By Author
  By Title
  By Keywords

March 1998, Volume 48, Issue 3

Special Communication

A Comparative Review of Two Screening Instruments; The Aga Khan University Anxiety and Depression Scale and The Self Reporting Questionnaire

Badar Sabir Ali  ( Department of Community Health Sciences, The Aga Khan University, Karachi. )
Siraj Amanullah  ( Department of Community Health Sciences, The Aga Khan University, Karachi. )

The Prevalence of Menial Illness:
Globally mental illness is on the rise. WHO in 1993 reported a world wide prevalence of 500 million cases and has further quoted that the ratio between psychialrists and patients in developing countries is 1:1 million1. The psychiatrists that are present arc concentrated in the cities. The majority of the population is deprived of health care in general and mental health care is non existent. Goldberg2 has reported a prevalence of psychiatric illness of 30% in general medical patients and further states that out of 250 psychiatric patients per 1000 per year 230 were accessible to primary health care. Rieger states that an overwhelming majority of mentally ill patients are treated by general practitioners and allied health professionals3.
The Role of Community Health Workers in Identifica­tion of Mental Illness:
WHO has clearly designated case finding as the responsibility of Community Health Workers (CHW). Buesenberg states that currently CHWs miss 2 out of 3 cases of mental disorder encountered during theirroutine work. This situation of wide spread morbidity and dismal access to proper care led WHO to developing a screening instrument that could be used by the CHWs for case finding.
The WHO Screening Instrument:
Harding et al in 19804 under the aegies of WHO, carried outa multinational collaborative study in which they reviewed several already existing screening instruments and selected 20 items by mutual agreement and developed a questionnaire. Initially psychiatrists, public health workers and other experts involved were from India, Senegal. Columbia and Sudan. Later they were joined by representatives from Egypt, Brazil and Philippines. This instrument was named the Self Reporting Questionnaire (SRQ) as it was planned to use it as a self administered instrument. It was soon realized that due to the low literacy level in the developing countries, it would have to be used as a verbally administered questionnaire by the CHWs. Buesenberg in 1993 developed a training manual for the CHWs, which has been published by the mental health division of WHO titled” A User’s Guide to Tile Self Reporting Questionnaire”1. The initial SRQ had 24 stems. Four of these were to assess psychosis, which were dropped, as psychosis is not as common as the affective illnesses (the life time prevalence for schizophrenia is 1.3%5, for depression it is 20% in women and 10% in men6) and that psychosis is more easily identified. Out of the 20 items of SRQ, 12 are psychiatric and 8 are somatic. Two stems address the same attribute triedness, this is done ‘to find out whether the answers obtained are reliable or not. In. reality the SRQ evaluates 19 attributes of anxieiy and depression. The CHWs inquires about the presence of symptoms listed over a period of 4 weeks. It is a dichotomous scale seeking ‘yes’ or ‘no’ answers. Each ‘yes’ response to astern has to be followed by open ended questions. Instructions are provided for CHWs regarding when to interpret an affirmative answer as significant and when not to do so. The actual questions asked are many more than the 20, as each question is to be followed by more questions and paradoxically it no longer remains a self-reporting questionnaire and is vulnerable to rater bias. There is no possibility of recording an answer for which the patient is not sure. The instructions are that when in doubt the CHW should leave it to the supervisor to review1. Availability of a competent supervisor is again not guaranteed inaPHC setting.
Development of Aga Khan University Anxiety and De­pression Scale (AKUADS):
The Aga Khan University Anxiety and Depression Scale (AKUADS) is an indigenous screening scale developed at the Aga Khan University (AKU), Karachi, Pakistan. Karachi is a mega city with a cosmopolitan population, estimated to be around 120 million7. 40% of the inhabitants live in urban squatter settlements7. Aga Khan University (AKU) has established several PHC centers in the uthan squatter settlements of Karachi. AKUADS is in Urdu language which is widely understood and spoken in Pakistan and India. It has been developed from a list of complaints collected by a retrospective file review of symptoms mentioned in Urdu by patients of anxiety and depression coming to the Community Health Center (CHC) of AKU. It is a 25 item questionnaire which includes 12 psychiatric and 13 somatic symptoms. It is a differential scale which is rank ordered for severity and inquiries about the presence and severity of psychiatric and somatic symptoms of anxiety and depression over a period of last 2 weeks. The CHWs are not required to formulate theirjudgement requiring the validity of symptoms as the subject himself is required to provide a direct quantifiable estimate of his/her distress, i.e. whether the symptoms have been present all the time, most of time, some time or have not been there at all, There is a do not know column provided if the patient is unsure of the response. It has been validated in a community setting keeping the psychiatrist’s diagnosis as the gold standard8.
Comparison of Characteristics of AKUADS, SRO and DSMIV:
This paper compares the critenon validity of AKUADS and SRQ with DSM IV and with each other. The criterion indices of sensitivity, specificity, positive predictive value, negative predictive value and overall misclassification rate of both the instruments is also documented.

Table I shows that AKUADS and SRQ criteria for depression are quite similar to DSMIV, neither has stems itlatingto psychomotor symptoms and AKUADS does not inquire about fatigue and guilt. SRQ covers 5 criteria for anxiety and AKUADS covers 6 criteria out of the 8 mentioned inDSM IV (Table II).

DSM IV lists 13 criteria for panic attacks, SRQ covers 4 and AKUADS covers 9 of these criteria (Table III).


Table IV compares the criterion validity of AKUADS and SRQ with DSM IV for somatization. AKUADS and SRQ both do not have any stem addressing sexual symptoms and SRQ do not inquire about pseudoneurological symptoms while AKUADS does.

Table V compares the content validity of AKUADS and SRQ. Thirteen attributes are common to both the instruments. Headaches, appetite, sleep. fear, tremor, wony, digestion, thinking, unhappiness, crying, anhedonia, loss of interest and suicide i.e. 70% of the stems are the same in SRQ and AKUADS. Out of the 7 remaining stems of SRQ, 2 assess the same characteristic ‘tiredness’. The remaining 6 stems of SRQ that are not present in AKUADS are related to difficulty in making decisions, suffering of work, inability to play a useful part in life, feelings of worthlessness and tiredness all related to cognitive and emotional domains. The attributes inquired in AKUADS that are not present in SRQ are, heartburn, nausea, constipation, numbness, tension in neck and shoulders, body aches and frequency of micturition, which are somatic in nature, it is known that most people with mental illness tend to somatise their distress particularly in the developing countries9.
AKUADS has been validated in an urban squatter settlement of Karachi. The interviewers were CHWs who were trained to administer AKUADS verbally to consenting adults who could understand Urdu. The gold standard for establishing concurrent validity was apsychiatrist’s interview. The sample size was 487. Both males and females had a near equal distribution, all decades of respondents between the ages of 16 and 60 were represented. Sampling was continuous i.e. all consenting adults who volunteered forthe interview and met the inclusion criteria for age and language were included. The psychiatrists classified the participants as having anxiety, depression, both anxiety and depression or as normal according to a structured interview schedule based onDSM Ill critena(atthe timeoftheinterviews, DSMlV had not arrived). The salient feature ofAKUADS is that it is not a translation. It has been developed from actual symptoms of patients as reported by them inUrdu and hence, is representative of local symptom complexes. In this population based study comprising of consenting adults AKUADS at a cut off of 19 demonstrated a sensitivity of 74% and a specificity of 81%, a positive predictive value of 61%, negative predictive value of 88% and an overall misclassification rate of 21%8. Buesenberg (1993) has reviewed 28 studies done to establish the reliability and validity of SRQ. Most of the studies reported are not actually population based, the cut off point used is not consistent, varying between 5-12 and the results of criterion indices of sensitivity and specificity reported are widely different, the range for sensitivity is 62.9% - 90% and that for specificity is from 58% - 95.2%. The gold standard used for comparison was either GHQ-l2, ICD-9, CIS depression scale, Hamilton depression scale or the psychiatrist’s perception but in the later event either the sample of low scores interviewed in the second stage was small (Harding et al4) or the psychiatrists were not blind to questionnaire scores (Dhaphale et al)10.
A population based study using SRQ reported by Buesenberg (1993) are, by Rahim and Cederblad(1989)11 ma suburban area of Khartoum, Sudan. The SRQ was administered to only 174 young adults and the cut off score is not mentioned. A positive correlation with the psychiatric assessment (r=0.69) is reported.
In another study carried out by Sen et al12 in 1987 the sample size was 202, out of which only 48 were males, the criteria used were the CIS and the Hamilton scales. At a cutoff of 11/12 a sensitivity of 79%, a specificity of 75%, a positive predictive value of 76% a negative predictive value of 78% and an overall misclassification rate of 23% was reported. Compared to the SRQ- 20, AKUADS has the following advantages. AKUADS has been developed from the actual symptoms of patients and does not represent only the views of experts. It is not a translation and hence is representative of local symptom complexes and local expressions. It has a good content validity as compared to DSM IV criteria for diagnosis of the common neurotic disorders found in the community, i.e. anxiety, depression, panic attacks and somatization. It has been validated in a population based study on a sample of 487 adults as a verbally administered questionnaire by trained CHWs, the criterion being the psychiatrists interviews. At a cut off 19 it has a reasonable sensitivity of 74%, a high specificity of 81%, a positive predictive value of 63%, a high negative predictive value of 88% and a low misclassification rate of 21%. The desirable attributes of a screening questionnaire are a high specificity, a high negative predictive value and a low misclassification rate. If the questionnaire is over inclusive, the non cases are screened out on subsequent clinical evaluation, but if a case is not identified as a case, then there is the likelihood that the CHW will neitherfollow-up that case norrefer. AKUADS isadifferential scale rank ordered for severity, it obtains an estimate of the distress caused by the attribute being assessed from the respondent directly, ona four point scale varying from present all the time, most of the time, some times only arid not present at all. It thus minimizes inter-rater bias and makes establishing inter-rater reliability relatively easy. This makes AKUADS cost effective in terms of time required fortmining of CHWs, intemis of CHWs time in administration and the respondents’ time in answering the questions. SRQ-20 on the other hand is a dichotomous scale which necessitates that the CHW has to assess the actual significance of a symptom by asking further questions even alter receiving an affirmative answer. Dichotomous scales are advocated for dichotomous variables like day and night, or male and female , the attributes of psychiatric morbidity lie on a continuum and the challenge is to be able to assess where the normal emotions and somatic distress change to pathological states. Leaving this decision to the CHWs with limited education and a lot of other responsibilities make SRQ-20 more susceptible to inter-rater bias. The SRQ-20 inquires as to the presence of symptoms in the preceding four weeks. It is advocated that recall bias is exemplified alter a period of two weeks and for this reason AKUADS was designed to inquire about the presence of symptoms during the preceding 2 weeks only. DSM IV also requires the presence of symptoms for 2 weeks to be significant.
Which screening instrument is likely to be most useful in our Setting?
AKUADS not a translation, is representative of local symptom complexes as expressed in urdu, is a differential scale rank ordered for severity and it is easy to train lay but literate interviewers in its use. it is cost effective regarding the time required by the interviewers and respondents and has the desirable attributes of a screening questionnaire, i.e., a higher specificity and a higher negative predictive value as compared to the SRQ. It is suggested that AKUADS should be included in the training manual for the national health workers. It is not suggested that the CHWs would administer AKUADS to all members in the community, but educating CHWs about the common attributes of mental illness would sensitize them to its presence. Then they could selectively administer AKUADS and refer the high scorers to health centers where doctors could evaluate. AKUADS has a potential tobe used by researchers interested in Psychiatric epidemiology and in PHC Psychiatiy in Pakistan, India and in countries that have a large number of Urdu speaking immigrants.

