Journal of Affective Disorders 120 (2010) 149157

Contents lists available at ScienceDirect

Journal of Affective Disorders

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j a d

Research report

A prospective study of religion/spirituality and depressive symptoms among adolescent psychiatric patients
R.E. Dew a,, S.S. Daniel b, D.B. Goldston a, W.V. McCall c, M. Kuchibhatla d, C. Schleifer e, M.F. Triplett f, H.G. Koenig g
a

Department of Psychiatry and Behavioral Sciences, Duke University Medical Center, Durham, NC, United States University of North Carolina at Greensboro, Greensboro, NC, United States c Department of Psychiatry and Behavioral Medicine, Wake Forest University Health Sciences, Winston-Salem, NC, United States d Department of Biostatistics and Bioinformatics, Duke University Medical Center, Durham, NC, United States e Duke Divinity School, Durham, NC, United States f Wake Forest University Department of Psychology, Winston-Salem, NC, United States g Department of Psychiatry and Behavioral Sciences, Duke University Medical Center, Geriatric Research, Education, and Clinical Center, VA Medical Center, Durham, NC, United States
b

a r t i c l e

i n f o

a b s t r a c t
Objective: Previous research has uncovered relationships between religion/spirituality and depressive disorders. Proposed mechanisms through which religion may impact depression include decreased substance use and enhanced social support. Little investigation of these topics has occurred with adolescent psychiatric patients, among whom depression, substance use, and social dysfunction are common. Method: 145 subjects, aged 1218, from two psychiatric outpatient clinics completed the Beck Depression Inventory-II (BDI-II), the Fetzer multidimensional survey of religion/spirituality, and inventories of substance abuse and perceived social support. Measures were completed again six months later. Longitudinal and cross-sectional relationships between depression and religion were examined, controlling for substance abuse and social support. Results: Of thirteen religious/spiritual characteristics assessed, nine showed strong cross-sectional relationships to BDI-II score. When perceived social support and substance abuse were controlled for, forgiveness, negative religious support, loss of faith, and negative religious coping retained signicant relationships to BDI-II. In longitudinal analyses, loss of faith predicted less improvement in depression scores over 6 months, controlling for depression at study entry. Limitations: Self-report data, clinical sample. Conclusions: Several aspects of religiousness/spirituality appear to relate cross-sectionally to depressive symptoms in adolescent psychiatric patients. Findings suggest that perceived social support and substance abuse account for some of these correlations but do not explain relationships to negative religious coping, loss of faith, or forgiveness. Endorsing a loss of faith may be a marker of poor prognosis among depressed youth. 2009 Elsevier B.V. All rights reserved.

Article history: Received 18 January 2009 Received in revised form 11 April 2009 Accepted 27 April 2009 Available online 17 May 2009 Keywords: Religion Spirituality Depression Adolescents

1. Introduction Adolescent depression is an increasingly recognized and concerning public health problem. Carrying a point preva Corresponding author. DUMC Box 3492, 718 Rutherford St., Durham, NC, 27705, United States. Tel.: +1 919 286 5260; fax: +1 919 286 7081. E-mail address: rachel.dew@duke.edu (R.E. Dew). 0165-0327/$ see front matter 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jad.2009.04.029

lence of 515%, depression affects up to 1 in 5 people before adulthood (Bhatia and Bhatia, 2007; Zuckerbrot and Jensen, 2006). Acute and long-term outcomes may include recurrent depression, psychosocial impairment, substance abuse, and suicide. Recent research has uncovered relationships linking adult and adolescent depression to religion and spirituality, prevalent phenomena in American and international life. The study of

150

R.E. Dew et al. / Journal of Affective Disorders 120 (2010) 149157

religion in the context of health and medicine has been rapidly increasing over the last thirty years (Koenig, 2004). Although denitions of these terms may vary, this manuscript will use religion to refer to an organized system of beliefs, rituals, practices, and community, oriented toward the sacred; spirituality will refer to more personal experiences of or search for ultimate reality or the transcendent that are not necessarily institutionally connected (Josephson and Dell, 2004). Studies among adults reveal fairly consistent relationships between level of religiousness/spirituality and depressive disorders that are signicant and inverse (Koenig et al., 2001). Research on this topic among adolescent subjects, however, has yielded mixed ndings. Dew and colleagues (Dew et al., 2008a), in a review of 21 studies on religion and depression in adolescents, reported that most described conicting results: some comparisons showed, as commonly found among adults, that religiousness related to lower levels of depression, but some showed no relationship. In addition, four of the 21 studies found religious variables (including personal importance of religion (Cotton et al., 2005) and church attendance (Sorenson et al., 1995)) correlated with higher levels of depression. Few studies on religion/depression relationships among adolescents include psychiatric populations. Osman et al. (1996) found that moral objections to suicide, which referenced religious concepts, correlated negatively with depression in adolescent psychiatric inpatients. Cole, however, reported that, in juvenile delinquents in psychiatric treatment, these same moral objections did not relate to depression (Cole, 1989). Miller et al. (2002) followed a clinical sample of depressed children longitudinally; when religious variables and depression were measured in adulthood, the relationship varied by childhood depression status. This raises the possibility that early depression affects religious development, determining whether or not religiousness is protective against adult psychopathology. Dew and colleagues found that in adolescent psychiatric outpatients, scores on the Beck Depression Inventory related positively to negative religious coping (perceptions of God as punishing or abandoning), positively to negative religious support (perceptions of the religious community as critical or demanding), and inversely to forgiveness (self-reported tendency to practice forgiveness or feel that God is forgiving) (Dew et al., 2008b). Population-based studies hold value due to the wide generalizability of their results. However, a low case rate can limit the ability to nd correlations in a general sample. Psychiatric samples feature high rates of depressive disorders, as well as important covariates such as substance use, family pathology, and social disadvantage. Therefore, clinical samples add much to research in this eld. This article describes the rst study, known to the authors, of religion and depression in adolescent psychiatric patients, using a longitudinal design and accounting for psychosocial covariates. Variability in ndings of religion/depression research with adolescents complicates consideration of its clinical utility. Several factors likely contribute to this inconsistency. One such factor is the diversity of denitions of religion. This complex subject has been contemplated in myriad elds such as sociology, anthropology, theology, and psychology. Even within medical research, which has overwhelmingly occurred

