Table 3

Hierarchical Modeling of the Association Between Team Social Networks and Outcomes for Panel of Patients With Cardiovascular Disease

VariableBP Control OR (95% CI)LDL Cholesterol Control OR (95% CI)UC Visits RR (95% CI)ED Visits RR (95% CI)Hospital Days RR (95% CI)Cost ($)β (SE)
Reduced form model
 Team social network densitya1.14 (0.98–1.33)1.13 (0.97–1.31)0.28b (0.19–0.40)0.58c (0.42–0.80)0.64b (0.51–0.80)−516b (129)
 Team social network centralizationa1.00 (0.84–1.19)0.94 (0.80–1.11)3.07b (2.09–4.51)1.70c (1.23–2.35)1.47c (1.14–1.89)519c (143)
Full modeld
 Team social network densitya1.15 (0.99–1.34)1.14 (1.00–1.31)0.95 (0.55–1.66)0.98 (0.50–1.89)0.62b (0.50–0.77)−556b (115)
 Team social network centralizationa1.03 (0.85–1.25)0.93 (0.79–1.08)1.20 (0.79–1.81)1.33 (0.83–2.13)1.45c (1.09–1.94)506c (155)
  • BP = blood pressure; ED = emergency department; LDL = low-density lipoprotein; OR = odds ratio; RR = rate ratio; SE = standard error; UC = urgent care.

  • Note: There were 31 primary care teams and 7,457 patients who had cardiovascular disease.

  • a Team social network interaction related to daily face-to-face and/or EHR communication connections about patient care in a team.

  • b P <.001.

  • c P <.01.

  • d Patient-level covariates entered in the full model were sex, age, age squared, race/ethnicity, insurance, and comorbidity.