A mortality follow-back survey was conducted. Potential respondents were the next-of-kin listed as the contact on the death certificates for decedents whose recorded cause of death was dementia. Death certificates from 2006/07 were sampled, with the interview conducted on average 23.8 months after the nursing home resident's death. States were sampled based on the following criteria: prevalence of feeding tube use, inclusion of different geographic regions, variety of minority representation, and lack of restriction in access to death certificates for research purposes. Based on previous work,11
states were purposely sampled based on the prevalence of feeding tubes in nursing homes. Two states were selected for their lower rates (MN and MA), and three were selected for having higher feeding tube prevalence (AL, FL, TX). In Texas and Florida, only hospital referral regions with the highest rates of feeding tube prevalence were sampled.
Family members were interviewed only if they stated that they were the person who knew the decedent best and that they were or would have been involved in medical decision-making. To qualify for this survey, the family member reported that a feeding tube was inserted, a decision was made not to insert a feeding tube, or the family member reported at least one of three risk factors: an eating problem that resulted in the patient not taking enough nourishment, choking on food, and visible weight loss from not eating enough food.
A total of 1,111 death certificates were sampled. Four hundred fifty cases were excluded because the next-of-kin could not be contacted (n = 277), the next-of-kin was ineligible because they could not communicate in English or Spanish (n = 50), or the next-of-kin stated that the decedent did not need assistance in eating (n = 64) or did not have at least one of the risk factors noted above (n = 59). Of the remaining 661 next-of-kin, 486 (73.5%) agreed to participate in the survey. The survey was conducted by telephone and took on average of 39.1 ± 20.6 minutes.
The survey asked about feeding tube decision-making, the outcomes of feeding tubes, and bereaved family members’ perceptions of the quality of end-of-life care. Decision-making regarding feeding tube insertion was based on modified questions from a previously developed survey of feeding tube decision-making8
and from the Toolkit of Instruments to Measure End of Life Care.13
The outcomes of feeding tube insertion focused on whether the feeding tube bothered the patient, whether physical and pharmacological restraints were used, whether artificial feeding was withdrawn, and overall satisfaction with end-of-life care. When decedents had feedings stopped without any other nourishment or fluids, the family members were asked how long the patient survived after feedings were stopped and about the family's distress in the patient's last days of life. For the purpose of this analysis, a single item was used to examine the quality of end-of-life care that asked the respondent to rate the overall quality of care (excellent, very good, good, fair, and poor.
The frequency of decisions, quality of decision-making, and outcome of feeding tube insertions are described. Analyses were weighted to account for the oversampling of minorities and nonresponse bias. To examine whether respondents of dementia decedents with a feeding tube reported higher ratings of quality of care, a multivariate logistic model used the weighted data to examine this association after adjustments for age, sex, race, education, use of hospice services, and nursing home organizational characteristics (whether the nursing home was hospital based, had an Alzheimer's unit or special dementia unit, and had disproportionate share of Medicaid patients). Analyses were completed in Stata 10.0 (Stata Corp, College Station, TX).