References

1. Buesenberg, M. A user’s guide to the Self Reporting Questionnaire. WHO document. Geneva, WHO, 1993.
2. Goldberg, D.P. Identifying psychiatric illness among general medical patients. Br.Med. J., 1985;291:161-2.
3. Reiger, D.A., Goldberg, ID., Bums, B.J. et al. Specialist/generalist division of responsibility for patients with mental disorders. Arch Gen. Psychiatry, 1982;39:219.224.
4. Harding, T.W.. DeArango, MV., Baltzar. J. eta!. Mental disorders in primary health care: A study of their frequencies and diagnosis in four developing countries. Psychol. Med., 1 980;1 0:231-241.
5. Reiger, DA., Boyd, J.H., Burke, J.D. et al. One month prevalence of mental disorders in the United States. Arch. Gen. Psychiatry, 1988,45:977-979.
6. Paykel, ES. Depression in women. Br. J. Psychiatry, 1991 ;1 58(Suppl. 10): 22-29.
7. The challenge of Urbanization. The Worlds large cities. Department for Economic and Social Information and Policy Analysis: New York. Population Division. United Nations, 1995, pp. 115-118.
8. Ali, B.S. Validation of the Aga Khan University Anxiety and Depression Sca6le. FCPS dissertation, 1996.
9. Ndetei, D.M. and Muhangi, 3. The prevalence and clinical presentation of psychiatric illness in a rural setting in Kenya. Br. J. Psychiatry, 1979;145:269-72.
10. Dhadphale, M. and Ellison, RH. The frequency of mental disorders in the outpatients of two Nyarsza hospitals. Cent. Afr. 1. Med., 1983;29:29-32.
11. Rahim, 3.1, and Cederblad, M. Epidemiology of mental disorders in young adults of a newly urbanized area in Khartoum, Sudan. Br. J.Psychiatry, 1989; 155:44-7.
12. Sen, B. An analysis of the nature ofdepressive phenomena in primary health care utilising multivariate statistical techniques. Acta Psychiatr. Scand., 1987;76:28-32.
13. Streiner, DL. and Norman, G.R. Health measurement scales - A practical guide to their development and use. 1St ed. Oxford, Oxford Press, 1992, pp. 2 1-37.

Journal of the Pakistan Medical Association has agreed to receive and publish manuscripts in accordance with the principles of the following committees: