Seminars in Fetal & Neonatal Medicine (2008) 13, 341e356

ava i lab le at www.sc ienced i rec t . com

journa l homepage : www.e lsev ie r . com/ loca te /s iny

Effectiveness of bereavement interventions inneonatal intensive care: A review of the evidence

Sheila Harvey a, Claire Snowdon a,b,*, Diana Elbourne a

a Medical Statistics Unit, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UKb Centre for Family Research, University of Cambridge, Free School Lane, Cambridge CB2 3RF, UK

KEYWORDSNeonatal intensivecare;Bereavement care;Bereaved parents;Grief;Effectiveness

* Corresponding author. Centre for FCambridge, Free School Lane, Cambr1223 334510; fax: þ44 1223 330574.

E-mail address: cms1000@cam.ac.

1744-165X/$ - see front matter ª 200doi:10.1016/j.siny.2008.03.011

Summary The provision of bereavement care is an important part of neonatal intensive care.This systematic review of the effectiveness of interventions to support families and facilitateemotional adjustment following the death of a baby suggests that, while these are largelyappreciated by parents who have participated in research, there has been little rigorous eval-uation of their effectiveness. This review reflects on possible reasons for this; for example:NICU-led bereavement care is changing, the effectiveness of bereavement care is difficultto measure, concepts of effectiveness are not static, and ethical concerns complicate exper-imental research. Bereavement interventions are compassion-led and generally considered tobe beneficial. New research questions and new methodological challenges are discussed withreference to two examples of evolving practice: bereavement photography and the use of rit-ual. Future research using innovative and sensitive RCTs and consensus amongst relevantstakeholders is suggested.ª 2008 Elsevier Ltd. All rights reserved.

Introduction

In many Western societies the death of a child is a rareevent, mostly occurring in hospital, and usually in anintensive care setting.1 When babies die in neonatal in-tensive care units (NICUs), parents experience eventsand an environment that differ in important ways fromthose associated with other types of bereavement, suchas the death of a spouse. Some aspects of death in theNICU are very context-specific, and the associated

amily Research, University ofidge CB2 3RF, UK. Tel.: þ44

uk (C. Snowdon).

8 Elsevier Ltd. All rights reserved

experiences can differ even from other forms of parentalbereavement that occur in relatively similar circum-stances. Unlike parents experiencing earlier perinatal los-ses, those bereaved in a NICU have a period precedingdeath during which they may engage with their child asa born and living individual. Unlike parents of older chil-dren who die in paediatric intensive care, most have noopportunity to experience home life with their child. Ex-periences associated with the NICU setting and the op-portunities and limitations that it generates, shape andcontextualise the bereavement interventions availableto parents. There have been calls for research on the ef-fectiveness of bereavement interventions more generallyin the paediatric field,2e5 but we suggest that NICU-ledbereavement interventions warrant investigation in theirown right.

.

342 S. Harvey et al.

Objectives

The aim of this review is to consider systematically theevidential base for the effectiveness of bereavementinterventions or packages of care for parents around thetime of and following the death of a baby in a NICU. Wereflect on the nature of this evidence, identify gaps in theliterature, and make suggestions for further research.

Methods

Studies evaluating any specific NICU-led intervention orpackage of care designed to improve the level of support orpsychological well-being of bereaved parents and otherfamily members, measuring any outcomes, were consid-ered. A hierarchy of study designs to assess effectivenesswas distinguished, ranging from randomized controlledtrials (RCTs), non-randomized controlled studies, and stud-ies without controls. Qualitative or broadly descriptivestudies were also included for their insights into parentalviews of the effectiveness of bereavement interventions.

Full details of the search strategy are available from theauthors. In brief, appropriate electronic databases weresearched on keywords including bereavement, bereaved,death, mortality, grief, end of life, withdrawal, infant,newborn, child, baby, neonate, neonatal, paediatric, in-tensive care, critical care, parent, sibling and family.Although studies involving only miscarriage, terminationof pregnancy, or stillbirths were excluded, studies whichconsidered the broad range of perinatal bereavementincluding death within the first month of life were includedin the review. Reference lists of studies, reviews andguidelines for professionals were checked for additionalstudies. Given that there were very few quantitativestudies, no statistical analysis was conducted.

Results

The searches yielded 1588 citations. Although few concen-tred exclusively on NICU parents, 109 full papers wereevaluated in depth, including six previous reviews2e7; 21studies1,8e27 met the inclusion criteria and are briefly sum-marized in Tables 1 and 2. Of the 21 studies, 14 were con-ducted in the US8e11,13,14,17e19,21,24e27 and seven inEurope.1,12,15,16,20,22,23 One was an RCT,12 four were cohortstudies with non-randomized controls,10,14,16,21 one withhistorical controls,14 one was a longitudinal case series,19

and 15 were qualitative or broadly descriptive stud-ies.8,9,11,13,15,17,18,20,22e27 The 21 studies focused either onspecific interventions or on broad packages of care.

Evaluation of specific interventions (n Z 7 studies)

Bereavement support groups or counsellingOf the four studies that investigated the impact of supportgroups or counselling, one was an RCT12 in which motherswere randomly allocated to receive planned support andcounselling, or to receive routine hospital care immediatelyfollowing stillbirth or early neonatal death. At 6 monthsthere was a lower rate of psychiatric disorder (measuredby the General Health Questionnaire) and less anxiety and

depression (Leeds Scale) in the support and counsellinggroup, but by 14 months these differences were no longerstatistically significant. Similarly, a cohort study with non-randomized controls16 found that parents whose babydied in an NICU or after sudden infant death syndrome(SIDS) in the first year of life who received counsellingexperienced significantly less anxiety at 1 and 13 monthspost-bereavement (State-Trait Anxiety Inventory) thanthose who had not received counselling. They did howeverexperience significantly more intrusive thoughts at 1 month(Impact of Event Scale) and more bodily symptoms at 13months (Bodily Symptom Scale) than the non-counsellinggroup. These findings contrast with those of another non-randomized study which found no difference in grief reac-tions (Hogan Grief Reaction Checklist) between parentswho had and had not attended a support group.21

Reilly-Smorawski and colleagues asked parents directlyfor their views of a support group programme, attended byapproximately half of the bereaved parents in theircentre.24 Questionnaires completed at the end of each12-week period of support consistently suggested that par-ents viewed the support groups as very helpful, particularlyas an aid to learning to tolerate their grief. No comparisonswere made between parents who did and did not attend thesupport groups.

Giving information on parental griefTwo studies described and assessed methods of informingnewly bereaved parents about parental grief and the copingstrategies that they might employ.

A small cohort study using non-randomized controlsinvestigated the impact of telephone follow-up by a physi-cian.10 Around the time of death a physician spoke to par-ents in person or by telephone. The physician explainedsome of the problems commonly experienced by bereavedparents and gave advice on topics such as sources ofstrength and support, talking to siblings, and organizing fu-nerals. A follow-up telephone call approximately 1 weeklater repeated this information, reviewed post-mortemfindings, and offered the opportunity to discuss the causeof death. Parents were encouraged to contact the physicianwith any further questions. At a later interview (9e27weeks post-bereavement), 18 parents who had receiveda follow-up telephone call reported significantly fewergrief-related problems, particularly loneliness and depres-sion, guilt feelings, questions about heredity, and concernsabout the cause of death, compared to the 11 parents whowere not called.

