Prevalence and Socio-demographic Correlates of Cigarette Smoking, Alcohol Use, and Unsafe Sexual Behavior among Ethnic Fijian Secondary Schoolgirls

Abstract

Introduction

The HEALTHY Fiji Study examines the impact of social transition on health risk behaviors among school-going ethnic Fijian adolescent girls. The primary aim of the present study was to assess prevalence and socio-demographic correlates of three risk behaviors, alcohol use, cigarette smoking, and unsafe sexual behavior in the study population.

Methods

We used an adapted version of the Global School-based Health Survey (GSHS) to assess health risk behaviors in a school-based sample of ethnic Fijian girls (n=523) in June and July 2007. We calculated prevalence of risk behaviors and then examined their relation to socio-demographic variables in logistic regression models.

Results

Prevalence estimates for any current alcohol use and cigarette smoking (20.1% and 17.6%) and lifetime history of sexual intercourse (20.8%) indicate that substantial percentage of this study sample has engaged in one of these health risk behaviors. Alcohol use was associated with two other risk behaviors, recurrent cigarette smoking and lifetime history of sexual intercourse. Although prevalence of alcohol use was lower than in several other Pacific populations, it was higher than previously reported among Fijian girls.

Conclusions

The prevalence of alcohol use, cigarette smoking, and unsafe sexual behaviors in this study population warrants concern. Comparison with estimates from previous health behavior surveys in Fiji suggest that mode of assessment may impact prevalence estimates for health risk behaviors.

Introduction

Tobacco use, alcohol use, and sexual behaviors that hold risk of sexually transmitted diseases (STDs) and unplanned pregnancy contribute to a substantial global burden of disease¹ and also to the leading causes of adolescent mortality.² These behaviors, which are frequently initiated in adolescence,^3,4,5,6,7 present opportunities to reduce risk for the preventable illness, death, and adverse social consequences associated with them. Globally, these three behaviors are among the top five risk factors for premature mortality and disability, topped only by malnutrition and poor water, sanitation or hygiene.¹ Substance use and poor reproductive and sexual health are also associated with poor mental health,⁸ in young people. Finally, tobacco use, alcohol use, and early sexual risk behaviors are often interrelated and are gateway behaviors for other health risk behaviors.^9,10,11, The well-documented social, economic, and health correlates of these health risk behaviors^{12,13,14, 15, 16,17, 18} are particularly challenging to address in resource-poor populations. Therefore, locally effective strategies to prevent and reduce youth health risk behaviors will benefit these same populations by protecting the health and economic productivity of their youth.¹⁹

Comparative prevalence estimates of cigarette smoking, alcohol use, and sexual risk behaviors among teenage girls in Fiji and the Western Pacific

Several sources of data are available on youth behaviors contributing to health risk in Fiji and in the Western Pacific. The Global Youth Tobacco Surveillance (GYTS), Youth Risk Behavioral Surveillance (YRBS), and the World Health Organization’s Global School-based Health Survey (GSHS) assess cigarette smoking with self-reported responses to similarly worded items. The YRBS and GSHS also assess alcohol use and sexual risk behaviors, and facilitate comparison of prevalence estimates across diverse populations in which they are implemented, although there are no published data from these on Fiji. These surveys are also all school-based, but cover different student age ranges. Finally, cigarette smoking and alcohol use data have been also collected in interviews with a sample of community-based youth and adults in the 2002 Fiji NCD STEPS survey.²⁰ Comparisons of definitions, phrasing, population age ranges, and year, and methods for these assessments are presented in Table 1.

Table 1.

