R esearch

A rticle

T h e S o w e t o S t r o k e Q uestionnaire

ABSTRACT A questionnaire was designed for a recent survey into the outcome LA HALE CJ EALES VU FRITZ of stroke patients in Soweto, named the Soweto Stroke Questionnaire (SSQ). It was based on the Barthel ADL Index (BI) but modified to suit the local context. This paper introduces the SSQ, and reports on its inter-rater reliability and its concurrent validity. Fifty-four subjects, in the age range 30 to 75 years, were interviewed and nineteen re-interviewed using the SSQ. Four different scores were calculated: a total score, a Barthel Index score, an Impairment score, and a Quality of Life score. The Pearson’s Correlation Coefficient was found to be high between the total score and the BI score. (r=0.948) which supports the concurrent validity of the developed questionnaire. In assessing the reliability of the SQQ, the Wilcoxin Test showed that there was no signifi­ cant difference between the initial and repeat interviews for the total score, the Barthel Index score, and the Impairment score (p<0,05). The Quality of Life Score came closer to a difference, but not statistically significantly so. These tests were collaborated by Bland and Altman graphs which showed that in 95% of the time, the questions were repeatable. Me Nemar’s Test of Symmetry showed that 34 out of 38 questions asked were found to have over 70% correlation. Four questions showed a lower correlation, the lowest being 63.16%. The SSQ was found to have inter­ rater reliability, and to be concurrently valid to the Barthel Index. It is quick and easy to use, requiring no sophisti­ cated equipment or training. It still requires to be investigated for sensitivity and predictiveness, and to be validated in a more general South African stroke population.

KEY WORDS: STROKE OUTCOME, BARTHEL INDEX, SOUTH AFRICA

Although stroke is ranked as the to hygiene, dressing, transfers, resources, as well as dealing with second most common cause of ambulation, continence and societal limitations (Whiteneck, death in South Africa its prevalence feeding) include the Barthel Index 1994). The CIQ (Community in South Africa is not well known (Collin et al, 1988) and the Katz Integrated Questionnaire) designed (Fritz, 1997). The Central Statistical Index of ADL (Brorsson and Asberg, by Wilier et al (1994) is an example ) . Services (1980) reported that cere­ 1984). Instrumental ADL, for of a scale which measures handicap. 3

1 brovascular accident accounted for example the Functional The CIQ was developed for persons 0 2

298/100,000 deaths in Coloured Independence Measure (FIM) with traumatic head-injury and d e South Africans aged 35-74 years, (Ham ilton et al, 1994; Glott, 1995), deals primarily with instrumental t a and 100/100,000 deaths in the explores higher degrees of ADL ADL and handicap. Although it d (

White population. The statistical such as domestic chores, transporta­ measures the success of outcome, r e data for the Black population of tion and communication. The FIM is namely, social integration, it gives h s i South Africa is imprecise (Fritz gaining popularity in South Africa. little assistance in the planning of an l b 1997). Putteril et al (1984) predicted However, it requires appropriate intervention (Wilier et al, 1994). In u

P that the incidence of non-fatal training and trainer testing before a order for a rehabilitation measure­ e stroke in South Africans over high level of reliability can be ment scale to be versatile and effi­ h t 75 years of age was likely to increase achieved (Hamilton et al, 1994). cient, it needs to measure y b

from 100-200/100,000 to Data can be converted from ordinal impairment (in order to plan the d e 2 000/ 100,000. ratings to a linear scale with interval intervention), disability and hand­ t n A review of the literature revealed properties using Rasch analysis. icap (to measure the outcome of the a r numerous stroke outcome This data can be plugged into a data intervention). g

e measuring scales. Some scales bank built up of over 100 000 FIM Dr Derek Wade of the Rivermead c

n predominantly measure impair­ ratings in order to predict manage­ Rehabilitation Centre in Oxford e c

i m ent and m otor recovery (Carr et al, ment, length of hospital stay and suggests that one of the best "buys" l

r 1972; Gowland et al, 1993). outcome (Glott, 1995). However, as on the market is the Barthel Index e d Disability measures of basic activi­ the FIM scale appears not to have (Collin et al, 1988). It is has been n

u ties of daily living (ADL) (attention been validated in South Africa, this used extensively in stroke outcome

y data prediction facility is invalid as studies. This ordinal scale measures a

w one cannot equate data collected in basic functional outcome, requires e t the United States of America to the very little training and has been a CORRESPONDENCE:

