- Email: [email protected]
Validation of the Hospital Anxiety and Depression Scale in patients with epilepsy
Epilepsy & Behavior 58 (2016) 97–101
Contents lists available at ScienceDirect
Epilepsy & Behavior journal homepage: www.elsevier.com/locate/yebeh
Validation of the Hospital Anxiety and Depression Scale in patients with epilepsy Mariusz S. Wiglusz a,⁎, Jerzy Landowski a, Lidia Michalak b, Wiesław J. Cubała a a b
Department of Psychiatry, Medical University of Gdańsk, Poland Regional Epilepsy Outpatient Unit, Copernicus Hospital, Gdańsk, Poland
a r t i c l e
i n f o
Article history: Received 19 December 2015 Revised 25 February 2016 Accepted 4 March 2016 Available online xxxx Keywords: Major depressive disorder Epilepsy DSM-IV-TR SCID-I Hospital Anxiety and Depression Scale Sensitivity Specificity
a b s t r a c t Objective: Despite the fact that depressive disorders are the most common comorbidities among patients with epilepsy (PWEs), they often go unrecognized and untreated. The availability of validated screening instruments to detect depression in PWEs is limited. The aim of the present study was to validate the Hospital Anxiety and Depression Scale (HADS) in adult PWEs. Methods:: A consecutive group of 118 outpatient PWEs was invited to participate in the study. Ninety-six patients met inclusion criteria, completed HADS, and were examined by a trained psychiatrist using Structured Clinical Interview (SCID-I) for DSM-IV-TR. Receiver operating characteristic (ROC) curves were used to determine the optimal threshold scores for the HADS depression subscale (HADS-D). Results: Receiver operating characteristic analyses showed areas under the curve at approximately 84%. For diagnoses of MDD, the HADS-D demonstrated the best psychometric properties for a cutoff score ≥ 7 with sensitivity of 90.5%, specificity of 70.7%, positive predictive value of 46.3%, and negative predictive value of 96.4%. In the case of the group with ‘any depressive disorder’, the HADS-D optimum cutoff score was ≥6 with sensitivity of 82.5%, specificity of 73.2%, positive predictive value of 68.8%, and negative predictive value of 85.4%. Conclusions: The HADS-D proved to be a valid and reliable psychometric instrument in terms of screening for depressive disorders in PWEs. In the epilepsy setting, HADS-D maintains adequate sensitivity, acceptable specificity, and high NPV but low PPV for diagnosing MDD with an optimum cutoff score ≥7. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Depressive disorders are the most common psychiatric comorbidities in patients with epilepsy (PWEs). The detection of depression is of particular clinical importance in PWEs as, despite extensive data on its occurrence, it is often still underdiagnosed and untreated. A key reason for this is the lack of well-validated, self-report, screening psychometric instruments in PWEs which could be easily implemented in a clinical setting. Reliable screening instruments for depression are crucial in PWEs especially since antiepileptic drug (AED) side effects as well as periictal symptomatology might affect the accuracy of psychiatric diagnosis in epilepsy [1]. When choosing a psychometric instrument for screening purposes, it is important to optimize cutoff points for the population
⁎ Corresponding author at: Department of Psychiatry, Medical University of Gdańsk, Dębinki 7 St., Build. 25, 80-952 Gdańsk, Poland. Tel.: +48 58 349 26 50; fax: +48 58 349 27 48. E-mail address: [email protected] (M.S. Wiglusz).
http://dx.doi.org/10.1016/j.yebeh.2016.03.003 1525-5050/© 2016 Elsevier Inc. All rights reserved.
with epilepsy [2–4]. At the moment, there is only a limited number of validation studies concerning screening instruments for depression in epilepsy. In some studies, self-report scales, already used in the general population and in other medical illnesses, were validated for use in PWEs, namely the Beck Depression Inventory (BDI) [5,6] and the Hospital Anxiety and Depression Scale (HADS) [2,5–10]. Recently, a new sixitem screening instrument, the Neurological Disorders Depression Inventory for Epilepsy (NDDI-E), was developed specifically for use in PWEs [10–14]. This instrument was designed to minimize the potential for confounding factors related to AEDs or epilepsy itself. Similarly, HADS is the scale that has no items relating to somatic symptoms that may confound the diagnosis in PWEs and therefore reduce the sensitivity in screening for depression. It was developed in the early 1980s as a tool to identify anxiety and depressive disorders in nonpsychiatric patients within a hospital setting [15,16] and was broadly used in the general population and in many populations with different somatic illnesses. There were only a few validation studies in PWEs, with some confounding results [2,7,8]. The aim of this study was to validate the psychometric properties of the HADS depression subscale in PWEs in order to find optimal specificity, sensitivity, and cutoff scores for identifying depressive disorders.
98
M.S. Wiglusz et al. / Epilepsy & Behavior 58 (2016) 97–101
2. Methods
Table 1 Demographic and clinical characteristics of study total population.
2.1. Study sample
N = 96 (%) Male sex (%)a Age, in years (SD)b Age of seizure onset (SD)
31 (32.3) 36.6 (12.0) 19.5 (11.6) 17.0 (11.8) 3 (2.5)
The study population selection and psychometric evaluation has been described in detail elsewhere [17]. Briefly, over a 1-year period, a consecutive series of 118 PWEs from a regional epilepsy outpatient unit was screened for the study, with 96 patients meeting inclusion/ exclusion criteria and enrolled. All individuals underwent a complete neurological examination at study entry. Inclusion criteria were as follows: (1) confirmed diagnosis of active epilepsy according to the International League Against Epilepsy criteria (ILAE) [18] by a trained epileptologist, (2) aged 18–65 years, (3) stable antiepileptic treatment in the last 2 months, and (4) willing to provide a written informed consent to undergo the experimental procedures. Exclusion criteria included (1) neurologic somatic-related factors: last seizure within 24 h prior to examination, more than 10 seizures in the last month, major brain damage with mass effect, neurosurgical treatment of epilepsy, unstable somatic disease or serous neurological disorder, psychogenic nonepileptic seizures and (2) psychiatry-related factors: mental retardation, dependence on or abuse of alcohol and/or other drugs in the past 6 months, and diagnosis of borderline, antisocial, or schizotypical personality disorder. The study protocol was approved by the local bioethics committee at the Medical University of Gdańsk. All participants provided written informed consent for participation in the study.
continuous data, Mann–Whitney's U-test for nonnormally distributed data, and Fisher's exact test for categorical data. A value of p b 0.05 was considered to be statistically significant. Statistical procedures were performed using Statistica 10.0.1011.