Comparison of Assessments for Tobacco Use, Alcohol Use, and Sexual Risk Behaviors in Pacific Adolescents

	Global School-based Student Health Survey	Youth Risk Behavior Surveillance	Global Youth Tobacco Survey	Fiji Ministry of Health: NCD STEPS Survey 2002
Age range surveyed	Generally ages 13 and up; ages 15–20 in this study	Generally grades 9–12 (high school), though also used in middle school	Grade levels generally corresponding to ages 13–15 years old	Age 15–64 (stratified by age including 15–24 years old)
Method	Self-report	Self-report	Self-report	Face-to-face interview
Setting	School-based	School-based	School-based	Community-based
Phrasing to assess key risk behaviors
Current smoking	“During the past 30 days, on how many days did you smoke cigarettes?”	“During the past 30 days, on how many days did you smoke cigarettes?”	“During the past 30 days (one month), on how many days did you smoke cigarettes?”	“Do you currently smoke any tobacco products such as cigarettes, cigars or rolled tobacco?”
Current drinking	“During the past 30 days, on how many days did you have at least one drink containing alcohol?”	“During the past 30 days, on how many days did you have at least one drink of alcohol?”	--	“In the past 12 months, how frequently have you had at least one alcoholic drink?”
Lifetime history of sexual intercourse	“Have you ever had sexual intercourse?”	“Have you ever had sexual intercourse?”	--	--
Condom use during last intercourse	“The last time you had sexual intercourse, did you or your partner use a condom?”	“The last time you had sexual intercourse, did you or your partner use a condom?”	--	--
Age at first intercourse	“How old were you when you had sexual intercourse for the first time?”	“How old were you when you had sexual intercourse for the first time?”	--	--
Lifetime sexual partners	“During your life, with how many people have you had sexual intercourse?”	“During your life, with how many people have you had sexual intercourse?”	--	--

In the 2005 GYTS in Fiji, 3.1% of female students reported current cigarette use compared with a global prevalence of 6.5% of female students estimated with this assessment.²¹ In contrast, the 2002 Fiji NCD STEPS survey supports a prevalence estimate of 13.6%^* in 15–24 year old Fijian females for current cigarette smoking.²⁰ Neither of these surveys has published results stratified by ethnicity within Fiji. Prevalence estimates for current cigarette smoking for female adolescents generated by the YRBS or GSHS are relatively high in some Pacific populations, including 51.2% in the Northern Mariana Islands (Figure 1).^22,11

Figure 1.

Comparative prevalence estimates of current smoking in adolescent girls across country-specific populations

In contrast to several YRBS-based relatively high prevalence estimates of current alcohol use among female high school students in the Western Pacific Region (including 35.4% in the Marshall Islands and 46.2% in Republic of Palau in 2003²²) the 2002 NCD STEPS Survey reported a 7.8% prevalence among 15–24 year old females in Fiji (Figure 2).²⁰

Figure 2.

Comparative prevalence estimates of alcohol use in adolescent girls in country-specific populations

We could find no published data on the prevalence of sexual risk behaviors among Fijian youth, although available data support that the incidence of teenage childbirth in Fiji is comparable to rates reported in both the U.S. and in other Pacific Island nations.²³ Data from the Western Pacific populations surveyed elsewhere suggest there is a high prevalence of sexual intercourse among youth. ¹¹ Among females, YRBS data reported lifetime prevalence of sexual intercourse of 52.1% in the Mariana Islands, 46.3% in the Marshall Islands, and 29.1% in Palau. In the Mariana Islands and Marshall Islands, these numbers are similar to 45.7% of high school females who have ever had intercourse in the U.S.¹¹ (Figure 3). However, condom use among sexually active adolescents during their most recent episode of sexual intercourse was much higher in the U.S. (55.9% of females ¹¹ compared with only 27.79% for females in the Mariana Islands, 48.5% of all students in the Marshall Islands, and 44.6% of all students in Palau).²²

Figure 3.

Comparative lifetime prevalence estimates of sexual intercourse in females across country-specific populations

Youth Health Risk Behavior Assessment in Fiji

Excellent progress has been made to date with assessment of tobacco and alcohol use in Fiji with the GYTS and Fiji Non-Communicable Diseases STEPS surveys.²⁰ In this study, we propose several strategies to augment these important data. Specifically, additional assessment will be invaluable for identifying local sexual risk behaviors, reconciling discrepant prevalence estimates in cigarette smoking, as well as investigating socio-demographic risk correlates of these behaviors. Discrepancies across survey prevalence estimates suggest validity challenges for assessment of youth risk behavior, which can be addressed though additional collection of psychometric data. Moreover, the assessment of socio-demographic data allows stratified analysis to identify populations in greater need of intervention as well as suggesting modifiable risk correlates for further investigation.

