NATURAL HISTORY OF

SEVERE SYMPTOMS IN
BORDERLINE WOMEN
TREATED IN AN
INCEST GROUP

Jean M. Goodwin, M.D., M.P.H.

Nancy Wilson, M.D.
Virginia Connell, R.N., M.S.

Jean M. Goodwin, M.D., M.P.H. is Professor in the Department the high level of baseline severity and the tendency of borderlines to
of Psychiatry and Behavioral Science at the University of respond negatively to any form of treatment contributed to our erro-
Texas Medical Branch, Galveston, Texas. neous impression that group treatment exacerbated symptoms.
Dissociative diagnoses were associated with poor outcome.
Nancy Wilson, M.D. is a psychiatrist at St. Mary's Hospital
in Rhinelander, Wisconsin.
INTRODUCTION
Virginia Connell, R.N., M.S. is a nurse with Ancillary Home
Health Care in Milwaukee, Wisconsin. In 1986, two of the authors (Goodwin and Connell)
initiated a time-limited group for adult women with a his-
For reprints write jean M. Goodwin, M.D., M.P.H., University tory of incest who had experienced at least one prior psy-
of Texas Medical Branch, Department Psychiatry and chiatric hospitalization.
Behavioral Science, Galverston, Texas 77555-0428. In 1989, we described severe symptoms and severe abuse
histories in the first 10 participants (Goodwin, Cheeves &
This paper was just presented at the 9th International Connell, 1988). All but one suffered borderline personali-
Conference on Multiple Personality and Dissociative States, ty disorder. All had experienced multiple prior hospitaliza-
November 11-15, 1992 in Chicago, Illinois. This research tions, multiple suicide attempts and multiple diagnoses; all
was partially funded by a grant from the Frieda Brunn Fund, were disabled. All described multimodal child abuse by mul-
Milwaukee County Mental Health Complex. We gratefully tiple perpetrators. Modes of abuse included physical, sexu-
acknowledge the help of the former medical students at the al and emotional abuse, neglect and witnessed violence. A
Medical College of Wisconsin: Fortunato Milano, Rod replication study with the next 10 group participants con-
McVeety, and Maureen Leahy. firmed these patterns of severity, both in symptoms and in
childhood histories (Goodwin, Connell & Cheeves, 1990).
In the second study we proposed the acronym BAD FEARS
ABSTRACT as a way to list the multiple symptom complexes in these
patients: Borderline disorders, Affective disorders, Dissociative
In 1989 the authors described a cluster of severe symptoms in 10 symptoms, Fears (anxiety and other post-traumatic symp-
women treated in a time-limited group for incest survivors who had toms) , Eating disorders, Alcoholism or other substance abuse,
been psychiatrically hospitalized. All had experienced multiple prior Revictimizations, Somatization disorders and Suicidality
hospitalizations and multiple suicide attempts. Multiple diagnoses usually with compulsive self-mutilation.
including borderline and affective disorders were present in the first The present study used chart review and a convenience
l0 and in a replication sample. The present study explores two clin- control group to explore two clinical issues that arose dur-
ical questions raised in the treatment and follow-up of these patients: ing treatment of these fragile patients. First, we noted that
1) Did the trauma focused groups exacerbate severe symptoms in patients often became extremely distressed during group
some patients? and 2) On long-term follow-up did group treatment and later self-mutilated or were hospitalized. We were con-
lead to a greater likelihood of recovery ? We used chart review to fol- cerned that trauma-focused treatment had exacerbated
low emergency contacts and hospitalizations through three inter- severe symptoms. Later as we followed these patients, we
vals: the two years that preceded treatment, and the two years that observed that many relinquished acute psychiatric symp-
included treatment, and a two year follow-up interval. Acute con- tomatology, completed individual treatment, and resumed
tacts actually decreased during the treatment interval and on fol- functional work and family roles. We wondered if the trau-
low-up only one group-treated patient remained severely ill and sui- ma focus of treatment had led in the long term to a greater
cidal. In comparison a control group of hospitalized borderline women likelihood of recovery.
showed increased rather than decreased acuity in the treatment inter- To clarify these questions, chart review data were col-
val, but a similarly high level of pre-treatment acuity and a similar lected both for group members and controls for three inter-
"
50% likelihood of "recovery by the follow-up interval. Better out- vals: 1) the two years prior to initiating treatment; 2) the two
come far group participants was most evident when we compared years that encompassed the treatment (group for probands
the most severe cases in the two groups. The only suicide in the study and an index hospitalization for controls) and 3) a a two-
sample occurred in the control group. Lack of appreciation both of year follow-up interval.

