The article titled "Impact of Copays in Vulnerable Populations" on the November supplement of 

makes an important contribution to the literature through thecomprehensive review of the healthcare experiences of a wide range of different vulnerable populationsand their sensitivity to increasing copayments.

This review describes the experiences of thepoor, the uninsured, the chronically ill, the elderly, the mentally ill, and others. What is missing fromthis review, however, is a reflection that these vulnerable groups overlap considerably (eg, the elderlyare more likely to be chronically ill and also be poor), and that their experiences are likely toreflect the multiple risk factors that they experience. It is likely that increasing cost sharing mayhave exaggerated effects on prescription filling and use among those who are both elderly and poor.

Such overlap of populations is nothing new, but rarely do we as researchers explicitly recognizethe concomitant and potentially combinatorial effects of risk factors, such as poverty and lack ofinsurance. Shi and Stevens are only among the latest to argue that we view most vulnerable populationsby the collection of health or healthcare risk factors they possess, and work harder to understandtheir interactivity.

Previous studies have indeed counted the number of risk factors a personhas and have shown clear gradients in health services use, quality of care, and health status.

At itssimplest and arguably finest, however, the vulnerability construct can be measured by examiningthose with multiple risk factors versus 1 or none (as the authors smartly did in reviewing the experiencesof dual eligibles for Medicare and Medicaid).

The general lack of discussion concerning multiple risk factors from the review is not the fault ofthe authors, as it is something that has been missing almost entirely from the literature that was availablefor review. While each of the studies reviewed fills an important gap in knowledge, it is time thatwe begin to rethink vulnerable populations as comprehensive wholes, not as just distinct categories,however a difficult challenge. Researchers must identify strategies to comprehensively measure vulnerablepopulations; administrators need to be flexible enough to design or modify programs toaccount for multiple, overlapping risks; and practitioners need information and tools for how to servepatients who have multiple risks or comorbidities.

The supplement article provides an important first step because it effectively defines the range ofresearch knowledge on the experiences of many different vulnerable groups. The next important stepis to reassemble this information and recognize that in pharmacy and clinic practice, as elsewhere,vulnerability is considerably more than the sum of its parts.

University of Southern California Keck School of Medicine

1. Impact of copays on vulnerable populations. 

Vulnerable Populations in the United States. 

3. Stevens GD, Seid M, Mistry R, Halfon N. 

Disparities in primary care for vulnerable children: the influence of multiplerisk factors. 

Vulnerability and unmet health care needs. The influence of multiple risk factors. 

Vulnerability and the receipt of recommended preventive services: the influence of multiplerisk factors. 

Articles

LETTERS TO THE EDITORS

Reassembling Our View of Vulnerable Populations

To examine the impact of geographic variation on racial differences in 7 of 15 Health Plan Employer Data and Information Set (HEDIS) measures that assess the quality of the Medicare managed care program (also known as Medicare+Choice). 

Cross-sectional analysis using the 2004 individual-level HEDIS for Medicare managed care plans and 2003 Medicare enrollment and demographic (ie, denominator) data for more than 5.1 million Medicare+Choice enrollees. 

Individual-level HEDIS data were linked with Medicare enrollment data. Hierarchical generalized linear models were used to assess statistical significance of region and race. Direct standardization was used to estimate the rate of meeting each HEDIS standard while controlling for differences in age and sex. 

Quality of care for white Medicare+Choice enrollees was strongly correlated with the racial composition of the geographic area. Except for cholesterol management after an acute cardiac event, between-region racial variation was consistently greater than within-region racial variation. 

Removing within-region racial variation while ignoring geographic differences will not equalize the experiences of black and white elders. Rather, both racial and geographic components of healthcare quality must be addressed if the Medicare managed care program is to provide care of equal quality to all elders regardless of race. 

Analysis of quality of care measures for more than 5.1 million Medicare+Choice enrollees in 2003 revealed consistent patterns of racial and geographic disparity. The geographic disparity was greatest in areas with the highest concentration of black elders suggesting that:

A strategy that focuses on equalizing geographic disparity may disproportionately benefit black elders and would be a strong step toward the reduction of racial disparity.

Racial disparities are a broad issue for the US healthcare system. The importance of racial disparity as a policy issue is apparent from 

, which has the elimination of racial disparities in health and healthcare by 2010 as an overarching goal.

Similarly, a goal of the US Department of Health and Human Service's Initiative on Racial and Ethnic Disparities is elimination of racial disparities in health and healthcare.

The Centers for Medicare & Medicaid Services (CMS) is challenged to identify ways to eliminate or at least reduce these disparities in the Medicare program for both managed care

populations, where the existence of racial disparities in health, receipt of healthcare, and health outcomes is well documented and recognized. Despite the consistency of the evidence for the existence of racial disparities in the Medicare program, the best strategy for eliminating this variation has not been established. 

Although often treated as a completely separate issue, geographic variation in patterns of healthcare is also well recognized and documented.

Like racial disparity, the causes of geographic variation are poorly understood and are attributed to factors such as physician supply, local practice patterns, and patient culture. 

A challenge to studying both race and geography in healthcare is the strong geographic concentration of our racial/ethnic populations.

