Association of ACEI/ARB and statin prescribing patterns with mortality after Transcatheter Aortic Valve Replacement (TAVR): Findings from real-world claims data.

BACKGROUND: Transcatheter aortic valve replacement (TAVR) has become the standard of care for most patients with severe aortic stenosis (AS), but the impact of medical therapy prescribing patterns on post-TAVR patients has not been thoroughly investigated. METHODS: We analyzed Optum claims data from 9,012 adults who received TAVR for AS (January 2014-December 2018). Pharmacy claims data were used to identify patients who filled ACEI/ARB and/or statin prescriptions during the study's 90-day landmark period post-TAVR. Kaplan-Meier and adjusted Cox Proportional Hazards models were used to evaluate the association of prescribing patterns with mortality during the 3-year follow-up period. Subgroup analyses were performed to examine the impact of 11 potential confounders on the observed associations. RESULTS: A significantly lower adjusted 3-year mortality was observed for patients with post-TAVR prescription for ACEI/ARBs (hazard ratio [HR] = 0.82, 95% confidence interval [CI] 0.74-0.91, P = .0003) and statins (HR = 0.85, 95% CI 0.77-0.94, P = .0018) compared to patients who did not fill prescriptions for these medications post-TAVR. Subgroup analyses revealed that the survival benefit associated with ACEI/ARB prescription was not affected by any of the potential confounding variables, except preoperative ACEI/ARB prescription was associated with significantly lower risk of mortality vs postoperative prescription only. No other subgroup variables had significant interactions associated with survival benefits, including preoperative use of statins. CONCLUSIONS: In this large-scale, real-world analysis of patients undergoing TAVR, the prescription of ACEI/ARB and statins was associated with a significantly lower risk of mortality at 3-years, especially in those where the medications were initiated preoperatively.

Outpatient visits before and after Lyme disease diagnosis in a Maryland employer-based health plan.

BACKGROUND: Insurance claims data have been used to inform an understanding of Lyme disease epidemiology and cost of care, however few such studies have incorporated post-treatment symptoms following diagnosis. Using longitudinal data from a private, employer-based health plan in an endemic US state, we compared outpatient care utilization pre- and post-Lyme disease diagnosis. We hypothesized that utilization would be higher in the post-diagnosis period, and that temporal trends would differ by age and gender. METHODS: Members with Lyme disease were required to have both a corresponding ICD-9 code and a fill of an antibiotic indicated for treatment of the infection within 30 days of diagnosis. A 2-year 'pre- diagnosis' period and a 2-year 'post-diagnosis period' were centered around the diagnosis month. Lyme disease-relevant outpatient care visits were defined as specific primary care, specialty care, or urgent care visits. Descriptive statistics examined visits during these pre- and post-diagnosis periods, and the association between these periods and the number of visits was explored using generalized linear mixed effects models adjusting for age, season of the year, and gender. RESULTS: The rate of outpatient visits increased 26% from the pre to the post-Lyme disease diagnosis periods among our 317-member sample (rate ratio = 1.26 [1.18, 1.36], p < 0.001). Descriptively, care utilization increases appeared to persist across months in the post-diagnosis period. Women's care utilization increased by 36% (1.36 [1.24, 1.50], p < 0.001), a significantly higher increase than the 14% increase found among men (1.14 [1.02, 1.27], p = 0.017). This gender difference was mainly driven by adult members. We found a borderline significant 17% increase in visits for children < 18 years, (1.17 [0.99, 1.38], p = 0.068), and a 31% increase for adults ≥ 18 years (1.31 [1.21, 1.42], p < 0.001). CONCLUSIONS: Although modest at the population level, the statistically significant increases in post-Lyme diagnosis outpatient care we observed were persistent and unevenly distributed across demographic and place of service categories. As Lyme disease cases continue to grow, so will the cumulative prevalence of persistent symptoms after treatment. Therefore, it will be important to confirm these findings and understand their significance for care utilization and cost, particularly against the backdrop of other post-acute infectious syndromes.

Incidence of Lyme Disease Diagnosis in a Maryland Medicaid Population, 2004-2011.