2.2. Instruments
3. Results
All subjects were assessed using Structured Clinical Interview (SCID-I) [19] and HADS at the same visit by the same psychiatrist (MSW). Structured Clinical Interview is a semistructured interview used for the identification of DSM-IV-TR psychiatric disorders [19]. The Hospital Anxiety and Depression Scale (HADS) was developed by Zigmond and Snaith in 1983 [15,16] to identify caseness (possible and probable) of anxiety disorders and depression among patients in nonpsychiatric hospital clinics. The tool includes 14 items, seven related to anxiety (HADS-A) and seven related to depression (HADS-D), each scored between 0 and 3. The scale authors recommended that a score N8 on an individual scale should be regarded as a possible case. This threshold was found to be optimal for HADS-A and HADS-D in the general population as well as in samples of patients with somatic symptoms. For analyses, patients were assigned either to a diagnostic group, ‘major depressive disorder’, or to a comprehensive group, ‘any depressive disorder’. The group with ‘any depressive disorder’ was comprised of MDD and mood disorders with depressive features that do not meet the criteria for major depressive disorder (depressive disorder not otherwise specified [DD-NOS]: minor depression, recurrent brief depressive disorder, dysthymic disorder, mood disorder due to a general medical condition, substance-induced mood disorder).
Clinical and demographic characteristics are shown in Table 1. According to the SCID-I, the diagnosis of major depressive disorder (current episode) was established in 21 (22%) patients; ‘any depressive disorder’ was found in 40 (41.6%) patients. Mean HADS-D total scores for study groups are shown in Table 2. Receiver operator characteristic values for the HADS-D are shown in Table 3. For diagnoses of MDD, the HADS-D demonstrated the best psychometric properties for a cutoff score of 6 with sensitivity of 90.5%, specificity of 70.7%, AUC of 84.9% (Fig. 1), positive predictive value of 46.3%, and negative predictive value of 96.4% (Table 4). In the case of the group with ‘any depressive disorder’, the HADS-D showed the best cutoff score of 5 with sensitivity of 82.5%, specificity of 73.2%, AUC of 83% (Fig. 1), positive predictive value of 68.8%, and negative predictive value of 85.4% (Table 4).
2.3. Statistics In order to determine the diagnostic sensitivity and specificity of the HADS for the DSM-IV depressive disorder diagnoses and determine an optimal cutoff point, a receiver operator characteristic (ROC) curve was obtained for HADS-D. Area under the curve (AUC) values were interpreted according to the following guidelines: 0.9–1, excellent; 0.8–0.9, good; 0.7–0.8, fair; and 0.6–0.7, poor. Cutoff values were established with the (0, 1) minimum distance method giving equal weight to sensitivity and specificity. There were no missing data or outliers. Frequencies and descriptive statistics were analyzed for each variable. Comparisons between patients with current MDD and patients without MDD were made using Student's t-tests for normally distributed
Duration of epilepsy (SD) Number of seizures/last month — median (IQR) Seizure type (%) Generalized Simple partial Complex partial Partial evolving to general Tonic–clonic Absence Myoclonic Atonic Number of AEDs (IQR) Drug-resistant (%) Polytherapy (%) a b
15 (15.6) 7 (7.3) 27 (28.1) 47 (49.0) 10 (10.4) 2 (1.0) 1 (1.0) 2 (2.1) 2 (1.2) 70 (72.9) 46 (47.9)
Student's t-test. Fisher's exact test.
4. Discussion The clinical profile of the study group is similar to other studies performed in specialized centers for epilepsy treatment (Table 5). In order to produce valid diagnoses, we used the complete version of SCID-I as a gold standard in psychiatric research. As previously observed, we found a high frequency of major depression (22%) and other forms of depressive disorders in PWEs [17]. A good screening method for the diagnosis of major depression must be practical and reliable, exhibiting an adequate balance between sensitivity and specificity. In the study group, the HADS-D showed significant ability as a screening tool for indicating depressive disorder categories in PWEs using ROC as compared with SCID-I. For major depression Table 2 Psychometric characteristic of analyzed group. Rating scale
Diagnostic category
(+)
(−)
Median (IQR)
Mann–Whitney p Z
HADS-D MDD 9 (7; 11) 4 (1; 6) 4.889 Any depressive 8 (5; 10) 2 (1; 5) 5.541 disorder
Difference (95% CI)
b0.0001 5 (4; 7) b0.0001 5 (4; 6)
M.S. Wiglusz et al. / Epilepsy & Behavior 58 (2016) 97–101
99
Table 3 ROC analysis of the HADS-D according to categorical diagnosis.
MDD (+) vs. MDD (−) Any depressive disorder (+) vs. any depressive disorder (−)
HADS-D HADS-D
diagnosis, a cutoff score ≥7 resulted in an optimal value with high sensitivity (90.5%) and medium specificity (70.7%). Interestingly, Fiest et al. [2] obtained similar results not only in terms of cutoff score but also for other measures of diagnostic accuracy. The Hospital Anxiety and Depression Scale, at a cutoff point of 7, classified the optimum balance between sensitivity (80.8%) and specificity (85.7%), and for optimizing sensitivity, a cutoff point of 6 was best, with sensitivity of 84.6% and specificity of 77.9% [2]. As suggested by others, for screening purposes, sensitivity should be prioritized over specificity, and it is recommended that sensitivity should be at least 75%, whereas specificity should also ideally be at least 75% [8,20]. In both studies, HADS showed a higher predictive value for the absence of depression with sensitivity much higher than recommended and NPV of approximately 96%. Similarities between both studies may result from the analogous methodology (SCID-I for DSM-IV-TR). We strongly agree with Fiest et al. [2] that, for validation studies, scales must be compared head to head against a gold standard. However, in the study by Fiest et al. [2], the mood disorder section of SCID-I was administered by telephone (Table 5). Although some previous research suggests that there is good agreement between in-person and telephone administration of SCID-I [2], it is important to keep in mind that SCID-I should be administered in a face-to-face format, and any deviation from administration procedure may produce a systematic error. The results from both studies suggest cutoff points lower than suggested before [5,7] for depression in PWEs. Other studies in PWEs reported on different optimal cutoff points. In a recent study by Zingano et al. [7], a cutoff ≥ 8 for the HADS depression subscale showed the best balance between sensitivity (72.7%) and specificity (79.7%) (PPV: 56.3, NPV: 88.7). In this study, Axis I disorder diagnoses were classified according to a semistructured interview for the diagnostic algorithm of the DSM-IV-TR to determine the current presence of MDD. In another study with depression diagnosis based on the MINI Plus, HADS-D performed optimally at a cutoff ≥8 with specificity of 80.2% and sensitivity of 85.7% (PPV: 62.5%, NPV: 93.5%) [5]. On the other hand, a recent study by Gandy et al. [8] found that the HADS did not perform optimally at a
AUC
95% CI
SE
p
0.849 0.831
0.750–0.948 0.744–0.918
0.051 0.044
b0.001 b0.001
cutoff ≥8 in a population of PWEs (42% sensitivity from MINI diagnosis of major depression), although the ROC analysis was performed only for this one cutoff point [8]. Discrepancies between these studies could result from different reference standards employed (SCID-I vs. MINI) [2,17, Table 5]. In the studies on validation for depression screening tools in PWEs, MINI is more often employed than the gold-standard SCID-I mainly because it is less time- and resource-consuming [6]. However, the MINI, apart from MDD and dysthymic disorder, does not cover all subthreshold forms of depression (depressive disorder not otherwise specified [DD-NOS] category) that are common in epilepsy. In fact, the same problem concerns epidemiological studies on depressive disorders in PWEs, often producing inconsistent results due to different methodologies applied [17]. A study validating the MINI against the SCID-I in PWEs reported only fair-to-good agreement for all mood disorder modules [2,6]. Moreover, as suggested by many authors, atypical mood disorders specific to epilepsy, namely interictal dysphoric disorder (IDD) [21,22], may not be precisely identified with DSM-IV criteria [1,4] or may also overlap with DD-NOS criteria [17]. Therefore, for screening depression in PWEs, apart from MDD, the identification of such subthreshold forms of depression is of particular importance. In the HADS validation studies in PWEs published so far, the ROC analysis mostly referred to major depression only (Table 5). It should be stressed that almost all studies used only the mood section of diagnostic structured interviews (Table 5). In psychiatric studies, it is very important to perform a whole psychiatric examination in order to exclude other psychiatric disorders (e.g., psychotic disorder), which may significantly influence the HADS results. As psychiatric comorbidity of more than one disorder is very common, it is mandatory to the entire SCID-I or MINI before including patients into the study. Apart from MDD, we separately applied ROC analysis to a comprehensive group with ‘any depressive disorder’ (including DD-NOS) in which a cutoff of ≥6 resulted in the optimal balance between sensitivity (82.5%) and specificity (73.2%). In this subgroup, PPV (68.5%) was relatively higher as related to the group with MDD (46.3%), but because PPV
1
sensitivity
0.75
0.5
0.25
0 0
0.25
0.5
0.75
1
1 - specificity Fig. 1. ROC for HADS — Depression: patients with MDD vs. patients without MDD (red color) and patients with ‘any depressive disorder’ vs. patients with no depressive disorder (black color). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
0.792 72.7 ≥8
79.7
0.869 0.849 0.831 80.2 90.5 82.5 ≥8 ≥7 ≥6
85.7 70.7 73.2
42 ≥8
97
0.9
1 0.87 77.9 85.7 0.9 0.99 84.6 80.8 =7 =8 =6 =7
CIDI major depressive disorder and any anxiety disorders categories DSM-IV-TR The telephone-conducted mood Optimal sensitivity section of SCID-I Greatest balance of sensitivity and specificity DSM-IV-TR MINI 5 — depression and suicide Statistics only for cutoff point 8 section DSM-IV-TR MINI Plus DSM-IV-TR Complete SCID-I MDD Any depressive disorder — MDD and DD-NOS DSM-IV-TR Semistructured interview for the diagnostic algorithm of the DSM-IV-TR 25.2
24 7 27.8 22.3 41.6
14 100.0 36.4 MTLE-HS 103 57.3 Zingano et al. [7]
Oliveira et al. [5] PWEs Wiglusz et al. [17] PWEs
126 54.7 96 67
39.6 36.6
88.6 70.0
N/A 0
MDE Dysthymia MDD MDD Any depressive disorder MDD N/A 47.6 40 147 59 PWEs
185 51.4
40.3
N/A
12
MDD
14.6
ICD-10 27 MDD N/A
Gandy et al. [8]
The study results indicate that HADS-D is a valid and reliable psychometric instrument and will perform consistently in terms of screening for depressive disorders in PWEs. In the epilepsy setting, HADS-D maintains adequate sensitivity, acceptable specificity, and high NPV but low PPV for diagnosing MDD, with an optimum cutoff score ≥7.
PWEs
6. Conclusions
Fiest et al. [2]
The methodology used may contribute to the conclusions drawn. The study may be underpowered because of its relatively small sample size. The results refer to outpatients treated in the tertiary referred unit who were at risk of a complicated course of epilepsy and a high percentage of patients with drug-resistant seizures. In order to minimize the influence of periictal and ictal psychiatric symptoms on interictal depressive disorders, subjects experiencing more than 10 seizures in the last month before participation were excluded. Thus, the results may underscore the depressive symptomatology and ‘atypical’ presentations of depression.
N/A
5. Study limitations
28.4
depends on the prevalence, which for ‘any depressive disorder’ (41.6%) was almost twice as high as for MDD (22%), it does not change overall PPV, which was low in the study group. Our findings also corroborate the suggestion of other authors [2] that the cutoff points used for epilepsy should be adapted (e.g., a different cutoff point for MDD) in order to reflect differences in depressive symptomatology for diagnosing depression in PWEs. We found an even lower cutoff score for the ‘any depressive disorder’ category as compared to MDD diagnosis. According to our results, the HADS depression subscale maintains, in the epilepsy setting, a very good sensitivity, an acceptable specificity, and an excellent negative predictive value but a low positive predictive value (Table 4). These findings are similar to those reported from other studies on HADS in PWEs, and it seems evident that HADS-D largely fails in terms of the positive predictive value. This could suggest that depressive disorders in PWEs present atypical features that are not efficiently captured by the HADS depression subscale. On the other hand, HADSD has a high efficiency in predicting the absence of depressive disorders in PWEs and presents adequate properties for screening depression in people with epilepsy. After scoring positive, patients should be referred to a specialist for further evaluation and for appropriate treatment. Of interest, a recent systematic review showed HADS's inability to consistently differentiate between the constructs of anxiety and depression [23], which could be due to different methodologies applied [2] but also due to lingual and cultural differences related to translations of HADS into different languages [24]. The original developers of the HADS intended to make the items easy to translate into other languages [16]. But some argue whether they succeeded [24] especially because of the use of colloquial British English for the construction of items, which are hard to adequately translate into other languages [25]. In an analysis of this issue, the authors concluded that, apart from methodology, it could be an important reason for not having consistent results from different studies [24]. Our results seem to suggest otherwise as, despite differences in language and culture (Canada and Poland), HADS performed consistently in both populations.
