They administered a survey to 56 adult patients with relapsing-remitting MS and 57 adult members of the general public. The health gamble asked patients and community members to choose between two drugs of differing effectiveness in reducing or ending a projected 30-relapse. Patients and community members did not differ consistently by risk attitudes. Both were predominately risk neutral with respect to health outcomes and predominately risk averse with respect to money. Risk preferences may depend more on the characteristics of the choice than on respondent type.

Schneeweiss, S. (2007, August). "Developments in post-marketing comparative effectiveness research." (AHRQ grant HS10881). Clinical Pharmacology and Therapeutics 82(2), pp. 143-56.

Research on the comparative effectiveness of new drugs against existing drugs is of importance to physicians seeking the best treatment for patients and to health plans seeking to combine quality care with effective but less costly medicines. Since the initiation of Medicare Part D prescription drug coverage, it is also of great interest to the U.S. government, which now has a much larger stake in finding the best evidence on the comparative effectiveness of medications. Nonrandomized postmarketing comparative effectiveness research assessing the relative value of two or more drugs under conditions of routine care is needed to supplement randomized controlled trials of drugs under ideal conditions.

A valuable source of comparative effectiveness information is the large administrative databases containing claims data compiled by Medicare and major private insurers. However, nonrandomized studies utilizing claims databases or other sources may yield different results than randomized trials because of a confounding bias due to insufficient clinical information or to differing patient characteristics that are not adjusted for by the study. To reduce confounding bias with respect to observed patient characteristics, two fundamental strategies, restriction and stratification, can make patient groups more comparable and therefore comparisons less biased.

Schneeweiss, S. (2007). "Understanding secondary databases: a commentary on 'Sources of bias for health state characteristics in secondary databases'." (AHRQ Grant HS010881). Journal of Clinical Epidemiology 60, pp. 648-50.

This article is a commentary on another article in the same journal issue on sources of bias for health state characteristics in secondary databases such as electronic medical records and health insurance claims. The author makes a number of key points. First, it is critical to fully understand the process that has generated a database, not simply official documentation and guidelines but also the realities of how health care encounters translate into standardized codes. Second, coded information can be understood and analyzed as a set of proxies that indirectly describe the health status of patients through the lenses of health care providers and coders operating under the constraints of a specific health care system. Furthermore, these issues have serious implications for the internal validity of studies as well as their generalizability to specific patient subgroups, health care systems, or jurisdictions. Finally, changes in coding patterns or differences in the codes themselves introduce difficulties in comparing health services use and health outcomes over time and between jurisdictions and health plans.

Schoenman, J.A., Sutton, J.P., Elixhauser, A., and others. (2007, August). "Understanding and enhancing the value of hospital discharge data." (AHRQ contract 282-98-0024). Medical Care Research and Review 64(4), pp. 449-68.

The authors of this paper document the range of uses of hospital discharge data, present examples of its uses, and identify ways to improve both the data and how it is used in the future. They conducted a systematic review of the published literature to identify research studies using discharge data.

The aim was to capture the most prevalent uses of discharge data. These included public safety and injury surveillance and prevention; public health, disease surveillance, and disease registries; community health assessments and health planning; quality assessment and performance improvement; public reporting for informed purchasing and comparative reports; health services and health policy research applications; and private sector and commercial applications. They identified the following strengths of hospital discharge databases: they are already centrally collected and readily available; they are more reliable than data collected through other means; and they are larger and more representative of the population than other databases. Discharge data limitations include problems with the quality of included data elements, missing data elements, and excluded populations. For example, race and ethnicity data are sometimes missing or unreliably reported.

The authors recommend including new data elements to improve the discharge databases, building more comprehensive data systems, and providing technical assistance to improve the use of existing databases.

Reprints (AHRQ Publication No. 08-R001) are available from the AHRQ Publications Clearinghouse.

Setoguchi, S., Solomon, D. H., Glynn, R. J., and others (2007). "Agreement of diagnosis and its date for hematologic malignancies and solid tumors between Medicare claims and cancer registry data." (AHRQ grant HS10881). Cancer Causes Control 18, pp. 561-69.

Large databases are needed to study rare outcomes, such as the risk of certain cancers due to use of the new biological immunomodifying drugs such as tumor necrosis factor—alpha antagonists. However, neither the SEER-Medicare dataset nor cancer registry data typically include pharmacy information. The alternative is to use large health care utilization databases to identify incident cancers.

To determine the agreement between Medicare claims data and cancer registry data on the diagnosis and date for certain cancers, the authors of this study linked Pennsylvania Medicare claims and drug benefit program data on cancer patients with the state cancer registry. Their goal was to examine the accuracy of claims-based definitions for lymphoma, leukemia, and four other cancers.

The researchers concluded that claims data can identify the hematological malignancies and solid tumors with high specificity but with low to moderate sensitivity and positive predictive values. Also, they found sufficient agreement in the first dates of diagnosis between the two databases. They caution that, due to the possibility of bias, the utility of the claims-based definitions has to be assessed for each study setting.

Terris, D.D., Litaker, D. G., and Koroukian, S. M. (2007). "Health state information derived from secondary databases is affected by multiple sources of bias." (AHRQ HS00059). Journal of Clinical Epidemiology 60, pp. 734-41.

Patient encounters with the health care system may lead to sources of bias for health state characteristics derived from secondary databases. These biases may come from organizational and environmental factors. This article describes sources of bias encountered during the generation and recording of data in an individual health care encounter, through processing and storage in a secondary database.

The authors offer a conceptual framework integrating theory and empirical evidence to illustrate the ways in which the validity of health state information may be influenced by multiple dynamic factors. These include the patient's propensity to access services, community and system-based factors, characteristics of the health care encounter, and the provider's propensity to detect, treat, and record. Finally, there are the factors influencing the processing and storage of health state information in a secondary database.

These are the origin, purpose, input structure, and system-level efficiencies of the database. The authors believe that a greater awareness of these biases may lead to innovative strategies to reduce the limitations inherent in analyses of secondary databases and improve the accuracy and application of insights derived from this work.

Thomas, M.B., Simmons, D., Graves, K. and others (2007, June). "Practice/regulation partnerships: The pathway to increased safety in nursing practice, health care systems, and patient care." (AHRQ grant HS11544). Nurse Leader 5(3), pp. 50-54.

There is a need for a program that documents adverse medical errors and addresses human performance and systems factors in hospital care. This paper describes the Healthcare Alliance Safety Partnership (HASP), a nonpunitive reporting program, based on the airline industry's successful Aviation Safety Action Partnership program.

With the support of the Board of Nursing Examiners (BNE), a Texas State agency, three Texas hospitals implemented HASP as a pilot program. There are three stages to the HASP program evaluation method: discovery, analysis, and resolution. In the discovery phase, a report is received from a nurse or there is a referral from the nurse's institutional peer review committee or the BNE.

Interviews are conducted with the nurse and other involved parties. In the analysis phase, HASP nurse analysts identify and cluster causal factors of the event and an event review committee (ERC) crafts an action plan. In the resolution phase, the nurse and the hospital respond regarding prescriptive recommendations until resolution is complete and approved by the ERC.

Return to Contents

AHRQ Publication No. 08-0001
Current as of November 2007


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


AHRQ Advancing Excellence in Health Care