Study aims

The primary aim of this paper was to assess prevalence and socio-demographic correlates of cigarette smoking, alcohol use, and sexual risk behaviors in a sample of female ethnic Fijian secondary school students with an adapted and translated version of the GSHS. To our knowledge, this study is the first to report GSHS prevalence data for Fiji. Because Fiji has recently announced plans to implement an English language version of the GSHS, a post-hoc aim of this study is to contribute to the data base that will be developed in Fiji. A secondary aim was to test hypotheses that enrollment in a peri-urban versus a rurally located school, availability of weekly spending money (which could increase access to alcohol and cigarettes), and concomitant substance use (kava, alcohol, and/or tobacco) were associated with higher risk for these health risk behaviors.

Methods

Study population

Study participants were drawn from all available ethnic Fijian adolescent girls, ages 15–20 inclusive, enrolled in Forms 3–6 in the 12 secondary schools registered within one administrative area designated by the Fiji Ministry of Education as of October 2006.²⁴ A total of 523 eligible study participants were enrolled from 12 schools.

Study procedures

Study participants completed a battery of self-report assessments on topics relating to demographic data, health behaviors, psychological health, and social environment. Informed parental (or guardian) consent and youth assent were obtained for each study participant. The surveys were administered at school sites where they were proctored by study staff who encouraged candid responses and clarified item content as needed. This study protocol was part of a larger study that was approved by the Fiji National Research Ethical Review Committee (FN-RERC), the Partners Healthcare Human Subjects Committee, and the Harvard Medical School Committee on Human Studies. Further details about this study protocol are available from the corresponding author.

Assessment of cigarette smoking, alcohol use, and sexual risk behaviors

Primary outcomes were assessed with the GSHS, adapted and translated for an ethnic Fijian population.^† This version included 71 questions from eight modules, including three assessing tobacco use, alcohol use, and sexual risk behaviors. The validity of our translation into a Fijian vernacular language was evaluated by comparing a back-translated version with the English original and editing for consistency, grammar, and syntax as described elsewhere.²⁵ There are no published studies on the validity of the GSHS, but its U.S. analog, the YRBS has been shown to have acceptable retest reliability for the majority of its items.²⁶ A related study reports acceptable test-retest reliability for the HEALTHY Fiji Study version of the GSHS, and substantial reliability²⁷ for the items assessing alcohol use, cigarette smoking, and sexual risk in this study except for one item assessing condom use which had moderate reliability.²⁸

Kava use

Four items relating to kava use were taken verbatim from the 2002 Fiji Non-Communicable Diseases STEPS survey²⁰ for use in the self-report survey. These items assessed lifetime and one-month prevalence of kava use, as well as perceived likelihood of smoking or drinking during or after kava use.

Socio-demographic characteristics, activities, and future plans

Primary ethnic identification as ethnic Fijian was determined by self-report. Age was calculated by the time elapsed from the birth date given and the day of the self-report study assessment. Rural or peri-urban school locations were determined by the study team based on proximity and access to a town in the province. Several additional items assessing activities and future plans were developed for this study and are reported in this paper for additional social context.

Data analyses

Prevalence of youth risk behaviors

One-month prevalence was assessed for the following outcomes: any cigarette use; recurrent or frequent cigarette use, any alcohol use; and recurrent or frequent alcohol use. Recurrent use of cigarette and alcohol was operationalized as use on at least 3–9 days in the past 30 days, and frequent use was operationalized as use on 10 or more days in the past month. In addition, both one-year and lifetime prevalence of sexual intercourse were assessed. Condom use during last intercourse was assessed only among respondents who affirmed a history of sexual intercourse in both an item on lifetime history and an item about condom use. In calculations of prevalence, the denominator was comprised of the number of non-missing responses to the given item (missing responses always occurred in less than 2% of the study sample).