150 44.6
96.4 85.4
PWEs
46.3 68.8
Al-Asmi et al. [9]
70.7 73.2
Notes
90.5 82.5
Diagnostic instrument
NPV
Diagnostic system
PPV
N (%)
≥7 ≥6
Specificity
Female Mean Drug-resistant ATDs Diagnosis (%) age (%) (n)
MDD Any depressive disorder
Sensitivity
N
HADS-D
Cutoff score
Study group
Diagnostic category
Table 5 Estimated sensitivity and specificity of the HADS depression subscale at optimal cutoff values.
Rating scale
Optimal Sensitivity (%) Specificity (%) AUC (%) cutoff values
Table 4 ROC analysis of the HADS-D (presence of MDD or depression).
0.989
M.S. Wiglusz et al. / Epilepsy & Behavior 58 (2016) 97–101
Reference
100
M.S. Wiglusz et al. / Epilepsy & Behavior 58 (2016) 97–101
Disclosure of conflicts of interest We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. No conflicts of interest exist. AUC — area under curve. ATDs (n) — number of subjects receiving antidepressant treatment. CIDI — Composite International Diagnostic Interview. DD-NOS — depressive disorder not otherwise specified. DSM-IV-TR — Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision. ICD-10 — International Statistical Classification of Diseases and Related Health Problems 10th Revision. MDD — major depressive disorder. MTLE-HS — mesial temporal lobe epilepsy with hippocampal sclerosis. N/A — not available. PWEs —patients with epilepsy.
References [1] Kanner AM, Barry JJ. Is the psychopathology of epilepsy different from that of nonepileptic patients? Epilepsy Behav 2001;2:170–86. [2] Fiest KM, Patten SB, Wiebe S, Bulloch AG, Maxwell CJ, Jetté N. Validating screening tools for depression in epilepsy. Epilepsia 2014;55:1642–50. [3] Krishnamoorthy ES. The evaluation of behavioral disturbances in epilepsy. Epilepsia 2006;47:3–8. [4] Mula M, Iudice A, La Neve A, Mazza M, Mazza S, Cantello R, et al. Validation of the Hamilton Rating Scale for Depression in adults with epilepsy. Epilepsy Behav 2014;41:122–5. [5] de Oliveira GN, Lessa JM, Gonçalves AP, Portela EJ, Sander JW, Teixeira AL. Screening for depression in people with epilepsy: comparative study among Neurological Disorders Depression Inventory for Epilepsy (NDDI-E), Hospital Anxiety and Depression Scale Depression Subscale (HADS-D), and Beck Depression Inventory (BDI). Epilepsy Behav 2014;34:50–4. [6] Jones JE, Hermann BP, Woodard JL, Barry JJ, Gilliam F, Kanner AM, et al. Screening for major depression in epilepsy with common self-report depression inventories. Epilepsia 2005;46:731–5. [7] de Lemos Zingano B, Guarnieri R, Diaz AP, Schwarzbold ML, Bicalho MA, Claudino LS, et al. Validation of diagnostic tests for depressive disorder in drug-resistant mesial temporal lobe epilepsy. Epilepsy Behav 2015;50:61–6. [8] Gandy M, Sharpe L, Perry KN, Miller L, Thayer Z, Boserio J, et al. Assessing the efficacy of 2 screening measures for depression in people with epilepsy. Neurology 2012;79: 371–5.
101
[9] Al-Asmi A, Dorvlo AS, Burke DT, Al-Adawi S, Al-Zaabi A, Al-Zadjali HA, et al. The detection of mood and anxiety in people with epilepsy using two-phase designs: experiences from a tertiary care centre in Oman. Epilepsy Res 2012;98:174–81. [10] Rampling J, Mitchell AJ, Von Oertzen T, Docker J, Jackson J, Cock H, et al. Screening for depression in epilepsy clinics. A comparison of conventional and visual-analog methods. Epilepsia 2012;53:1713–21. [11] Gilliam FG, Barry JJ, Hermann BP, Meador KJ, Vahle V, Kanner AM. Rapid detection of major depression in epilepsy: a multicentre study. Lancet Neurol 2006;5:399–405. [12] Mula M, Iudice A, La Neve A, Mazza M, Bartolini E, De Caro MF. Validation of the Italian version of the Neurological Disorders Depression Inventory for Epilepsy (NDDI-E). Epilepsy Behav 2012;24:329–31. [13] Zis P, Yfanti P, Siatouni A, Tavernarakis A, Gatzonis S. Validation of the Greek version of the Neurological Disorders Depression Inventory for Epilepsy (NDDI-E). Epilepsy Behav 2013;29:513–5. [14] Thomson AE, Calle A, Fontela ME, Yepez L, Muñoz Giacomelli F, Jáuregui A, et al. Screening of major depression in epilepsy: the Neurologic Depression Disorders Inventory in Epilepsy — Spanish version (Argentina). Epilepsia 2014;55:331–4. [15] Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatr Scand 1983;67:361–70. [16] Snaith RP. The Hospital Anxiety and Depression Scale. Health Qual Life Outcomes 2003;1:1–4. [17] Wiglusz MS, Landowski J, Michalak L, Cubała WJ. Reevaluating the prevalence and diagnostic subtypes of depressive disorders in epilepsy. Epilepsy Behav 2015;53: 15–9. [18] Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia 1989;30:389–99. [19] First MB, Spitzer RL, Gibbon M, Williams JBW. Structured Clinical Interview for DSMIV-TR Axis I Disorders, Research Version, Patient Edition. (SCID-I/P) New York: Biometrics Research, New York State Psychiatric Institute; 2000. [20] Löwe B, Spitzer RL, Gräfe K, Kroenke K, Quenter A, Zipfel S. Comparative validity of three screening questionnaires for DSM-IV depressive disorders and physicians' diagnoses. J Affect Disord 2004;78:131–40. [21] Blumer D, Montouris G, Davies K. The interictal dysphoric disorder: recognition, pathogenesis, and treatment of the major psychiatric disorder of epilepsy. Epilepsy Behav 2004;5:826–40. [22] Amiri M, Hansen CP. The interictal dysphoric disorder in patients with epilepsy: a doubtful disorder lacking diagnostic tools. Seizure 2015;24:70–6. [23] Cosco TD, Doyle F, Ward M, McGee H. Latent structure of the Hospital Anxiety and Depression Scale: a 10-year systematic review. J Psychosom Res 2012;72:180–4. [24] Maters GA, Sanderman R, Kim AY, Coyne JC. Problems in cross-cultural use of the Hospital Anxiety and Depression Scale: “no butterflies in the desert”. PLoS One 2013;9:e70975. [25] Brislin RW, Lonner WJ, Thorndike RM. Cross-cultural research methods. New York: J. Wiley; 1973.