The epidemiology of Lyme disease has been examined utilizing insurance claims from privately insured individuals; however, it is unknown whether reported patterns vary among the publicly insured. We examined trends in incidence rates of first Lyme disease diagnosis among 384,652 Maryland Medicaid recipients enrolled from July 2004 to June 2011. Age-, sex-, county-, season-, and year-specific incidence rates were calculated, and mixed-effects multiple logistic regression models were used to study the relationship between Lyme disease diagnosis and these variables. The incidence rate in our sample was 97.65 cases per 100,000 person-years (95% confidence interval (CI): 91.53, 104.06), and there was a 13% average annual increase in the odds of a Lyme disease diagnosis (odds ratio = 1.13, 95% CI: 1.09, 1.17; P < 0.001). Incidence rates for males and females were not significantly different, though males were significantly more likely to be diagnosed during high-season months (relative risk (RR) = 1.24, 95% CI: 1.06, 1.44) and less likely to be diagnosed during low-season months (RR = 0.63, 95% CI: 0.46, 0.87) than females. Additionally, adults were significantly more likely than children to be diagnosed during low-season months (RR = 1.59, 95% CI: 1.19, 2.12). While relatively rare in this study sample, Lyme disease diagnoses do occur in a Medicaid population in a Lyme-endemic state.

Going Beyond Clinical Care to Reduce Health Care Spending: Findings From the J-CHiP Community-based Population Health Management Program Evaluation.

BACKGROUND: Addressing both clinical and nonclinical determinants of health is essential for improving population health outcomes. In 2012, the Johns Hopkins Community Health Partnership (J-CHiP) implemented innovative population health management programs across acute and community environments. The community-based program involved multidisciplinary teams [ie, physicians, care managers (CM), health behavior specialists (HBS), community health workers, neighborhood navigators] and collaboration with community-based organizations to address social determinants. OBJECTIVES: To report the impact of a community-based program on cost and utilization from 2011 to 2016. DESIGN: Difference-in-difference estimates were calculated for an inclusive cohort of J-CHiP participants and matched nonparticipants. The analysis was replicated for participants with a CM and/or HBS to estimate the differential impact with more intensive program services. SUBJECTS: A total of 3268 high-risk Medicaid and Medicare beneficiaries (1634 total J-CHiP participants, 1365 with CM and 678 with HBS). OUTCOME MEASURES: Paid costs and counts of emergency department visits, admissions, and readmissions per member per year. RESULTS: For Medicaid, costs were almost $1200 per member per year lower for participants as a whole, $2000 lower for those with an HBS, and $3000 lower for those with a CM; hospital admission and readmission rates were 9%-26% lower for those with a CM and/or HBS. For Medicare, costs were lower (-$476), but utilization was similar or higher than nonparticipants. None of the observed Medicaid or Medicare differences were statistically significant. CONCLUSIONS: Although not statistically significant, the results indicate a promising innovation for Medicaid beneficiaries. For Medicare, the impact was negligible, indicating the need for further program modification.

3-Year Outcomes for Degenerative Mitral Regurgitation Repair in a Medicare Population.

OBJECTIVE: Mitral valve repair (MVr) has become the standard therapy for degenerative mitral regurgitation (DMR), but real-world late mortality, reintervention, and readmission data are lacking. This study estimates MVr outcomes for DMR to 3 years in the Medicare fee-for-service population. METHODS: There were 4,219 DMR patients older than 65 years undergoing MVr within the Medicare 100% standard analytic file from October 2015 to December 2018 who were evaluated. Outcomes were analyzed for isolated MVr patients (n = 2,433) and patients undergoing MVr with certain concomitant procedures: MVr + tricuspid valve surgery (TVS; n = 619), MVr + cardiac ablation (CA; n = 540), and MVr + left atrial appendage closure (n = 627). Outcomes over a 3-year period included all-cause mortality, reintervention, rehospitalization, and common complications. All outcomes were modeled with adjustments for patient demographics and comorbid conditions. RESULTS: The average age for all patients was 71.9 ± 5.2 years. Adjusted all-cause mortality and MV reintervention (surgery or transcatheter) at 3 years for the primary cohort of isolated MVr was 3.5% and 1.6%, respectively. Directionally higher mortality at 3 years was observed in patients with concomitant TVS or CA. All-cause readmission and cardiac readmission for isolated MVr was 37.0% and 14.1%, with the highest rates for those with concomitant TVS or CA. Acute kidney injury and stroke/transient ischemic attack were the most common adverse events over 3 years for all patients. CONCLUSIONS: The 3-year mortality and reintervention rates in Medicare patients undergoing degenerative MVr are low. Those undergoing concomitant TVS or CA had directionally higher mortality and cardiac readmission rates. These results help refine outcome benchmarks as new transcatheter MVr procedures continue to emerge.

