Developing a Mood Scale Suitable for Use with

Developing a Mood Scale Suitable for Use with Aphasic Stroke Patients Paul Barrows

Stroke Incidence & Prevalence • In England Wales it is estimated that 130, 000 people have a stroke every year (Office for National Statistics, 2001). • There are over 1. 1 million people living with stroke in the UK, but by 2020 it is expected to double, mainly due to the increasing proportion of older people (Scarborough et al. , 2009).

Stroke Mortality • After heart disease and cancer, stroke is the third most common cause of death in the UK and worldwide (Warlow et al. , 2003) • Stroke is responsible for over 60, 000 deaths annually in the UK. (The Stroke Association, 2006) • Over the last decade, stroke has accounted for an average of around 7. 5% of deaths in men and 10% of deaths in women (Office for National Statistics, 2010).

Stroke Outcome • Of those people who have a stroke, about third are likely to die within the first ten days, about third can be expected to make a recovery within one month and about a third will likely be left severely disabled (Bosanquet & Franks, 1998).

Depression after Stroke Approximately one third of stroke survivors suffer from poststroke depression (PSD) (Hackett et al. , 2005). This can lead to: • Increased morbidity and mortality • Poorer outcomes in recovery of physical and cognitive function • Impeded recovery and longer hospital stays • Increased caregiver stress.

Depression after Stroke Clinical guidelines recommend screening of stroke patients, and assessment of those identified as at risk of depression (The Royal College of Physicians, 2012).

Measuring Mood in Stroke Patients Though many measures of depression and low mood exist, stroke offers challenges to assessing patients for co-morbid mental health problems. Assessment of mood in this group is problematic because: • Mood states are often masked or mimicked by neurological consequences of stroke, such as dysprosodia and dyspraxia • Impaired language and cognition – commonly in the form of aphasia – makes assessment by self-report measures problematic.

Exclusion of Aphasic Stroke Patients in Studies • A study of adaptations to measuring depression in stroke patients with aphasia revealed that 63% of studies examining PSD excluded patients whose aphasia was too severe for them to be amenable to standard measures (E. Townend et al. , 2007) • Hackett and Anderson (2005) report that only 3 of 20 studies they reviewed in which predictors of PSD were examined included aphasia as a potential risk factor. This underlines an overwhelming tendency for patients with communication impairments to simply be omitted from such studies.

Stroke, Depression & Aphasia affects around 20– 38% of stroke patients (E. Townend et al. , 2007). • One of the two most important predictors of immediate major depression is aphasia (p<0. 001) (Astrom et al. , 1993) • “Communication impairment was the greatest predictor of depression severity and prognosis” (Thomas & Lincoln, 2006) • Two-thirds of patients with aphasia met the DSM-III-R criteria for depression in the first year following stroke, a figure that was significantly higher than in those without aphasia (Kauhanen et al. , 2000).

Measuring Mood in Stroke Patients Observer-Rated • Poststroke Depression Rating Scale (PSDS) (Gainotti et al. , 1997) • Stroke Aphasic Depression Rating Scale (SADQ H 10/21) (Sutcliffe & Lincoln, 1998) • Aphasic Depression Rating Scale (ADRS) (Benaim et al. , 2004). Self-Report • Visual Analogue Mood Scales (VAMS) (Stern et al. , 1997) • Visual Analogue Self Esteem Scales (VASES) (Brumfitt & Sheeran, 1999) • Depression Intensity Scale Circles (DISCS) (Turner. Stokes et al. , 2005).

Need for Screening Measures There is evidence to suggest that observer-rated scales like the SADQ-H 10 and the ADRS are of clinical use (Bennett et al. 2006; Hacker et al. , 2010; Benaim et al. 2010). However self-report measures have proven more problematic. There is therefore a need for self report screening tools that can more accurately assess patients in the stroke population, particularly those with communication difficulties.

