Translating the Semi-Structured Assessment for Drug Dependence and Alcoholism in the Western Pacific: Rationale, Study Design and Reliability of Alcohol Dependence

Abstract

Aims: The aims of this study were to develop a bilingual version of the Semi-Structured Assessment for Drug Dependence and Alcoholism (SSADDA) in English and Samoan and determine the reliability of assessments of alcohol dependence in American Samoa. Methods: The study consisted of development and reliability-testing phases. In the development phase, the SSADDA alcohol module was translated and the translation was evaluated through cognitive interviews. In the reliability-testing phase, the bilingual SSADDA was administered to 40 ethnic Samoans, including a sub-sample of 26 individuals who were retested. Results: Cognitive interviews indicated the initial translation was culturally and linguistically appropriate except items pertaining to alcohol tolerance, which were modified to reflect Samoan concepts. SSADDA reliability testing indicated diagnoses of DSM-III-R and DSM-IV alcohol dependence were reliable. Reliability varied by language of administration. Conclusion: The English/Samoan version of the SSADDA is appropriate for the diagnosis of DSM-III-R alcohol dependence, which may be useful in advancing research and public health efforts to address alcohol problems in American Samoa and the Western Pacific. The translation methods may inform researchers translating diagnostic and assessment tools into different languages and cultures.

INTRODUCTION

Heavy drinking is an important health risk factor in the Western Pacific Region, including the U.S. Territory of American Samoa (World Health Organization (WHO), 2002; Lim et al., 2012). Despite the burden of disease due to alcohol in the region, there currently exists no formal alcoholism treatment system in American Samoa. There is also a lack of information on alcohol-related diagnoses and a paucity of research about drinking behaviors.

The aims of this study were to develop a linguistically and culturally appropriate bilingual version of the Semi-Structured Assessment of Genetics for Drug Dependence and Alcoholism (SSADDA) in Samoan and English and determine the diagnostic reliability of the bilingual version. Because the attitudes of different societies to alcohol use have been found to affect the criteria used for diagnostic decision-making (Room, 1996) and the discrepancy between clinical and cultural definitions of behavioral health problems (Held et al., 2010), the translation of complex diagnostic assessments from one language to another may require linguistic and cultural adaptation of concepts in addition to psychometric accuracy of literally translated items (Geisinger, 1994; Hambleton, 1994; van de Vijver and Hambleton, 1996). Achieving cultural equivalency of certain concepts has been identified as an issue in translating research documents into Samoan, and a collaborative approach to Samoan translation has been recommended (Siaki, 2011).

Psychiatric diagnostic instruments, such as the Structured Clinical Interview for DSM (SCID) and the related Semi-Structured Assessment for the Genetics of Alcoholism (SSAGA) and SSADDA, have been shown to be reliable and valid for the assessment of substance use and related diagnoses and traits across several languages and cultures (Keppel et al., 2001; Dick et al., 2005; Pierucci-Lagha et al., 2005; Malison et al., 2011). These interviews allow a trained (non-clinician) interviewer to identify DSM-III and DSM-IV substance dependence and Axis I and II psychiatric disorders by collecting data on the onset, severity and duration of symptoms.

Methods to translate psychiatric assessments optimally include iterative procedures to permit the adaption of complex cultural and linguistic concepts, such as disease state or associated symptoms, from one language to another. These procedures include translation and back translation (Kalayasiri et al., 2010), multiple translations and expert review (Garcia-Campayo et al., 2010), and key informant interviews and focus groups (Room et al., 1996). The process developed in this study and concurrent research in American Samoa (DePue et al., 2010) utilizes these procedures as well as the technique of cognitive interviewing (Willis, 2005). We used cognitive interviews to assess the linguistic translation and cultural relevance of the alcohol module of the SSADDA and determine how participants interpreted key questions and constructed their answers to these questions.

The development of a reliable and linguistically and culturally appropriate bilingual SSADDA could enhance the acquisition of useful information on alcohol and alcohol-related diagnoses in American Samoa and provide a new method for researchers translating diagnostic and assessment tools into different languages and cultures.

METHODS

Study setting

American Samoa was selected as the setting for this study based on a 35-year collaboration that one of us (STM) has had focusing on non-communicable diseases, especially obesity, type 2 diabetes and hypertension, in the context of rapid modernization that has led to a nutrition and health transition (McGarvey and Baker, 1979; Baker et al., 1986; Keighley et al., 2007; DiBello et al., 2009) and recent translational intervention research on diabetes (DePue et al., 2010, 2013). Alcohol consumption patterns from those surveys over many years showed increases in the proportion of adults consuming alcohol. Samoan collaborators also reported increases in the negative social and health consequences of alcohol use and abuse. Finally, alcohol consumption is a risk factor for a variety of non-communicable diseases, including cardiovascular and psychiatric illness, providing an additional rationale to conduct systematic alcohol research among Samoans (Rehm et al., 2003).