G South African context. found to be both reliable and valid

t Mrs L A Hale e Measuring handicap is a lot more for stroke (Wade and Hewer, 1987). n Physiotherapy Department i complex as it has to take into However, it's limitations include the b Faculty of Health Sciences a account the many individual char­ fact that it only measures disability. S

University of the Witwatersrand y acteristics that affect an individual's The BI has not, to the best of my b 7 York Road, Parktown, social role fulfilment, for example, knowledge, been validated in South d e 2193, South Africa education, family support, financial Africa; it is unable to measure c u d o r p 16 SA J o u r n a l o f Physiotherapy 1998 V ol 54 No 4 e R quality of recovery; and has been The aim of this paper is to intro­ scored on an ordinal scale, giving a found to be insensitive in the higher duce the Soweto Stroke Question­ total score of fifty points. This total scores with a low ceiling of measure naire, and to report on its inter-rater score can then be subdivided into (Smith, 1993). reliability and its concurrent four sections, there being no The Soweto Stroke Questionnaire validity. weighting to the distribution of the was designed for a recent survey into scoring (see Table 1): the problems facing Sowetan stroke THE SOWETO STROKE QUESTIONNAIRE 1. Data relating to the patients' patients post-discharge from Chris Although the questionnaire has home circumstances: whether their Hani (CH) Baragwanath Hospital both qualitative and quantitative toilet is located inside the house or and may prove to be a basis from aspects (see Table 1), only the quan­ outside; do they have running water which to design a more specific titative side will be dealt w ith in this and electricity in the home. This South African measuring scale. paper. The quantitative section is data was deemed necessary to collect so that the context of the home could be invisualised. TABLE 1: SOWETO STROKE QUESTIONNAIRE 2. Data relating to the known complications of stroke, for QUANTITATIVE QUALITATIVE example, pain in the hemiplegic shoulder and knee; pressures sores Total score (50) (coded) and contactures. This is known as the Impairment Score. Muscle tone, strength and sensory status were Home Impairment Quality of Life Expanded not measured. Circumstances Score Score Barthel Index 3. Data relating to the patient's (8) (31) (Embedded BI = 20) (3 ) (8) disability - the Barthel Index (BI) is incorporated into the questionnaire, TABLE 2: CORRELATION OF TOTAL SCORES (n = 19) and is expanded in order to gain more insight regarding functional Total Score - Initial (50) Total Score - Repeat (50) ability, the BI being shown to be insensitive in the higher scores Mean 25.95 26.42 ) . known as the expanded Barthel 3 Std. Dev + /- 14.41 + /- 13.10 1 Index score. 0

2 Range 5-46 4-47

4. Data relating to handicap: d WILCOXIN 0.93 e ability to help with household t a chores, socialisation, outings, ability d (

to catch a minibus taxi - known as r TABLE 3: CORRELATION OF IMPAIRMENT SCORES (n = 19) e the Quality of Life score. h s i l Total Score - Initial (8) Total Score - Repeat (8) INTER-RATER RELIABILITY AND b u Mean 4.05 4.16 CONCURRENT VALIDITY P

e Std. Dev + /- 2.29 + /- 2.36 h t METHODOLOGY Range 1 - 8 0 - 8 y b

WILCOXIN 0.87 All cerebrovascular accident d

e (CVA) patients admitted to the t

n medical admissions ward of CH a r TABLE 4: CORRELATION OF QUALITY OF LIFE SCORES (n = 19) Baragwanath Hospital over a four g

e and a half month period were c Total Score - Initial (8) Total Score - Repeat (8) n screened for admission into the e c Mean 3.16 2.68 study. The inclusion criteria of the i l

r Std. Dev +/- 1.92 +/- 1.73 study required that patients, of e d Range 0 - 7 0 - 7 either sex, who had their first and n only CVA in the region of the u

WILCOXIN 0.084 y middle cerebral artery, and who a

w resided in Soweto be included. e t TABLE 5: CORRELATION OF BARTHEL INDEX SCORES (n = 19) Patients had to be in the age range a

G 30 to 75 years. The patients' names t e Total Score - Initial (20) Total Score - Repeat (20) were acquired from the ward's n i Mean 12.26 12.68 admission register, and information b a regarding their inclusion suitability

