British Journal ofPsychiatry (1992), 161,211—216
Women Whose Mental Illnesses Recur after Childbirth and Partners' Levels of Expressed Emotion During Late Pregnancy M. N. MARKS,A. WIECK, A. SEYMOUR,S. A. CHECKLEYand R. KUMAR Expressedemotion(EE)inthepartners of25 pregnantwomen witha history ofpsychosis orseveredepression and in13 pregnantcontrol subjects withoutany previous psychiatric
disorderwas assessedin the ninth monthof pregnancy.At this time, no patientpre sented as a case accordingto RDC. Elevensubjectswith a history of psychiatricdisorder experienceda furtherepisodeof illnessinthe sixmonthsfollowingdelivery.Partnersof women who becameill had madefewer criticaland positivecommentsabouttheir wives duringthe pregnancythanthepartners ofwomen who remainedwell. Poorself-rated social adjustment in the partnerswas also predictiveof recurrenceof illnessafter delivery.
Women with a history of functional psychosis or severe depression (puerperal or non-puerperal) are at a high risk of relapse after childbirth. Estimates of risk range from 20—30% (Brockington et a!, 1982) to 50% (Schopf et a!, 1984; Davidson & Robertson, 1985; Dean et a!, 1989). Many investigators have
last month of pregnancy, i.e. shortly before the time of greatestlikelihood of relapse, but when the women were all well.
commented on a higher incidence of postpartum psychoses in women without partners, i.e., those
The subjectswere25 multiparous,marriedor cohabiting womenandtheirpartners.A womanwasdefmedashaving
unmarried
a partner if she had been living with the man for at least
(Jones,
1902; Tetlow,
1955; Kendell
Method
eta!,
1981), separated, widowed or divorced (Kendell eta!, 6 months at 36 weeks into her pregnancy, and if the partner 1987), or whose husbands were absent (at war) said he was the father of the new baby. The women were duringthepregnancy and delivery (Jacobs, 1943). drawn from a larger sample of 47 pregnant women with Kendell et a! (1987) noted that single women were a historyof affectivepsychosisor non-psychoticdepressive also more likely to make out-patient
disorder. Nine women did not have partners, and four of
contacts with
the remaining38 were first-timemothers.They are not
psychiatric services, suggesting that the higher admissionratespostpartumwere not simplya consequence
of preferential
admissions
included
in this report because
there were different
proportionsof first-timemothersin theindexandcontrol groupsandwithinthe subgroupsof womenwho relapsed andthosewhodidnot. Ourinitialanalysessuggestedthat
for single
women. It seems therefore that having a partner present can reduce the likelihood of postnatal relapse
first-time mothers would be better studied separately from
in otherwise vulnerable women, and the aim of this study was to determine whether certain characteristics of spouses predicted psychiatric outcome after childbirth inwomen atriskof relapse.
multiparae because their partners tended tomakefewer criticalcommentsaboutthemandof course,unlikeallbut oneof themultiparae,noneof theirpreviousillnesseswere associatedwith childbirth. Threemultiparaedeliveredbeforethe interviewerhad completedherEEtraining,threebabieswerebornbefore the partnerscould be interviewed,and three partners
Many studies have reported
that high expressed
emotion (EE) in key relatives is predictive of psychiatric relapse, in particular in schizophrenia
refused to be interviewed.
(Vaughn & Leff, 1976; Vaughn eta!, 1984) but also
There were 47 control mothers,
and 13partnersof the23 multiparaein thisgroupagreed in bipolar affective disorder (Miklowitz et a!, 1988) to beinterviewed; thisrate(57'lo)is too lowforourcontrol and neurotic depression (Vaughn& Leff,1976; sampleto be regardedas representativebutis comparable Hooley eta!, 1986). In the latter two studies spouses
to rates found in other studies involving couples (for
were assessed while their partners were in hospital for depression, and in both studies a further relapse in the nine-month follow-up was associated with high levels of criticism by the spouse during the index episode. We investigated the partner's EE and also his psychiatric history, emotional stability, and social adjustment. These factors were assessed during the
interested fathers.
