Abstract
When a randomized evaluation finds null results, it is important to understand why. We investigated two very different explanations for the finding from a randomized evaluation that the Camden Coalition's influential care management program-which targeted high-use, high-need patients in Camden, New Jersey-did not reduce hospital readmissions. One explanation is that the program's underlying theory of change was not right, meaning that intensive care coordination may have been insufficient to change patient outcomes. Another explanation is a failure of implementation, suggesting that the program may have failed to achieve its goals but could have succeeded if it had been implemented with greater fidelity. To test these two explanations, we linked study participants to Medicaid data, which covered 561 (70 percent) of the original 800 participants, to examine the program's impact on facilitating postdischarge ambulatory care-a key element of care coordination. We found that the program increased ambulatory visits by 15 percentage points after fourteen days postdischarge, driven by an increase in primary care; these effects persisted through 365 days. These results suggest that care coordination alone may be insufficient to reduce readmissions for patients with high rates of hospital admissions and medically and socially complex conditions.