A survey investigated the impact of sending a book toparents a few days to several weeks post-bereavement.13

The book covered topics such as funeral arrangements,marital strain, reactions of others, effects on the couple’sphysical and emotional relationships, siblings, religious is-sues, guilt, learning to have fun again, and getting onwith life. It was accompanied by a letter extending theNICU staff’s sympathies and encouraging parents to callwith any problems or if they wished to review the baby’s ill-ness and death. The letter also reviewed the nature of griefand the problems some families might experience. Ina questionnaire, most respondents indicated that thebook was helpful and suggested continuation of theproject.

Table 1 Description of the studies

Leadauthor

Year Setting Aim Outcomes Types ofparticipants

Intervention Design/methods

Rowe 1978 US e single

universityhospital

To examine how

effectively informationabout

the baby’s deathwas transmitted

to parents by thephysicians at a

perinatal referralcentre

Morbid grief reactions, level

of understanding, degree ofsatisfaction, sources of

support and information

Mothers

(perinataldeath)

Not a specific

intervention

Qualitative study e interviews

conducted 10e22 months after thedeath

Cohen 1978 US e singleuniversity

hospital

Aim not stated Parents’ views on viewingthe body of their dead baby

Mothers(perinatal

death)

Not a specificintervention

Descriptive study e interviews at aperinatal bereavement clinic around

1 month after the deathSchreiner 1979 US e single

tertiaryhospital

To examine the

outcome of a physiciantelephone call to

parents approximately1 week after the death

of their baby

Persistence and severity of

nine different grief-relatedproblems

Familiesa (death

in a NICU)

Telephone call from

a neonatologist 3e

19 days after death

Non-randomized study comparing

parents who received a telephonecall with those who did not.

Semi-structured interviews con-ducted 2e3 months after the death

to assess outcomeMahan 1981 US e single

university

hospital

To evaluate methods of

helping families

through the birth,illness and death of

their baby and to see ifthe staff’s ideas of

parental supportsystems concur with

the feelings of thoseaffected by care

practices

Parents’ views about the

care they received

Sets of parents

(death in a

NICU)

Not a specific

intervention

Qualitative study e postal question-

naires sent to mothers and fathers

(time since the death not reported).Effect on response of socioeconomic

status, whether mother or fatheranswered, and parental age

examined

Forrest 1982 UK e single

universityhospital

To test the hypothesis

that psychologicalrecovery from stillbirth

or early neonatal death(death of the baby

within the first 7 daysafter birth) is enhanced

by a plannedprogramme of support

and counselling

Psychological recovery at 6

and 14 months

Mothers

(perinataldeath)

Planned support

and counselling

RCT e mothers randomly allocated

immediately after the death/still-birth to either planned support and

counselling or to routine hospitalcare.

Interviews conducted at 6 and 14months to assess the outcome using

the General Health Questionnaireand Leeds scales

Mahan 1983 US e single

universityhospital

Not specifically stated Parents’ views on the value

of being sent a book, The

Bereaved Parent

Parents (death

in a NICU)

Bibliotherapy e

book sent a few daysto several weeks

after the death

Descriptive study e postal

questionnaire sent to mothers andfathers approximately 5e6 months

after the death

(continued on next page)

Bere

ave

ment

inte

rventio

ns

343

Table 1 (continued)

Leadauthor

Year Setting Aim Outcomes Types ofparticipants

Intervention Design/methods

Harmon 1984 US e single

regionaltransport NICU

To describe the

reactions of motherswhose baby had died in

a regional transportNICU

Maternal grieving and family

functioning; mothers’impressions of interventions

and staff involvement dur-ing their baby’s hospitaliza-

tion and following death

Married mothers

(death in aNICU)

A neonatal hospice

programme

Non-randomized comparison (using

historical controls) comparing theimpact of a neonatal hospice

programme on mothers before andafter its implementation e

telephone interviews conducted 3e9

months after the deathWhite 1984 UK e single

hospital

To examine whether

various objectives toestablish normal

grieving were beingachieved and if they

accorded with parentswishes

Parents’ views about their

care; grief response

Familiesa

(perinataldeath)

Not a specific

intervention

Descriptive study e interviews

conducted 2e13 months after thedeath. Grief response assessed using

the Leeds scale

Dyregrov 1991 Norway e singlehospital

To examine the griefreactions of mothers

and fathers during thefirst year of

bereavement

Grief response Familiesa (deathin a NICU or

following SIDS)

Grief counselling(although this was

not the main aim ofstudy)

Non-randomized comparison of fam-ilies who did and did not receive

grief counselling.Postal questionnaires sent to parents

at 1, 6 and 13 months after the deathto assess outcome using: Impact of

Event Scale; Goldberg GeneralHealth Questionnaire; State-Trait

Anxiety Inventory; Bodily SymptomScale; Beck Depression Inventory

Calhoun 1994 US e

bereavement

support group

To evaluate parents’perceptions of the

nursing interventions inthe initial stages of

grief from neonatal loss

Perceived helpfulness ofinterventions

Parents(perinatal

death)

Not a specificintervention

Descriptive study e questionnairesdistributed to members of a support

group during group meetings (timesince the death not reported)

Harper 1994 US e

bereavementsupport group

To investigate the

relationship betweenrecommended

physician treatmentsand parent satisfaction

following the death ofa baby and to

determine whichphysician behaviours

are perceived by

parents as most helpful

Relationships between

satisfaction with care,‘degree of occurrence

continuum’ and ‘perceivedhelpfulness scale’

Parents

(perinataldeath)

Seven categories of

physician actions

Case series study e parents referred

to SHARE (a bereaved parents’support group) completed

questionnaires 2e60 months afterthe death: Ware’s Short Form

Patient Satisfaction Questionnairemeasured satisfaction; 34-item

questionnaire developed byinvestigators to assess degree to

which physician actions occurred

(‘degree of occurrence continuum’)and perceived helpfulness

344S.

Harve

yet

al.

Lasker 1994 US e fivehospitals

To evaluate parents’responses to specific

interventions and toassess parents’ overall

satisfaction with thecare they received at

the time of their loss

Whether intervention wasexperienced; level of

satisfaction; value ofinterventions; grief

response

Parents(perinatal

death)

25 recommendedinterventions

Longitudinal case series e interviewsconducted with parents

approximately 2 months and 1 and 2years after the death. Perinatal

Grief Scale used to measure griefresponse. Relationships between

experience of a given interventionand parents’ satisfaction with their

general care, satisfaction withspecific intervention, and grief

outcomes examinedDent 1996 UK e

11 healthdistricts

To establish what

parents thought of thecare they had received

after their child haddied suddenly and

unexpectedly fromaccident or illness

Parents’ views and

satisfaction with the carethey had received

Familiesb

(sudden,unexpected

death of childaged between 1

week and 12years)

Not a specific

intervention

Descriptive study e postal

questionnaire sent to families (timesince the death not reported)

McHaffie 2000 UK e threeregional referral

centres

To provide insights intothe issues which might

influence a chaplain’srole in supporting

parents whentreatment is withdrawn

Parents’ views about therole of the chaplain

Familiesa (deathin a NICU)

