Patient-reported quality of life and depression in women with ovarian cancer - why ignoring the patterns of missing data can bias research conclusions — ASN Events
  1. Schedule
  2. Best of the Best Orals - Epidemiology
  3. Patient-reported quality of life and depression in women with ovarian cancer - why ignoring the patterns of missing data can bias research conclusions

Patient-reported quality of life and depression in women with ovarian cancer - why ignoring the patterns of missing data can bias research conclusions (#53)

Rebecca Mercieca-Bebber 1 2 , Melanie Price 1 3 , Melanie L Bell 1 4 , Madeleine King 1 2 , Penelope Webb 5 , Phyllis Butow 1 3
  1. Psycho-oncology Co-operative Research Group, University of Sydney, NSW, Australia
  2. Central Clinical School, Sydney Medical School, The University of Sydney, NSW, Australia
  3. Centre for Medical Psychology and Evidence-Based Decision-Making, University of Sydney, NSW, Australia
  4. Mel and Enid Zuckerman College of Public Health, University of Arizona, Tucson, Arizona, USA
  5. QIMR Berghofer Medical Research Institute, Brisbane, QLD, Australia

Aims
Patient-reported outcomes (PROs) are patients’ self-reports of symptoms, functioning or multi-dimensional constructs, including health-related quality of life (HRQOL). A major barrier to high-quality PRO data collection is monotone missing data; i.e. where patients stop completing scheduled PRO questionnaires during the study due to study attrition or some other reason. Previous research has shown that cancer patients with worse HRQOL overtime are more likely to drop out. To test the hypothesis that ovarian cancer patients with worse baseline PRO scores would drop out earlier, we examined how patients differed by time of dropout on HRQOL, depression, anxiety, optimism and insomnia.

Methods
Six-hundred-and-nineteen participants with Stage II-IV ovarian cancer who participated in the population-based Australian Ovarian Cancer Study (AOCS) Quality of Life (QOL) sub-study were included in our analysis. Participants completed eight PRO assessments over 21 months. Participants were stratified by time of dropout and Pearson r correlations were calculated to examine the relationship between time of dropout and mean HRQOL (FACT-O questionnaire), depression (HADS), anxiety (HADS), optimism (LOT-R) and insomnia (ISI).

Results
Participants who dropped out earlier had significantly worse baseline HRQOL r=.20, p<.0001, 95% CI [0.13, 0.28] and higher depression scores r=-.17,p<.0001, 95% CI [-0.24, -0.09]. HRQOL and depression scores declined more rapidly over time in dropouts than for participants who remained in the study longer. Similar, but not statistically significant patterns were evident for anxiety, and to a lesser extent, optimism. There was no relationship between insomnia and dropout.

Conclusions
Poorer HRQOL and greater depression at baseline predicted time of dropout in this sample of ovarian cancer patients. These results highlight the potential for bias if complete case analysis is used for PRO data, as well as the importance of collecting auxiliary data to inform careful and considered handling of missing PRO data during analysis, interpretation and reporting.

#COSA2015