A STRUCTURAL EQ U ATIO N M O D EL OF A LC O H O L M ISUSE AM O N G CLIENTS OF TTIIELI3]3^\IJIlEC0E2;TI3j\I3IE;S PICXSSIEEHLE IPROGItvUVI by Louise Spaulding B .A ., The U niversity o f Alberta, 1990 THESIS SUBM ITTED IN P A R T IA L FU LFILM E N T OF THE REQUIREMENTS FOR THE DEGREE OF M ASTER OF SCIENCE in PSYCHOLOGY © Louise Spaulding, 2002 THE U N IVER S ITY OF NORTHERN BR ITIS H C O LU M BIA August 2002 A ll rights reserved. This w ork may not be reproduced in whole or in part, by photocopy or other means, w ithout the permission o f the author. NmlkxiW Ubwy c(C«n#d# Biblioxhèque nmtionml# du Canada AcauWon# #nd Bib#ogrmph#eS#*vic## 305 iWHW WngkwINMMM AcquWbonaat menâaalAGognyAkp*# aaS.wWWmglon ONmwmON K1A0N4 Cmmmdm OammON K1A0N4 Canada I o*m TtemuthorliasgpanüKlaiK*^ I^'amlettr maccomdé une IkaenAciKNi eoBcdkigisMBpenneMmnîàla (BMdhaÔMBl&oeooeiükMwnygdke )%*d€guü]La%nuycf()BWMhiü) f qpKXh c e, kM Kk Grade 12 23.8% Note: Marital Status categories were collapsed into two groups, the first group labelled, ‘With Partner’, includes those clients who were married or common-law, and the second group labelled, ‘Without Partner’ includes clients indicating they were single, divorced, or separated. 60% 43% w 40% uJ 30% o. 20% 9% 4-12wks 1 3 -2 8 wks 29 - 37 wks GESTATIONAL AGE Figure 4 Distribution of HBP clients by gestational age at intake. Ü1 in 80% 72% 70% 60% - □ 50% ü u. O lU 40% g z g 30% QC UJ Q. 21% 20% 18% 13% 7% 10% 7% 3% 0% SELF PHYSICIAN FRIEND c cr',- * . z , Figure 5 Client sources of referral to HBP progarm. COMMUNITY HEALTH UNIT GROUP PREVIOUS CLIENT SOCIAL SERVICES SOURCE OF REFERRAL cn CD 70% 62% 60% 50% g u 40% u. 0 u ; 1 30% 5 EL 20% 16% 10% 10% 9% 3% 0% Social Assistance No Income Figure 6 Client financial circumstances. Inadequate Income- Depend on Parents No Assistance Adequate Income cn N 58 Forty-one percent of the women were experiencing their first pregnancy at the time of their enrolment in the program. The remaining clients had an average of three total pregnancies and one term delivery. Of the women going through their second or subsequent pregnancies, one third (32.8%) had previous elective abortions, and another third (34.5%) had previous spontaneous abortions. 8% of these multiparous women delivered low birth weight babies, and 10% reported pre-term deliveries. Of these multiparous clients, 55.1% attended prenatal classes for their previous pregnancies and 46.2% represented previous clients of the Healthiest Babies Possible program. Table 2 presents pertinent psychosocial risk factors. Almost 40% were recorded as having an inability to cope with the pregnancy, and/or suffering some anxiety regarding the pregnancy and having a baby according to the counsellors' assessment. The vast majority of the Healthiest Babies Possible clients used tobacco (72.5%), and 41% were considered at risk for alcohol misuse according to the counsellors' assessment. 59 Table 2 Prenatal Risk Factors for HBP Clients for the 1996 Fiscal Year. Number of Responses to Item Number of Affirmative Counsellor Responses Percentage of Affirmative Counsellor Responses Cigarette Smoking 142 103 72.5 Alcohol Use 142 58 40.8 Illegal Drug Use 140 53 37.9 Low Self Esteem 137 104 75.9 Relationship Problems or Family Violence 120 63 52.5 Inability to Cope / Anxiety Regarding Pregnancy and Baby 122 47 38.5 Family History of Abuse or Neglect 104 32 30.8 Refusal or Resistance to Appropriate Services 127 28 22.0 Delayed Access to Prenatal Care 135 26 19.3 SUBSTANCE ABUSE or MISUSE PSYCHOSOCIAL and ECONOMIC FACTORS The Intake Assessment Questionnaire The Healthiest Babies Possible program's Intake Assessment Questionnaire (lAQ) was designed and used to collect information about clients at intake and at various stages of their pregnancies (Browne, Thio-Watts, & Spaulding, 1996). The questionnaire was comprised of 11 sections. Because not all sections provided data for the present study only those that were used will be described here. Data were collected during a period that began at intake and often continued over a period of two client 60 visits due to the length of the lAQ and the client’s availability for intake. The Client Characteristics section was used to record demographic information including age, marital and ethnic status, education and financial levels. The Referral Data section was used to record information about the client's access to the program such as source of referral and gestational age at intake. One section pertained to Prenatal Risk Factors and was completed at various stages of the client’s relationship with the program as the counsellors became better acquainted with each client. It included a checklist of factors that are known to have an effect on the fetus or the client during the course and/or outcome of their pregnancy. The checklist of prenatal risk factors was divided into three subsections which included ’physical" factors, substance abuse/misuse' factors, and 'psychosocial and economic factors'. Only the last two subsections were of interest in this study (see Appendix A). Responses to the Prenatal Risk Factors section reflected the client's response and the counsellors' opinions as to whether or not a particular issue should be considered a significant problem for the client. This section was updated as the counsellors obtained more information about the client during the course of the pregnancy and was one of the last sections to be thoroughly completed before the client file was closed after delivery. For example, a counsellor might have been able to check off ‘single motherhood’ at their first visit with the client; however, determining whether or not a client is exhibiting drug or alcohol problems often took more time to verify. Questionnaires are updated as information is presented. Included in the prenatal risk factor section were variables used in this study for the analysis. From the 'psychosocial and economic factors' subsection, clients were 61 assessed by checklist Items where an affirmative response indicated psychological conditions such as low self-esteem and/or an inability to cope with the pregnancy and/or baby. Client behaviours were also recorded in this section, such as delaying access to prenatal care and resistance to appropriate services. Additionally, there was one item from the Substance Misuse section checked off by the counsellor if, in their opinion, the client was thought to misuse alcohol. The Alcohol Data section gathered various information regarding the client’s history with alcohol. The purpose of this exhaustive section of the intake tool was to identify clients who may have had problems with reducing or eliminating their alcohol intake during the course of their pregnancy. There were three alcohol use related variables used for the present study including consumption, the T- ACE Score and the counsellor's decision. It is explained to clients that 1 drink is the equivalent of 1 beer, a 5 ounce of wine or a 1 ounce high ball. The alcohol consumption variable asked clients what is the maximum number of drinks they hold. Clients were asked four questions contained in the T-ACE screening tool for alcohol risk and the total score was entered for each client. Finally, the counsellor made a decision based on the above variables and their observations and experience with the client as to whether alcohol misuse during pregnancy was a potential concern. Procedure As implied by the foregoing, completion of the lAQ was flexible and evolved over the course of the client’s contact with the program. The lAQ was administered by the lay counsellors employed at the Healthiest Babies Possible program. Since the counsellors’ main priority was with regard to the client’s situation, and many of these 62 high-risk clients were enduring crises at the time of intake, the completion of the lAQ assessment in itself was not the first priority. For many clients the initial assessment using the lAQ took at least two visits. In addition, due to the client-centred philosophy at the Healthiest Babies Possible program, clients were not asked certain questions if they were thought to be disturbing. The client could also refuse to answer any questions. Therefore, the number of missing values for this data set was expected to be a slightly higher than is generally desirable. The lAQ was designed for the use of the Healthiest Babies Possible program staff for program planning, evaluative and administrative purposes, and was not specifically designed for this thesis. However, close contact between the researcher, the program co-ordinator and counsellors was established two years before data analysis. Presentations and discussions regarding the importance of consistent data were an important part of ensuring that client records were accurate and reliable between and within counsellors. Data Analysis The original data were entered into a prepared file using Epilnfo 5.0 data entry software. This program is shareware provided by the World Health Organization (WHO), and is easily downloaded from the internet. Since the capacity for analysis of data using this program is limited, the data file was converted, and the descriptive analyses were conducted using SPSS® version 6.1.2 (1995), while LISREL 8.50 (Student Version, June 2001) was used to perform the structural equation analyses. Measures used for structural equation model. All measures for this study were extracted from the aforementioned lAQ data. The percentages of clients who 63 experienced the social, psychological and economic prenatal risk factors pertinent to this study included in the last section of the prenatal risk factors checklist are presented in Table 2. Half of the Healthiest Babies Possible clients reported relationship problems and/or violence with their partner (52.2%). Thirty-one percent reported a history of family abuse or neglect. The vast majority of the clients experienced low self-esteem (75.9%) and 22% exhibit resistance or refusal of appropriate services. There were 38.5% of clients reporting an inability to cope and/or anxiety about the pregnancy and baby. The 19.3% of clients recorded as delaying access to prenatal care entered the program after 24 weeks gestation. Appropriate scales were constructed from dichotomous response items, and are used as indicator variables in the model. The latent factors, social influences, personal resources, and risky behaviour, were constructed from the indicator variables and guided by the model presented in Figure 7. ENVIRONMENTAL FACTORS COGNITIVE PROCESSES BEHAVIOUR SOCIAL INFLUENCES PERSONAL RESOURCES RISKY BEHAVIOUR LATENT CO NSTRUCT FAMILY HISTORY DENSITY & INDICATORS YES 1 1 FATHER MOTHER PARTNER 1 SIBLING 1 SUM NO 0 0 0 0 LATENT CONSTRUCT Self-efficacy Inability to Cope Delayed Access to Prenatal Care & INDICATORS YES NO 1 1 0 0 SUM LATENT CO NSTRU CT & INDICATORS T-ACE SCORE Range = 0 -5 Counsellor's Assessment YES NO 1 0 Self-esteem ADVERSITY DENSITY 0 Low Self-esteem Resistance to Help Hx of Abuse/Neglect Consumption Log Function 0 SUM Relationship Violence SUM MEDIATED EFFECT PERSONAL RESOURCES ' SELF-EFFICACY / SELF-ESTEEM SOCW^ WFLUENCES \ FAMILY HISTORY DENSITY RISKY BEHAVIOUR \ J ^ T-ACE DIRECT EFFECT ADVERSITY Figure 7 Latent and Indicator variables for ttie proposed model. ASSESSMENT CONSUMPTION 65 Social Influences Factor. The Social Influences factor represents negative social influences surrounding the client currently and in the past. There were two composite variables used to indicate social influences labelled Family History Density and Adversity. Family History Density was the total number of immediate family members who, according to the client, misused alcohol. Clients were asked if their father, mother, siblings, and/or partner misuse alcohol. There were no questions regarding extended family or peers. The total number of affirmative responses was totalled and labelled Family History Density. The family history density variable was used by Stoltenberg and Mudd, (1998) who found this method of operationalization to be reliable and valid. Scores ranged from 0 to 4, with the higher scores reflecting a greater density of drinking behaviours within the client's family. The second variable in the social influences factor is the Adversity variable which was designed to assess the presence of neglectful, abusive, and violent environments that clients were experiencing currently or had in the past. Clients were asked if they had experienced any abuse and/or neglect in childhood, and whether or nor they were suffering current relationship problems and/or family violence issues. The adversity variable was the sum of affirmative responses for these two dichotomous variables. Scores ranged from 0 to 2 with higher scores reflecting a greater level of perceived adversity. Personal Resources Factor. The Personal Resources factor was created from two composite indicator variables and was intended to reflect specific client inner cognitive resources or psychological characteristics. The composite indicator variables for this factor were self-efficacy and self-esteem as assessed by the counsellor based 66 on the client's manifest behaviour. The self-efficacy variable combined two questionnaire items from the prenatal risk factors section and is intended to reflect the client's self-efficacy regarding their pregnancy. The first item, ‘Inability to Cope/Anxiety Regarding Pregnancy and Baby’ reflected the client's feelings about her current challenges. The second item indicated that the client had delayed accessing prenatal care (defined as longer than 24 weeks gestational age at first intake). In conceiving this variable, such behaviour was taken as a marker of avoidant style coping behaviour. Scores ranged from 0 to 2 with higher scores reflecting lower levels of self-efficacy. The self-esteem variable was constructed in a similar manner to self-efficacy. The first item of this composite variable indicated whether, in the counsellor's assessment, the client suffered from low self-esteem. The second item used in this variable indicated refusal of or resistance to appropriate services as observed by the counsellor. This resistance behaviour is comparable to the 'threat appraisal' concept put forth by Nyamathi et al. (1995), which was found to be a function of self-esteem. Affirmative responses to the counsellor's perception of low self-esteem and to the resistance behaviour item were scored 1. Scores ranged from 0 to 2 with higher scores reflecting lower levels of self-esteem. Risky Alcohol Behaviour. The Risky Behaviour factor was an index of the level of risk for alcohol consumption during pregnancy. It was comprised of three indicator variables - the T-ACE score, alcohol consumption and the counsellor's assessment. The T-ACE scale is a well-established screening tool to assess the risk of alcohol problems. According to Russell's (1994) study on new screening tools for risky 67 drinking while pregnant, the four-item T-ACE test is just as sensitive as the large, well established Michigan Alcoholism Screening Test (MAST) (Selzer, as cited in Russell, 1994). T-ACE is an acronym for its component variables. All clients were informed of what was considered a standard drink (see Appendix A). 'T stands for tolerance and is an index of how many drinks the client consumes before they feel the effects of the alcohol. If the client’s response to the tolerance question is one drink or less it is scored zero, if the response is 2 drinks, it is scored as 'T, and any response greater than 2 drinks is scored as '2'. T-ACE tolerance scores range from 0 to 2. 'A' stands for annoyed, and is an index of whether other people have ever annoyed them by criticizing their drinking. 'C stands for cut down, and is an index of whether they believe they should cut down on their drinking. 'E' stands for eye-opener and is an index of whether they have ever consumed a drink in the morning to relieve stress or quell a hangover. Affirmative responses to the last three questions receive scores of 1, thus the T-ACE score ranges from 0 - 5. A total score of two or more on the T-ACE test suggests problems with alcohol. The second risky behaviour variable reflected the counsellors' cumulative understanding of the client, both through the lAQ questions regarding alcohol use, and personal interviews held during the course of the client's relationship with the program. If, in the counsellor's assessment, the client was thought to abuse or misuse alcohol during pregnancy, this item in the prenatal risk factor subsection was scored as "1", otherwise a score of 0 was given. This indicator variable is labelled Counsellor's Assessment. The last indicator of the risky behaviour factor was alcohol consumption. For 68 consumption, the clients were asked, ‘How many drinks does it take until you feel you have consumed too much?’ Unlike the T-ACE score, this variable records the actual number of drinks as quoted by the client. This last continuous variable had a highly skewed distribution and was therefore, subjected to a log transformation for the analysis. The Proposed Structural Equation Model One of the advantages of structural equation modelling (SEM) over other methods is that multiple indicators are permitted for any given theoretical concept (i.e., latent factor). This reduces measurement error, and allows the testing of mediating factors inherent in psychological theories such as the social learning theory. Since SEM is an iterative type of analysis with complex decision conventions, it will be necessary in this section to discuss technical and conceptual details that guided this analysis. Measurement model. The measurement model deals with the relationships between indicator variables and latent factors. Hayduk (1996) recommends using fewer ‘gold standard’ indicators per concept, rather than greater than three indicators which is typically recommended for a factor analysis. Hayduk recommends that indicators represent the intellectual and theoretical goals of the study. It is recommended that the variance of the 'best' indicator per concept be fixed at 1 for the purposes of scaling and identification (Hayduk, 1987; Hoyle, 1995). The social influences factor has two indicators, family history density and adversity. According to theory, family history density is the 'best' available indicator of this factor. The personal resources factor also has two indicators, self-efficacy and self­ esteem. Self-esteem is the 'best' indicator for this factor. Finally, the risky behaviour 69 factor is comprised of three indicator variables with the T-ACE score as the 'best' indicator. The variances for these three 'best' indicators were fixed at 1 for the purposes of analysis. Another advantage of structural equation modelling is that the researcher may fix' or 'free' model parameters as theory dictates. Since the T-ACE score was a function of consumption and the counsellor's decision was a function of the T-ACE score, the error covariances between these variables were allowed to vary freely. Structural model. The structural model refers to the relationships among the latent factors or theoretical constructs. Independent latent variables in SEM are referred to by the Greek letter 'eta'. Eta variables have no paths leading towards them. Latent factors with paths leading to them are referred to by the Greek letter 'ksi', and may be outcome or mediating factors. In the present model, the social influences factor is the eta factor, risky behaviour is the outcome factor and personal resources is the mediating factor. Personal resources and risky behaviour are ksi variables. Method of estimation. In SEM the maximum likelihood extraction method was used in all analyses. Maximum likelihood is a robust method of estimating structural equation modelling parameters particularly recommended when the sample size is relatively small (i.e., 150 or less; Tabachnick & Fidell, 1996). Model Fit There is not a definitive measure to determine if the hypothesized model fits the data, therefore researchers use a combination of measures to evaluate a structural equation model. There are several types of fit indices used to evaluate models. In addition, path and squared multiple correlation coefficients are examined. Finally, similar to regression, residuals are analyzed. Many researchers consider four types of 70 fit indices: absolute, relative, parsimony-based and those based on the non-centrality parameter (Tanaka, 1993). Each type of measure is designed to take various aspects of the empirical data into consideration. Most goodness of fit indices compare hypothesized models relative to the best and worst models possible with a particular data set. The worst model is where all variables are independent (i.e., no relationships) and is referred to as the independence model. Conversely, the best model is referred to as the saturated or unconstrained model, where all possible paths between variables have been drawn and 100 % of the variance is accounted for. The hypothesized model falls somewhere in between. Goodness of fit indices are basically a measure of how far away the hypothesized model is from either or both of these theoretical benchmarks (Hoyle, 1995). Chi-square. The chi-square statistic was the original goodness of fit measure because it is derived directly from the maximum likelihood estimation minimization function (i.e., 'Fit' function). In structural equation modelling, chi-square statistics are used to compare the hypothesized model to the independence model (i.e., worst) or to the unconstrained model (i.e., best). The unconstrained model is the theoretical model with all pathways unconstrained, that is, variance between factors are free. If there is a significant difference between the two models, it is considered a bad fit. However, chisquare is very sensitive to sample size, model size (i.e., number of parameters), and the distribution of variables. As a result, chi-square itself is rarely used any more, however, new indices have been developed that adjust the chi-square value for the degrees of freedom, normality, and the number of parameters in the model (Hoyle, 1995). 