221
DISSOCIATION. I'M. C. Nu. 4, December 1992
 11'41 '1 v!
I T
1 ` Al l ' ROUP

These data, presented below, helped place our clini- to patients with five to nine acute contacts over two years,
cal questions in the context of an emerging natural history and a "high acuity" is defined as 10 or more acute contacts.
of severe symptoms which seems to include both a prolonged "Acute contacts per year per patient" is a self-explanatory
period of severe acuity (Carpenter, Gunderson & Strauss, calculation devised to maintain comparability despite the
1975) and a high frequency of recovery (Stone, Stone & changes in denominator as patients were lost in the follow-
Hurt, 1987), regardless of treatment modality. up period.

METHODS CONTROLS

Intervention Controls were a convenience sample of 10 women with

Group treatment involved the following elements: I ) a primary or secondary diagnosis of borderline personality
two co-therapists; 2) a 12-session time-limited format with disorder hospitalized in the institution in the same year that
the possibility of pursuing more than one sequence; 3) 90- the group began.
minute sessions held in a Mental Health Center with ade- This sample had originally been chart reviewed for two
quate security; 4) fewer than 6 patients per group; 5) a require- related purposes: 1) to determine whether other hospital-
ment for ongoing individual therapy; and 6) encouragement ized borderlines met the high severity criteria found in incest
to stop therapeutic work in a particular group and resume group members; and 2) to determine the frequency with
a later sequence if sessions proved stressful. The therapeu- which child abuse histories were charted on inpatient units.
tic focus emphasized collecting information both about symp- Nineteen (about 10%) of the 194 patients admitted to
toms and childhood experiences and understanding how one of the index institution's psychiatric teaching wards dur-
responses to childhood experiences influenced present ing 1986 were discharged with a primary or secondary diag-
symptoms and coping strategies. nosis of borderline. For the present study, we excluded the
The number of group sessions utilized by a single patient five males in this sample and the two women who appeared
ranged from one to forty; in seven patients there was ongo- in this sample but who later participated in the incest group.
ing consultation between group and individual therapists; We included as controls the remaining 8 white females (all
six patients contacted the group therapists in the follow-up women in the incest group were white) and the two black
period and were interviewed; eight completed questionnaires women in the sample who had a secondary diagnosis (for
about symptoms and lifetime violence history. comparability with the probands, all of whom had at least
dual diagnoses).
Chart Review for Group Participants Although the presence of two probands in the original
Chart review for group participants collected 1) life- control group argues some comparability, the results of the
ti me severity indicators and 2) acuity indicators for pre-treat- earlier study indicated that hospitalized borderlines had lower
ment, treatment, and follow-up intervals. severity and less data available about childhood abuse. In
Lifetime severity indicators included total number of that study we generated chart review indicators of the BAD
psychiatric hospitalizations, age at first hospitalization, and FEARS or severe symptoms; all probands produced at least
presence of a life-threatening suicide attempt. six indicators while only two of the ten women included as
Acuity indicators for each time interval included: 1) controls had that many indicators of severe syndrome sta-
number of emergency visits; 2) number of hospitalizations; tus. Only eight of the ten controls showed charted infor-
3) days of psychiatric hospitalization; and 4) months of out- mation on childhood abuse or neglect; two described phys-
patient therapy (any month was counted in which at least ical and sexual abuse (the same two who showed severe
one outpatient contact was logged). symptoms); one physical and emotional abuse; one emo-
The index hospital was a large public teaching facility tional abuse with parental alcoholism; and three said a par-
which provided both general medical and psychiatric ser- ent had been psychotic. Of the two incest group members
vices. In these analyses we do not distinguish medical from who appeared in the inpatient sample, only one had child
psychiatric contacts. For example, an overdose leading to a abuse experiences charted.
medical emergency visit, a medical hospitalization, a psy-
chiatric emergency visit, and a psychiatric hospitalization RESULTS
was tabulated as two emergency visits and two hospitaliza-
tions. Severity Measures in Probands and Controls
Records from other institutions or outside therapists The 10 probands were white women ranging in age
were present or alluded to in some charts; however, these from 20 to 44 with a median age of 28. All had three or more
data were incomplete and inconsistent, and we did not tab- prior hospitalizations and all had made life threatening prior
ulate them. All patients were receiving medications during suicide attempts. Number of lifetime hospitalizations ranged
the study interval, but we did not tabulate this either. from three to 13 with a median of 10. Age at first hospital-
"Total acute contacts" refers to the sum of emergency ization ranged from 13 to 28 with a median of 21. Duration
visits and hospitalizations in the index institution in a par- of illness (calculated as the years between first psychiatric
ticular interval. hospitalization and entry into treatment) ranged from one
"Low acuity" is defined as four or fewer acute contacts to 18 years with a median of eight years. Nine had a bor-
for a patient in a two-year interval; " moderate acuity" refers derline diagnosis. The tenth was diagnosed as having a mixed