in relatively homogenous white American Christian samples, there is no consensus on how to dene or measure religion. Investigators have conceptualized religiousness variously as attendance at services, self-reported devotion, frequency of prayer, using religion to cope with adversity, and other constructs. A newer literature describes the concept of negative religious coping, dened as expressions of conict, question, and doubt regarding matters of faith, God, and religious relationships (McConnell et al., 2006) p.1470). This body of work has found a higher rate of depression in those who feel abandoned or rejected by God or the religious community (Ano and Vasconcelles, 2005). Because several of the many constructs called religion correlate with mental health outcomes, clarication of these questions will require use of multidimensional measures. Inconsistency may also relate to geographic variability between studies. Concentrations of specic denominations, as well as local culture in general, may cause religion/depression relationships to vary by region. Strong social cohesion seen in certain religious minorities may be helpful in some contexts and more stigmatizing in others. Regional differences have proved important in previous studies examining religiosity in relation to longevity (la Cour et al., 2006), delinquency (Stark et al., 1982), and suicide (Zhang and Jin, 1996). An additional issue to consider is the inconsistent use of multivariate analyses in religion/health studies. While most researchers account for demographics, psychosocial variables such as family cohesion or social support are not uniformly controlled for (or examined as explanatory/mediating variables). In studies that consider such variables as family closeness or school stress, these constructs often weaken or eliminate zero-order correlations of religion to mental health outcomes (Benda and Corwyn, 1997; Nooney, 2005; Stewart et al., 1999). Furthermore, previous studies have rarely considered substance use as a mediator, despite its consistent negative relationship to teenage religiosity (Koenig et al., 2001) and its status as a proposed mechanism through which spirituality positively impacts mental health (George et al., 2002). A major limitation of current religion/health research is the dearth of longitudinal studies. Extrapolation of crosssectional correlations to longitudinal or predictive relationships is questionable. A correlation between depressive symptoms and religious variables at a single point in time may indicate any of several long-term relationships between these constructs. For example, depression could cause a temporary emergence of certain religious traits, such as seeing God as more punitive, or the decline in other traits, such as religious service attendance. Alternatively, certain religious tendencies could either raise or lower risk for depression. Clarication of such temporal relationships is crucial to our understanding of the intersection of religion and mental health. 2. Methods In order to address the above gaps in the religion/ depression literature, the following prospective study was carried out at two outpatient psychiatric clinics in North Carolina. Counties containing the two clinics are predominantly Christian and Protestant, with the largest denominations being Southern Baptist and United Methodist.

R.E. Dew et al. / Journal of Affective Disorders 120 (2010) 149157 Table 1 Subscales analyzed from the brief multidimensional measure of religiousness/spirituality. Subscale Daily spiritual experiences Forgiveness Private religious practices Positive religious coping Negative religious coping Positive religious support # items 5 3 5 4 2 2 Item range/directionality 16, lower = more frequent experiences 14, lower = more forgiving 18, lower = more frequent practices 14, lower = greater use 14, lower = greater use 14, lower = greater support Sample item I feel God's presence I have forgiven those who hurt me How often do you pray privately in places other than at a church, synagogue, or temple? I look to God for strength, support, and guidance I feel God is punishing me for my sins or lack of spirituality If you had a problem or were faced with a difcult situation, how much comfort would the people in your congregation be willing to give you? How often are the people in your congregation critical of you and the things you do? I try hard to carry my religious beliefs over into all my other dealings in life Have you ever had a signicant gain in your faith? Have you ever had a signicant loss in your faith? How often do you go to religious services? To what extent do you consider yourself a spiritual person? I have a sense of mission or calling in my own life Mean 15.8 5.7 24.2 9.5 6.4 3.8

151

SD 6.8 2.2 8.1 3.3 1.7 2.0

Negative religious support Commitment Positive religious history Loss of faith Organizational religiousness Overall self-ranking Meaning

2 1 2 1 2 2 2

14, lower = more negative support 14, lower = stronger agreement 12, higher higher = more religious or spiritual 12, higher = more religious or spiritual 16, lower = more frequent attendance 14, lower = more religious or spiritual 14, lower = greater agreement

