PROBLEM GAMBLING SYMPTOMATOLOGY AND ALCOHOL MISUSE AMONG ADOLESCENTS A PARALLEL-PROCESS LATENT GROWTH CURVE MODEL Seema Mutti-Packer, Ph.D. University of Calgary Mutti-Packer, S., Hodgins, D.C., el-guebaly, N., Casey, D.M., Currie, S.R., Williams, R.J., Smith, G.J., & Schopflocher, D.P. (in press). Problem gambling symptomatology and alcohol misuse among adolescents: A parallel-process latent growth curve model. Psychology of Addictive Behaviors. AGRI Annual Meeting April 8 th, 2017
DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST I have no potential conflicts of interest for my presentation.
RESEARCH ON ADOLESCENT GAMBLING Adult prevalence of problem gambling (PG) 0.5% to 7.6% vs. Adolescent prevalence 0.2% to 12.3% Williams et al., 2012; Calado, Alexandre, & Griffiths, 2016 Based on two reviews from 2010: Research on adolescent gambling mostly descriptive and cross-sectional in nature Volberg et al., 2010 and Blinn- Pike et al., 2010 Since 2010, research on adolescent gambling has carried on in the same manner Delfabbro et al., 2016
RESEARCH ON ADOLESCENT GAMBLING Exceptions to this; earlier longitudinal studies based on aggregate analysis Winters et al., 2002 More recent studies have found gambling is not stable over time Carbonneau et al., 2016, and Delfabbro et al., 2009, 2014 Overall, longitudinal work on adolescent gambling is limited.
RESEARCH ON ADOLESCENT GAMBLING + ALCOHOL MISUSE Previous research has established the co-morbidity between problem gambling and alcohol misuse Barnes et al., 2002, 2009; Lorains et al., 2011 Behaviors thought to be part of a larger problem behavior syndrome Jessor and Jessor, 1977 Limited longitudinal research on adolescent gambling and alcohol misuse Barnes et al., 2002, 2009
STUDY OBJECTIVES The current study sought to examine: 1. the independent growth curves for problem gambling and alcohol misuse; 2. the directionality of the association between problem gambling and alcohol misuse from adolescence through to young adulthood.
HYPOTHESIZED ASSOCIATIONS 1. Baseline levels of alcohol misuse may be positively associated with the overall trajectory of problem gambling; Barnes et al., 2002 2. These two behaviors may mutually influence each other over time; 3. These two behaviors may not directly influence one another, but are instead related through other common underlying risk factors. Jessor and Jessor, 1977
DATA THE LEISURE, LIFESTYLE, & LIFECYCLE PROJECT Nady el-guebaly 1, David M. Casey 2, Shawn R. Currie 2, David C. Hodgins 1, Don P. Schopflocher 3, Garry J. Smith 3, Robert J. Williams 4 1. University of Calgary 2. Alberta Health Services 3. University of Alberta 4. University of Lethbridge
DATA THE LEISURE, LIFESTYLE, & LIFECYCLE PROJECT Seven-year prospective study 1,808 participants from four locations in Alberta Five critical age ranges: 13 to 15 18 to 20 23 to 25 43 to 45 63 to 65 years.
BACKGROUND CONCEPTUAL MODEL OF CONSTRUCTS IN THE LLLP
DATA ADOLESCENT COHORT Between 13-16 years at baseline Equal proportions of males and females High-risk group over-sampled TIME 1 n=436 TIME 2 n=350 TIME 3 n=312 TIME 4 n=312 RETENTION RATE = 72%
MEASURES PROBLEM GAMBLING SYMPTOMATOLOGY DSM-IV-MR-J (Diagnostic and Statistical Manual- Fourth Edition-Multiple Response-Juvenile) Based on the DSM-IV criteria for adult pathological gambling, and includes nine domains associated with problem gambling Conservative classification of problem gambling 0 = non-problem gambler 1 to 3 = at-risk gambler 4 or more = problem gambler
MEASURES ALCOHOL MISUSE Defined as binge drinking. How often in the past 12 months have you had 5 or more drinks on one occasion? 0 = Never 1 = Less than once a month 2 = Once a month 3 = Two to three times a month 4 = Once a week 5 = More than once a week
ANALYTIC PLAN A parallel-process latent growth curve model (LGCM) was specified: Step 1 Unconditional LGCM specified to assess change over time in problem gambling symptomatology. Step 2 Unconditional LGCM specified to assess change over time in alcohol misuse. Step 3 These two LGCM s are then combined into one unconditional parallel-process model. Step 4 Covariates are added to the model.