Correlates of risk behaviors

Logistic regression models were used to examine socio-demographic correlates of the following outcomes: any tobacco use, recurrent and/or frequent tobacco use, any alcohol use, recurrent and/or frequent alcohol use. Models also examined correlates of lifetime history of sexual intercourse and failure to use a condom during the last episode of sexual intercourse. Estimates of variance were adjusted for the potential non-independence of responses from participants attending the same schools using the method of generalized estimating equations,²⁹ as implemented in SAS version 9.1.³⁰

Results

The participants’ mean age was 16.7 (1.1) years, and there was an approximately even distribution between enrollments in peri-urban versus rurally located schools. Approximately half of the sample reported receiving weekly spending money. Sixty-nine percent reported a history of lifetime kava use. Respondents reported high levels of socially desirable attitudes and behaviors; for example, the majority were weekly church goers and spent at least two evenings every week engaged in Bible study or church activities. Moreover, 80% had a professional occupational ambition and 84% expressed an ambition to either finish form 7 or a university education. Table 2 summarizes data on socio-demographic and positive social behaviors in participants.

Table 2.

Selected Socio-demographic Characteristics and Risk Behaviors of Study Participants

Mean age in years (SD)	16.7 (1.1)
Enrollment by form, % (no.)
Form 3	8.0 (42)
Form 4	30.8 (161)
Form 5	30.6 (160)
Form 6	30.6 (160)
Enrollment by school location, % (no.)
Peri-urban	49.9 (261)
Rural	50.0 (262)
Reporting weekly spending money, % (no.)
Any	50.7 (265)
None	43.8 (229)
Lifetime kava use, % (no.)
Any	69.4 (363)
None	30.6 (160)
Any kava use, past month, % (no.)
Any	31.0 (162)
None	68.8 (360)
Likely to smoke during/after drinking yaqona, % (no.)
Yes	7.3 (38)
No	92.5 (484)
Likely to drink alcohol during/after drinking yaqona, % (no.)
Yes	9.2 (48)
No	90.1 (471)
Participation in community functions, % (no.)
Not regularly	64.1 (335)
1 – 3 times/month	23.7 (124)
Every week	11.7 (61)
Sunday church attendance, % (no.)
Not regularly	11.3 (59)
1 – 3 times/month	7.3 (38)
Every week	81.3 (425)
Spend most of an evening at church, choir practice or Bible study, % (no.)
0 nights per week	11.1 (58)
1 night per week	12.1 (63)
2 to 3 nights per week	29.8 (156)
4 to 6 nights per week	29.3 (153)
7 nights per week	17.6 (92)
Educational ambition, % (no.)
Complete Form 7 or university	83.9 (439)
Attend trade school	12.6 (66)
Finish through Form 6, or do not know	3.3 (17)
Occupational ambition, % (no.)
Profession (e.g. teacher, nurse)	79.5 (416)
Retail or service industries	18.4 (96)
Help at home only	1.34 (7)

Response rates

All schools identified by the Ministry of Education as meeting the location eligibility for this study graciously agreed to participate and facilitated data collection. Within these schools, 71% of individuals meeting eligibility criteria were enrolled and completed their participation in the self-report assessments. Virtually all participants answered the questions relating to the primary study outcomes (cigarette use, alcohol use, sexual behaviors); the percent of item-level missing data ranged from 0.2% to 1.9% (Table 3).

Table 3.

Response Rates to Risk Behavior Questions

Item	Response Rate, % (no.)
Current cigarette use	99.8 (522)
Current alcohol use	99.8 (522)
Ever been drunk	99.8 (522)
Sexual intercourse ever	98.5 (515)
Sex in past 12 months	99.0 (518)
Condom during last sexual intercourse	98.1 (513)
Age at first intercourse	98.7 (516)
Lifetime sexual partners	98.9 (517)
Perception of peer cigarette use	99.4 (520)
Perception of peer alcohol use	99.6 (521)
Perception of peer sexual activity	98.3 (514)

Prevalence of cigarette smoking use, alcohol use, and sexual risk behaviors

17.6% of study participants reported at least some cigarette smoking in the 30-day period prior to the study, with a range of 3% to 27% across schools. With respect to use patterns, we found a much lower mean prevalence of recurrent use (3.8%) and an even lower mean prevalence of frequent use (0.8%). Results stratified by peri-urban versus rurally located school enrollment are presented in Table 4; prevalence estimates comparing the present study data with previous studies are presented in Figures 1 and 4.

Table 4.