6.3 2.3 3.0 1.6 7.0 5.0 4.1

1.8 0.9 0.9 0.5 3.1 1.6 1.6

Adolescent participants were asked to complete self-report measures of depressive symptoms, substance abuse, perceived social support, and a multidimensional measure of religiousness/spirituality. The same measures were administered a second time approximately six months later. Parents/guardians provided treatment history and demographics, and diagnosis made by the treating clinician was recorded. Because of possible reliability problems with nonstandardized diagnosis, chart diagnosis was used to characterize the sample only. The primary aim of the study was to assess the cross-sectional and longitudinal relationships between religious/spiritual characteristics and depression in adolescent psychiatric outpatients. We hypothesized that negative religious coping, negative religious support, and loss of faith would relate positively with depressive symptoms, while the remaining variables would relate inversely with depression. Additionally, we hypothesized that social support and substance abuse would serve as mediators between religion/spirituality and depression. The study was approved by the Institutional Review Boards of Duke University Medical Center and Wake Forest University Health Sciences. Participants were recruited consecutively from clinic schedules. Potential subjects were approached by their treating clinicians. If subjects were interested, study staff explained the study and obtained written informed consent from their parent or legally responsible person. Subjects could be included if they were aged 1218, were presenting for treatment, and were able to understand and complete the questionnaires. Assistance with reading was given as needed. Those with mental retardation, either reported by the treating clinician or documented in the medical record, were excluded. Of 267 potential subjects identied from clinic schedules, 75 were excluded because they were unable to understand and complete the questionnaire, had no available parent or guardian

to give consent, or were otherwise deemed inappropriate for the study by their treating clinician. Fifty-one potential subjects refused participation, usually citing lack of time. One hundred forty seven subjects gave consent/assent and completed the rst data collection. Two were subsequently excluded due to diagnosis of mild mental retardation found on chart review. 3. Procedures Following consent and assent, parents/guardians of participants completed a demographic/treatment history form. In a separate room, adolescent enrollees completed four instruments: the Beck Depression Inventory-II (BDI-II, (Osman et al., 2004), the Brief Multidimensional Measure of Religiousness/Spirituality (BMMRS, (Fetzer, 1999), the Multidimensional Scale of Perceived Social Support (MSPSS, (Canty-Mitchell and Zimet, 2000), and the Problem Oriented Screening Instrument for Teenagers Substance Abuse Subscale (POSIT,(Knight et al., 2001). Study personnel reviewed the clinical chart to obtain the diagnosis documented by the subject's primary mental health care provider as near as possible to the day of data collection. Approximately six months later, participants completed the measures again at the clinic, by mail or by telephone interview. Subjects were paid ve dollars for participation in each data point. 4. Measures Participants completed the Beck Depression Inventory-II (BDI-II, (Osman et al., 2004), a 21-item self-report scale (possible scores 063). This scale has been validated in adolescent psychiatric patients. Cronbach's alpha of the BDI-II in this study was 0.90. Adolescents completed the Brief Multidimensional Measure of Religiousness/Spirituality (BMMRS, (Fetzer, 1999), a 40-item, 14-subscale inventory querying various aspects of religion and spirituality. For this

152

R.E. Dew et al. / Journal of Affective Disorders 120 (2010) 149157

study, 34 items comprising 14 subscales were analyzed. Subscales include: daily spiritual experiences, forgiveness, private religious practices, positive religious coping, negative religious coping, positive religious support, negative religious support, commitment, positive religious/spiritual history, loss of faith, organizational religiousness, overall self-ranking, and meaning. These subscales and sample items for each are presented in Table 1. Unanalyzed items include: the item I feel deep inner peace and harmony, as this item could articially inate correlations with depression; an item asking for a total number of hours spent weekly in religious activity, due to excessive missing data; the values/beliefs scale, due to poor internal reliability; the coping item I try to make sense of a situation and decide what to do without relying on God, due to poor correlation with either the positive or negative religious coping items, and an item querying monetary donations to religious organizations, which was not felt to be age appropriate. This instrument has been recently validated in an adolescent clinical sample (Harris et al., 2008) and previously used in adolescent and adult populations (Fetzer, 1999; Knight et al., 2007; Pearce et al., 2003). Cronbach's alpha for subscales in this study ranged from 0.59 to 0.87. The Problem Oriented Screening Instrument for TeenagersSubstance Abuse Subscale (POSIT) contains 17 items related to drug and alcohol use, related behaviors, and consequences. It has been shown to be internally consistent and reliable (Knight et al., 2001). Cronbach's alpha for the scale in this study was 0.88. Social support has been theorized to mediate relationships between religiousness and mental health (George et al., 2002). Perceived social support was measured via the Multidimensional Scale of Perceived Social Support (MSPSS, (Canty-Mitchell and Zimet, 2000; Clara et al., 2003). The MSPSS contains 12 items, and has been found reliable and valid in adolescent populations. Cronbach's alpha for the scale in this study was 0.88. Demographics and treatment history were obtained from the consenting parent or guardian. Demographics included age, race/ethnicity, gender, and religious preference of the adolescent. Parents/guardians were asked to provide their own highest level of education and family income. Treatment history obtained included length and type of services used. 5. Analysis Data imputation was performed on scales if at least 80% of the total items were completed, substituting the mean of the remaining items; if less than 80% of the scale was complete, the item remained missing. Imputed values represent b0.01% of the data. Regression models were created with BDI-II score at time 1 as the dependent variable and demographic and religious/ spiritual variables as predictors. Models were then controlled for substance abuse and social support. Our nal model included all religious/spiritual, control, and explanatory variables. Due to skew in the distribution of BDI-II scores at time 2, a change score was created using the difference between BDI-II at time 1 and time 2. This change score was modeled using linear regression and controlled for BDI-II at time 1. Analyses were repeated excluding ve signicant

outliers with no appreciable change in results. All analyses were performed using SAS Enterprise Guide 3.0 (Cary, NC). 6. Results Seventy-eight subjects (54%) were recruited from Duke University and 67 (46%) from Wake Forest University Health Sciences. The 145 subjects had a mean age of 14.3 years (SD 1.8). Sixty-one (42%) were female. Ethnic composition of the sample was as follows: Caucasian n = 82 (56%), African American n = 50 (35%), and other n = 13 (9%). Fifty-two percent reported family incomes of less than $40,000/year and 12% had incomes over $100,000. 92% of parents/guardians reported having at least a high school education. Religious afliations of subjects, as reported by parents/ guardians, were classied as Conservative Protestant (Baptist, non-denominational Christian, Church of Jesus Christ of Latter Day Saints (LDS), and others; 71% of sample), Liberal Protestant (Methodist, Presbyterian, and others; 15% of sample), Roman Catholic (7%), other (5%), or none (5%). Although LDS youth could represent a unique category based on theological and social distinctions, only two subjects endorsed this preference; thus, based on shared views toward substance use, these subjects were included in the Conservative Protestant category. 6.1. Illness and treatment parameters Participants reported a mean substance use score of 0.8, with a median score of 0, on a possible scale of 0 to 17. Reported substance use scores ranged 013. Due to this skew in the data, scores on the POSIT were dichotomized. Thirtytwo subjects (22%) scored one or greater on this inventory. The sample had a mean BDI-II of 13.5 (SD 9.6). Forty-four subjects reported past admissions to psychiatric hospitals (31%), and 116 (81%) had used psychotropic medications. Seventy-two subjects (51%) endorsed having been prescribed antidepressant medications. Depression (including major depression, dysthymia, adjustment disorder with depressed mood, and depressive disorder not otherwise specied) was diagnosed in 52 patients (38%). Other mood disorders (including bipolar disorder and mood disorder not otherwise specied) were noted in 15 patients (10%). ADHD, diagnosed in 75 subjects (55%), was the most common diagnosis. No clinical diagnosis was available for 7 subjects. 6.2. Religious and social characteristics Sample means of each BMMRS subscale are presented in Table 1. Mean item scores indicate that subjects generally endorsed having spiritual experiences on most days, often used religion to cope with problems, and participated in organized religious activities once or twice per month. The mean score on the MSPSS was 66.5 (SD 13.5; range 1784). This indicates that the average respondent expressed mild to strong perception of good social support. 6.3. Follow-up One hundred four subjects (72%) of the original sample completed measures approximately six months following

R.E. Dew et al. / Journal of Affective Disorders 120 (2010) 149157 Table 2 Bivariable and multivariable correlates of BDI-II score at Time 1. Uncontrolled analysis Independent variable Site Age Female gender Caucasian African American Other race SES No religion Conservative Protestant Liberal Protestant Catholic Other religion Substance abuse Social support Daily spiritual experiences Forgiveness Private religious practices Positive religious coping Negative religious coping Positive religious support Negative religious support Commitment Positive religious history Loss of faith Organizational religiousness Overall self-ranking Meaning 1.47 0.32 3.15 0.84 4.15 0.79 1.91 1.33 1.06 0.74 4.20 6.05 0.29 0.35 1.84 0.12 0.60 2.00 0.98 1.00 1.08 0.12 5.81 0.71 1.41 0.93 p 0.3772 0.4795 0.0599 0.6353 0.1595 0.2194 0.5940 0.4696 0.6576 0.8280 0.3069 0.0022 b .0001 0.0032 b .0001 0.2248 0.0176 b 0.0001 0.0171 0.0360 0.2521 0.8963 0.0006 0.0066 0.0063 0.0760 Controlled for gender, social support, and substance abuse 0.01 1.14 0.13 0.09 1.53 0.14 1.22 0.73 1.62 3.73 0.22 0.28 0.24 p Religious variables compete in single model p

153

0.9157 0.0034 0.2018 0.7442 0.0006 0.7268 0.0045 0.4185 0.0612 0.0185 0.3941 0.5791 0.6573

0.06 1.18 0.23 0.33 1.14 0.06 0.59 2.26 0.89 4.33 0.55 0.87 0.10

0.8054 0.0221 0.1228 0.4956 0.0223 0.9026 0.1936 0.0787 0.3806 0.0086 0.0975 0.2190 0.8789

Religious variables entered into base model (gender, social support, and substance abuse).

study entry. t-tests comparing demographics and depression severity of completers and non-completers revealed no signicant differences between the groups. 6.4. Regression analyses Uncontrolled cross-sectional correlations with BDI-II scores are displayed in Table 2. Among demographic measures, only gender showed a near-signicant correlation with BDI-II scores. No treatment variables related to BDI-II. Religious characteristics which negatively correlated with BDI-II scores included the following: daily spiritual experiences, forgiveness, positive religious coping, positive religious support, loss of faith, organizational religiousness, and selfranking as religious/spiritual. Negative religious coping, negative religious support, and loss of faith were related positively to BDI-II, such that higher levels of these characteristics related to greater depressive symptoms. Both substance abuse and perceived social support were signicantly related to BDI-II score, such that greater levels of substance abuse and lower levels of social support correlated with increased depression. Multivariable results are summarized in Table 2. When perceived social support and substance abuse measures were added to models predicting BDI-II score, daily spiritual experiences, positive religious coping, organizational religiousness, and self-ranking became non-signicant (implying that social support and substance abuse explained these relationships with depression). Only forgiveness, negative

religious support, negative religious history, and negative religious coping remained signicant and could not be explained by social support or substance abuse. When all religious variables were allowed to compete in the model, forgiveness, negative religious coping, and loss of faith were retained, each contributing unique variance. When gender, substance abuse, and social support were added to this parsimonious model, forgiveness and loss of faith retained signicance, and negative religious coping became nonsignicant ( = 0.87, p = 0.07). Next, baseline predictors were sought of change in BDI-II score over time, controlling for the initial BDI-II. Only loss of faith predicted change in depression, such that endorsing a loss of faith at time 1 predicted less improvement in BDI-II score six months later ( = 4.69, p = 0.007). This effect persisted when controlled for gender, social support, and substance abuse. To further characterize the meaning of loss of faith in this sample, an attempt was made to correlate it with a lack of religiosity. Disagreeing with the BMMRS item I believe in a God who watches over me did not relate to a loss of faith, nor did endorsing no religion. 7. Discussion We hypothesized that negative religious coping, negative religious support, and loss of faith would relate positively to depressive symptoms, while other aspects of religiousness/ spirituality would relate negatively to depression. We expected

