Quality of care for gout in the US needs improvement†
The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs.
To examine evidence-based quality indicators (QIs) in US veterans with gout diagnosis, and to examine the effect of demographics, heath care utilization/access, comorbid conditions, or physican characteristics as predictors of quality of gout care.
Using the Minneapolis Veterans Affairs electronic medical record system, we identified a cohort of veterans receiving medication to treat gout between January 1, 1999 and December 31, 2003, and evaluated 3 recently published evidence-based QIs for gout management: QI 1 = allopurinol dose <300 mg in gout patients with renal insufficiency, QI 2 = uric acid check within 6 months of starting a new allopurinol prescription, and QI 3 = complete blood count and creatine kinase check every 6 months for gout patients receiving prolonged colchicine therapy. We calculated the proportion of patients whose therapy adhered to each QI and to all applicable indicators (overall physician adherence). Logistic regression analysis examined association of overall physician adherence with sociodemographics, health care utilization, comorbidity, and provider characteristics.
Of 3,658 patients with a diagnosis of gout, 663 patients qualified for examination of ≥1 QI. Of these 663 patients, therapy in only 144 (22%) adhered to all applicable QIs; 59 (78%) of 76 adhered to QI 1, 155 (24%) of 643 adhered to QI 2, and 18 (35%) of 52 adhered to QI 3. Overall physician adherence to QIs was significantly lower in older veterans and in those with more inpatient visits per year, but was higher in those with more primary care visits or more health care providers.
Suboptimal physician adherence to QIs was seen for all 3 QIs tested in this cohort of veterans with gout. These findings can guide quality improvement efforts.
Gout is a common disease causing significant morbidity. Its self-reported prevalence in the US is ∼2% (1, 2) and the prevalence increases with age, reaching 9% in men and 6% in women older than 80 years of age (2). Prevalence in the US has increased recently (1, 3), nearly doubling between 1990 and 1999 in a study of health maintenance organization enrollees (3). The increase is thought to be partially due to aging of the population and to the increasing prevalence of renal failure, hypertension, diuretic therapy, and obesity (4, 5).
Treating gout reduces morbidity, but the extent to which patients receive effective high-quality care is unknown. Low adherence to quality indicators for treating various diseases has been found in many US health care systems (6, 7), particularly for treatment of musculoskeletal diseases including rheumatoid arthritis (8), acute low back pain (6), and osteoarthritis (9), with quality scores ranging from 54% to 65%. MacLean et al (8) studied patients with rheumatoid arthritis enrolled in a nationwide US insurance company and reported quality scores of 62% for arthritis care, 52% for comorbid disease care, and 42% for health care maintenance. Higher scores were noted in patients who received relevant specialist or primary care compared with those who did not (8).
There is widespread evidence of undertreatment (10), overtreatment (10, 11), and inappropriate medication use in patients with gout (12–14). Mikuls et al recently published evidence-based quality of care indicators for gout management (15). These quality indicators (QIs) are minimum acceptable standards of care that have been directly or indirectly linked to outcomes in published clinical studies, and that are believed by the experts to produce good outcomes. Mikuls et al examined 3 QIs in the UK General Practitioner Research Database (UKGPRD) and found low (25–57%) physician adherence (16). Studies in US health plans have found gaps in quality care for gout (17, 18). It is unknown whether care provided to US veterans with gout meets these minimum quality standards.
The Department of Veterans Affairs (VA) is the largest health care delivery system in the US, providing care to 5.2 million veterans with an annual medical care budget of more than $30 billion (19). Although no formal gout prevalence studies in US veterans have been reported, gout is probably at least as common in veterans as in the general population, because veterans are older, more likely to be male, and have greater comorbidity (20). We used VA electronic medical records to assess 1) in veterans receiving care at the Minneapolis VA Medical Center, if therapy met the standard of care as measured by physician adherence to QIs, and 2) if demographic characteristics, health care utilization/access, comorbidities, or physician characteristics are associated with physician adherence to QIs.