example, in Birtchnell's study (1988) of depression in womenin the community,in two samplesof couples,64% and62%of husbandsagreedto aninterview).Itseemsthat the controlgroup may be biasedto more compliantor Psychiatricassessmentsof patientswerecarriedout by the secondauthor,a psychiatrist,who was not involved in spouseinterviews.ResearchDiagnosticCriteria(RDC)
211
diagnoses (Spitzer et a!, 1978) were made of the woman's
212
MARKS ET AL
historyof majordepressivedisorder,it was2/9(22%),both the 12monthsprecedingexpecteddeliverydate,at36weeks relapses non-psychotic. All of the women admitted to a antenataland from then until six monthsafter delivery. psychiatrichospitalafter this deliveryhad a previoushistory Diagnoses were based on case-notes; SADS (Schedule for of bipolaror schizoaffectivedisorder(38%of thisgroup). previous psychiatric history, psychiatric condition during
Affective Disorder and Schizophrenia; Endicott & Spitzer,
There were no statistically significant differences between
1978) interviews at 36 weeks antenatal and 6 months postpartum; and PSE (Present State Examination; Wing
womenwho becameill and those who did not in age at delivery;parity; age at first illness;total time admitted to
eta!,1974) interviews at36weeksantenatal and4 days, a psychiatric hospital; time since last admitted; time since last on neuroleptic or antidepressant medication; and 6 weeks, and 6 monthspostpartum. EE was assessedusing the abbreviatedversionof the numberof puerperalpsychiatricadmissions,non-puerperal CamberwellFamilyInterviewSchedule(Vaughn& Leff, admissions, and puerperal and non-puerperal admissions 1976), which was modified to cover issues involving the combined.Therewereno differencesbetweenindexand woman'spregnancy,as well as any previousand current control subjects in age and parity. psychiatric illness orsymptoms. Theinitial section ofthe interview,thatconcerningepisodesof psychiatricillness, Partners' EE was omitted for control partners. Interviews and ratings were carried out by the first author, who had completed There were no differencesbetweenpartnersof women theMRCSocialPsychiatryUnit(C. Vaughn)EEtraining who becameill and those of womenwho remainedwell course and had achieved inter-raterreliability coefficients and control partnersin their occupation, duration of of greaterthan0.80 on eachdimensionof EE. Onthetwo marriage/cohabitation, psychiatric history(RDCdiagnosis), dimensionswhichwereimportant in this study, criticaland and EPQ neuroticismscoresat 36 weeksantenatal(see positive comments, the inter-rater reliability coefficients Table 1). were greater than 0.90. Thenumberof positivecommentswasthebestpredictor EE interviewswerecarriedout in the couple'shomeat of outcome.Partnersof high-riskwomenwho remained 36 weeksantenatally.Also assessedat this timewerethe well were most positive and those of high-riskwomen who man's psychiatric history and current psychiatric state (RDC diagnoses
based
on a SADS-Lifetime
interview).
Social
became ill were least positive, while the mean for control partners lay between the two index groups. The pattern for
adjustment was measured through the Social Problems Questionnaire (SPQ; Corney & Clare, 1985), which is
criticalcommentswassimilarbutnotstatistically significant.
a 33-item self-report
according to different levels of critical and positive comments is summarised in Table 2. It can be seen that
schedule derived from the Social
MaladjustmentSchedule(Clare&Cairns, 1978),and which assesses the subject's satisfaction with five social domains:
housing,work,fmancialsituation,maritalrelationship, and other interpersonal relationships. The spouse also completed the Eysenck Personality Questionnaire Eysenck, 1975).
(EPQ; Eysenck &
The rate of relapse when partners were categorised at all cut-off points a high number of positive or critical Table1 Characteristicsof partners(mean(s.e.m.)) Well (n=11)Control(n=13)x2 ANOVAsOccupation5:95:65:8NSmanual (n=14)Relapsed
and
Results Sixteen of the high-risk group had a history of bipolar or
schizoaffectivedisorder, and nine womenhad a history of major depression. No patient presentedas an RDC case at 36 weeks antenatal. Two women were RDC cases during the 12 months preceding their delivery. One was manic
(bipolardisorder)duringthepregnancybuthadrecovered by 7 monthsantenatal,andthe otherwomanhada panic disorderattheverybeginningof herpregnancy.Allpatients
: non manualDuration of5.97.47.7NScohabitation(0.7)(1.4)(1.5)Previous
RDC6/144/114/13NSdiagnosis(43%)(36%)(31%)Neuroticism9.0810.8210.17NS(E
0.42F(2,33)=3.57(SPQ)1(0.44)(0.90)(0.34)P