Telephonic Care Coordination In High-Need, High-Cost Patients

Key Takeaways

• No significant difference was seen in the prespecified coprimary outcomes over 12 months.
• The same overall pattern was seen in the diabetes subsample and in instrumental variable analyses.
• Interpretation was limited by low uptake, 5% control-arm contamination, and limited clinical and process data.

Over 12 months, a JAMA Network Open randomized clinical trial found no significant differences in the prespecified coprimary outcomes among commercially insured adults identified as high-need, high-cost. The trial compared telephonic nurse-led care coordination with usual care in adults selected through insurer claims modeling. Investigators observed no significant difference in emergency department visits, inpatient hospitalizations, or total plan cost during follow-up. In this trial, the program was not associated with measurable reductions in acute care use or spending.

This national pragmatic two-arm randomized clinical trial evaluated commercially insured adults aged 18 years or older, although the analytic sample excluded patients aged 65 years or older and those with Medicare as primary insurance. Eligibility required high-need, high-cost status from a proprietary claims model tied to the top 5% of spending in a rolling 12-month window. The model also projected persistent top-tier spending during the next 12 months, and the analytic sample included 93,379 patients with a mean age of 46 years; 54% were female. Patients were randomized monthly in a 60:40 ratio from January 2018 through October 2019 to a registered nurse–delivered telephonic care coordination program or usual care. The intervention combined medication review, assessment of barriers or clinical needs, urgent coordination, a case plan, and outreach over 60 days, reflecting a relatively low-intensity insurer-based model.

Prespecified coprimary outcomes were mean monthly inpatient hospitalizations, emergency department visits, and total plan cost, including medical and pharmacy spending, over 12 months. Emergency department visits were 0.033 versus 0.033, with a mean difference of 0, a 95% CI from -0.001 to 0.002, and P=.69. Inpatient hospitalizations were 0.009 versus 0.010, with a mean difference of 0.001, a 95% CI from 0 to 0.002, and P=.06. Total plan cost was $2507 versus $2568, with a mean difference of $60, a 95% CI from -$20 to $140, and P=.14. In the intention-to-treat analysis using multivariate linear regression with multiple-comparison adjustment, no significant reductions were seen in utilization or spending.

Secondary analyses likewise did not show statistically significant reductions in the diabetes subsample or in instrumental variable analyses. In the diabetes subsample, the cost difference was $37, with a 95% CI from -$121 to $196 and P=.64. Engagement reached 26% in the intervention arm, while 5% of control participants also engaged, and detailed clinical, process, and fidelity data after randomization were unavailable.

Investigators also noted fragmentation of care, difficulty identifying persistently high-need, high-cost patients, and some sensitivity analyses showing a statistically significant but not clinically significant increase in hospitalizations. In this commercially insured national program, the telephonic care coordination model was not associated with lower acute care use or total plan cost over 12 months.