The study was approved by the Institutional Review Boards of Brown University and the American Samoa Department of Health.

Eligibility criteria

Eligibility for each phase's sample was determined with the following criteria: (a) not participating in ongoing translational diabetes intervention research; (b) ≥18 years old; (c) mentally competent and willing to give informed consent; (d) not intoxicated (breath alcohol <0.01 g/L); (e) unlikely to leave American Samoa for >6 months during the study; (f) one member per household. Additionally, focus group, cognitive interview and reliability-testing participants (described below) were eligible only if they self-reported Samoan ethnicity. In-depth interview participants (described below) were selected because they had specific experiences and expertise relevant to understanding the local context of alcohol-related services. Therefore, interviewees were eligible if they were not Samoan; however, all in-depth interviews addressed the local context of alcohol consumption. Each participant enrolled in the study after signing a written consent form.

Development phase procedure

The first aim of this study was to translate the SSADDA linguistically and culturally to identify alcohol diagnoses accurately. The SSADDA was adapted to the Samoan culture and language through the following steps: (a) translation and back translation of assessments into Samoan and expert review, (b) interviews and focus groups to identify culturally salient Samoan terms and concepts (e.g., portion size of a drink, effects of drinking and/or being drunk, depression, blackout) and (c) cognitive interviews to evaluate the translation and subject comprehension of relevant SSADDA items.

Translation (English to Samoan) and back translation (Samoan to English) with expert review

The alcohol module of the SSADDA was first translated into Samoan and then back translated into English by bilingual Samoans with experience in healthcare. Back translation was completed by a translator who did not participate in the original translation.

The translation and back translation were reviewed by a Samoan member of the research team (FA) for consistency. The back translation was also reviewed by the principal investigator (RS) to ensure the integrity and meaning of the content and diagnostic criteria. Throughout the process, a bilingual American Samoan psychiatrist practicing in Hawai'i reviewed translations and consulted with the research team.

Interviews and focus groups

Two of us (A.E.Q and F.A.) conducted 12 individual interviews with key informants and four focus groups with low-to-moderate and heavy alcohol users to understand the cultural context of alcohol use and to identify culturally salient terms and concepts. Focus groups and interviews were 60–90 min in length and conducted primarily in English. They were designed to identify the social context of drinking and obtain information on constructs in the SSADDA that are challenging to translate into Samoan, such as alcohol tolerance. The focus groups and interviews included discussion of terms and social cues used to communicate about alcohol.

The in-depth interview participants (N = 12) were predominately Samoan men aged 30–69. All had completed high school and half received education past college. These 12 individuals represented key stakeholders in either the distribution of alcohol, including bar and hotel owners, or the treatment of alcohol use and its social and legal consequences, such as legal professionals, policy makers and treatment providers. They were identified through formal and informal meetings with Samoan collaborators.

The focus group participants (N = 16) were ethnic Samoans between the ages of 20 and 29 and 44% were women. Participants were assigned to focus groups based on gender and the amount of alcohol that they reported typically drinking. One focus group was held with male, low-to-moderate alcohol users; one with male heavy drinkers; one with female, low-to-moderate alcohol users; and one with female heavy drinkers. We defined heavy drinking as >7 drinks per week for women and >14 drinks per week for men (Saitz, 2005; National Institute on Alcohol Abuse Alcoholism, 2007). Participants were recruited through newspaper and radio advertisements and posters placed in bars, grocery stores, churches and health centers. A description of the development phase sample is presented in Table 1.

Table 1.

Demographic characteristics of subjects in the development phase

	In-depth interviews (N = 12)	Focus groups (N = 16)	Cognitive interviews (N = 9a)
Age range
20–29	–	7 (44%)	2 (22%)
30–39	2 (17%)	3 (18%)	3 (33%)
40–49	2 (17%)	3 (18%)	3 (33%)
50–59	4 (33%)	2 (13%)	1 (11%)
60–69	4 (33%)	1 (6%)	–
Sex
Male	11 (92%)	9 (56%)	2 (22%)
Female	1 (8%)	7 (44%)	7 (78%)
Education range
Less than high school	–	2 (13%)	1 (11%)
High school	2 (17%)	3 (18%)	2 (22%)
Some college	2 (17%)	6 (38%)	1 (11%)
College	2 (17%)	2 (13%)	1 (11%)
More than college	6 (50%)	1 (6%)	2 (22%)
Missing	–	2 (13%)	2 (22%)
Ethnicity
Samoan	9 (75%)	16 (100%)	9 (100%)
White	2 (17%)	–	–
Tongan	1 (8%)	–	–

^aTotal N = 34 because three of the cognitive interview participants also participated in the focus groups.