Not a specificintervention

Qualitative study e semi-structuredinterviews conducted 3 and 13

months after the death

McHaffie 2001 UK e threeregional referral

centres

To explore parents’experiences of

bereavement careafter withdrawal of

intensive care

Parents’ perceptions andexperiences of their care

Familiesa (deathin a NICU)

Not a specificintervention

Qualitative study e see above(McHaffie 2000)

DiMarco 2001 US e perinatalloss support

newslettermailing list

To determine if asupport group

intervention makes adifference in grief

reactions of parentswho have experienced

a prenatal loss

Grief reaction Parents(perinatal loss)

Support groupparticipation

Non-randomized comparison of par-ents who did and did not participate

in a support group.Postal questionnaire sent to a conve-

nience sample of families who re-ceived a perinatal loss support

newsletter. Time lapse since deathwas 1 month to 13 years. Grief re-

action assessed using the HoganGrief Reaction Checklist

Lundqvist 2002 Sweden e threehospitals

To focus further on andilluminate mothers’

lived experiences ofthe professional care

they received whilefacing the threat and

reality of losing theirbaby

Mothers’ experiences of thecare they received

Mothers (deathin a NICU)

Not a specificintervention

Qualitative study e interviewsconducted 14e32 months after the

death

(continued on next page)

Bere

ave

ment

inte

rventio

ns

345

Table 1 (continued)

Leadauthor

Year Setting Aim Outcomes Types ofparticipants

In on Design/methods

Reilly-

Smorawski

2002 US e single

hospital

To gain feedback from

parents who hadattended a 12-week

support groupprogramme

Parents’ views Parents (death

in a NICU)

1 pport

g ion

Descriptive study e brief informal

questionnaire given to parents (timesince the death not reported) at the

end of the 12-week programme

Pector 2004 North America e

internet

networks

To assess theexperiences of

bereaved parents ofmultiples with

resuscitation and lifesupport decisions

Views and experiences ofparents

Parents(perinatal loss e

multiples)

N ificin n

Qualitative study e internet surveyof parents who were members of

internet support groups andorganizations for multiple birth loss

(time since the deaths not reported)

Kavanaugh 2005 US e threehospitals and

throughnewspaper

advertisement

To examine theexperience of low-

income, AfricanAmerican parents

surrounding perinatalloss and to describe

how other life stressorsinfluenced the parents’

responses and caring

needs

Views and experiences ofparents

Parents(perinatal loss)

N ificin n

Qualitative study e parentsinterviewed within 4 months of the

death

Brosig 2007 US e single

hospital

To identify factors

important to parents intheir baby’s end-of-life

care

Views and experiences of

parents; perceptions of grief

Familiesb (death

in a NICU)

N ific

in n

Qualitative study e semi-structured

interviews with parents, mean of1.91 years after the death.

Parents completed the Revised GriefExperience Inventory

a Investigators did not report numbers of mothers/fathers, therefore it is not clear whether ‘family’ referred to one or bot responding jointly.b Family e either mother or father or both responded. Where both responded this was treated as a joint response.

346S.

Harve

yet

al.

terventi

2-week su

roup sess

ot a specterventio

ot a specterventio

ot a spec

terventio

h parents

Table 2 Summary of study results

Lead author Year Number ofparticipants

Response/participation rate

Summary of results Comments

Rowe 1978 26 mothers 89.7% 6/26 mothers had a prolonged grief reaction (12e20 months);

13/26 obtained information during hospitalization only;13/26 received additional information weeks or months later;

60% mothers with adequate understanding and no prolongedgrief reaction felt totally dissatisfied or only partially satisfied

with the information they received;follow-up contact (phone or in-person) increased understanding

significantly;mothers who had in-person follow-up were more likely to be

satisfied with the information they received

Additional 30 families had moved andwere not invited to participate

Cohen 1978 80 mothers Not reported Mothers were generally in favour of being given the option toview the body of their dead baby;

18/80 refused or were not given the option

Study poorly described.

Schreiner 1979 29 familiesa

(number ofmothers/fathersnot reported)

Not reported Follow-up telephone call (n Z 18) e 3/18 had ten moderate/

major grief-related problems;No follow-up telephone call (n Z 11) e 11/11 had 26 moderate/

major grief-related problems

Same neonatologist recruited families,did follow-up call and assessed outcomes

Mahan 1981 23 sets of parents 25% Responses were not affected by socioeconomic status or pa-

rental age but were significantly affected by whether themother or father answered the question;

Time in NICU e topics covered were seeing/touching/holdingthe baby; perception of baby’s chances; maternal desire for

company; level of information (incl. at time of death); help-fulness of staff; source of help at time of death; receipt of

a picture of the babyPost-death e topics covered were seeing/touching/holding the

baby; explaining death to siblings; funeral; contact with physi-cian; relationship with spouse; desire for further children

Questionnaire responses of mothers andfathers presented with a narrativedescription of the results

Forrest 1982 50 mothers Not reported Planned support and counselling (n Z 25) e 16/25 completed

6-month follow-up; 2/16, General Health Questionnaire score12þ indicating psychiatric disorder and 5/16, Leeds scale score

7þ indicating pronounced symptoms of depression and anxietyControl e 19/25 completed 6-month follow-up; 10/19, General

Health Questionnaire score 12þ and 12/19, Leeds scale 7þP < 0.01, Fisher’s exact test.

At 14-month follow-up differences not statistically significant

Recruitment of mothers, andrandomization and allocation procedurenot described. High rate of loss tofollow-up

(continued on next page)

Bere

ave

ment

inte

rventio

ns

347

Table 2 (continued)

Lead author Year Number ofparticipants

Response/participation rate

Summary of results Comments

Mahan 1983 40 parents (39mothers, onefather)

33% Two mothers who responded did not read the book, and onemother read parts. In all, 36 mothers, four fathers andseven significant others read the book; 37/40 felt the bookwas helpful, and all but one suggested continuation of theproject; 16/40 felt book was less helpful as it was about anolder child; 21/40 felt that this made no difference

Harmon 1984 38 mothers Not reported Maternal grieving e no significant differences between the

groups;Before hospice programme e 33% contacted by a member of

staff to offer support; 60% contacted about post-mortem resultsAfter hospice programme e 75% contacted by a member of staff

to offer support; 92% contacted about post-mortem results.Indication of increased husband involvement following hospice

programme

Additional descriptive findings alsoreported, such as pictures of the baby

White 1984 12 familiesa

(number ofmothers/fathersnot reported)

70.6% 6/12 families scored 8þ on the Leeds scale, of whom five alsohad high grief scores;

no difference between maternal and paternal grief scores;grief scores did not correlate with length of mourning period;

topics covered e handling of the death; mementos; funeral;communication; recurrence risk; support counselling;

GPs

Methods of study poorly described

Dyregrov 1991 37 familiesa

(number ofmothers/fathersnot reported)

62.7% n Z 29 received grief counselling e experienced lessanxiety at 1 and 13 months, more intrusive thoughtsat 1 month, more bodily symptoms at 13 monthscompared to those who received no grief counselling(n Z 8)

Examination of the impact of griefcounselling not one of the original aimsof the study

Calhoun 1994 23 (number ofmothers/fathersnot reported)

Not reported Results of descriptive analyses presented covering thefollowing topics: photographs and other mementoes;holding the baby; general emotional support; informationabout support groups; funerals, expected response of familyand friends

Harper 1994 37 parents (23mothers, 14fathers)

Not reported The seven categories of recommended physician actionswere: method of informing parents; provide medical

information; demonstrate compassion; provide support; beavailable; efforts to overcome denial; grief

counselling.Significant correlations were found between satisfaction scores

and the use of most of the recommended actions (5/7 cate-gories). Being available, providing medical information and grief

counselling were the categories with the greatest correlation

348S.

