Deaths among children are rare, but the effect on family members is profound. Compared with adult deaths, information about grief, recovery, and quality of care is sparse.
To describe aspects of bereavement for parents who had experienced the death of a child and to compare these aspects by parent sex, type of death, and overall experience.
In-person interviews with families, primarily parents, a mean of 21.8 months after the child's death.
Academic, tertiary care, faith-based children's hospital.
Fifty-nine child deaths and 79 parents or guardians.
Main Outcome Measures
In-person interviews, including standard instruments for bereavement and quality of care: the Texas Revised Inventory of Grief and the Comprehensive Assessment of Satisfaction With Care–Short Form.
Fathers and mothers had similar levels of grief. Mothers who experienced the sudden death of their child had somewhat more intense grief reactions than those whose child died of a chronic condition. Grief scores did not vary according to satisfaction with treatment. Comprehensive Assessment of Satisfaction With Care–Short Form scores were high for parents and similar between mothers and fathers and between sudden and unexpected deaths.
Although there were some differences in grief responses among parents, satisfaction-with-care scores were high. Further studies should examine the role of satisfaction with care in parental grief response and incorporate the reporting of experiences rather than simple ratings to measure satisfaction with care.
Nearly 53 000 children die each year in the United States1 leaving parents to cope with the death of their child. Approximately 19% of the adult population has experienced the death of a child.2 The literature suggests that parental reaction to child death is complex and poorly understood.3-8 Studies3,4,9 show that parents have more intense grief concerning the death of their child than the death of their partner or than adult children concerning the death of their parent. Several studies regarding parental grief at the death of a child have shown that mothers have a more intense and longer grief reaction than fathers7,8,10 and that parents of children who have a sudden, unexpected death have a stronger grief reaction than parents of children whose death is expected.3,11 However, these studies focused on parents with children in specific age groups such as neonatal death7,8 or adolescents,3,5 and some research included parents with adult children.9 To date, parental assessment of quality of care has shown that the primary concerns are pain management and communication issues for hospital care in general12-14 and for end-of-life care in particular.15 In addition, there is little quantitative research regarding parental experiences with treatment and assessment of satisfaction with their child's medical care10,15,16 and whether this has a relationship to the level of grief for their child.
To assess parental grief and satisfaction with their child's care, we used a qualitative phase to design an appropriate pro tocol and to choose among published validated instruments for the quantitative phase. There are no existing instruments to assess the quality and satisfaction with pediatric palliative care; therefore, we sought parents' opinions about existing surveys and instruments on general quality of life, end-of-life care, grief, and spirituality. Based on this first phase, we selected the Comprehensive Assessment of Satisfaction With Care–Short Form17 (CASC) questionnaire and the Texas Revised Inventory of Grief18 (TRIG) as appropriate surveys to use.19,20 We investigated the potential for differences in grief and assessment of quality of care by parent sex and by circumstances of death, sudden vs anticipated.
We conducted interviews of parents of children who died in our 162-bed faith-based teaching hospital from January 1, 1998, through December 31, 2000. The supervising university's institutional review board approved all study procedures. Not all families could be contacted, and among those we contacted, some families refused the interview or were not scheduled for an interview during the time frame of the data collection. From 1998 through 2000, 373 patients died. Among these families, 169 (45.3%) did not have current contact information, and repeated tracing efforts failed to locate a current address or telephone number. A small number of the child cases had been investigated in the legal system and were excluded, and some parents were willing to be interviewed but failed to keep an appointment (n = 25) or could not be scheduled during the funded research time frame. Of 193 families contacted about the interview, 59 (30.6%) were interviewed and 64 (33.2%) refused. Seventy-nine parents or legal guardians were interviewed. Based on advice from bereaved parents in the qualitative phase,19 we conducted interviews at a location chosen by the subjects: at their home (37 subjects [46.8%]), at our hospital office (25 subjects [31.6%]), at another neutral location (2 subjects [2.5%]), or by telephone (15 subjects [19.0%]). For analyses using a single parent, we combined mothers and other female guardians (2 aunts, 2 grandmothers, and 1 stepmother). Similarly, we included a case with a grandfather in the category of male guardian for sex comparisons. For analyses comparing sexes, we included 19 cases with paired responses from male and female parents or guardians. We excluded 1 pair of respondents who were a female guardian and a young mother.
