Skip to main content

Table 3 Main Results from Generalized Estimating Equations modeling (with Interaction Terms) for the depression and anxiety datasets (note: therapist and organization fixed effects included but are not detailed due to presentation complexity)

From: Action recommendations review in community-based therapy and depression and anxiety outcomes: a machine learning approach

Variable

PHQ9

GAD7

Coefficient (p-value)

Coefficient CI [0.025 0.975]

Coefficient (p-value)

Coefficient CI [0.025 0.975]

Intercept

-1.93 (ns)

[-6.40, 2.55]

-2.07 (ns)

[-5.56, 1.41]

Log (Review Percentage)

0.61 (0.032*)

[0.05, 1.17]

0.23 (ns)

[-0.11, 0.56]

Number of Dialogues/Session

0.04 (0.007*)

[0.01, 0.07]

0.01 (ns)

[-0.01, 0.04]

Number of Action Recommendations/Previous Session

0.01 (ns)

[-0.05, 0.07]

0.03 (ns)

[-0.04, 0.09]

Client's Initial Score

0.09 (0.006*)

[0.03, 0.015]

0.16 (0.000*)

[0.09, 0.22]

Days Examined

0.02 (ns)

[-0.06, 0.11]

0.02 (ns)

[-0.04, 0.08]

Log (Review Percentage): Initial Score

-0.02 (.045*)

[-0.04, -0.00]

-0.02 (ns)

[-0.05, 0.004]

Log (Review Percentage): Number of Dialogues/Session

-0.01 (ns)

[-0.03, 0.01]

NA

NA

  1. * denotes p < 0.05, "ns" denotes lack of statistical significance, “NA” denotes non-applicable. Variables transformed: review percentage logged, score change cube-rooted (the dependent variable). All significant associations were verified via bootstrap resampling (1000 samples)