In-Hospital Outcomes of Heart Failure Patients with Valvular Heart Disease: Insights from Real-World Claims Data.

Purpose: Heart failure (HF) is a serious public health burden that is rapidly increasing in the aging population. Valvular heart disease (VHD) is a known etiology of heart failure (HF); however, the impact of VHD on outcomes of patients with HF has not been well-studied in Japan. This study aimed to determine the rates of VHD in Japanese patients admitted for HF and explore associations of VHD with in-hospital outcomes through a claim-based analysis. Patients and methods: We analyzed claims data from 86,763 HF hospitalizations (January 2017 through December 2019) from the Medical Data Vision database. Common etiologies of HF were examined, then hospitalizations were categorized into those with VHD and those without. Covariate-adjusted models were used to explore the association of VHD with in-hospital mortality, length of stay, and medical cost. Results: Of 86,763 hospitalizations for HF, 13,183 had VHD and 73,580 did not. VHD was the second most frequent etiology of HF (15.2%). The most frequent type of VHD was mitral regurgitation (36.4% of all hospitalizations with VHD), followed by aortic stenosis (33.7%) and aortic regurgitation (16.4%). There was no significant difference in in-hospital mortality between hospitalizations with VHD vs those without (9.0% vs 8.9%; odds ratio [95% CI]: 1.01 [0.95-1.08]; p=0.723). Hospitalizations with VHD were associated with significantly longer length of stay (26.1 vs 24.8 days; incident rate ratio [95% CI]: 1.05 [1.03-1.07]; p<0.001) and higher medical costs (1536 vs 1195 thousand yen; rate ratio [95% CI]: 1.29 [1.25-1.32]; p<0.001). Conclusion: VHD was a frequent etiology of HF that was associated with significant medical resource use. Future studies are needed to investigate whether timely VHD treatment could reduce HF progression and its associated healthcare resource utilization.

Cost and Clinical Outcomes Evaluation Between the Endoaortic Balloon and External Aortic Clamp in Cardiac Surgery.

OBJECTIVE: Endoaortic balloon occlusion facilitates cardioplegic arrest during minimally invasive surgery (MIS). Studies have shown endoclamping to be as safe as traditional aortic clamping. We compared outcomes and hospital costs of endoclamping versus external aortic occlusion in a large administrative database. METHODS: There were 52,882 adults undergoing eligible cardiac surgery (October 2015 to March 2020) identified in the Premier Healthcare Database. Endoclamp procedures (n = 419) were 1:3 propensity score matched to similar procedures using external aortic occlusion (n = 1,244). Generalized linear modeling measured differences in in-hospital complications (major adverse renal and cardiac events, including mortality, new-onset atrial fibrillation, acute kidney injury [AKI], myocardial infarction [MI], postcardiotomy syndrome, stroke/transient ischemic attack [TIA], and aortic dissection) and length of stay (LOS). RESULTS: The mean age was 63 years, and 53% were male (n = 882). The majority (93%, n = 1,543) were mitral valve procedures, and 17% of procedures (n = 285) were robot-assisted. Total hospitalization costs were not statistically significantly different between the 2 groups ($52,158 vs $49,839, P = 0.06). The median LOS was significantly shorter in the endoclamp group (incident rate ratio = 0.87, P < 0.001). Mortality, atrial fibrillation, AKI, and stroke/TIA were similar between the 2 groups. MI and postcardiotomy syndrome were lower in the endoclamp group (odds ratio [OR] = 0.14, P = 0.006, and OR = 0.27, P = 0.005). There were no aortic dissections in the endoclamp group. CONCLUSIONS: Aortic endoclamping in MIS was associated with similar costs, shorter LOS, no dissections, and comparably low mortality and stroke rates when compared with external clamping in this hospital billing dataset. These results demonstrate the clinical safety and efficacy of endoaortic balloon clamping in a real-world setting. Further studies are warranted.

Impact of Managing Provider Type on Severe Aortic Stenosis Management and Mortality.

