Borderline Personality Disorder and Migraine


  • John Rothrock MD,

  • Ivan Lopez MD,

  • Richard Zweilfer MD,

  • Diane Andress-Rothrock BA,

  • Renay Drinkard C-NP,

  • Naomi Walters

  • From the University of South Alabama Headache Center, University of South Alabama College of Medicine, Mobile, AL.

Address all correspondence to Dr. John Rothrock, University of Alabama at Birmingham, RWUH M226, 1530 3rd Ave. South, Birmingham, AL 35294-3280.


Background.—Borderline personality disorder (BPD) may be disproportionately common in the migraine patient population, but specific migraine features in the BPD subgroup remain incompletely characterized.

Purpose.—To define more clearly the clinical characteristics of migraine patients with BPD, to evaluate their clinical response to aggressive headache management, and to assess the sensitivity of an instrument intended to screen for the presence of BPD.

Methods.—We evaluated 50 consecutive patients with migraine and previously documented BPD, 50 consecutive patients with migraine of all types and no history of BPD, and 50 patients with migraine and no history of BPD matched to the first group for age, gender, and headache frequency. We assessed a variety of demographic and clinical variables at baseline, and to all patients we administered the McLean Screening Instrument for Borderline Personality Disorder (MSI-BPD). Patients in Groups 1 and 3 were treated for their headaches via a uniform pharmacologic management program. Treatment outcome was assessed after 6 months.

Results.—In this series of migraine patients, coexisting BPD was associated with female gender, more pervasive headache, more migraine-related disability, a higher prevalence of medication overuse headache, more unscheduled visits for acute migraine treatment, a higher prevalence of active, self-reported depression, and a lower likelihood of responding to pharmacologic therapy intended for headache management. The MSI-BPD was insensitive in detecting BPD.

Conclusions.—Migraine patients with coexisting BPD are clinically distinct from the migraine patient population as a whole; in particular, they are more severely affected by their headache disorder and more treatment refractory. Simple screening instruments used to detect BPD may be ineffective in the headache clinic setting.

Borderline personality disorder (BPD) is characterized by impulsivity, unstable and intense interpersonal relationships, an unstable self-image, frantic efforts to avoid perceived abandonment, inappropriate anger, self-destructive threats and behavior, transient paranoid ideation, dissociative symptoms, affective instability, or some combination thereof.1 BPD is relatively common in the general population, with a reported prevalence of about 2%, and BPD appears to be disproportionately prevalent in the migraine patient population.2,3 The treatment of migraine in patients with BPD can be effortful, costly, and ultimately fruitless. In this study, we sought to characterize more clearly the clinical characteristics of migraine in BPD patients presenting to our university-based headache clinic, to determine the prevalence of self-reported depression in those patients, to evaluate the impact of pharmacologic treatment prescribed to ameliorate their headaches, and to assess the sensitivity of a self-report instrument intended to screen for the presence of BPD.


We evaluated a consecutive series of 50 patients with migraine and BPD in whom the psychiatric diagnosis previously had been made by a psychiatrist or psychologist and who subsequently presented to a university-based headache clinic. We compared this group (Group 1) with another 50 consecutive migraine patients with no history of BPD presenting to that clinic (Group 2) and with a third group of 50 patients with migraine but no BPD matched to the BPD group for gender, age, and headache frequency (Group 3). We made the diagnosis of migraine according to existing IHS criteria, and we classified patients with migraine and ≥15 days of headache monthly according to criteria suggested by Silberstein et al.4,5 We further subtyped migraine as “episodic migraine” (EM ≤ 15 “headache days”[HDs] per month), frequent episodic migraine (FEM ≥ 15 HDs per month but not daily headache), “transformed migraine” (TM = daily headache), and “new daily headache with migrainous features” (NDPH w/MF).

The demographic and clinical variables we specifically examined at baseline in all 3 groups included age, gender, migraine subtype, “headache profiles” (ie, HDs per month/functionally incapacitating HDs per month), migraine-related disability (via Migraine Disability Assessment Scores [MIDAS]), prevalence of depression (via Quick Inventory of Depressive Symptomatology-Self Report [QIDS-SR] scores), and presence of presumed medication overuse headache (MOH).6.7 Regarding the last, we defined MOH as the self-reported use ≥3 days per week for the previous 60 days of any single medication or class of medications intended for acute headache treatment. For all patients, we also recorded scores from the McLean Screening Instrument for Borderline Personality Disorder (MSI-BPD). The instrument consists of 10 questions to be answered by the examinee, and a positive response to 7 or more is considered indicative of BPD; the questions are listed in Table 1. The instrument previously has been validated in a group of subjects who previously or currently were receiving some form of psychotherapy.8

Table 1.—.  McLean Screening Instrument for Borderline Personality Disorder
  1. Form entitled “Difficulty in Living Questionnaire” for the purposes of this study.