Prevalence of Cigarette Smoking in Past 30 Days on the GSHS by School Location

	Rural, % (no.) (n = 261)	Peri-urban, % (no.) (n = 261)	Total, % (no.) (n = 522) [95% CI]
Any cigarette use	18.4 (48)	16.9 (44)	17.6 (92) [14.4, 20.9]
Infrequent or no cigarette usea	95.8 (250)	95.0 (248)	95.4 (498) [93.6–97.2]
Recurrent cigarette useb	3.8 (10)	3.8(10)	3.8 (20) [2.0–5.4]
Frequent cigarette usec	0.4 (1)	1.1 (3)	.77 (4) [0.0–1.5]

^aInfrequent or no use defined as use on 0–2 days of past 30 days.

^bRecurrent use defined as use on 3–9 of past 30 days.

^cFrequent use defined as use on ≥ 10 days of past 30 days.

Figure 4.

Comparative prevalence estimates of current smoking in young women in Fiji across three assessments

The vast majority of students (94%) reported either no drinking or only infrequent drinking. Approximately one-fifth of study participants (20.1%) reported having at least one alcoholic drink in the past 30 days, but the prevalence estimates for any alcohol use ranged markedly across participating schools (from 7% to 37%). Although 23% of respondents reported having ever been drunk, less than 6% reported recurrent or frequent alcohol use. Results stratified by peri-urban versus rural participants are presented in Table 5; prevalence estimates comparing the present study data with previous studies are presented in Figures 2 and 5.

Table 5.

Prevalence of Alcohol Use on the GSHS by School Location

	Rural, % (no.) (n=261)	Peri-urban, % (no.) (n=261)	Total, % (no.) (n=522) [95% CI]
Any alcohol use (in the past 30 days)	17.6 (46)	22.6 (59)	20.1 (105) [16.7, 23.6]
Infrequent or no alcohol use†	97.3 (254)	91.6 (239)	94.4 (493) [92.5, 96.4]
Recurrent or frequent alcohol use‡	2.7 (7)*	8.4 (22)*	5.6 (29) [3.6, 7.5]
Ever been drunk (lifetime)	18.4 (48)*	28.4 (74)*	23.4 (122) [19.7, 27.0]

^*Between group difference statistically significant at p<.01.

^†Infrequent use defined as use on 0–2 days in the past 30 days.

^‡Recurrent or frequent use defined as use on ≥ 3 days in the past 30 days.

Figure 5.

Comparative prevalence of alcohol use in young women in Fiji between two assessments

Prevalence of sexual risk behaviors

One-fifth of girls reported that they had ever had sexual intercourse (20.8%); 11% reported having had sexual intercourse in the past 12 months. Among sexually active girls, 40.3% reported not using a condom during their last sexual intercourse. Approximately 4% of participants (n = 19) reported first sexual intercourse at age 14 or younger and 7% (n = 37) reported having had sexual intercourse with two or more partners. Results stratified by participant school location are presented in Table 6; prevalence estimates comparing the present study data with previous studies are presented in Figure 3.

Table 6.

Prevalence of Sexual Risk Behaviors on the GSHS by School Location

	Peri-urban, % (no.)	Rural, % (no.)	Total, % (no.) [95% CI]
Lifetime sexual intercourse	23.6 (61)	17.9 (46)	20.8 (107) [17.3, 24.3]
Sexual intercourse, past 12 months	14.6 (38)**	7.4 (19)**	11.0 (57) [8.3,13.7]
Failure to use a condom during last sexual intercourse	45.8 (27)*	32.6 (14)*	40.3 (41) [31.2,49.9]
Age of first sexual intercourse 14 or younger	5.8 (15)**	1.6 (4)**	3.68 (19) [2.1,5.3]
Two or more lifetime sexual partners	9.6 (25)*	4.7 (12)*	7.16 (37) [4.9,9.4]

^*Between group difference statistically significant at p<.05.

^**Between group difference statistically significant at p<.01.

Correlates of risk behaviors

We investigated correlates of risk behaviors by fitting separate logistic regression models for smoking (Table 7), alcohol use (Table 8), and sexual risk behaviors (Table 9). A higher odds ratio indicates a stronger association between the covariate and the risk behavior.

Table 7.