For example, the black population is geographically concentrated in the South Central and southeastern United States (

the disproportionate use of lower quality providers (be they physicians, hospitals, or managed care plans) by black elders compared with white elders complicates studies of racial disparity. As a result, researchers have 2 related challenges: first, an analytic issue and second, a policy issue. As an analytic issue, techniques such as multiple regression are frequently used to "remove" or "adjust for" the effects of potentially confounding factors such as geography. Typical multivariable approaches to studying race after adjusting for geography may underestimate the magnitude of the disparity by controlling for elements (ie, geography) that are causally related to the disparity. Second, clustering of racial or ethnic groups into poor-quality hospitals or health plans might point to interventions aimed at helping elders make better choices–such as choosing a higher quality hospital. However, if the inequality has a strong regional component, the policy options will be quite different and may need to focus on changing regional care disparities. 

With these challenges in mind, the objective of this study was to examine the impact of geographic variation on racial differences in 7 Health Plan Employer Data and Information Set (HEDIS) measures that assess the quality of the Medicare managed care program (also known as Medicare+Choice [M+C]). 

Data sources for this study are (1) the individual-level HEDIS data submitted by Medicare managed care plans for reporting year 2004 (based on 2003 experience) as a condition for continuing their M+C contract; (2) CMS denominator (ie, enrollment) files for 2003; and (3) US Census data. 

HEDIS data were merged with CMS denominator files using the approach previously described.

Briefly, individual HEDIS records, each containing the Health Insurance Claim (HIC) number, a unique identifier assigned by Medicare, were merged with the Medicare denominator file to obtain information on age, race, sex, and ZIP code of residence. Individuals were excluded from this analysis if they did not have a valid HIC, if their race was not classified as black or white, or if they were younger than age 65 years in 2003. Plans were excluded from this analysis if their submitted records failed to achieve at least a 90% match rate on the HIC. 

Overall, 148 of 162 M+C plans submitted individual-level data that were linkable with Medicare demographic information (91.4% of plans). These plans represent the experience of 81.4% of the more than 5.1 million 2003 M+C enrollees. Of 8 regions, 6 had plans that were excluded because of lack of identifiers. Finally, plans were excluded from specific measures if the audited summary reporting for the measure was not similar to the plan summary calculated from the individual records. No region had more than 3 plans excluded. 

shows the characteristics of Medicare managed care plans that submitted individual-level HEDIS data and the number of persons included in the sample for each HEDIS quality indicator. The number of plans and subjects included in these analyses varied from measure to measure because of differing reporting requirements. 

The analysis focused on 7 of the 15 HEDIS 2004 measures related to quality and outcomes. We confirmed that the patterns we reported held for the remaining 8 measures (analysis available on request). 

Race was obtained directly from the 2003 Medicare denominator files. The categories included in this analysis were white and black. All but 1 plan had at least some black members, with a median of 5.5% black and a maximum of 68% black. We imputed household income indirectly, based on figures from the US Census on the median disposable household income by ZIP code for households with persons age 65 years and older. This income estimate was grouped into 4 categories: (1) <$15 000; (2) $15 000 to <$30 000; (3) $30 000 to =$45 000; and (4) >$45 000. 

Area of residence was grouped into the 8-level Census division designation: New England, East North Central, Middle Atlantic, South Central, South Atlantic, West North Central, Pacific, and Mountain. 

Health plan size was classified as fewer than 10 000 members, 10 000 to 49 999 members, and 50 000 or more members. 

All analyses were conducted using SAS version 9.13 (SAS Institute Inc, Cary, NC). We used a hierarchical generalized linear model to account for the nesting of enrollees within health plans. For each measure, we estimated the within-region quality of care separately for white and black enrollees. We used multiple regression models to assess the impact of factors on quality of care: demographic variables (age, sex, and area income), plan size, percentage of the region that was black, race, geography, and a race/geography interaction. Adjusted rates were calculated for each measure. 

The study was approved by the Chesapeake Research Review, Inc, institutional review board (institutional review board for the National Committee on Quality Assurance), protocol number CRRI 0204002. 

The racial composition of the Medicare managed care population varied considerably across census divisions, ranging from 2% black in New England to 13.8% black in the South Central division (

). Likewise, there was considerable variation across geography in mean level of HEDIS measures for white populations. We used quality for whites as an overall measure of local background quality, which allowed us to separate quality differences that can be attributed to race alone, geography alone, and a combination of race and geography. For example, the spread in performance across geography ranged from 6.4% for glycosylated hemoglobin (A1C) testing to more than 20% for diabetic eye exams and followup after mental health hospitalizations. For all measures, divisions with higher percentages of blacks had significantly lower HEDIS quality scores (all 

illustrates the cumulative effects of race and geography on the experience of black M+C enrollees compared with white enrollees. With few exceptions, the geographic disadvantage is greater than the within-area racial disparity. For example, the geographic disparity for controlling high blood pressure was 11.9%, whereas racial disparities within a geographic area ranged from 3.1% to 10.7%. The only consistent exception to the pattern of across-geography variation in quality for whites exceeding the within-region racial variation is cholesterol management; for 4 of 8 regions, the white/black variation was higher within the region than between regions. Other specific exceptions to this pattern are ÃŸ-blocker use in the Mid-Atlantic region and A1C testing in the West North Central region. We found significant geography/race interactions for all measures except controlling high blood pressure. This pattern is consistent with the information in Table 3, which shows considerable variation in the magnitude of the white/black difference in measures by region. 

