Relationship Between TCPI Practice Progress in Transformation and Changes in ED Visits for Medicare FFS Beneficiaries in the 2 Follow-Up Years
| PAT Score or Domain | Percentage Change in ED Visits by Follow-Up Year 1a | Percentage Change in ED Visits by Follow-Up Year 2a Practices That Had Implemented | |||
|---|---|---|---|---|---|
| All Practices | All Practices | <10% at First PAT | 10% to 49% at First PAT | >50% at First PAT | |
| Primary care practices | |||||
| Number of TCPI practicesb | 1,567 | 1,067 | 168 | 561 | 338 |
| Baseline mean of ED visits (per 1,000 beneficiaries per year)b | 549 | 549 | 509 | 545 | 555 |
| Overall PAT score, % | −2 | −6c | 2 | −6c | −7d |
| Patient and family engagement, % | <1 | −1 | −2 | −2 | <1 |
| Population management, % | −1 | −5c | −4 | −5c | −5e |
| Practice as a community partner, % | −<1 | −<1 | −<1 | −<1 | −2 |
| Coordinated care, % | −<1 | −3e | −9 | −2 | <1 |
| Enhanced access, % | <1 | −3e | −4 | −4 | −2 |
| Behavioral health integration, % | −2 | −4e | <1 | −6d | −2 |
| Measuring and documenting value and QI strategy, % | −3d | −5c | −4 | −5c | −6e |
| Committed and engaged leadership, % | −<1 | −<1% | −<1 | <1 | −7 |
| Specialty care practices | |||||
| Number of TCPI practicesb | 2,397 | 1,016 | 121 | 539 | 356 |
| Baseline mean of ED visits (per 1,000 beneficiaries per year)b | 539 | 506 | 365 | 501 | 516 |
| Overall PAT score, % | −<1 | −4e | −3 | −1 | −9e |
| Patient and family engagement, % | <1 | −2 | <1 | −3 | −4 |
| Population management, % | −<1 | −5d | −<1 | −5d | −7e |
| Practice as a community partner, % | <1 | <1 | 1 | −3 | −3 |
| Coordinated care, % | 3d | −2 | <1 | −4e | −1 |
| Enhanced access, % | 1 | −3e | −4 | <1 | −6d |
| Measuring and documenting value and QI strategy, % | 1 | −2 | −4 | <1 | −11c |
| Committed and engaged leadership, % | −<1 | 2 | 4e | −5 | 3 |
ED = emergency department; FFS = fee-for-service; PAT = practice assessment tool; QI = quality improvement; TCPI = Transforming Clinical Practice Initiative.
Notes: Sources are author’s analysis of PAT data and Medicare administrative data. This table shows the estimated percent change in ED visits in follow-up years 1 and 2 associated with a 40-percentage point increase in the overall PAT score or domain score between the first and last PAT submission. Each row corresponds to a separate model. We used 1 regression for the last 3 columns, in which we allowed the association to vary based on the extent of each practice’s implementation of the Change Package at the time of first PAT. Because we excluded practices that had fully implemented processes measured in the overall PAT or domain at the first PAT, each regression could have a different set of practices in the sample. Statistically significant favorable estimates are shown in green cells with bolded text, and statistically significant unfavorable estimates are shown in orange cells with italicized text. To interpret the estimates, for example, the results on the “Overall PAT” row in the last columns under the “Primary care practices” panel indicates that a 40-percentage point change in the overall PAT score was associated with a 6% reduction in ED visits for practices that had implemented 10-49% of the Change Package at the time of their first PAT.
↵a We calculated the percentage change estimates by dividing the estimated coefficient by the baseline mean outcome from each respective regression.
↵b The sample sizes and the baseline means shown are based on the set of practices included in the analysis for the overall PAT (that is, excluding those that scored 100% on the first PAT).
↵c The underlying estimate (which is visits per 1,000 beneficiaries per year) was statistically significant at the 0.01 level, 2-tailed test.
↵d The underlying estimate (which is visits per 1,000 beneficiaries per year) was statistically significant at the 0.05 level, 2-tailed test.
↵e The underlying estimate (which is visits per 1,000 beneficiaries per year) was statistically significant at the 0.10 level, 2-tailed test.