Implementation of the Quality Blue Primary Care program in Louisiana was associated with a shift in primary care delivery and reductions in overall cost.
Objectives: This study aimed to investigate the role of the Quality Blue Primary Care (QBPC) program on healthcare utilization and overall cost among the beneficiaries of Blue Cross and Blue Shield of Louisiana (BCBSLA).
Study Design: Retrospective observational cohort study using claims data from adults residing in QBPC-implemented regions continuously enrolled through BCBSLA from June 2012 to December 2014 (N = 89,034).
Methods: Controlling for age, gender, and risk score by propensity score weighting, inpatient, outpatient, and corresponding medical expenditures were each compared between the QBPC group and the control group using a difference-in-differences regression model.
Results: Average total cost increased in both the QBPC and control groups in 2014, but the increase was significantly less in the intervention group—a difference of $27.09 per member per month (PMPM) (P ≤.001). Savings in total cost were derived largely from a decrease in hospitalizations—a difference of $18.85 PMPM (P = .0023). Furthermore, savings were associated with shifts in healthcare utilization by the intervention group toward proactive management, including increased primary care physician visits (P = .0106) and higher screening rates for diabetes (P = .0019). Inpatient admissions also decreased in the QBPC group, most significantly among those with chronic conditions (P <.05). Conversely, an unexpected increase was observed in emergency department visits.
Conclusions: The QBPC program was associated with a shift in primary care delivery and reductions in overall cost. Savings were achieved largely through reductions in hospitalization costs.
Am J Manag Care. 2017;23(12):e402-e408Takeaway Points
The United States spent in excess of $3.0 trillion, or $9523 per person, on healthcare in 2014, reflecting an increase of 5.3% from 2013.1,2 Louisiana ranks among the lowest states in healthcare quality and among the highest in healthcare expenditures per capita.3-5 The national trend is further exacerbated by the rising number of aging individuals with chronic conditions, estimated to account for more than 75% of total healthcare costs.6 Patients with more than 1 chronic condition are estimated to account for 95% of all Medicare spending.7 The concentration of healthcare expenditures involved with chronic conditions is a major concern for individuals, insurance companies, and policy makers alike, and understanding how to effectively care for individuals with multiple chronic conditions is one of the most important challenges our healthcare system is facing.8,9
Various initiatives have targeted health management and quality improvement for patients with chronic conditions. Among the components in the chronic care model,10 delivery system design and clinical information systems were proven effective in improving the management of chronic conditions.11 Beyond those, innovations toward enhanced primary care performance12 and payment systems also play important roles. In 2009, Blue Cross Blue Shield of Massachusetts launched a global payment system, the Alternative Quality Contract, to improve healthcare quality by offering substantial financial incentives to primary care providers based on performance against quality measures. Improvement in quality and reduced spending growth were observed 1 year after implementation.13 Based on the success observed in Massachusetts, we recognize a significant potential in working with primary care networks to improve healthcare quality and expenditures.
The cornerstone of effective management of chronic conditions is collaborative team-based care.14 In recognition of these solutions, Blue Cross and Blue Shield of Louisiana (BCBSLA) is taking a lead role in engaging and supporting primary care physicians (PCPs) to redesign healthcare. In 2012, BCBSLA piloted a population health and quality improvement program called Quality Blue Primary Care (QBPC) (see eAppendix: Design and Implementation of Quality Blue Primary Care Program [eAppendix available at ajmc.com]) to improve patient care delivery for PCPs in any of 3 specialties: family medicine, internal medicine, and general practice. In the QBPC program, BCBSLA contracts with PCPs and provides a free Web-based patient-centric health information exchange tool to support the existing standard of care. This tool was designed to identify and manage chronic conditions that are prevalent and burdensome, while providing practices with data and resources that enable proactive, efficient, and high-quality care. QBPC leverages the current framework of the chronic care model and aims to create a minimally disruptive, efficient, and active care management process, whereby a BCBSLA-employed case manager acts as the care coordinator. Integration of the Web-based tool across multiple medical providers, including physicians, nurses, and care coordinators, enables all team members to act on timely key patient data. The program also equips primary care providers with an outcomes-based reimbursement structure that supports increased value and drives cost reduction through care coordination.15
QBPC was designed to improve the current quality of care. In this study, we identify changes in healthcare utilization and expenditures associated with implementation of the QBPC
program in Louisiana.