There were 3 components to the interview. First, we asked demographic questions about the child and the parent respondent. Second, we administered 2 existing questionnaires (detailed in the next 2 subsections). Third, we asked a series of questions designed to elicit more in-depth experiences about the quality of the child's palliative care. Three of these questions were used to provide an overall indicator of experience with care (described herein). We also reviewed the medical records of children and coded the primary diagnosis that contributed to their death.
Texas revised inventory of grief
The TRIG contains 21 items in 2 scales.18,21 The Past Behavior scale contains 8 questions and the Present Feelings scale contains 13 questions (α = .82 for both scales in this sample). Each item is posed as a statement with 5 response categories, ranked from 1 ("completely true") to 5 ("completely false"). The items are summed for a total score for each scale, for which a higher score means less grief and a lower score means more grief. In addition, scores provide a grief classification scheme, with each scale ranked as low grief (below the 50th percentile) or high grief (above the 50th percentile).18 Respondents who had high grief on both scales were classified as having prolonged grief, and those with low grief on both scales were classified as having low grief (termed as "absence of grief P in the original TRIG). Low grief on the Past Behavior scale and high grief on the Present Feelings scale placed the respondent in the delayed grief group. Having high grief on the Past Behavior scale and low grief on the Present Feelings scale placed the respondent in the acute grief group. We excluded the participant's score if 3 or more items were missing for the Past Behavior scale and if 7 or more items were missing for the Present Feelings scale. For participants who were not excluded, we input the mean of other items of the scale to replace the missing items.
Comprehensive assessment for satisfaction with care–short form
The CASC17,22-24 contains 35 items and 10 scales. Four scales describe physicians, including Technical Skills (4 questions, α = .90 in this sample), Interpersonal Skills (4 questions), Information (3 questions, α = .61), and Availability (3 questions). Three of the scales are about nurses. These include Technical Skills (3 questions, α = .82), Information (3 questions, α = .90), and Availability (6 questions). The last 3 scales assess satisfaction as a Care Organization (5 questions, α = .71), Access and Comfort (3 questions, α = .54), and General Satisfaction (1 question). Each item has 5 rating response options, with 1 being "poor" and 5 being "excellent," so that a higher score means better care. To score each scale, we took the sum of the values for each question within that scale and divided by the number of questions in that scale. We excluded subjects if 2 or more items were not answered for those scales with 3 or 4 items and if 3 or more items were not answered for those scales with 5 or 6 items. The mean values were recalibrated as the mean of answered items in these cases.
Descriptive statistics of individual questions and scales included means and standard deviations or proportions. We computed 95% confidence intervals for proportions.25 We also compared scaled scores for groups defined by parent sex and the circumstances of the child's death. Our clinician colleagues classified child deaths as sudden or anticipated. Sudden deaths were those from trauma and deaths within 24 hours of birth or case presentation (eg, newborns). Deaths among children with a history of life-limiting illness were classified as anticipated. In post hoc analysis, we also examined differences between mothers grouped by their overall care experience (best possible vs less favorable). As described in more detail in the "Comment" section, the similarity of scores for the expected groups (parent sex and circumstances of death) suggested that the questionnaires might not discriminate between experiences, in part because of skewed (positive) responses on the CASC questionnaire. To determine factors affecting the quality of the parents' experience in cases in which the CASC suggested highly positive responses, we classified the experience as best possible or less favorable based on responses to the following 3 questions from the in-depth experiences questionnaire: (1) Would you recommend this hospital to relatives or friends [an overall marketing satisfaction question]? (2) Do you have confidence in the physician26,27? and (3) Were you able to identify one physician in charge of your child's care15,28-31? Respondents who answered yes to question 1, who rated their confidence as 1 or 2 (having a great deal or moderate amount of confidence) in response to question 2, and who answered no to question 3 were classified as having a best possible experience. We classified any other response to these questions as a less favorable experience.
For each analysis, we calculated scores, including all parent and guardian respondents, and then restricted these analyses to mothers or female guardians for each child in cases in which 2 parents answered questions. For sex comparisons, we analyzed the 19 matched sets in which male and female parents or guardians were interviewed for the same child. Except for 95% confidence intervals, analyses (descriptive statistics, t tests for mean comparisons, and χ2 for comparison of proportions) were conducted using SPSS.32
Seventy-nine parents or guardians completed the postdeath interview for the 59 children in the study. Interviews were conducted 7 to 39 months (mean, 21.8 ± 7.5 months) after the child's death. For 20 children, we had responses from 2 family members. However, one of these sets included 2 women (a young mother and a female guardian), who were not included in the matched analysis. In 50 cases (84.7%), the child's death was anticipated. Nearly half of the children were male, and two thirds were white. Twenty-seven percent of the children were aged 13 years and older and 40.7% were younger than 6 months at the time they died. Table 1 gives descriptive results of these children in more detail.