Cognitive interviews

Cognitive interviews were conducted to evaluate translation and subject comprehension of the SSADDA alcohol module, which included items for which the translations and back translations were ambiguous and items that were identified as linguistically or culturally challenging based on the interviews and focus group data. A cognitive interview (CI) is a set of specific inquiries about the way participants understand assessment questions. Respondents answered questions from the alcohol module and were asked to narrate their thought processes about those answers, in order to evaluate whether the translation was effective. Alternative wordings of questions, written by the research team, were provided for some items. If participants indicated that the provided translation was not effective, they were asked to discuss the concept further and invited to suggest other language. This information was used to evaluate the translated item and decide whether an alternative was needed.

The cognitive interview participants (N = 9) were ethnic Samoans and predominately women between the ages of 20 and 59. Three of the cognitive interview subjects were recruited from among the focus group participants. The other six participants were recruited by word of mouth.

Reliability-testing procedure

The reliability phase consisted of interviews with 40 participants using the bilingual SSADDA to estimate its reliability. The participants in the reliability phase were ethnic Samoans recruited as an opportunity sample through newspaper and radio advertisements, posters, and word-of-mouth. Participants were able to choose to have the SSADDA alcohol module administered to them in English, Samoan, or both languages because the bilingual instrument included both English and Samoa. The interview was administered to 32% of the sample in both English and Samoan, to 20% in Samoan only and to 40% in English only (7.5% missing). The median age of the reliability participants was 24 years. The majority of the sample was female (60%), never married, had completed high school, earned less than $30,000 annually, and identified themselves as being in good health. A description of the sample is presented in Table 2. Twenty-six participants underwent repeat testing. The test sample and retest sample were not significantly different on key sociodemographic characteristics. The repeat interview was administered at least two weeks after the first interview.

Table 2.

Demographic characteristics of subjects in the reliability phase

	Total sample (N = 40) (%)	Test-retest (N = 26) (%)
Language used in SSADDA administration*
English	40	46
Samoan	20	39
Both English and Samoan	32.5	12
Missing	7.5	4
Age (years)
Median	24	25
Range	18–59	20–59
Sex
Male	40	34.6
Female	60	65.4
Religion has rules forbidding use of alcohol
No	45	38.5
Yes	47.5	50
No religion or missing	7.5	10.8
Marital status
Married	20	2.9
Separated	5	7.7
Never Married	72.5	65.4
Missing	2.5	–
Education (years)
Mean	12.8	12.9
Range	9–16a	9–16a
Currently employed
No	45	42.3
Yes	55	57.7
Income
$10,000–$19,999/year	10.3	11.5
$20,000–$29,999/year	40	38.5
$30,000–$39,999/year	18	15.4
$40,000–$49,999/year	15.4	15.4
>$50,000/year	7.8	6.8
Don't know/Refused/Missing	10.3	11.5
Self-reported health status
Excellent	17.5	15.4
Very good	22.5	2.9
Good	50	50
Fair	10	7.7
Poor	–	–

^a16 years of education represent attending 4 years of college or achieving a BA/BS degree.

*Test and retest sample are significantly different (P = 0.011).

Interviewers were trained and certified to administer the SSADDA. Training was conducted in American Samoa by a SSADDA trainer from the University of Connecticut Health Center.

Statistical methods

Chi square analysis and ANOVA were used to compare the sociodemographic characteristics of the test and retest samples. Data were scored using algorithms to generate DSM-III-R and DSM-IV alcohol dependence diagnoses. Following the literature (Fountoulakis et al., 2002), the test-retest reliability was determined with the phi coefficient using the test-retest sample (N = 26). The phi coefficient was used to estimate the categorical diagnostic agreement for DSM-III-R alcohol dependence and DSM-IV alcohol dependence. Spearman correlations were used to test the agreement between the number of criteria met during the first and second interviews (which varied from 0 to 7), corresponding to the DSM-IV criteria for alcohol dependence (American Psychiatric Association (APA), 2000). Both coefficients take values from −1 (total disagreement) to +1 (total agreement). The value 0 indicates chance agreement.