222
l1ISSOC1.ATION, vol. V. No, 4. December 1992
 GOODWIN/WILSON/CONNELL

personality disorderwith dependent features. All had acoex- accounted for 410 days of hospitalization in the last inter-
isting affective diagnosis. Seven had eating disorders and val.
seven gave addiction histories. All gave histories of physical Only one group participant was lost to follow-up; she
and sexual abuse in addition to other types of abuse and had entered the group with a diagnosis of fugue state and
neglect. fuged again one year into the treatment interval.
Of the controls eight were white and two were black. Looking case-by-case rather than at group data, we found
They ranged in age from 23 to 40 with a median of 35. Nine that four patients had more than doubled their acute con-
had three or more hospitalizations prior to the treatment tacts during the treatment period; in two the increases took
interval but only four had made a life-threatening suicide them into the moderate and high acuity ranges while the
attempt_ Number of lifetime hospitalizations ranged from other two remained in the low acuity range.
0 to 22 with a median of five. Age at first hospitalization Nine probands had some acute contact both in the pre-
ranged from 15 to 34 with a median of 27. Duration of ill- treatment and treatment intervals (eight were hospitalized
ness ranged from 0 to 19 years with a median of 9.5. All had in each of those intervals). In the follow-up interval only four
borderline diagnoses. Six had a coexisting major affective had acute contacts and three of those patients had low acu-
disorder. Two had eating disorders and one gave an addic- ity, logging a total of four contacts for the following corn-
tion history. Two had no Axis 1 diagnoses at all. Only two plaints: anxiety attack, pelvic pain related to pelvic inflam-
gave histories of combined physical and sexual abuse; six matory disease, nausea diagnosed as secondary to hepatitis,
others had other types of adverse childhood experiences. and elective surgery to repair ear damage secondary to phys-
Although lifetime chart review indicated lower acuity ical abuse. Only one proband remained acutely ill and chron-
in controls, when we looked only at the pre-treatment inter- ically suicidal. This woman had high acuity throughout and
val, five controls and four probands were at high or mod- was diagnosed as having multiple personality disorder in the
erate levels of acuity. treatment interval. She accounted for five of the six hospi-
In summary probands tended to differ from controls, talizations, 410 of the 411 hospital days and nine of the 12
as follows: emergency visits logged by the probands in the follow-up
interval. Chart review, interviews and consultation indicat-
1. They were younger. ed a great deal of conflict between inpatient and outpatient
therapists regarding her dissociative disorder diagnosis.
2. They had a younger age of onset of psychiatric This may have exacerbated acuity. Her 410 days hospital-
symptoms ized in a two-year interval was the highest seen in either
probands or controls. The next highest was 155 days per
3. They had more previous hospitalizations. interval logged respectively by one proband and one con-
trol.
4. A higher percentage had made life-threatening Of the three probands diagnosed with a dissociative
suicide attempts. disorder, one fugued and was lost to follow-up, one remained
acutely ill and one (diagnosed as having multiple personal-
5. A higher percentage had multiple diagnoses. ity in the pre-treatment interval) was the most ill of the three
patients still showing low acuity at follow-up.
In some areas the two groups were quite comparable. Five probands had attained zero acuity at follow-up
About half in both groups had mod-
erate or high acuity in the pre-treat-
ment interval, and both groups had
a median duration of illness of eight TABLE 1
to ten years. Acuity Measures for Group Participants in Pre-Treatment, Treatment
and Post Treatment Intervals.*
Acuity of Probands in the Pre-
Treatment, Treatment
and Follow-up Intervals Total
For group participants, total Acute Emergency Days in
acute contacts declined at each Contacts Visits Hospitalizations Hospital
interval, with 59 contacts logged pre-
treatment, 51 during treatment and
18 post-treatment (See Table 1).
Pre-Treatment 59 37 22 383
Emergency visits declined more
steadily than hospitalizations, going Treatment 51 28 23 155
from 37 to 28 to 12 while hospital- Post-Treatment 18 12 6 411
izations went from 22 to 23 to six.
Days in hospital changed from 383 *One patient was last to follow-up in the treatment interval.
to 155 to 411, but a single patient