For each question, please indicate whether “yes” or “no” is the best answer for you:
1. Have any of your closest relationships been troubled by a lot of arguments of repeated breakups?1 = yes 0 = no
2. Have you deliberately hurt yourself physically (eg, punched yourself, cut yourself, burned yourself)? How about made a suicide attempt?1 = yes 0 = no
3. Have you had at least 2 other problems with impulsivity (eg, eating binges and spending sprees, drinking too much and verbal outbursts)?1 = yes 0 = no
4. Have you been extremely moody?1 = yes 0 = no
5. Have you felt very angry a lot of the time? How about often acted in an angry or sarcastic manner?1 = yes 0 = no
6. Have you been distrustful of other people?1 = yes 0 = no
7. Have you frequently felt unreal or as if things around you were unreal?1 = yes 0 = no
8. Have you chronically felt empty?1 = yes 0 = no
9. Have you often felt that you had no idea of who you are or that you have no identity?1 = yes 0 = no
10. Have you made desperate efforts to avoid feeling abandoned or being abandoned (eg, repeatedly called someone to reassure yourself that he or she still cared, begged them not to leave you, clung to them physically)?1 = yes 0 = no

A neurologist blinded as to the diagnosis of BPD versus no BPD evaluated, treated, and followed all patients under study and managed the Group 1 and 3 patients according to a uniform pharmacologic treatment plan that featured antiepileptic drug therapy (divalproex sodium, topiramate, zonisamide, and gabapentin) and botulinum toxin type A for prophylaxis and a combination of triptans and NSAIDs for acute headache. For the entire 6 month follow-up period all patients in Groups 1 and 3 were asked to keep daily headache diaries wherein they recorded headache occurrence, the maximum intensity of any headache experienced that day (“mild” vs “moderate/nondisabling” vs “disabling”), any symptomatic medication used and any unscheduled office, clinic, or ED visits for acute headache treatment. The treating neurologist examined all patients at baseline and the Group 1 and 3 patients at 1 month, 3 months, and 6 months. Diaries were reviewed with the patients at each follow-up visit.

At the 6-month follow-up we reevaluated all patients in Groups 1 and 3. We reassessed variables related to headache frequency, severity, and disability and again administered the QIDS-SR. We also assessed clinical outcome; we defined a positive treatment response as a ≥50% reduction in HDs per month relative to the pretreatment baseline month, with that reduction enduring for a minimum of 2 consecutive months within any portion of the 6-month follow-up period; this served as the primary endpoint for the study, and the total number of positive responders in Groups 1 and 3 were compared via χ2 analysis. We also assessed the total number of unscheduled visits for acute headache treatment within the follow-up period.

Prior to its initiation, this study was approved by our university's investigational review board. Written informed consent was obtained from all subjects.


The baseline demographic and clinical characteristics of all groups are outlined in Table 2. In summary, patients in Group 1 (ie, migraine patients with coexisting BPD) were more likely than those in Group 2 (all migraine patients) to be female, to have more pervasive forms of migraine, to record QIDS-SR scores indicative of depression of at least “moderate” severity, to report more presumed MOH, and to report greater migraine-related disability.

Table 2.—.  Baseline Characteristics of Groups 1, 2, and 3
 n (%)Age*FemaleEMFEMTMNOPH w/MFQIDS-SR ≥ 11Headache profile*MIDAS*MOH
  1. *Means.

  2. 31(62%) patients with scores >100.

  3. ‡22 (44%) with scores >100.

  4. Group 1 = migraine/history of borderline personality disorder (BPD).

  5. Group 2 = all migraine/no history of BPD.

  6. Group 3 = migraine/no history of BPD but matched to Group 1 for age, gender, and headache frequency.

  7. EM = episodic migraine; FEM = frequent episodic migraine; TM = transformed migraine/daily headache; NDPH w/MF = new daily persistent headache with migrainous features; QIDS-SR = Quick Inventory of Depressive Symptomatology (Self-Report), 0–27 scale, score ≥11 = moderate, severe or very severe depression; MIDAS = Migraine Disability Assessment score, 0–270 scale; MOH = medication overuse headache.

Group 150 (100)32.950 (100)4 (8)18 (36)22 (44) 6 (12)42 (84) 26.2/18.1111 (29-270)37 (74)
Group 250 (100)38.144 (88) 16 (32)14 (28)19 (38)1 (2)14 (28)16.3/4.542.4 (8-110)   15 (30)
Group 350 (100)32.250 (100)4 (8)18 (36)25 (50)3 (6)21 (42)25.7/5.960.7 (25-270)31 (62)

At baseline, all demographic and most clinical features of Groups 1 and 3 were similar (Table 2; Group 3 = migraine patients without BPD matched to the migraine/BPD Group 1 patients for age, gender, and headache frequency). While the means for HDs per month were similar for the 2 groups (26.2 for Group 1 vs 25.7 for Group 3), the mean for functionally incapacitating HDs per month was 3-fold higher in the BPD group (18.1 vs 5.9). Similarly, the mean baseline MIDAS score was higher in the BPD patients (111 vs 60.7), as was the prevalence of self-reported “moderate,”“severe,” or “very severe” depression (42/50 = 84% vs 21/50 = 42%). Finally, the prevalence of presumed MOH was higher in the BPD group (37/50 =74% vs 31/50 = 62%). Only 7 (14%) of the BPD patients recorded scores on the MSI-BPD consistent with the diagnosis of BPD.