Contents lists available at ScienceDirect
Epilepsy & Behavior journal homepage: www.elsevier.com/locate/yebeh
Validation of the Hospital Anxiety and Depression Scale in patients with epilepsy Mariusz S. Wiglusz a,⁎, Jerzy Landowski a, Lidia Michalak b, Wiesław J. Cubała a a b
Department of Psychiatry, Medical University of Gdańsk, Poland Regional Epilepsy Outpatient Unit, Copernicus Hospital, Gdańsk, Poland
a r t i c l e
i n f o
Article history: Received 19 December 2015 Revised 25 February 2016 Accepted 4 March 2016 Available online xxxx Keywords: Major depressive disorder Epilepsy DSM-IV-TR SCID-I Hospital Anxiety and Depression Scale Sensitivity Specificity
a b s t r a c t Objective: Despite the fact that depressive disorders are the most common comorbidities among patients with epilepsy (PWEs), they often go unrecognized and untreated. The availability of validated screening instruments to detect depression in PWEs is limited. The aim of the present study was to validate the Hospital Anxiety and Depression Scale (HADS) in adult PWEs. Methods:: A consecutive group of 118 outpatient PWEs was invited to participate in the study. Ninety-six patients met inclusion criteria, completed HADS, and were examined by a trained psychiatrist using Structured Clinical Interview (SCID-I) for DSM-IV-TR. Receiver operating characteristic (ROC) curves were used to determine the optimal threshold scores for the HADS depression subscale (HADS-D). Results: Receiver operating characteristic analyses showed areas under the curve at approximately 84%. For diagnoses of MDD, the HADS-D demonstrated the best psychometric properties for a cutoff score ≥ 7 with sensitivity of 90.5%, specificity of 70.7%, positive predictive value of 46.3%, and negative predictive value of 96.4%. In the case of the group with ‘any depressive disorder’, the HADS-D optimum cutoff score was ≥6 with sensitivity of 82.5%, specificity of 73.2%, positive predictive value of 68.8%, and negative predictive value of 85.4%. Conclusions: The HADS-D proved to be a valid and reliable psychometric instrument in terms of screening for depressive disorders in PWEs. In the epilepsy setting, HADS-D maintains adequate sensitivity, acceptable specificity, and high NPV but low PPV for diagnosing MDD with an optimum cutoff score ≥7. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Depressive disorders are the most common psychiatric comorbidities in patients with epilepsy (PWEs). The detection of depression is of particular clinical importance in PWEs as, despite extensive data on its occurrence, it is often still underdiagnosed and untreated. A key reason for this is the lack of well-validated, self-report, screening psychometric instruments in PWEs which could be easily implemented in a clinical setting. Reliable screening instruments for depression are crucial in PWEs especially since antiepileptic drug (AED) side effects as well as periictal symptomatology might affect the accuracy of psychiatric diagnosis in epilepsy [1]. When choosing a psychometric instrument for screening purposes, it is important to optimize cutoff points for the population
⁎ Corresponding author at: Department of Psychiatry, Medical University of Gdańsk, Dębinki 7 St., Build. 25, 80-952 Gdańsk, Poland. Tel.: +48 58 349 26 50; fax: +48 58 349 27 48. E-mail address: [email protected] (M.S. Wiglusz).
http://dx.doi.org/10.1016/j.yebeh.2016.03.003 1525-5050/© 2016 Elsevier Inc. All rights reserved.
with epilepsy [2–4]. At the moment, there is only a limited number of validation studies concerning screening instruments for depression in epilepsy. In some studies, self-report scales, already used in the general population and in other medical illnesses, were validated for use in PWEs, namely the Beck Depression Inventory (BDI) [5,6] and the Hospital Anxiety and Depression Scale (HADS) [2,5–10]. Recently, a new sixitem screening instrument, the Neurological Disorders Depression Inventory for Epilepsy (NDDI-E), was developed specifically for use in PWEs [10–14]. This instrument was designed to minimize the potential for confounding factors related to AEDs or epilepsy itself. Similarly, HADS is the scale that has no items relating to somatic symptoms that may confound the diagnosis in PWEs and therefore reduce the sensitivity in screening for depression. It was developed in the early 1980s as a tool to identify anxiety and depressive disorders in nonpsychiatric patients within a hospital setting [15,16] and was broadly used in the general population and in many populations with different somatic illnesses. There were only a few validation studies in PWEs, with some confounding results [2,7,8]. The aim of this study was to validate the psychometric properties of the HADS depression subscale in PWEs in order to find optimal specificity, sensitivity, and cutoff scores for identifying depressive disorders.
98
M.S. Wiglusz et al. / Epilepsy & Behavior 58 (2016) 97–101
2. Methods
Table 1 Demographic and clinical characteristics of study total population.
2.1. Study sample
N = 96 (%) Male sex (%)a Age, in years (SD)b Age of seizure onset (SD)
31 (32.3) 36.6 (12.0) 19.5 (11.6) 17.0 (11.8) 3 (2.5)
The study population selection and psychometric evaluation has been described in detail elsewhere [17]. Briefly, over a 1-year period, a consecutive series of 118 PWEs from a regional epilepsy outpatient unit was screened for the study, with 96 patients meeting inclusion/ exclusion criteria and enrolled. All individuals underwent a complete neurological examination at study entry. Inclusion criteria were as follows: (1) confirmed diagnosis of active epilepsy according to the International League Against Epilepsy criteria (ILAE) [18] by a trained epileptologist, (2) aged 18–65 years, (3) stable antiepileptic treatment in the last 2 months, and (4) willing to provide a written informed consent to undergo the experimental procedures. Exclusion criteria included (1) neurologic somatic-related factors: last seizure within 24 h prior to examination, more than 10 seizures in the last month, major brain damage with mass effect, neurosurgical treatment of epilepsy, unstable somatic disease or serous neurological disorder, psychogenic nonepileptic seizures and (2) psychiatry-related factors: mental retardation, dependence on or abuse of alcohol and/or other drugs in the past 6 months, and diagnosis of borderline, antisocial, or schizotypical personality disorder. The study protocol was approved by the local bioethics committee at the Medical University of Gdańsk. All participants provided written informed consent for participation in the study.
continuous data, Mann–Whitney's U-test for nonnormally distributed data, and Fisher's exact test for categorical data. A value of p b 0.05 was considered to be statistically significant. Statistical procedures were performed using Statistica 10.0.1011.