Visual Analogue Self-Esteem Scale (VASES) Brumfitt & Sheeran (1999) created a 10 item, self-esteem scale for use with aphasic patients. It is made up of 10 pairs of bipolar pictures of concepts related to self-esteem, which are judged on a 5 -point Likert scale below them. Often used as an indirect indicator of depression. Brumfitt & Sheeran (1999)

The Depression Intensity Scale Circles (DISCs) Essentially a Likert Scale, but using black disks of increasing size to enable cognitively impaired patients to quantify their mood state more easily. Turner-Stokes et al. (2005)

Visual Analogue Mood Scales (VAMS) Stern et al. (1997) developed a set of 8 VAS-based mood scales in pictographic form. Happy, Sad, Energetic, Tired, Angry, Afraid, Tense, Confused. Each is comprised of two cartoon faces at either end of a 100 mm line. The user indicates their mood by making a mark along the length of the line. VAMS ‘Sad’ Item

Visual Analogue Mood Scales (VAMS) VAMS validated in several populations, including stroke and dementia patients (Stern et al. 1997; Arruda et al. 1999; Temple 2004). However: • Researchers report patients being ‘bewildered’ by the VAMS (Townend et al. , 2007). • Though VAMS offers a broad measure of severity of depression, it is of little use as a screening instrument (Bennett et al. , 2006) • “The use of Visual Analogue Mood Scales amongst patients with aphasia {. . . } cannot be recommended. ” (Berg et al. , 2009).

Visual Analogue Mood Scales (VAMS) • Typically, studies use only a single VAMS scale – the ‘sad’ item – to assess mood • VAMS are separate, single-item scales, so a combined total score is not possible • VAMS not underpinned by any theory of mood • Evidence suggests that stroke patients with cognitive impairments cannot use VASs effectively (Price et al. , 1999).

Addressing Limitations of VAMS Cartoon faces used to denote affect Poor realism Use photos of human faces

1 a. Using Photos Of Human Faces Why use Human Faces? VAMS ‘Sad’ Item Study Participant • Facial expression is arguably the most powerful, universal communicator of mood states (Ekman, 1971, 1993) • Photographs of actual emotional expressions offer a more nuanced, detailed and accurate portrayal of mood states • “. . . use of realistic looking pictures {. . . } may usefully support communication about mood with people with aphasia”– Townend et al (2007)

1 b. Using Photos Of Human Faces Why use Human Faces? • Aphasia can impact upon a person’s ability to decode symbols, such as those in written language, and this may also affect their ability to recognise such graphics • Evidence suggests that recognition of emotion in facial expressions occurs primarily in the right hemisphere (Adolphs et al. , 1996; Philippi et al. 2009). Most aphasic patients have left hemisphere lesions, therefore affect recognition in faces is unlikely to be affected.

Addressing Limitations of VAMS Cartoon faces used to denote affect Poor realism Use photos of human faces Scales separate and unconnected No underlying theory Adopt explicit structural theory

2 a. Explicit Structural Theory The Valence-Activation Model of Affect Larsen & Diener (1992)

2 b. Explicit Structural Theory Using Scales Across 2 -factor Affect Space Larsen & Diener (1992)

2 b. Explicit Structural Theory Using Scales Across 2 -factor Affect Space

Addressing Limitations of VAMS Cartoon faces used to denote affect Poor realism Use photos of human faces Scales separate and unconnected No underlying theory Adopt explicit structural theory VAS requires cognitive interpolation Cognitively impaired can not use VASs Adopt explicit graphical interpolation

Cognitive Interpolation Sca l e. M axi mu m lt u s e R Scaled M e l Sca m u inim Reference State ?

Explicit Interpolation Dynamic Visual Analogue Scale (DVAS) Reference State A slider control dynamically animates a picture to explicitly anchor points of the scale to transitional images

Addressing Limitations of VAMS Cartoon faces used to denote affect Poor realism Use photos of human faces Scales separate and unconnected No underlying theory Adopt explicit structural theory VAS requires cognitive interpolation Cognitively impaired can not use VASs Adopt explicit graphical interpolation

Study 1 Part 1 – Producing a. Objectives Suitable Set of Photographs 1. To select and test a pool of mood words for their ability to evoke recognisable facial expressions, based on which a smaller pool of candidate descriptors will be selected for further study 2. To confirm that judgments of facial expression are consistent with a two-factor, circumplex model of affect, and to establish a coordinate system for mood words and corresponding expressions 3. To identify ‘actors’ who are good at producing recognisable facial expressions, and whose photos may be candidates for the final scales.

Study 1 Part 1 – Producing of Photographs Studya 1 Suitable - Three Set Parts 1. Producing a suitable set of photographs representing a range of candidate mood words 2. Judging the photographs 3. Eliminating weaker items, and identifying candidate mood words/faces for the final scales

Study 1 Part 11 –– Producing aa Suitable Set of of Photographs Participants Gender: 10 male, 10 female (n=20) Age: mean 20. 6 years; S. D. = 1. 7 years English: 12 (60%) spoke English as a first language, 8 (40%) spoke English fluently as a second language. Ethnicity: 12 (60%) were ethnically European, 6 (30%) were Asian or East Asian and 2 (10%) were of African descent.

Study 1 Part 11 –– Producing aa Suitable Set of of Photographs Methods Recruitment: Advertisements posted about University of Nottingham and Community Centres Selection criteria: >18 years; fluent in English; have no facial hair, tattoos; capable of posing facial expressions Incentive: £ 10 for session up to approx 1 hour Task: Participants photographed posing expressions based on 26 mood words.

Study 1 Part 11 –– Producing aa Suitable Set of of Photographs Methods Stimulus words 26 mood words from previous research in the area which: 1) span circumplex well, 2) represent persistent mood states, and 3) include items used in VAMS.

Study 1 Part 11 –– Producing aa Suitable Set of of Photographs Results Photographs: Total of 1, 560 photographs taken of 20 ‘actors’ posing 26 mood states (3 photos each). Pre-screening: ‘Best of three’ task independently completed by experimenter and 2 staff. 520 (20 x 26) photographs selected for final pool.

Study 1 Part 11 –– Producing aa Suitable Set of of Photographs Results 26 Posed Moods Pleased Happy Excited Energetic Enthusiastic Aroused Afraid Anxious Tense Angry Nervous Confused Sad Distressed Miserable Disappointed Depressed Bored Tired Sleepy Relaxed Content Peaceful Neutral Calm Satisfied

Study 1 Part 1 – Part Producing a Suitable Set of Photographs 2 – Judging the Photographs Participants Gender: 21 male, 23 female (n=44) Age: mean 24. 6 years; S. D. = 6. 2 years English: 25 (56%) spoke English as a first language, 20 (44%) spoke English fluently as a second language. Ethnicity: 15 (33%) were ethnically European, 25 (56%) were Asian or East Asian, 2 (4%) were of African descent and 3 (7%) were mixed race.

Study 1 Part 1 – Part Producing a Suitable Set of Photographs 2 – Judging the Photographs Methods Recruitment: Advertisements posted about University of Nottingham campus Selection criteria: >18 years; fluent in English Incentive: £ 4 per dataset Task: Online judgement task. Each of 26 photographs judged on 26, 7 -point Likert scales corresponding to the stimulus mood words.

Study 1 Part 1 – Part Producing a Suitable Set of Photographs 2 – Judging the Photographs Methods Data Collection Project portal website and database created to manage data collection at www. xvams. com. 100 (26 x 26) datasets collected (n=44), Total of 67, 600 judgements. *XVAMS = Extended Visual Analogue Mood Scales.

Study 1 Part 1 – Part Producing a Suitable Set of Photographs 2 – Judging the Photographs Results A principle component analysis demonstrates a clear, 2 -factor solution (Eigenvalue cut-off 1. 75– 2. 4). Factor 1 identified as valence (70%). Factor 2 identified as activation (20%). Extraction Sums of Squared Loadings Component 1 2 3 Total 18. 279 4. 963 1. 278 % Variance Cumulative % 70. 303 19. 088 89. 391 4. 917 94. 307

Study 1 Part 1 – Part Producing a Suitable Set of Photographs 2 – Judging the Photographs Results . 80 Factor 2 Plot of factor loadings yields clear evidence of circumplexical structure, with mood words following predicted pattern. Tense Confused Distressed Angry 1. 50 Aroused . 60 Nervous Anxious Afraid . 40 Happy Pleased Satisifed Content . 20 Disappointed Miserable 1. 00. 50 Depressed Sad . 00 -. 50 Peaceful Relaxed -. 60 Tired Sleepy -1. 00 -. 20 -. 40 Bored Energetic Excited Enthusiastic Neutral -. 80 -1. 00 Factor 1 Calm -1. 50