At the 6-month follow-up the Group 3 (ie, no BPD) patients were significantly more likely to have experienced a positive treatment response than those in Group 1 (29/50 = 58% vs 7/50 = 14%; P < .03); 2 Group 1 patients and 7 Group 2 patients were lost to follow-up and counted as treatment failures in this intent-to-treat analysis. MIDAS scores had decreased in both groups but declined by only 4 points (to 106.8) in the Group 1 patients versus 17 points (to 43.4) in Group 3 patients. The prevalence of “moderate,”“severe,” or “very severe” depression was unchanged in the BPD group but had declined to 19% (9/48) in the Group 3 patients. The prevalence of presumed MOH remained higher in the BPD group (33/43 = 77% vs 18/48 = 38%), and the difference between the 2 groups was greater than at baseline. The mean for total unscheduled visits for acute headache treatment over the 6-month follow-up period was higher in Group 1 (3.1 vs .6).


In our clinic-based study of patients with migraine, coexisting BPD was associated with female gender, more pervasive headache, greater migraine-related disability, a higher prevalence of MOH, more unscheduled visits for acute headache treatment, a higher prevalence of moderate-to-severe depression, and a lower likelihood of responding to pharmacologic treatment intended for migraine prophylaxis. Many of these findings previously have been reported, albeit not necessarily in the same context. As examples, others have reported that a significant majority of BPD patients are female, and severe headache was noted to be highly prevalent in one study of BPD patients actively receiving psychiatric therapy.3,9 A number of investigators have reported a high prevalence of substance abuse in psychiatric patients with BPD, and in a study of BPD patients with migraine it consequently should not be surprising to find—as we did—a disproportionately high prevalence of overuse of symptomatic medications.10,11 Finally, there exist data indicating that psychiatric comorbidity of various types is higher in patients with TM than in those with EM, and our data suggest that BPD specifically may be associated with more pervasive and disabling headache.12

What advantages could result from early identification of BPD in a headache clinic population? For one, the treating clinician then may anticipate the eventual emergence of behavior disruptive to the therapeutic alliance and so implement proactively strategies intended to establish clear boundaries and minimize that behavior (eg, treatment contracts).13 For another, when BPD is present, the treating clinician may wish to refer the patient for psychiatric comanagement; psychotherapy combined with psychopharmacologic treatment may impact BPD positively.14 As our own data reflect, depression is comorbid with BPD, and prescription of an antidepressant may be appropriate for those patients with true depression of moderate-to-severe proportions.15 Finally, identification of BPD should guide the clinician to avoid the prescription of certain medications; given the high prevalence of benzodiazepine abuse in the BPD population, this class of agents is at least relatively contraindicated, and—aside from a similar concern regarding abuse potential—opioid use may aggravate the behavioral symptomatology associated with BPD.10,16

The MSI-BPD is a self-report screening instrument for detection of BPD that has exhibited high sensitivity and specificity when administered to individuals with a psychiatric treatment history recruited via a newspaper advertisement and posters placed around a hospital campus; its diagnostic efficiency was highest in patients 25 years old or younger.8 In our study the MSI-BPD was not effective in identifying BPD in patients previously diagnosed as having that disorder. Interestingly, even those of our BPD patients who “passed” the MSI-BPD undetected were highly likely to record scores on the QIDS-SR that correlated with moderate-to-severe depression. Although the relatively high age of our BPD population (32.9 years) may have reduced the efficiency of the MSI-BPD, it seems unlikely that this factor by itself accounted for the instrument's low sensitivity. Because the instrument currently regarded by many as the “gold standard” for BPD diagnosis—ie, the Structured Clinical Interview for DSM-IV Axis II Personality Disorders, or SCID-II—had not been administered uniformly to all patients with presumed BPD, that diagnosis may have been inaccurate in at least a portion of our Group 1 subjects. While it is also conceivable that the instrument's insensitivity may have reflected some qualitative difference in the psychiatric symptomatology experienced by our patients relative to those tested in the validation study by Zanarini et al (eg, our patients were “higher functioning”), it may be that the clinical setting within which the instrument was administered—a headache clinic as opposed to a specific psychiatric clinical trial restricted to subjects with BPD and a history of psychiatric treatment—was the factor most decisively influencing outcome.8 If true, this last may have implications for other screening instruments used to detect psychiatric disease, ie, does the efficiency of these instruments vary according to the clinical setting within which they are administered?

In summary, our data suggest the following:

  • 1In the headache clinic populations with migraine, BPD is associated with
    • (a) more pervasive headache
    • (b) greater migraine-related disability
    • (c) a high prevalence of self-reported depression
    • (d) a higher prevalence of presumed MOH
    • (e) more unscheduled visits for acute headache treatment
    • (f) a lower likelihood of responding to a chronic migraine treatment strategy, and
  • 2The MSI-BPD may not be diagnostically sensitive in the headache clinic setting.

Conflict of Interest:  None.