Abstract
Background: Although the risk of non-response bias is a serious concern for retrospective surveys, there is little empirical evidence on the extent of this problem, particularly in the context of end-of-life care research in nursing homes using proxy respondents.
Methodology: Both the Dying Well with Dementia study (DWD, 2010) and the ongoing PACE study 1 (2015) used randomized cluster-sampling to identify a representative sample of nursing homes in Flanders, Belgium. In PACE, all residents of these facilities who died over a 3 month period were reported (n=328); in DWD, only residents with dementia were reported (n=205). For each resident, the nurse and relative most involved in care and the general practitioner (GP) were given a structured questionnaire. Demographic and care-related characteristics of the resident extracted from the nurses’ questionnaire (response rate DWD: 88%, PACE: 80%) were used to compare residents for whom the GP and/or relative did not complete the questionnaire with those for whom they did.
Results: Response rates for relatives and GPs were 53% in DWD and 61% for relatives and 67% for GPs in PACE. In DWD, there were no differences in demographic characteristics, stage of dementia or nursing care received between residents for whom the GP and/or the relative did or did not complete the questionnaire. Nurses perceived more consensus on care among relatives in cases where the GP completed a questionnaire (p=.004) than in cases where the GP did not. Non-response of relatives was not related to this characteristic. Results of PACE will be presented at the conference.
Conclusion: For both GPs and relatives there appears to be no non-response bias based on resident and care characteristics, though non-response by GPs may lead to bias based on consensus among relatives. These results suggest that non-response bias is not a barrier for the use of retrospective studies in research on end-of-life care in nursing homes.
PACE funded by EUFP7 GA no 603111.
Methodology: Both the Dying Well with Dementia study (DWD, 2010) and the ongoing PACE study 1 (2015) used randomized cluster-sampling to identify a representative sample of nursing homes in Flanders, Belgium. In PACE, all residents of these facilities who died over a 3 month period were reported (n=328); in DWD, only residents with dementia were reported (n=205). For each resident, the nurse and relative most involved in care and the general practitioner (GP) were given a structured questionnaire. Demographic and care-related characteristics of the resident extracted from the nurses’ questionnaire (response rate DWD: 88%, PACE: 80%) were used to compare residents for whom the GP and/or relative did not complete the questionnaire with those for whom they did.
Results: Response rates for relatives and GPs were 53% in DWD and 61% for relatives and 67% for GPs in PACE. In DWD, there were no differences in demographic characteristics, stage of dementia or nursing care received between residents for whom the GP and/or the relative did or did not complete the questionnaire. Nurses perceived more consensus on care among relatives in cases where the GP completed a questionnaire (p=.004) than in cases where the GP did not. Non-response of relatives was not related to this characteristic. Results of PACE will be presented at the conference.
Conclusion: For both GPs and relatives there appears to be no non-response bias based on resident and care characteristics, though non-response by GPs may lead to bias based on consensus among relatives. These results suggest that non-response bias is not a barrier for the use of retrospective studies in research on end-of-life care in nursing homes.
PACE funded by EUFP7 GA no 603111.
Original language | English |
---|---|
Article number | P84 |
Pages (from-to) | NP114-115 |
Number of pages | 2 |
Journal | Palliative Medicine |
Volume | 30 |
Issue number | 6 |
Publication status | Published - 11 May 2016 |
Event | EAPC Research Congress 2016: 9th World Research Congress of the European Association for Palliative Care - Dublin, Dublin, Ireland Duration: 9 Jun 2016 → 11 Jun 2016 http://www.eapcnet.eu/research2016/ |