Table 2.
Studies validating instruments for screening and measurement of depression among African youth
| Reference | Study design | Instrument(s) assessed | Evidence of reliability or validity |
|---|---|---|---|
| Abubakar and Fischer (2012) | A convenience sample of 427 working adults (25-43 years), 108 university students (19-23 years) and 696 secondary school students (14-19 years) in urban Kenya self- administered the English version of the General Health Questionnaire (GHQ-12). |
GHQ-12 | Confirmatory factor analysis was used to test five different models of the GHQ-12. The best-fitting model was a three-dimensional model of anxiety/depression, loss of confidence, and social dysfunction. The multi-dimensionality appeared to be substantively related to negative wording. |
| Adewuya et al. (2006a) | A probability sample of 512 students (15-40 years) in Nigeria self-administered English versions of the 9-item Patient Health Questionnaire (PHQ) and 21-item Beck Depression Inventory (BDI). Research assistants administered the Mini International Neuropsychiatric Interview (MINI) in English to establish the reference criterion of major and minor depressive disorder. |
PHQ-9 BDI-21 |
The internal consistency of the PHQ-9 was 0.85. The PHQ-9 had a statistically significant correlation with the BDI (Spearman’s rho = 0.67, P<0.001). PHQ-9 scores obtained 4 weeks apart had a statistically significant correlation with each other (Spearman’s rho = 0.89, P<0.001). PHQ-9 ≥5 had 0.90 sensitivity and 0.99 specificity for detecting combined major and minor depressive disorder (AUC = 0.991). PHQ-9 ≥10 had 0.85 sensitivity and 0.99 specificity for detecting major depression (AUC = 0.985). |
| Adewuya et al. (2007) | A stratified random sample of 1,095 Nigerian adolescents (13- 18 years) in secondary school completed the 21-item BDI. The entire high-morbidity group (≥ 10) and 10% of those in the low-morbidity group (<10) were administered the Schedule for Affective Disorders and Schizophrenia for School-aged Children-Epidemiological Version 5 (K-SADS-E) by psychiatrists blinded to the BDI scores to establish the reference criterion diagnosis of major depressive disorder (MDD). |
BDI-21 | The internal consistency of the BDI was 0.82. BDI scores obtained 2 weeks apart had a statistically significant correlation (Spearman’s rho = 0.72, P<0.001). BDI ≥18 had 0.91 sensitivity and 0.97 specificity for detecting major depressive disorder (AUC = 0.985). In a separately reported analysis (Adewuya and Ologun, 2006b), significant correlates of depressive symptoms were: parental depression, interpersonal problems, self-esteem, and drinking. |
| Ambaw (2011) | A randomly selected sample of 804 orphans (11-18 years) receiving care at 16 selected orphan support organizations in Addis Ababa, Ethiopia were administered the Amharic version of the 14-item Hospital Anxiety and Depression Scale (HADS) by trained interviewers. |
HADS | Factor analysis revealed two factors (anxiety, depression) that explained 46% of the total variance. In the overall sample, consistency of the HADS depression and anxiety sub-scales were 0.76 and 0.81 respectively. In the subsample of orphans aged 11- 15 years, the internal consistency of the depression and anxiety sub-scales were 0.77 and 0.80 respectively. |
| Bekhet and Zauszniewski (2010) | A convenience sample of 170 adolescents (17-20 years) studying at a nursing school in Egypt self-administered the 8- item Arabic version of the Depression Cognition Scale (DCS). |
DCS | The internal consistency of the DCS was 0.86. Factor analysis confirmed the presence of a single factor. The DCS had a statistically significant positive correlation with a scale measuring alienation (r=0.51, P<0.01). |
| Betancourt et al. (2009a) | A convenience sample of 178 adolescents (14-17 years) and their caregivers in two IDP camps in Gulu were administered the 60-item Acholi Psychosocial Assessment Instrument (APAI) to identify local mental health syndromes, three of which (two tam, par, and kumu) overlap with western concepts of mood disorders. Participants were randomly selected for a second interview, either 1-3 days later to determine test-re-test reliability (N = 30), or by other interviewer to determine inter- rater reliability (N = 19). Caseness was by determined by agreement between both the adolescent and the caregiver. |
