Acta Psychiatr Scand 2016: 1–7 © 2016 John Wiley & Sons A/S.

© 2016 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
All rights reserved ACTA PSYCHIATRICA SCANDINAVICA
DOI: 10.1111/acps.12643

The International Mood Network (IMN)

Nosology Project: differentiating borderline
personality from bipolar illness
V€
ohringer PA, Barroilhet SA, Alvear K, Medina S, Espinosa C, P. A. V€
ohringer1,2,3,4,
Alexandrovich K, Riumallo P, Leiva F, Hurtado ME, Cabrera J, S. A. Barroilhet3,5,6, K. Alvear2,7,
Sullivan M, Holtzman N, Ghaemi SN. The International Mood S. Medina2, C. Espinosa2,
Network (IMN) Nosology Project: differentiating borderline K. Alexandrovich2, P. Riumallo2,
personality from bipolar illness F. Leiva2, M. E. Hurtado2,
J. Cabrera2, M. Sullivan3,
Objective: The differential diagnosis of bipolar illness vs. borderline
personality is controversial. Both conditions manifest impulsive
N. Holtzman3, S. N. Ghaemi3
1 
Unidad de Trastornos del Animo, Clínica Psiquiatrica
behavior, unstable interpersonal relationships, and mood symptoms.
Universitaria, Hospital Clínico Universidad de Chile,
This study examines whether and which mood clinical features can
Facultad de Medicina Universidad de Chile, Santiago,
differentiate between both conditions. 2 
Clínica de Trastornos del Animo, Instituto Psiquiatrico
Method: A total of 260 patients (mean
standard deviation age “Dr. Jose Horwitz B”, Santiago, Chile, 3Mood Disorders
41
13 years, 68% female) attending to a mood clinic were examined Program, Tufts Medical Center, Boston, MA, USA,
for diagnosis of bipolar illness and borderline personality disorder using 4
Millenium Institute for Depression and Personality
SCID-I, SCID-II, and clinical mood criteria extracted from Mood Research, Ministry of Economy, Chile, Macul, Santiago,
Disorder Questionnaire (MDQ). They were analyzed using diagnoses as 5
Escuela de Psicología, Universidad de los Andes,
dependent variables. Predictors of bipolar and borderline diagnoses Santiago, 6Unidad de Psiquiatría de Enlace, Clínica
were identified by multivariable logistic regressions, and predictive Psiquiatrica Universitaria, Hospital Clínico Universidad
validity of models was assessed using ROC curve analysis. de Chile, Facultad de Medicina Universidad de Chile,
Santiago and 7Universidad Diego Portales, Santiago,
Results: Bipolar illness was strongly predicted by elevated mood
Chile
(OR = 4.02, 95% CI: 1.80–9.15), increased goal-directed activities
(OR = 3.90, 95% CI: 1.73–8.96), and episodicity of mood symptoms
(OR = 3.48, 95% CI 1.49–8.39). This triad model predicted bipolar Key words: bipolar illness; differential diagnosis;
illness with 88.7% sensitivity, 81.4% specificity, and obtained an borderline personality disorder; misdiagnosis
auROC of 0.91 (95% CI: 0.76–0.96) and a positive predictive value of Paul V€ohringer, Mood Disorders Program, Tufts Medical
85.1%. For borderline personality disorder, only female gender was a Center, 800 Washington Street #1007, Boston, MA
statistically significant predictor (OR = 3.41, 95% CI: 1.29–13.7), and 02111, USA. E-mails: PVohringer@tuftsmedicalcenter.
org, pvohringer@gmail.com
the predictive model obtained an auROC of 0.67 (95% CI: 0.53–0.74).
Conclusion: In a mood disorder clinic setting, manic criteria and
episodic mood course distinguished bipolar illness from borderline
personality disorder. Accepted for publication August 22, 2016

Significant outcomes
• Bipolar illness (BI) and borderline personality disorder (BPD) can be distinguished using clinical cri-
teria in a mood clinical setting
• Elevated mood, increased goal-directed activities, and episodicity are the strongest predictors of BI in
a mood setting
• Female gender is the only predictor of BPD in a mood setting.
Limitations
• Sample was obtained from a tertiary mood clinic, therefore enriched for mood symptoms.
• Based on a mood clinic sample, the above-mentioned clinical predictors may rule out borderline and
not viceversa.
• These results cannot be generalized to personality disorder settings.