2.2. Instruments
3. Results
All subjects were assessed using Structured Clinical Interview (SCID-I) [19] and HADS at the same visit by the same psychiatrist (MSW). Structured Clinical Interview is a semistructured interview used for the identification of DSM-IV-TR psychiatric disorders [19]. The Hospital Anxiety and Depression Scale (HADS) was developed by Zigmond and Snaith in 1983 [15,16] to identify caseness (possible and probable) of anxiety disorders and depression among patients in nonpsychiatric hospital clinics. The tool includes 14 items, seven related to anxiety (HADS-A) and seven related to depression (HADS-D), each scored between 0 and 3. The scale authors recommended that a score N8 on an individual scale should be regarded as a possible case. This threshold was found to be optimal for HADS-A and HADS-D in the general population as well as in samples of patients with somatic symptoms. For analyses, patients were assigned either to a diagnostic group, ‘major depressive disorder’, or to a comprehensive group, ‘any depressive disorder’. The group with ‘any depressive disorder’ was comprised of MDD and mood disorders with depressive features that do not meet the criteria for major depressive disorder (depressive disorder not otherwise specified [DD-NOS]: minor depression, recurrent brief depressive disorder, dysthymic disorder, mood disorder due to a general medical condition, substance-induced mood disorder).
Clinical and demographic characteristics are shown in Table 1. According to the SCID-I, the diagnosis of major depressive disorder (current episode) was established in 21 (22%) patients; ‘any depressive disorder’ was found in 40 (41.6%) patients. Mean HADS-D total scores for study groups are shown in Table 2. Receiver operator characteristic values for the HADS-D are shown in Table 3. For diagnoses of MDD, the HADS-D demonstrated the best psychometric properties for a cutoff score of 6 with sensitivity of 90.5%, specificity of 70.7%, AUC of 84.9% (Fig. 1), positive predictive value of 46.3%, and negative predictive value of 96.4% (Table 4). In the case of the group with ‘any depressive disorder’, the HADS-D showed the best cutoff score of 5 with sensitivity of 82.5%, specificity of 73.2%, AUC of 83% (Fig. 1), positive predictive value of 68.8%, and negative predictive value of 85.4% (Table 4).
2.3. Statistics In order to determine the diagnostic sensitivity and specificity of the HADS for the DSM-IV depressive disorder diagnoses and determine an optimal cutoff point, a receiver operator characteristic (ROC) curve was obtained for HADS-D. Area under the curve (AUC) values were interpreted according to the following guidelines: 0.9–1, excellent; 0.8–0.9, good; 0.7–0.8, fair; and 0.6–0.7, poor. Cutoff values were established with the (0, 1) minimum distance method giving equal weight to sensitivity and specificity. There were no missing data or outliers. Frequencies and descriptive statistics were analyzed for each variable. Comparisons between patients with current MDD and patients without MDD were made using Student's t-tests for normally distributed
Duration of epilepsy (SD) Number of seizures/last month — median (IQR) Seizure type (%) Generalized Simple partial Complex partial Partial evolving to general Tonic–clonic Absence Myoclonic Atonic Number of AEDs (IQR) Drug-resistant (%) Polytherapy (%) a b
15 (15.6) 7 (7.3) 27 (28.1) 47 (49.0) 10 (10.4) 2 (1.0) 1 (1.0) 2 (2.1) 2 (1.2) 70 (72.9) 46 (47.9)
Student's t-test. Fisher's exact test.
4. Discussion The clinical profile of the study group is similar to other studies performed in specialized centers for epilepsy treatment (Table 5). In order to produce valid diagnoses, we used the complete version of SCID-I as a gold standard in psychiatric research. As previously observed, we found a high frequency of major depression (22%) and other forms of depressive disorders in PWEs [17]. A good screening method for the diagnosis of major depression must be practical and reliable, exhibiting an adequate balance between sensitivity and specificity. In the study group, the HADS-D showed significant ability as a screening tool for indicating depressive disorder categories in PWEs using ROC as compared with SCID-I. For major depression Table 2 Psychometric characteristic of analyzed group. Rating scale
Diagnostic category
(+)
(−)
Median (IQR)
Mann–Whitney p Z
HADS-D MDD 9 (7; 11) 4 (1; 6) 4.889 Any depressive 8 (5; 10) 2 (1; 5) 5.541 disorder
Difference (95% CI)
b0.0001 5 (4; 7) b0.0001 5 (4; 6)
M.S. Wiglusz et al. / Epilepsy & Behavior 58 (2016) 97–101
99
Table 3 ROC analysis of the HADS-D according to categorical diagnosis.
MDD (+) vs. MDD (−) Any depressive disorder (+) vs. any depressive disorder (−)
HADS-D HADS-D
diagnosis, a cutoff score ≥7 resulted in an optimal value with high sensitivity (90.5%) and medium specificity (70.7%). Interestingly, Fiest et al. [2] obtained similar results not only in terms of cutoff score but also for other measures of diagnostic accuracy. The Hospital Anxiety and Depression Scale, at a cutoff point of 7, classified the optimum balance between sensitivity (80.8%) and specificity (85.7%), and for optimizing sensitivity, a cutoff point of 6 was best, with sensitivity of 84.6% and specificity of 77.9% [2]. As suggested by others, for screening purposes, sensitivity should be prioritized over specificity, and it is recommended that sensitivity should be at least 75%, whereas specificity should also ideally be at least 75% [8,20]. In both studies, HADS showed a higher predictive value for the absence of depression with sensitivity much higher than recommended and NPV of approximately 96%. Similarities between both studies may result from the analogous methodology (SCID-I for DSM-IV-TR). We strongly agree with Fiest et al. [2] that, for validation studies, scales must be compared head to head against a gold standard. However, in the study by Fiest et al. [2], the mood disorder section of SCID-I was administered by telephone (Table 5). Although some previous research suggests that there is good agreement between in-person and telephone administration of SCID-I [2], it is important to keep in mind that SCID-I should be administered in a face-to-face format, and any deviation from administration procedure may produce a systematic error. The results from both studies suggest cutoff points lower than suggested before [5,7] for depression in PWEs. Other studies in PWEs reported on different optimal cutoff points. In a recent study by Zingano et al. [7], a cutoff ≥ 8 for the HADS depression subscale showed the best balance between sensitivity (72.7%) and specificity (79.7%) (PPV: 56.3, NPV: 88.7). In this study, Axis I disorder diagnoses were classified according to a semistructured interview for the diagnostic algorithm of the DSM-IV-TR to determine the current presence of MDD. In another study with depression diagnosis based on the MINI Plus, HADS-D performed optimally at a cutoff ≥8 with specificity of 80.2% and sensitivity of 85.7% (PPV: 62.5%, NPV: 93.5%) [5]. On the other hand, a recent study by Gandy et al. [8] found that the HADS did not perform optimally at a