This study was a retrospective observational cohort study of BCBSLA enrollees. Pre- and postintervention data were collected and balanced by propensity score weighting for both the QBPC and control groups, and a difference-in-differences (DID) multivariate regression analysis was used to identify changes in healthcare utilization and expenditures associated with the implementation of the QBPC program.
Our study used medical and pharmacy claims data from BCBSLA. The study population included adults who were continuously enrolled in BCBSLA medical and pharmacy insurance from July 2012 through December 2014. All QBPC providers were enrolled in QBPC from July 2013 to December 2013 (enrollment period) and remained in QBPC during all of 2014 (outcome period). The members who visited the QBPC providers in 2014 were defined as the intervention group, and the members who visited non-QBPC providers in 2014 were defined as the control group. The baseline period was defined as 1 year before the QBPC start date (July 2012-June 2013).
The selected members were 18 years and older and were required to be residents in regions with QBPC providers (Baton Rouge, Lake Charles, Monroe, New Orleans, and Shreveport). BCBSLA needed to be the primary payer for the selected members. Members with supplementary plans only (BlueChoice 65, Variable Income Plan, Cancer and Serious Disease plan, dental, vision, or part D) were excluded from our sample. We also excluded members who crossed over between the comparison groups in the outcome period and members who had extremely high annual expenditures on inpatient care (≥$100,000 per year) (Figure 1 shows the flow chart of sample selection).
Outcomes and Key Covariates
Inpatient admissions, office-based visits, and emergency department (ED) visits per 1000 members were estimated as utilization outcomes. Inpatient care included total admissions and admissions with any one, or more, of the following diagnoses: cardiovascular disease (CVD), hypertension (HTN), diabetes, and chronic kidney disease (CKD). Diagnoses were determined by International Classification of Diseases, Ninth Revision, Clinical Modification codes, independent of the individual order of diagnosis. Total office-based visits were estimated and further specified as visits to PCPs/nurse practitioners (NPs) and visits to specialists. ED visits were defined as total ED visits, ambulatory ED visits, and admitted ED visits (ie, ED visits followed by inpatient care).
Health expenditures in this study were defined as the allowed amount paid by BCBSLA, presented as dollars per member per month (PMPM). Total costs were summed by total medical costs and total prescription costs, estimated and shown in result table separately. The costs linked to utilization were captured and categorized by ED (ambulatory ED and admitted ED), inpatient admissions and admissions with chronic conditions, and office-based visits (eg, PCPs/NPs, specialists).
Diabetes management outcomes were measured by screening test rates of glycated hemoglobin (A1C), low-density lipoprotein cholesterol, and microalbuminuria.
Age was defined as the age at the end of the baseline period. The DxCG risk score was classified into 5 levels: healthy, stable, at risk, struggling, and in crisis. Insurance types were defined as the specific products members enrolled in through BCBSLA, listed as preferred provider organization (PPO), health maintenance organization (HMO), and Community Blue/Blue Connect (designed for lower monthly premiums).
The demographic characteristics for the QBPC and control groups were described at the baseline period by means and percentages. The statistical differences between the 2 groups were compared by t test for continuous variables and χ2 test for categorical variables.
To mitigate differences in members’ baseline characteristics across QBPC and control groups, propensity score weights (PSWs) were estimated by age, gender, risk score, residential region, and insurance type in a logistic regression. The propensity score (PS) was predicted for both the QBPC and control groups, and the inverse and normalized PSs were used as PSWs in the outcomes analysis.16
Multivariate regression analysis of a DID model with PSW was used to estimate the impact of QBPC on healthcare utilization and expenditures and the quality of diabetes management, controlling for age, gender, product type, and categorized risk score. Generalized linear model (GLM) was used with Poisson distribution and log link function for outcome of utilization. Gamma distribution and log link function for outcome of expenditure, and binomial distribution and logit link function for lab test rate, were assigned in GLM.
Rate ratios (RRs) and 95% CIs from multivariate regression models were presented, and a 2-tailed alpha level of 0.05 was used to determine statistical significance. SAS software version 9.4 was used to conduct statistical analyses (SAS; Cary, North Carolina).
A total of 89,034 BCBSLA members were included in the study sample, with 13,914 enrollees in the QBPC group and 75,120 enrollees in the control group (Table 1). An average age of 46.9 years was observed in the QBPC group compared with 45.2 years in the control group. A total of 54.1% of enrollees in the QBPC group were female compared with 52.5% in the control group. The general health of enrollees measured by risk score was worse in the QBPC group compared with the control group. Higher proportions of enrollees in the QBPC group were defined as at risk, struggling, and in crisis compared with the control group, which contained higher proportions of enrollees defined as healthy and stable. Enrollees in the QBPC group were primarily from Baton Rouge (58.6%), New Orleans (19.6%), and Shreveport (15.2%), whereas enrollees in the control group were primarily from New Orleans (34.7%), Baton Rouge (27.9%), and Lake Charles (16.0%). A total of 67.4% of enrollees in the QBPC group held PPO plans and 32.3% were HMO plan members, whereas 78.4% of enrollees in the control group held PPO plans and 21.51% were HMO plan members. There were no significant differences in gender, risk categories, and residential regions between the intervention and control groups after propensity score weighting (P >.05).
Total office-based visits increased in both the QBPC and control groups (Table 2). The increase in the intervention group was significantly less than the increase in the control group (RR, 0.99; P = .0066) due to increased visits to specialists by members of the control group (RR, 0.97; P ≤.0001). However, visits to PCPs/NPs increased in both the QBPC and control groups by 60.26 and 7.59 per 1000 members, respectively. The increase in the QBPC group was significantly greater than the increase in the control group (RR, 1.02; P = .0106).
The change in total admissions was not significant between groups. However, admissions for chronic conditions, including CVD, HTN, diabetes, and CKD, significantly decreased in the QBPC group compared with the control group (P ≤.05).
Total ED visits increased in both the QBPC and the control group by 13.86 and 1.84 per 1000 members, respectively, but the increase was significantly higher in the QBPC group (RR, 1.07; P = .0245). Ambulatory ED visits were significantly increased in the QBPC group compared with the control group (RR, 1.08; P = .0130). Admitted ED visits increased in both groups, but no significant difference was observed.
In the QBPC and control groups, total allowed amounts increased by $55.15 and $82.24 PMPM, respectively, but the QBPC group had a significantly lower increase compared with the control group (RR, 0.92; P ≤.0001) (Table 3). Total medical cost also increased in both groups, but again, the increase in the QBPC group was significantly less than in the control group (RR, 0.87; P ≤.0001).
Total allowed amounts for office-based visits and specialists were both reduced in the QBPC group compared with the control group (RR, 0.97; P = .0047; and RR, 0.95; P = .0002, respectively). However, the difference of allowed amount for visits to PCPs/NPs was not significant between groups (RR, 1.01; P = .4595).
Total costs for admissions decreased in the QBPC group and increased in the control group by $6.10 and $12.75 PMPM, respectively. The decrease in the QBPC group was significant compared with the control group (RR, 0.87; P = .0023), but the cost for admissions with chronic conditions was not significant between groups (P ≥.05).
