Meta-analysis of non-occupational physical activity and cardiovascular disease, cancer and mortality outcomes

Short URL = https://t.ly/z5SO

Ali Abbas

MRC-Epidemiology Unit, University of Cambridge

May 5, 2023

Structure

Agenda

  • Meta-Analysis
  • Physical Activity (PA) and its relationship with diseases/mortality
  • Outcomes of interest
  • Interactive Interface
  • Code to access the dose-response curves

Meta-Analysis

What is meta-analysis?

  • Statistical technique for combining data from multiple studies on a particular topic.
  • Synthesizing evidence and summarizing the results of individual studies

The Evidence-Based Medicine Pyramid

https://uk.cochrane.org/news/meta-analysis-what-why-and-how

Diseases and PA

  • Globally
    • Cardiovascular diseases cause 17.9 million deaths per year in 2019
    • Cancers are responsible for 9.6 million deaths in 2017
  • Physical Activity is known to reduce the risk of cardiovascular diseases, cancers and early deaths

Physical Activity Meta-Analysis

  • The shape of the dose-response curves has not been established for a range of chronic diseases (@garcia2023non)
  • Key finding: 1 in 10 premature deaths can be avoided if everyone achieves 75 minutes of vigorous-intensity activity a week
  • Higher level of PA is associated with lower risk of all outcomes dose-response paper

Eligibility Criteria

  • Prospective cohort studies with general population samples of greater than 10K adults
  • 3 or more physical activity categories
  • Risk measures and Confidence Interval (CIs) for host of measures/disease end points

Studies

  • 196 studies
  • 94 cohorts
  • > 30 Million participants

Interactive Interface

Outcomes - overall

https://shiny.mrc-epid.cam.ac.uk/meta-analyses-physical-activity/

Outcomes - Level 1

https://shiny.mrc-epid.cam.ac.uk/meta-analyses-physical-activity/

Outcomes - Level 2

https://shiny.mrc-epid.cam.ac.uk/meta-analyses-physical-activity/

Outcomes - Level 3

https://shiny.mrc-epid.cam.ac.uk/meta-analyses-physical-activity/

Outcome category and type

https://shiny.mrc-epid.cam.ac.uk/meta-analyses-physical-activity/

Meta-analysis and dose-range

https://shiny.mrc-epid.cam.ac.uk/meta-analyses-physical-activity/

Dose distribution and Potential Impact Fraction (PIF)

https://shiny.mrc-epid.cam.ac.uk/meta-analyses-physical-activity/

Code

R Package

https://meta-analyses.github.io/drpa/drpa.html

  • An R package called dose-response physical activity (drpa) is created
  • It is open source and hosted on GitHub
  • You may need to install it used remotes::install_github("meta-analyses/drpa")
# Specify the cause, outcome_type and the dose
rr <- drpa::dose_response(cause = cause_name, 
        outcome_type = outcome_type, 
        dose = float)

R Package use

https://meta-analyses.github.io/drpa/drpa.html

  • An R package called dose-response physical activity is created
  • It is open source and hosted on GitHub . . .
  • You may need to install it used remotes::install_github("meta-analyses/drpa")
# Specify the cause, outcome_type, dose and the confidence interval
rr <- drpa::dose_response(cause = "all-cause-mortality", 
        outcome_type = "fatal", 
        dose = 40,
        confidence_interval = TRUE)
An example output
Relative Risk Lower Bound Upper Bound
0.68 0.63 0.72

R Package arguments

https://meta-analyses.github.io/drpa/drpa.html

Arguments for the dose_response function

Lookup tables

Download from the Interactive Interface

Lookup table for All-cause mortality

https://shiny.mrc-epid.cam.ac.uk/meta-analyses-physical-activity/

Lookup on GitHub repo

Lookup tables for all the outcomes

https://github.com/meta-analyses/drpa/tree/master/inst/extdata

Lookup on GitHub repo for All-cause-mortality

Lookup tables for all the outcomes

https://github.com/meta-analyses/drpa/blob/master/inst/extdata/all-cause-mortality-fatal.csv

Access the presentation by using https://t.ly/z5SO