1
V€
ohringer et al.

Introduction disorder as assessed with Structured Clinical Inter-

view for DSM-IV interview (31).
In clinical practice, patients with maladjusted
behavior, unstable interpersonal relationships, and
intermittent affective symptoms are common and Methods
can be difficult to accurately diagnose (1–3). This This cross-sectional study was carried out in a ter-
difficulty in part arises because such symptoms can tiary clinic specialized in mood disorders (MDC),
indicate the presence of either bipolar illness, bor- after receiving institutional review board approval.
derline personality disorder, or both conditions This clinic attends nearly 500 patients, primarily
concurrently (4). those with BI. It receives nearly 200 new referrals
Bipolar illness (BI) is a chronic and recurrent annually.
disease that often goes undiagnosed in primary
care and psychiatric settings (5–8). The same has
Measures
been claimed for borderline personality disorder
(BPD) (9, 10). Both entities entail significant psy- Patients with BI and BPD were diagnosed using
chosocial morbidity, mortality, poor quality of life, Structured Clinical Interview for DSM-IV Axis I
and pose a heavy burden on health providers and Disorders: Clinical Version (SCID-I CV) (32) and
relatives (10). Structured Clinical Interview for DSM-IV Axis II
In one review, approximately one in every seven Disorders (SCID-II) (33) respectively. This method
patients with BI also met diagnostic criteria for was used as a ‘gold standard’ to differentiate
BPD (11). However, high comorbidity may not be between BI and BPD. Although no psychiatric
true comorbidity given that these two conditions diagnostic method is flawless, SCID interviews are
share overlapping phenomenological characteris- well-validated and often-used tools in psychiatric
tics like mood instability, impulsiveness, and inap- research (34, 35).
propriate anger (12–14). Additionally, the Mood Disorder Questionnaire
This overlap of DSM criteria has produced clini- (MDQ) (36) was used as a routine mood criteria
cal controversy. Some focus on the overlap and assessment. The MDQ is a simple, brief question-
argue that one of the two syndromes is more naire that identifies manic or hypomanic symptoms.
important or more prominent than the other (15, In question 1 (Q1), participants are asked to respond
16). Others suggest that it is important to focus on ‘yes’ or ‘no’ to 13 items (Q1.x) designed to identify
clinical features that are different, rather than hypomanic/manic criteria as defined by the Diag-
shared, between the two diagnoses (17). nostic and Statistical Manual of the American Psy-
We identified only one study that provided data chiatric Association, 4th version (DSM-IV) (12, 23)
that could identify symptoms differentiating the as well as from clinical experience. Question 2 (Q2)
two conditions (18). In that report, BPD features examines whether these symptoms were experienced
of dissociative symptoms, fears of abandonment, concurrently. Question 3 (Q3) rates functional
identity disturbance, and parasuicidal behaviors impairment due to these symptoms by asking sub-
did not occur in BI (19), although others have jects to rate the difficultly in completing various
reported some evidence of self-harm in BI (20). A actions on a four-point scale. Question 4 (Q4)
few other studies reported that mood instability in assesses history of BI in blood relatives. Finally,
BI had a different character than mood instability question 5 (Q5) asks about previous suspicion of BI
in BPD (21–23), or that the former are less psy- by other health professionals. The Chilean Spanish
chosocially influenced than the latter (3, 23–25). version used in this study was previously validated
Direct comparisons of the two groups are infre- by the study authors (37). Manic/hypomanic symp-
quent, however (21, 26–30). toms were retrieved from this questionnaire.
MDC investigators were trained to conduct the
Aim of the study
SCID-I CV to detect manic symptoms and diag-
nose BI, and SCID-II to diagnose personality dis-
Given the paucity of clinical research on this orders. Clinical interviews also identified other axis
important diagnostic question, the International I disorders in addition to BI within the sample.
Mood Network Nosology Project aimed to Personality disorders diagnosis using clinical and
directly compare borderline personality disorder SCID-II interviews were highly correlated between
and bipolar illness patients in a mood disorder raters, with a kappa = 0.9. The same degree of cor-
clinic to identify which self-reported Mood Disor- relation was obtained regarding axis I disorders
der Questionnaire criteria and clinical features pre- other than BI, obtaining a high degree of inter-
dict bipolar illness and borderline personality rater reliability of kappa = 0.8.