Characteristics of 59 Children Who Died and Their Families
The mean TRIG scores for all of the parents or guardians were in the middle of the TRIG score ranges, with a score of 22.9 for the Past Behavior scale and 33.9 for the Present Feelings scale, indicating intermediate grief levels. Scores for mothers and fathers separately were also in the middle of the TRIG score ranges for both scales. Mothers tended to have slightly lower levels of grief (higher scores) on both scales. Fathers were more likely to be in the prolonged grief group, and mothers were more likely to be in the low grief group. However, none of these differences were statistically significantly different. In addition, TRIG grief categories were not related to the time between the child's death and the interview. The time since death was similar for low vs high grief classifications for the Past Behavior scale (21.5 vs 22.0 months) and the Present Feelings scale (21.8 vs 21.7 months).
All parents demonstrated high levels of satisfaction with care, as shown by the scale scores of the CASC, and there were few differences between fathers and mothers. In instances in which some small differences occurred, mothers reported a higher level of satisfaction than fathers. Table 2 gives a complete report of these results.
Results of the TRIG and CASC Among Parents* of Children Who Died
In comparing responses of mothers whose child had a sudden death vs those whose child had an anticipated death, the sudden death of a child led to a stronger grief reaction for mothers in the Past Behavior and Present Feelings scales (18.6 vs 23.1 and 28.8 vs 35.5, respectively). In addition, 66.7% of mothers experiencing the sudden death of their child were in the prolonged grief group, compared with only 29.2% of mothers who anticipated the death of their child. Both groups of mothers gave high satisfaction-with-care scores for the items in the CASC, and the scores were similar between the 2 groups. From the analysis of groups based on best possible and less favorable experiences, the grief reactions were similar. Mothers who had a best possible experience also reported higher levels of satisfaction with care than mothers with a less favorable experience on all of the CASC scales. These higher levels of satisfaction were statistically significantly higher for 3 of 4 scales on physicians and the Care Organization scale. Table 2 summarizes these results.
Most parents in this study scored in the middle range of grief scores and in the high range of satisfaction-with-care scores. There were some small differences in grief scores by parental sex, circumstances of death (sudden vs anticipated), and type of experience (best possible vs less favorable); however, few of the score differences were statistically significant. In this study, mothers had a higher level of past grief and fathers had a higher level of present grief. Our research does not support the results of previous studies5-9 that have found that mothers have a more intense grief reaction than fathers. Mothers experiencing the sudden, unexpected death of a child have a more prolonged grief response than parents who anticipated the death of their child. These parents do not have a chance to psychologically prepare for their child's death,3,4 and that may explain the differences we observed. We are unaware of any research regarding parents' evaluation of their child's care as it relates to their grief responses, and there were no differences among respondents' assessments in our analyses. The intensity of grief over the loss of a child may supersede the circumstances of the death and perception of the quality of and satisfaction with medical treatment. The satisfaction-with-care scores were similar between mothers and fathers, and between mothers who experienced sudden and anticipated deaths. Despite the high level of satisfaction and overall similarities, mothers with a best possible experience had higher satisfaction scores than those who had a less favorable experience.
Research on quality-of-care issues has identified several areas as increasingly important to parents concerning the death of their child. These include pain management,12,15,28 adequate communication,12,15 and bereavement follow-up after death.15 Consumer health care satisfaction questionnaires have had problems with positive skewed responses, or ceiling effects,33-35 and in the development of new questionnaires, this is a central issue.36-39 When most responses are skewed toward positive reports, it is difficult to examine potential differences by group experiences. In particular, parent assessment of health care for their children is complex,40 comprising child and parent experiences. Cognitive evaluation of the problems in satisfaction questions has focused on the importance of asking people to report experiences, rather than asking global rating questions.41 Constructing scales for measuring satisfaction with care in the difficult situation of a child's death may require more theoretical, cognitive, and psychometric work before determining the important issues that affect the quality of palliative care for children.
