|Year : 2012 | Volume
| Issue : 4 | Page : 203-211
A comparison of two screening instruments in detecting psychiatric morbidity in a Nigerian pediatric primary care service: Assessing clinical suitability and applicability
Mosunmola F Tunde-Ayinmode1, Babatunde A Ayinmode2, Olushola A Adegunloye1, Olatunji A Abiodun1
1 Department of Behavioural Sciences, University of Ilorin/University of Ilorin Teaching Hospital Ilorin, Nigeria
2 Department of Family Medicine, University of Ilorin Teaching Hospital Ilorin, Nigeria
|Date of Web Publication||24-Oct-2012|
Mosunmola F Tunde-Ayinmode
Department of Behavioural Sciences, University of Ilorin, Ilorin
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Background: To improve poor attendance and underutilization of the child and adolescent psychiatric service in the University of Ilorin Teaching Hospital, Ilorin, Nigeria, we compared a child behavior questionnaire (CBQ) with reporting questionnaire for children (RQC) for use in our primary care unit where the bulk of referrals come from to determine which is more applicable.
Methods: This was a cross-sectional two-stage study on the prevalence and pattern of psychiatric disorders in children attending the primary care unit of University of Ilorin Teaching Hospital, over a 6-month period. CBQ and RQC were completed by the mothers of 350 children aged 7-14 years in the first stage; in the second, a stratified subsample of 157 children based on scores on CBQ were interviewed using the children's version of the schedule for affective disorders and schizophrenia (Kiddie-SADS-PL). The receiver operating characteristics (ROC) analysis was carried out to determine the screening properties of CBQ and RQC.
Results: The optimal cutoff score for CBQ and RQC were 7 and 1, respectively. Sensitivity, specificity, and misclassification rates at the cutoff for CBQ was 0.8 (80%); 0 96 (96%); 0.083 (8.3%), and RQC was 0.90 (90%); 0.78 (78%); 0.19 (19%). The discriminating ability of CBQ indicated by the area under the curve (AUC) in the ROC was 0.93 while RQC was 0.88.
Conclusion: Even with the evident marginal superior discriminating ability of CBQ in our study in primary care, RQC has the advantage of brevity and ease of application for workers at this level. In the child and adolescent clinic CBQ may still be preferred.
| Abstract in French|| |
Contexte: Pour améliorer le peu d'affluence et la sous-utilisation de l'enfant et l'adolescent service psychiatrique de l'hôpital universitaire de l'Université d'Ilorin, Ilorin, Nigeria, nous avons comparé un questionnaire de comportement des enfants (CBQ) par rapport au questionnaire pour les enfants (RQC) pour utilisation dans notre unité de soins de première ligne où la majeure partie des références viennent de déterminer qui est le plus applicable.
Méthodes: C'est une étude transversale de deux étages sur la prévalence et le patron des troubles psychiatriques chez les enfants qui fréquentent l'unité de soins de première ligne de l'Université d'Ilorin Teaching Hospital, sur une période de 6 mois. CBQ et RQC ont été achevés par les mères de 350 enfants âgés de 7 à 14 ans dans la première étape ; dans le second, un sous-échantillon stratifié de 157 enfants basé sur les scores sur CBQ ont été interrogés à l'aide de la version pour les enfants de l'annexe pour les troubles affectifs et la schizophrénie (Kiddie-SADS-PL). Le récepteur d'exploitation de l'analyse des caractéristiques (ROC) a été effectué afin de déterminer les propriétés de contrôle de CBQ et RQC.
Résultats: Note coupure de optimale pour CBQ et RQC était de 7 et 1, respectivement. Taux de classification erronée, la spécificité et la sensibilité à la coupure pour CBQ était de 0,8 (80%); 0 96 (96%); 0,083 (8,3%) et RQC 0,90 (90%); de 0,78 (78%); de 0,19 (19%). La capacité de discrimination des CBQ indiqué par l'aire sous la courbe (AUC) dans le ROC était 0,93 RQC alors 0,88.