ANALYTIC PLAN SPSS 24.0 used for descriptive statistics; Mplus 6.1 used for longitudinal analysis. FIML for missing data Enders, 2001 MLR for non-normality and skewness of variables Muthen and Muthen, 2010 Model fit Indices Good fit: CFI and TLI >0.95, RMSEA <0.06, SRMR <0.08 Hu and Bentler, 1999
CONCEPTUAL MODEL Step 1
RESULTS. STEP 1: PROBLEM GAMBLING SYMPTOMATOLOGY Time 1 (n=436) n (%) Time 2 (n=350) n (%) Time 3 (n=312) n (%) Time 4 (n=312) n (%) DSM-IV-MR-J a 0 missing 0 missing 0 missing 0 missing Non-gambler 197 (45.2) 212 (60.6) 198 (63.5) 148 (47.3) Non-problem gambler 191 (43.8) 115 (32.8) 100 (32.0) 152 (48.8) At-risk gambler 46 (10.5) 22 (6.3) 13 (3.7) 12 (3.8) Problem gambler 2 (0.5) 1 (0.3) 1 (0.3) 0 (0.0) M ean (SD) 0.70 (0.79) 0.49 (0.73) 0.43 (0.65) 0.58 (0.65) a Categories listed for descriptive purposes only. The raw score was used for analyses.
STEP 1: PROBLEM GAMBLING SYMPTOMATOLOGY MEANS p-value Intercept 0.70 <0.001 Slope -0.24 <0.001 Observed individual trajectories Quadratic 0.05 <0.001 VARIANCES Intercept 0.13 0.07 Slope 0.22 0.005 Quadratic 0.01 0.09 WITHIN-PROCESS CORRELATION (r) Int <--> Slope -0.61 0.008 Int <--> Quad 0.36 0.269 Slope <--> Quad -0.92 <0.001 Model fit indices for gambling: X 2 =0.17 df=1, p=0.673; CFI=1.00, TLI=1.00, RMSEA=0.00; 90% CI=0.00-0.10, SRMR=0.01.
CONCEPTUAL MODEL Step 2
RESULTS. STEP 2: ALCOHOL MISUSE Time 1 (n=436) Time 2 (n=350) n (%) Time 3 (n=312) n (%) Time 4 (n=312) n (%) n (%) ALCOHOL MISUSE a 0 missing 0 missing 1 missing 0 missing Never 380 (87.2) 221 (63.1) 118 (37.9) 73 (23.4) < Once a month 33 (7.6) 74 (21.1) 82 (26.4) 102 (32.6) Once a month 11 (2.5) 23 (6.6) 40 (12.9) 46 (14.7) 2-3 times per month 7 (1.6) 22 (6.3) 48 (15.4) 54 (17.3) Once a week 3 (0.7) 8 (2.3) 21 (6.8) 30 (9.6) More than once a week 2 (0.5) 2 (0.6) 2 (0.6) 7 (2.2) M ean (SD) 1.22 (0.70) 1.65 (1.06) 2.28 (1.32) 2.64 (1.38) a Categories listed for descriptive purposes only. The raw score was used for analyses.