Study 1 Part 1 – Part Producing a Suitable Set of Photographs 2 – Judging the Photographs Results Examination of correlation matrix and S. D. values identified stronger items that could be retained for Part 3. Energetic Pleased Excited Happy Enthusiastic Sad Afraid Disappointed Tense Angry Nervous Distressed Anxious Aroused Calm Bored Tired Miserable Depressed Satisfied Relaxed Confused Sleepy Content Peaceful Neutral

Study 1 Part 1 – Part Producing a Suitable Set of Photographs 2 – Judging the Photographs Results Examination of correlation matrix and S. D. values identified stronger items that could be retained for Part 3. Satisfied Happy Excited Afraid Angry Distressed Sad Miserable Bored Sleepy Calm Peaceful

Study 1 Part a Suitable Set of Photographs Part 13––Producing Judging the 12 Item Photograph Subset Repeat of Part 2, with 12 Item Subset • After eliminating weaker items, results should be more accurate • Smaller datasets mean more images sets can be completed (5 sets instead of just one); many more datasets per actor required to allow comparison • 44% of Part 2 participants spoke ESL. Higher proportion of native English speakers desirable • Most were students at Nottingham, need more geographically diverse for external validity.

Study 1 Part a Suitable Set of Photographs Part 13––Producing Judging the 12 Item Photograph Subset Participants Gender: 38 male, 26 female (n=64) Age: mean 33 years (18– 72 years); S. D. = 14. 8 years English: 53 (83%) spoke English as a first language, 11 (17%) spoke English fluently as a second language. Ethnicity: 46 (72%) were ethnically European, 6 (9%) were Asian or East Asian, 1 (2%) was of African descent and 3 (5%) were mixed race.

Study 1 Part a Suitable Set of Photographs Part 13––Producing Judging the 12 Item Photograph Subset Participants – Locations Canada & U. S. U. K. Europe (non-U. K. ) S. Afr/Austr/Asia 37 9 5 4 (67%) (16%) (9%) (7%)

Study 1 Part a Suitable Set of Photographs Part 13––Producing Judging the 12 Item Photograph Subset Methods Recruitment: Advertisements posted in U. S. and Canada, and on online forums Selection criteria: >18 years; fluent in English Incentive: £ 4/$6/0. 05 BTC per experiment (5 datasets) Task: Online judgement task. Each of 12 photographs judged on 12, 7 -point Likert scales corresponding to the stimulus mood words.

Study 1 Part a Suitable Set of Photographs Part 13––Producing Judging the 12 Item Photograph Subset Methods Data Collection Project portal website and database created to manage data collection at www. xvams. com. 540 (12 x 12) datasets collected (n=64), Total of 77, 760 judgements.

Study 1 Part a Suitable Set of Photographs Part 13––Producing Judging the 12 Item Photograph Subset Results As before, principle component analysis demonstrates a clear, 2 -factor solution (Eigenvalue cut-off 1. 45). Factor 1 identified as valence (60. 2%). Factor 2 identified as activation (24. 7%). Extraction Sums of Squared Loadings Component 1 2 Total 7. 228 2. 964 % Variance Cumulative % 60. 232 24. 704 84. 937

Study 1 Part a Suitable Set of Photographs Part 13––Producing Judging the 12 Item Photograph Subset Results -. 80 Excited -. 60 Afraid Happy -. 40 Angry Distressed Satisfied -. 20 Factor 2 As in Part 2, plot of factor loadings yields evidence of circumplexical structure, with mood words following predicted pattern. -1. 00 1. 50 1. 00 . 50 Miserable . 00 -. 50 -1. 00 . 20 Sad Peaceful . 40 Calm . 60 Bored . 80 Sleepy 1. 00 Factor 1 -1. 50

Study 1 Part a Suitable Set of Photographs Part 13––Producing Judging the 12 Item Photograph Subset Candidate Scales -1. 00 Prospective bipolar scales and correlation coefficients (r) Pole 2 Afraid Angry Distressed 1. 50 1. 00 . 50 . 00 -. 95 -. 94 Distressed Peaceful -. 91 -. 93 Sad Happy -. 89 -. 82 . 40 Bored Excited -. 83 -. 8 . 60 Angry Peaceful -. 75 -. 73 -. 7 -. 62 Satisfied -. 20 Pt 2 Pt 3 Satisfied Calm Happy -. 40 Miserable Afraid Excited -. 60 Factor 2 Pole 1 -. 80 Miserable -. 50 -1. 00 . 20 Sad Bored . 80 Peaceful Calm Sleepy 1. 00 Factor 1 -1. 50

What Next? Part 1 –Part Producing a Suitable Set of Photographs 4 – Building the Candidate Scales 1) Recall two highest scoring ‘actors’ for prototype scales 2)Get images for high activation, neutral valence (alert/surprised) for sleep-alert scale 3) Get images for poses of transitions along each scale vector, i. e. : at least two reference images between scales end-points and ‘neutral’ (origin) – validate through judgement study 4) Apply ‘morphing’ to neighbouring images along each scale to render scale to many finely graded images.

What Next? Part 1 –Part Producing a Suitable Set of Photographs 4 – Building the Candidate Scales -1. 00 -. 80 Collecting transitional images along scale vectors. Excited -. 60 -. 40 Satisfied Factor 2 -. 20 1. 50 1. 00 Miserable . 50 . 00 . 20 Examples: 1) Excited–Bored 2) Miserable–Satisfied . 40 . 60 Bored . 80 1. 00 Factor 1 -. 50 -1. 00 -1. 50

References Adolphs, R. , Damasio, H. , Tranel, D. , & Damasio, R. (1996). Cortical systems for the recognition of emotion in facial expressions. The Journal of Neuroscience, 16(23), 7678 -7687. Astrom, M. , Adolfsson, R. , & Asplund, K. (1993). Major depression in stroke patients. A 3 -year longitudinal study. Stroke, 24(7), 976 -982. Benaim, C. , Cailly, B. , Perennou, D. , & Pelissier, J. (2004). Validation of the aphasic depression rating scale. Stroke, 35(7), 1692 -1696. Benaim, C. , Decavel, P. , Bentabet, M. , Froger, J. , Pelissier, J. , & Perennou, D. (2010). Sensitivity to change of two depression rating scales for stroke patients. Clinical Rehabilitation, 24(3), 251 -257. Bennett, H. E. , Thomas, S. A. , Austen, R. , Morris, A. M. , & Lincoln, N. B. (2006). Validation of screening measures for assessing mood in stroke patients. British Journal of Clinical Psychology, 45(Pt 3), 367 -376. Berg, A. , Lonnqvist, J. , Palomaki, H. , & Kaste, M. (2009). Assessment of depression after stroke: a comparison of different screening instruments. Stroke, 40(2), 523529. Brumfitt, S. M. , & Sheeran, P. (1999). The development and validation of the Visual Analogue Self-Esteem Scale (VASES). British Journal of Clinical Psychology, 387400.

References II Ekman, P. , & Friesen, W. V. (1971). Constants across cultures in the face and emotion. Journal of Personality and Social Psychology, 17(2), 124 -129. Gainotti, G. , Azzoni, A. , Gasparini, F. , Marra, C. , & Razzano, C. (1997). Relation of lesion location to verbal and nonverbal mood measures in stroke patients. Stroke, 28(11), 2145 -2149. Hacker, V. L. , Stark, D. , & Thomas, S. (2010). Validation of the stroke aphasic depression questionnaire using the brief assessment schedule depression cards in an acute stroke sample. British Journal of Clinical Psychology, 49(Pt 1), 123 -127. Hackett, M. L. , Yapa, C. , Parag, V. , & Anderson, C. S. (2005). Frequency of depression after stroke: a systematic review of observational studies. Stroke, 36(6), 1330 -1340. Kauhanen, M. L. , Korpelainen, J. T. , Hiltunen, P. , Maatta, R. , Mononen, H. , Brusin, E. , . . . Myllyla, V. V. (2000). Aphasia, depression, and non-verbal cognitive impairment in ischaemic stroke. Cerebrovascular Diseases, 10(6), 455 -461. Office for National Statistics. (2001). Stroke incidence and risk factors in a populationbased prospective cohort study. Office for National Statistics Health Statistics Quarterly, 12. Office for National Statistics. (2010). Mortality Statistics: Deaths registered in 2010 (Series DR) Tables 1– 4 and Tables 6– 14. Retrieved from http: //www. ons. gov. uk/ons/rel/vsob 1/mortality-statistics--deaths-registered-inengland-wales--series-dr-/2010/dr-table 5 -2010. xls