The total allowed amount for ED visits increased in both the QBPC and control groups by $5.07 and $2.62 PMPM, respectively; however, the increase was significantly greater in the QBPC group compared with the control group (RR, 1.10; P = .0031). Cost for ambulatory ED visits increased significantly in the QBPC group compared with the control group (RR, 1.08; P = .0196). There was no significant difference between groups in the allowed amount for admitted ED visits (RR, 0.96; P = .6580).
The allowed amount per admission decreased by $35.63 in the QBPC group and increased by $91.24 in the control group. The decrease in the QBPC group was significant compared with the control group (RR, 0.92; P = .0484). The allowed amount per admission with chronic condition increased in both the QBPC and control groups, but the difference in increase was not significant between groups (RR, 1.08; P = .2988).
Screening test rates for A1C increased in the QBPC group by 3.92% and decreased in the control group by 1.66% (Figure 2). The increase in the QBPC group was significant
(P = .0019). Screening test rates for lipids increased in the QBPC group by 1.36% and decreased in the control group by 1.63%. The increase in the QBPC group was not significant compared with the control group (P = .1081). Screening test rates for microalbuminuria increased in the QBPC and control groups by 3.53% and 1.32%, respectively. The increase in the QBPC group was not significant compared with the control group (P = .2536).
The QBPC program was associated with a shift in healthcare utilization toward proactive management and reductions in overall cost during the first year after implementation. These changes were associated with a significant difference in total cost savings between the QBPC and control groups of $27.09 PMPM (Table 3). Savings were derived largely through reductions in total admissions, where we observed a cost difference between the QBPC and control groups of $18.85 PMPM (Table 3). In addition, savings in expenditures were associated with shifts in healthcare utilization by QBPC enrollment toward cost-effective prevention practices. We observed increases in the QBPC group in visits to PCPs and NPs and decreases in visits to specialists (Table 2). We observed a cost reduction in total office-based visits, a difference between the QBPC and control group in total cost of $2.32 PMPM (Table 3). Furthermore, we observed increases in the QBPC group in screening test rates for chronic conditions like diabetes (Figure 2).
The unexpected increase in overall ED visits observed in the QBPC group was associated with a significant increase in ambulatory ED visits (Table 2). QBPC enrollment was associated with a decrease in ED admissions, but this decrease was not significant (Table 2). An increase in ED use, especially for ambulatory ED visits, can be due to multiple factors. First, other study results have shown that recent changes in health insurance status under the Affordable Care Act for newly insured adults and newly uninsured adults were associated with greater ED use. As policy and economic forces create disruptions in health insurance status, new surges in ED usage should be anticipated.17 Second, increased access to primary care but failure to provide timely care has been shown to increase preventable ED visits (ie, visits for conditions likely preventable by timely outpatient care). By contrast, study findings have shown no significant change in emergent, nonpreventable visits.18 Delayed primary care, defined as a wait of more than 2 weekdays to access a PCP, has been observed to be associated with a higher rate of self-referred ED usage and subsequent discharge.19,20 These data suggest that the increase in ED visits observed in the QBPC group can be attributed to factors (eg, longer wait time to see PCP) beyond the scope of the QBPC program. Furthermore, these findings support our observed increase in ambulatory ED visits, although there was a decrease in total admissions.
Our study has several limitations. It was designed as a retrospective database analysis using BCBSLA claims data, which include limited clinical information. Due to insufficient data in reference to partial attribution information for baseline characteristics, the attribution model of patients to providers was defined by the information attained in 2014, which was after the QBPC program was implemented. Baseline characteristics were comparable after adjusting for PS, with the exception of significant differences in age and product type. These differences may reflect imperfect weighting, and thus age and product type were also included in our regression models for utilization and costs. The evaluation of QBPC was limited to those early adopter providers in Louisiana. Therefore, the results may not be generalizable to other insurance policies (eg, Medicare/Medicaid population) or to other states. Results may also not be generalizable to the control group (ie, late adopters or providers that refused to adopt QBPC). Our cost analysis accounted solely for the amount paid by the primary payer (BCBSLA) and assumed that additional payer (out-of-pocket) behavior was independent of the QBPC program. Furthermore, we did not examine the details of each QBPC contract, which varied to some degree, or collect information on clinical procedures and outcomes of enrollees. Although we identified associated improvements and cost reductions, these measures do not consider qualitative feedback provided by enrollees and healthcare providers. The long-term effect of QBPC on improving the quality of care at a lower total cost remains contingent on future financial incentives toward preventive care and providers’ ability to further improve synergies between physicians and their chronic condition management teams.