Harve

yet

al.

Lasker 1994 194 parents (138mothers, 56fathers)

84.6% Parents grouped by whether they experienced: spontaneousabortion or ectopic pregnancy; early fetal death; late fetal

death; or neonatal death. 25 recommended interventions pre-sented giving proportions of parents in each group who experi-

enced the intervention, felt the intervention to be essential andsatisfaction; in all groups for more than 70% of interventions,

those who experienced the intervention were significantly moresatisfied (P< 0.05);

neonatal death parents e no association between more inter-ventions and general satisfaction; no association between sat-

isfaction and intensity of grief;all parents e no relationship between number of interventions

and the intensity of grief; satisfaction with care not associated

with demographic characteristics

Of the 194 parents, 27 had experiencedneonatal death. Results are presented bytype of loss

Dent 1996 42 familiesb (40mothers, 27fathers)

58% Descriptive results presented on the following topics:emergency services; hospital service (including care byhospital staff, communication, mementoes etc); coroner’soffice; media contact; dealing with governmentdepartments; post-mortem examinations; funeral; support(e.g. follow-up care); other children in the family

McHaffie 2000 59 familiesa

(number ofmothers/fathersnot clear)

72.8% Investigators reported families’ views on and experiences ofthe chaplain’s role. Five factors were identified:reassurance; comfort; sense of control; officiating atceremonial functions; ongoing support and concern inbereavement

The report forms part of the studybelow. Insights reported mainly from asubset of six families

McHaffie 2001 59 familiesa

(number ofmothers/fathersnot clear)

72.8% Parents highlighted a number of specific needs. Follow-upappointments should be scheduled soon after the death(within 2 months), irrespective of autopsy results beingavailable, with the named neonatologist, away from thehospital if possible. Parents indicated that they valuedefforts to find out how they are coping, full and frankinformation given sensitively and reassurances wherepossible

DiMarco 2001 121 parents (88mothers, 33fathers)

30.3% No statistically significant difference in the Hogan GriefReaction Checklist scores between parents who did and didnot attend a support group

Lundqvist 2002 16 mothers 76.2% The primary themes identified were feeling empowered andpowerless. Related issues explored included: attitude andactions of health-care staff; information andcommunication; touching/holding the baby; photographs ofthe baby; follow-up care

(continued on next page)

Bere

ave

ment

inte

rventio

ns

349

ble 2 (continued)

ad author Year Number ofparticipants

Response/participation rate

Summary of results Comments

illy-Smorawski 2002 Not reported Not reported 54% of parents attended the 12-week support groups. Parents

who did not attend were usually single mothers.Couples reported that the support groups were helpful overall

and specifically aided learning to tolerate the grief and painof having lost a baby. Parents also reported that they were

able to get through particularly difficult moments by tellingthemselves to ‘save it for the group’. Some parents reported

no longer feeling afraid to talk to each other aboutthe baby

Methods of evaluation poorly reported.No comparison between parents who didand did not attend the support groupsessions.

ctor 2004 71 parents (67mothers, fourfathers)

Not reported Topics covered were: resuscitation decisions; influenceof multiple-birth status on decision-making, style ofdecision-making (e.g. collaborative, parent-initiated;support and criticisms of parental choices); the deathprocess (e.g. importance and meaning of time withdeceased baby, desired experiences, multiple memories,such as photographs); parent preferences for beinginformed about the death; and discussing death (e.g.follow-up)

vanaugh 2005 23 parents (17mothers, sixfathers)

Not reported Main themes which emerged were: recognizing problemsand responding to the loss (e.g. misreading symptoms ofpregnancy complications, maintaining hope during medicaltreatment, feeling intense emotions after the death);dealing with stressful life events (e.g. feeling abandoned orunsupported, suffering economic hardship, experiencingother’s serious illness or untimely deaths, receiving unfairtreatment); creating and cherishing memories of their baby(e.g. being with their baby, noticing family resemblances,gathering keepsakes of and remembering their baby,deciding on burial); living with the loss (e.g. encounteringdifficult situations, relying on their spirituality, seekingdiversions and support, making sense of the loss,contemplating future pregnancies)

350S.

Harve

yet

al.

Ta

Le

Re

Pe

Ka

Bro

sig

2007

19fa

milie

sb(1

8m

oth

ers

,11

fath

ers

,one

aunt)

37.3

%There

was

no

sign

ifica

nt

dif

fere

nce

insc

ore

sbetw

een

pare

nts

who

had

hosp

ice/p

allia

tive

care

invo

lvem

ent

wit

hth

ose

who

had

not.

Mean

Post

-Death

Adap

tati

on

Score

sin

dic

ate

dth

at

the

inve

sti-

gato

rsvi

ew

ed

pare

nts

as

adapti

ng

posi

tive

ly.

Topic

sco

vere

dw

ere

:honest

y;em

pow

ere

ddeci

sion-m

aki

ng;

pare

nta

lca

re;

envi

ronm

ent;

fait

h/t

rust

innurs

ing

care

;phys

i-

cians

beari

ng

wit

ness

;su

pport

from

oth

er

hosp

ital

care

pro

-vi

ders

;th

ings

that

help

ed

copin

g,fa

mily

support

,ke

epin

gth

e

mem

ory

alive

;sp

irit

uality

/fait

h;

alt

ruis

m;

refo

cusi

ng

on

life

;va

lidati

on

of

deci

sion;

bere

avem

ent

support

groups;

pare

nta

l

reco

mm

endati

ons

aIn

vest

igato

rsdid

not

report

num

bers

of

moth

ers

/fath

ers

,th

ere

fore

itis

not

clear

wheth

er

‘fam

ily’

refe

rred

toone

or

both

pare

nts

resp

ondin

gjo

intl

y.b

Fam

ily

eeit

her

moth

er

or

fath

er,

or

both

resp

onded.

Where

both

resp

onded

this

was

treat

ed

as

ajo

int

resp

onse

.

Bereavement interventions 351

Neonatal hospice programmeA cohort study using historical controls investigated theimpact of a neonatal hospice programme by comparing twogroups of parents before and after its implementation.14

Staff were given training to implement the programmewhich included the use of a ‘family room’, frequent familyconferences, and the provision of ongoing support. Staffwere encouraged to take pictures to offer to the family,discuss the possibility of a post-mortem and subsequentlyexplain the results and possible implications, provide infor-mation about funeral arrangements, and systematically fol-low up bereaved families, including assessment of currentfamily functioning. Telephone interviews conducted at 3and 6 months post-bereavement revealed that after theprogramme there was a substantial increase in the propor-tion of bereaved parents (from 33% to 75%) who were con-tacted by a member of the hospital staff, as well as anincrease in the numbers contacted about post-mortem re-sults. Most mothers reported that post-mortem informationhelped them understand why their baby had died andexpressed gratitude that someone cared enough to talk tothem directly about their experiences. Increased involve-ment from their partners appeared to be another benefitof the programme.

