Skip to main content

Table 5 Interaction analyses assessing differential weight change from baseline to 6 months between the face-to-face and digital-only cohorts. Subgroup effects are shown, where relevant

From: The effectiveness of digital delivery versus group-based face-to-face delivery of the English National Health Service Type 2 Diabetes Prevention Programme: a non-inferiority retrospective cohort comparison study

 

Change in weighta (n=31064)

B

95% CI

p-value

Sex

 Male (ref)

0.472

(-0.055, 0.998)

 

 Female

-0.623

(-1.116, -0.131)

 

 Interaction

-1.095

(-1.588, -0.602)

<0.001

Age at referral

 Interaction

0.012

(-0.011, 0.034)

0.305

Ethnicityb

 White (ref)

-0.396

(-0.879, 0.086)

 

 Mixed

-0.403

(-2.473, 1.666)

 

 Asian

0.443

(-0.383, 1.269)

 

 Black

1.010

(-0.133, 2.152)

 

 Other

-2.864

(-8.202, 2.474)

 

 Interaction (Mixed)

-0.007

(-2.084, 2.069)

0.994

 Interaction (Asian)

0.839

(-0.026, 1.704)

0.057

 Interaction (Black)

1.406

(0.232, 2.580)

0.019

 Interaction (Other)

-2.468

(-7.813, 2.877)

0.366

IMDc

 1 (most deprived) (ref)

-0.519

(-1.273, 0.236)

 

 2

0.334

(-0.302, 0.971)

 

 3

-0.164

(0.767, 0.439)

 

 4

-0.511

(-1.191, 0.169)

 

 5 (least deprived)

-0.016

(-0.807, 0.775)

 

 Interaction (2)

0.853

(0.007, 1.698)

0.048

 Interaction (3)

0.355

(-0.485, 1.195)

0.408

 Interaction (4)

0.008

(-0.887, 0.902)

0.987

 Interaction (5)

0.503

(-0.488, 1.494)

0.320

  1. How to read the table – using example of sex: Males, in the face-to-face cohort lost, on average, 0.472kg more than males in the digital-only cohort, and females, in the face-to-face cohort lost, on average, 0.623kg less than females in the digital-only cohort. The interaction effect of -1.095 is the difference between the effect in females and the effect in males. As the interaction effect is statistically significant (p<0.001), there is strong evidence that the difference in change in weight between the digital-only and face-to-face cohorts varies by sex. Furthermore, the 95% confidence intervals around the subgroup effects provide evidence that females lose more weight on the digital programme but, as the 95% confidence interval around the effect for males contains 0, there is insufficient evidence to suggest males lost more weight on one programme than the other
  2. aAll models adjust for age at referral, sex, ethnicity (white/mixed/black/Asian/other), IMD quintile, time since baseline (in months) as fixed effects and CCG (site - Clinical Commissioning Group) nested within STP (Sustainability and Transformation Partnership) as random effects
  3. bAsian’ comprises those reporting Indian, Pakistani, Bangladeshi, Chinese or ‘other Asian’ ethnicity; ‘Black’ comprises those reporting Caribbean, African or ‘other Black’ ethnicity; ‘Mixed’ comprises people with a Mixed ethnic background and ‘Other’ comprises those reporting any other ethnicity
  4. c‘IMD’ – Index of Multiple Deprivation (English, 2015)
Back to article page

BMC Health Services Research

ISSN: 1472-6963

Contact us