Results

Linguistic and cultural appropriateness

In most cases, the cognitive interviews confirmed that the initial translation was the most appropriate, because either it was understood as written or the alternatives were not an improvement. In one case, the translation was changed because the literal Samoan translation of the word ‘tolerance’ did not convey the equivalent diagnostic concept in English. A Samoan phrase that translates into English as ‘to feel the effect of alcohol, did you need to drink more of it?’ was preferred and the original translation was changed accordingly.

Reliability testing of bilingual SSADDA

Description of alcohol use and behaviors related to alcohol dependence

All 40 SSADDA participants reported a history of alcohol consumption. Our sample is similar to the American Samoan general population in terms of educational attainment, household income distribution, and age (U.S. Census Bureau, 2004). The median age of first use was 18 years with an interquartile range of 11. The median largest number of drinks consumed in a 24-h period was 12 with an interquartile range of 21. However because of the loose definition of a ‘drink’ in American Samoa, it is unlikely that this represents the largest number of ‘standard’ drinks consumed. As defined in the USA, a standard drink contains 14 g of alcohol, which is ∼12 oz of beer containing 5% alcohol, 5 oz of wine containing 12% alcohol or 1.5 oz of liquor containing 40% alcohol (Saitz, 2005; National Institute on Alcohol Abuse Alcoholism, 2007). Interview and focus group participants described ‘a drink’ in American Samoa to be a local 26 oz bottle of beer with 6.7% alcohol (Rosen et al., 2008), which is 2.9 standard drinks. Five participants reported having gone on binges/benders, and 14 participants reported having had blackouts. Two participants reported experiencing health problems from drinking. Although no participants had ever talked to a professional about their alcohol use, five had attended a self-help group. Seven participants were heavy drinkers in the past month, which was defined as drinking >4 drinks for women and >5 drinks for men per drinking occasion at least once in the past 30 days (Saitz, 2005; National Institute on Alcohol Abuse Alcoholism, 2007).

Alcohol diagnoses

The results of the diagnostic assessment are shown in Table 3. The SSADDA alcohol module was administered to 12 participants in English at both the test and retest sessions. It was administered to 14 participants in Samoan or in both Samoan and English at the test and/or retest sessions. Five participants had a diagnosis of DSM-III-R alcohol dependence at the first interview, but only four received that diagnosis at the second interview. Three participants received a diagnosis of DSM-IV alcohol dependence at the both interviews.

Table 3.

Categorical diagnostic reliability of alcohol dependence diagnoses, N = 26

Diagnosis	Proportion, % (N)		Reliability (phi coefficient)
	First interview	Second interview
DSM-III-R alcohol dependence
Total	19.23 (5/26)	15.38 (4/26)	0.87
English	25.00 (3/12)	25.00 (3/12)	1.00
Samoan and both languages	14.26 (2/14)	7.14 (1/14)	0.68
DSM-IV alcohol dependence
Total	11.54 (3/26)	11.54 (3/26)	0.62
English	16.67 (2/12)	16.67 (2/12)	1.00
Samoan and both languages	7.14 (1/14)	7.14 (1/14)	0.08a

^aThese were not the same participants.

Test-Retest reliability of categorical diagnostic agreement

For the total sample, the diagnoses of DSM-III-R alcohol dependence and DSM-IV alcohol dependence had good to excellent reliability with phi equal to 0.87 and 0.62, respectively. Because participants selected the language that was used during administration, phi coefficients were also calculated for the diagnostic reliability in English and in Samoan/bilingual versions. For the English-only sample, the diagnoses of DSM-III-R alcohol dependence and DSM-IV alcohol dependence were both perfectly reliable with phi equal to 1.00. The diagnosis of DSM-III-R alcohol dependence (phi = 0.68) in the Samoan/bilingual sample was good. However, the diagnosis of DSM-IV alcohol dependence was not reliable (phi = 0.08) in the Samoan/bilingual sample.

Test-Retest reliability of diagnostic criteria

The DSM-IV alcohol dependence diagnostic criteria met were consistent at the first and second interviews: two participants met three criteria and one participant met four criteria. The Spearman correlations between the two interviews for alcohol dependence (Spearman = 0.61, P = 0.001) reflected good reliability.