References III Ekman, P. , & Friesen, W. V. (1971). Constants across cultures in the face and emotion. Journal of Personality and Social Psychology, 17(2), 124 -129. Gainotti, G. , Azzoni, A. , Gasparini, F. , Marra, C. , & Razzano, C. (1997). Relation of lesion location to verbal and nonverbal mood measures in stroke patients. Stroke, 28(11), 2145 -2149. Hacker, V. L. , Stark, D. , & Thomas, S. (2010). Validation of the stroke aphasic depression questionnaire using the brief assessment schedule depression cards in an acute stroke sample. British Journal of Clinical Psychology, 49(Pt 1), 123 -127. Hackett, M. L. , Yapa, C. , Parag, V. , & Anderson, C. S. (2005). Frequency of depression after stroke: a systematic review of observational studies. Stroke, 36(6), 1330 -1340. Kauhanen, M. L. , Korpelainen, J. T. , Hiltunen, P. , Maatta, R. , Mononen, H. , Brusin, E. , . . . Myllyla, V. V. (2000). Aphasia, depression, and non-verbal cognitive impairment in ischaemic stroke. Cerebrovascular Diseases, 10(6), 455 -461. Office for National Statistics. (2001). Stroke incidence and risk factors in a populationbased prospective cohort study. Office for National Statistics Health Statistics Quarterly, 12.

References IV Office for National Statistics. (2010). Mortality Statistics: Deaths registered in 2010 (Series DR) Tables 1– 4 and Tables 6– 14. Retrieved from http: //www. ons. gov. uk/ons/rel/vsob 1/mortality-statistics--deaths-registered-inengland-wales--series-dr-/2010/dr-table 5 -2010. xls Philippi, C. L. , Mehta, S. , Grabowski, T. , Adolphs, R. , & Rudrauf, D. (2009). Damage to association fiber tracts impairs recognition of the facial expression of emotion. The Journal of Neuroscience, 29(48), 15089 -15099. Price, C. I. , Curless, R. H. , & Rodgers, H. (1999). Can stroke patients use visual analogue scales? Stroke, 30(7), 1357 -1361. Royal College of Physicians. (2012). National Clinical Guidelines for Stroke, 4 th edition. Prepared by the Intercollegiate Stroke Working Party. London: RCP. Scarborough, P. , Peto, V. , Bhatnagar, P. , Kaur, A. , Leal, J. , Luengo-Fernandez, R. , . . . Allender, S. (2009). Stroke Statistics: 2009 Edition. British Heart Foundation. Stern, R. A. , Arruda, J. E. , Hooper, C. R. , Wolfner, G. D. , & Morey, C. E. (1997). Visual analogue mood scales to measure internal mood state in neurologically impaired patients : description and initial validity evidence. Aphasiology, 11(1), 59 -71. Sutcliffe, L. M. , & Lincoln, N. B. (1998). The assessment of depression in aphasic stroke patients: the development of the Stroke Aphasic Depression Questionnaire. Clinical Rehabilitation, 12(6), 506 -513.

References V The Stroke Association. (2006). R 11 Stroke Statistics Resource Sheet. Retrieved from http: //www. stroke. org. uk/document. rm? id=877 Thomas, S. A. , & Lincoln, N. B. (2006). Factors relating to depression after stroke. British Journal of Clinical Psychology, 45(Pt 1), 49 -61. Townend, E. , Brady, M. , & Mc. Laughlan, K. (2007). A systematic evaluation of the adaptation of depression diagnostic methods for stroke survivors who have aphasia. Stroke, 38(11), 3076 -3083. Turner-Stokes, L. , Kalmus, M. , Hirani, D. , & Clegg, F. (2005). The Depression Intensity Scale Circles (DISCs): a first evaluation of a simple assessment tool for depression in the context of brain injury. Journal of Neurology, Neurosurgery and Psychiatry, 76(9), 1273 -1278. Warlow, C. , Sudlow, C. , Dennis, M. , Wardlaw, J. , & Sandercock, P. (2003). Stroke. Lancet, 362(9391), 1211 -1224.