Use and effectiveness of non-pharmaceutical interventions for mild to moderate depression

Date of Approval
Application Number
20_000187
Technical Summary

This study focuses on the patterns of use and real-life effectiveness of non-pharmaceutical interventions in UK primary care. Previous studies investigating the health effects of non-pharmaceutical interventions administered to people diagnosed with non-severe depression are largely limited to controlled clinical trials and small-scale observational studies. First, this study explores the use of non-pharmaceutical interventions in UK primary care in adherence to current guidelines for non-severe depression, for which they are recommended as a first line option, by studying the proportion of patients who received a non-pharmaceutical intervention upon being diagnosed with non-severe depression. Second, while previous studies have pointed towards the efficacy of non-pharmaceutical interventions in improving key outcomes, they might not be able to fully capture treatment effectiveness during routine care and frequently lack the time horizon to study long-run benefits and risks. Thus, we aim to explore the use of non-pharmaceutical interventions on short-, mid-, and long-term physical and behavioural outcomes in a routine care set-up. For these analyses, we apply an advanced matching technique relying on machine learning - DAME (developed by the Duke Almost Matching Exactly Lab) - which is built to learn the proper metric to be used by weighting those covariates more which directly contribute towards the treatment effect more. Upon matching our treatment and control groups and to determine whether outcomes of patients differ between patients in each subgroup, we apply mean comparison tests. Finally, we test for heterogenous treatment effects by stratifying our sample by sex, age, socioeconomic status, ethnicity and comorbidities. Multiple testing is accounted for using the Benjamini-Hochberg method to take into consideration the number of subgroups and primary outcomes. The findings of this study are expected to provide novel insights into the use and effectiveness of non-pharmaceutical interventions in a real-life setting and can directly inform clinical practice.

Health Outcomes to be Measured

We will explore the effects on the following primary outcome:
- Persistence of non-severe depression (i.e., repeated depression diagnosis and/or uptake of medication, see section ‘Exposures, outcomes and covariates’)

We will also explore the following secondary outcomes
- Number of all-cause emergency hospitalizations
- Number of severe adverse health events (stroke, heart attack [myocardial infarction] – each event type evaluated separately)
- All-cause mortality
- Suicide
- Number of GP visits
- Number of medications
- BMI
- Blood pressure
- Smoking status
- Recorded hazardous alcohol consumption

To address missingness of data, we will use multiple imputation for missingness in the covariates and pairwise deletion for the outcome variables (see relevant section below for more detail).

Collaborators

Till Bärnighausen - Chief Investigator - University of Heidelberg
Julia Lemp - Corresponding Applicant - University of Heidelberg
Anant Jani - Collaborator - University of Oxford
Justine Davies - Collaborator - University of Birmingham
Maximilian Schuessler - Collaborator - University of Heidelberg
Pascal Geldsetzer - Collaborator - University of Heidelberg

Linkages

2011 Rural-Urban Classification at LSOA level;HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation