Multisystemic Therapy for Adolescents With Poorly Controlled Type 1 Diabetes: Reduced diabetic ketoacidosis admissions and related costs over 24 months

A total of 127 adolescents with CPMC and their families were recruited from endocrinology clinics at Children's Hospital of Michigan between 1999 and 2004. Eligible youth were diagnosed with type 1 diabetes for at least 1 year, had an average A1C of ≥8% during the year before study entry, and were aged 10–17 years. Mean ± SD A1C at study entry was 11.3 ± 2.3%. A total of 92% of the subjects used injected insulin, and 8% used insulin pumps. Mean ± SD age was 13.2 ± 2.0 years, and 63% of participants were African American.

A total of 64 participants were randomly assigned to MST and 63 to a control group. All families received quarterly visits with a multidisciplinary diabetes team. MST-treatment families also received 6 months of therapy (mean 5.7 months). Families were followed for 24 months total. MST targeted adherence-related problems within the family and broader community systems (1,3). These systems included family (e.g., poor parental supervision and oversight of the youth's diabetes care completion), school (e.g., inadequate communication between parents and school personnel regarding the youth's health needs), and health care system factors (e.g., barriers to keeping clinic appointments due to problems with transportation or family disorganization).

The number of DKA admissions was obtained from the treating hospital's information system for the 6-month window before study entry (baseline [T1]) and for the follow-up periods (baseline to 6 months [T2], 6–12 months [T3], 12–18 months [T4], and 18–24 months [T5]). Criteria used to diagnose DKA were hyperglycemia (blood glucose >16.65 mmol/l), serum acetone positive at greater than 1:2 dilution of serum, acidosis (pH <7.30 and bicarbonate <15 mmol/l), ketonuria, and glucosuria.

As decreases in A1C at treatment termination for MST-treated youth were not maintained at 6 months (2), cost-effectiveness evaluation was not appropriate. Cost savings were evaluated by reductions in admissions. A total of 45 youth (21 MST-treated and 24 control subjects) had at least one admission during the study. Costs were estimated by obtaining direct hospital costs and revenues from the hospital financial database and calculating an average cost for DKA admissions during the study. Revenues reflected third-party payor reimbursements.

Costs of MST for youth with CPMC were estimated from MST costs in the “real world” rather than costs in the trial, which might underestimate true implementation costs. Because MST as originally developed for delinquent youth is widely disseminated, real-world estimates could be calculated. Costs included salary and benefits; overhead for therapists, supervisors, and program staff; therapist mileage; travel for training; MST licensing fees; and quality assurance costs. Costs per youth were estimated at 6,934 USD.

Figure 1 shows cumulative DKA admissions for MST-treated and control youth. Change in admission frequency over the 24-month trial was evaluated using repeated-measures Poisson regression. Analyses were performed with generalized estimating equations by specifying the Poisson distribution for the response variable (4,5); this accounted for correlated error. Time effects were partitioned into time-ordered contrasts comparing T1 with each follow-up period (T2–T5). The effects of MST were evaluated by the group × time interaction, followed by simple effect tests of MST within-group changes from baseline. Single-parent status was used as a covariate due to its relatedness to outcome in prior studies. The group × time interaction was significant (χ² = 11.77, 4 d.f., n = 127; P = 0.019), indicating MST-treated youth had significantly fewer admissions than control subjects. Simple effect contrasts showed that MST youth had significantly fewer DKA admissions relative to baseline frequency at T2 (P = 0.004), T3 (P = 0.021), T4 (P = 0.046), and T5 (P = 0.034). Drops in admissions per youth were obtained over constant 6-month intervals and, hence, are measures of effect size when expressed as rates. The rates were 0.31 (95% CI 0.09–0.54), 0.27 (0.03–0.50), 0.23 (0.01–0.47), and 0.23 (0.01–0.47) for T2–T5, respectively. Control subjects had fewer admissions only at T5 (P = 0.026); rate 0.24 (0.02–0.45).

Hospital direct costs were 4,237 USD, and revenues were 5,446 USD per admission. The 24 control youth with any admission in 24 months had 85 DKA admissions. The 21 MST-treated youth had 45 admissions. Hence, DKA admissions resulted in 360,145 and 190,665 USD in hospital costs and 462,910 and 245,070 USD in third-party payor costs for control and MST-treated youth, respectively. Costs to provide MST for 21 youth were estimated at 145,614 USD (21 × 6,943). Therefore, MST was estimated to potentially save a total of 23,886 (institutional perspective) or 72,226 (third-party payor perspective) USD.

MST produced lasting reductions in postdiagnostic DKA hospital admissions, which occur most commonly due to insulin noncompliance (6,7). Reduced admissions rates in the MST group at follow-up were consistent with those reported in recent general population studies of youth with diabetes (8). The only other intervention with effects on DKA in youth with CPMC is residential psychiatric treatment (9), a costly intervention with unknown long-term impact. Costs to provide MST to youth with CPMC were relatively high. However, preliminary evaluation suggests that control youth with DKA admissions accumulated sufficient costs over 24 months and that expenditures on MST may be justified by potential for savings. MST could produce cost savings for the subset of youth with CPMC and a recent history of DKA admissions if admissions are occurring frequently. The study demonstrates the potential for intensive behavioral interventions to reduce serious consequences of medication noncompliance in high-risk youth.

This project was supported by grant R01 DK59067 from the National Institute of Diabetes and Digestive and Kidney Diseases.