Background Many patients with symptomatic severe aortic stenosis do not undergo aortic valve replacement (AVR) despite clinical guidelines. This study analyzed the association of managing provider type with cardiac specialist follow-up, AVR, and mortality for patients with newly diagnosed severe aortic stenosis (sAS). Methods and Results We identified adults with newly diagnosed sAS per echocardiography performed between January 2017 and March 2019 using Optum electronic health record data. We then selected from those meeting all eligibility criteria patients managed by a primary care provider (n=1707 [25%]) or cardiac specialist (n=5039 [75%]). We evaluated the association of managing provider type with cardiac specialist follow-up, AVR, and mortality, as well as the independent association of cardiac specialist follow-up and AVR with mortality, within 1 year of echocardiography detecting sAS. A subgroup analysis was limited to patients with symptomatic sAS. Patient characteristics and comorbidities at baseline were used for covariate-adjusted cause-specific and multivariable Cox proportional hazard models assessing group differences in outcomes by managing provider type. An adjusted Cox proportional hazard model with additional time-dependent covariates for follow-up and AVR was used to assess these practices' association with mortality. Within 1 year of echocardiography detecting sAS, data revealed that primary care provider management was associated with lower rates of cardiac specialist follow-up (hazard ratio [HR], 0.47 [95% CI, 0.43-0.50], P<0.0001) and AVR (HR, 0.58 [95% CI, 0.53-0.64], P<0.0001) and with higher 1-year mortality (HR, 1.45 [95% CI, 1.26-1.66], P<0.0001). Cardiac specialist follow-up and AVR were independently associated with lower mortality (follow-up: HR, 0.55 [95% CI, 0.48-0.63], P<0.0001; AVR: HR, 0.70 [95% CI, 0.60-0.83], P<0.0001). Results were similar for patients with symptomatic sAS. All analyses were adjusted for baseline patient characteristics and comorbidities. Conclusions For patients newly diagnosed with sAS, we observed differences in rates of cardiac specialist follow-up and AVR and risk of mortality between primary care provider- versus cardiologist-managed patients with sAS. In addition, a lower likelihood of receiving follow-up and AVR was independently associated with higher mortality.

Utilization of Medical Codes for Hypotension in Shock Patients: A Retrospective Analysis.

PURPOSE: To evaluate the utilization of hypotension diagnosis codes by shock type and year in known hypotensive patients. PATIENTS AND METHODS: Retrospective analysis of the Medicare fee-for-service claims database. Patients with a shock diagnosis code between 2011 and 2017 were identified using the International Classification of Diseases, Ninth and Tenth Revision, Clinical Modification (ICD-9-CM and ICD-10-CM). Based on specific ICD codes corresponding to each shock type, patients were classified into four mutually exclusive cohorts: cardiogenic shock, hypovolemic shock, septic shock, and other/unspecified shock. Annual proportion and counts of cases with at least one hypotension ICD code for each shock cohort were generated to produce 7-year medical code utilization trends. A Cochran-Armitage test for trend was performed to evaluate the statistical significance. RESULTS: A total of 2,200,275 shock patients were analyzed, 13.3% (n=292,192) of which received a hypotension code. Hypovolemic shock cases were the most likely to receive a hypotension code (18.02%, n=46,544), while septic shock cases had the lowest rate (11.48%, n=158,348). The proportion of patients with hypotension codes for other cohorts were 18.0% (n=46,544) for hypovolemic shock and 16.9% (n=32,024) for other/unspecified shock. The presence of hypotension codes decreased by 0.9% between 2011 and 2014, but significantly increased from 10.6% in 2014 to 17.9% in 2017 (p <0.0001, Z=-105.05). CONCLUSION: Hypotension codes are remarkably underutilized in known hypotensive patients. Patients, providers, and researchers are likely to benefit from improved hypotension coding practices.

Relation of Institutional Mitral Valve Surgical Volume to Surgical and Transcatheter Outcomes in Medicare Patients.

There are limited data to support proposed increases to the minimum institutional mitral valve (MV) surgery volume required to begin a transcatheter mitral valve repair (TMVr) program. The current study examined the association between institutional MV procedure volumes and outcomes. All 2017 Medicare fee-for-service patients who received a TMVr or MV surgery procedure were included and analyzed separately. The exposure was institutional MV surgery volume: low (1 to 24), medium (25 to 39) or high (40+). Outcomes were in-hospital mortality and 1-year postdischarge mortality and cardiovascular rehospitalization. For MV surgery patients, in-hospital mortality rates were 6.4% at low-volume, 8.7% at medium-volume and 9.8% at high-volume facilities. Rates were significantly higher for low-volume [OR = 1.50, 95% CI (1.23 to 1.84)] and medium-volume [OR = 1.33, 95% CI (1.06 to 1.67)] compared with high-volume facilities. There was no statistically significant relationship between institutional MV surgery volume and in-hospital mortality for TMVr patients, either at low-volume [OR = 1.52, 95% CI (0.56, 4.13)] or medium-volume [OR = 1.58, 95% CI (0.82, 3.02)] facilities, compared with high-volume facilities. Across all volume categories, in-hospital mortality rates for TMVr patients were relatively low (2.3% on average). For both cohorts, the rates of 1-year mortality and cardiovascular rehospitalizations were not significantly higher at low- or medium-volume MV surgery facilities, as compared with high-volume. In conclusion, among Medicare patients, there was a relation between institutional MV surgery volume and in-hospital mortality for MV surgery patients, but not for TMVr patients.