Conclusion: Même l'évidente marginal supérieur discriminant capacité du CBQ dans notre étude dans les soins primaires, RQC a l'avantage de la concision et la facilité d'application pour les travailleurs à ce niveau. À la clinique des enfants et des adolescents CBQ peut toujours être préférée.
Mots clés: Morbidité psychiatrique, dépistage, instrument
Keywords: Psychiatric morbidity, screening, instrument
|How to cite this article:|
Tunde-Ayinmode MF, Ayinmode BA, Adegunloye OA, Abiodun OA. A comparison of two screening instruments in detecting psychiatric morbidity in a Nigerian pediatric primary care service: Assessing clinical suitability and applicability. Ann Afr Med 2012;11:203-11
|How to cite this URL:|
Tunde-Ayinmode MF, Ayinmode BA, Adegunloye OA, Abiodun OA. A comparison of two screening instruments in detecting psychiatric morbidity in a Nigerian pediatric primary care service: Assessing clinical suitability and applicability. Ann Afr Med [serial online] 2012 [cited 2017 Nov 22];11:203-11. Available from: http://www.annalsafrmed.org/text.asp?2012/11/4/203/102849
| Introduction|| |
Mental health of children in the hospital setting is gradually receiving attention in Nigeria. This may be related to the increasing number of child and adolescent psychiatric facilities being established in the country. , Primary care is the main source of referral to these clinics. Child psychiatric problems are known to be prevalent in primary care settings in Nigeria ,,, but research works in this area are still too few.
Early identification and management are still major problems. , There is evidence to suggest that many primary care physicians may either be too busy or lacked the requisite knowledge and attitude to detect children with mental health problems.  Many of the child and adolescent psychiatric units being set up in Nigeria need referrals from primary care or be faced with poor attendance and underutilization. , Screening instruments are required in primary care to make case detection easy and simple. Two screening instruments developed for the purpose of quick detection of cases are child behavior questionnaire (CBQ) and reporting questionnaire for children (RQC).
CBQ is an instrument developed by Rutter  to screen for child psychiatric disorders in children and adolescent populations and has been validated and found to be acceptable, applicable, and reliable in many countries of the world including Nigeria. ,,, CBQ has two versions, parents (Scale A) and teachers (Scale B). The teachers' version consists of 26 items and parents version 31 items, investigating the child's behavior at school and at home respectively during the previous 12 months. Both scales have undergone modification a few times and the latest forms are referred to as Scale A2 and Scale B2. 
With Scale A2 parents are asked to indicate the extent to which the statement applies to the child, the frequency of occurrence of the behavior or the degree of its severity. Each item is rated 0-2 (0; does not apply, 1; apply somewhat, 2; certainly apply). These produce a total score within the range of 0-62. It has a neurotic subscale consisting of items 3, 13, 19, and 28 and antisocial subscale consisting of items 11, 16, 26, 30, and 31.  According to this original work, selection of children into clinical diagnostic categories is a two-stage procedure: Firstly, children with a total score of 13 (which is the optimal cutoff point for the presence of probable psychiatric morbidity) or more are designated as having probable psychiatric disorder or classified as cases. Secondly, children whose neurotic sub-score exceed the antisocial are designated as "neurotic"; those with antisocial sub-score exceeding the neurotic are designated "antisocial"; while those whose sub-scores are equal are classified as having undesignated (mixed) disorder. 
The reporting questionnaire for children (RQC) is a 10-item questionnaire developed by the WHO for use specifically in primary health care setting to screen for speech, neurotic, conduct, and learning disorders;  it has been used extensively in developing countries. , Each item on the questionnaire carries a score of 0 or 1 with a cutoff point of 1. The RQC has been found to be a highly sensitive and specific instrument, a high score (that is 1 and above) suggest that the child is having a psychiatric problem.  In the validation study by Omigbodun et al.  aforementioned RQC was found to have identical ability with CBQ at differentiating children with specific disorders from those without. The optimal threshold in their study was 1. At this threshold sensitivity was 0.73 and specificity was 0.61. The misclassification rate was 0.35. Their result was similar to the original validation studies of the questionnaire by Giel et al. 