71 Absolute fit indices. The Goodness of Fit Index (GFI) is simply a ratio of the maximum likelihood fit function before and after the model is fit to the data. The Adjusted Goodness Of Fit Index (AGFI) is the GFI adjusted for degrees of freedom. These two measures are not explicitly a function of sample size, but their distributions are affected by sample size. Values of .90 or greater indicate a good fitting model. The final absolute index of fit is the Root Mean Square Residual (RMR) which is the square root of the average squared amount, by which sample variances and covariances differ from the estimates under the model. Values for the RMR and RMSEA should be less than .05. There are several other absolute fit indices, however they are all based on chisquare and suffer from similar shortcomings. Relative fit indices. Relative fit indices compare the hypothesized model with the independence model (i.e., the worst model). Values for these indices are calculated as ratios of the model chi-square and the independence model chi-square and degrees of freedom for the model. These indices indicate how much better the hypothesized model is over the independence model. There are several relative fit indices. The Rentier Bonnett Nonnormed Fit Index (NFI) is based on a ratio of the hypothesized model fit to the independence model fit. An NFI value of .92 would indicate that the model is 92% of the way between the worst model and the best model. Bollen's Incremental Fit Index (I FI) is similar to the NFI, however, the I FI is relatively independent of sample size. The Comparative Fit Index (GFI) takes the noncentrality parameter into consideration. Again, values greater than .90 indicate a good fit. Noncentrality-based indices. The noncentrality parameters are calculated by subtracting the degrees of freedom for the model from the chi-square («^ - df). 72 Noncentrality measures are based on population discrepancies. That is, in the real world, there are always variables missing from the model, therefore 100% of the variance can really never be explained. These fit indices compare the hypothesized model with a saturated (i.e., best) model limited to the degrees of freedom in the hypothetical model. Bentler's Comparative Fit Index (GFI) and the Steiger-Lind Root Mean Square Error of Approximation (RMSEA) reflect the lack of fit of a model in proportion to the sample size. However, Hayduk (1997) suggests strongly that adhering to theory is of greater importance than arbitrary numerical constructions representing data fit, that is, instead of changing the theoretical model to ‘fit’ data, the data should eventually fit the model. In classical statistics, effect sizes characterize the fit of a model (e.g., a fixed-effects factorial ANOVA model) to data. Similarly, in structural equation modeling (SEM) goodness of fit indices may be thought of as effect sizes. Path model diagrams. The primary question in this study is whether social influences are predictive of alcohol use in the clientele of the Healthiest Babies Possible program. In addition, this study seeks to determine whether clients' personal resources have a mediating effect on the relationship between social influences and risky drinking behaviour during pregnancy. Personal resources, as indicated by self-efficacy and self­ esteem, are thought to mediate the relations between social influences and alcohol use during pregnancy. Figure 7 presents the fundamental structural model used to guide the present study. The individual variables (e.g., family history density, adversity, self-efficacy, etc.) are displayed in the top part of the top part of the figure, which indicates how they are 73 measured and what latent constructs they are associated with. The bottom part displays the proposed model. 74 Results Structural Equation Model - Indicator variables Table 3 presents descriptive statistics for the indicator variables included in the proposed model. Skewness and kurtosis was examined using 'z' formulas for significance testing (Tabachnick & Fidell, 1996, pp. 72 - 73). According to an alpha value of .05, the skewness and kurtosis levels of all indicators in the SEM model were within acceptable limits. The only exception to normality was in the self-efficacy indicator where skewness was significant with a z score of 4.1. Table 3 Descriptive Statistics for Model Variables Mean Model Variables Standard Deviation Range Family History Density 1.8 1.4 0 -4 Adversity 0.8 0.8 0 -2 Self-esteem 0.9 0.7 0 -2 Self-efficacy 0.5 0.7 0 -2 T-ACE Score 2.0 1.6 0 -5 Counsellor Assessment 0.4 0.5 0 —1 Consumption 0.9 0.3 Log Fen The guiding model for this study is presented in Figure 3. SEM procedures allow and encourage the assessment of alternative models. Some procedures compare competing models and choose the model that best fits the data, whereas the present study is only confirmatory. The analysis of the present study follows the structure 75 developed by Baron and Kenny (1986) and elaborated for SEM analysis by Holmbeck (1997). This analytic strategy allows for the testing of direct and mediating influences of predictor variables on outcomes. Table 4 presents the Pearson correlation coefficients among the indicator variables. Indicator variables within each factor are all above .30. Table 4 Pearson's r Correlation Matrix for Indicator Variables. CONCEPT: 1 2 INDICATOR VARIABLES SOCIAL INFLUENCES: 1. FAMILY DENSITY 1 1 2. ADVERSITY .36** PERSONAL RESOURCES: 3. SELF-EFFICACY .14 .42** 4. SELF-ESTEEM .16 .38** RISK: .15 5. T-ACE .36“ .32** 6. ASSESSMENT .31** 7. CONSUMPTION .15 .36** Note: ‘ Correlation is significant, p< .05; “ Correlation is significant, p< .01. 3 4 1 .47** 1 .13 .15 .23** .02 .20* .10 5 6 1 .56** .59** 1 .34** Structural equation models are illustrated as path models. In path model diagrams, circles represent latent variables or constructs, and squares represent the measured or indicator variables. The connecting lines between circles and squares indicate relationships between variables, and the arrowheads indicate the direction of the relationship. Path coefficients that reflect the strength of these relationships are positioned adjacent to the connecting lines. Holmbeck (1997) suggested that four conditions of mediation be tested when 76 using structural equation modelling. These four conditions are sequential and depend on adequate model fit along the way. Holmbeck suggested first assessing the fit of a direct effects model (see Figure 8). Second, a full model should be examined for fit (see Figure 9). The path coefficients of the full model should also be examined. These path coefficients should be significant in all predicted directions. Finally, model fit should be compared when the direct path is constrained to zero (i.e., a mediated model; see Figure 10) versus when it is not (i.e., a Full model; see Figure 9). If the full model does not show a significant improvement over the mediation model, then mediation is present. This final step is equivalent to Baron and Kenny's (1986) regression protocol whereby mediation is present if the significant impact of the predictor variable (i.e, social influences) on the dependent measure, risky behaviour, is lessened after controlling for the mediator (Holmbeck, 1997). , \ PERSONAL ^ RESOURCES ; T-ACE 0.63. SOCIAL INFLUENCES RISKY ALCOHOL BEHAVIOUR 0.77 ASSESSMENT 0.78 0.73/ 0.55 0.53 CONSUMPTION FAMILY ADVERSITY DENSITY e = .47 e = .69 Figure 8 Direct model. e - .44 e = .40 S ELF - SELF­ ESTEEM EFFICACY 0.71 1.78 PERSONAL ; RESOURCES A 0.72 0.20 T-ACE e = .76 ASSESSMENT e = .78 CONSUMPTION e = .26 0.49 SOCIAL INFLUENCES RISKY ALCOHOL BEHAVIOUR 0.86 0.40 0.41 0.84 FAMILY DENSITY ADVERSITY 0.47 e = .77 Figure 9 Full model N 00 e = .46 e = .40 SELFEFFIC A C Y SELF­ ESTEEM t).78 0.7: PERSONAL i RESOURCES A 0.68 .82 0.53 0.43 RISKY ALCOHOL BEHAVIOUR SOCIAL INFLUENCES 0.841 ASSESSMENT Constrained to Zero 0.48 0.4: CONSUMPTION FAMILY DENSITY I a d ver sity ' e = .10 e = .82 Figure 10 Mediated model. N CD 80 In SEM there are a variety of fit measures, most controversial, and all are based on the chi-square statistic or some derivation of it. Overall, results for all three tested models contained several indicators of a working theoretical model. For example, there were no negative error variances, and the path coefficients reflected the intended causal paths. In addition, particular fit indices were very high. On the other hand, certain standardized residuals, the RMR and RMSEA were very high in the full and mediated models. However, since the purpose of this study was confirmatory only, none of the suggested model modifications were done in an effort to statistically 'tweak' the model. Table 5 Values for the Goodness of Fit Indices Calculated by LISREL. DIRECT FULL INDICES OF FIT MODEL MODEL CHI-SQUARE: Model 5.57 36.31 2 9 Degrees of Freedom Ratio: 2.8 4.0 SINGLE SAMPLE FIT INDICES: Absolute Fit Indices .94 Goodness of Fit Index - GFI .99 Root Mean Square Residual .08 .03 RMR RELATIVE FIT INDICES Non-normed Fit Index - NFI .90 .98 .92 Incremental Fit Index - IFI .98 NON-CENTRALITY-BASED FIT INDICES Rentier Comparative Fit Index-CFI .92 .98 .14 Root Mean Square Error of .11 Approximation - RMSEA MEDIATED MODEL 38.67 10 3.9 .93 .10 .89 .92 .91 .14 Direct Model Fit indices for all three conditions of Holmbeck’s procedure are shown in Table 5. The direct model is presented in Figure 8. In the direct model, the direct effects of the 81 social influences factor on the outcome factor of risky behaviour was the only relationship examined. According to Holmbeck (1997), the direct model must fit and the path coefficients must be significant before continuing to look for mediating variables. This model converged in 16 iterations. In the direct model, there was an apparent good fit with a significant path coefficient between social influences and risky alcohol behaviour (r= .78 p, < 001) (see Figure 8). The relationship between social influences and risky drinking behaviour had an value of .61. In other words, social influences explain 61% of the variance in risky behaviour. In addition, the social influences factor explained 53% (r= .73) of the variance of the family history density indicator and 31% (r= .55) of the adversity indicator. The risky behaviour factor explained 39% of the T - ACE score variance, 28% of the consumption indicator, and 59% of the counsellor's decision indicator. The largest standardized residuals for the direct model were between the T-ACE score variable and both family history density and adversity, 2.0 and -2 .0 respectively. Residual values within the ±3 range are considered acceptable. Although the RMR for this model (.03) fell below the .05 criterion, the RMSEA was large at.11. The most important finding in the direct model was that there was a strong significant relationship between the predictor factor, social influences, and the outcome factor, risky drinking behaviour. Full Model The next step in testing for mediation is to test the full model (see Figure 9). This model maintains the direct effect paths, and introduces paths from the social influences factor to the mediating factor, personal resources, and from personal resources to the risky behaviour factor. This model converged in 30 iterations. Although the goodness of 82 fit indices indicate a relatively good fit (i.e., .90 or above) (see Table 5), the introduction of the personal resources factor changed the path coefficients between factors and indicators considerably. In the full model, the relationship between social influences and personal resources was very strong (r= .72, p < .001) and the path from social influences to risky behaviour failed to reach significance despite the fact that the value of the path coefficient was fairly large, r = .40, f (9) = 1.43 ns. Further, the path coefficient from personal resources to risky behaviour also seemed high r = .20, t (9) = .99, ns, yet failed to reach statistical significance. One interesting change was the reversal of variance explained by the social influences factor. In the full model, the social influences factor explains 70% of variance for the adversity indicator and only 23% of the variance of family history density indicator. Conversely, in the direct model the social influences factor explained only 29% of the variance in the adversity variable and 31% of family history density. The personal resources factor accounted for 56% and 60% of self-efficacy and self-esteem respectively. The pattern of variance from the risky behaviour factor to its indicators was similar to the direct model and the values changed only slightly. In the full model, the risky behaviour factor accounted for 74% of the variance in the counsellor assessment, an increase of 15%. On the other hand, there was a decrease in variance accounted for in the consumption and T - ACE score indicators, 22% and 24% respectively. In the full model there were also significantly high residuals. The largest residuals were between the family history density variable and all three of the risky behaviour indicators, the T - ACE score (z = 4.12), the counsellor's decision (z = 3.92), and the consumption indicator (z = 3.29). There was also a significant residual between 83 the adversity and self-efficacy indicators (z = 3.13). The RMSEA (.14) and RMR (.08) values were too large according to accepted criteria of .05 (Hoyle, 1995). High residuals in SEM indicate that more paths need to be added to the model. The modification indices for the full model suggested adding an error covariance between the T - ACE score and the consumption indicator with the self-esteem indicator. Since the intent of this study was to test for mediation and not to compare competing hypothetical models, no modifications were made to the hypothesized full model. Mediated Model In spite of the residual problems in the full model, it may nevertheless prove valuable to carry on with the exercise of evaluating the mediated model. Therefore, the comparison of the full model with the mediated model was also done. The direct paths from social influences to risky behaviour were constrained to zero. The goodness of fit indices were similar to those of the full model and indicated a borderline fit (refer to Table 5). However, in this model the path coefficient between personal resources and risky behaviour was significant. In the mediated model, social influences accounted for 47% of the variance in the personal resources factor and personal resources accounted for 28% of the variance in risky behaviour. There was also a weak indirect effect of social influences on risky behaviour (R^ = .13). There are three residuals with z-values higher than 3. Again, the highest residuals are between the family density indicator and the all the risky behaviour indicators, the T - ACE score (z = 4.40), the counsellor's decision (z = 4.38), and the consumption indicator (z = 3.56). The RMSEA (.14) and RMR (.10) are very large compared to acceptable criteria. Finally, LISREL provides modification indices based on 84 error covariance. These SEM statistics point to alternative paths for the model. In this case, the modification indices suggest an error covariance be added from the self­ esteem indicator and the consumption indicator. Since the goal was to test a theoretical mediation model, no exploratory analysis was carried out. According to Holmbeck's (1997) four conditions, full mediation of the relation between social influences and risky alcohol behaviour during pregnancy by personal resources did not occur in this data set. However, there are two critical premises of mediation shared by Holmbeck (1997) and Baron and Kenny, (1986). First criteria is a good model fit according to the goodness of fit indices. Secondly, there must be a significant direct relationship between social influences and risky behaviour, and this direct relationship diminishes when the mediating factor personal resources is entered into the model. For this data, the direct relationship was highly significant, and this direct relationship diminished when the mediating variable is entered into the model, however the full and mediated models failed to obtain a good fit to this data. 85 Discussion The outcomes of this study were in some ways consistent and in other ways inconsistent with expectations. As expected, social influences, as indicated by clients’ reports of parental drinking and a history of abuse and neglect had a direct effect on the index of risky drinking behaviour provided by the Healthiest Babies Possible counsellors. In the direct model, social influences of an adverse nature explained most of the variance in risky behaviour. These social influences were also found to predict the level of clients’ personal resources, as indicated by their self-efficacy and selfesteem. Clients experiencing higher levels of negative social influences were more likely to suffer from deficits in personal resources and to exhibit risky drinking behaviour. However, findings that were inconsistent with expectations emerged from the SEM analysis. In particular, it was expected that personal resources would mediate the effects of negative social influences on drinking behaviour. The outcome of the SEM analysis did not support this expectation within the strict guidelines set by Holmbeck (1997). However, the data did provide sufficient collateral support for the expectation that negative social influences and deficits in personal resources would be associated with risky drinking behaviour. The primary objective of this project was to test a relatively simple theoretical model of alcohol use among the clients of the Prince George Healthiest Babies Possible program. Although many of the fit indices used to assess a structural equation model indicated an acceptable fit for the full and mediated models, the residuals and RMSEA values were unacceptably high. According to Hoyle (1995), high RMSEA values may be the result of indicator distributions, scale size, and/or model 86 misspecification. Although there were no real departures from normality, most indicators in the model were based on small three point scales. Larger scales are better suited to structural equation modeling. A high value on the RMSEA index also indicates that the overall model is low in statistical power (Hoyle, 1995). Many researchers support the all or nothing assessment of structural models. That is, the model passes or fails according to strict cut-off criteria. Others, such as Hayduk (1997), suggest further examination and more latitude from a substantive perspective. The present discussion favours Hayduk's perspective. According to the goodness of fit indices, the direct model fits the data. As expected, in the direct model, there was a significant direct effect between a history of living in an at-risk home environment (i.e., social influences) and increased risk of alcohol use. However, in the full model, this relationship became non-significant and the residuals became significantly high which suggests model misspecification. In addition, the direct path from social influences to personal resources was the only statistically significant path coefficient in the full model. This may indicate that the negative social influences faced by these clients have a greater direct effect on their deficits in personal resources than directly on alcohol abuse. For the mediated model the RMSEA was very high again indicating a less than adequate fit. The path coefficients between social influences and personal resources and between personal resources and risky behaviour were significant in this model. In terms of model parameters, one could say that the mediated model worked better with these data than the full model. Since the main criterion of mediation is a reduction in the strength of the relationship between the predictor and outcome variables in the 87 presence of a third variable, one could loosely conclude that personal resources do indeed alter the relationship between social influences and risky drinking behaviour. However, examination of residuals and modification indices suggests that there may be changes required to the model itself. Examination of modification indices provided by LISREL software suggested that there may be other important relationships between the variables under study. Theoretically, the suggestions made sense in that the modification indices pointed to a very strong error covariance between the self-esteem indicator and the consumption variable. This relationship would present a recursive aspect to the model, where low levels of self-esteem are seen not only as a result of poor social influences but also as a result of excessive consumption of alcohol. It is understandable that the consequences of excessive alcohol use itself could decrease self-esteem, particularly during pregnancy when alcohol use is viewed as socially unacceptable. Analysis of the residuals suggests that there are other recursive aspects to this model. The high residuals between the family history density indicator and all of the alcohol indicators suggested that the women in this sample who were at risk for alcohol abuse were more likely to keep the company of other risky drinkers. Changes suggested to the model by the residuals and the RMSEA make sense theoretically. Nyamathi et al. (1999) and Quigley and Collins (1997) report strong support for the effects of family environment and simple modeling on alcohol use behaviour. Future investigation into these recursive relationships for this particular group of women may prove valuable. 