AUC
95% CI
SE
p
0.849 0.831
0.750–0.948 0.744–0.918
0.051 0.044
b0.001 b0.001
cutoff ≥8 in a population of PWEs (42% sensitivity from MINI diagnosis of major depression), although the ROC analysis was performed only for this one cutoff point [8]. Discrepancies between these studies could result from different reference standards employed (SCID-I vs. MINI) [2,17, Table 5]. In the studies on validation for depression screening tools in PWEs, MINI is more often employed than the gold-standard SCID-I mainly because it is less time- and resource-consuming [6]. However, the MINI, apart from MDD and dysthymic disorder, does not cover all subthreshold forms of depression (depressive disorder not otherwise specified [DD-NOS] category) that are common in epilepsy. In fact, the same problem concerns epidemiological studies on depressive disorders in PWEs, often producing inconsistent results due to different methodologies applied [17]. A study validating the MINI against the SCID-I in PWEs reported only fair-to-good agreement for all mood disorder modules [2,6]. Moreover, as suggested by many authors, atypical mood disorders specific to epilepsy, namely interictal dysphoric disorder (IDD) [21,22], may not be precisely identified with DSM-IV criteria [1,4] or may also overlap with DD-NOS criteria [17]. Therefore, for screening depression in PWEs, apart from MDD, the identification of such subthreshold forms of depression is of particular importance. In the HADS validation studies in PWEs published so far, the ROC analysis mostly referred to major depression only (Table 5). It should be stressed that almost all studies used only the mood section of diagnostic structured interviews (Table 5). In psychiatric studies, it is very important to perform a whole psychiatric examination in order to exclude other psychiatric disorders (e.g., psychotic disorder), which may significantly influence the HADS results. As psychiatric comorbidity of more than one disorder is very common, it is mandatory to the entire SCID-I or MINI before including patients into the study. Apart from MDD, we separately applied ROC analysis to a comprehensive group with ‘any depressive disorder’ (including DD-NOS) in which a cutoff of ≥6 resulted in the optimal balance between sensitivity (82.5%) and specificity (73.2%). In this subgroup, PPV (68.5%) was relatively higher as related to the group with MDD (46.3%), but because PPV
1
sensitivity
0.75
0.5
0.25
0 0
0.25
0.5
0.75
1
1 - specificity Fig. 1. ROC for HADS — Depression: patients with MDD vs. patients without MDD (red color) and patients with ‘any depressive disorder’ vs. patients with no depressive disorder (black color). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
0.792 72.7 ≥8
79.7
0.869 0.849 0.831 80.2 90.5 82.5 ≥8 ≥7 ≥6
85.7 70.7 73.2
42 ≥8
97
0.9
1 0.87 77.9 85.7 0.9 0.99 84.6 80.8 =7 =8 =6 =7
CIDI major depressive disorder and any anxiety disorders categories DSM-IV-TR The telephone-conducted mood Optimal sensitivity section of SCID-I Greatest balance of sensitivity and specificity DSM-IV-TR MINI 5 — depression and suicide Statistics only for cutoff point 8 section DSM-IV-TR MINI Plus DSM-IV-TR Complete SCID-I MDD Any depressive disorder — MDD and DD-NOS DSM-IV-TR Semistructured interview for the diagnostic algorithm of the DSM-IV-TR 25.2
24 7 27.8 22.3 41.6
14 100.0 36.4 MTLE-HS 103 57.3 Zingano et al. [7]
Oliveira et al. [5] PWEs Wiglusz et al. [17] PWEs
126 54.7 96 67
39.6 36.6
88.6 70.0
N/A 0
MDE Dysthymia MDD MDD Any depressive disorder MDD N/A 47.6 40 147 59 PWEs
185 51.4
40.3
N/A
12
MDD
14.6
ICD-10 27 MDD N/A
Gandy et al. [8]
The study results indicate that HADS-D is a valid and reliable psychometric instrument and will perform consistently in terms of screening for depressive disorders in PWEs. In the epilepsy setting, HADS-D maintains adequate sensitivity, acceptable specificity, and high NPV but low PPV for diagnosing MDD, with an optimum cutoff score ≥7.
PWEs
6. Conclusions
Fiest et al. [2]
The methodology used may contribute to the conclusions drawn. The study may be underpowered because of its relatively small sample size. The results refer to outpatients treated in the tertiary referred unit who were at risk of a complicated course of epilepsy and a high percentage of patients with drug-resistant seizures. In order to minimize the influence of periictal and ictal psychiatric symptoms on interictal depressive disorders, subjects experiencing more than 10 seizures in the last month before participation were excluded. Thus, the results may underscore the depressive symptomatology and ‘atypical’ presentations of depression.
N/A
5. Study limitations
28.4
depends on the prevalence, which for ‘any depressive disorder’ (41.6%) was almost twice as high as for MDD (22%), it does not change overall PPV, which was low in the study group. Our findings also corroborate the suggestion of other authors [2] that the cutoff points used for epilepsy should be adapted (e.g., a different cutoff point for MDD) in order to reflect differences in depressive symptomatology for diagnosing depression in PWEs. We found an even lower cutoff score for the ‘any depressive disorder’ category as compared to MDD diagnosis. According to our results, the HADS depression subscale maintains, in the epilepsy setting, a very good sensitivity, an acceptable specificity, and an excellent negative predictive value but a low positive predictive value (Table 4). These findings are similar to those reported from other studies on HADS in PWEs, and it seems evident that HADS-D largely fails in terms of the positive predictive value. This could suggest that depressive disorders in PWEs present atypical features that are not efficiently captured by the HADS depression subscale. On the other hand, HADSD has a high efficiency in predicting the absence of depressive disorders in PWEs and presents adequate properties for screening depression in people with epilepsy. After scoring positive, patients should be referred to a specialist for further evaluation and for appropriate treatment. Of interest, a recent systematic review showed HADS's inability to consistently differentiate between the constructs of anxiety and depression [23], which could be due to different methodologies applied [2] but also due to lingual and cultural differences related to translations of HADS into different languages [24]. The original developers of the HADS intended to make the items easy to translate into other languages [16]. But some argue whether they succeeded [24] especially because of the use of colloquial British English for the construction of items, which are hard to adequately translate into other languages [25]. In an analysis of this issue, the authors concluded that, apart from methodology, it could be an important reason for not having consistent results from different studies [24]. Our results seem to suggest otherwise as, despite differences in language and culture (Canada and Poland), HADS performed consistently in both populations.