The relative impact of eliminating the impact of geographic and racial disparity can be illustrated by considering the effect of eliminating within-census division racial disparity but not equalizing performance across census divisions. As can be seen in 

, in such a case, the national level of b-blocker use after heart attack would move from 85.1% to 87.1% for blacks and remain unchanged at 93.4% for whites. However, if racial variation within areas were allowed to remain but regional variation were removed and all regions were to achieve the levels experienced by the highest performing area (in this case, New England), performance would improve to 96.3% for blacks and 97.9% for whites. The remaining racial variation would be similar to that seen with a fully regression-adjusted, race-based approach (about 1.4%). Notice that removing area variation would result in improvements in performance for both blacks and whites, but would particularly benefit blacks (an absolute increase of 12.8% vs 4.5%). 

Our results are consistent with prior literature showing considerable racial and geographic variability in quality of care in the Medicare managed care sector. We observed small but significant levels of racial disparity within all regions of the country. The estimates we present are somewhat conservative because we did adjust for differences in area income. Because of the correlations among race, geography, and poverty, some of the effect of area variation was removed by our regression models. Despite these adjustors, both racial and geographic disparities remained. 

The presence of a strong correlation between the racial composition of an area and the level of quality should not be seen as evidence of a causal relationship–variation in quality is complex and due to multiple factors. With all trends, exceptions also exist. The Mountain states have a very small percentage of black beneficiaries, but are among the lowest in quality as measured by HEDIS. Likewise, for breast cancer screening, the measure with both the highest overall performance and the greatest consistency across divisions, blacks sometimes have minimally higher-quality scores than whites. Despite the fact that the correlation should not be interpreted as indicating a causal relationship between racial composition and quality, it may point to a strategy to reduce race-based inequality that will ultimately improve the experience of both black and white beneficiaries. 

We were unable to assess whether the racial disparities were different in plans that did not submit data or submitted data without identifiers that could be linked with CMS enrollment and demographic data. Likewise, because the HEDIS data do not contain information about the treating physician, we cannot comment on the role individual physicians play in racial disparity. However, the effects that we describes are broad and consistent, suggesting that they are the result of more than the behavior of a single physician. Likewise, all regions have 9 or more plans, so the experience of a single plan would not explain a regional effect. 

CMS has an ongoing commitment to eliminating racial disparity that is both commendable and just. It would be rational for such a policy to focus on within-area interventions. However, this approach would not necessarily remove or lessen overall racial disparity because it fails to address the component of racial disparity that is tied to geographic disparity. A purely within-area approach is unlikely to equalize the average experience of black and white beneficiaries. The importance of considering both racial and regional variation is illustrated in Table 4, which highlights the disproportionate impact of geographic disparity on black populations and the need to recognize geographic variation as an important contributor to existing racial disparity. 

The obstacles associated with efforts targeted to a particular racial group are numerous and include problems with identifying eligible individuals. As a result, a geography-centered approach to quality improvement may be a strong step toward achieving the goal of lessening racial disparity in Medicare managed care. We believe that this approach is particularly well suited for improving performance in areas that have poorer quality for both black and white populations. 

 Portions of this work were presented at AcademyHealth's 2005 Annual Research Meeting, Boston, Mass, June 26-28, 2005. The authors would like to thank Drs Trent Haywood, MD, Ignatious Bau, JD, and Alan Zaslavsky, PhD, for their helpful comments and suggestions, and Russell Mardon, PhD, and Rich Mierzejewski, MS, for their assistance with analysis. 

From the Division of Health Policy and Management, University of Minnesota School of Public Health, Minneapolis, Minn (BAV); the National Committee on Quality Assurance, Washington, DC (SHS, SS); and the Department of Health Administration & Policy, College of Public Health, University of Oklahoma, Oklahoma City, Okla (AFC). 

This work was supported by a grant from The California Endowment (Targeted Capacity Expansion [TCE] grant 20032907). 

Beth A. Virnig, PhD, MPH, Associate Professor, Division of Health Policy and Management, University of Minnesota School of Public Health, 420 Delaware St SE, MMC 729 A365-Mayo, Minneapolis, MN 55455. E-mail: virni001@umn.edu. 

POLICY

Efforts to Reduce Racial Disparities in Medicare Managed Care Must Consider the Disproportionate Effects of Geography

To assess adherence with guidelines for testing and treatment of 

 infection and upper gastrointestinal symptoms. 

Retrospective longitudinal cohort analysis of patient-level medical and pharmacy claims from 75 US managed care plans. 

Persons with new claims for antisecretory medication, H pylori tests, or endoscopies were selected from among 2 million plan members continuously enrolled from 2001 to 2004 and were grouped by initial clinical diagnosis (3456 with peptic ulcer disease [PUD], 14 593 with nonulcer dyspepsia [NUD], and 36 233 with gastroesophageal reflux disease [GERD]). Diagnostic procedures, medications received, and sequencing of specific procedures and medications were examined relative to published guidelines by initial diagnosis, age, and physician specialty. 

While guidelines recommend testing before treatment, one third of persons receiving antibiotics for 

infection had not first been tested for the infection. In one third of all posttreatment testing, primary care practitioners incorrectly used serologic tests, despite their inability to distinguish cured from active infection. Eighteen percent of patients with GERD were tested for 

, although there is no causal link between them. Only two thirds of patients aged 50 to 64 years with presumed PUD underwent endoscopy, which guidelines recommend for older adults; yet one third of patients with PUD aged 18 to 49 years, for whom prompt endoscopy is generally not recommended, had an endoscopy within 30 days of their index date. 

Substantial noncompliance with widely disseminated guidelines calls for better understanding of appropriate indications for H pylori testing and endoscopy to improve patient care and conserve healthcare resources. 

Guidelines recommend that patients undergo a diagnostic test or endoscopy for determination of Helicobacter pylori infection before eradication therapy. Serologic tests only detect antibodies and cannot reliably distinguish between ongoing and cured infections. Urea breath tests and fecal antigen tests are recommended as nonendoscopic pretreatment

Guidelines call for endoscopy for patients aged 50 years or older with new-onset dyspepsia. Prompt endoscopy is generally not recommended for patients aged 18 to 49 years with peptic ulcer disease. Prompt endoscopy is reserved for older patients (aged ≥45-55 years) and for patients with alarm features.

Dyspepsia is a highly prevalent symptom complex defined as pain or discomfort centered in the upper abdomen.

Newonset dyspepsia in a previously asymptomatic patient may indicate peptic ulcer disease (PUD), nonulcer dyspepsia (NUD), or (rarely) gastric cancer. It is distinct from gastroesophageal reflux disease (GERD), typically characterized by heartburn or acid regurgitation, although differentiation may be difficult in some patients. Symptoms of dyspepsia and GERD are experienced by about 40% of the population annually

and account for 2% to 5% of primary care visits

 and $14 billion in annual prescription drug expense.

infection as the single most common cause of PUD revolutionized management of dyspepsia; the significance of that discovery resulted in a Nobel Prize for discoverers Barry J. Marshall and J. Robin Warren.

infection be eradicated in patients with PUD and that this be accomplished using antibiotics and antisecretory medication. Several national and international gastroenterological organizations have since released more detailed guidelines for the diagnosis and treatment of dyspepsia as the understanding of the contribution of 

One of the key recommendations of various guidelines has been that clinicians use nonendoscopic tests (such as serologic or breath tests) for 

infection, unless endoscopy is clinically indicated, and that patients testing positive receive eradication therapy. Compared with early endoscopy and treatment based on endoscopic findings, this less costly test-and-treat approach has demonstrated similar outcomes such as number of physician visits and days lost to sick time.

Prompt endoscopy is reserved for older patients (≥45-55 years) and for patients with alarm features, such as persistent vomiting, bleeding, weight loss, dysphagia, and anemia. 

Clinical practice guidelines for dyspepsia have been in place for more than 10 years and have been widely disseminated to primary care clinicians.

However, little is known about the medical community's adherence to these recommendations. Surveys of physician-reported practice have been published,

of treatment patterns at the patient level have been limited to narrow populations or to analyses of drug utilization. The release of updated US guidelines in 2005

marked an appropriate time to assess current clinical practice. 

This study describes recent patterns for managing patients with dyspepsia or GERD symptoms based on analysis of a national managed care claims database. While lacking the precision of medical record review, analysis of claims data offers insight into actual patterns of community care resulting from the interaction of physician and patient behavior, as well as organizational factors. We characterize current practice and, where possible, assess the consistency of these treatment patterns with published clinical practice guidelines. 

We performed a retrospective longitudinal cohort study of nonelderly patients treated for dyspepsia or GERD from 2002 to 2004. The study was based on analysis of paid medical and pharmacy claims for more than 15 million persons enrolled in 75 commercial managed care plans across the United States (Pharmetrics, Watertown, Mass). To help ensure documentation of a patient's complete course of care, the study was limited to individuals continuously enrolled for 3 years. 

of Diseases, Ninth Revision, Clinical Modification 

Diseases, Ninth Revision, Clinical Modification 

Individuals were selected if they had paid claims for any of the following: test for 

, endoscopy with biopsy, claims that included a diagnosis for dyspepsia or GERD, or prescription for antisecretory therapy (proton pump inhibitor [PPI] or H

RA]) or for eradication therapy (antibiotics plus PPI or H

diagnosis codes were 531-534 (PUD), 535-536 (NUD), and 530 (GERD/esophagitis). Codes and drug names for inclusion criteria are given in 

). Additional clinical events were identified as potential markers for the beginning of treatment. These included endoscopy without biopsy and a general code often applied to patients who present with abdominal signs and symptoms of undetermined cause (

. The first appearance of any of these events or the previously described inclusion events defined the patient's index date. 

The study database spanned the 3-year period from October 1, 2001, to September 30, 2004. To limit the analysis to patients presenting with new-onset symptoms, the sample was restricted to patients in whom no index event appeared during the first 6 months. A minimum of 6 months' follow-up after the index event was also required. Patients were followed up until the end of the study period or until discontinuation of enrollment. 

The final analysis was limited to individuals who had at least 1 qualifying prescription or diagnostic procedure; patients with diagnosis only (no intervention) were excluded. Herein, the goal was to eliminate patients whose symptoms may have been recorded only as a rule-out code or secondary diagnosis or those whose symptoms were apparently too mild to warrant follow-up treatment or testing. 

For each patient, claims were scanned in chronological order to identify the index event and subsequent events potentially related to diagnosis and treatment of dyspepsia or GERD. Outcomes included the percentage of patients receiving specified procedures and treatments and the percentage of patients undergoing noninvasive testing or endoscopy before or after a prescription fill for eradication therapy. The date of a diagnostic test is based on the recorded date of test analysis rather than the date of test administration. Coding ambiguities prevent us from knowing if every endoscopy included an endoscopic test for 

(such as a biopsy urease test). To err in favor of compliance, we assumed that endoscopy included an 

test. Similarly, when the data reported a test and prescription fill on the same day, we assumed that the test occurred first. We also checked the effect of moving back the diagnostic test date by 2 days to allow for situations in which the physician receives test results by telephone before the date appearing on the claim. 