The QBPC program was associated with shifts in healthcare utilization toward proactive management and reductions in overall cost. During the first year of implementation in Louisiana, savings were achieved largely through reductions in office-based visits to specialists and inpatient care. The long-term implications of the QBPC program on improving primary care and patient outcomes at lower total costs warrant additional research.
This study expresses the opinions of the Tulane Research Team, including Drs Shi and Shao. Blue Cross Health Analytics Group provided their reviews on the results. Blue Cross Health Care Analytics Group provided members’ and providers’ claims data to Tulane. This study is funded by Blue Cross. In addition to research funding from Blue Cross, Dr Shi receives funding from Patient-Centered Outcomes Research Institute, Agency for Healthcare Research and Quality, National Institute of Health, Bristol-Myers Squibb, Chiasma, and Cepheid.Author Affiliations: Department of Global Health Management and Policy, School of Public Health and Tropical Medicine (QS, LS), and School of Medicine (TJY), Tulane University, New Orleans, LA; Blue Cross and Blue Shield of Louisiana (PL, PM, XY, HS, WHB, SL, DCan, DCar, JS), Baton Rouge, LA.
Source of Funding: Blue Cross and Blue Shield of Louisiana (BCBSLA).
Author Disclosures: Ms Shi received payment for involvement in the preparation of this manuscript. Mr Lee is employed as vice president of healthcare analytics at BCBSLA, which is the payer for the QBPC program. Dr Murphree is medical director of BCBSLA. Ms Yuan works as a healthcare informatics consultant at BCBSLA. Dr Bestermann is a consultant for BCBSLA and works at the COSEHC Practice Transformation Network, a quality improvement organization. Ms Loupe is employed by BCBSLA. Dr Shi has received grants and honoraria from BCBSLA. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (QS, PL, PM, XY, WHB, SL, DCan, DCar, JS, LS); acquisition of data (QS, PL, PM, XY, WHB, SL); analysis and interpretation of data (QS, TJY, PL, XY, HS, WHB, SL, LS); drafting of the manuscript (QS, TJY, PM, WHB, JS, LS); critical revision of the manuscript for important intellectual content (TJY, PL, HS, WHB, DCan, LS); statistical analysis (QS, TJY, XY, HS, LS); provision of patients or study materials (PL, PM); obtaining funding (JS); administrative, technical, or logistic support (PL, PM, XY, SL, DCan, DCar, LS); and supervision (PL, PM, DCar).
Address Correspondence to: Lizheng Shi, PhD, MSPharm, Tulane University, 1440 Canal St, Ste 1900, New Orleans, LA 70112. E-mail: firstname.lastname@example.org.REFERENCES
1. CMS releases 2014 national health expenditures [press release]. Baltimore, MD: CMS; December 2, 2015. cms.gov/Newsroom/MediaReleaseDatabase/Press-releases/2015-Press-releases-items/2015-12-02.html. Accessed December 2, 2015.
2. Smith C, Cowan C, Sensenig A, Catlin A; Health Accounts Team. Health spending growth slows in 2003. Health Aff (Millwood). 2005;24(1):185-194. doi: 10.1377/hlthaff.24.1.185.