S Std. Dev + /- 7.22 + /- 6.44 obtained from their medical files. y 0-20 b Range 0-20 The selected subjects were then d WILCOXIN 0.86 e interviewed in their homes three c u d o r

p SA J o u r n a l o f Physiotherapy 1998 V o l 54 No 4 17 e R months post-discharge from showed that in 95% of the time, the a home, the subjects had often being hospital using the Soweto Stroke SSQ questions were repeatable. locked into their homes by their Questionnaire (SSQ), after informed In order to show concurrent relatives for their own safety; the consent had been obtained from validity the Total score of the SSQ assistants often had to travel far and each subject. The interviews were was correlated to the BI scores using with many taxis in order to reach conducted by physiotherapy assis­ Pearson's correlation co-efficient some of the homes. These are real tants trained in the use of the SSQ. (Table 7), bearing in m ind that the BI problems, and as stroke is a chronic Subjects were then randomly has already been shown to have reli­ disorder, unlikely to change rapidly chosen by an "out-the-hat" method ability and validity internationally. with time, the time periods used in for repeat interviews. These repeat The Pearson's Correlation the study were felt not to be unreal­ interviews were performed by Coefficient was found to be high istic. different interviewers in order to between the total score and the BI The Barthel Index has been shown evaluate inter-rater reliability. score (r=0.948). in the past to have both reliability and validity. The SSQ is based on RESULTS DISCUSSION the BI, expanded with additional A total of 361 subjects w ere The criteria for a South African questions felt to be relevant for a admitted during the selection outcome measuring tool has already specific research study. The SSQ was period. Ninety-three patients died been outlined. In many respects, the found to be highly correlated to the and 214 did not fulfill the selection SSQ has been shown to fulfill these BI, indicating that the additional criteria. Fifty-four subjects requirements. questions did not alter the validity comprised the final study group Reliability has been established at of the Index. who were interviewed in their the inter-rater level for the Total Furthermore, the SSQ demon­ homes. Nineteen subjects were scores, Impairment scores, and the strates versatility in that it provides interviewed twice within a period expanded Barthel Index score. This both a measure of functional of sixty days. The following scores reliability was less evident in the outcome and information on which were calculated from each inter­ Quality of Life score. This is not to plan intervention: it measures view: Total score, Impairment score, surprising as "Quality of Life" is a impairment, disability and hand­ Quality of Life score, and a Barthel dynamic construct, and can not be icap. In a country where resources Index score. expected to be the same at different are limited, a tool which measures ) . In order to test inter-rater relia­ times during a disease process. The more than one thing in a simple 3

1 bility, the above mentioned scores of reduced correlation may be due to

0 manner is a very valuable 2 the initial interviews were corre­ the initial support offered by the commodity. d e lated with those of the repeat inter­ subject's friends and family which Two important criteria were that t a views using the Wilcoxin test (see may have waned as time passed the tool should be quick and simple d ( Tables 2, 3, 4, 5). A high degree of and the interest grew less, as often is to administer, and not require r e correlation was shown in the case of the case. The initial interviews may sophisticated equipment. Physio­ h s i the Total scores, the Impairment have been conducted while the therapy assistants were taught how l b scores and the Barthel Index scores social support system was still to use the SSQ in two training u

P (p<0,05), indicating that there was a good, and the repeat interviews sessions, and were able to conduct e

h degree of correlation for these scores performed after this phase was over. the interview in less than an hour, t between the initial interview and Question for question, the SSQ with no problems. The assistants y b

the repeat interview. showed a high degree of reliability. were all residents of Soweto, and d e The Quality of Life Score, did not The physiotherapy assistants had when asked their opinion regarding t n show a statistically significant logistical difficulties in revisiting which languages the questionnaire a r difference between the initial and subjects. For example, the lack of should be translated into (remem­ g

e repeat interview mean scores, and telephones prevented appointments bering that South Africa has eleven c n its degree of correlation was not as from being set-up; when arriving at official languages), they were e c i high as with the other scores. l

r The SSQ was then tested for the e TABLE 6: Me NEMAR'S TEST OF SYMMETRY (n = 19) d degree of reliability question for n (Question for question correlation between initial and repeat interviews) u question using Me Nemar's test for y sym m etry (Table 6). This test a % Correlation Number of Questions w showed a greater than 70% correla­ e t tion between the initial and repeat 100% 5 a

