Medicaid Managed Care Reduces Readmissions for Youths With Type 1 Diabetes

An analysis of the largest cohort available reveals that youths with type 1 diabetes, on a Medicaid managed care plan, are less likely to be readmitted within 90 days of discharge.
Published Online: April 20, 2016
Kathleen Healy-Collier, CSSBB, DHA; Walter J. Jones, PhD; James E. Shmerling, DHA, FACHE; Kenneth R. Robertson, MD, MBA; and Robert J. Ferry, Jr, MD, FAAP


Objectives: To determine whether the likelihood of readmission (adjusted for severity on first admission) for pediatric type 1 diabetes (T1D) differs between Medicaid managed care and non–managed care.

Study Design: De-identified patients were retrospectively selected from the Pediatric Health Information Systems database of the Children’s Hospital Association (CHA). The cohort of 42 hospitals across 25 states included discharges between 2008 and 2011 for patients who were receiving Medicaid at the time of service and had T1D as their diagnosis.

Methods: Multiple factors and co-variants for readmission were analyzed by logistic regression, including age, race, gender, severity of illness, and state of admission.

Results: Of 14,544 T1D discharges with Medicaid, 4985 were readmitted, including 1792 readmitted for diabetic ketoacidosis (DKA). Despite similar rates of DKA between the managed care and non–managed care cohorts, overall 90-day readmission was 1.12 times more likely for Medicaid patients on non–managed care plans than those on managed care (odds ratio, 1.12; range = 1.04-1.20; both adjusted for severity of illness). Significant contributors were race, age, and gender; the relationship of location (state) and days between readmissions was also significant. The conservative estimate of cost reduction from Medicaid managed care related to lower readmission rate for pediatric T1D across CHA institutions between 2008 and 2011 was $2.6 million.

Conclusions: From the largest, national, defined cohort available for contemporary study, youths with T1D on Medicaid managed care plans were less likely to be readmitted within 90 days of discharge.

Am J Manag Care. 2016;22(4):250-256

Take-Away Points
This large national study of 42 free-standing children’s hospitals across 25 states compared youths with type 1 diabetes (T1D) on Medicaid insurance who received a managed care product with peers who did not. Key findings were: 
  • Youths with T1D on Medicaid managed care were less likely to be readmitted within 90 days, adjusting for severity, despite similar rates of diabetic ketoacidosis (DKA). 
  • States displayed wide variation in overall readmission rates for diabetes and those presenting in DKA. 
  • Policy makers should build on specific successes with the use of managed care tools (eg, case management, health information technology) to reduce preventable readmissions related to pediatric T1D.
As many as 3 million US individuals were affected by type 1 diabetes (T1D) in 2011, with youths aged under 20 years comprising a large group at particular risk for healthcare disparities.1 Over 15,000 US children are newly diagnosed with T1D each year,1 and for unclear reasons, the incidence of T1D in children under the age of 14 has been reported to be increasing by 3% annually worldwide.2,3 Most alarming, the incidence of diabetic ketoacidosis (DKA) in the United States continues to increase even more rapidly than the incidence of patients being newly diagnosed with diabetes.4-6 Since readmission for pediatric DKA is expensive, associated with significant morbidity and mortality, and largely preventable, active research has been focusing on variation in resource use.6

A significant proportion of youths are enrolled in a state Medicaid program, and lack of standardization among these programs can contribute to disparities in access and quality of care for those with diabetes.7 Many state Medicaid programs have had some level of managed care penetration for over 30 years, with mixed evidence for reduced healthcare costs in adults.8-11  Medicaid programs have been set up as traditional indemnity plans, or as various managed care organizations (MCOs) with mixed levels of implied risk. The potential benefits of MCOs include provision of comprehensive healthcare at reduced cost.11 Economic and clinical strategies within managed care plans also directly affect physician compliance with best practices, efforts to improve outcomes, and coordination of care across clinical services through case management.12-15 Tools applied within MCOs have included: 1) maintaining a “gatekeeper” in efforts to improve coordination of care, 2) limiting access to services with prior authorizations and pre-certifications, 3) limiting formularies, 4) sharing risk by arrangement, and 5) controlling access to supplies and equipment.11

Although each state coordinates its own Medicaid program, the federal government now pays an average of 57% of the total Medicaid program costs.10,16 As leaders of federal and state Medicaid programs have recognized the rising cost, they have implemented programs to manage care in order to reduce expenditures; this trend has accelerated over the past 15 years, despite mixed evidence of actual cost reduction (vs traditional fee-for-service).17 Medicaid expansion and reforms related to the Affordable Care Act (ACA) will hopefully yield more meaningful analyses in this regard.

Compounding regional workforce challenges for youths with diabetes,18 recent studies have demonstrated that those with public or no insurance were more likely to be hospitalized than those on private insurance.19 Market efforts to evaluate the effects of competition and risk models through managed care have yielded mixed evidence for optimizing clinical outcomes.20 To our knowledge, no study to date has focused directly on the national impact of managed care on pediatric diabetes.