3. Improving Louisiana’s healthcare rankings: a report prepared in response to SCR 111 of the 2012 Regular Session. Louisiana Department of Health website. dhh.louisiana.gov/assets/docs/LegisReports/SCR111-2012R.S..pdf. Published March 5, 2013. Accessed March 5, 2013.
4. Special population estimates for impacted counties in the Gulf Coast area. United States Census Bureau website. ftp.census.gov/topics/preparedness/pdf/impacted_gulf_map.pdf. Accessed June 2006.
5. Report on Louisiana healthcare delivery and financing system. Louisiana Recovery Authority website. lra.louisiana.gov/assets/docs/searchable/reports/PwChealthcarereport42706l.pdf. Accessed March 2006.
6. Vogeli C, Shields AE, Lee TA, et al. Multiple chronic conditions: prevalence, health consequences, and implications for quality, care management, and costs. J Gen Intern Med. 2007;22(suppl 3):391-395. doi: 10.1007/s11606-007-0322-1.
7. Anderson G. Chronic conditions: making the case for ongoing care. Robert Wood Johnson Foundation website. rwjf.org/content/dam/farm/reports/reports/2010/rwjf54583. Accessed February 2010.
8. Orszag PR, Ellis P. The challenge of rising health care costs—a view from the Congressional Budget Office. N Engl J Med. 2007;357(18):1793-1795. doi: 10.1056/NEJMp078190.
9. Chernew ME, Baicker K, Hsu J. The specter of financial armageddon—health care and federal debt in the United States. N Engl J Med. 2010;362(13):1166-1168. doi: 10.1056/NEJMp1002873.
10. Bodenheimer T, Wagner EH, Grumbach K. Improving primary care for patients with chronic illness. JAMA. 2002;288(14):1775-1779. doi: 10.1001/jama.288.14.1775.
11. Bodenheimer T, Wagner EH, Grumbach K. Improving primary care for patients with chronic illness: the chronic care model, part 2. JAMA. 2002;288(15):1909-1914. doi: 10.1001/jama.288.15.1909.
12. Rothman AA, Wagner EH. Chronic illness management: what is the role of primary care? Ann Intern Med. 2003;138(3):256-261. doi: 10.7326/0003-4819-138-3-200302040-00034.
13. Song Z, Safran DG, Landon BE, et al. Health care spending and quality in year 1 of the Alternative Quality Contract. N Engl J Med. 2011;365(10):909-918. doi: 10.1056/NEJMsa1101416.
14. Katon WJ, Lin EH, Von Korff M, et al. Collaborative care for patients with depression and chronic illnesses. N Engl J Med. 2010;363(27):2611-2620. doi: 10.1056/NEJMoa1003955.
15. BlueCross BlueShield of Louisiana. Quality Blue Primary Care: a framework for value-based innovation in healthcare delivery. Patient-Centered Primary Care Collaborative website. pcpcc.org/resource/quality-blue-primary-care. Published January 2015. Accessed January 2015.
16. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399-424. doi: 10.1080/00273171.2011.568786.
17. Ginde AA, Lowe RA, Wiler JL. Health insurance status change and emergency department use among US adults. Arch Intern Med. 2012;172(8):642-647. doi: 10.1001/archinternmed.2012.34.
18. Taubman SL, Allen HL, Wright BJ, Baicker K, Finkelstein AN. Medicaid increases emergency-department use: evidence from Oregon’s Health Insurance Experiment. Science. 2014;343(6168):263-268. doi: 10.1126/science.1246183.
19. Cheung PT, Wiler JL, Lowe RA, Ginde AA. National study of barriers to timely primary care and emergency department utilization among Medicaid beneficiaries. Ann Emerg Med. 2012;60(1):4-10.e2. doi: 10.1016/j.annemergmed.2012.01.035.
20. Cowling TE, Cecil EV, Soljak MA, et al. Access to primary care and visits to emergency departments in England: a cross-sectional, population-based study. PLoS One. 2013;8(6):e66699. doi: 10.1371/journal.pone.0066699.