Skip to main content

Table 3 Efficacy outcomes

From: Improving adherence in mental health service users with severe mental illness in South Africa: a pilot randomized controlled trial of a treatment partner and text message intervention vs. treatment as usual

Outcome

n

%

Risk ratio (ITT)

p value

95% CI

Unadjusted

(n = 77)

Adjusteda

(n = 77)

Intention-to-treat analysis (ITT): non-adherence to first clinic visit, re-admission over 9 months

 Non-adherence to first clinic appointment

  Intervention (n = 42)

14

33.3

0.72

0.79

0.419

0.44 to 1.39

  Treatment as usual (n = 35)

16

45.7

–

–

–

–

 Any re-admission over 9 months

  Intervention (n = 42)

5

11.9

0.83

0.86

0.713

0.39 to 1.87

  Treatment as usual (n = 35)

5

14.3

–

–

–

–

Outcome

Complete case analysisc (intervention vs. TAU)

ITT (intervention vs. TAU)

Mean differenced

p value

95% CI

Mean differencee (n = 77)

p value

95% CI

Unadjusted

Adjusted

Unadjusted

Adjusted

Complete case and intention to treat analysis (ITT) of other efficacy outcomes at 3 months

 PANSS score

  Total score

− 9.4

− 14.7

0.052

− 29.71 to 0.16

− 13.4

− 13.1

0.062

− 27.00 to 0.73

  Positive subscale

− 3.8

− 6.4

0.011

− 11.20 to 1.60

− 5.6

− 5.4

0.060

− 11.16 to 0.25

  Negative subscale

− 2.6

−4.4

0.059

− 8.99 to 0.18

− 3.5

− 3.5

0.078

− 7.52 to 0.43

  General subscale

− 2.8

− 3.9

0.350

− 12.6 1 to 4.68

− 4.4

− 4.2

0.248

− 11.67 to 3.19

 MARS

− 0.21

− 0.75

0.425

− 2.68 to 1.17

0.36

0.49

0.603

− 1.44 to 2.43

 CGI

− 0.8

− 0.58

0.346

− 1.84 to 0.67

–

–

–

–

 GAF

7.5

4.1

0.440

− 6.90 to 15.17

–

–

–

–

 CAN

    

–

–

–

–

  Total needs

–

–

–

–

–

–

–

–

  Unmet needs

− 3.2

− 3.6

0.029

− 6.74 to 0.49

–

–

–

–

  Met needs

–

–

–

–

–

–

–

–

EUROQUEL-VAS

16.1

15.2

0.124

− 4.59 to 34.99

–

–

–

–

  1. aLog-Poisson regression model with robust variance estimation, ITT; adjusted for age, sex, substance use disorder, baseline scores of PANSS total score, GAF score, MARS, EUROQUEL VAS scale, CAN unmet needs score
  2. bLog-Poisson regression model with robust variance estimation, ITT; sex dropped from model, adjusted for adjusted for age, substance use disorder, baseline scores of PANSS total score, GAF score, MARS, EUROQUEL VAS scale, CAN unmet needs score. Baseline data were imputed
  3. cMultiple linear regression models adjusted for age, sex, substance use disorders, baseline scores of PANSS total score, GAF, EUROQUEL, MARS, CAN unmet needs. Models with violation of linear regression assumptions omitted
  4. dSample size varied due to list-wise deletion
  5. eOnly models for which missing data < 55% are reported