Socio-demographic and Risk Behavioral Correlates of Cigarette Use, in the Past 30 days, in Multivariate Logistic Regression Models

	Any cigarette use† OR (95% CI)	At least recurrent cigarette use‡ OR (95% CI)
Age	1.07 (0.81, 1.42)	1.04 (0.79, 1.36)
Peri-urban	0.79 (0.45, 1.39)	1.09 (0.46, 2.62)
Spending money	0.84 (0.44, 1.60)	0.73 (0.40, 1.33)
Any kava use	2.23** (1.26, 3.97)	1.63 (0.46, 5.78)
Any alcohol use	6.80**** (3.77, 12.29)	10.75** (1.81, 63.94)

Note: OR=Odds Ratio, CI=Confidence Interval

^**P<.01

^****P<.0001

^†Defined as cigarette use on ≥ 1 day

^‡Defined as cigarette use on ≥ 3 days

Table 8.

Socio-demographic and Risk Behavioral Correlates of Alcohol Use, in the Past 30 Days, in Multivariate Logistic Regression Models

	Any alcohol use† OR (95% CI)	At least recurrent alcohol use‡ OR (95% CI)
Age	1.14 (0.90, 1.46)	1.03 (0.70, 1.50)
Peri-urban	1.52 (1.00, 2.33)	4.00*** (1.94, 8.26)
Spending money	1.28 (.83, 1.99)	0.89 (0.44, 1.83)
Kava use	3.37**** (2.23, 5.09)	2.38 (0.73, 7.77)
Cigarette use	6.82**** (3.76, 12.39)	13.21**** (5.03, 34.74)

Note: OR=Odds Ratio, CI=Confidence Interval

^**p < .01

^***p < .001

^****p < .0001

^†Defined as alcohol use on ≥ 1 day

^‡Defined as alcohol use on ≥ 3 days

Table 9.

Socio-demographic and Risk Behavioral Correlates of Lifetime Sexual Intercourse and No Condom Use

	Lifetime sexual intercourse OR (95% CI)	No condom use during last intercourse OR (95% CI)
Age	0.94 (0.78, 1.14)	0.94 (0.73, 1.21)
Peri-urban	1.45 (0.94, 2.23)	2.11* (1.02, 4.34)
Spending money	0.64* (0.44, 0.94)	0.64 (0.32, 1.29)
Kava use	1.31 (0.85, 2.01)	0.80 (0.29, 2.22)
Alcohol use	3.10** (1.48, 6.50)	2.20 (0.74, 6.50)
Cigarette use	1.43 (0.80, 2.56)	1.56 (0.61, 3.94)

Note. OR = odds ratio; CI = confidence interval.

^*P<.05

^**P< .01

Cigarette smoking

Both kava use and alcohol use (within the last month) were significantly associated with any cigarette smoking in the past month. Any kava use in the past month was associated with a twofold greater odds of smoking (OR: 2.23, CI: 1.26,3.97) and any alcohol use in the past 30 days was associated with a substantially higher odds of cigarette smoking (OR: 6.80, CI: 3.77,12.29). In contrast, alcohol use--but not kava use--was significantly associated with recurrent or frequent cigarette smoking (≥ 3 days/month). Age, spending money, and peri-urban school location were not statistically significantly associated with cigarette smoking (Table 7).

Alcohol use

Both kava use and cigarette use were associated with a higher odds of any alcohol use (Table 8). Kava use (any in the past 30 days) was associated with a threefold higher odds of alcohol use (OR: 3.37, CI 2.23, 5.09). Cigarette use (any use in the past 30 days) was associated with an almost sevenfold higher odds of alcohol use (OR: 6.82, CI: 3.76, 12.39). School location, age, and spending money were not associated with any alcohol use.

In contrast, kava use in the past month was not associated with recurrent or frequent alcohol use (≥ 3 days in the past month) in this study population. On the other hand, cigarette use was associated with a higher odds of recurrent alcohol use (OR: 13.21, CI: 5.03,34.74). In addition, peri-urban school location (as compared with rural location) was associated with a fourfold higher odds of recurrent drinking (OR: 4.00, CI: 1.94, 8.26, p = .0002), whereas age and spending money were not associated with recurrent or frequent alcohol use.