APAI | For the 16-item two tam subscale, internal consistency was 0.87, split halves reliability (Spearman-Brown) was 0.88, test-retest reliability was 0.79, and inter-rater reliability was 0.86. For the 13- item kumu subscale, internal consistency was 0.87, split halves reliability was 0.88, test-retest reliability was 0.83, and inter-rater reliability was 0.92. For the 17-item par subscale, internal consistency was 0.84, split halves reliability was 0.83, test-retest reliability was 0.79, and inter-rater reliability was 0.78. Mean subscale scores were greater among adolescents identified as having those syndromes (P<0.001 for each). In a subsequent study of 667 youth, the APAI was refined using item response theory and reconfigured into a shorter, 41-item African Youth Psychosocial Assessment designed for use in assessing mental health among African youth more broadly (Betancourt et al., in press). |
| Betancourt et al. (2009b) | This was a qualitative study of 56 children (10-17 years) and 47 adult key informants living in two IDP camps in Gulu, Uganda. |
Not applicable | Key informants identified three local syndromes that overlap with mood and depressive disorders: two tam (having “lots of thoughts”), kumu (persistent grief and par (having many worries). |
| Betancourt et al. (2011) | A purposive sample of 31 adults and 43 children (10-17 years) in southwestern Rwanda was asked to free-list problems faced by HIV-affected children. A snowball sample of 90 adults (including 10 clinicians) and 38 children participated in in- depth key informant interviews to explore specific local syndromes. |
Not applicable | Participants identified local syndromes that overlap with DSM-IV criteria for dysthymia and major depressive disorder, including guhangayika (constant anxiety/stress), agahinda kenshi (persistent sorrow or sadness), and kwiheba (severe hopelessness). Umishiha (persistent irritability or anger) emerged as the syndrome most heavily influenced by repeated experience of loss and stigma due to HIV/AIDS. |
| Betancourt et al. (2012) | The Center for Epidemiological Studies-Depression scale for Children (CES-DC) was adapted by including parenthetical reminders of the conceptually equivalent Kinyarwanda symptom terms identified in a qualitative study. The modified CES-DC underwent cognitive testing with a convenience sample of 46 children and adolescents. The Pearson correlation coefficient was used to estimate test-retest reliability in a convenience sample of 34 children (10-17 years) who were re- interviewed 1-3 days after initial assessment. The intra-class correlation coefficient was used to estimate inter-rater reliability in a convenience sample of 30 children and adolescents (10-17 years). A purposive sample of 467 children and adolescents (10-17 years) in southeastern Rwanda were administered the modified CES-DC. Psychologists blind to the CES-DC scores administered the MINI for Children and Adolescents (MINIKID) to establish the reference criterion diagnosis for depressive disorder. |
CES-DC | The CES-DC had an internal consistency of 0.86. The Pearson coefficient for test-retest reliability was 0.85. The intra-class correlation within participants was 0.82. CES-DC ≥30 had 0.82 sensitivity and 0.72 specificity for detecting depression (AUC = 0.83). The CES-DC had a statistically significant association with a measure of functional disability (Pearson’s r=0.46; P<0.001). |
| Cherian, Peltzer, and Cherian (1998) | A random sample of 622 grade 11 secondary school students (17-24 years) in Northern Province, South Africa were administered the 20-item Self-Reporting Questionnaire (SRQ) by trainee teachers. |
SRQ | The SRQ had an internal consistency of 0.9. Factor analysis revealed four factors (anxiety/depression, depression, anxiety, and somatic complaints) accounting for 51% of the total variance. |
| El-Missiry et al. (2012) | A probability sample of 602 girls (14-17 years) in secondary schools in Cairo, Egypt self-administered the Arabic version of the Children’s Depression Inventory (CDI). A researcher blind to the CDI scores administered the Structured Clinical Interview for DSM-IV Axis I Diagnosis Research Version, Non-Patient Edition (SCID-I/NP) to establish the reference criterion, a combined diagnosis of major depression, dysthymia, and adjustment disorder. |
CDI | CDI ≥24 had 0.75 sensitivity and 0.98 specificity for detecting depressive disorders. CDI scores had statistically significant associations with poor academic achievement (P<0.001), termination of romantic relationships (P<0.001), a quarrelsome home environment (P<0.001), and negative life events (P=0.01). |
| Ertl et al. (2010) | A random sample of 1,114 war-affected adolescents and young adults (12-25 years) living in IDP camps in Northern Uganda were administered the 15-item HSCL. A randomly selected subset of 68 participants underwent expert validation interviews, 4-18 days after the initial interview, by blinded psychologists who administered the MINI to establish the reference criterion diagnosis for major depressive disorder. |