Evaluation of broader aspects of bereavementcare (n Z 14 studies)

Fourteen studies examined the impact of bereavement caremore broadly by seeking the views of parents about thecare they had received. Data were collected either throughface-to-face interviews,8,9,15,17,19,20,22,23,26,27 postal ques-tionnaires,1,11,18 or an internet survey.25 Several themesemerged from these studies.

Communication strategiesPoor communication was frequently mentioned by parents,particularly around the time of and following thedeath.11,15 Parents were said to want frank, honest infor-mation about their child’s condition even if this meant re-ceiving bad news.27 An appointment to discuss the deathwith a neonatologist or paediatrician was judged to bevery important.1,27 The provision of medical informationwas found to greatly influence parental satisfaction.18,19

Information which answers parents’ questions, clarifiesexactly what happened, and provides reassurances abouttheir baby’s care was reported as being important in help-ing parents to cope with their grief.22,23,27 Explanations, in-cluding written information, about the grieving process andhow families might be affected were also considered veryhelpful by parents17,18 and were associated with greaterlevels of satisfaction.19 Mothers were more likely to feeldissatisfied and have less understanding of the cause ofdeath if there was no follow-up contact with a physician.9

Touching and holding the babyParents seem to have mixed views about touching orholding their baby. Many were positive about viewing8 orholding their baby,17 particularly while still alive11,19; thiswas viewed as essential by almost all parents in one study.19

There seemed to be some reluctance to have physicalcontact with their baby during the dying process or after

352 S. Harvey et al.

death.11,15,23 White and colleagues, for example, reportedthat seven of the ten parents invited to touch or hold theirlive baby did so, whereas after death only four out of the 10parents did so.15 For some, being repeatedly urged to touchtheir baby created unpleasant feelings.23 In the most re-cent of the studies, however, it was reported that ‘mothersreadily embraced their infants’ and all but one of the 17study mothers held their baby at the time of death; formost, this was the only opportunity they had to hold theirbaby.26

MementoesIt was important to parents to receive mementoes such asphotographs, handprints, footprints, casts, or clothes thatthe baby had worn in hospital.27 Investigators report that, ingeneral, the response to such gestures was positive,17e19

and that parents who were not offered or who declined me-mentoes often regretted this subsequently.1,15,23 Parentswanted photographs to be taken before their baby’s deathwithout any equipment, particularly on the face.26 Parentsof multiples felt that it was important to have an image oftheir babies together, even if it was taken shortly after onebaby had died.25

Parental careParents need staff to take care of them, as well as theirbaby, both physically and emotionally.17,19,23,27 Parents re-ported finding it very helpful when nurses made themselvesavailable to listen, allowing them to express their feelings.They also appreciated nurses talking openly and honestlywith them about their loss.17 Compassion from staff was im-portant,17,18 and parents described feeling upset when theyfelt staff members were insensitive to their feelings.19,27

Non-clinical hospital staff, particularly chaplains, werefound to be very important sources of parental support.20,27

Methodological issues in the identified studies

Different methodological approaches were adopted to in-vestigate the effectiveness of bereavement interventions inthe 21 studies. Only one was an RCT,12 and four studiesincluded a non-randomized control group.10,14,16,21 Evenin the RCT there were threats to internal validity, includingthe possibility of contamination in the control group, anda differential and low rate of follow-up of patients (36%of mothers in the intervention group and 24% in the controlgroup were lost to follow-up).

In terms of external validity, most of the studies involvedrelatively small samples of parents, and where responserates to surveys or invitations to participate were reported,these were often quite low. It is not clear, therefore, howrepresentative these study participants are of all parentsbereaved in the NICU. Important perspectives are beingmissed if the parents who decline research participation arethose who are more distressed, or who dislike or benefitleast from bereavement interventions. Some studies fo-cused solely on mothers,8,9,12,14,23 but even in studies whichdid invite both parents, the proportion of fathers who tookpart compared with mothers was generally much lower.

It is well recognized that the grieving process does notfollow a series of sequential phases. The lack of an obviousoptimal time for assessment is reflected in the wide

variation across and within studies in the time betweenbereavement and assessment. In one study the time sincebereavement ranged from 1 month to 13 years.21

There is perhaps also a general lack of consensus as towhat might be appropriate outcomes to measure followingbereavement intervention(s). For example, several inves-tigators measured grief responses at various time pointsusing the range of available tools.12,15,16,21,27 Other studiesexamined level of satisfaction, level of understanding, theimpact on relationships with partners and other familymembers, and parents’ views and experiences of theircare generally and regarding specific interventions.

Discussion

Most NICUs carry out bereavement interventions in variousforms. If we judge the available evidence for theireffectiveness by the number of RCTs, then this reviewsuggests that the empirical basis of current practice islimited. This does not sit well in an age of evidence-basedmedicine, particularly in such a research-active specialty asneonatology. The difficulties which act as impediments toresearch in this area are likely to be complex andmultifaceted.

Factors which may impede evaluation ofNICU-led bereavement interventions

Research may be limited by a number of factors, such as:

� NICU-led bereavement care is changing;� effectiveness of bereavement interventions is difficult

to measure, and concepts of effectiveness are notstatic;� effectiveness is often assumed;� ethical concerns complicate experimental research.

NICU-led bereavement care is changingResearch in this area is undoubtedly complicated, espe-cially as the field itself is changing in a variety of ways, withmodifications to existing approaches and the introductionof novel interventions. These changes may be subject toa high degree of local variability, making research difficultto conduct and the results difficult to apply. This raises newresearch questions and new methodological challenges, ashighlighted by two examples of evolving practice: develop-ments in the use of bereavement photography, and thewider introduction of ritual into bereavement care.

Bereavement photographyCommentators11,23,25,28,29 suggest that parents value be-reavement photographs, but with the caveat that forsome they have been ‘too clinical’.30 Images of babies onventilators can be difficult for parents,26 and many onlyhave instantly developed photographs which can be starkand fade with time.25 Digital cameras are, however, in-creasingly available, and it has been suggested that nursesshould develop skills in bereavement photography.25,29

This aspect of bereavement care is changing. Some NICUsnow utilize the specialist skills of medical photographers.

Bereavement interventions 353

Professionals can produce long-lasting sensitive images, andsome can offer creative services, manipulating details orcombining images of siblings who were not photographedtogether. Websites offer the services of independent pho-tographers, some for free (www.news.therecord.com/Life/article/279525; www.toddhochberg.com/see.cfm; www.nowilaymedowntosleep.org/pageDisplay.php?pageZ3).Some produce DVDs of images and music for parents to sharewith friends and relatives and to use on websites andat funerals (http://bereavementservices.org/newsletters/summer2007.pdf). When death is imminent, photographersmay take pictures with parents before their baby dies. Thesecan be deliberately posed to leave parents with positiveimages, or they can be a broad record of events arounddeath, including images of parents in their grief.

