A 'missing not at random' (MNAR) and 'missing at random' (MAR) growth model comparison with a buprenorphine/naloxone clinical trial. (16th October 2014)

• Record Type: Journal Article
• Authors: McPherson, Sterling; Barbosa‐Leiker, Celestina; Mamey, Mary Rose; McDonell, Michael; Enders, Craig K.; Roll, John
• Is Part Of: Addiction. Volume 110:Number 1(2015:Jan.)
• Page Start: 51
• Page End: 58
• Publication Date: 2014-10-16
• Subjects: Alcoholism -- Periodicals; Drug addiction -- Periodicals; 616.86
• DOI: 10.1111/add.12714
• Languages: English
• ISSNs: 0965-2140

Abstract

<abstract abstract-type="main"> <title>Abstract</title> <sec id="add12714-sec-0001" sec-type="section"> <title>Aims</title> <p>To compare three missing data strategies: (i) the latent growth model that assumes the data are missing at random (MAR) model; (ii) the Diggle–Kenward missing not at random (MNAR) model, where dropout is a function of previous/concurrent urinalysis (UA) submissions; and (iii) the Wu–Carroll MNAR model where dropout is a function of the growth factors.</p> </sec> <sec id="add12714-sec-0002" sec-type="section"> <title>Design </title> <p>Secondary data analysis of a National Drug Abuse Treatment Clinical Trials Network trial that examined a 7‐day versus 28‐day taper (i.e. stepwise decrease in buprenorphine/naloxone) on the likelihood of submitting an opioid‐positive UA during treatment.</p> </sec> <sec id="add12714-sec-0003" sec-type="section"> <title>Setting</title> <p>11 out‐patient treatment settings in 10 US cities.</p> </sec> <sec id="add12714-sec-0004" sec-type="section"> <title>Participants</title> <p>A total of 516 opioid‐dependent participants.</p> </sec> <sec id="add12714-sec-0005" sec-type="section"> <title>Measurements</title> <p>Opioid UAs provided across the 4‐week treatment period.</p> </sec> <sec id="add12714-sec-0006" sec-type="section"> <title>Findings</title> <p>The MAR model showed a significant effect (B = −0.45, <italic>P</italic> < 0.05) of trial arm on the opioid‐positive UA slope (i.e. 28‐day taper participants were less likely to submit a positive UA over time) with a small effect size (<italic>d</italic> = 0.20). The MNAR Diggle–Kenward model demonstrated a significant (B = −0.64, <italic>P</italic> < 0.01) effect of trial arm on the slope with a large effect size (<italic>d</italic> = 0.82). The MNAR Wu–Carroll model showed a significant (B = −0.41, <italic>P</italic> < 0.05) effect of trial arm on the UA slope that was relatively small (<italic>d</italic> = 0.31).</p> </sec> <sec id="add12714-sec-0007" sec-type="section"> <title>Conclusions</title> <p>This performance comparison of three missing data strategies (latent growth model, Diggle–Kenward selection model, Wu–Carrol selection model) on sample data indicates a need for increased use of sensitivity analyses in clinical trial research. Given the potential sensitivity of the trial arm effect to missing data assumptions, it is critical for researchers to consider whether the assumptions associated with each model are defensible.</p> </sec> </abstract>