PATIENTS AND METHODS
We identified all patients who received an International Classification of Diseases, Ninth Revision (ICD-9) code for gout at the Minneapolis VA Medical Center between January 1, 1999 and December 31, 2003 by searching the purpose-of-visit file and medical problem list files in the local Veterans Information System Technology Automation computerized databases. The purpose-of-visit file includes diagnoses coded at each patient encounter; the medical problem list includes diagnoses from the computerized patient clinical chart. We searched for the following ICD-9 codes and the earliest date when each was recorded: gouty arthropathy (274.0); gout, unspecified (274.9); gout with other specified manifestations (274.89); tophaceous gout of ear (274.81) or other sites (274.82); gouty nephropathy (274.19, 274.10); and uric acid nephrolithiasis (274.11). We assessed the accuracy of these ICD-9 codes in a random sample from a separate patient cohort seen at the Minneapolis VA rheumatology clinic between January 1, 2001 and July 31, 2002; of 23 patients both with and without a gout ICD-9 code, the ICD-9 codes had sensitivity of 90%, specificity of 100%, and positive and negative predictive values of 100% and 87%, respectively. Although the accuracy of gout ICD-9 codes may differ slightly when considering all clinics, we doubt results will be dramatically different. The Minneapolis VA Medical Center's Human Studies Committee approved the study, which conformed to the VA Privacy Act and the Health Insurance Portability and Accountability Act.
We also obtained this cohort's demographic, health care utilization, comorbidity, laboratory, and pharmacy data for the same period (January 1, 1999 to December 31, 2003). Demographic information including age, sex, and race was obtained from VA administrative data (patient treatment and outpatient files). Race data were available for only 47% of the patients. These data sets capture all visits at VA facilities with primary and secondary diagnoses related to the visit by specialty. We extracted the number of outpatient visits and inpatient admissions, and for each of the latter we determined whether gout was the primary diagnosis. Each outpatient visit was categorized as primary care, rheumatology, or other. We obtained the number of days with any outpatient visits, since patients may have one or more clinic visits on the same day (total outpatient visit days/year). Charlson Comorbidity Index score was calculated by aggregating diagnoses from all visits (21). For each veteran, we obtained the percent service connection, which ranges from 0% to 100%. Veterans receive service connection for disabilities related to disease conditions that occur during or due to active duty service. Finally, we used the means test as an indicator of socioeconomic status, categorizing veterans as most needy but not service connected, most in need of health care but service connected, and not most needy. For the most frequent provider of gout care, we obtained information on sex, age in 2000, and type of provider (rheumatology, nonrheumatology, or neither, usually nurse/nurse practitioner).
Laboratory data included blood hematocrit, white blood cell count, platelet count, creatine kinase (CK), serum uric acid, serum creatinine, and creatinine clearance. For patients without measured creatinine clearance, we calculated this information from the most recent height, weight, and age before each 90-day prescription using the Cockcroft-Gault calculation (22). We examined creatinine and/or creatinine clearance immediately preceding each 90-day prescription, because a physician is most likely to examine these to make dose adjustments.
A research pharmacist (JPT) searched the local VA pharmacy database for filled prescriptions for gout-specific medications (allopurinol, colchicine, and probenecid) occurring during the study period (January 1, 1999 to December 31, 2003). We only analyzed prescription data if ≥3 consecutive months of prescription supply had been dispensed. For each prescription, we retrieved information on drug strength, quantity, number of days supplied, original fill date, number of refills, and refill dates. The medication and laboratory files were merged with the diagnoses using social security numbers.
Using the merged file of diagnoses and medications, we identified a cohort of patients with gout diagnoses who filled a prescription for at least 1 gout medication for at least 3 months. This constituted the cohort for examination of physician adherence to the QIs. If a patient had >1 gout diagnosis, we used the earliest such date as the date of gout diagnosis. For all analyses, veterans were included only if, over the entire study period (including the index prescription year), they averaged ≥2 outpatient visits per year to a VA clinic, to ensure that VA clinics had assumed responsibility for their care. Of 715 patients eligible for determination of ≥1 QI, 52 were excluded for having <2 visits per year, leaving 663 patients eligible for examination of QIs.
Process measures (i.e., QIs) must be linked to outcomes to be valid, but direct measurement of outcomes is not always needed when evaluating quality of care using valid process indicators. Published gout treatment QIs have been linked to outcomes based on retrospective and prospective studies (15). Of the 10 QIs described by Mikuls et al (15), 7 pertain to uric acid–lowering therapy (allopurinol and probenecid), 2 to antiinflammatory medications (colchicine and nonsteroidal antiinflammatory drugs [NSAIDs]), and 1 to lifestyle modification. We focused on QIs related to medication management. Because we lacked radiographic data, details of diagnoses other than gout, and data about over-the-counter NSAIDs and other non–gout-specific medications, we could only examine the following 3 QIs: QI 1 indicates that if a patient with gout has significant renal impairment (serum creatinine ≥2 mg/dl or measured/estimated creatinine clearance ≤50 ml/minute), then the initial daily allopurinol dosage should be <300 mg/day to avoid risk of toxicity; QI 2 indicates that if a patient with gout is given a prescription for xanthine oxidase inhibitor, then serum urate level should be checked at least once during the first 6 months of continued use, such levels being necessary for appropriate dose adjustments; and QI 3 indicates that if a patient with gout receives long-term prophylactic oral colchicine (≥0.5 mg/day for ≥6 months) and has significant renal impairment (serum creatinine ≥2 mg/dl or measured/estimated creatinine clearance ≤50 ml/minute), then complete blood cell count (CBC) and CK should be checked at least every 6 months of continued use, because risk of colchicine-related myopathy and myelosuppression appears to increase when renal function is reduced. The numerators and denominators for the 3 QIs were as follows. For QI 1, the denominator was the number of patients receiving an initial prescription for allopurinol who also had significant renal impairment, and the numerator was the number of patients among those included in the denominator with an initial daily allopurinol dosage <300 mg/day; for QI 2, the denominator was the number of patients receiving an initial prescription for allopurinol, and the numerator was the number of patients among those in the denominator who had at least 1 urate level check during the first 6 months of allopurinol use; and for QI 3, the denominator was the number of patients receiving long-term prophylactic oral colchicine who also had significant renal impairment, and the numerator was the number of patients among those in the denominator who had CBC and CK checked for every 6 months of continued colchicine use.
Definitions of medication use, renal insufficiency, and laboratory test monitoring.
The following were defined a priori.
New medication prescription.
A gout medication prescription was considered new if the patient had no prescription for the same medication in the previous 180 days, if the prescription was filled no more than 31 days before the earliest gout diagnosis, and if the prescription was filled on or before July 1, 2003, and for allopurinol the new prescription had to be 180 days or longer. We required no prescriptions in the preceding 180 days to exclude patients ordering late requests for refills, and to make our selection more specific. Prescriptions filled after July 1, 2003 were not included because QIs 2 and 3 required at least a 6-month followup.
A patient with the most recent creatinine ≥2 mg or creatinine clearance ≤50 ml/minute was considered to have renal insufficiency.
For QI 3, laboratory monitoring was recommended every 6 months. Because we wanted to be specific rather than sensitive, we considered laboratory monitoring adequate as long as no more than three-quarters of a year (270 days) elapsed between the prescription and the first laboratory examination or between laboratory examinations.
Overall physician adherence was calculated as the percentage of eligible patients for whom all applicable QIs were satisfied. To determine factors associated with physician adherence, we performed multivariable logistic regression with overall physician adherence as a dichotomous outcome (did versus did not satisfy all applicable QIs). We examined the following patient covariates: race; age on January 1, 1999; inpatient stays per year; inpatient stays per year with gout as the primary diagnosis; primary care, rheumatology, and other outpatient visits per year; percent service connection and Charlson Comorbidity Index during the followup period; and age, sex, and specialty type (rheumatologist, nonrheumatologist, other) of the most frequent gout care provider. Variables that were significant in the univariate analyses were entered in the multivariate model. For service connection and Charlson Comorbidity Index, we used the maximum value if the period examined had ≥2 values. We did not consider patient sex because the cohort included only 4 women (<1%). The means test was highly correlated with percent service connection and total outpatient visit days/year with other outpatient visits/year; therefore we excluded the means test and total outpatient visit days/year from the logistic regression model to avoid collinearity.
A total of 3,658 unique patients had an ICD-9 gout diagnosis: 3,588 (98%) with gouty arthropathy, 111 (3%) with tophaceous gout, and 73 (2%) with gout with renal involvement. Of these, 2,061 (56%) received 1 of the 3 medications: 1,702 (83%) received allopurinol, 871 (42%) received colchicine, and 94 (5%) received probenecid.
A total of 663 patients who were receiving allopurinol and/or colchicine and were eligible for testing at least 1 QI constituted the quality analytic data set. These patients had a mean ± SD age of 67.9 ± 9.7 years, 99% were men, and among those with known race, 96% were white. Demographic and clinical characteristics of the larger cohort of 2,061 patients and of the QI-eligible cohort are shown in Table 1.
Table 1. Baseline cohort characteristics*
|Age in years||67.8 ± 10.5||67.9 ± 9.7|
|Male sex, %||99.3||99.4|
|Race, %|| || |
|Charlson Comorbidity Index (maximum)||2.4 ± 2.2||2.5 ± 2.3|
|Inpatient visits/year||0.37 ± 0.73||0.34 ± 0.61|
|Inpatient visits/year with gout as primary diagnosis||0.077 ± 0.22||0.077 ± 0.18|
|Primary care visits/year||3.26 ± 2.44||3.60 ± 2.67|
|Rheumatology visits/year||0.12 ± 0.45||0.15 ± 0.52|
|Other outpatient visits/year||12.92 ± 13.24||13.35 ± 12.49|
|Total outpatient visit days/year†||14.08 ± 13.33||14.6 ± 12.6|
|Means test, %‡|| || |
|Percent service connected (maximum)§||17.9 ± 30.4||18.4 ± 30.7|
The number of patients receiving high-quality care according to the 3 QIs is shown in Table 2. Overall physician adherence was very low, with only 144 (22%) of 663 patients receiving high-quality care, i.e., meeting all QIs for which they were eligible. Of 76 patients with creatinine ≥2 mg/dl, 59 (78%) were prescribed <300 mg/day of allopurinol (adherent to QI 1), i.e., 22% received doses higher than recommended given their renal function. Of greater concern, only 24% of the 643 patients starting allopurinol had their uric acid checked within 6 months of initiating allopurinol therapy (adherent to QI 2). Only 35% of the 52 patients with creatinine levels ≥2 mg/dl, who were prescribed daily colchicine for ≥6 months, had both CBC and CK checked every 6 months (adherent to QI 3).
Table 2. Prevalence of physician adherence to 3 quality indicators (QIs) of gout management*
|QI 1: lowering of allopurinol dose to <300 mg/day in patients with gout with renal insufficiency||59/76||78 (68–87)|
|QI 2: uric acid check in patients with gout within 6 months after allopurinol initiation||155/643||24 (21–27)|
|QI 3: CBC and CK check every 6 months in patients with gout and renal insufficiency who are receiving prophylactic therapy with colchicine for ≥6 months||18/52||35 (22–48)|
|All applicable QIs satisfied||144/663||22 (19–25)|
The clinical characteristics of patients whose therapy did meet each QI versus those whose therapy did not are summarized in Table 3. In the logistic regression model with overall physician adherence to QIs as the outcome (Table 4), nonwhite race (other), younger age, fewer inpatient admissions (with or without a primary diagnosis of gout), more primary care visits, and more providers were significantly associated with higher overall physician adherence. An increase in age of 1 SD (9.7 years) was associated with 22% lower odds of overall physician adherence to QIs (odds ratio 0.78). Increasing the number of hospitalizations by 1 SD (0.61 hospitalizations per year) reduced the odds of overall physician adherence by 43%. In contrast, increasing the number of outpatient primary care visits by 1 SD (2.7 visits per year) increased the odds of overall physician adherence by 28%. Increasing the number of providers by 1 SD (1.8 providers) increased the odds of meeting all quality criteria by 69%. This association analysis did not allow us to determine whether this protective effect occurred because shared care tends to have higher quality or because more eyes on the problem creates more opportunities to provide better quality. Outpatient rheumatology visits per year, other outpatient visits per year, Charlson Comorbidity Index, percent service connection, and most frequent provider's age, sex, or specialty type were not associated with adherence.