Sexual Risk Behaviors

Finally, we examined socio-demographic correlates of lifetime prevalence of sexual risk behaviors (Table 9). Alcohol use in the past month was associated with a threefold higher odds of sexual intercourse (OR: 3.10, CI: 1.48, 6.5). In contrast, spending money was associated with a lower odds of lifetime history of sexual intercourse (OR: .64; CI: .44, .94). Peri-urban school location, age, kava use, and tobacco use were not associated with sexual intercourse.

We then examined the socio-demographic correlates of not using a condom during sexual intercourse. Peri-urban school location was associated with a twofold higher odds of not using a condom during the most recent episode of sexual intercourse (OR: 2.11, CI: 1.02, 4.34). However, alcohol use, cigarette smoking, and kava use were not associated with failure to use a condom in this model. Nor were either age or spending money associated with failure to use a condom.

Discussion

In this study we present the first report of the prevalence estimates of cigarette use, alcohol use, and sexual risk behaviors assessed by an adapted version of the GSHS in an ethnic Fijian adolescent study population. Assessment in the vernacular language and support for the retest reliability of this version of the GSHS may have enhanced the validity of these findings. Study results augment data on cigarette smoking and alcohol use previously reported in Fijian populations with overlapping but distinct demographic characteristics. Prevalence estimates for sexual risk behaviors reported in this study also address a gap in the public health literature in Fiji. The substantial percentage of respondents who have engaged in at least one of these health risk behaviors warrants concern and consideration of strategies for intervention to support the healthy development of Fiji’s youth.

Several comparisons between prevalence estimates reported here and elsewhere are noteworthy. First, although comparability of data cannot be assumed, the prevalence estimates for all three health-risk behaviors (current cigarette smoking, current alcohol use, and lifetime history of sexual intercourse) appear to be lower than estimates generated with the YRBS in several neighboring Pacific adolescent populations (Figures 1, 2, & 3). A greater understanding of the potential for resilience among Fijian youth will be essential to protecting the future health of Fijian adolescents.

Second, there are discrepancies in prevalence estimates for cigarette smoking and alcohol use generated from other studies of Fijian populations as well as some reassuring convergence. Although our prevalence estimate of current smoking at 17.6% was very similar to the 13.6% estimate reported by the NCD STEPS survey, it is substantially higher than the 3.1% reported by the GYTS in Fijian 13–15 year old girls in 2005. We note that the GSHS and GYTS surveys use nearly identical wording to assess this behavior, although our survey allowed respondents to choose to respond in the vernacular. The different age ranges surveyed in these studies may partially explain the discrepancy.

The discrepancy in the prevalence of alcohol use between the present study and the NCD STEPS survey data (20.1% vs. 7.8%; Figure 5) was unexpected, since the latter survey included older respondents (ages 15–24) and 12 month prevalence (as compared with only a one-month prevalence) of alcohol consumption, both which would be expected to increase prevalence among the latter study population. Possible explanations for the discrepancy include regional differences, ethnic differences, changes between 2002 and 2007, and mode of assessment. Because these behaviors are viewed as socially undesirable, it is possible that youth were more likely to report their behavior in a self-report format that preserves greater anonymity.³¹

Because it is unlikely that students would falsely report engaging in a stigmatized behavior,³² we believe that prevalence estimates from any of these studies are likely to be conservative. Whereas differences in study populations do not allow interpretation of changes over time, we cannot exclude the possibility that the prevalence of cigarette smoking and alcohol use is increasing among ethnic Fijian adolescent girls.

Key findings presented in this study show: (1) Risk behaviors are present in both peri-urban and rurally located schools; (2) Although this study’s prevalence estimate for current alcohol use is higher than reported in the NCD STEPS survey, it is substantially lower than estimates assessed by similar measures in other Pacific Island adolescent girls as well as in the U.S. and in Australia; (3) Peri-urban versus rural school location is associated with a higher odds of at least recurrent alcohol use and failure to use a condom, but not with cigarette smoking in multivariate models; (4) kava use in the past month is not associated with a higher prevalence of recurrent or frequent cigarette smoking or recurrent or frequent alcohol use; (5) any alcohol use is associated with both cigarette smoking and with lifetime history of sexual intercourse. Although a causal link between behavioral and socio-demographic correlates cannot be inferred in this cross-sectional study, correlational data suggest areas for further investigation relevant to prevention strategies.