88 Comparison to Nyamathi's Constructs This study examined a conceptual model of risk factors for alcohol use during pregnancy using concepts similar to that of Nyamathi et al. (1995) - placed within the simple structure of social learning theory. One major difference between Nyamathi's (1995) model and the model under study here was that Nyamathi's model reflects positive social influences (i.e., social support), whereas the social influences factor in the present study reflects detriments in the home environment (i.e., abuse and high density of drinking behaviour). Nyamathi (1995) found that increased social support was associated with increased self-esteem, decreased emotional disturbance, increased use of active as opposed to avoidant coping strategies, and ultimately less risky behaviour. The direct model best illustrates how a history of abuse or neglect coupled with a high density of family drinkers leads to increased likelihood of alcohol consumption during pregnancy. Despite the fact that the hypothesized model failed to fit the data satisfactorily according to the aforementioned residuals and RMSEA criteria, examination of the first order correlations adds further support to the overall theoretical utility of the model. The strong significant relationships (Cohen, 1977) between the family density score and all alcohol use indicators support Nyamathi et al.'s (1999) conclusion that role models and influences within a family are extremely important in the development of a woman's pattern of substance use, abuse, or abstinence. Despite the significant influence of role models on client drinking behaviour, there was no direct relationship between family history density and either self-efficacy or self-esteem. In other words. 89 according to present data, simply being surrounded by a high density of alcohol abusers does predict risky alcohol use, but does not have an effect on the individual's psychological resources. In isolation, this finding gives strong support for the effects of modelling. However, as expected, the family history density indicator was strongly related to the adversity indicator. These results support Califano's (1999) assertion that children of parents who abuse alcohol are also more likely than parents who do not abuse alcohol to have experienced traumatic or neglectful family environments, but most likely, because "adversity" includes current domestic violence. Although the adversity indicator in the present study is in contraposition to Nyamathi et al's (1995) social resources construct, in that it reflects relative lack of past or present positive social support or even deprived conditions, the relationships are still comparable. In Nyamathi et al.'s (1995) model, social support was positively related to self-esteem, and inversely related to emotional disturbance and avoidant-style coping. Further, emotional disturbance and avoidant-style coping were positively related to risky behaviour. Similar relationships were found in this study sample. The adversity indicator was strongly related to both the self-efficacy and self-esteem indicators. Moreover, the adversity indicator was also strongly and directly related to the counsellor's decision as to whether a client should be classified as at risk for alcohol misuse (Cohen, 1977). There were no significant relationships between adversity and the T - ACE score or consumption. Although family density and adversity were highly correlated, and both had influence on alcohol use, they seem to work somewhat independently. In other words, alcoholic modelling may predict alcohol use, but adversity may influence both alcohol use and psychological deficits. This suggests that the concept of adversity 90 should be evaluated as a mediating variable, or perhaps set apart as a separate construct. The biggest difference between Nyamathi et al's (1995) approach and this study is that the present study is basically reactive and looks at negative social influences in an attempt to predict the risk of alcohol misuse. On the other hand, Nyamathi et al's approach is more proactive. Although her model is also intended to be predictive, she focuses on how positive social support can help reverse the negative cycles of low self­ esteem, self-efficacy or general coping strategies and ultimately, risky behaviour. Overall, the similarities in the variable relationship patterns of between Nyamathi et al.'s (1995) study and the present study suggest that recommendations made by Nyamathi (1989,1995, & 1999), may also be relevant for the clients of the Healthiest Babies Possible Program. Clients of the Healthiest Babies Possible Program The demographic profile for the sample of women in the present study, suggested that the Healthiest Babies Possible Program is serving their target population. The majority of these women are from very low-income backgrounds, have attained relatively low education levels, are victims of trauma, and suffer from poor health. Most of the clients have low self-esteem, report feelings of an inability to cope with their pregnancy, show resistance to seeking appropriate social services, and often use and abuse substances during their pregnancy. Decreased self-esteem is one of the most common results of living in an abusive or neglectful environment (Malinosky-Rummel & Hanson, 1993). For these women, having little or no source of income, minimal education, being a single parent, and 91 having concerns for housing must exacerbate low self-esteem, feelings of helplessness, unworthiness and perceptions of external threat. On the other hand, as per Nyamathi (1989), if the self-esteem levels in these women could be improved, selfefficacy in certain situations may also improve in a recursive manner. According to Nyamathi, self-esteem levels increase as social support increases. Since the staff within the Healthiest Babies Possible program believe in a holistic approach with their clients, their first priority should be focussed on providing social support aimed toward building higher self-esteem levels and teaching active coping strategies. Almost 40% of these women reported feeling isolated, either socially or culturally. In an effort to teach clients certain basic skills, the Healthiest Babies Possible program provided a cooking class for their clients. There were additional benefits to this class. Counsellors noted that this class turned out to be markedly more important for the social benefits achieved than for the actual cooking lessons. The women shared their stories and advice with one another, but most important, the cooking class provided social support. Within the context of the Healthiest Babies Possible program, a heavy presence of alcohol abusers in a family may predict alcohol use, but a history of abuse and/or neglect also predicts the extent to which psychological wellness is achieved. Since the clients of this program are future mothers, their psychological wellness should be a top priority. Teaching life skills and facilitating independence through practical programs such cooking, sewing and child care classes should increase a client's ability to cope with the pregnancy and infant. As noted above, self-efficacy (i.e., the ability to cope) is strongly related to self-esteem. 92 Finally, clients suffering from the effects of abuse and/or neglect should be provided psychological counselling. People can physically remove themselves from a heavy drinking environment. However, the recursive relationship between self-efficacy, low self-esteem and alcohol misuse can be deadly. Recall the multidimensionality of alcohol abuse in women. Alcoholic women are more likely than men to suffer from depression, exhibit ineffective coping strategies, suffer low self-esteem and anxiety, as well as other affective and compulsive disorders, such as, bulimia or anorexia nervosa. Women who abuse alcohol are also more likely to abuse more than one other substance, including cigarettes and a variety of prescription drugs (Beckman, 1994; Blume, 1990; Gomberg, 1993; Wilsnack & Wilsnack, 1990; Yaffe, Jenson, & Howard, 1995). Further, the Straus and Kantor (1994) study found that those who experienced corporal punishment as children were significantly more likely to suffer from depression, experience suicide ideation, develop problems with alcohol, and/or abuse their own children or partners. Limitations The objective of this study was to determine whether personal resources mediate the relationship between environment and the risk for alcohol consumption. Some support for this model was established, however, personal resources did not appear to mediate the influence of a poor social environment according to Holmbeck's (1997) criteria. On the other hand, the hypothesized model in this study is extremely simple and basic. The general patterns of the model path coefficients are reasonable and are further supported by the relationships among the indicator variables. The data used for an analysis such as testing a structural equation model should 93 be of much better quality than data collected by an intake assessment questionnaire. However, using the longer, albeit more appropriate measures, such as Coopersmith's (1967) Self-esteem Inventory (Nyamathi, Wayment et al., 1993), or the Ways of Coping (Revised) scale (Folkman, Lazarus, Dunkel Schetter, Delongis, & Gruen, as cited in Easley & Epstein, 1991) would heavily tax already stressed clients. In addition, the Healthiest Babies Possible program assists approximately 150 clients per year. In light of the high amount of data already missing from the collection of intake information and, considering that the clients would have the right to refuse to participate in a survey study, the collection of enough data would take years (Browne, 1998). Conclusion Issues To Consider In Future Research According to the research, excessive alcohol use and continued alcohol use during pregnancy among the clients of HBP is strongly influenced by their social environments. Therefore, investigating the psychological and social conditions that these women face is very important in understanding why certain women may drink to excess. This study examines two social influences considered to have a negative effect on psychological resources and subsequent drinking behaviour. According to Nyamathi’s (1989) paradigm, the client-centred, supportive social environment provided by the HBP program may be having very positive psychological and behavioural effects. Therefore, future research should focus on measuring the outcomes of their programs. Finding out what type of interventions are effective and to what to degree, would help the progress of intervention research. The 1996 version of the intake assessment questionnaire (lAQ) for the Prince 94 George HBP has been revised at least once since the present study has been completed. Even without the revision, the lAQ provides adequate information for researchers to investigate circumstances surrounding this special population of women. In addition, there is a client monitoring section and a pregnancy outcome section within the lAQ. The client monitoring section tracks information on the client's nutrition, their use of substances (e.g., tobacco, alcohol and other drugs), number of visits to HBP and their medical progress through the pregnancy. Therefore, new research is needed that compares the intake and intervention variables to these available outcome measures. The outcome measures could be used for evaluation research projects. For example, future research could examine the relationship between the number of visits and interventions provided and a reduction in drinking. Further, the intake, intervention and outcome information should be compared across years of data compiled within the Prince George HBP program. This type of research would reveal trends in clients and provide a baseline for evaluation research and support the reliability of the intake tool. The lAQ must not only be comprehensive but non-threatening and clientcentred. In addition, the lAQ should also be conducive to doing research, however, not to the invasion of the client’s privacy. This can be a difficult balance to achieve. For example, for the purposes of research, in the prenatal risk factors section of the lAQ, one could ask the women not only whether they suffered from an inability to cope, but to what degree. Although this change to the lAQ may produce a continuous measure convenient to researchers, it effectively doubles the number of questions the clients are required to answer. As an alternative, clarifying current items on the lAQ and using creative scaling 95 would provide useful variables for research analysis - and remain respectful of patients. For example, the lAQ item asking clients if they had a history of abuse and/or neglect could be broken into separate items, one for past abuse and another for neglect. Additional items may have to be added in order to identify specific concerns. For example, include separate items that inquire into the client's history of childhood emotional abuse, physical abuse, sexual abuse, and/or history of child neglect. Certainly, the addition of clearer and more direct yes/no items would allow for a variety of derived variables, which is more informative to researchers and less invasive to clients. In other words, the 'creative' scaling (i.e., combining appropriate dichotomous variables) used for this study may be used to investigate other issues of concern to HBP. Separating depression from the general mental illness item on the lAQ is particularly important with regard to alcohol abuse (Gomberg, 1994). One goal of this thesis was to learn the basic concepts of a novel, somewhat complex statistical technique, structural equation modelling. Structural equation modelling allows researchers to test social models of behaviour. Another goal of this project was to apply this advanced statistical technique to data from a special population. Although the present study only tested a relatively simple theoretical model, future research should test explanatory models. Results from this analysis indicate that expansions to future models on alcohol use, social and personal factors should include recursive aspects to the social models tested. For example, the recursive nature between alcohol use and self-esteem or family drinking could be investigated. Structural equation modelling can also be employed as an evaluation technique. For example, using the aforementioned client monitoring section of the lAQ, the 96 differences in outcome substance consumption levels could be placed within a model of various combinations of interventions and risk factors. For example, treatment program 'A' may be more effective for women with histories of abuse, whereas treatment B' may be more effective for women living in a high density alcohol environment. One limitation of using SEM is that a sample size of 150 minimum. Although the Healthiest Babies Possible program serves over 150 women per year, evaluation research usually consists of at least 2 experimental groups which would require data for at least 300 clients. On the other hand, comparisons from year to year would be possible. Finally, a qualitative study similar to Nyamathi et al.'s (1999) study may prove invaluable in gaining a deeper understanding of the issues the at-risk women have to face everyday. However, instead of burdening the clients, perhaps the peer counsellors could be interviewed. Due to the fact that HBP is designed to help women who are at-risk, the intake assessment questionnaire tends to seek out problems that may need addressing. Thus, the social influences and psychological variables in this study measure problems or deficits. According to Nyamathi (1989), the British Columbia FAS report (BO Children & Families, 1998), and the very recent and comprehensive Health Canada report (2001), the most important goal for any program attempting to provide prenatal service to at-risk women is social support. The Healthiest Babies Possible program in Prince George exemplifies the recommendations of the aforementioned report. HBP provides a single access to a comprehensive system of social support. Another successful strategy used by HBP is the employment of peer counsellors. Non-professionals are perceived as less 97 threatening and provide a much more comfortable environment. A comfortable environment is primary because you cannot help someone who will not return. Once clients are comfortable, HBP provides prenatal education: education in nutrition, phases of pregnancy and the risk factors to their pregnancy. Counsellors also provide advocacy to other community agencies that may serve individual clients and remove barriers to treatment. Findings from this study, which may build on the existing strengths of HPB, indicate that providing support for the whole family, and extending the length of support for the mother after delivery of the baby would be advantageous. For example, it would be difficult for a woman to abstain from alcohol when her partner keeps drinking around her and/or they are having relationship problems. In addition, several well-respected, well-researched prenatal projects extend the social support long past the delivery of the baby. Seattle has a program of support until the child is 3 years old (Streissguth, 1997). Their program focuses on supporting both the mother and the child. In New Zealand, the Plunket Society provides support for mother, babies and preschool children. This program provides parenting education and support through office visits, home visits and courses on cooking for healthy children (Plunket Society, 2002 Retrieved from www.plunketsociety.com). Providing guidance aimed at helping the entire family through the crucial preschool years has proved very successful in New Zealand. Extending support for the mothers not only helps them with the new baby, but is also beneficial to the rest of the children. In terms of alcohol misuse, the risk of relapse would be reduced if the social support is maintained. As a result, children see less modeling of alcohol misuse, and 98 family dysfunction may also be reduced. In turn, the risks for alcohol misuse are less for the next generation. In conclusion, this study tested a social model of relationships between the social and psychological risk factors of a special population and their degree of risk for alcohol use during pregnancy. Although the data did not support the stated hypothesized model, relationships between psychological and social risk factors and risk for alcohol use during pregnancy were present for these clients. The results of this study contribute to the scarce research focusing on pregnant women and alcohol by building on earlier research, and providing several implications for the special population enrolled in the Healthiest Babies Possible program. Not only is there a scarcity of research with respect to this topic, there is even less research that attempts to examine unique female populations. The results from this study support the idea that there is an important connection between social influences and alcohol use. The ability to cope well and have a positive perception of oneself are valuable skills when having to deal with the strong influence of others' negative behaviours, particularly, when interacting with family members, peers and partners. A positive self-image allows one to think for oneself and have the confidence to follow one’s own decisions, regardless of the behaviour and opinions of others. Finally, the Healthiest Babies Possible program is successfully providing single access to a broad range of social support for pregnant women from disadvantaged environments in Prince George. 99 References Bandura, A. (1969). Principles of Behavior Modification. New York: Holt, Rinehart & Winston. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51 (6), 1171-1182. BC Children's Commission (1999). Press Release Beckman, L. J. (1994). Treatment needs of women with alcohol problems. Alcohol Health and Research World, 18 (3), 206-211. Blume, S.B. (1990). Chemical dependency in women: Important issues. American Journal of Drug & Alcohol Abuse, 16 (3-4), 297-307. British Columbia, Ministry of Children and Families. (1998). Community Action Guide: Working Together for the Prevention of Fetal Alcohol Syndrome. Publication Number: C98-960050-5. Browne, A. (1998). An Evaluation Of Solution Focused Approaches To Smoking Reduction With Pregnancy Outreach Clients. Health Canada, Project #65542826. Browne, A. J., Thio-Watts, M. & Spaulding, L. (1996). Intake Assessment Questionnaire for the Prince George Healthiest Babies Possible Program. Prince George, BC. Burgess, D. M. & Streissguth, A. P. (1996). Fetal Alcohol Syndrome and Fetal Alcohol Effects: Principles to educators. In G. Schmidt & J. Turpin (Eds.), Fetal Alcohol Syndrome/Fetal Alcohol Effects: A Resource Manual (pp. 45-51). Prince George, BC, University of Northern British Columbia, Child Welfare Research Centre. 100 Califano, J. A. Jr. (1999). The least among us: children of substance abusing parents. America, 14, 10-12. Canadian Centre on Substance Abuse, Canadian Profile 1999. Alcohol Highlights. www.ccsa.ca/cp99alc.htm. Cohen, J (1977). Statistical power analysis for the behavioral sciences. New York: Academic Press. Cooper, M. L., Russell, M., & George, W . H. (1988). Coping, expectancies, and alcohol abuse: a test of social learning formulations. Journal of Abnormal Psychology, 97 (2), 218-230. Cooper, M. L., Russell, M., Skinner, J. B., Prone, M. R., & Mudar, P. (1992). Stress and alcohol use: moderating effects of gender, coping and alcohol expectancies. Journal of Abnormal Psychology, 101 (1), 139-152. Easley, M. J., & Epstein, N. (1991). Coping with stress in a family with an alcoholic parent. Family Relations, 40 (2), 218-224. El-Guebaly, N. (1995). Alcohol and polysubstance abuse among women, Canadian Journal of Psychiatry, 40 (2), 73-79. Fleming, J., Mullen, P. E., Sibthorpe, B., Attewell, R., & Bammer, G. (1994). The relationship between childhood sexual abuse and alcohol abuse in women: a case-control study. Addiction, 93(12), 1787-1799. Gomberg, E.S. (1993). Recent Developments in Alcoholism: Gender issues. Recent Developments in Alcoholism, 11, 95-107. Gomberg, E. S. L. (1994). Risk factors for drinking over a woman’s life span. Alcohol Health & Research World, 18 (3), 220-227. 101 Gove, T. (1995). Mathew’s Legacy: Report of the Gove Inquiry into Child Protection^, British Columbia: Ministry of Social Services. Hayduk, L. A. (1987). Structurai equation modeling with LISREL: essentials and advances. Baltimore : Johns Hopkins University Press, cl 987. Hayduk, L. A. (1996). LISREL issues, debates, and strategies. Baltimore : Johns Hopkins University Press. Health and Welfare Canada. (1993). Prenatal Health Promotion, Project Report. Health and Welfare Canada. Health Canada (1995). Hewitt, D., Vinje, G., & MacNeil, P. (Eds.). Horizons Two: Canadian women’s alcohol and other drug use: increasing our understanding. Health Canada Cat. No. H39-307/2-1996E; Published under Canada’s Drug Strategy. Health Canada (2001). Best Practices: Fetal Alcohol Syndrome/Fetal Alcohol Effects and the Effects of Other Substances During Pregnancy. Health Canada Cat. No. H49-156/1-2001E Holmbeck, G. N. (1997). Toward terminological, conceptual, and statistical clarity in the study of mediators and moderators: examples from child-clinical and pediatric psychology literatures. Journal of Consulting and Clinical Psychology, 65 (4), 599-610. Hoyle, Rick. H. ed. (1995). Structural equation modeling: concepts, issues, and applications. Thousand Oaks : Sage Publications. Kelloway, E. K. (1998). Using LISREL for Structural Equation Modeling: a Researcher’s Guide 102 Langeland, W, & Hartgers, C. (1998). Child sexual abuse and alcoholism: a review. Journal of Studies on Alcohol, 59, 336-348. Lazarus, R., & Folkman, S. (1984). Stress, Appraisal and Coping^ Springer, New York, pp. 177-214. Lex, B. W. (1994). Alcohol and other drug abuse among women. Alcohol Health & Research World, 78(3), 212-219. Little, B. B., Snell, L. M., & Rosenfeld, 0 ., Gllstrap, L., & Gant, N. (1990). Failure to recognize Fetal Alcohol Syndrome In newborn Infants. American Journal of Diseases in Children, 144, 1142-1146. Malsto, S. A., Carey, K. B., & Bradlzza, C. M. Social Learning Theory^ In Leonard, K. E., & Blane, H. T. (Eds.) Psychoiogical Theories of Drinking and Alcoholism, (2nd ed.). (1999) pp. 106-163. Mallnosky-Rummell, R., & Hanson, D. J. (1993). Long-term consequences of childhood physical abuse. Psychological Bulletin, 114 (1), 68-79. Nyamathi, A. (1989). Comprehensive health seeking and coping paradigm. Journal of Advanced Nursing, 14, 281 -290. Nyamathi, A., Bayley, L., Anderson, N., Keenan, C., & Leake, B. (1999). Perceived factors Influencing the Initiation of drug and alcohol use among homeless women and reported consequences of use. Women & Health, 2 9 (2), 99-114. Nyamathi, A., Stein, J. A., & Brecht, M. (1995). Psychosocial predictors of AIDS risk behavior In homeless and drug addicted women of color. Health Psychology, 14 (3), 265-273. Patterson, D. G. (1995). Alcoholism: treating ‘the second sex’. Irish Journal of 103 Psychological Medicine, 12 (2), 46-47. Plunket Society (2002). The Plunket Society Home Page. Retrieved June 20, 2002. from http://www.plunket.org.nz Quigley, B. M. & Collins, R. L. (1997). The modeling of alcohol consumption: a metaanalytic review. Journal of Studies on Alcohol, 60, 90-98. Robbins, 0 . (1989). Sex differences in psychosocial consequences of alcohol and drug abuse. Journal of Health and Social Behavior, 30(1), 117-130. Russell, M. (1994). New assessment tools for risk drinking during pregnancy. Alcohol Health and Research World, 73(1), 55-61. Schmidt, G., & Turpin, J. (1996). Fetal Alcohol Syndrome/Fetal Alcohol Effects: A Resource Manual. Child Welfare Research Centre, Faculty of Health and Human Sciences, University of Northern British Columbia, June 1996. Single, E. W., MacLennan, A., & MacNeil, P. (1994). Horizons 1994: Alcohol and Other Drug Use in Canada. Studies Unit, Health Promotion Directorate, Health Canada; and the Canadian Centre on Substance Abuse. Smitherman, C. H. (1996). In G. Schmidt & J. Turpin (Eds.), Fetal Alcohol Syndrome/Fetal Alcohol Effects: A Resource Manual (pp. 5-11). Prince George, BC, University of Northern British Columbia, Child Welfare Research Centre. Spaulding, L. (1996). [Annual Report: Healthiest Babies Possible, Prince George, BC.] Unpublished raw data. Statistics Canada (1999). Control and sale of alcoholic beverages. The Daily. Thursday, June 22, 2000. Stoltenberg, S. F., & Mudd, S. A. (1998). Evaluating measures of family history of 104 alcoholism: Density versus dichotomy. Addiction, 93(10), 1511-20. Straus, M. A., & Kantor, G. K. (1994). Corporal punishment of adolescents by parents: a risk factor in the epidemiology of depression, suicide, alcohol abuse, child abuse, and wife beating. Adolescence, 29 (115), 543-562. Streissguth, A. (1997). Fetal Alcohol Syndrome: A Guide for Families and Communities. Baltimore, MD: Paul H. Brooks Publishing. Tabachnick, B. G., & Fidell, L. S. (1996). Using Multivariate Statistics (3rd ed.). New York: Harper Collins. Tanaka, J. S. (1993). Multifaceted conceptions of fit in structural equation models. In K. A. Bollen, & J. S. Long (eds.), Testing structural equation models. Newbury Park, CA: Sage. Tweed, S. H., & Ryff, C. D. (1991). Adult children of alcoholics: profiles of wellness amidst distress. Journal of Studies on Alcohol, 52 (2), 133-141. Widom, C. S., Ireland, T., & Glynn, P. J. (1995). Alcohol abuse in abused and neglected children followed-up: Are they at increased risk? Journal of Studies on Alcohol, 56, 207-217. Wilsnack, S. C., & Wilsnack, R. W. (1990). Women and substance abuse: research directions for the 1990s. Psychology of Addictive Behaviors, 4 (1), 46-49. Wilsnack, S. C., & Wilsnack, R. W. (1994). How women drink: epidemiology of women’s drinking and problem drinking. Alcohol Health & Reserarch World, 18 (3), 173-181. Wilsnack, S. C., Vogeltanz, N. Klassen, A. D., & Harris, T. R. (1997). Childhood sexual abuse and women’s substance abuse: national survey findings. Journal of 105 Studies on Alcohol, 58, 264-271. Yaffe, J., Jenson, J. M., & Howard, M. O. (1995). Women and substance abuse: implications for treatment. Alcoholism Treatment Quarterly, 13 (2), 1-15. 106 Appendix Intake Assessment Questionnaire 107 B.c. Pregnancy Outreach Program INTAKE & EPI - INFO DATA BLANK FORM for USE IN INTAKE INTERVIEW Revised June 27,1996 Revised by: A. Browne R.N., M.S.N. Marlene Thio-Watts R.N., B.N. L. Spaulding B.A. M.Sc. Candidate Ilona Schaufler B.A. M.Sc. Candidate This Project was supported by a contribution received from the Community Action Initiatives Program, Tobacco Demand Reduction Strategy, Health Canada. 1. INDIVIDUAL PRENATAL RISK IDENTIFICATION (PR) 108 Yes No N/A No! Assessed PR 1 Physical Factors PF a) Previous pregnancy loss o a a o □ o PF b) Illness / condition with Impact on pregnancy a o PF o) Pre-pregnancy weight - body mass index (BMI) o o PF d) Rate of weight gain a o PF e) Inadequate nutrition a □ PF f) Previous child with anomaly a o o a PF g) Previous child requiring neonatal intensive care o o □ a PF h) Multiple pregnancy a a a PF i) Birth interval a □ a a a PF j) Grand multipara - 5 or more a o a a PF k) Established genetic risk a □ o a a a a o o a □ □ a a □ □ □ a □ □ □ □ □ a □ a □ a o PE a) Single parenthood □ □ a a PE b) Delayed access to prenatal care PE c) Refusal of / resistance to appropriate services PE d) Isolation - ethnic, language & social PE e) Limited learning ability / illiterate PE f) Marital problems / unstable relationship / family violence PE g) Mental health problems PE h) Low self-esteem PE i) Inability to cope / anxiety regarding pregnancy and baby PE j) Unrealistic expectations PE k) Unwanted pregnancy / denial of pregnancy PE 1) Financial problems PE m) Inadequate housing PE n) Family history of abuse / neglect □ a o o □ □ □ □ □ □ n □ □ □ □ a □ a □ □ □ □ a □ n □ a □ a □ □ □ o □ □ □ □ □ o n o □ □ a □ □ □ □ o a PF 1) Age 17 and younger / 36 and older PR 2 Substance Abuse / Misuse SA a) Cigarette smoking SA b) Alcohol use SA o) inappropriate use of over the counter and prescription drugs SA d) Other drug use o PR 3 Psychosocial & Economic Factors n □ 2. REFERRAL DATA (RD) RD 1. Referral Date (D D /m m a t) RD 2. Source of Referral (Check All That Apply) a) Health Unit O Yes a No b) Physician o Yes o No c) Social Services a Yes o No d) Alcohol & Drug □ Yes □ No e) Community Group a Yes □ No f) Self o Yes □ No g) Detox a Yes o No h) Needle Exchange o Yes □ No i) Street Workers □ Yes a No i) Other (soecifv) RD 3. Weeks Gestation at Referral: RD 4. Intake Assessment Date / / (DD/MM/YY) RD 5. Due Date: __ (DD/MMAA') RD 6. Client Began Program (Yes indicates Client W as Enrolled/intake Assessment W as Completed. ) □ Yes □ No RD 7. If No, Why Did Client Not Begin Program? a) □ Not High Risk b) □ Reason: (Check Only One) □ Refused/Not Interested □ Moved □ Not Pregnant □ Pregnancy Terminated Before Assessment □ Unable to Contact c) □ Other (specify):_________________ 109 110 3. CLIENT CHARACTERISTICS (CC) CC 1. a) Client’s Date of Birth : ( D D /M M A 'Y ) _ /_ _ Z _ b) Age at Due Date:_ CC 2. Marital Status (Check Only One) □ Married □ Common Law □ Single □ Relationship □ Separated □ Widowed □ Don’t Know □ Divorced CC 3. First Language * a) English □ Yes □ No b) Other (specify ) ___________________ 4c) If Ethnic Background is First Nations, Select Status £7 On Reserve £7 Off Reserve £7 fi/letis □ Non-status Indian £7 Unknown CC 4. Ethnic Background a) n Caucasian (Not Identifiable) □ Vietnamese O First Nations (Go to 4 c) □ Latin American n Indo-Canadian b) Other (specify) □ Chinese _________________ CC 5. Education □ Grade 8 Or Less □ Grade 9 - 1 1 □ Completed Grade 12 □ Some Post-secondary □ University Graduate/College CC 6. Employment Status (Check All That Apply) □ Student □ Homemaker □ Unemployed □ Employed Occupation (specify):___________________________ CC 7. Financial Situation □ □ No Income □ Receiving Income Assistance □ Low/inadequate Income, But No Social Assistance Not Low Income/Adequate □ Dependent on Parents CC 8. Spouse’s/partner’s Financial Situation □ No Income □ Receiving Income Assistance O Low/inadequate Income, But No Social Assistance □ Not Low Income/Adequate □ Dependent on Parents □ N/A First Language: most comfortable language for communicating. Ill 4. OBSTETRICAL DATA (OB) OB 1. Number Of: a) Total Pregnancies* b) Pre Term Deliveries (<37 Weeks) c) Term Deliveries (> 38 Weeks) d) Elective Abortions e) Spontaneous Abortions (< 20 Weeks) f) Living Children g) Stillbirths h) Low Birth Weight Babies (< 2500 gms or 5 ibs, s oz) * Including current oreanancv/past miscarriaaes/past abortions. 0 8 2. Attended Prenatal Classes During A Previous Pregnancy? □ Yes □ No □ Not Applicable 0 8 3. During A Previous Pregnancy, Has Client EverBeen A POP Client? □ Yes □ No □ Not Applicable 112 4. OBSTETRICAL DATA (OB) a) b) c) d) e) f) g) h) OB 1. Number Of: Total Pregnancies* Pre Term Deliveries (^ 7 Weeks) Term Deliveries (k 38 Weeks) Elective Abortions Spontaneous Atwrtions (< 20 Weeks) Living Children Stillbirths Low Birth Weight Babies (<2500 gms or 5 ibs, 8 oz) * Including current preanancv/past miscarriages/past abortions. OB 2. Attended Prenatal Classes During A Previous Pregnancy? a Yes O No O Not Applicable OB 3. During A Previous Pregnancy, Has Client Ever Been A POP Client? O Yes O No O Not Applicable 113 6. ALCOHOL DATA (AD) AD 1. History of Misuse By: (Check All That Apply) a) Siblings O Yes O No O Don't Know a N/A b) Biological Mother O Yes a No 0 c) Biological Father a Yes o No d) Spouse/partner/boyfriend a Yes 0 e) Extended Family Members O Yes o No No Don't Know O N/A o Don't Know a N/A o Don't Know O N/A o Don't Know O N/A f) Other Caretakers (While growing up) O Yes a No a Don't Know a N/A g) Adoptive Mother O Yes a No □ Don't Know O N/A h) Adoptive Father 0 Yes o No O Yes O No (If No, Go To Drug History' Sec. 7) o Don't Know O N/A AD 2. History of Alcohol use in lifetime: AD 3. Age Started Drinking Alcohol AD 4. Triggers (Check All That Apply) a) Social O Yes 0 No b) Emotions o Yes o No c) Stress a Yes a No d) To Relax o Yes a No e) When Bored o Yes a No 0 When Having Fun 0 Yes a No 9) When Smoking 0 Yes o No h) When Family Drinks Around Me a Yes o No i) When Friends Drink Around Me a Yes o No j) When Boyfriend Drinks Around Me a Yes o No k) Other (specify) 6. ALCOHOL DATA (AD) (confd) 114 AD 5. Coping Methods Most Effective For Avoiding Aicohoi in the Past (Check Aii That Apply) a) Avoid Places 1am Likely to Drink a Yes O No b) Get Support of Friends o Yes O No c) Get Support of Family 0 Yes O No d) Awareness of Triggers/temptations a Yes a No e) Substitute With Non-alcoholic Beverages 0 Yes a No f) Learned Self-control/stop At 1 Drink a Yes a No 9) "Not a Problem for Me" (client feels they can stop any time) 0 Yes No h) Change Routines/Keep Busy a Yes o o No a No o o o No a No o No i) Think of Benefits of Cutting Down or Quitting j) Stay Away From Stressful Situations k) Exercise 1) Try to Cut Down m) Try to quit n) Quit Cold Turkey o Yes o Yes a Yes o Yes o Yes o Yes No No o) Other AD 6. T- Ace Score (specMy)_ T • ACE Scoring Key: 1. How many drinks does it take to make you feel high? (Tolerance) 0 less than or equal to 2 drinks 2 more than 2 dinks 2. Have people Annoyed you by oriticBing your drinking? 0 no 1 yes 3. Have you felt you ought to Cut down on your drinking? 0 no 1 yes 4. Have you ever had a drink fkst thing in the morning to steady your nerves o r to get rid o f a hangover? (Eye Opener) 0 no _____ I_____ _____________________________________________________________________ AD 7. Drinking Patterns (1 drink = 1 Oz. Hard Liqueur = 5 Oz. Wine = 12 Oz. Beer) a) How Many Drinks Does it Take For You to Feel The Effects of Alcohol?_______ b) How Many Drinks Can You Hold Until You've Had Too M uch?_________ 115 6. ALCOHOL DATA (AD) (confd) AD 8 (F). Drinking Frequency (1 drink = 1 Oz. Hard Liqueur = 5 Oz. Wine = 12 Oz. Beer) 1) Daily a) O Yes O No b) Average # of Drinks/Day: Average # Drinks/Occasion 2) Social (< 5 Drinks) O Yes O No 3) Binge (> 5 Drinks) O No O Yes AD 9. Past History of Treatment a) O Yes O No b) If Yes, When? (DD\MM\YY) /___/_ c) Where?________________ AD 10. Current History of Treatment a) O Yes O No b) If Yes, When Started? (DD\IVIM\YY). / / c) Where?________________ AD 11. Personal Goals for This Pregnancy (Check Only One) a Cut Down O Abstain/Quit During Pregnancy O Keep Cutting Down O Stay Quit a No Desire to Change Drinking Patterns O Don't Know Other (specify) _________________________ AD 12. Personal Goals After the Baby is Bom (Check Only One) a Cut Down O Abstain/Quit During Pregnancy O Keep Cutting Down O Stay Quit O No Desire to Change Drinking Patterns a Don't Know Other (specify) _______________________ Number of Occasions/wk 116 7. DRUG DATA (DD) DD 1. History of Drug Use in Lifetime: O Yes Type(s) of Drugs Used (DD) (Check All That Apply) O No (If No, Go To Client Monitonng' Sec. 8) Average Amount/Occasion Frequency (^ aoCharactw») (wks)* DD 2. Marljuana/THC O Yes o No DD 3. Crack Cocaine O Yes 0 No DD4. Cocaine (IV) O Yes a No DD 5. Cocaine (other) a Yes o No DD 6. LSD/Add O Yes o No DD 7. Mushrooms O Yes a No DD 8. Heroin (IV) O Yes a No DD 9. Heroin (other) O Yes o No DD10 Tylenol a Yes 0 No DD11. Codeine OYes a No DD12. Tdwin O Yes a No DD 13. Valium O Yes o No DD 14. Ritalin O Yes o No DD15. Barbituates & O Yes Other Tranquaiizers o No DD 16. Gravoi □ Yes a No DD 17. inhalants a Yes o No DD 18. Other (specify) "Enter In number of weeks where, 52wks = year, 4 wks/month. [If less than 1 wk Enter 999] DD 19. Past Treatment a) □ Yes □ No If Yes, b)When: / / (DD/MM/YY) c) Where (specify):____________________ DD 20. Current Treatment a) O Yes a No If Yes, b) When Started:__ / __ / __ (DD/MM/YY) n i W hA TA rmnAnHvV Time Since Last Use 117 8. CLIENT MONITORING (CM) CM 1. a) Date of Intake Interview: CM 1. b) Last Visit before Delivery: / / / / DD/MM/YY Pre-pregnancy CM 2. Weight (kgs) CM 3. Body Mass lndex(BMI) Substance Use Summary CM 4. Average # Cigarettes Per Day CM 5. Average # Drinks Per Week (@ 1 OZ per Drink) CM 6. Average # Drinks Per Month (@ 1 OZ per Drink) CM 7. Average No. Of Times Drugs Used Per Week CM 8. Average No. Of Times Drugs Used Per Month Meal Pattern CM 9. Number of Meals Per Day CM 10. Number of Snacks Per Day (Nb: A Meal Includes Foods From 3 to Food Groups; A Snack includes Foods from 1 to 2 Food Groups) Food Intake (Number of Serwngs Per Day Based on 24 Hour Recall) CM 11. Grain Products CM 12. Vegetables and Fruit CM 13. Milk Products CM 14. Meat and Alternatives Fluids (Number of 250 ML (8 Oz) Cups Per Day) CM 15. Coffee (Perc Or Drip, Caffeinated) CM 16. Coffee (Instant, Caffeinated) CM 17. Tea (Caffeinated) CM 18. Colas (Caffeinated) CM 19. Chocolate (Bars, etc.) Program Intake Last Vblt (if None, Leave Blank) Before Delivery 118 CM 20. Other Pops & Sweetened Fruit Drinks (Eg, Koolaid, Tang - Excluding Fmlt Juices) CM 21. Water Key Nutrients (Number of Servings Per Day Based on 24 Hour Recall) Iron Rich Foods CM 22. Excellent Sources CM 23. Other Sources Folate Rich Foods CM 24. Good Sources CM 25. Other Sources 119 10. PROGRAM CONTACT (PC) Number of Counselling Contacts Client's Home Program Site Phone Calls PC 1. Health Professionals PC 2. Outreach Workers PC 3. Other Locations of Contact with Health Professionals a) Location (specify): b) No. of Visits: PC 4. Other Locations of Contact with Outreach Workers a) Location (specify): b) No. of Visits:, PC 5. Number of Case Management Consultations, PC 6. Number of Attendances At POP Drop-In:___ PC 7. Number of Appointments Cancelled by Client/Attempts made to contact client: (Includes Not Home, No Show, and Other Attempts Made) PC 8. Receiving Program Food Supplements? a) a Yes Cl Don't Know O No -> If No, why not?_____ b) (specify)________________ PC 9. Seeing A Physician for Prenatal Care? a) O Yes O No b) Date of first Physician C ontact / __ / (DD/MM/YY) 120 11. PROGRAM OUTCOME (PO) P 0 1. Outcome of Present Pregnancy (Check Only One) a) O Single Live Birth O Multiple Live Birth O Therapeutic Abortion O Stillbirth (fetus bom dead > 20 wks) O Spontaneous Abortion (Miscarriage) b) Gestational Age (wks):_ d) Infant Birth Date (D D /M M /Y Y ) / ___/ _ c) Birth Weight (Grams):, for each Child e) If Multiple Births,Complete PO 1. (a) to (d) PO 2. Neonatal Complications (Check All That Apply) a) Premature Rupture of Membranes O Yes a b) Meconium Aspiration O Yes O No c) Forceps Assisted Delivery a O Yes No No d) Fetal Distress O Yes O No e) Prolapsed Cord O Yes O No f) Precipitate Delivery O Yes O No a) Active Genital Herpes O Yes O No b) Failure to Progress □ O No g) Other (specify)_________________ PO 3. Maternal Complications (Check All That Apply) c) Placenta Previa/Abruptio Yes O Yes a No O Yes O No Pregnancy Induced Hypertension (Toxemia) (PIH) a Yes O No Incompetent Cervix a Yes O No g) Prolonged labour O Yes No h) Cephalopelvic Disproportion (CPD) O Yes No i) Oligohydraminos O Yes O No j) Other (specify)______________ _ d) Gestational Diabetes e) f)