223
[ASSOC 1:)TION. Vol. V. No. -1. Drcrmlrr 1 1 192
 NI 111 ~i►r '4W'i I 11► All •1#

although four had logged outpatient visits in that two-year ple life-threatening suicide attempts. Both completed 24 or
interval. Of these five patients, two had low acuity both in more months of psychotherapy with termination and resumed
pre-treatment and treatment intervals despite a prior his- jobs.
tory oflife-threatening overdoses. Both pursued greater than In four of these five cases of recovery both patient and
24 months of outpatient treatment with termination and psychiatrist felt there had been a positive response to med-
both resumed jobs. Both remained well at follow-up despite ications.
major stresses: one had married and one had been diag-
nosed as having a life-threatening physical illness. A third Comparison of Acuity and Outcome in Probands and Controls
patient had moderate acuity pre-treatment which increased The 10 Control women showed slightly higher rates of
to high acuity in the treatment intervals with physical fights, acuity in the pre-treatment interval (3.3 acute contacts per
psychotic symptoms, medication toxicity and somatic symp- patient year versus 3.0 acute contacts per patient year in
toms. She was stabilized in a home health care program but probands) . They also logged more outpatient months in the
remained disabled for work. Two had high acuity with 18 index institution than did probands (108 for controls ver-
and 22 contacts in the pre-treatment interval due to multi - sus 28 for probands), indicating that their treatment was
more concentrated in the index insti-
tutions.
Excess acuity in controls was also
TABLE 2 observed in the treatment and fol-
Total Acute Contacts in Probands and Controls During Pre-Treatment, low-up intervals (See Table 2.) The
Treatment and Follow-up Intervals.* difference attained significance only
during the treatment interval when
controls averaged 4.5 acute contacts
Probands Controls per year and group members logged
2.7. Comparison of total acute con-
Total *Rate Per Total *Rate Per tacts yielded a chi square of8.94with
Per Year Patient Per Year Patient P less than .01.
One control suicided one year
into the treatment interval. Asecond
Pre-Treatment 59 (3.0) 66 (3.3) control moved away at the beginning
of the follow-up interval.
Treatment 51 (2.7) 86 (4.5) Despite these dropouts and the
very long hospitalization described
Follow-Up 18 (1.0) 30 (1.9) in one proband, controls outstripped
probands in total days in hospital over
*Rates are based on a denominator of l0 for both groups pre-treatment, 9.5 for both the 6 years, logging 1026 days as com-
groups during treatment, and 9 probands and 8 controls during follow-up. pared to 949 for probands. Controls
also utilized more months of outpa-
tient therapy over the 6 years with
297 months versus 233 in probands.
Only in the follow-up interval did
TABLE 3 probands outstrip controls in out-
Outcome in High Severity* Probands and Controls patient utilization, 103 months to 70
months. The control patient who sui-
cided logged only five months of out-
Probands Controls patient therapy, the second lowest in
this group.
Looking at individual patterns
Number 4 4 we find that five controls more than
doubled their acute contacts in the
Dead 0 1 treatment interval; two of these
remained at low acuity levels despite
Ill 1 2 the increase, one entered the mod-
erate range and two entered the high
Well 3 1 acuity range.
Overall, eight of 10 controls were
*High severity is defined as presence of a major affective diagnosis, a history of suicide hospitalized pre-treatment. All were
attempt and moderate or high acuity in the pre-treatment interval. hospitalized during the treatment
interval (this was a requirement for
their being defined as controls).