In the healthcare system primary care sees and treats the largest number of patients and thus generates the highest number of referrals. Our recently established child and adolescent psychiatric unit (CAMH) hampered by inadequate referral and underutilization despite sensitization efforts was deemed to receive most of its referrals from the primary care unit under study. The objective of this study is to assess comparatively the clinical and screening applicability and suitability of two instruments (CBQ and RQC) in our primary care unit as part of an intervention strategy of involving primary care physicians in early detection, management, and referral of child psychiatric cases. It was also to serve for their sensitization and training.
| Materials and Methods|| |
The study took place at the primary care unit that is Family Medicine Department of University of Ilorin Teaching Hospital (UITH), Ilorin Nigeria. It was part of a larger investigation to assess the prevalence and pattern of psychiatric morbidity of children and their mothers attending the primary care unit of the hospital. Information gathered by this investigation was to assist in dealing with the problem of poor attendance at our recently setup child and adolescent psychiatric unit. A two-stage screening procedure was employed for the study population consisting of children aged 7-14 years and their accompanying mothers attending our hospital. In the first stage the 350 pair children and mothers were assessed for the presence of probable psychiatric morbidity. The children were screened with CBQ and RQC screening instruments and their mothers with GHQ-12; because of the generally low level of literacy in the population the screening instruments were read to these mothers in English or Yoruba the local language of the study area; the Yoruba version of the instruments produced by a concordance driven twofold vice versa translation from English to Yoruba by independent proficient translators (back translation) are available and have been used in this locality.  Trained research assistants administered the instruments and a semi-structured questionnaire was used for data on sociodemographic characteristics.
In the first stage we used the score of 7 as a cutoff on CBQ as suggested by Omigbodun et al. in their validation study  which gave the best trade-off between a high sensitivity and a low false-positive rate; the sensitivity was 0.61, specificity 0.74, and misclassification rate 0.29. Their study and ours are similar in many respects: first both were conducted in an urban primary care setting, the Yoruba ethnic group was predominant, and Islam and Christianity were the religions predominant in both settings.
The second stage was designed to identify children with DSM IV psychiatric diagnosis; here we selected a stratified subsample of 157 children consisting of all those who were CBQ screened positive (scored≥7) and 30% (one of every three) of those who were screened negative (scored<7) selected by systematic random sampling, included to enhance the potential of identifying as many DSM IV cases as possible in the study population. They were interviewed jointly with their mothers using the children's version of the schedule for affective disorders and schizophrenia (Kiddie-SADS-PL).  The interviews were carried out by three senior registrars in psychiatry who were unaware of the scores of the children in stage 1.
Diagnoses were assigned using DSM IV by the interviewing senior registrars. Where a child had multiple diagnoses, the most appropriate single diagnosis chosen for the purpose of the analysis was the one highest in the order of an arbitrary hierarchy  in which mental retardation was first followed by ADHD, conduct disorders, depression, anxiety-related disorders, and enuresis in that order.
The study assessed the validity coefficients of RQC and CBQ in those children assigned a DSM IV diagnosis to compare the screening ability of the two instruments.The ability of the two instruments to discriminate between cases(presence of DSM IV diagnosis) and noncases (absence of DSM IV diagnosis) was assessed by the application of the receiver operating characteristics (ROC) analysis. To determine the validity coefficient for every possible threshold score of the screening instruments a 2 by 2 table of cases identified in those who scored the threshold being considered (i.e., high scorers) and noncases identified in those who scored below (i.e., low scorers) was derived. Sensitivity, specificity, misclassification rate, positive-predictive value, negative-predictive values, and false positive rates were calculated from the contingency tables.
An ROC curve is obtained by plotting sensitivity against the false-positive rate (1-specificity) for all possible cutoff points of the screening instrument. The area under the ROC curve (AUC) is an index of the discriminating ability of the instruments; in the present study SPSS  was used to determine AUC. The ROC curve has also been used in many studies to determine the optimal cutoff score for instruments. ,, Also the point on the ROC curve of an instrument that is furthest from the diagonal is suggested to be the best cutoff for that instrument; this is because at this point, there are equal rates of false negatives and false positives. ,
We also assessed the case or noncase differentiating ability of individual items and group of items of the instruments. The frequency of positive items rating for cases and noncases was determined, compared, and statistical differences determined by the Chi-square statistics. In situations where the Chi-square statistics was not valid because the frequency was too low a Fishers exact probability test was used instead.
| Results|| |
The Chi-square test for trends did not show any significant difference between the children in the first and second stages with regard to sociodemographic characteristics [Table 1]. Also there was no significant difference between the mean age of the children in stage 1 (9.75 SD=2.11) and those in stage 2 (9.78; SD=2.14); P=0.899. The distribution of RQC and CBQ scores in the two stages was assessed and found to be similar; all score points in the first were represented in the second. All put together, the two stages are comparable and a fair representation of each other with good degree of internal consistency.
|Table 1: Comparison of sociodemographic characteristics of children in the first and second stages|
Click here to view
Prevalence of psychiatric morbidity
There were 40 children identified on Kiddie- SADS- PL to have a DSM IV diagnosis; on the basis of this the overall prevalence of psychiatric morbidity in the children was 11.4%. The order of frequency of occurrence was enuresis (6%); conduct disorders (1.5%); ADHD (1.3%); anxiety disorders (1.0%) depression (0.5%), and mental retardation (0.5%). The overall prevalence of probable psychiatric morbidity in the children using RQC was 30.9% (108/350); that of CBQ was 10.29 (36/350).
The best tradeoff between a high sensitivity and a low false positive rate (1-specificity) on CBQ was at the threshold of 7. At this threshold sensitivity was 0.8 (80%) and specificity was 0 96 (96%) and misclassification rate was 0.083 (8.3%). For RQC the optimal threshold was 1; at this threshold sensitivity 0.90 (90%), specificity 0.78 (78%), and misclassification rate was 0.19 (19%) [Table 2], [Table 3] and [Table 4].
|Table 2: Validity coefficient of CBQ at cutoff of 7 and RQC at cutoff of 1 in the study|
Click here to view
|Table 3: Validity coefficient of CBQ at a threshold score of 4--10 for all the children|
Click here to view
|Table 4: Validity coefficient of RQC at a threshold score of 1--3 for all the children|
Click here to view
For CBQ the discriminating ability as indicated by the area under the curve was 0.93; for RQC it was 0.88. The greatest perpendicular distance from the diagonal on the ROC curve for CBQ was at the threshold of 7 and for RQC it was at threshold score of 1 [Figure 1] and [Figure 2].
|Figure 1: The Receiver Operating Characteristic curve for the Child Behaviour Questionnaire (CBQ), plotted point for CBQ scores 4-10 is asterisked (see table 3)|
Click here to view
|Figure 2: The Receiving Operating Characteristic for the Reporting Questionnaire for Children (RQC), plotted points for RQC 1-3 is asterisked (see table 4)|
Click here to view
Item analysis is meant to assess what individual or group of factors differentiate cases from noncases. For CBQ, 16 of the 31 items were significantly more frequent in cases, 4 more frequent in noncases; 6 were more frequent in noncases but not statistically significantly so; the other items were equally frequent and not significant [Table 5]. For RQC 5 of the 10 items were significantly more frequent in cases; 3 in noncases but not significantly so; the 2 remaining (having a fit and stealing) items were equally frequent [Table 6].