G 90 - 99% 5

interviews for 34 out of 38 ques­ t e tions asked. Four questions 80 - 89% 12 n i showed a lower correlation, the b 70 - 79% 12 a low est being 63%. S 60 - 69% 4 y These results were collaborated by b Bland and Altman graphs which <60% 0 d e c u d o r

p 18 SA J o u r n a l o f Physiotherapy 1998 V ol 54 No 4 e R adamant that it should be in tion measured on the Barthel Index. predictive validity, and responsive­ English. The only equipment In Soweto many houses have ness still need to be established. required was the questionnaire and outside toilets accessed by dirt These items are presently being a pen. paths. This requires a greater degree researched both in the Sowetan The final criteria was that the SSQ of ability to manage independently community with an upgraded ques­ should have credibility in the inter­ Stairs are often not encountered by tionnaire, as well as in a rural national community. This has been residents of Soweto who are setting of South Africa. achieved by having the Barthel confined to their single-storied The SSQ is an ordinal scale in that Index em bedded within the SSQ. homes and surroundings, and it hierarchally classifies stroke Still to be shown is the test-retest therefore this question is of doubtful patients. It has been argued that reliability of the scale; and its value. The ability to wash one's ordinal scales only give a relative, content, construct and predictive body may require a greater degree and not a definite, measure, and validity. It must be validated to the of ability if one has to fetch one's thus should not be added to a South African stroke population in w ater from a com m unity source and sumscore unless specifically justi­ general. The time taken to admin­ then use it standing in a tin basin. fied (Sedring, 1995). However, the ister the questionnaire and its Therefore a subject in Soweto with SSQ is predominantly a "yes/no" responsiveness must still be specifi­ the same Barthel Score as a subject scale. Thus it exhibits ratio-scale cally established. in England could actually be func­ properties: there are only two So w here to from here? If all that is tioning from a different disability possible states. These can be repre­ required is a functional baseline level. Within a given population sented statistically as "0,1" and can measurement, then the Barthel group of similar home conditions thus be used for ANOVAs or corre­ Index is probably the quickest and the scoring will be standard, unfor­ lations (Krebs, 1987). Dichotomous easiest method to use. tunately these conditions are not scales show, on the one hand, high However, the Barthel Index has always met in Soweto, as people reliability, however, on the other been criticised for not having the live under different conditions, hand, the lack of choice can often ability to measure the quality of the from squatter camps to fair sized decrease their sensitivity (Sedring, recovery of functional movement, houses with inside bathroom facili­ 1995). Em bedded w ithin the SSQ, is and in the higher scores has been ties. the Barthel Index. The Barthel found to be insensitive in discerning The Barthel Index only measures Index, also an ordinal scale, has ) . the degree of recovery (Smith, 1993). disability. This makes its usefulness been added to sumscores which 3

1 The highest functional levels in designing a rehabilitation have frequently been used in statis­ 0 2

assessed are whether the subject can programme limited as the reasons tical analysis. Therefore, statistical d

e walk independently (with or (the impairments) for the disability analysis of the SSQ data for research t a without an appliance) and manage are not identified. In rehabilitation is justifiable. d (

stairs. Walking in different environ­ these reasons are addressed in order The scoring of the SSQ carries no r e mental conditions is not ascer­ to improve the degree of disability. weighting, for example, the impair­ h s

i tained. This often requires a greater Improvement in disability, hope­ ment and quality of life subsections l b degree of recovery than walking fully but not necessarily, will then are scored out of eight, whereas u

P indoors. translate into an improvement of ADL is scored out of twenty. This by

e The Barthel Index has been widely handicap or quality of lifestyle. no means reflects a perceived h t used in the United Kingdom but its The SSQ probably provides more importance of one subsection over y b

suitability for the South African information on which to base another. Further investigation into d

e stroke population has not been management and intervention than the scoring is required. t

n investigated, and needs to be estab­ the BI. It can thus be argued that its a CONCLUSION r lished. For example, the ability to go construct validity has been demon­ g

e to the toilet independently is a func­ strated. However, content and The SSQ is in its infancy, however c

n it does provide both a functional e c status and an impairment measure. i l

TABLE 7: CORRELATION OF TOTAL SCORE TO BARTHEL INDEX

r Although it was designed to meet e SCORES (n= 54) d the criteria of a specific research n study, it does fulfill the require­ u

y Initial (Total Score = 50) Repeat (Total Score = 20) ments for an outcome measuring a Mean 31.73 15.21 scale suitable for South Africa. w e t Std. Dev + /- 12.49 Jules Rothstein, editor of Physical a + /- 5.84