2
 Differential diagnosis between bipolar and borderline

statistic, along with its 95% confidence interval,

Subjects
was used to ascertain discrimination capacity. Two
The study sample consisted of 260 subjects recruited models were assessed: The full one and ‘the triad’
between April 2009 and December 2012. Of total one. The latter comprises the three stronger predic-
sample, 177 (68%) were female, and mean age was tors from the full model. Hypothesis testing analy-
41
14 years. Patients referred to MDC were con- ses between them were made using a chi-square
secutively asked about participation by an investi- test in order to detect whether a more parsimo-
gator upon arrival. The study was not otherwise nious and easy-to-use model could be applied by
advertised. All patients screened for the study clinicians. Sensitivity, specificity, positive predic-
agreed to participate and provided written consent. tive value (PPV), negative predictive value (NPV),
The sample was divided, regarding their diagno- and the proportion of correctly classified patients
sis, in four mutually exclusive groups: patients with of the triad model were also reported.
BI, patients with BPD, patients that have both Data were analyzed using R statistical package
conditions (comorbidity), and patients that have (38).
neither.
Results
Statistical analysis
Using SCID-I and SCID-II, 45% of the total sam-
Chi-square or Fisher’s exact test was used to per- ple was diagnosed exclusively with BI (type I or
form a bivariate analysis in order to select which II), 20% with BPD, 9% with both, and 25% with
clinical mood criteria could predict better the two neither.
main outcomes: bipolar illness diagnosis and bor- As seen in Table 1, BI-only patients were 100%
derline personality disorder diagnosis. Effect esti- more episodic and 114% more familial than
mates were reported as odds ratios (OR) along patients with BPD, while female gender was more
with their 95% confidence intervals and P-values. common in BPD than in BI. Both conditions led to
Items that obtained P-values equal to or lower similar functional impairment.
than 0.1 were then included in logistic regression Table 2 shows unadjusted bivariate analyses.
modeling. Gender and age were also included as Bolded clinical predictors were significant at
clinical covariates in the models. BI diagnosis was P < 0.1. Those predictors meeting that criterion
the outcome (dependent) variable in the first model were entered into the multivariate regression mod-
and BPD diagnosis was the outcome variable in els. Tables 3 and 4 show the final multivariate
the second model. A stepwise/backward AIC crite- regression models for BI and BPD diagnoses
rion selection procedure plus clinical criteria were respectively. Bipolar illness was strongly predicted
applied in order to decide which variables would by the MDQ0 s triad of elevated mood (Q1.1: ‘you
be included in the final predictive models. felt so good or so hyper that other people thought
you were not your normal self or you were so
hyper that you got into trouble?’), increased goal-
Predictive validity of the models: ROC analysis
directed activities (Q1.9: ‘you were much more
Model discrimination was assessed by the auROC active or did many more things than usual?’), and
statistic given by a ROC curve. The auROC episodicity of mood symptoms (Q2: ‘If you

Table 1. Clinical and demographic description of subgroups, and comparison of bipolar illness and borderline personality disorder groups (no comorbidity)

Comorbidity bipolar Bipolar

Total sample illness and borderline Neither illness Borderline personality
(N = 260) personality (N = 24) (N = 66) (N = 118) disorder (N = 52) Effect estimate*