Whilst studies suggest support for the use of bereave-ment photographs,11,23,25 they largely report on issues ofsatisfaction in small research populations. It is importantthat the short- and long-term effects of such a potentiallypowerful tool should be properly assessed to providea sound basis for NICU policies on a practice which is push-ing new boundaries in care. Experimental research maywell be difficult, but not beyond the realms of possibility.A cluster trial involving randomization at the NICU levelcould be coupled with a qualitative study. A mixed-methodsapproach such as this would provide data on effectivenessas well as preferences, and could be used to explore widerpractical and ethical questions, such as who should take thephotographs, how, when and where? When should they begiven to parents: at an early stage, on leaving the hospital,or at a follow-up visit? Who should have responsibility forthis: photographers, nurses, bereavement counsellors orconsultants? Style and content of photographs can differ,and reactions to these should be explored. It would beappropriate to consider not only whether parents want tohave these images, but also whether partners, siblings orwider family members need and respond to them in thesame way.

The introduction of ritual into NICU bereavement careThe use of ritual around death in the NICU is not new. Manyparents have participated in religious rites if death is likely,and most NICUs have tiny robes for such ceremonies. SomeNICUs are giving greater prominence to rituals as anelement of bereavement care. It has been suggested thatbathing or holding a baby can become an important ritual,with the introduction of music or scents to trigger latersensory associations.31 Some care-givers will help parentsto create unique rituals with specific significance withintheir particular family context.31 Ceremonial removalfrom a ventilator, or formally carrying a baby to handover to a funeral director are novel approaches whichsome parents might appreciate.31 In some centres parentsmay be offered the use of a hospital chapel for a memorialservice and the body of their baby may be prepared forviewing at that time.32 A variety of cultural and religiousrituals exist around death which can be accommodated inNICUs. Non-bereaved Muslim women who gave their viewsof neonatal end-of-life care suggested that parents shouldappoint individuals from their community to help formallyprepare their baby’s body for burial. This task is gender-specific: women should prepare the body of a girl and

men should prepare the body of a boy.33 Care-givers oftenattend the funerals of babies who die in NICUs, so extend-ing the parameters of care into non-hospital family-basedrituals. There is support and enthusiasm for ritualistic ele-ments of care in the literature, but detailed research onparental and staff experiences and studies assessing theireffectiveness are not yet available.

Effectiveness of bereavement interventions is difficult tomeasure, and concepts of effectiveness are not staticAssessing and interpreting the effectiveness of NICU-ledbereavement interventions is particularly difficult, as pa-rental grief is such a complex and highly variable phenom-enon. Researchers must ensure that their psychometrictools are valid,34 and can that they can distinguish betweengrief and other effects on well-being such as depression35

and the familial disturbances that can follow the death ofa child.36 Bereavement exerts a profound effect on bothparents, but men and women can respond in very different,if not ‘incongruent’ ways.37 The complex, ‘multifaceted’36

dynamic of grief within couples is inevitably part of thephenomenon under assessment in experiments measuringeffectiveness. Assessments of effectiveness must alsotake into account the likelihood of parents being exposedto a range of potentially confounding interventions suchas family support, support groups, memorial ceremonies,internet communities38 and help-lines.

There are also theoretical difficulties, hinging uponwhich particular effects on grief are seen as desirable andappropriate, and which methods are considered most likelyto achieve those effects. All research in this area has a basisin culturally and temporally specific models of grief,grieving and bereavement care. Davies suggests that thereare fundamental differences between traditional and newmodels, with older models premised on the importance ofpeople ‘letting go of their emotional relationships withthose who have died’, and new models rooted in theimportance of ‘holding on’.39 The ethos of neonatal carehas also moved over time from ‘protective to supportive’.32

The widely endorsed practice of offering extended contactbetween parents and dying or deceased babies, and collec-tion of mementoes, reflect contemporary models of grief intheir affirmation of the reality of a child.

It is very tempting to view shifting attitudes as linear,representing progress and increasingly sophisticated un-derstanding of grief, but how we now judge effectivenessand by what standards will change again, and not neces-sarily in ways that we might expect. One study, forinstance, has suggested the possibility of negative longer-term psychological sequelae for some women and theirsubsequent babies after seeing and holding their stillbornbaby,40 and this has shaped both practice (http://www.nice.org.uk/nicemedia/pdf/CG45fullguideline.pdf) and thedirection of subsequent research.41

Effectiveness is often assumedChanges to the culture of specialties such as neonatalintensive care bring changes in practice which can becomedeep-seated and highly valued. Many centres have pub-lished details of their philosophy of care, describing a rangeof interventions from support groups to smaller acts ofcommemoration.24,42,43 One centre gives parents a memory

354 S. Harvey et al.

box to store mementoes; another gives two inscribedceramic hearts, one for burial or cremation with a child,the other for parents to keep.

Bereavement care is clearly driven by compassion. It istherefore not surprising that practices such as listening toparents, avoiding minimizing their loss, and attendingfunerals, can seem obviously and intuitively beneficial.They reflect the type of non-professional support thatdecent and caring individuals might offer. Other practices,such as facilitating time with a baby after death, initiatingrituals or photographs, or providing information on sourcesof support, arise from professional roles and are also widelyendorsed. Professional responsibilities and gestures ofhumanity and empathy combine with other structuredaspects of care such as follow-up visits and the availabilityof bereavement counsellors to become formalized bereave-ment programmes with mission statements and protocols.The use of experimental research to assess interventionswithin such complex situations can be challenging, both forresearchers and for those whose practice or caring envi-ronment comes under scrutiny. Certainly the failure ofpotentially informative trials has been directly attributedto professional discomfort with research and convictionsabout best practice.44,45 Cautionary tales warn against as-sumptions of benefits for interventions which later proveto be ineffective or harmful.46 It may be that the chancesof this occurring seems unlikely for compassion-led non-pharmacological interventions, but they are interventionsnonetheless and so require rigorous evaluation.

Ethical concerns complicate experimental researchResearch involving bereaved parents is difficult and re-quires careful ethical consideration.47 It is perhaps signifi-cant that much of the research considered here isqualitative or broadly descriptive rather than directly inter-vening in care. Qualitative research is particularly suited tosensitive situations, emphasizing the value and variety ofindividual experiences and perspectives, and can providean excellent springboard for experimental work.48 The in-sights gained have not, however, fuelled further evalua-tion; only five of the reviewed studies areexperimental,10,12,14,16,21 they are of variable methodolog-ical quality, and only one was conducted this century.21 TheRCT, so widely used to generate evidence to guide care ineven the most difficult of clinical situations, is a rarity inthis field.

Equipoise is fundamental to the successful conduct ofRCTs but care-givers and parents are likely to havepreferences and beliefs about the value of many bereave-ment interventions. Certain interventions, such as seeingand holding a baby around the time of death, involvedeeply personal experiences; they cannot be undone, and ifthe moment is lost, it cannot be repeated. They mighttherefore be viewed as especially difficult to randomize.Statham49 cites Zeanah’s50 discussion of the likely difficul-ties for trials of such interventions with parents of stillbornbabies: ‘Giving parents the opportunity to see and to touchstillborn infants has become so widely accepted that ran-dom assignment in order to test this theoretically appealingand anecdotally supported hypothesis is probably no longerpossible’. Zeanah’s also argues that a trial allocating par-ents to no contact with a baby would ‘probably be as rare

and problematic now as including them in the process mighthave been 25 years ago’.