Each patient was categorized according to the earliest recorded disease-related diagnosis (PUD, NUD, GERD, or SIGN); if none of these appeared, the category assigned was "other." For comparison, patients were also assigned a secondary category based on the last disease-related diagnosis. Consultation with a gastroenterologist was recorded if the patient had a claim with the specialty coded as gastroenterology or if the patient underwent endoscopy. Patients and outcomes were categorized by age, sex, type of health plan, and region of the country. 

The study sample was derived by nonrandom selection of health plans of various types and regions across the country. To develop nationally representative findings of the nonelderly privately insured population, results were weighted by region and by whether the health plan was a health maintenance organization (HMO). The nationwide distribution of individuals by region and by plan type was taken from the Kaiser Family Foundation 2003 Annual Employer Health Benefits Survey,

with corroborating information from the Agency for Healthcare Research and Quality Medical Expenditure Panel Survey.

Differences (with 95% confidence intervals) in proportions of outcomes between categories of patients were analyzed. Standard errors were computed accounting for the sample weighting using SAS software version 9.1 (SAS Institute Inc, Cary, NC). 

Among more than 1.9 million continuously enrolled health plan members, 129 179 had a qualifying event and met other inclusion criteria. Another 38 334 had no qualifying therapies or tests, although they had at least 1 qualifying diagnosis. The final cohort included 3456 patients with first diagnosis recorded as PUD, 14 593 with NUD, and 36 233 with GERD. Another 44 129 patients were first given a SIGN diagnosis. Almost 31 000 other patients received none of these diagnoses but had dyspepsia or GERD-related treatments or tests. The 

gives cohort characteristics by initial diagnostic category. For most patients, the last qualifying diagnosis was the same as the first. However, half of those initially diagnosed as having PUD had a different last diagnosis. Two percent to 3% of the other groups were eventually diagnosed as having PUD, so that the total number with PUD as a final diagnosis was 3786, which was 9% larger than the initial diagnosis total. 

Ten percent of the sample were younger than 18 years, while 37% were aged 50 to 64 years (Table 1). Patients with PUD tended to be older and were more likely to be male than patients with NUD or SIGN. Two thirds of the study population were enrolled in HMOs; almost half were located in the Midwest census region (Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, Ohio, North Dakota, South Dakota, and Wisconsin). 

Although a minimum follow-up period of 6 months was required for sample inclusion, 86% actually had at least 1 year of follow-up, and 40% had at least 2 years (Table 1). Two thirds of the cohort had no prior claims for at least 12 months before their index date. 

gives the sample distribution by region and by HMO coverage, along with the distribution from a more representative national sample.

Our sample substantially underrepresents non-HMO plans in the West and South. Subsequent results presented herein are tabulations weighted to the regional and HMO composition of the national sample. 

shows the number of patients recorded as having received a particular test or treatment during the study period. Among the patients with presumed PUD, 27% were tested noninvasively for 

infection, while 54% underwent endoscopy; neither occurred for about 25% of patients. About 30% received eradication therapy. Of the patients initially suspected to have NUD, 31% had endoscopy, and 27% were tested for 

infection by other means. Eighteen percent of the patients with a first diagnosis of GERD were tested for 

infection, and 6% received eradication therapy. 

Thirty-nine percent of patients with GERD had prescription and refill claims for PPIs totaling at least a 4-month supply (Figure). Fewer patients with PUD and NUD had at least a 4-month supply of PPIs (24% and 22%, respectively). Not shown in the Figure is that about 76% of patients with PUD and NUD received at least 1 prescription for a PPI or an H

Complementary to these results is the proportion of patients who filled only a PPI or an H

RA prescription (ie, they had no endoscopy, test, or eradication therapy). This proportion was 22% for patients with presumed PUD, 44% for patients with NUD, and 53% for patients with GERD. 

These data were also analyzed with patients categorized according to the last observed diagnosis rather than the initial diagnosis. Similar results were obtained. 

The use of specific nonendoscopic tests for 

). While serologic and whole-blood antibody tests still dominated in 2004, the use of tests of active infection (urea breath tests and fecal antigen tests) increased from 7% to 13% (difference, 6% [95% confidence interval, 1%-11%]) of the tests between 2002 and 2004. 

Guidelines recommend that patients undergo a diagnostic test or endoscopy for determination of 

Among the study sample, 7863 patients received eradication therapy, but only 71% (95% confidence interval, 67%-75%) of those were pretested (

). Patients who had been seen by a gastroenterologist were more likely to have been pretested than those who had seen another type of physician (88% vs 59%). The proportion was higher for those with a presumed diagnosis (eg, 87% for PUD). Year-to-year differences in testing rates were not statistically significant. Varying the diagnostic test date or prescription fill date as recorded on the claim form by 2 days did not affect the results. 

Guidelines are mixed as to recommendations for routine posttreatment testing to document successful eradication of 

 In our study, only 14% (95% confidence interval, 11%-17%) of patients who received eradication therapy were tested after treatment. Posttreatment testing was performed more frequently in those patients who had been seen by a gastroenterologist (29% vs 3%). We also found that one third of posttreatment tests performed by primary care physicians (PCPs) were serologic tests, contrary to guidelines. The rates of pretreatment and posttreatment testing varied little by type of health plan, region, or year. 