Using the Pediatric Health Information Systems (PHIS) database from the Children’s Hospital Association (CHA), the present study compares managed care to non–managed care by evaluating all readmissions within 90 days for youths with T1D receiving Medicaid. The goal of this research is to understand the effects of managed care for youths admitted on their initial visits with T1D, those readmitted for T1D, and those readmitted for DKA, while adjusting for admissions related to nondiabetic issues.

Forty-two institutions across 25 states comprise the CHA, the nation’s largest consortium of children’s hospitals. De-identified administrative and clinical data were retrieved from the CHA’s PHIS database between 2008 and 2011. The primary sample was the multi-state cohort of all patients from the CHA data set, including 25 states and 42 children’s hospitals that had consistently submitted data between 2008 and 2011; 1 hospital was removed from the data analysis due to inconsistencies with data from that site. Data for analysis included all patients who were discharged from any participating CHA institution between 2008 and 2011. Data were filtered for those patients on a Medicaid program, and then for T1D; all episodes were grouped using the All Patients Refined Diagnosis-Related Group approach, with severity of illness assigned on the first admission. Data for the sample were based on the following final diagnoses for any patient in the data set: 250.01 (T1D without mention of complication; 50%); 250.03 (uncontrolled T1D; 6%); 250.11 (T1D with ketoacidosis, not stated as uncontrolled; 3%); and 250.13 (T1D with ketoacidosis, uncontrolled; 41%).

Across the 42 hospitals contributing to the data set, 14,544 patients with diabetes on in-state Medicaid were flagged as either “managed care” (7835) or “other” (6709). We excluded out-of-state Medicaid patients due to concern about potential confounding of the analysis if the patient was unlikely to return to the same hospital. Although the analysis included all patients with these diagnoses, data were classified as either: 1) all diabetic readmissions with a coded diabetes diagnosis, or 2) all diabetes, whether readmitted or not with DKA. Across the data set, 9633 patients with diabetes were not readmitted (66.6%) (Table 1), and of the 4985 readmissions, 1792 were diagnosed with DKA. For this retrospective longitudinal cohort, all participants were continuously enrolled over the time period studied and were not excluded if they did not have a readmission.

The dependent variable was diabetic readmission for diabetes (Yes/No) or for diabetes with DKA (Yes/No). The independent variables were managed care flag (primary as a categorical variable) and age, gender, race, and severity of illness on first admission. Covariates were measured at the time of discharge. The blinded data set lacked personal health information linked to any specific patient. Severity of illness was determined by CHA institutions, as previously described.21 All analyses were performed using SAS version 9.2 (SAS Institute, Cary, North Carolina). The CHA and the Medical University of South Carolina Institutional Review Board approved the study.

To estimate the cost savings associated with reduced readmissions, we adapted our previously reported method.22 Based on 1 representative CHA hospital’s average daily charge ($4695) for ward admission of pediatric diabetes, we multiplied this average daily charge by the CHA’s reported average for readmission length of stay and by the calculated difference in readmission rates with managed care. Cost estimates are reported in 2013 US dollars.

During the study period, 14,544 qualifying discharges for pediatric diabetes among 42 hospitals met our entry criteria, and these data represented 12,618 individual patients. Table 1 illustrates the total numbers of readmissions by category. Patients are not unique in each category because those who are readmitted for diabetes may also be readmitted due to DKA. Overall, 14.9% of the 12,618 patients were readmitted with a diagnosis of T1D within 90 days, and 12.1% of those were readmitted with DKA. Time to readmission ranged from 0 to 90 days, with a mean of 38.5 days between readmissions.

Crude readmission rates for diabetes, sorted by state (Table 1), ranged from 4.08% to 24.82% (mean = 13.57%), with readmission rates for DKA ranging from 3.69% to 25.33% (mean = 12.10%). For the overall diabetic admission rate, the patient’s race (P <.0001), age (P <.0001), and gender (P <.03) were all significant factors (Table 2). When combined with any of these factors, the primary insurance type contributed significantly to the model. Adjusting for severity of illness, logistic regression revealed that overall readmissions at 90 days were 1.12 times more likely for Medicaid patients on non–managed care plans than for those on managed care plans (odds ratio, 1.12; 95% CI, 1.04-1.20) (Table 3). When both the overall diabetic readmission rate and the DKA readmission rate were introduced as factors into the model, primary insurance type became more significant. Analysis demonstrated a highly significant relationship between the US state and the number of days between readmissions (P <.0001), even after adjusting for severity. For those patients readmitted with DKA, we observed no difference in readmissions between managed care and non–managed care groups. No significant differences were observed for DKA readmissions at 7 days or 30 days.

PDF is available on the last page.
Adult ADHD Compendium
COPD Compendium
Dermatology Compendium
Diabetes Compendium
Hematology Compendium
Immuno-oncology Compendium
Lipids Compendium
MACRA Compendium
Neutropenia Compendium
Oncology Compendium
Pain Compendium
Reimbursement Compendium
Rheumatoid Arthritis Compendium
Know Your News
HF Compendium
Managed Care PODCAST