Mode fit indices for alcohol misuse: X 2 =1.16 df=2, p=0.558; CFI=1.00, TLI=1.00, RMSEA=0.00; 90% CI=0.00-0.08, SRMR=0.02. STEP 2: ALCOHOL MISUSE Observed individual trajectories MEANS p-value Intercept 1.22 <0.001 Slope 1.50 <0.001 VARIANCES Intercept 0.47 <0.001 Slope 1.52 <0.001 WITHIN-PROCESS CORRELATION (r) Int <--> Slope -0.29 0.009
STEP 3: BASELINE PARALLEL PROCESS MODEL
STEP 3: BASELINE PARALLEL-PROCESS MODEL Baseline levels of alcohol misuse positively associated with Baseline levels of problem gambling symptoms r=0.31, p=0.026 Baseline alcohol misuse negatively associated with change over time in problem gambling symptoms r= -0.25, p=0.028 r=0.31 p=0.026 Average trend line for Alcohol Misuse r=-0.25 p=0.028 Average trend line for Problem Gambling Symptomatology Slopes not correlated.
STEP 4: ADD COVARIATES Gender Parental household income Smoking status Past-year illicit drug use Age Ethnicity Location (urban vs. rural)
STEP 4: CONDITIONAL PARALLEL-PROCESS MODEL Average trend line for Alcohol Misuse The positive association between baseline levels of alcohol misuse and problem gambling symptoms dropped significance. r=0.31 p=0.026 r=-0.25 p=0.028 The negative association between baseline levels of alcohol misuse and the change over time in problem gambling symptoms dropped significance. Average trend line for Problem Gambling Symptomatology
STEP 4: COVARIATES TIME-INVARIANT COVARIATES (β) GENDER (male ref.) GAMBLING p-value ALCOHOL p-value --> Intercept -0.22* 0.034-0.02 0.631 --> Slope 0.03 0.732-0.16* 0.017 --> Quadratic 0.05 0.637 -- -- INCOME --> Intercept 0.26* 0.016 0.05 0.309 --> Slope -0.17 0.092-0.00 0.962 --> Quadratic 0.13 0.266 -- -- SMOKING STATUS --> Intercept -0.12 0.286 0.53*** <0.001 --> Slope -0.13 0.221-0.20 0.062 --> Quadratic 0.09 0.442 -- -- ILLICIT DRUG USE --> Intercept 0.22* 0.033 0.28** 0.001 --> Slope -0.16 0.154 0.09 0.280 --> Quadratic 0.16 0.247 -- -- Unstandardized estimates used for means/variances. Standardized estimates for correlations/coefficients.
HYPOTHESIZED PATHWAYS 1. Baseline levels of alcohol misuse may be positively associated with the overall trajectory of problem gambling; Barnes et al., 2002 2. These two behaviors may mutually influence each other over time; 3. These two behaviors may not directly influence one another, but are instead related through other common underlying risk factors. Jessor and Jessor, 1977
CONCLUSIONS Findings support problem behavior syndrome; Jessor and Jessor, 1977 Shared risk factors may include impulsivity, deviant friends; Vitaro, Brendgen, Ladouceur, & Tremblay, 2010 Prevention efforts should be broad-based, addressing skills to deal with multiple potential problems. Foxcroft et al., 2011; Griffin & Botvin, 2010; Williams, West, & Simpson, 2012
FUTURE RESEARCH Cohort-sequential analysis to examine the stability of problem gambling over the life course; Latent class analysis to address the significant variance found in trajectories of problem gambling symptomatology and alcohol misuse; The slope variance highlights the episodic, transient nature of problem gambling symptoms. However there is limited research on what underlies these transitions over time.
THANK YOU Seema Mutti-Packer, Ph.D. Addictive Behaviors Lab Department of Psychology, University of Calgary 200 University Avenue NW, T2N 1N4 seemamutti@ucalgary.ca seemamutti.com
QUESTIONS?
PROBLEM GAMBLING SYMPTOMATOLOGY AND ALCOHOL MISUSE AMONG ADOLESCENTS A PARALLEL-PROCESS LATENT GROWTH CURVE MODEL Seema Mutti-Packer, Ph.D. University of Calgary Mutti-Packer, S., Hodgins, D.C., el-guebaly, N., Casey, D.M., Currie, S.R., Williams, R.J., Smith, G.J., & Schopflocher, D.P. (in press). Problem gambling symptomatology and alcohol misuse among adolescents: A parallel-process latent growth curve model. Psychology of Addictive Behaviors. AGRI Annual Meeting April 8 th, 2017