Study limitations

Our study has a number of limitations. First, our sample was limited to ethnic Fijian girls in just one region of Fiji and is therefore not representative of Fiji’s diverse youth. Although we do not have reason to believe that this region of Fiji has distinct behavioral risk from other areas, we anticipate that future studies of social and economic risk correlates may allow a more sophisticated understanding of local social environments and exposures that place certain youth at higher relative risk. Moreover, because we believe that health risk data on adolescent boys and on adolescents of all ages and regions in Fiji would be informative, we hope that this study’s findings and template for methods and design can be expanded upon for application for other populations in Fiji.

One of our concerns in presenting data from just one region and demographic of Fiji is that our findings may be misconstrued as evidence that the prevalence of risk behaviors is particularly high in this study population. However, there is no reason to believe that this region of Fiji is unique, given the nearly identical ratios of males to females and ethnic Fijians to Indo-Fijians in this area compared to Fiji as a whole. We also present information about selected behaviors and attitudes–church attendance, participation in community functions, professional ambitions -- that reflect the positive motivation and behaviors reported in this study sample (summarized in Table 2). We believe these data demonstrate that risk behaviors are prevalent despite the positive engagement in school and church activities and aspirations for future employment.

This study is also limited by its self-report nature. Although we believe that reliability support that youth understood the intention of these questions in a consistent way, self-report is also subject to recall bias, misinterpretation of questions, and social desirability.³² Although face-to-face interviews have the advantage of clarification of concepts and responses, they may also introduce bias related to perceived social desirability.^33,34 Finally, observational data would be difficult to collect for these risk behaviors as well.

Summary

In summary, the data presented here complement previous behavioral risk assessment data in Fiji in several ways. First, they extend the age range of self-report risk behavior data available in Fiji from the 13–15 year old age group surveyed in the Fiji Global Youth Tobacco Survey to a 15–20 year old age group. Our data also complement the data collected during the 2002 Fiji NCD STEPS survey by providing self-report data on tobacco and alcohol use. Next, the validity of GSHS data collected in this study is supported by reliability data;²⁸ the data also support the feasibility of use of a vernacular language for assessment. Finally, the data provide assessment of sexual behavioral risk, which has not previously been published for Fijian adolescents.

The GSHS collected in this study also provide two additional future potential benefits. In collecting concomitant data on psychological and social environmental risk and resilience factors, future analyses can identify correlates of risk and resilience, if any, with health risk behaviors that can inform prevention programs. Second, collection of these data in the GSHS format allows comparison of behavioral health risk among youth from Fiji with youth from diverse populations in other parts of the world.

This study suggests that concerning youth health risk behaviors are prevalent among ethnic Fijian schoolgirls. These behaviors have serious health consequences in other populations and warrant public health concern in Fiji as well. In addition, these behaviors have potential social consequences—a concern that has been echoed by ethnic Fijian leadership.³⁵ The apparently lower prevalence of some health risk behaviors in this study sample compared with other Pacific populations may also suggest some resilience among Fijian girls which needs to be better understood. Prevalence estimates reported in the present study also differ from previous reports in similar populations obtained with different assessments, but it is unclear whether demographic differences in the study populations, the method of assessment, or both have contributed to these discrepancies. Differences between the GSHS findings and NCD STEPS survey may also be due to the respondents’ greater comfort with reporting stigmatizing behaviors in self-report format. If this is borne out by other findings, the self-report format -- which is also easier and less time-consuming to administer -- may be a preferable option for future surveys. Finally, it is possible that an option to respond in the vernacular encouraged greater candor about stigmatizing behaviors. Clarity about the relative strengths of different modes of assessment will be enhanced by additional psychometric analyses accompanying future surveys. This study was only a first step in augmenting previous data to understand the prevalence and risk correlates of health risk behaviors among Fiji’s youth. Future research on these behaviors and their risk correlates in other Fijian communities can lay the foundation for community and school-based prevention strategies.