224
DISSOCIATION, Von.' No. 4. Decemher 1992
 GOODWIN/WILSON/ CONNELL

Eight had emergency contacts pre-treatment and 10 during ing to make statistical inferences which would overstep the
the treatment interval. data. Secondly, we tried to think through the problems, many
In the post-treatment interval, two patients remained of which tended to bias against finding a positive effect of
quite ill. One logged seven hospitalizations and 16 emer- group treatment in probands. Probands were younger. We
gency visits, the other three hospitalizations and three emer- noted that all five women under 30 in the combined study
gency contacts. Both remained acutely suicidal. As mentioned, group did poorly. The four who were probands failed to
another patient. had died by suicide i n the treatment inter- reach zero acuity and the one who was a control suicided.
val. One patient remained at a low level of acuity. A fifth Probands had higher severity. Also they were less attached
patient moved and was lost to follow-up. to the institution pre-treatment and might have been drawn
Of the two controls who told inpatient staff about sex- into greater institutional contacts through their entry to the
ual abuse in childhood, one suicided and one is the most group, thus leading to increased acute contacts in the treat-
acutely ill patient at follow-up. These were the only two patients ment and post-treatment intervals.
who met severe syndrome criteria on the chart review study. A design problem that biased towards a false finding
The sexually abused control still alive but ill was scored as of better outcome in probands was the requirement that
having dissociative symptoms on chart review because of voic- controls be hospitalized, which made their treatment con-
es described as inside her head and problems with finding dition also a measure of poor outcome. However, baseline
possessions that she could not remember having purchased. rates of hospitalization were high (eight in the pre-treatment
Both controls with eating disorders also had poor outcomes; interval for both groups); and eight probands were hospi-
one was the control who suicided and the other was the sec- talized during the treatment interval. Even if 10 acute con-
ond most ill control at follow-up. tacts are subtracted from the controls during the treatment
Five controls had attained zero acuity at follow-up. Two interval, their acuity during treatment remains higher than
had low acuity throughout and had never made a suicide found in probands with a chi square value of 4.9 and P still
attempt. Two others had low acuity pre-treatment which less than .05. A more serious problem with using standard
increased to moderate and severe levels during treatment. treatment as a comparison is that controls did not share in
Both had made life threatening suicide attempts in the past. the extra attention and hopefulness associated with partic-
The fifth patient had maximal acuity pre-treatment but had ipating in a research project.
never made a life-threatening suicide attempt. She was the
only patient of the 20 to log 72 months of outpatient treat- Applying the Results to Clinical Questions
ment over the six-year study interval. One focus of her treat- Does incest group treatment in this format exacerbate
ment concerned her feelings about her own mother ' s psy- severe symptoms? These data do not support this observa-
chosis, which had resulted in numerous foster care placements tion. For group participants, acute contacts actually decreased
for the patient in childhood. The patient wanted to be a bet- during the treatment interval, while that of controls increased.
ter parent to her own children. Although four group members showed increased acuity dur-
ing treatment, five controls showed similar increases. The
Comparison of Outcome in High Severity Probands and Controls one completed suicide occurred during the treatment inter-
We also compared outcome in probands and controls val but in a control. Contrary to expectation, acuity was sig-
focusing only on individuals with high severity. We defined nificantly higher during the treatment interval among con-
high severity as the presence of a major affective diagnosis trols than in group participants.
in addition to borderline, a history of at least one life-threat- In retrospect, our concern about exacerbation can be
ening suicide attempt, and moderate or high acuity in the understood as follows: 1) We underestimated the high pre-
pre-treatmen t interval. Four patients in each group met these treatment levels of acuity in these patients. 2) We failed to
criteria. Three of the high severity probands attained zero appreciate the importance of negative reactions to treatment
acuity at follow-up, one was still quite ill. Only one of the in borderlines which were observed in controls as well as
high severity controls achieved zero acuity. Two were still group participants. This subject is much discussed in the lit-
quite ill at follow-up and one had died by suicide. erature (Kernberg, 1984) . 3) There may be a prolonged phase
of high acuity in the natural history of this syndrome that
DISCUSSION occupies about four years of the time course regardless of
treatment.
Problems with the Study New questions are raised by the finding that contrary
There are obvious problems with these data. The study to expectation, the treatment focus on child abuse was asso-
sample is quite small. The controls are a convenience sam- ciated with decreased acuity. It may be that the negative respons-
ple, known only by chart review (as opposed to the much es to treatment in borderlines are made more intelligible
richer clinical data available for probands) . Controls are dif- and workable when they can be understood in the light of
ferent from probands in age and severity. There is a great earlier childhood trauma and patients' concerns about
deal of missing data as only contacts at the index institution repeated abandonment or abuse.
were tabulated. Does a specific psychotherapy focus on child abuse favor
The authors have tried to address these problems in recovery? Perhaps. Five patients in each group had attained
two ways. First we present the data descriptively without try- zero acuity by the follow-up interval. However, when we looked

225
DISSOCIATION. Vol, V. No. 4, Du.iml ..r 1492
 ; 1'0 '1 ' "1 ' '1 # 1 I ll ' E T ROUP

only at the four patients in each group with the highest sever-
ity; three in the intervention group were "well" while only
one with standard treatment had improved to that level and
another control had died.
There was some evidence both in probands and con-
trols that those with dissociative disorders in addition to bor-
derline and mood disorders remained ill longer.
We could not rule out a specific focus on child abuse
in the psychotherapy of controls. Indeed in the recovered
control for whom records were most complete, this was doc-
umented as a therapeutic focus.

CONCLUSION

The group of psychiatric patients defined in this and

previous studies show high utilization of inpatient and emer-
gency services, high risk for completed suicide and prolonged
course, especially when dissociation complicates borderline
and mood disorders. However, about half recover in terms
of reaching zero acuity levels despite continued use of out-
patient treatment.
This study on group treatment, like a previous study
on screening (Goodwin, Attias, McCarty, et at., in press),
showed no evidence that a carefully designed focus on child
abuse exacerbated acuity. Indeed a favorable outcome for
group intervention was most demonstrable when we looked
only at those patients with the highest severity, patients who
might be screened out of trauma-focused therapy if one
assumed that such a focus would exacerbate symptoms. More
data are needed about this high-risk group of patients. ■

REFERENCES

Goodwin, J., Attias, R., McCarty, T., Chandler, S., & Romanik, R. (In
press) . Effects of routine questioning about childhood sexual abuse
in psychiatric inpatients. Victimology.

Goodwin, J. Cheeves, K, & Connell, V. (1988). Defining a syn-

drome of severe symptoms in survivors of severe incestuous abuse.
DISSOCIATION (1 (4), 11-16.

Goodwin, J. Cheeves, K., & Connell, V. (1990). Borderline and

other severe symptoms in adult survivors of incestuous abuse. Psychiatric
Annals, 20, 22-32.

Gunderson,J.G., Carpenter, W.T., &Strauss,J.S. (1975). Borderline

and schizophrenic patients: A comparative study. American Journal
of Psychiatry, 132, 1257-1264.

Kernberg, 0. (1984). Severe Personality Disorder.- Psychotherapeutic

Strategies. New Haven, CT: Yale.

Stone, M.H., Stone, D.K. & Hart, S. (1987). The natural history of
borderline patients: Global outcome. Psychiatric Clinics ofNorthAmerica,
10, 185-205.

226
l)lssoCI:11 ION,1 n1. V. No. I, Dr( mho r 1992