Certainly considerations fundamental to the conduct ofall RCTs may raise greater concerns in this particular field.For any RCT which randomizes potentially desirable butunevaluated interventions, informing possible participantsabout the research can be a demanding if not dauntingtask. There might also be concerns over a ‘deprivationeffect’ as observed by Oakley51 amongst women who wereinformed of, but not allocated to, additional social sup-port during high-risk pregnancy. An RCT of a communica-tion strategy and brochure for relatives of dying patientshas been successfully completed in adult intensivecare.52 In the NICU setting, however, the vulnerability ofrecently delivered, recently bereaved parents, coupledwith the compassionate drive of those who care forthem, may render experimental research particularlydifficult.

Future research

There is a need not only to improve understanding of theimpact of practice as it currently stands, but also to guideits evolution before unevaluated changes become firmlyentrenched as new standard practice. Possibly the greatestchallenge will be to find the means to conduct studieswhich are methodologically and ethically sound, andacceptable both to care-givers and to bereaved parents.

A first step towards meeting such a challenge is gainingan understanding of the views of those most closelyinvolved in research in this area. Forte and colleaguessuggest a consensus-building conference involving stake-holders and investigators to define a research agenda, toidentify which outcomes are valued by bereaved individ-uals, and to consider which interventions might achievethose outcomes.6 Such a consultation may also be con-ducted through qualitative research processes. Attitudinalresearch uncovering the barriers and facilitators to good-quality bereavement-related studies in the NICU mightalso be informative. It is important to ensure that thedesign of research assessing bereavement interventions isnot only grounded in existing high-quality theoretical andempirical research, but also reflects stakeholder prioritiesand addresses their methodological and logistical concerns.

Qualitative research might be used to further explorethe perspectives of parents and care-givers with the aim ofimproving the management of bereavement interventions.Qualitative approaches may also be incorporated into thedesign of RCTs.53 Here parental responses to allocatedinterventions might be sensitively explored, taking intoaccount contextual elements of individual experiences.

It is likely that the design of future trials necessarily willbe non-standard and will require some creativity. Patientpreference trials, for instance, might allow parents tofollow a course which they feel will suit them; however,such trials can be difficult to interpret because of unknownand uncontrolled confounders.54 Cluster RCTs would workwell for interventions not in current practice as they wouldallow an intervention to be introduced into some NICUs butnot others, and they have the advantage of reducing con-tamination between experimental and control groups.55

Where there is reluctance to withhold an intervention, an

Bereavement interventions 355

individually or cluster-randomized stepped-wedge trial de-sign might be used.56 This involves the sequential random-ized roll-out of an intervention to participants over timeuntil all participants have received the intervention. Thisdesign, like other non-standard designs, is a compromise;contamination may still be an issue, analysis is complex,and it is likely take longer than a conventional design.56

Such methodological compromises may, however, be themeans by which the evidence base is improved and NICU-led bereavement interventions are given a firmer empiricalfooting than they currently have.

Conclusions

A range of bereavement interventions has been adopted aspart of neonatal intensive care. Although there is evidencethat these are largely appreciated and valued by parentswho have participated in research, there has been littlerigorous evaluation of their effectiveness. The authors ofseveral of the papers discussed here acknowledge the needfor research, but often do so from the perspective ofrecommending interventions or packages of care as evi-dently beneficial. We suggest that it is important toquestion a wide range of NICU-led bereavement practices,not with the aim of disproving their value, but with a neutralline of inquiry and the intention to refine and improve theiruse if beneficial. Conducting bereavement-related RCTs insuch a difficult setting is challenging but not impossible.Neonatal intensive care has a good record of collaborativeresearch evaluating new and existing technologies. Be-reavement care is, however, a sensitive subject, and thevulnerability of potential research participants is an impor-tant consideration which must be taken into account. Thiscreates ethical dilemmas for the conduct of RCTs. Eventhough it is widely recognized that interventions have notbeen adequately evaluated, and the possibility of deliver-ing potentially ineffective or even deleterious interventionsraises ethical issues in its own right, the lack of RCTssuggests an understandable reluctance to discuss researchand randomize bereavement interventions for parents at anextremely stressful time in their lives.

In this review we have suggested ways in which researchmight move forward, using less traditional RCT designs.Whilst these have their limitations, they do offer alterna-tive approaches to evaluating bereavement interventionsthat may be more acceptable to parents and clinicians.Before designing new studies, however, there needs to beconsensus amongst the relevant stakeholders as to theappropriate outcomes and time points at which theseshould be measured. To rise to the challenges inherent inthis field, researchers should seek out the questions ofimportance, and design high-quality studies e whetherquantitative or qualitative e which are capable of address-ing those questions appropriately and with sensitivity.

Acknowledgement

This review forms part of the BRACELET Study which isbeing funded by the NIHR Health Technology AssessmentProgramme.

Disclaimer: the views and opinions expressed in thisarticle do not necessarily reflect those of the NHS or theDepartment of Health.

References

1. Dent A, Condon L, Blair P, Fleming P. A study of bereavementcare after a sudden and unexpected death. Arch Dis Child1996;74:522e6.

2. Schneiderman G, Winders P, Tallett S, Feldman W. Do childand/or parent bereavement programs work? Can J Psychiatry1994;39(4):215e8.

3. Chambers HM, Chan FY. Support for women/families after peri-natal death. Cochrane Database Syst Rev 2000;(2). CD000452.

4. Rowa-Dewar N. Do interventions make a difference tobereaved parents? A systematic review of controlled studies.Int J Palliat Nurs 2002;8(9):452e7.

5. Gold KJ, Dalton VK, Schwenk TL. Hospital care for parentsafter perinatal death. Obstet Gynecol 2007;109(5):1156e66.

6. Forte AL, Hill M, Pazder R, Feudtner C. Bereavement care inter-ventions: a systematic review. BMC Palliat Care 2004;3(1):3.

7. Widger KA, Wilkins KL. What are the key components of qualityperinatal and pediatric end-of-life care? A literature review.J Palliat Care 2004;20(2):105e12.

8. Cohen L, Zilkha S, Middleton J, O’Donnohue N. Perinatal mor-tality: assisting parental affirmation. Am J Orthopyschiatry1978;48(4):727e31.

9. Rowe J, Clyman R, Green C, Mikkelsen C, Haight J, Ataide L.Follow-up families who experience a perinatal death. Pediat-rics 1978;62(2):166e70.

10. Schreiner RL, Gresham EL, Green M. Physician’s responsibilityto parents after death of an infant. Beneficial outcome ofa telephone call. Am J Dis Child 1979;133(7):723e6.

11. Mahan CK, Perez RH, Ratliff M, Schreiner RL. Neonatal death:parental evaluation of the NICU experience. Issues ComprPediatr Nurs 1981;5(5e6):279e92.

12. Forrest GC, Standish E, Baum JD. Support after perinataldeath: a study of support and counselling after perinatal be-reavement. Br Med J (Clin Res Ed) 1982;285:1475e9.