While guidelines call for endoscopy for patients aged 50 years or older presenting with new-onset dyspepsia,

only 63% of patients aged 50 to 64 years having an initial presumptive diagnosis of PUD underwent endoscopy, compared with 40% of those initially diagnosed as having NUD and 38% of those having GERD. The proportions of adults aged 18 to 49 years undergoing endoscopy were 46%, 29%, and 27% for PUD, NUD, and GERD diagnoses, respectively; the endoscopies occurred within 30 days of their index date for 34%, 11%, and 12%, respectively, of the patients. 

Assessment of current practice is a prerequisite for improving outcomes and cost-effectiveness of patient care. As far as we can determine, this is the first study of clinical testing and treatment patterns for dyspepsia and GERD based on a large national sample of claims data. 

MANAGERIAL

Practice Patterns for Managing Helicobacter pylori Infection and Upper Gastrointestinal Symptoms

To computerize indicators of potential drug-related emergency department and hospital admissions and to report the incidence of these potential drug-related morbidities for a managed care organization. 

Retrospective review of healthcare organizations' pharmacy and administrative claims databases. 

Thirty-nine indicators were coded and were used in an automated search of claims data. The indicators of potential drug-related morbidities comprised a pattern of care and an associated adverse outcome. Poisson distribution regression analysis was performed to assess the association of patient factors with indicator positives. 

The incidence densities for indicator positives were 1.96 (95% confidence interval, 1.60-2.40) per 1000 patient-years in the general population and 13.6 (95% confidence interval, 8.8-20.2) per 1000 patient-years among older persons. Age, male sex, number of medical conditions, and number of medications from different classes were associated with an increased rate of indicator positives. 

Indicators of potential drug-related morbidities can be fully automated and used to search through medical and pharmacy claims. The indicators investigated in this study show promise as a quality improvement tool and should be further developed and evaluated. 

Healthcare quality and safety are serious public health concerns. 

Computerized indicators of potential drug-related emergency department and hospital admissions show promise as a quality improvement tool and should be further developed and evaluated.

Healthcare quality is a serious public health concern. The prevalence of drug-related admissions in the United States has been reported to be 3% to 12% of hospital admissions.

 Annual costs of drug-related problems in the United States are estimated to be more than $150 billion, and hospital admissions account for most of these costs.

has proposed a 50% reduction in medical errors. To achieve this in medications use, the following are needed: an efficient measurement technique to gather baseline information about drug-related morbidity (DRM), an assessment of system-related causes and rational interventions that target system failures, and follow-up measurements to gauge the effects of the interventions. 

The use of computerized indicators to search electronic databases for DRM has been proposed as a method that healthcare organizations can use to obtain efficient baseline and follow-up measurements.

used a Delphi approach to develop and validate a set of indicators of potential DRMs among older persons, and they used a combination of manual and computerized search criteria to identify potential DRMs through administrative data and information from the Personal Welfare Profile Senior Assessment Survey.

 the indicators were revalidated and coded to computerize the search method for use in administrative databases. 

Reducing DRM would significantly improve the safety and quality of medical care provided by a managed care organization and at the same time potentially reduce the mean per-patient costs.

The indicators used in this study were designed to link possible suboptimal patterns of care with their subsequent adverse outcomes. The objectives of this study were to computerize indicators of potential drug-related emergency department (ED) and hospital admissions and to report the incidence of these potential DRMs in a managed care organization. 

This study used a historical open cohort design and was based on a retrospective review of a healthcare purchasing coalition's claims data. Enrollees had to have at least 1 pharmacy or outpatient claim to be included in the analysis. Person-time was calculated for each individual based on his or her enrollment start and end dates during the study period from July 1999 to June 2001. Ethics approval for the study was obtained from the University of Florida Health Science Center Institutional Review Board. 

The database contained administrative data about enrollees in a preferred provider organization. The preferred provider organization administered coverage to families of employees of corporate members of the healthcare coalition. The population included patients of all ages. The data consisted of all claims made by professionals and facilities for outpatient and inpatient care, as well as data on all ambulatory prescriptions for the plan enrollees provided by the pharmacy benefit manager. A unique patient identifier was used to link the 3 data sets; patients were deidentified to protect patient confidentiality. 

Thirty-nine of 49 potential DRM indicators by Faris

were used in this study; the 10 not used were deemed outdated or impractical to automate. These indicators are believed to represent potentially preventable cases of DRM because a Delphi panel of reviewers agreed that the indicators fit the criteria for preventability.

For a DRM to be considered preventable, the following 4 elements are required: (1) the drug-therapy problem has to be recognizable, (2) the potential adverse outcome associated with the drug therapy problem (DTP) has to be foreseeable, (3) the immediate cause has to be identifiable, and (4) the immediate cause must be controllable.

translated the potential DRM scenarios into the medical event codes used to construct the computerized indicators and verified them with expert review. The list of 39 indicators used in this study is available in online 

Each indicator had the following format: a pattern of care and an associated adverse outcome (eg, the use of warfarin sodium 

prothrombin time or international normalized ratio [for anticoagulant monitoring] not evaluated every month 

hospital admission for hemorrhagic event). The pattern of care is the process component of the indicator, and it represents a possible DTP.