Key Design Considerations When Calculating Cost Savings for Population Health Management Programs in an Observational Setting.

OBJECTIVE: To illustrate the impact of key quasi-experimental design elements on cost savings measurement for population health management (PHM) programs. DATA SOURCES: Population health management program records and Medicaid claims and enrollment data from December 2011 through March 2016. STUDY DESIGN: The study uses a difference-in-difference design to compare changes in cost and utilization outcomes between program participants and propensity score-matched nonparticipants. Comparisons of measured savings are made based on (1) stable versus dynamic population enrollment and (2) all eligible versus enrolled-only participant definitions. Options for the operationalization of time are also discussed. DATA COLLECTION/EXTRACTION METHODS: Individual-level Medicaid administrative and claims data and PHM program records are used to match study groups on baseline risk factors and assess changes in costs and utilization. PRINCIPAL FINDINGS: Savings estimates are statistically similar but smaller in magnitude when eliminating variability based on duration of population enrollment and when evaluating program impact on the entire target population. Measurement in calendar time, when possible, simplifies interpretability. CONCLUSION: Program evaluation design elements, including population stability and participant definitions, can influence the estimated magnitude of program savings for the payer and should be considered carefully. Time specifications can also affect interpretability and usefulness.

Association of a Care Coordination Model With Health Care Costs and Utilization: The Johns Hopkins Community Health Partnership (J-CHiP).

Importance: The Johns Hopkins Community Health Partnership was created to improve care coordination across the continuum in East Baltimore, Maryland. Objective: To determine whether the Johns Hopkins Community Health Partnership (J-CHiP) was associated with improved outcomes and lower spending. Design, Setting, and Participants: Nonrandomized acute care intervention (ACI) and community intervention (CI) Medicare and Medicaid participants were analyzed in a quality improvement study using difference-in-differences designs with propensity score-weighted and matched comparison groups. The study spanned 2012 to 2016 and took place in acute care hospitals, primary care clinics, skilled nursing facilities, and community-based organizations. The ACI analysis compared outcomes of participants in Medicare and Medicaid during their 90-day postacute episode with those of a propensity score-weighted preintervention group at Johns Hopkins Community Health Partnership hospitals and a concurrent comparison group drawn from similar Maryland hospitals. The CI analysis compared changes in outcomes of Medicare and Medicaid participants with those of a propensity score-matched comparison group of local residents. Interventions: The ACI bundle aimed to improve transition planning following discharge. The CI included enhanced care coordination and integrated behavioral support from local primary care sites in collaboration with community-based organizations. Main Outcomes and Measures: Utilization measures of hospital admissions, 30-day readmissions, and emergency department visits; quality of care measures of potentially avoidable hospitalizations, practitioner follow-up visits; and total cost of care (TCOC) for Medicare and Medicaid participants. Results: The CI group had 2154 Medicare beneficiaries (1320 [61.3%] female; mean age, 69.3 years) and 2532 Medicaid beneficiaries (1483 [67.3%] female; mean age, 55.1 years). For the CI group's Medicaid participants, aggregate TCOC reduction was $24.4 million, and reductions of hospitalizations, emergency department visits, 30-day readmissions, and avoidable hospitalizations were 33, 51, 36, and 7 per 1000 beneficiaries, respectively. The ACI group had 26 144 beneficiary-episodes for Medicare (13 726 [52.5%] female patients; mean patient age, 68.4 years) and 13 921 beneficiary-episodes for Medicaid (7392 [53.1%] female patients; mean patient age, 52.2 years). For the ACI group's Medicare participants, there was a significant reduction in aggregate TCOC of $29.2 million with increases in 90-day hospitalizations and 30-day readmissions of 11 and 14 per 1000 beneficiary-episodes, respectively, and reduction in practitioner follow-up visits of 41 and 29 per 1000 beneficiary-episodes for 7-day and 30-day visits, respectively. For the ACI group's Medicaid participants, there was a significant reduction in aggregate TCOC of $59.8 million and the 90-day emergency department visit rate decreased by 133 per 1000 episodes, but hospitalizations increased by 49 per 1000 episodes and practitioner follow-up visits decreased by 70 and 182 per 1000 episodes for 7-day and 30-day visits, respectively. In total, the CI and ACI were associated with $113.3 million in cost savings. Conclusions and Relevance: A care coordination model consisting of complementary bundled interventions in an urban academic environment was associated with lower spending and improved health outcomes.

A method for estimating cost savings for population health management programs.

OBJECTIVE: To develop a quasi-experimental method for estimating Population Health Management (PHM) program savings that mitigates common sources of confounding, supports regular updates for continued program monitoring, and estimates model precision. DATA SOURCES: Administrative, program, and claims records from January 2005 through June 2009. DATA COLLECTION/EXTRACTION METHODS: Data are aggregated by member and month. STUDY DESIGN: Study participants include chronically ill adult commercial health plan members. The intervention group consists of members currently enrolled in PHM, stratified by intensity level. Comparison groups include (1) members never enrolled, and (2) PHM participants not currently enrolled. Mixed model smoothing is employed to regress monthly medical costs on time (in months), a history of PHM enrollment, and monthly program enrollment by intensity level. Comparison group trends are used to estimate expected costs for intervention members. Savings are realized when PHM participants' costs are lower than expected. PRINCIPAL FINDINGS: This method mitigates many of the limitations faced using traditional pre-post models for estimating PHM savings in an observational setting, supports replication for ongoing monitoring, and performs basic statistical inference. CONCLUSION: This method provides payers with a confident basis for making investment decisions.

Predictive modeling in practice: improving the participant identification process for care management programs using condition-specific cut points.

The objective of this study was to optimize predictive modeling in the participant selection process for care management (CM) programs by determining the ideal cut point selection method. Comparisons included: (a) an evidence-based "optimal" cut point versus an "arbitrary" threshold, and (b) condition-specific cut points versus a uniform screening method. Participants comprised adult Medicaid health plan members enrolled during the entire study period (January 2007-December 2008) who had at least 1 of the chronic conditions targeted by the CM programs (n = 6459). Adjusted Clinical Groups Predictive Modeling (ACG-PM) system risk scores in 2007 were used to predict those with the top 5% highest health care expenditures in 2008. Comparisons of model performance (ie, c statistic, sensitivity, specificity, positive predictive value) and identified population size were used to assess differences among 3 cut point selection approaches: (a) single arbitrary cut point, (b) single optimal cut point, and (c) condition-specific optimal cut points. The "optimal" cut points (ie, single and condition-specific) both outperformed the "arbitrary" selection process, yielding higher probabilities of correct prediction and sensitivities. The condition-specific optimal cut point approach also exhibited better performance than applying a single optimal cut point uniformly across the entire population regardless of condition (ie, a higher c statistic, specificity, and positive predictive value, although sensitivity was lower), while identifying a more manageable number of members for CM program outreach. CM programs can optimize targeting algorithms by utilizing evidence-based cut points that incorporate condition-specific variations in risk. By efficiently targeting and intervening with future high-cost members, health care costs can be reduced.

Chronic care improvement in primary care: evaluation of an integrated pay-for-performance and practice-based care coordination program among elderly patients with diabetes.

OBJECTIVE: To examine the effects of an intervention comprising (1) a practice-based care coordination program, (2) augmented by pay for performance (P4P) for meeting quality targets, and (3) complemented by a third-party disease management on quality of care and resource use for older adults with diabetes. DATA SOURCES/STUDY SETTING: Claims files of a managed care organization (MCO) for 20,943 adults aged 65 and older with diabetes receiving care in Alabama, Tennessee, or Texas, from January 2004 to March 2007. STUDY DESIGN: A quasi-experimental, longitudinal study in which pre- and postdata from 1,587 patients in nine intervention primary care practices were evaluated against 19,356 patients in MCO comparison practices (>900). Five incentivized quality measures, two nonincentivized measures, and two resource-use measures were investigated. We examined trends and changes in trends from baseline to follow-up, contrasting intervention and comparison group member results. PRINCIPAL FINDINGS: Quality of care generally improved for both groups during the study period. Only slight differences were seen between the intervention and comparison group trends and changes in trends over time. CONCLUSIONS: This study did not generate evidence supporting a beneficial effect of an on-site care coordination intervention augmented by P4P and complemented by third-party disease management on diabetes quality or resource use.