154

R.E. Dew et al. / Journal of Affective Disorders 120 (2010) 149157

this relationship to be apparent cross-sectionally and longitudinally. Our secondary hypothesis stated that social support and substance abuse would mediate or explain the observed relationships between religion/spirituality and depression. The primary hypothesis was partially conrmed. Nine of 13 religious subscales related signicantly to BDI-II score in uncontrolled analysis. Greater use of negative religious coping and endorsing negative support from the religious community correlated with higher depression scores, as did endorsing having experienced a loss of faith. In contrast, daily spiritual experiences, forgiveness, positive religious coping, positive religious support, organizational religiousness, and self-ranking as religious/spiritual were inversely related to depressive symptoms. These ndings are consistent with previous research (Ano and Vasconcelles, 2005; Dew et al., 2008b; Knight et al., 2007; Koenig et al., 2001), and argue for a more complex model of the religion/depression relationship; all religious beliefs and experiences are not necessarily related to better mental health. Contrary to our hypothesis, longitudinal analysis showed that only endorsing a loss of faith predicted less improvement in depressive symptoms over time. This argues against the hypothesis that religious beliefs or behaviors cause or prevent depression, but rather correspond to concurrent level of depression. In accordance with our secondary hypothesis, that substance abuse and social support may serve as mechanisms through which religiousness impacts depression, ve of the nine subscales initially correlating with BDI-II score became non-signicant when substance abuse and social support were controlled. However, these two control variables failed to completely explain the observed relationships: four subscales forgiveness, negative religious coping, negative religious support, and loss of faith, continued to relate signicantly to BDI-II score despite control for substance abuse and social support. The prospective contribution of loss of faith to the variance in BDI-II change score also retained signicance when controlled for these potential explanatory variables. Accordingly, it appears that religion and spirituality may relate to adolescent depression both indirectly through social support and substance abuse, and also directly. These ndings are consistent with other psychiatric research. First, as in the adult literature, several aspects of religiosity related inversely to depressive symptoms. Among adults, similar relationships have been found cross-sectionally (Baetz et al., 2004; Hahn et al., 2004; Koenig et al., 2001), and longitudinally (Braam et al., 2004; Koenig et al., 1998). Also consistent is the observed association between negative religious coping and higher levels of depression. This relationship has been conrmed among adults in a recent meta-analysis (Ano and Vasconcelles, 2005). The nding that baseline loss of faith predicted less improvement in depressive symptoms over time is consistent with the idea that faith somehow protects against depression. However, loss of faith does not seem to be the same as not having faith, as this construct did not correlate with measures of atheism or having no religion. It may be that it is specically the loss rather than the absence of faith that is important. Previous literature supportive of this idea includes a study which found a decrease in religious faith predicted greater use of mental health services among veterans with

PTSD (Fontana and Rosenheck, 2004), and a longitudinal survey which found that a decrease in religiosity predicted conduct problems equaling or exceeding the conduct problems of those who were never religious at all (Peek et al., 1985). It is possible that adolescents completing this survey interpreted loss of faith to mean loss of hope. Hopelessness is a concept highly relevant to the study of depression, and has been found to relate to suicidality (Nrugham et al., 2008). Although there was no direct measure of hopelessness in this study, it may be noted that loss of faith did not correlate independently with suicidality. Further investigation, likely involving qualitative methodology, will be needed to clarify this issue. The ndings of this study also mirror results of several studies in adolescents (Miller and Gur, 2002; Schapman and Inderbitzen-Nolan, 2002; Wright et al., 1993; Pearce et al., 2003). Unlike our study, Harker found that religious attendance predicted lower depression scores one year later among adolescent participants of the Add Health Study (Harker, 2001). This study, however, did not control for baseline depression. As stated above, previous literature on religion and depression among young psychiatric patients has been sparse and inconsistent. The current study provides partial replication of previous results found among a sample of 117 adolescent psychiatric outpatients (Dew et al., 2008b). Cross-sectionally related to lower scores on the BDI-II in that sample, after controlling for substance use, were forgiveness, negative religious coping, and negative religious support; in the current study, these scales, in addition to loss of faith, retained signicant relationships to depression even controlling for potential mediators. 7.1. Interpretation of ndings Why might these religious variables relate to depressive symptoms? Theories on how religion might impact mental health abound (Koenig et al., 2001). The fact that ve of the nine subscales lost signicance when substance abuse and social support were added to the model gives us insight into possible mechanisms. It has been consistently found that religious variables relate inversely with teenage substance abuse (Brown et al., 2001b; Heath et al., 1999; Nonnemaker et al., 2003; Wallace et al., 2003). It is also known that substance use relates to depression (Conway et al., 2006; Currie et al., 2005; Goldstein et al., 2007). It is therefore likely that one way religion relates to depression is that those who embrace religion use fewer substances, protecting them against depression. Social support has also been found to relate to depression (Buist-Bouwman et al., 2004; Dahlem et al., 1991; Denny et al., 2004a,b). Possible explanations of this relationship include the theory that social support prevents or alleviates depression, as well as the idea that non-depressed persons are better able to engage with others socially. Religion and spirituality can provide social support in several ways. Church membership may allow teens to develop extra-familial relationships with stable adult gures, sometimes providing guidance and affection lacking in their own families. Church youth groups likely help establish healthy peer bonds. Trends

R.E. Dew et al. / Journal of Affective Disorders 120 (2010) 149157

155

in contemporary American evangelical Christianity toward intense individual relationships with God, who is perceived to accompany the adherent throughout daily life, have been observed in anthropological research (Luhrmann, 2004). A close and nourishing relationship with the Divine may mitigate the effects of negative life experiences or other risks for depression. Possibly, adolescents in this study who were not depressed were better able to connect and bond with social contacts. Alternatively, those who endorsed religious connections may also tend, independently, to have close, supportive families and friends. Four subscales contributed unique variance to the BDI-II score beyond that of substance abuse and social support: lack of forgiveness, negative religious coping, negative religious support, and loss of faith. These phenomena can be understood in several ways. Such feelings could precede and contribute to the onset of depression. Alternatively, they may represent depressive symptoms negative thoughts of those looking at the world through the depressed lens. Cognitions about the harsh and punitive character of God may represent a way to cope with depression; e.g., some depressed subjects may understand their illness as divine retribution for past sins. It is also possible that these phenomena represent the type of religion that emerges in someone who has experienced depression in youth; adolescent depression may leave a scar on the personality or worldview that distorts mainstream religious experience. Finally, these religious variables and depression may in truth be unrelated, but instead both related to a third variable not assessed in this study. Only one variable, loss of faith, related to change in depressive symptoms prospectively. Independent of the level of depressive symptoms at Time 1, those endorsing a loss of faith showed less improvement over time. Thus loss of faith may represent a marker of poor prognosis in treatmentseeking adolescents. Most other religious variables correlated with depressive symptoms cross-sectionally, but did not predict change in depression over time. This is noteworthy, in that the eld to this point has often supposed causal relationships between religion and depression. Rather, these results support the hypothesis that such variables as lack of forgiveness, negative religious coping, and negative religious support are depressive symptoms. These may to some extent represent state, rather than trait, attributes. 7.2. Clinical implications Given associations with depression, religious/spiritual characteristics may require assessment in the mental health care setting. Inquiries can be made about perceptions of the relationship to the Divine and to the religious community, about inability to forgive oneself for past wrongs, and about loss of previously held religious faith. As these phenomena may represent clinical symptoms, they may be monitored over time as such. If they represent risk factors, then they may be useful in terms of prognosis. Direct intervention into religious matters is more controversial. It is not clinically accepted at present for therapists to attempt to reshape a client's relationship with God, but referral to a chaplain or pastoral counselor may be helpful and appropriate (Koenig,

2001). If it is discovered that interpersonal interactions between the patient and his or her religious congregation are in fact negative and stressful, discussion with parents may help alleviate this. 7.3. Limitations This study has several features that limit its generalizability. First, subjects were all psychiatric patients. Differences between depressed and non-depressed psychiatric patients will not necessarily generalize to non-clinical populations. Second, the data presented here are observational and cannot prove or disprove causal relationships between any of the measured variables. Thirdly, both clinical sites are in the Southeast United States, and the sample was dominated by those with conservative Protestant religious afliations; thus, results may not generalize to samples from other geographic areas or religious groups. Religious groups growing out of other cultures as well as newer movements in American spirituality may differ vastly in emphasis and relationship to health. Important information may be gleaned by comparing these results to future studies in more religiously-diverse areas. Fourth, the sample size of 145 is relatively small given the number of comparisons made; thus replication with larger samples is needed. Larger studies may be better able to explore how religious variables interact to predict health outcomes; for example, some studies nd that level of religiousness interacts with denomination to predict substance use (Dew et al., 2008a). Larger studies will also be better equipped to examine non-linear relationships, which are sometimes found in substance abuse (Shedler and Block, 1990) and religion/health research (Brown et al., 2001a). In addition, large studies will be able to analyze important subgroups such as lesbian/gay/bisexual youth. Fifth, the substance abuse measure showed low variability necessitating use of a dichotomized score; it is possible that if more variance in substance abuse could be captured, results would change. However, approximately 20% of the sample reported some substance abuse, a gure consistent with previous literature on rates of substance abuse in mental health care settings (Rush and Koegl, 2008). Sixth, recruitment for this study required the assistance of subjects' treatment providers; providers were approached by study personnel about all identied eligible candidates and exclusion criteria were explained. It is unknown if any treatment provider introduced bias by encouraging or discouraging participation based on any of our salient variables, e.g., encouraging more religious or less depressed subjects to participate. Despite these limitations, this study represents the most extensive longitudinal analysis to date of the relationship of religiousness/spirituality to depression among adolescent psychiatric patients, and the rst such study (known to the authors) to investigate the possibility of mediation through substance abuse and social support. Future studies should attempt to replicate these ndings. In addition, more qualitative research should be done to further understand the meaning and function of spiritual concepts to adolescents. Such research will pave the way for more concrete clinical applications.

156

R.E. Dew et al. / Journal of Affective Disorders 120 (2010) 149157 among adolescents in an urban emergency department. J. Adolesc. Health 40, 276279. Hahn, C.Y., Yang, M.S., Yang, M.J., Shih, C.H., Lo, H.Y., 2004. Religious attendance and depressive symptoms among community dwelling elderly in Taiwan. Int. J. Geriatr. Psychiatry 19, 11481154. Harker, K., 2001. Immigrant generation, assimilation, and adolescent psychological well-being. Soc. Forces 79, 969. Harris, S.K., Sherritt, L.R., Holder, D.W., Kulig, J., Shrier, L.A., Knight, J.R., 2008. Reliability and validity of the brief multidimensional measure of religiousness/spirituality among adolescents. J. Relig. Health 47, 438457. Heath, A.C., M.P., Grant, J.D., McLaughlin, T.L., Todorov, A.A., Bucholz, K.K., 1999. Resiliency factors protecting against teenage alcohol use and smoking: inuences of religion, religious involvement and values, and ethnicity in the Missouri Adolescent Female Twin Study. Twin Res. 2, 145155. Josephson, A.M., Dell, M.L., 2004. Religion and spirituality in child and adolescent psychiatry: a new frontier. Child Adolesc. Psychiatr. Clin. N. Am. 13, 115 (v). Knight, J.R., Goodman, E., Pulerwitz, T., DuRant, R.H., 2001. Reliability of the Problem Oriented Screening Instrument for Teenagers (POSIT) in adolescent medical practice. J. Adolesc. Health 29, 125130. Knight, J.R., Sherritt, L., Harris, S.K., Holder, D.W., Kulig, J., Shrier, L.A., Gabrielli, J., Chang, G., 2007. Alcohol use and religiousness/spirituality among adolescents. South Med. J. 100, 349355. Koenig, H.G., 2001. Religion and medicine IV: religion, physical health, and clinical implications. Int. J. Psychiatry Med. 31, 321336. Koenig, H.G., 2004. Religion, spirituality, and medicine: research ndings and implications for clinical practice. South Med. J. 97, 11941200. Koenig, H.G., George, L.K., Peterson, B.L., 1998. Religiosity and remission of depression in medically ill older patients. Am. J. Psychiatry 155, 536542. Koenig, H.G., McCullough, M.E., Larson, D.B., 2001. Handbook of Religion and Health. Oxford, New York. la Cour, P., Avlund, K., Schultz-Larsen, K., 2006. Religion and survival in a secular region. A twenty year follow-up of 734 Danish adults born in 1914. Soc. Sci. Med. 62, 157164. Luhrmann, T.M., 2004. Metakinesis: how God becomes intimate in contemporary U.S. Christianity. Am. Anthropol. 106, 518528. McConnell, K.M., Pargament, K.I., Ellison, C.G., Flannelly, K.J., 2006. Examining the links between spiritual struggles and symptoms of psychopathology in a national sample. J. Clin. Psychol. 62, 14691484. Miller, L., Gur, M., 2002. Religiosity, depression, and physical maturation in adolescent girls. J. Am. Acad. Child Adolesc. Psychiatry 41, 206214. Miller, L., Weissman, M., Gur, M., Greenwald, S., 2002. Adult religiousness and history of childhood depression: eleven-year follow-up study. J. of Nerv. Ment. Dis. 190, 8693. Nonnemaker, J.M., McNeely, C.A., Blum, R.W., 2003. Public and private domains of religiosity and adolescent health risk behaviors: evidence from the National Longitudinal Study of Adolescent Health. Soc. Sci. Med. 57, 20492054. Nooney, J.G., 2005. Religion, stress, and mental health in adolescence: ndings from Add Health. Rev. Relig. Res. 46, 341354. Nrugham, L., Larsson, B., Sund, A.M., 2008. Specic depressive symptoms and disorders as associates and predictors of suicidal acts across adolescence. J. Affect. Disord. 111, 8393. Osman, A., Kopper, B.A., Barrios, F.X., Osman, J.R., Besett, T., Linehan, M.M., 1996. The Brief Reasons for Living Inventory for Adolescents (BRFL-A). J. Abnorm. Child Psychol. 24, 433443. Osman, A., Kopper, B.A., Barrios, F., Gutierrez, P.M., Bagge, C.L., 2004. Reliability and validity of the Beck depression inventory-II with adolescent psychiatric inpatients. Psychol. Assess. 16, 120132. Pearce, M.J., Little, T.D., Perez, J.E., 2003. Religiousness and depressive symptoms among adolescents. J. Clin. Child Adolesc. Psychol. 32, 267276. Peek, C., Curry, E., Chalfant, H., 1985. Religiosity and delinquency over time: deviance deterrance and deviance amplication. Soc. Sci. Q. 65, 120131. Rush, B., Koegl, C.J., 2008. Prevalence and prole of people with co-occurring mental and substance use disorders within a comprehensive mental health system. Prvalence et prot de personnes souffrant de troubles d'utilisation de substance et de troubles mentaux co-occurrents dans un systme complet de sant mentale, 53, pp. 810821. Schapman, A.M., Inderbitzen-Nolan, H.M., 2002. The role of religious behaviour in adolescent depressive and anxious symptomatology. J. Adolesc. 25, 631643. Shedler, J., Block, J., 1990. Adolescent drug use and psychological health: a longitudinal inquiry. Am. Psychol. 45, 612630. Sorenson, A.M., Grindstaff, C.F., Turner, R.J., 1995. Religious involvement among unmarried adolescent mothers: a source of emotional support? Sociol. Relig. v56, 71(11). Stark, R., Kent, L., Doyle, D.P., 1982. Religion and delinquency: the ecology of a lost relationship. J. Res. Crime Delinq. 19, 424. Stewart, S.M., Lam, T.H., Betson, C., Chung, S.F., 1999. Suicide ideation and its relationship to depressed mood in a community sample of adolescents in Hong Kong. Suicide Life Threat Behav. 29, 227240.

Role of funding source Support for this manuscript provided by a grant from the John Templeton Foundation, which had no further role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication.

Conict of interests All authors declare that they have no conicts of interest.