G Therapy recently wrote in an edito­ Median 34 18 t

e rial entitled "Break on Through": n Range 5-48 0 - 20 i "Accountability, evidence-based b a Pearson's practice, and out-come based S Correlation management are not ploys of an evil y b Coefficient 0.95 empire, just concepts that are diffi­ d e cult to implement - and almost c u d o r SA J o u r n a l o f Physiotherapy 1998 V o l 54 No 4 p 19 e R impossible to logically argue Fritz VU. Stroke incidence in South Africa. S Smith A. Beware of the Barthel. against" (Rothstein, 1997). Afr Med J 1997; 87(5): 584 - 585 Physiotherapy 1993; 79 (12): 843 - 844 It is im perative that stroke rehabil­ Glott T. Functional independence measure S0dring KM. The S0dring motor evaluation itation in South Africa be measured, scale on stroke patients. Physiotherapy in of stroke patients. Physiotherapy in Stroke and thus an appropriate tool is Stroke Management. Edited by M. A. Management. Edited by M.A. Harrison, highly desirable. Harrison, Churchill Livingstone, Edinburgh, Churchill Livingstone, Edinburgh, 1995, 1995, Chapter 16, 139- 144 Chapter 14, 119- 123

REFERENCES Gowland C, Stratford P, Ward M, et al. Wade DT, Langton-Hewer R. Functional abil­ Measuring physical impairment and disability Brorsson B, Asberg KH. Katz Index of inde­ ities after stroke measurement, natural history pendence in ADL: reliability and validity in with the Chedoke-Mc Master stroke assess­ and prognosis. J Neuro Neurosurg Pysch short-term care. Scand J Rehabil Med 1984: ment. Stroke 1993; 24 (1): 58 - 63 1987; 50: 177 - 182 16: 125 - 132 Hamilton BB, Laughlin JA, Roger CF, Whiteneck GG. Measuring what matters: key Granger CV. Inter-rater reliability o f the Carr JH, Shepherd RB, Nordholm L, Lynne rehabilitation outcomes. The 44th annual seven level Functional Independence D. Investigation of a new motor assessment John Stanley Coulter lecture. Arch Phys Med scale for stroke patients. Phys Ther 1972; 65: Measure (FIM). Scand J Rehabil Med 1994; Rehabil 1994; 75: 1073 - 1076 175 - 178 26:115 - 119 Wilier B, Ottenbacher KJ, Coad ML. The Krebs DE. Measurement theory. Phys Ther Central Statistical Services. Report on community intergration questionnaire. Am J 1987; 67 (12): 1834- 1839 Deaths: Whites, Coloureds and Asians, 1980: Phys Med Rehabil 1994: 73 (2): 103 - 111 Report 07-03-19. Government Printer, Putteril JS, Disler PB, Jacka E. Coping with Pretoria chronic illness. Part II. Cerebrovascular acci­ This study was granted ethical clearance from dents. S Afr Med J 1984; 65: 891 - 893. the Committee for Research on Human Collin C, Wade DT, Davies S, Home V. The Subjects, University of the Witwatersrand: Barthel Index: a reliability study. Int. Disabil. Rothstein JM. Break on through (Editorial) . No. 950 111. The study was funded by the Studies 1988; 10 (2): 61 - 63 Physical Therapy 1997; 77 (12): 1680 Medical Research Council of South Africa.

) The College of . 3 1

0 Physiotherapists of 2 d e t South Africa (CPSA) a d ( r

e The CPSA is an independent examining body for h s i

l clinical specialisation. It is currently also in the process b u of developing an accreditation and certification system

P ' i i n, ■' s:-'; 'i ■ ’> i - " ■ e

h for Continuing Physiotherapy Education (CPE) t

y courses. Certificates issued by a body such as the b -1 *.) ■ ' ■ I- If.-.ii ; . I., I . d

e CPSA have more national and international recogni­ t n

a tion than certificates issued by a Special Interest r g

e Group or private companies. c n e c ^ B

i Any one running or organising CPE courses can l r e apply for accreditation of such courses. It is envisaged d n that the process of accreditation and certification will u

y <_ -it I i'‘ i"ltr a commence early in 1999. w e t a For further information please contact: The Registrar G t e .-j \r i or President of the CPSA, Physiotherapy Department, n i

b Medical School, 7 York Road, Parktown, 2193 a S Johannesburg. y b d e c u d o r 20 SA J o u r n a l o f Physiotherapy 1998 V ol 54 No 4 p e R