This study has limitations. One limitation is that the results may not be fully generalizable to other pediatric hospitals. Results from a tertiary care center may be somewhat different than in other hospitals. However, these teaching hospitals are the primary institutions that give children the specialty care needed for serious injuries or life-limiting conditions. Our hospital was developing an end-of-life care program,19 which may have led to better response to pain management and other aspects of care by health care teams than in other hospitals that do not have this type of program. This may, in part, explain the overall high level of satisfaction and the positive ratings of provider care from the CASC. However, one third of the families we contacted refused the interview, and these parents may have had more negative care experiences. Finally, this study included interviews about a small number of children who died during the 3 years of the project. In addition to affecting the power of our study to detect statistical differences between our comparison groups, the results are probably not representative of all parents and guardians of children who die. Without data on these parents, it is difficult to suggest the potential nature of this biased sample. We also restricted group comparisons based on the type of death and overall experience to mothers and female guardians to avoid dependence of our observations, which excluded 20 (male) observations. However, there were no significant differences between men and women matched on child deaths. The difficulty of this research lies in garnering participation of people who are grieving over the death of their child. Also, it is difficult to get the participation of single-parent families or teenage mothers, who may be harder to locate after the death of a child. In the qualitative phase of this study, parents had advised that a minimum of 6 months was needed before contacting parents for interviews.19 This time frame was longer in our study, and it is possible that a much longer follow-up (and more funding) is needed to include higher levels of response. Although our level of participation of families is similar to that of other studies5,8,9,15,42 on this subject, more effort is required to include larger and more representative parents in future research.
Quality of care is an underresearched aspect of end-of-life care for children and families. Prior work has been more substantial for adult deaths, including finding conceptual domains for quality of dying based on evaluations of others.43 Future directions for research should include finding and evaluating end-of-life care domains for children and considering the lessons learned from the broader field of measuring consumer satisfaction with health care. In addition, more research should be done to evaluate if experiences of death, including quality of and satisfaction with care, affect grief reactions in a larger group of parents of deceased children. Enhancing the quality of end-of-life care has become a priority not only for patients and their families but also for health care professionals. This study may help evaluate the work of the health care team, thereby providing a way to improve pediatric end-of-life care.
Although parents experience substantial grief in the death of a child, it does not differ strongly by parental sex. There are some differences in grief reactions for sudden deaths vs those from chronic conditions. Parental satisfaction with the care of their children is generally high, but questionnaire responses may be skewed, suggesting the need to revise satisfaction instruments used in pediatric palliative care. A low response rate to this kind of research hampers generalizability of the results.
Corresponding author and reprints: Elena M. Andresen, PhD, Department of Community Health, School of Public Health, Saint Louis University, Salus Center, 3545 Lafayette Ave, Suite 300, St Louis, MO 63104 (e-mail: firstname.lastname@example.org).
Accepted for publication July 2, 2003.
This work was funded, in part, by grant 035489 from The Robert Wood Johnson Foundation's Promoting Excellence in End-of-Life Care, Missoula; by administrative support of the SSM Cardinal Glennon Children's Hospital; and by grant U48CCU710806 from the Centers for Disease Control and Prevention (CDC) Prevention Research Center at Saint Louis University.
We thank the many health care providers who gave their valuable time and experiences to this work. Our special thanks to our FOOTPRINTS team members: Judy Carron, RN; Patricia Codden, RN, PNP; May Ann Collins, MHA; Drew Gerald; Lisa Griebel; Upasana Nanda, MPH; Dennis O'Connor, MD; Kim Pacatte, RN; and Jenna Putzel.
KD Deaths: final data for 1999. Natl Vital Stat Rep.
2001;491- 113PubMedGoogle Scholar
N A longitudinal study comparing bereavement phenomena in recently bereaved spouses, adult children and parents. Aust N Z J Psychiatry.
1998;32235- 241PubMedGoogle ScholarCrossref
FE Life transitions of parents at the unexpected death of a school-age and older child. J Pediatr Nurs.
1993;841- 49PubMedGoogle Scholar
SB Similarities and differences in mothers' and fathers' grief following the death of an infant. Scand J Psychol.
1987;281- 15PubMedGoogle ScholarCrossref
SB Parental grief following the death of an infant: a follow-up over one year. Scand J Psychol.
1991;32193- 207PubMedGoogle ScholarCrossref
S Family satisfaction with end-of-life care for cancer patients in a cluster randomized trial. J Pain Symptom Manage.
2002;2453- 63PubMedGoogle ScholarCrossref
K Parental responses to sudden accidental death of a child. CCQ.
1985;873- 84Google Scholar
et al. Quality of care at a children's hospital: the parent's perspective. Arch Pediatr Adolesc Med.