|Table 5: Frequency of positive items on CBQ: Comparison of cases and noncases for the second-stage sample (N=157)|
Click here to view
|Table 6: Frequency of positive items on RQC: Comparison of cases and noncases for the second-stage sample (N=157)|
Click here to view
| Discussion|| |
In our child and adolescent psychiatric unit we routinely used CBQ in screening and evaluating the children. This clinic is usually poorly attended and underutilized despite our sensitization efforts. Most of the referrals came from the hospital's primary care unit. Child psychiatric problems are putatively prevalent in primary care settings; the problem here could be under detection and inadequate attention. Screening instruments as a tool for quick and easy detection of child psychiatric problems in primary care settings especially in developing countries have been previously researched and recommended. ,,
We therefore compared the psychometric properties of CBQ and RQC, two common, well-researched instruments to assess their suitability and applicability in our primary care setting. The pattern of discriminatory power as evident in validity coefficients and ROC analysis was consistent and similar to studies in our environment. ,, The two instruments were quite similar in their discriminating abilities for the presence or absence of at least one DSM IV disorder; however subtle differences found could influence clinical suitability and applicability in our setting.
In this study a cutoff of 7 was used in selecting the patients for the second stage based on the experience from a local validation study in a culturally similar environment.  This study had suggested that for primary care setting this cutoff was optimal, it offered best tradeoff between false positive and false negative rates, and higher cutoff would most likely produced higher level of false negatives. It is important to state, however, that many factors are reported to influence cutoff scores of instruments. Some of these include sex, age, clinical state of subjects, mode of selection of subjects, structure of diagnostic interview, and cultural factors.  The original validation study  used children attending a psychiatric clinic and required a very high cutoff of 13 to discriminate cases from noncases; but a primary care clinic with undifferentiated cases is likely to require lower cutoff to obviate high false negative rates.
Since the choice of optimal cutoff is highly dependent on the distribution of scores in the study sample, in samples with a high level of psychopathology higher cutoff point will be required to identify optimal number of cases.  Children presenting in a primary care clinic would have a less number of children with severe psychopathology because of dilution from other cases thus greater overlapping of scores, therefore requiring lower scores. The fact that normal children were contrasted with children attending a well-known psychiatric clinic was the reason given for the comparatively high cutoff of 13 used in the original validation study. ,
We tried to determine a new cutoff score using the validity coefficient for a range of scores on the two instruments and still got 7 for CBQ and 1 for RQC which is identical to the local study by Omigbodun et al. The similarity between our study and theirs may be due to the fact that sample characteristic were similar in many respects. First the children age group 7-14 years was used; gender ratio (51:49) was also similar. Similarity in cultural and religious background could have contributed. Lastly Kiddie-SADS semi-structured questionnaire was used in both studies to make DSM IV diagnosis in the children. Although, there may be slight methodological differences but they were unlikely to affect the cutoff point.
There were 40 children identified to have at least one DSM IV diagnosis in this study. The detailed report of prevalence, risk factors, and functional status of the children is being presented elsewhere. Summing up, 11.4% of the 350 children had at least a DSM IV diagnosis this is lower than 19.6% recorded in the study by Omigbodun et al.  that guided our initial cutoff. Comparing our study with theirs there was a remarkable trend in sensitivity and specificity pattern; in both, RQC appeared to be more sensitive than CBQ, recognizing 90% and 73% of cases, respectively, compared to CBQ 80% and 61%, respectively. Conversely, CBQ appears to be more specific (recognizing noncases) 96% and 74%, respectively compare to RQC 78% and 61% respectively. Furthermore, misclassification rates for RQC were higher (19% and 35% respectively) compared to CBQ (8% and 29% respectively). It can be observed herein that we recorded higher sensitivity and specificity rates and lower misclassification rates compared to the other study  on the two instruments. It is important to mention that 10 (arbitrarily chosen) was the cutoff used to select children into the second stage in their study  while we used 7 (based on the final outcome of Omigbodun et al. study). This could also have contributed to the differences in rates, but further study is needed to reach a more valid conclusion since our sample size need improvement.
In further comparing the ability of the two instruments, by ROC analysis, the area under the curve for the 31-item CBQ was larger than that of RQC, suggesting that it had a higher discriminating ability across the total spectrum of morbidity. This is consistent with the pattern in the validity coefficient analysis (very high specificity, lower misclassification rates, lower false positive rates, and higher positive predictive rates) but contrast the study by Omigbodun et al.  in which the AUC was identical for the two instruments.
The first five most frequent positive items of the 31 items on CBQ in both cases in decreasing order are "bedwetting"; "very restless"; "headache"; "stomach ache /vomiting"; and "temper tantrums." For noncases are "headache"; "very restless"; "stomach ache /vomiting"; "temper tantrums" and "bedwetting"; "Headache"; and "stomach ache/vomiting" are common somatic symptoms often reported by mothers and children attending primary care clinic. This may explain why it was the most common among cases and noncases. "Bedwetting ;" and "temper tantrums" were more frequent in cases; restlessness which was the second most frequent item was almost equally distributed in cases and noncases since it could be a feature of both physical and psychological disorders.
On RQC the five most frequent items in both cases and noncases combined are as follows: "bedwetting" ;"frequent headache"; "fits/fall for no reason" "backward and slow to learn" and "sacred/nervous for no reason." Expectedly enuresis was the most common in cases while frequent headache was the most frequent in noncases. The trend where physical symptoms items were more frequent in noncases and psychological symptoms items in cases was also observed with RQC. This seems to ascertain the content validity of the two instruments in our locality; the content of the scale is representative of what is being tested, i.e., its test representativeness is good.
The comparison of the two instruments on their ability to differentiate children into subcategories or subscales was not carried out. This was partly because the ability to differentiate between conduct and emotional disorders was not evident with RQC as was previously reported elsewhere,  making comparison difficult. In addition the items representative of the sub classifications in CBQ , and RQC  were too infrequent to allow for any meaningful analysis of category-based discriminating abilities of the instruments. Although these were limitations, it may not be particularly important in the primary care setting where detection, not elaborate diagnostic grouping takes priority. 
| Conclusions|| |
The two instruments compared, discriminated children with one or more DSM IV disorders from those without in our primary care setting. The CBQ had higher specificity, lower false positive rates, higher positive predictive rates, and half the misclassification of RQC. On the basis of this it appears that CBQ has a marginally superior discriminating power despite RQC's higher sensitivity. RQC with its higher sensitivity and negative predictive value may be more appropriate for screening in the primary care setting; aside it is shorter and likely to be easier to administer by primary care worker. The CBQ could be more appropriate in the child and adolescent clinics where it is already in use.
| References|| |
|1.||Shaw RJ, Bassols AM, Berelowitz M, Eapen V, Frank R, Gao H. Chapter 1. Paediatric Psychosomatic Medicine. In: Shaw RJ, DeMaso DR, editors. Textbook of Paediatric Psychosomatic Medicine. USA: America Psychiatric Publishing Inco.; 2010. p. 15-6. |
|2.||Oshodi YO, Erinfolami AR. Presentation and follow-up patterns of child and adolescent patients attending a teaching hospital psychiatric clinic in Lagos, Nigeria. Niger J Psychiatry 2010;8. |
|3.||Abiodun OA. Emotional illness in a paediatric population in Nigeria. J Trop Pediatr 1993;39:49-51. |
|4.||Adelekan ML, Ndom RJ, Ekpo M, Oluboka O. Epidemiology of child behavioural disorders in Ilorin, Nigeria -Findings from parental reports. West Afr J Med 1999;18:39-48. |
|5.||Gureje O, Omigbodun OO, Grater R, Acha RA, Ikuesan BA, Morris J. Psychiatric disorders in a paediaric primary clinic. Br J Psychiatry 1994;165:527-30. |
|6.||Ohaeri JU, Odejide OA, Gureje O, Olatawura MO. Psychiatry morbidity among children attending health facilities in primary health care in a rural Nigeria community. Niger Med J 1994;26:93-5. |
|7.||Adelekan ML. Psychiatric disorders of children and young adults in developing countries. In: Tantam D, Appleby L, Duncan A, editors. Psychiatry for the developing world. London: Imprints of Royal College of Psychiatrists; 1995. p. 311-40. |
|8.||Olatawura MO. Mental health services for children in African region. Int J Ment Health 1978;7:34-8. |
|9.||Bushell J, Mcleod D, Dowell A, Salmond C, Ramage S, Collings S, et al.; MaGple (Mental Health and General Practice Investigation) Research Group. Do patients want to disclose psychological problems to GPs? Fam Pract 2005;22:631-7. |
|10.||Makanjuola JD. Paediatric psychiatry at the neuro- psychiatric hospital Aro, Abeokuta, Nigeria - A five year review. West Afr J Med 1987;6:185-91. |
|11.||Olatawura MO, Odejide AO. Child psychiatric disorders in Ibadan. Niger J Paediatr 1976;3:9-14. |
|12.||Rutter M, Tifard J, Whitemore KL. Education health behaviour. London: Longmans; 1970. |
|13.||Rutter M, Yule W, Berger M, Yule B, Morton J, Bagley C. Children of West Indian Immigrants-1. Rates of behavioural deviance and of psychiatric disorder. J Child Psychol Psychiatry 1974;15:241-62. |
|14.||Boyle MH, Jones SC. Selecting measures of emotional and behavioural disorders of childhood for use in general populations. J Child Psychol Psychiatry 1985;26:137-59. |
|15.||Venables PH, Fletcher RP, Dalais JC, Mitchell DA, Schulsinger F, Mednick SA. Factor structure of the Rutter children's behaviour questionnaire in a primary school population in a developing country. J Child Psychol Psychiatry 1983;24:213-22. |
|16.||Minde KK. Children in Uganda: Rates of behavioural deviations and psychiatric disorders in various schools and clinic population. J Child Psychol Psychiatry 1977;18:23-37. |
|17.||Giel R, DeArango MV, Climet CE, Harding TW, Ibrahim HA, Ladrido-Ignacio L, et al. Childhood mental disorders in primary health care; Results of observations in four developing countries. A report from the WHO collaborative study on strategies for extending mental health care. Pediatrics 1981;68:677-83. |
|18.||Omigbodun O, Gureje O, Gater R, Ikuesan B, Adebayo E. Psychiatric morbidity in a Nigerian paediatric primary care service: A comparison of two screening instruments. Soc Psychiatry Psychiatr Epidemiol 1996;34:186-93. |
|19.||Abiodun OA. A validity study of the Hospital Anxiety and Depression Scale in general hospital units and a community sample in Nigeria. Br J Psychiatry 1994;165:669-72. |
|20.||Kaufman J, Birmaher B, Brent D, Rao U, Ryan N. Kiddie- SADS-present and Lifetime version (K-SADS-PL). Pittsburgh: Department of Psychiatry, University of Pittsburgh School of Medicine; 1996. |
|21.||Gureje O, Omigbodun OO. Children with mental disorders in primary care: Functional status and risk factors. Acta Psychiatr Scand 1996;92:310-4. |
|22.||Statistical Package for Social Sciences (SPSS) Version 15.0.1. Chicago Ill: SPSS Inc.; 2006. |
|23.||Frombonne I. The use of questionnaire in child psychiatry research: Measuring their performance and choosing an optimal cut-off. J Child Psychol Psychiatry 1996;32:677- 24. |
|24.||Mari I, Williams P. A comparison of validity of two psychiatric screening questionnaires (GHQ 12 and SRQ- 20) in Brazil using Relative Operating Characteristics (ROC) analysis. Psychology 1985;15:651-9. |
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6]