Age in years (mean
SD) 41
13.45 40
13.92 43.3
14.21 38
13.19 42
13.19 3.59 (7.92 to 0.74)
Gender (female) 177 (68%) 19 (79%) 46 (70%) 68 (58%) 46 (88%) 0.66 (0.07–0.28)
Family history of bipolar 123 (47%) 12 (50%) 29 (43%) 68 (58%) 14 (27%) 2.14 (1.16–2.32)
illness (Q4)
Previous suspicion of bipolar 178 (68%) 20 (83%) 28 (42%) 95 (81%) 35 (70%) 1.19 (0.02–2.90)
illness (Q5)
Significant impact in daily 207 (80%) 23 (95%) 44 (67%) 96 (82%) 44 (85%) 0.96 (0.01 to 2.63)
functioning of manic
symptoms (Q3)
Manic symptoms occur 184 (71%) 17 (71%) 34 (51%) 108 (92%) 25 (48%) 2.00 (1.03–2.52)
episodically (Q2)

*Relative risk for dichotomic variables or risk difference for continuous variables and 95% confidence intervals, comparing bipolar and borderline patients.

3
V€
ohringer et al.

Table 2. Bivariate analysis of mood criteria comparing bipolar illness and borderline checked YES to more than one of the above, have
personality disorder (in decreasing order of effect size)
several of these ever happened during the same per-
Bipolar illness Borderline personality iod of time?’). MDQ less strong predictors for BI
were presence of racing thoughts (Q1.6: ‘Thoughts
Mood criteria Odds Odds
raced through your head or you couldn’t slow your
(MDQ question.item) Ratio 95% CI Ratio 95% CI
mind down?’), less need for sleep (Q1.4: ‘you got
More goal active (Q1.9) 13.63 7.06–27.60 0.85 0.26–2.63 much less sleep than usual and found you didn’t
More energy (Q1.8) 10.91 5.81–21.28 0.94 0.30–2.90 really miss it?’), and increased self-confidence (Q1.3:
Elevated mood (Q1.1) 9.59 5.15–18.46 0.77 0.21–2.58
More talkative (Q1.5) 8.93 4.81–17.15 0.72 0.22–2.24
‘you felt much more self-confident than usual?’).
More social (Q1.10) 7.58 3.87–15.74 0.79 0.24–2.80 For BPD, only female gender was a statistically sig-
Episodicity (Q2) 6.59 3.44–13.16 0.85 0.26–2.63 nificant predictor.
More self-confidence (Q1.3) 5.84 3.28–10.61 0.81 0.26–2.47 Figure 1 (predictive capacity of the bipolar illness
More sexual (Q1.11) 5.32 2.97–9.78 0.69 0.22–2.17
Previous suspicion of bipolar 4.55 2.53–8.37 1.33 0.39–4.30 triad multivariate model) shows the ROC curve for
illness (Q5) the triad model predicting BI. The auROC statistic
Unusual behavior (Q1.12) 3.38 1.98–5.86 1.08 0.35–3.30 for this model was 0.91 (95% CI: 0.76–0.96) while
Less need for sleep (Q1.4) 3.06 1.75–5.40 0.28 0.05–1.14
for the full model was 0.95 (95% CI: 0.80–0.97).
Irritability (Q1.3) 3.01 1.67–5.50 0.90 0.03–1.55
Racing thoughts (Q1.6) 2.94 1.53–5.83 0.53 0.09–2.30 When comparing both auROC, no statistical differ-
Spend more money (Q1.13) 1.90 1.13–3.23 1.44 0.13–1.37 ence was found (chi2 = 2.36, P = 0.211). The triad
Easy distraction (Q1.7) 1.39 0.76–2.53 1.25 0.33–4.39 model predicted BI with 88.7% sensitivity and
Family history of bipolar 1.22 1.73–2.07 1.04 0.28–12.47
illness (Q4)
81.4% specificity and showed a positive predictive
value (PPV) of 85.1%, namely the triad model accu-
CI, confidence intervals. rately diagnosed BI in that proportion of the sam-
Bolded items have P-value ≤ 0.1 and were introduced into multivariate regression ple. Negative predictive value (NPV) was 85.7%,
modeling.
namely patients that did not have the triad did not
have BI 85.7% of times. When making differential
Table 3. Multivariate logistic regression model of clinical predictors of diagnosis of
diagnosis of BI and BPD in a mood disorder set-
bipolar illness* ting, the triad correctly classified 85.4% of BI
patients from the sample. On the other hand, the
Clinical features BPD model obtained an auROC statistic of 0.67
(MDQ question.item) Estimate Odds Ratio 95% CI
(95% CI: 0.53–0.74).
Elevated mood (Q1.1) 1.39120 4.02 1.80–9.15*
More goal active (Q1.9) 1.36208 3.90 1.73–8.96*
Episodicity (Q2) 1.24778 3.48 1.49–8.39* Discussion
More social (Q1.10) 1.08449 2.96 1.30–6.99*
Racing thoughts (Q1.6) 1.04769 2.85 1.28–6.53* In the setting of a mood disorder specialty clinic,
Less need for sleep (Q1.4) 1.01178 2.75 1.21–6.28* borderline personality and bipolar illness were dis-
More self-confidence (Q1.3) 0.88680 2.43 1.13–5.20* tinguishable based on a clinical triad of elevated
Age 0.00271 1.00 0.97–1.02
Gender 0.55485 0.57 0.24–1.30
mood, increased goal-directed activity, and episodic
course of illness. These results suggest that these
CI, confidence intervals. two conditions, despite overlap in criteria and
*Adjusted model built using stepwise/backward with AIC criterion selecting vari-
able process.

Table 4. Multivariate logistic regression model of predictors of diagnosis of border-

line personality disorder*

Clinical features
(MDQ question.item) Estimate Odds Ratio 95% CI

Gender 1.227094 3.41 1.29–13.70

Unusual behavior (Q1.12) 0.950204 2.58 0.73–10.05
Spend more money (Q1.13) 1.250713 1.28 0.06–1.02
Family history of bipolar illness (Q4) 1.688066 1.14 0.54–21.84
Age 0.002703 0.99 0.94–1.05
More talkative (Q1.5) 0.437456 0.64 0.17–2.24
Irritability (Q1.3) 1.350531 0.25 0.02–1.59

CI, Confidence intervals.

*Adjusted model built using stepwise/backward AIC criterion selecting variable Fig. 1. Predictive capacity of the bipolar illness triad multi-
process. variate model.

4
 Differential diagnosis between bipolar and borderline

reported common comorbidity, can be clinically notable that subtype of bipolar disorder (type I or
distinguished in a mood disorder clinic. These find- type II) was not assessed, which prevents making
ings may not generalize to other clinical settings differential assessments on the topic of type II
where the prevalence of both conditions could be bipolar illness vs. borderline personality. Other
lower, as in primary care clinics (39). The proposed important clinical features that also were not col-
triad may be combined with other non-DSM fac- lected include baseline depressive symptomatology,
tors which should be assessed for patients in whom years of being ill, age of onset, and number of pre-
bipolar illness is in the differential diagnosis. vious mood episodes. The lack of these clinical fea-
These results can be related to the previous liter- tures limits the generalizability of the results.
ature in a number of ways. Classic mania symp- While assessing personality in patients in an acute
toms, like euphoric mood and decreased need for mood state is controversial, some studies using
sleep, are not part of the diagnostic criteria of bor- SCID-II methodology have shown stability when
derline personality or other conditions. As pro- assessing mood state patients (42) in contrast with
posed in the DSM system, we confirmed that the other personality assessment tools (43, 44).
identification of mania criteria and episodicity of Overall, this study aimed for practitioners’ clini-
mood symptoms are specific to bipolar illness, and cal usefulness. In that regard, episodicity is proba-
can diagnose it positively. bly the only one BI predictor that does not directly
The clinical implication of these results is that if rely on mood. Therefore, it might be considered a
these criteria are present, then bipolar illness is more accurate predictor (17, 45). However, severe
likely to be present, and borderline personality is mixed states, as well as affective temperaments
less likely to be present. Additionally, this can be (such as cyclothymic and hyperthymia), have little
reinforced by the excellent predictive capacity (0.91) to no episodicity, with nearly constant mood
and PPV of the mood clinical triad (85.1%). In con- symptoms and no ‘well intervals’ to define ‘epi-
trast, mood criteria did not specifically help in pre- sodes’ (46). Our clinical observation of mood
dicting borderline personality. Only female gender patients goes in the same direction. Mood assess-
was associated with the latter. Again this can fur- ment depends on the observer and therefore is
ther be supported by the poor discriminative capac- more likely to be subject to clinical bias. The triad
ity of the BPD predictive model (0.67). Specific found in this study provides additional data to tie
borderline personality criteria were not assessed in diagnoses to objective findings; but clinical judg-
this study but obviously have been well demon- ment will still be required in interpreting them.
strated elsewhere to identify borderline personality
(40) even in samples with mood disorders (18, 19).
As this was a mood disorder specialty clinic setting, Funding
the patient population was not enriched for border- This study was supported by the Fund for Innovation and
line personality, but it was enriched for mood ill- Competitiveness (FIC) of the Chilean Ministry of Economy,
nesses. Thus, the absence of diagnostic criteria for Development and Tourism, through the Millennium Scientific
Initiative, Grant Number IS130005.
predicting borderline personality likely relates to
the fact that this sample was not obtained in a set-
ting where such patients were referred. Although References
current conceptions in psychiatric nosology are 1. Blacker D, Tsuang MT. Contested boundaries of bipolar
heading to dimensional models of disease rather disorder and the limits of categorical diagnosis in psychia-
than categorical ones, our daily clinical work places try. Am J Psychiatry 1992;149:1473–1483.
2. Tyrer SP, Brittlebank AD. Misdiagnosis of bipolar affec-
us in need of operational clinical criteria to work tive disorder as personality disorder. Can J Psychiatry
with. The triad may be useful in this respect. 1993;38:587–589.
It is noteworthy that while there were notable 3. Zimmerman M, Morgan TA. Problematic boundaries in the
subgroups of patients with bipolar illness only diagnosis of bipolar disorder: the interface with borderline
(45%) vs. borderline personality only (20%), fewer personality disorder. Curr Psychiatr Rep 2013;15:422.
4. Zimmerman M, Galione JN, Ruggero CJ et al. Screening for
subjects had the comorbidity of both (9%), in con- bipolar disorder and finding borderline personality disor-
trast to other clinical samples where higher comor- der. J Clin Psychiatry 2010;71:1212–1217.
bidities are reported (around 20%) (11, 41). 5. Ghaemi SN, Sachs GS, Chiou AM, Pandurangi AK, Goodwin
Referral bias (higher age, mood disorder specialty K. Is bipolar disorder still underdiagnosed? Are antide-
clinic setting) might explain the low comorbidity in pressants overutilized? J Affect Disord 1999;52:135–144.
6. Hirschfeld RM, Lewis L, Vornik LA. Perceptions and
this sample. This factor relates to the larger ques- impact of bipolar disorder: how far have we really come?
tion of generalizability. These results mainly apply Results of the national depressive and manic-depressive
to ruling out borderline personality in a mood dis- association 2000 survey of individuals with bipolar disor-
order sample, rather than vice versa. It is also der. J Clin Psychiatry 2003;64:161–174.

5
V€
ohringer et al.

7. Hirschfeld RM. Bipolar spectrum disorder: improving its 26. Benazzi F. A relationship between bipolar II disorder and
recognition and diagnosis. J Clin Psychiatry 2001;62(Suppl borderline personality disorder? Prog Neuropsychophar-
14):5–9. macol Biol Psychiatry 2008;32:1022–1029.
8. Vicente B, Kohn R, Rioseco P, Saldivia S, Levav I, Torres 27. Wilson ST, Stanley B, Oquendo MA, Goldberg P, Zalsman
S. Lifetime and 12-month prevalence of DSM-III-R disor- G, Mann JJ. Comparing impulsiveness, hostility, and
ders in the Chile psychiatric prevalence study. Am J Psy- depression in borderline personality disorder and bipolar
chiatry 2006;163:1362–1370. II disorder. J Clin Psychiatry 2007;68:1533–1539.
9. Paris J, Gunderson J, Weinberg I. The interface between 28. Zimmerman M, Martinez JH, Morgan TA, Young D,
borderline personality disorder and bipolar spectrum dis- Chelminski I, Dalrymple K. Distinguishing bipolar II
orders. Compr Psychiatry 2007;48:145–154. depression from major depressive disorder with comorbid
10. Zimmerman M. Improving the recognition of borderline borderline personality disorder: demographic, clinical, and
personality disorder in a bipolar world. J Pers Disord family history differences. J Clin Psychiatry 2013;74:880–
2016;30:320–335. 886.
11. Brieger P, Ehrt U, Marneros A. Frequency of comorbid 29. Galione J, Zimmerman M. A comparison of depressed
personality disorders in bipolar and unipolar affective dis- patients with and without borderline personality disorder:
orders. Compr Psychiatry 2003;44:28–34. implications for interpreting studies of the validity of the
12. Gunderson JG, Phillips KA. A current view of the interface bipolar spectrum. J Pers Disord 2010;24:763–772.
between borderline personality disorder and depression. 30. Berrocal C, Ruiz Moreno MA, Rando MA, Benvenuti A,
Am J Psychiatry 1991;148:967–975. Cassano GB. Borderline personality disorder and mood
13. Benazzi F. Borderline personality-bipolar spectrum rela- spectrum. Psychiatry Res 2008;159:300–307.
tionship. Prog Neuropsychopharmacol Biol Psychiatry 31. First MB, Gibbon M, Spitzer RL, Williams JBW, Benjamin
2006;30:68–74. LS. Structured clinical interview for DSM-IV axis II per-
14. Ghaemi SN, Dalley S, Catania C, Barroilhet S. Bipolar or sonality disorders, (SCID-II). Washington, DC: American
borderline: a clinical overview. Acta Psychiatr Scand Psychiatric Press, Inc. 1997.
2014;130:99–108. 32. Spitzer RL, Williams JBW, Gibbon M, First MB. Struc-
15. Gunderson EW, Coffin PO, Chang N, Polydorou S, Levin tured clinical interview for DSM-III-R, patient edition/
FR. The interface between substance abuse and chronic non-patient edition (SCID-P/SCID-NP). Washington,
pain management in primary care: a curriculum for medi- DC: American Psychiatric Press, Inc, 1990.
cal residents. Subst Abus 2009;30:253–260. 33. Spitzer RL, Williams JBW, Gibbon M, First MB. Struc-
16. Akiskal HS. Demystifying borderline personality: critique tured clinican interview for DSM-III-R axis II disorders
of the concept and unorthodox reflections on its natural (SCID-II). Washington, DC: American Psychiatric Press,
kinship with the bipolar spectrum. Acta Psychiatr Scand Inc, 1990.
2004;110:401–407. 34. Segal DL, Hersen M, Van Hasselt VB. Reliability of the
17. Ghaemi SN, Barroilhet S. Bipolar illness versus borderline structured clinical interview for DSM-III-R: an evaluative
personality: red skies versus red apples. In: Choi-Kain L, review. Compr Psychiatry 1994;35:316–327.