An ED visit or hospitalization was required to represent an adverse outcome of care. Outcome identification was restricted to the primary diagnosis because it is intended to represent the cause of an ED visit or hospital admission.

A potential DRM required the process and the outcome components of the indicator to be present (positive) in a patient. 

The indicators fit into the following 3 general categories: disease-drug interaction, drug monitoring, and drug-drug interaction. Each required different programming algorithms, which are described herein. 

The disease-drug algorithms required the presence of specific diagnosis codes and pharmacy claims before a claim for the associated outcome. A process positive was recorded when a diagnosis claim was present before or during the inferred use of the medication from pharmacy claims. A period of medications use was considered to have lasted up to 100 days after the last prescription refill date, to recognize that many patients received 90-day prescription quantities. A potential DRM required the outcome, the disease diagnosis codes, and a pharmacy claim for the medication within 100 days before the outcome. 

Drug monitoring algorithms required the presence of specific pharmacy claims and procedural codes. 

codes were used to indicate whether specified laboratory tests were conducted. Two search algorithms were required for monitoring indicators, one for the process component and another for the process and outcome simultaneously. Once the process was initiated by specific pharmacy claims, lags were calculated to represent the interval from first drug claim date to first laboratory test, from laboratory test to laboratory test, and from last drug claim date to last laboratory test. Patients were considered at risk and were included in the analysis if they were taking the medication for at least as long as the monitoring interval defined in the indicator. If any lag exceeded the defined monitoring interval, the member was recorded as having a process positive or potential DTP. 

A potential DRM for a monitoring-related indicator was recorded (1) when a patient had a pharmacy claim for the specified medication and the outcome and (2) the lag between the outcome and the laboratory test closest to the outcome exceeded the indicator-specific monitoring interval. If the lag was less than 4 days, the lag between the outcome date and the second-to-last laboratory test date was used to avoid false-negative results. The assumption is that monitoring within 4 days of hospitalization identified the impending injury and the need for emergency or hospital care. 

The third search algorithm considered multiple drugs. This included drug-drug interactions, drug-no drug situations (eg, the use of b-agonist without a disease-modifying medication in patients with asthma), and overuse of one drug and underuse of another. The process component is indicator specific. In the case of a simple drug-drug interaction in which drug A is a chronic medication and drug B is an acute medication, the process positives were recorded when a claim for drug B occurred between the first and last claims for drug A. A potential DRM was recorded when the ED or hospital claim occurred within the specified number of days for each lag. 

The number of possible DTPs per indicator was limited to 1 per individual for incidence rates and for statistical analyses. However, patients could have DTPs from multiple indicators. No limits were placed on the number of potential DRMs. Available information from the database was used to explain the indicator findings. Variables were selected based on availability in the database and on previous studies.

Age, sex, and numbers of drug classes, office visits, pharmacies, and prescribers were used as independent variables, and they were treated as continuous variables whenever possible. Poisson distribution regression analysis was used to evaluate their association with potential DRMs. Significance was set at 

were used to evaluate data quality and the overall integrity of the databases. Computerized search criteria and statistical analyses were performed using SAS version 8.2 (SAS Institute, Cary, NC). 

During the study period from July 1999 to June 2001, 37 063 members with at least 1 medical or pharmacy claim had a total observation time of 51 892 years; the mean + standard deviation length of enrollment was 511.04 + 201.73 days. Population demographics are given in 

Incidence Densities of Indicator Positives

There were 102 potential DRMs among 79 members. The incidence densities for potential DRMs were 1.96 (95% confidence interval [CI], 1.60-2.40) per 1000 patient-years in the general population and 13.6 (95% CI, 8.8-20.2) per 1000 patient-years among older persons. Twenty-one of 39 indicators had no DRM indicator positives. Of the remaining 18 indicators, the range of potential DRMs per indicator was 1 to 26. The indicator results are listed in 

In Poisson distribution regression analysis, age, male sex, number of medical conditions, and number of drug classes were significantly associated with an increased rate of potential DRMs. The number of outpatient clinic visits showed a significant negative association with the rate of potential DRM. The rate for men was 2.05 (95% CI, 1.33- 3.16) times the rate for women (

< .001). The rate increased 3% (95% CI, 1.02%-1.05%) for each 1-year increase in age (

= .001). Each additional medical condition increased the rate of potential DRMs by 15% (95% CI, 1.12%-1.19%) (

< .001). Each additional drug class increased the rate by 14% (95% CI, 1.10%-1.17%) (

< .001). Each additional outpatient physician visit decreased the rate of potential DRM by 5% (95% CI, 0.93-0.97). Although the numbers of different prescribers, and different pharmacies were significant in bivariate analysis, they were not independently associated with the rate of potential DRMs (

The incidence rate of potential DRMs from this study was lower than the incidence rates in previous studies

using indicators of potential DRMs and in studies

of preventable drug-related admissions. The lower rate found in our study population may be partially explained by variations in population demographics and by differences in search algorithms. 

However, a subgroup analysis among older persons produced results that were somewhat similar to previous findings among a Medicare managed care population.

The subgroup analysis among older persons in the present study indicated an incidence rate of 13.6 per 1000 patient-years. Faris

found an incidence proportion of 62.5 per 1000 patients, and Mackinnon and Hepler

noted an incidence proportion of 28.8 per 1000 patients. 