13. Mahan CK, Schreiner RL, Green M. Bibliotherapy: a tool to helpparents mourn their infant’s death. Health Soc Work 1983;8(2):126e32.

14. Harmon RJ, Glicken AD, Siegel RE. Neonatal loss in theintensive care nursery: effects of maternal grieving anda program for intervention. J Am Acad Child Psychiatry1984;23(1):68e71.

15. White MP, Reynolds B, Evans TJ. Handling of death in specialcare nurseries and parental grief. BMJ 1984;289:167e9.

16. Dyregrov A, Matthiesen SB. Parental grief following the deathof an infant-a follow-up over one year. Scand J Psychol 1991;32(3):193e207.

17. Calhoun LK. Parents’ perceptions of nursing support followingneonatal loss. J Perinat Neonatal Nurs 1994;8(2):57e66.

18. Harper MB, Wisian NB. Care of bereaved parents. A study ofpatient satisfaction. J Reprod Med 1994;39(2):80e6.

19. Lasker JN, Toedter LJ. Satisfaction with hospital care and inter-ventions after pregnancy loss. Death Stud 1994;8(1):41e64.

20. McHaffie HE. Supporting families when treatment is withdrawnfrom neonates: parental views on the role of the chaplain.Scottish J Health Care Chaplain 2000;3(2):2e7.

21. DiMarco M, Renker P, Medas J, Bertosa H, Goranitis JL. Effectsof an educational bereavement program on health care profes-sionals’ perceptions of perinatal loss. J Contin Educ Nurs 2002;33(4):180e6.

22. McHaffie HE, Laing IA, Lloyd DJ. Follow up care of bereavedparents after treatment withdrawal from newborns. Arch DisChild Fetal Neonatal Ed 2001;84(2):F125e8.

356 S. Harvey et al.

23. Lundqvist A, Nilstun T, Dykes AK. Both empowered and power-less: mothers’ experiences of professional care when theirnewborn dies. Birth 2002;29(3):192e9.

24. Reilly-Smorawski B, Armstrong AV, Catlin EA. Bereavement sup-port for couples following death of a baby: program develop-ment and 14-year exit analysis. Death Stud 2002;26(1):21e37.

25. Pector EA. Views of bereaved multiple-birth parents on lifesupport decisions, the dying process, and discussions surround-ing death. J Perinatol 2004;24(1):4e10.

26. Kavanaugh K, Hershberger P. Perinatal loss in low-incomeAfrican American parents. J Obstet Gynecol Neonatal Nurs2005;34(5):595e605.

27. Brosig CL, Pierucci RL, Kupst MJ, Leuthner SR. Infant end-of-lifecare: the parents’ perspective. J Perinatol 2007;27(8):510e6.

28. Primeau R, Recht Kahn. Professional bereavementphotographs: one aspect of a perinatal bereavement program.J Obstet Gynecol Neonatal Nurs 1994;23(1):22e5.

29. Jones B. Good grief: a medical illustrator’s view of bereave-ment photography. J Audiov Media Med 2002;25(2):69e70.

30. Sanchez NA. Mothers’ perceptions of benefits of perinatal losssupport offered at a major university hospital. J Perinat Educ2001;10(2):23e30.

31. Kobler K, Limbo R, Kavanaugh K. Meaningful Moments: the useof ritual in perinatal and pediatric Death. MCN Am J MaternChild Nurs 2007;32(5):288e95.

32. Capitulo KL. Evidence for healing interventions with perina-tal bereavement. MCN Am J Matern Child Nurs 2005;30(6):389e96.

33. Lundqvist A, Nilstun T, Dykes AK. Neonatal end-of-life care inSweden: the views of Muslim women. J Perinat NeonatalNurs 2003;17(1):77e86.

34. Toedter LJ, Lasker JN, Janssen HJ. International comparison ofstudies using the perinatal grief scale: a decade of research onpregnancy loss. Death Stud 2001;25(3):205e28.

35. Ritsher JB, Neugebauer R. Perinatal bereavement grief scale:distinguishing grief from depression following miscarriage.Assessment 2002;9(1):31e40.

36. Vance JC, Boyle FM, Najman JM, Thearle MJ. Couple distressafter sudden infant or perinatal death: a 30-month follow up.J Paediatr Child Health 2002;38(4):368e72.

37. Peppers LG, Knapp RJ. Maternal reactions to involuntaryfetal/infant death. Psychiatry 1980;43(2):155e9.

38. Capitulo KL. Perinatal grief online. MCN Am J Matern ChildNurs 2004;29(5):305e11.

39. Davies R. New understandings of parental grief: literaturereview. J Adv Nurs 2004;46(5):506e13.

40. Hughes P, Turton P, Hopper E, McGauley GA, Fonagy P. Disor-ganised attachment behaviour among infants born subsequentto stillbirth. J Child Psychol Psychiatry 2001;42(6):791e801.

41. Radestad I, Surkan PJ, Steineck G, Cnattingius S, Onelov E,Dickman PW. Long-term outcomes for mothers who have orhave not held their stillborn baby. Midwifery; 2007 Dec 5[epub ahead of print].

42. Ellard A. The memory box program. Can Nurse 1997;93(2):31e3.

43. deJong-Berg MA, deVlaming D. Bereavement care for familiespart 1: a review of a paediatric follow-up programme. Int JPalliat Nurs 2005;11(10):533e9.

44. Lumley J, Lester A, Renou P, Wood C. A failed RCT todetermine the best method of delivery for very low birthweight infants. Control Clin Trials 1985;6(2):120e7.

45. Klein MC. Studying episiotomy: when beliefs conflict withscience. J Fam Pract 1995;41:483e8.

46. Silverman WA. Personal reflections on lessons learned fromrandomized trials involving newborn infants from 1951 to1967. Clin Trials 2004;1(2):179e84.

47. Parkes CM. Guidelines for conducting ethical bereavementresearch. Death Stud 1995;19(2):171e81.

48. Pope C, Mays N. Reaching the parts other methods cannotreach: an introduction to qualitative methods in health andhealth services research. BMJ 1995;311:42e5.

49. Statham H. Prenatal diagnosis of fetal abnormality: thedecision to terminate the pregnancy and the psychologicalconsequences. Fetal Matern Med Rev 2002;13:213e47.

50. Zeanah CH. Adaptation following perinatal loss: a criticalreview. J Am Acad Child Adolesc Psychiatry 1989;28(4):467e80.

51. Oakley A. Social support and motherhood: The Natural Historyof a Research Project. Oxford: Blackwell; 1992.

52. Lautrette A, Darmon M, Megarbane B, et al. A communicationstrategy and brochure for relatives of patients dying in the ICU.N Engl J Med 2007;356(5):469e78.

53. Campbell M, Fitzpatrick R, Haines A, et al. Framework fordesign and evaluation of complex interventions to improvehealth. BMJ 2000;321:694e6.

54. Torgerson DJ, Sibbald B. Understanding controlled trials. Whatis a patient preference Trial? BMJ 1998;316:360.

55. Campbell MK, Elbourne DR, Altman DG, CONSORT Group. TheCONSORT statement: extension to cluster randomized trials.BMJ 2004;328:702e8.

56. Brown CA, Lilford RJ. The stepped wedge trial design: a system-atic review. BMC Med Res Methodol 2006;6(54):1e9.