Gunderson J, eds. Borderline personality and mood disor- 35. Williams JB, Gibbon M, First MB et al. The structured
ders: comorbidity and controversy. New York, Heidel- clinical interview for DSM-III-R (SCID). II. Multisite
berg, Dordrecht, London: Springer, 2015: 97–115. test-retest reliability. Arch Gen Psychiatry 1992;49:630–
18. Perugi G, Angst J, Azorin JM et al. The bipolar-borderline 636.
personality disorders connection in major depressive 36. Hirschfeld RM, Williams JB, Spitzer RL et al. Develop-
patients. Acta Psychiatr Scand 2013;128:376–383. ment and validation of a screening instrument for bipolar
19. Barroilhet S, Vohringer PA, Ghaemi SN. Borderline versus spectrum disorder: the Mood Disorder Questionnaire. Am
bipolar: differences matter. Acta Psychiatr Scand J Psychiatry 2000;157:1873–1875.
2013;128:385–386. 37. V€ohringer P, Alvear K, Medina S, Espinosa C, Cabrera J.
20. Inder ML, Crowe MT, Luty SE et al. Prospective rates of Validation study of Mood Disorder Questionnaire
suicide attempts and nonsuicidal self-injury by young peo- (MDQ), in a Chilean outpatient sample. Rev Gaceta de
ple with bipolar disorder participating in a psychotherapy Psiq Univ 2008;5:339–344.
study. Aust N Z J Psychiatry 2016;50:167–173. 38. R Core Team. R: a language and environment for statisti-
21. Henry C, Mitropoulou V, New AS, Koenigsberg HW, Sil- cal computing. Vienna, Austria: R Foundation for Statis-
verman J, Siever LJ. Affective instability and impulsivity in tical Computing, 2010.
borderline personality and bipolar II disorders: similarities 39. Phelps J, Ghaemi SN. The mistaken claim of bipolar ‘over-
and differences. J Psychiatr Res 2001;35:307–312. diagnosis’: solving the false positives problem for DSM-5/
22. Reich DB, Zanarini MC, Fitzmaurice G. Affective lability ICD-11. Acta Psychiatr Scand 2012;126:395–401.
in bipolar disorder and borderline personality disorder. 40. Gunderson JG. Borderline personality disorder: ontogeny
Compr Psychiatry 2012;53:230–237. of a diagnosis. Am J Psychiatry 2009;166:530–539.
23. Koenigsberg HW. Affective instability: toward an integra- 41. McCormick B, Blum N, Hansel R et al. Relationship of sex
tion of neuroscience and psychological perspectives. J Pers to symptom severity, psychiatric comorbidity, and health
Disord 2010;24:60–82. care utilization in 163 subjects with borderline personality
24. Russell JJ, Moskowitz DS, Zuroff DC, Sookman D, Paris J. disorder. Compr Psychiatry 2007;48:406–412.
Stability and variability of affective experience and inter- 42. Ouimette PC, Klein DN. Test-retest stability, mood-state
personal behavior in borderline personality disorder. J dependence, and informant-subject concordance of the
Abnorm Psychol 2007;116:578–588. SCID-Axis II Questionnaire in a nonclinical sample. J Pers
25. Gunderson JG, Lyons-Ruth K. BPD’s interpersonal hyper- Disord 1995;9:105–111.
sensitivity phenotype: a gene-environment-developmental 43. Wetzler S. The million clinical multiaxial inventory
model. J Pers Disord 2008;22:22–41. (MCMI): a review. J Pers Assess 1990;55:445–464.

6
 Differential diagnosis between bipolar and borderline

44. Wasek T, Endicott J. Assessing personality: effects of the 46. Mackinnon DF, Pies R. Affective instability as rapid
depressive state on trait measurement. Am J Psychiatry cycling: theoretical and clinical implications for borderline
1983;140:695–699. personality and bipolar spectrum disorders. Bipolar
45. Antoniadis D, Samakouri M, Livaditis M. The association Disord 2006;8:1–14.
of bipolar spectrum disorders and borderline personality
disorder. Psychiatr Q 2012;83:449–465.