The rate of potential DRMs in the general population in our study was also lower (15 preventable drug-related ED visits and hospitalizations per 1000 persons [detailed in online 

]) than the findings from a systematic review of preventable drug-related admissions would suggest. These articles

used a medical record review process to identify preventable drug-related admissions, while the present study used computer screening of administrative and pharmacy data. Explicit criteria for identifying potential DRMs may not identify as many cases as the less-specific medical record review. Different methods of case ascertainment yield different results. An inpatient study

 compared yields of medical record review, automated computer screening, and stimulated voluntary report and found that each method identified different types of adverse drug events and that the computer screening method only identified about 45% of the total adverse drug events. The computer screening methods are believed to increase efficiency and reliability; however, the yield is limited to the explicit indicators used. Implicit review methods allow for more flexibility and comprehensive assessments, but the limiting step is the knowledge and thoroughness of the reviewers and not the predefined search algorithms. 

The present study required potential DRMs to have an ED visit or hospitalization for the outcome, and only the primary diagnosis was used for outcome identification; the primary diagnosis is intended to represent the cause of admission. The previous studies

of potential DRM indicators did not have the same requirements. Their search algorithms did not require ED visits or hospitalizations for more than half of the indicators, and the analyses were not limited to primary diagnosis codes to signify the reason for seeking medical treatment.

The findings from our analysis of variables associated with the rate of potential DRM show important consistencies with previous inpatient

studies. Our Poisson distributed regression analysis showed that age, male sex, number of medical conditions and medications were also found to be independently associated with DRM. Moreover, MacKinnon and Helper

found that number of medical conditions and prescription medications, along with female sex, antihypertensive drug use, and number of prescribers had a positive association with the rate of potential DRM, and Faris

confirmed these findings. When considering the results from the present analysis in the context of previous research, the consistency of findings provides some evidence to support the use of the indicators to identify DRMs. 

Our results are limited by the fact that the potential DRM indicators have not been criterion validated. Therefore, indicator positives represent possible DRMs. However, the indicators are based on accepted pharmacology and empirical studies.

 They were content validated by independent expert panels using an accepted consensus-seeking procedure. The search algorithms took into account the proper sequence and the proximity of the potential DTP relative to the adverse outcome. Furthermore, Poisson distribution regression analysis supported findings from previous studies

of DRMs. Other work demonstrates that claims data are reasonably valid for identifying patients with specific diseases, especially when pharmacy claims that would be consistent with the disease are included.

The search algorithms were limited to severe outcomes (ie, ED visits or hospitalizations). Studies

show that many DRMs are considered significant but not severe. The incidence rates likely underestimate the true DRM rate, which would include less severe outcomes. In our study, clinic visits were not included in outcomes assessments because of ambiguities in determining the cause and severity of an ambulatory visit using claims data. 

Claims-based surveillance for potential DRMs could be enhanced by searching problem lists and clinical notes from electronic medical records. However, many clinics still do not have a computerized patient medical record; therefore, this approach is not universally feasible. Because some data in an electronic medical record are in free text rather than being coded, text-search algorithms or natural language processors would be needed to identify meaningful terms and concepts. Searching free text would support the identification of less severe DRMs (ie, adverse outcomes that were identified and resolved in the clinic), and it would provide a more comprehensive assessment of DRM. 

Further work is needed to refine the technical aspects of the potential DRM indicators. This would involve continuously updating the indicators to include newer therapies and more recent treatment evidence, evaluating their criterion validity with chart review, and adapting the search algorithms to include clinical data (eg, laboratory test results and problem lists) along with claims data. Careful and meaningful analysis of the recent literature is needed to add additional performance indicators and to remove indicators that do not represent current best practice. Institutions and investigators should conduct an internal review of available indicators, such as the potential DRM indicators presented herein, to determine the usefulness of each indicator and to develop indicators that fit the needs of their specific institutions. 

Indicators of DRMs can be automated and used to search administrative databases for cases of potential DRMs. In our study, the incidence densities for potential DRMs were 1.96 (95% CI, 1.60-2.40) per 1000 patient-years in the general population and 13.6 (95% CI, 8.8-20.2) per 1000 patient-years among older persons. Age, male sex, number of medical conditions, and number of medications from different classes were associated with an increased rate of potential DRMs. The information generated from the potential DRM indicators is the first step in a multistep process. Ongoing steps include identifying system problems, implementing solutions, and evaluating the effects of interventions. These indicators show promise as a quality assessment and as an improvement tool and should be further developed and evaluated. 

From the University of Florida College of Pharmacy, Gainesville (BCS, CDH), and Florida Health Care Coalition, Orlando (BC); and Emma Eccles Jones Nursing Research Center, College of Nursing, The University of Utah, Salt Lake City (JW). Dr Sauer is now medical informatics fellow at the Veterans Affairs Medical Center, Salt Lake City, Utah. 

This study was funded by the P. A. Foote Small Grants Program in Health Outcomes and Pharmacoeconomics Research from the University of Florida College of Pharmacy. 

January 2007

Reassembling Our View of Vulnerable Populations

Efforts to Reduce Racial Disparities in Medicare Managed Care Must Consider the Disproportionate Effects of Geography

Practice Patterns for Managing Helicobacter pylori Infection and Upper Gastrointestinal Symptoms

Computerized Indicators of Potential Drug-related Emergency Department and Hospital Admissions

Productivity Enhancement for Primary Care Providers Using Multicondition Care Management

Transition to Medicare Part D: An Early Snapshot of Barriers Experienced by Younger Dual Eligibles With Disabilities