HSCL-15 | The HSCL-15 had an internal consistency of 0.89. HSCL-15 ≥2.65 had 0.50 sensitivity and 0.83 specificity for detecting major depressive disorder (AUC = 0.76). The widely used cutoff ≥1.75 had 0.86 sensitivity and 0.44 specificity. HSCL-15 scores had statistically significant associations with the Posttraumatic Diagnostic Scale (P<0.001), a locally-derived measure of functional impairment (P<0.001), and suicide risk (P=0.002). |
| Flisher et al. (2012) | A sample of 105 parent/caregiver and child (12-17 years) pairs from a peri-urban South African clinic and community sample participated in the study. Trained research assistants administered the Xhosa version of the Diagnostic Interview Schedule for Children (DISC-IV) and then again two weeks later. |
DISC-IV | Test-retest reliabilities for parent informants were as follows: MDD (κ = 0.662), oppositional defiant disorder (ODD) (κ = 0.662), attention deficit hyperactivity disorder (ADHD) (κ = 0.559), anxiety (κ = 0.448) and agarophobia (κ = 0.789). Test- retest reliabilities youth informants were: MDD (κ = 0.661), ODD (κ = 0.385), ADHD (κ = 0.227), anxiety (κ = 0.145) and agarophobia (κ = 0.579). The test-retest reliabilities of the combined parent-child algorithm lay between the parent and youth findings but only MDD yielded substantial results (κ = 0.662). |
| Ibrahim, Kelly, and Glazebrook (2012) | A probability sample of 988 Egyptian undergraduate university students (16-26 years) self-administered a modified 46-item Arabic version of the Zagazig Depression Scale. |
Zagazig | The Zagazig Depression Scale had an internal consistency of 0.90 and a split-half reliability of 0.89. Internal consistency of the subscales ranged between 0.64-0.79. Factor analysis revealed an 11-factor solution that explained 62% of the variance: depression, suicidal ideation, guilty feelings, insomnia, agitation/ hypochondriasis, sleep maintenance, cognitive impairment, diminished energy, weight loss, and sexual symptoms. |
| Kebede et al. (2000) | A purposive sample of 255 children and adolescents (6-18 years) was obtained from the inpatient and outpatient wards of a psychiatric hospital, a school for mentally disabled children, and the surrounding community in Addis Ababa, Ethiopia. For children aged 6-11 years, the parent or primary caregiver was interviewed. One trained lay interviewer and one clinician participated in each interview; one administered the Revised Diagnostic Interview for Children and Adolescents (DICA-R) in Amharic, while both coded the responses independently. |
DICA-R | The kappa statistic for agreement on the DSM-III diagnosis of major depressive episode was 0.90. |
| Lowenthal et al. (2011) | A convenience sample of 509 HIV-positive children and adolescents (8-16 years) in two outpatient settings in Botswana were administered the Setswana version of the 35-item Pediatric Symptom Checklist-Youth Version (PSC-Y) and the CDI, while one parent/guardian was administered the PSC (i.e., adult version). The reference criterion for the PSC was “parent and clinic staff reports of concern about the child”, while the reference criterion for the PSC-Y was depressive disorder as diagnosed by the CDI. |
PSC CDI |
Internal consistency was 0.87 for the PSC-35 and 0.86 for the PSC-35-Y. PSC-35 ≥20 had 0.62 sensitivity and 0.86 specificity for detecting concern about the child (AUC = 0.85). PSC-35-Y ≥20 had 0.64 sensitivity and 0.88 specificity for detecting depression (AUC = 0.81). |
| Mels et al. (2010) | Focus group interviews with 66 key informants in the Democratic Republic of Congo were used to derive a list of locally observed symptoms. The 37-item HSCL was modified by removing two items that did not emerge in the qualitative interviews (“feeling trapped”, “using sleeping pills), condensing two items into a single item (“drinking alcohol”) and adding four frequently mentioned local idioms (“overburdened by worries”, ”talking to oneself”, “not interested in school”, “not following the rules”). The Swahili or Congolese French versions of the modified 38-item HSCL were administered to 1,046 adolescents (13-21 years) in a school-based survey. |
HSCL-38 | One item (“loss of sexual interes”) was excluded from analysis due to a high proportion of missing values, especially among participants in Catholic schools. The French version of the HSCL- 38 had an internal consistency of 0.90, with coefficients ranging from 0.76-0.89 on the four subscales (internalizing, depression, anxiety, externalizing). The Swahili version of the HSCL-38 had an internal consistency of 0.91, with a subscale coefficients ranging from 0.66-0.91. Exploratory factor analysis revealed two factors broadly categorized an internalizing and externalizing problems. The modified HSCL-38 total score had statistically significant associations with the Impact of Event Scale-Revised and its possible subscale scores, the Adolescent Complex Emergency Exposure Scale, and subjective psychological wellbeing(P<0.01). |