Table 3. Clinical and demographic features of patients eligible for quality indicator (QI) testing*
|Race, %|| || || || || || |
| Primary care||4.3||3.8||4.3‡||3.3||4.5||5.0|
| Other clinics||19.0||15.5||17.5‡||11.5||24.8||27.0|
|Inpatient stays/year|| || || || || || |
| Gout 1° diagnosis||0.02||0||0.01||0.01||0.02||0.04|
|Charlson Comorbidity Index||4.1||2.9||3.0‡||2.4||4.4||4.4|
|Percent service connected§||24.7||14.7||25.2‡||15.2||28.3||39.1|
|Means test, %¶|| || || || || || |
|Type of gout|| || || || || || |
| Gouty arthritis||98.5||100.0||99.4||99.6||100.0||100.0|
| Tophaceous gout||16.2||0||11.6||4.7||5.6||26.5|
| Gout, renal involvement||1.5||0||0.7||1.2||0||0|
Table 4. Predictors of odds of overall physician adherence to quality indicators from a multivariable logistic regression analysis*
|Demographic characteristics|| || || |
| Age on January 1, 1999||0.78‡||0.64–0.96‡||0.0207‡|
| Race|| || || |
| Other vs white||1.41‡||0.52–3.84‡||0.0353‡|
| Unknown vs white||0.63‡||0.36–1.10‡|| |
|Comorbidity and utilization characteristics|| || || |
| Charlson Comorbidity Index§||1.26||0.97–1.63||0.0805|
| Inpatient stays/year with gout as primary diagnosis||0.71‡||0.52–0.97‡||0.0151‡|
| Inpatient stays/year||0.57‡||0.40–0.81‡||0.0006‡|
| Primary care visits/year||1.28‡||1.02–1.62‡||0.0367‡|
| Rheumatology visits/year||0.97||0.65–1.45||0.88|
| Other outpatient visits/year||1.24||0.93–1.67||0.15|
| Percent service connection§||1.07||0.86–1.34||0.52|
| Number of health care providers||1.69‡||1.32–2.15‡||<0.0001‡|
We found deficiencies in the quality of care provided to veterans with gout diagnoses, as assessed by 3 indicators of medication dosing and monitoring. Deficiencies were most prominent in laboratory monitoring for efficacy and side effects, but were also remarkable for dose adjustment for renal failure. These results changed little using a more conservative definition of gout diagnosis, i.e., a patient received at least 2 gout diagnoses during the study period (or during/before the study). Considering provider characteristics (clinic type and, for the most frequent provider, provider type, age, and sex) also had little effect on the results. We believe this is the first study to examine evidence-based quality indicators in US veterans with gout diagnosis and to examine the physician characteristics as predictors of quality of gout care. These estimates likely understate the problem for the US as a whole, because recent studies demonstrate that the VA outperforms other systems in chronic disease management (9, 23). These results confirm the findings by MacLean et al (8) of gaps in care quality for patients with rheumatic diseases, and extend their findings to include patients with gout.
Results in this US VA population are similar to results from the UK. The overall rate of 22% physician adherence to all 3 QIs is similar to results from the UKGPRD study that also examined 3 QIs. The QIs in the UK study included allopurinol use in patients with asymptomatic hyperuricemia, allopurinol dosing in renal failure, and concomitant allopurinol use in patients receiving azathioprine or 6-mercaptopurine (16). Our finding that 22% of patients with renal insufficiency received higher-thanrecommended allopurinol doses confirms the analogous rate of 25% in the UK study. Similarly, we found that 24% of patients received uric acid monitoring within 6 months of new allopurinol prescription, and 2 studies in managed care plans found analogous rates of 37% (17) and 13% (18). The former study found that women were more likely than men to receive serum urate monitoring (17), implying that quality of gout care may vary by sex.
Our quality scores, ranging from 24% to 78%, cover the 62% adherence reported by MacLean et al in their study of patients with rheumatoid arthritis enrolled in a US health insurance plan between 1991 and 1995 (8), although in our study physician adherence was lower for 2 of 3 QIs (QI 2 and QI 3). This difference between our study and the study by MacLean et al in overall physician adherence to quality care may be secondary to differences in disease condition (gout versus rheumatoid arthritis), patient population (99% men, mean age 68 years versus 76% women, mean age 52 years), different time period (1999–2004 versus 1991–1995), likely greater comorbidity in our veteran cohort compared with the general US population (20), and differences in health care delivery structure (VA versus health insurance). However, low physician adherence to QIs for various rheumatic conditions underscores the great need to improve care for these chronic disabling conditions, which lead to huge morbidity and disability in the US population (24).
We cannot generalize our finding to non-VA patients because our sample comprised almost all men and tended to be older than the general population. However, it is unlikely that quality of gout care is much worse in VA than non-VA settings, because many studies have found better quality care for patients in the veteran health administration than patients in a national sample (9) or those covered by Medicare (23). Nonetheless, studies of physician adherence to gout QIs in the general US population are needed.
Our finding of low physician adherence to monitoring CBC and CK in patients with renal insufficiency taking colchicine adds to the current literature, because to our knowledge this QI has not been examined in any population. The widespread but variable nonadherence to gout QIs suggests room for improvement in many aspects of gout therapy, but there may be greater opportunity in some QIs than others.
Our study and other reports of poor quality care for patients with gout should motivate policy makers to design system-wide strategies for improving gout care. An example would be a computer system intervention to automatically order laboratory monitoring for each new allopurinol or colchicine prescription, and pharmacy reminders or autocorrection for inappropriate dosing of allopurinol and other medications. Additional interventions could include increasing patients' awareness of the disease and its management, increased patient involvement in their gout care (similar to patients asking physicians about their cholesterol level), or physician behavior change using multifaceted interventions (education, thought leaders, team approach) and performance-based incentives.
It is of concern that 22% of our patients with renal insufficiency received higher-than-recommended allopurinol doses, similar to the 25% noncompliance rate found in an earlier study (16). Two previous studies reported inappropriately high allopurinol doses in patients with renal insufficiency, 47% in elderly adults (10) and 39% in adults (11). Although the rate of inappropriate dosing was slightly lower in our study, most physicians would agree this is not acceptable. Our study differs from the previous studies in that we examined a different patient population, i.e., our study and an earlier study (11) included all ages, whereas a third study (10) included only hospitalized patients over 65 years of age; we evaluated a slightly different validated outcome; and we had a larger sample size with longer followup duration (663 patients in our QI-eligible group cohort were followed >5 years versus 46 or 73 patients followed for 4 or 22 weeks, respectively [10, 11]). Our findings confirm these earlier observations and extend them to include a larger population.
A study of quality of care in the US for chronic medical conditions (asthma, congestive heart failure, and diabetes) found that quality care was provided in 56% of examined instances, varying among the condition (6). Our study and the study by Mikuls et al (16) extend this observation of suboptimal care to include gout patients using evidence-based quality indicators.
Our study's strengths include a large sample size, use of computerized patient records, and ability to obtain all VA prescriptions. We analyzed an individual as a unit (adherent or nonadherent) rather than analyzing instances of physician behavior, because the latter disregards the correlation of repeated observations in the same patient. Our study has several limitations. We could not examine all gout quality indicators, and physician adherence to QIs may vary. However, we doubt that results would differ dramatically, because our results for these 3 indicators were not dramatically different from results in previous studies using these indicators (16). We obtained gout diagnoses from administrative and clinical databases and it may be argued that these are not as accurate as medical chart review. However, these databases have been found reliable for demographics and most common diagnoses (25) and valid for specific diagnoses (26, 27). Our observation of high specificity and positive predictive value of gout ICD-9 codes in a separate subsample indicates that gout ICD-9 codes in particular are accurate in these Minneapolis databases. Because our gout prevalence estimates of 5% are similar to the 4% in men from the Third National Health and Nutrition Examination Survey (NHANES-III) of the US general population, our findings may be generalizable at least to older men and our database diagnoses are likely to have accuracy similar to diagnoses in NHANES-III. We cannot account for non-VA prescriptions and diagnoses in non-VA settings. However, because we confined our analyses to veterans with ≥2 visits per year to the VA, the VA can reasonably be held accountable for these patients' quality of care, regardless of care received in non-VA settings. The low number of women and the frequently missing race information prevented us from testing for race- or sex-related disparities in care quality, and this needs further study. Further study is also needed to determine whether these findings are generalizable to other VA networks, because VA networks differ. Despite a large overall sample size and long study period, a smaller number of patients qualified for QI 1 and QI 3, resulting in larger confidence intervals for physician adherence rates. Another limitation is that we did not measure medication compliance, and some patients who filled a qualifying prescription may not have fully complied with the accompanying instructions.
In conclusion, we found suboptimal physician adherence to QIs for gout care in a cohort of veterans with gout diagnosis. Frequent lack of physician adherence to these QIs is problematic and identifies gout treatment and management as an area requiring intervention. Lower physician adherence to quality of care in older patients and in those with lower outpatient and higher inpatient utilization identifies groups at higher risk, which can guide future interventions to improve quality of care. Because of its sophisticated computerized medical record system and leadership in quality improvement, the VA is uniquely positioned to launch a quality improvement initiative for gout care.
Dr. Singh had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Singh, Hodges, Asch.
Acquisition of data. Hodges, Toscano.
Analysis and interpretation of data. Singh, Hodges, Asch.
Manuscript preparation. Singh, Hodges, Toscano, Asch.
Statistical analysis. Singh, Hodges, Asch.
We thank Tim McKenna (Information Resources Management) and Joanne Thomas (Data Analyst Group, Center for Chronic Disease Outcomes Center) of the Minneapolis VA Medical Center, Minneapolis, MN for performing computerized searches.