1999;1531123- 1129PubMedGoogle ScholarCrossref
SJ Qualitative analysis of the care of children in hospital in four countries: part 1. J Pediatr Nurs.
2001;16137- 145PubMedGoogle ScholarCrossref
H Weaving the seamless web of care: an analysis of parents' perceptions of their needs following discharge of their child from hospital. J Adv Nurs.
2000;31812- 820PubMedGoogle ScholarCrossref
H Family perspectives on the quality of pediatric palliative care. Arch Pediatr Adolesc Med.
2002;15614- 19PubMedGoogle ScholarCrossref
I A measure of the quality of dying and death: initial validation using after-death interviews with family members. J Pain Symptom Manage.
2002;2417- 31PubMedGoogle ScholarCrossref
C A comprehensive assessment of satisfaction with care for cancer patients. Support Care Cancer.
1998;6518- 523PubMedGoogle ScholarCrossref
TR Texas Revised Inventory of Grief Manual. Houston, Tex Honeycomb Publishing Co1981;
EM The roles of data collection, evaluation, and research in pediatric palliative care. Support Voice.
2002;811- 13Google Scholar
MA Measuring symptoms of grief and bereavement. Am J Psychiatry.
1982;1391590- 1592PubMedGoogle Scholar
JC A comprehensive assessment of satisfaction with care: preliminary psychometric analysis in an oncology institute in Italy. Ann Oncol.
1999;10839- 846PubMedGoogle ScholarCrossref
et al. A comprehensive assessment of satisfaction with care: preliminary psychometric analysis in French, Polish, Swedish and Italian oncology patients. Patient Ed Counseling.
2001;43243- 252Google ScholarCrossref
et al. Timing of patient satisfaction assessment: effect on questionnaire acceptability, completeness of data, reliability and variability of scores. Patient Ed Counseling.
2002;46131- 136Google ScholarCrossref
JL Statistical Methods for Rates and Proportions. 2nd ed. New York, NY Wiley & Sons1981;14- 17
et al. Preparing for the end of life: preferences of patients, families, physicians, and other care providers. J Pain Symptom Manage.
2001;22727- 737PubMedGoogle ScholarCrossref
JM Continuity of care and trust in one's physician: evidence from primary care in the United States and the United Kingdom. Fam Med.
2001;3322- 27PubMedGoogle Scholar
et al. Symptoms and suffering at the end of life in children with cancer. N Engl J Med.
2000;342326- 333PubMedGoogle ScholarCrossref
JA Families looking back: one year after discussion of withdrawal or withholding life-sustaining support. Crit Care Med.
2001;29197- 201PubMedGoogle ScholarCrossref
Committeeon Bioethics and Committee on Hospital Care, American Academy of Pediatrics, Palliative care for children. Pediatrics.
351- 357PubMedGoogle ScholarCrossref
Not Available, SPSS
[computer program]. Version 10.0.5
Chicago, Ill SPSS Inc1999;Google Scholar
JC Improving the assessment of (in)patients' satisfaction with hospital care. Med Care.
2001;39270- 283PubMedGoogle ScholarCrossref
S Using patient stories to inspire quality improvement within the NHS Modernization Agency collaborative programmes. J Clin Nurs.
2003;12422- 430PubMedGoogle ScholarCrossref
JD Assessment of medical care by elderly people: general satisfaction and physician quality. Health Serv Res.
1998;32741- 758PubMedGoogle Scholar
R Reliability and validity of a new measure of patient satisfaction with out of hours primary medical care in the United Kingdom: development of a patient questionnaire. BMJ.
1997;314193- 198PubMedGoogle ScholarCrossref
P Construction of a scale measuring inpatients' opinion on quality of care. Int J Qual Health Care.
1999;11507- 516PubMedGoogle ScholarCrossref
E A new, brief questionnaire (PEQ) developed in primary health care for measuring patients' experience of interaction, emotion and consultation outcome. Fam Pract.
2001;18410- 418PubMedGoogle ScholarCrossref
EA Assessing the quality of healthcare provided to children. Health Serv Res.
1998;331059- 1090PubMedGoogle Scholar
SF The use of cognitive testing to develop and evaluate CAHPS 1.0 core survey items: Consumer Assessment of Health Plans Study. Med Care.
MS10- MS21PubMedGoogle ScholarCrossref
MB Care of bereaved parents: a study of patient satisfaction. J Reprod Med.
1994;3980- 86PubMedGoogle Scholar