Use of the Fagerström test to assess differences in the degree of nicotine dependence in smokers from five ethnic groups: The HELIUS study

Highlights

• “Morning smoking” and “smoking patterns” explained variation in Fagerstrom Test for Nicotine Dependence (FTND) items.

• FTND items show differential item functioning across ethnic groups.

• Smokers of 3 out of 4 ethnic minorities score higher on ND compared to ethnic Dutch.

Abstract

Background

The prevalence of smoking varies across ethnic groups in developed countries, but little is known about ethnic variations in specific aspects of nicotine dependence (ND). We conducted item-response analyses in current smokers to compare ND factors across five ethnic groups.

Methods

Data were obtained from a population-based, multi-ethnic cohort study conducted in the Netherlands. The Fagerström Test for Nicotine Dependence (FTND) was assessed in 1147 Dutch, 991 South-Asian Surinamese, 1408 African Surinamese, 1396 Turkish, and 584 Moroccan smokers (N = 5526). We tested whether the factorial structure of the FTND was invariant across ethnic groups using a multi-group confirmatory factor analysis. FTND item and total scores and factor means were compared across groups.

Results

The two-factor model representing “morning smoking” and “smoking patterns” provided an adequate fit. The items “Cigarettes smoked daily” and “Time until first cigarette” showed differential item functioning (DIF) as a function of ethnicity. Three out of four ethnic minority groups scored significantly higher on both factors compared to the Dutch origin group (all p < 0.001) before and after taking DIF into account, while the African Surinamese scored higher only on “morning smoking” when DIF was accounted for.

Discussion

The factor structure of the FTND is not measurement invariant across ethnic groups in this population-based sample. Accounting for DIF affecting the nicotine dependence factor scores, although South-Asian Surinamese, Turkish, and Moroccan groups showed higher levels of dependence than the Dutch origin group, genetic as well as environmental factors may account for the observed differences.

Introduction

The prevalence of smoking varies greatly across countries. Western European countries, such as the Netherlands, have intermediate prevalence (Ng et al., 2014) while the estimated prevalence of daily smoking in men is relatively low in African countries. At a global level, women smoke consistently less than men (Ng et al., 2014). Similar differences across ethnic populations are observed within immigrant populations in high income countries. For example, one study performed in three immigrant populations in the Netherlands showed that the prevalence of smoking is higher in Turkish and Surinamese origin men (63% and 55%, respectively) than in Moroccan men (30%) (Nierkens et al., 2006). Furthermore, a study recently performed in the U.K. showed that some immigrant groups, especially those from Eastern European countries such as Turkey and Greece, had higher smoking rates than subjects born in the U.K. (Aspinall and Mitton, 2014).

Provided that there are large differences in smoking rates across ethnic populations, the question arises whether there are also ethnic differences in the levels of nicotine dependence (ND) among smokers. Liability for ND in smokers may differ across ethnic groups as there may be variations in sociocultural factors (e.g., religiosity, smoking acceptability, or education), psychological factors [e.g., perceived discrimination (Visser et al., 2017) and personality traits (Choi et al., 2017)], affordability, and biological factors (Benowitz et al., 2006; Sullivan and Kendler, 1999). In the present study, we will explore ethnic differences in the different components of ND in smokers across ethnic groups.

The HELIUS study (Healthy Life in an Urban Setting; a large multi-ethnic cohort study) investigates health differences in cardiovascular disease, infectious disease, and mental health among the six largest ethnic groups in Amsterdam, the Netherlands, i.e.: South-Asian Surinamese, African Surinamese, Ghanaian, Turkish, Moroccan, and Dutch (Stronks et al., 2013). Within the HELIUS study, Visser et al. (2017) showed that, compared to the Dutch origin group, smoking rates were higher in Turkish (OR = 1.59), South-Asian Surinamese (OR = 1.31), African Surinamese (OR = 1.59), but lower in Ghanaian (OR = 0.14) and Moroccan (OR = 0.47) origin groups. Compared to Dutch origin smokers, the prevalence of ND was higher in Turkish (OR = 2.09), Moroccan (OR = 1.85), and South-Asian Surinamese participants (OR = 1.79), whereas it was similar in African Surinamese (OR = 0.89) and Ghanaian (OR = 0.95) smokers.

In the HELIUS study, ND was assessed with the Fagerström Test For Nicotine Dependence (FTND), one of the most widely used instruments to assess dependence (Heatherton et al., 1991). The FTND has been shown to be a better predictor of ND and smoking quit rate compared to the number of cigarettes per day or the number of years of smoking (John et al., 2003). Among different scales aimed at assessing ND, it has the strongest correlation with withdrawal symptoms and relapse to smoking because of withdrawal symptoms (DiFranza et al., 2012; Fagerström and Furberg, 2008; Moolchan et al., 2002; Rios-Bedoya et al., 2008). In some studies, the FTND is composed of one factor (Chabrol et al., 2003; Heatherton et al., 1991), while in other studies, there are two factors, labelled as “Smoking Pattern” and “Morning Smoking,” (Steinberg et al., 2005; Richardson and Ratner, 2005; Radzius et al., 2003; Jhanjee and Sethi, 2010; Haddock et al., 1999).

In the present study, we will extend the earlier work by Visser et al. (2017) and explore ethnic differences in the factor structure and the item characteristics of the FTND. To be able to make a meaningful comparison regarding ND across different ethnic subgroups, it is essential to establish that the FTND is a valid instrument in all ethnic groups by testing whether the factor structure is invariant across groups. Earlier studies have indicated that the factor structure of the FTND may be different across ethnic groups. The factor structure of the FTND was found to be different in a sample of white and African-American adolescents (Schroeder and Moolchan, 2007) and in a community-based sample of European and African American current and former adult smokers (Johnson et al., 2008). As far as we are aware, no studies have investigated measurement invariance of the FTND in ethnic minority groups within Europe.

The aim of this study was to investigate whether the factorial structure of the FTND is similar across five different ethnic groups monitored by the HELIUS study group. Participants with Ghanaian background were excluded because of low smoking rates in the Ghanaian population (11.7%). Since the number of factors was found to be inconsistent in earlier studies, we will compare the fit of one- and two-factor models. After establishing whether the FTND is measurement invariant amongst ethnic groups, the second aim of this study is to investigate differences in the levels of ND at the level of the underlying factor(s).