| Pretorius (1991) | A sample of 450 undergraduate psychology students (19-53 years) in South Africa self-administered the CES-D. |
CES-D | The CES-D had an internal consistency of 0.90. Factor analysis revealed a four-factor solution. The internal consistencies of the factor subscales were as follows: depressed affect (0.85), somatic- retarded activity (0.71), positive affect (0.73), and interpersonal relations (0.70). The 57-item Life Experiences Survey had a statistically significant association with the CES-D total score (Pearson’s r=0.21, P<0.05), as well as with three of the factors: depressed affect (r=0.18, P<0.01), somatic-retarded activity (r=0.26, P<0.01) and interpersonal relations (r=0.15, p <0.01). |
| Pretorius (1998) | A sample of 213 undergraduate psychology students (19-53 years) in South Africa self-administered the CES-D. |
CES-D | The CES-D had an internal consistency of 0.90. The CES-D had a statistically significant association with the Life Experiences Survey-Negative (Pearson’s r=0.19, P<0.05). |
| Rothon et al. (2011) | A convenience sample of 237 adolescents (14-15 years) in Cape Town, South Africa self-administered the Afrikaans or isiXhosa versions of the 13-item Short Moods and Feelings Questionnaire (SMFQ) on two occasions one week apart. |
SMFQ | The SMFQ had an internal consistency of 0.85. The correlation between SMFQ scores one week apart was 0.32 (P-value not reported). |
| Sharp et al. (2011) | A focus group interview was held in English with 10 Sesotho- speaking clinicians (five clinical psychologists, five licensed social workers, and one clinical psychology intern) in Bloemfontein, South Africa. Data were grouped into broad thematic areas. |
DISC-IV | Participants identified a number of cultural considerations that could affect the utility of the DISC in the Sesotho context. These included its rigid response structure, “Americanisms,” problems in interpretation due to widespread socioeconomic adversity, language problems, and cultural norms about psychiatric symptoms, the expression of emotion and family structure. |
| Traube et al. (2010) | After a local work group translated the CDI, field workers provided further input to modify three scale items. The CDI was then administered to four groups of children and adolescents in southwestern Tanzania (3-19 years), including orphans living in a local residential facility vs. those who were not. |
CDI | The CDI had an internal consistency of 0.67, and the subscale reliability coefficients were lower: negative mood (0.31), interpersonal problems (0.24), ineffectiveness (0.11), anhedonia (0.58), and negative self-esteem (0.34). Spearman-Brown split half reliability was 0.66. The proportion of orphans with high-risk symptoms was lower among residents of the orphan facility compared to orphans not living in the facility (14.3% vs. 47.1%). |
| Ward et al. (2003) | A convenience sample of 104 students (12-18 years) in Cape Town, South Africa self-administered the 21-item BDI in English, Afrikaans, or Xhosa. Participants completed the questionnaire again 10-14 days after the initial self- administration. |
BDI-21 | Internal consistency of the BDI was 0.86. Test-retest reliability was described as “good” but the estimated kappa coefficients were not reported. |
ADHD = Attention Deficit Hyperactivity Disorder; AUC = area under the receiver-operating characteristics curve; BDI = Beck Depression Inventory; CDI = Child Depression Inventory; CES-DC = Center for Epidemiological Studies-Depression scale for Children; DCS = Depression Cognition Scale; DICA-R = Revised Diagnostic Interview for Children and Adolescents; DISC-IV = Diagnostic Interview Schedule for Children; DSM = Diagnostic and Statistical Manual of Mental Disorders; GHQ = General Health Questionnaire; HADS = Hospital Anxiety and Depression Scale; HSCL = Hopkins Symptom Checklist; IDP = internally displaced persons; K-SADS-E = Schedule for Affective Disorders and Schizophrenia for School-aged Children-Epidemiological Version 5; MDD = Major Depressive Disorder; MINI = Mini International Neuropsychiatric Interview; MINIKID = Mini International Neuropsychiatric Interview for Children and Adolescents; ODD = Oppositional Defiant Disorder; PHQ = Patient Health Questionnaire; PSC-Y = Pediatric Symptom Checklist-Youth version; SCID-I/NP = Structured Clinical interview for DSM-IV Axis I Diagnosis Research Version, Non-Patient Edition; SMFQ = Short Moods and Feelings Questionnaire; SRQ = Self-Reporting Questionnaire