DISCUSSION

We found the linguistic and cultural meanings and interpretations of the bilingual version of the SSADDA appropriate for the diagnosis of alcohol dependence in American Samoans. Diagnostic reliability analysis indicated that the bilingual version of the SSADDA can reliably diagnose DSM-III-R alcohol dependence in this small convenience sample of American Samoans, who had all consumed alcohol in the past. SSADDA modules have now been found to be reliable in English (Pierucci-Lagha et al., 2005), Thai (Malison et al., 2011) and bilingual Samoan/English. Having a linguistically, culturally, and psychometrically accurate assessment tool available in American Samoa is important for the diagnosis of alcohol dependence in American Samoa.

The bilingual version of the SSADDA did not reliably diagnose DSM-IV alcohol dependence when administered in Samoan and in both languages. When the instrument was administered in Samoan or both languages, one participant was diagnosed with DSM-IV alcohol dependence at the first administration and a different participant was diagnosed with DSM-IV alcohol dependence at the second administration. It is possible that this is erroneous due to survey administration or interviewer error. It is possible that in a larger sample we would not have found such a discrepancy between DSM-III-R and DSM-IV alcohol dependence diagnoses. In view of the recent publication of DSM-5 (American Psychiatric Association, 2013), subsequent research should examine the impact of the change in diagnostic system on reliability of the alcohol use disorder diagnosis.

Although diagnostic reliability varied by language, it is clinically relevant for American Samoans to select their preferred language of administration (English, Samoan or both) because many people in American Samoa speak both English and Samoan (2000 Census of American Samoa, 2003). Certain concepts are also understood differently in the two languages (Siaki, 2011). Thus, the bilingual version of the SSADDA most likely maximizes comprehension in the population.

The generalizability of our estimates of alcohol use and alcohol-related diagnoses to the general Samoan population must be treated with caution for two reasons. Firstly, the study was conducted with a small, opportunity sample. Secondly, a high proportion of women in our sample reported having drunk alcohol. All women in the reliability-testing sample (N = 24) reported consuming alcohol, which is not consistent with prior research (Galanis et al., 1995; Aberg et al., 2008) that found very low levels of regular alcohol use among women in both American Samoa and Samoa.

Translation of complex diagnostic assessments from one language to another may require linguistic and cultural adaptation of concepts in addition to psychometric accuracy (Geisinger, 1994; Hambleton, 1994; van de Vijver and Hambleton, 1996). In developing the bilingual SSADDA, we used traditional procedures coupled with cognitive interviews designed to address linguistic and cultural translation. These cognitive interviews allowed us to clarify ambiguous concepts, such as alcohol tolerance. The translational process developed in this study and concurrent research in American Samoa (DePue et al., 2010) improves upon the translation processes in previous studies using the SSADDA or SSAGA (Keppel et al., 2001; Dick et al., 2005; Kalayasiri et al., 2010; Malison et al., 2011), because it deliberately assesses the linguistic translation, cultural relevance, and participant interpretation of diagnostic questions through cognitive interviewing. This methodology is important in American Samoa because there appears to be a discrepancy between clinical and cultural definitions of behavioral health problems (Held et al., 2010) and an influence of cultural protocols on translation (Siaki, 2011).

The results of this study of alcohol diagnoses in American Samoa are of public health importance, as American Samoans are experiencing high levels of alcohol consumption (World Health Organization, 2007) and undergoing substantial environmental and behavioral changes with modernization. Because the SSADDA alcohol module appears to be reliable and culturally valid in the Samoan language, if replicated, this study can lay the groundwork for future research on the patterns of alcohol consumption, behavioral and social problems associated with excessive alcohol consumption, potential treatment interventions, and genetic epidemiology. The results may also advance research and public health efforts to address the burden of alcohol use and non-communicable diseases across the Western Pacific region (World Health Organization (WHO), 2002). Further, the methodology used to translate the instrument can inform researchers translating diagnostic and assessment tools into different languages and cultures.

Funding

Funding for this study was provided by the National Institute on Alcohol Abuse and Alcoholism (R21 AA016597).

Conflict of interest statement

A.E.Q., R.K.R., J.E.M., F.A., S.F. and S.T.M. have no conflicts of interest to disclose. R.M.S. is a member of the Advisory Board for D&A Pharma and received a consulting fee and travel expenses. He is a consultant to Pharmaceutico CT and Transcept Pharmaceuticals and received consulting fees. H.R.K. has been a consultant or advisory board member for the following pharmaceutical companies: Alkermes, Lilly, Lundbeck, Pfizer and Roche. He is also a member of the American Society of Clinical Psychopharmacology's Alcohol Clinical Trials Initiative, which is supported by Lilly, Lundbeck, AbbVie and Pfizer.