150 44.6
96.4 85.4
PWEs
46.3 68.8
Al-Asmi et al. [9]
70.7 73.2
Notes
90.5 82.5
Diagnostic instrument
NPV
Diagnostic system
PPV
N (%)
≥7 ≥6
Specificity
Female Mean Drug-resistant ATDs Diagnosis (%) age (%) (n)
MDD Any depressive disorder
Sensitivity
N
HADS-D
Cutoff score
Study group
Diagnostic category
Table 5 Estimated sensitivity and specificity of the HADS depression subscale at optimal cutoff values.
Rating scale
Optimal Sensitivity (%) Specificity (%) AUC (%) cutoff values
Table 4 ROC analysis of the HADS-D (presence of MDD or depression).
0.989
M.S. Wiglusz et al. / Epilepsy & Behavior 58 (2016) 97–101
Reference
100
M.S. Wiglusz et al. / Epilepsy & Behavior 58 (2016) 97–101
Disclosure of conflicts of interest We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. No conflicts of interest exist. AUC — area under curve. ATDs (n) — number of subjects receiving antidepressant treatment. CIDI — Composite International Diagnostic Interview. DD-NOS — depressive disorder not otherwise specified. DSM-IV-TR — Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision. ICD-10 — International Statistical Classification of Diseases and Related Health Problems 10th Revision. MDD — major depressive disorder. MTLE-HS — mesial temporal lobe epilepsy with hippocampal sclerosis. N/A — not available. PWEs —patients with epilepsy.
References [1] Kanner AM, Barry JJ. Is the psychopathology of epilepsy different from that of nonepileptic patients? Epilepsy Behav 2001;2:170–86. [2] Fiest KM, Patten SB, Wiebe S, Bulloch AG, Maxwell CJ, Jetté N. Validating screening tools for depression in epilepsy. Epilepsia 2014;55:1642–50. [3] Krishnamoorthy ES. The evaluation of behavioral disturbances in epilepsy. Epilepsia 2006;47:3–8. [4] Mula M, Iudice A, La Neve A, Mazza M, Mazza S, Cantello R, et al. Validation of the Hamilton Rating Scale for Depression in adults with epilepsy. Epilepsy Behav 2014;41:122–5. [5] de Oliveira GN, Lessa JM, Gonçalves AP, Portela EJ, Sander JW, Teixeira AL. Screening for depression in people with epilepsy: comparative study among Neurological Disorders Depression Inventory for Epilepsy (NDDI-E), Hospital Anxiety and Depression Scale Depression Subscale (HADS-D), and Beck Depression Inventory (BDI). Epilepsy Behav 2014;34:50–4. [6] Jones JE, Hermann BP, Woodard JL, Barry JJ, Gilliam F, Kanner AM, et al. Screening for major depression in epilepsy with common self-report depression inventories. Epilepsia 2005;46:731–5. [7] de Lemos Zingano B, Guarnieri R, Diaz AP, Schwarzbold ML, Bicalho MA, Claudino LS, et al. Validation of diagnostic tests for depressive disorder in drug-resistant mesial temporal lobe epilepsy. Epilepsy Behav 2015;50:61–6. [8] Gandy M, Sharpe L, Perry KN, Miller L, Thayer Z, Boserio J, et al. Assessing the efficacy of 2 screening measures for depression in people with epilepsy. Neurology 2012;79: 371–5.
101
[9] Al-Asmi A, Dorvlo AS, Burke DT, Al-Adawi S, Al-Zaabi A, Al-Zadjali HA, et al. The detection of mood and anxiety in people with epilepsy using two-phase designs: experiences from a tertiary care centre in Oman. Epilepsy Res 2012;98:174–81. [10] Rampling J, Mitchell AJ, Von Oertzen T, Docker J, Jackson J, Cock H, et al. Screening for depression in epilepsy clinics. A comparison of conventional and visual-analog methods. Epilepsia 2012;53:1713–21. [11] Gilliam FG, Barry JJ, Hermann BP, Meador KJ, Vahle V, Kanner AM. Rapid detection of major depression in epilepsy: a multicentre study. Lancet Neurol 2006;5:399–405. [12] Mula M, Iudice A, La Neve A, Mazza M, Bartolini E, De Caro MF. Validation of the Italian version of the Neurological Disorders Depression Inventory for Epilepsy (NDDI-E). Epilepsy Behav 2012;24:329–31. [13] Zis P, Yfanti P, Siatouni A, Tavernarakis A, Gatzonis S. Validation of the Greek version of the Neurological Disorders Depression Inventory for Epilepsy (NDDI-E). Epilepsy Behav 2013;29:513–5. [14] Thomson AE, Calle A, Fontela ME, Yepez L, Muñoz Giacomelli F, Jáuregui A, et al. Screening of major depression in epilepsy: the Neurologic Depression Disorders Inventory in Epilepsy — Spanish version (Argentina). Epilepsia 2014;55:331–4. [15] Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatr Scand 1983;67:361–70. [16] Snaith RP. The Hospital Anxiety and Depression Scale. Health Qual Life Outcomes 2003;1:1–4. [17] Wiglusz MS, Landowski J, Michalak L, Cubała WJ. Reevaluating the prevalence and diagnostic subtypes of depressive disorders in epilepsy. Epilepsy Behav 2015;53: 15–9. [18] Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia 1989;30:389–99. [19] First MB, Spitzer RL, Gibbon M, Williams JBW. Structured Clinical Interview for DSMIV-TR Axis I Disorders, Research Version, Patient Edition. (SCID-I/P) New York: Biometrics Research, New York State Psychiatric Institute; 2000. [20] Löwe B, Spitzer RL, Gräfe K, Kroenke K, Quenter A, Zipfel S. Comparative validity of three screening questionnaires for DSM-IV depressive disorders and physicians' diagnoses. J Affect Disord 2004;78:131–40. [21] Blumer D, Montouris G, Davies K. The interictal dysphoric disorder: recognition, pathogenesis, and treatment of the major psychiatric disorder of epilepsy. Epilepsy Behav 2004;5:826–40. [22] Amiri M, Hansen CP. The interictal dysphoric disorder in patients with epilepsy: a doubtful disorder lacking diagnostic tools. Seizure 2015;24:70–6. [23] Cosco TD, Doyle F, Ward M, McGee H. Latent structure of the Hospital Anxiety and Depression Scale: a 10-year systematic review. J Psychosom Res 2012;72:180–4. [24] Maters GA, Sanderman R, Kim AY, Coyne JC. Problems in cross-cultural use of the Hospital Anxiety and Depression Scale: “no butterflies in the desert”. PLoS One 2013;9:e70975. [25] Brislin RW, Lonner WJ, Thorndike RM. Cross-cultural research methods. New York: J. Wiley; 1973.