References
Ano, G.G., Vasconcelles, E.B., 2005. Religious coping and psychological adjustment to stress: a meta-analysis. J. Clin. Psychol. 61, 461480. Baetz, M., Grifn, R., Bowen, R., Koenig, H.G., Marcoux, E., 2004. The association between spiritual and religious involvement and depressive symptoms in a Canadian population. J. of Nerv. Ment. Dis. 192, 818822. Benda, B.B., Corwyn, R.F., 1997. Religion and delinquency: the relationship after considering family and peer inuences. J. Sci. Study Relig. 36, 8192. Bhatia, S.K., Bhatia, S.C., 2007. Childhood and adolescent depression. Am. Fam. Physician 75, 7380. Braam, A.W., Hein, E., Deeg, D.J., Twisk, J.W., Beekman, A.T., Van Tilburg, W., 2004. Religious involvement and 6-year course of depressive symptoms in older Dutch citizens: results from the Longitudinal Aging Study Amsterdam. J. Aging Health 16, 467489. Brown, T.L., Parks, G.S., Zimmerman, R.S., Phillips, C.M., 2001a. The role of religion in predicting adolescent alcohol use and problem drinking. (Statistical Data Included). J. Stud. Alcohol 62, 696 (610). Brown, T.N., Schulenberg, J., Bachman, J.G., O'Malley, P.M., Johnston, L.D., 2001b. Are risk and protective factors for substance use consistent across historical time?: national data from the high school classes of 1976 through 1997. Prev. Sci. 2, 2943. Buist-Bouwman, M.A., Ormel, J., de Graaf, R., Vollebergh, W.A.M., 2004. Functioning after a major depressive episode: complete or incomplete recovery? J. Affect. Disord. 82, 363371. Canty-Mitchell, J., Zimet, G.D., 2000. Psychometric properties of the multidimensional scale of perceived social support in urban adolescents. Am. J. Community Psychol. 28, 391400. Clara, I.P., Cox, B.J., Enns, M.W., Murray, L.T., Torgrudc, L.J., 2003. Conrmatory factor analysis of the multidimensional scale of perceived social support in clinically distressed and student samples. J. Pers. Assess. 81, 265270. Cole, D.A., 1989. Validation of the reasons for living inventory in general and delinquent adolescent samples. J. Abnorm. Child Psychol. 17, 1327. Conway, K.P., Compton, W., Stinson, F.S., Grant, B.F., 2006. Lifetime comorbidity of DSM-IV mood and anxiety disorders and specic drug use disorders: results from the national epidemiologic survey on alcohol and related conditions. J. Clin. Psychiatry 67, 247257. Cotton, S., Larkin, E., Hoopes, A., Cromer, B.A., Rosenthal, S.L., 2005. The impact of adolescent spirituality on depressive symptoms and health risk behaviors. J. Adolesc. Health 36, 529. Currie, S.R., Patten, S.B., Williams, J.V.A., JianLi, W., Beck, C.A., El-Guebaly, N., Maxwell, C., 2005. Comorbidity of major depression with substance use disorders. Can. J. Psychiatry 50, 660666. Dahlem, N.W., Zimet, G.D., Walker, R.R., 1991. The multidimensional scale of perceived social support: a conrmation study. J. Clin. Psychol. 47, 756761. Denny, S., Clark, T.C., Fleming, T., Wall, M., 2004a. Emotional resilience: risk and protective factors for depression among alternative education students in New Zealand. Am. J. Orthopsychiatr. 74, 137149. Denny, S., Clark, T.C., Fleming, T., Wall, M., 2004b. Emotional resilience: risk and protective factors for depression among alternative education students in New Zealand. Am. J. Orthopsychiatr. 74, 137149. Dew, R.E., Daniel, S.S., Armstrong, T.D., Goldston, D.B., Triplett, M.F., Koenig, H.G., 2008a. Religion/spirituality and adolescent psychiatric symptoms: a review. Child Psychiatr. Hum. Dev. 39 (4), 381398. Dew, R.E., Daniel, S.S., Goldston, D.B., Koenig, H.G., 2008b. Religion, spirituality, and depression in adolescent psychiatric outpatients. J. of Nerv. Ment. Dis. 196, 247251. Fetzer, 1999. Multidimensional Measurement of Religiousness/Spirituality for Use in Health Research. In: Institute, N.I.o.A.F. (Ed.). John E. Fetzer Institute, Kalamazoo, MI. Fontana, A., Rosenheck, R., 2004. Trauma, change in strength of religious faith, and mental health service use among veterans treated for PTSD. J. of Nerv. Ment. Dis. 192, 579584. George, L.K., Ellison, C.G., Larson, D.B., 2002. Explaining the relationships between religious involvement and health. Psychol. Inq. 13, 190200. Goldstein, A.L., Walton, M.A., Cunningham, R.M., Trowbridge, M.J., Maio, R.F., 2007. Violence and substance use as risk factors for depressive symptoms

R.E. Dew et al. / Journal of Affective Disorders 120 (2010) 149157 Wallace Jr., J.M., Brown, T.N., Bachman, J.G., LaVeist, T.A., 2003. The inuence of race and religion on abstinence from alcohol, cigarettes and marijuana among adolescents. J. Stud. Alcohol 64, 843848. Wright, L.S., Frost, C.J., Wisecarver, S.J., 1993. Church attendance, meaningfulness of religion, and depressive symptomatology among adolescents. J. Youth Adolesc. v22, 559(510).

157

Zhang, J., Jin, S., 1996. Determinants of suicide ideation: a comparison of Chinese and American college students. Adolescence v31, 451(417). Zuckerbrot, R.A., Jensen, P.S., 2006. Improving recognition of adolescent depression in primary care. Arch. Pediatr. Adolesc. Med. 160, 694704.