Psychometric Analysis of the Life Worries Scale for a New Generation of Sexual Minority Men: The P18 Cohort Study

Abstract

Sexual minority men (SMM) in the United States continue to experience adverse health problems and psychosocial burdens. However, there is limited psychometric research seeking to quantify the life worries of this population. Informed by syndemic theory, the Life Worries Scale (LWS) was developed to measure the concerns of young SMM. Analyses of the scale were undertaken using baseline data (n = 665) from an ongoing cohort study of emerging adult, SMM. Exploratory factor analyses (EFA) of an initial set of 24 Likert-type items followed by confirmatory factor analysis (CFA) and an exploratory structural equation model (ESEM), indicated a structure consisting of six domains of worries: financial stability, social stability, self esteem, loneliness, physical appearance, and physical health. These six subscales were highly correlated and also demonstrated high levels of internal consistency. Differences in life worries were noted across demographic states, specifically HIV serostatus, sexual attraction, housing status, and self-rated health. High levels of association were also detected between all six sub-scales with both depression and PTSD, while significant correlations were detected between suicidality and both self esteem and loneliness related worries. The results of our analyses provide evidence for the strong psychometric characteristics of the LWS. This newly developed instrument should be utilized in research to examine the extent to which life worries explain health outcomes and risk behaviors in sexual minority males, and may be potentially extended for use in other populations.

Lesbian, gay, bisexual and transgender (LGBT) populations experience greater health-related disparities, in part exacerbated by experiences of discrimination and prejudice (Institute of Medicine, 2011), compared to the general population. Within this population, young sexual minority men (SMM) including gay, bisexual, and other men who have sex with men are particularly vulnerable as they are at heightened risk for contracting HIV and other sexually transmitted infections (STIs) (Centers for Disease Control & Prevention, 2012; 2015). There is also strong evidence from prior studies indicating that young SMM are at increased risk for a range of negative health conditions including, but not limited to, anxiety (Cochran, Sullivan, & Mays, 2003; Pachankis & Goldfried, 2007), depression (Cochran et al., 2003), and substance use (Cochrna, Ackerman, Mays, & Toss, ), and suicidcality (Marshal et al., 2008; McCabe, Hughes, Bostwick, West, & Boyd, 2009), as compared to the general population. In addition, individuals who identify as members of a sexual minority experience economic and legal discriminatory practices. First, members of sexual minority groups are often paid less than their heterosexual counterpart (Elmslie & Tebaldi, 2007), they experience heightened workplace discrimination (Pfizer, Sears, Mallory, & Hunter, 2011) and laws prohibiting discrimination based on sexual orientation status are not uniformly applied across state and local jurisdictions (Cramer et al., 2017).

Despite a growing body of research highlighting health and economic disparities in sexual minority groups (Elmslie & Tebaldi, 2007; Halkitis, Moeller, et al., 2013; Mustanski, Garofalo, Herrick, & Donenberg, 2007), studies that seek to understand the life experiences of SMM continue to focus on HIV-related concerns. This narrow focus undermines our ability to fully understand the potentially more complicated life-related worries and their impact on the health in this new generation of gay and bisexual men as compared to the generation that preceded them, and for whom those formative years were defined by AIDS (Halkitis, 2013). Given this context, the goal of this manuscript is to provide a more holistic framework to understand the subjective life experiences and present day worries of a new generation SMM.

In the extant literature, syndemic theory is widely applied to understand the health of SMM (Halkitis, Wolitski, & Millet, 2013; Stall, Friedman, & Catania, 2008). This conceptualization of health as a means of examining the drivers of multiple overlapping and mutually reinforcing health epidemics (i.e., the syndemic) provides a robust framework for contextualizing the worries that young men may experience. Syndemic theory was originally conceived as a paradigm for understanding HIV as a health disparity in the urban poor (Singer, 1994) and over time extended to explaining the HIV health disparity more broadly (Singer & Clair, 2003). The underlying tenets of this conceptual model nest HIV within a constellation of interrelated health challenges (i.e., the syndemic) that are directed and exacerbated by psychosocial stressors. These psychosocial stressors, such as homophobia, prejudice and state sanctioned discrimination, negatively impact the health of YMSM as evidenced in evaluation of the framework using both additive (e.g., Stall et al., 2003) and latent construct models (Halkitis, Moeller, et al., 2013). Given this background of syndemic theory, its framework also informs the examination of worries in SMM and hence the measure described here.

The worries of SMM include health related concerns that extend beyond HIV to include problems with alcohol and substance use, concerns regarding social well-being and loneliness or isolation, self esteem, and financial issues. The extant literature over the last several decades has clearly documented these challenges and interconnectedness of HIV and sexual risk taking with substance use (Halkitis et al., 2011; McCarty-Caplan, Jantz, & Swartz, 2014; Mustanski, Newcomb, & Clerkin, 2011), loneliness and isolation (Halkitis, Siconolfi, Fumerton, & Barlup, 2008; Hubach et al., 2015; Li, Hubach, & Dodge, 2015), and self esteem (Chaney & Burns-Wortham, 2015; Hubach et al., 2015; Meyer, 2003) in SMM. Many of these complex challenges are exacerbated by the expectation of physical attractiveness (Moskowitz, Rieger, & Seal, 2009) and experiences of body dissatisfaction (Siconolfi et al., 2016) in SMM, attributable in part to the physical effects of the HIV epidemic (Halkitis, Green, & Wilton, 2004). In addition, concerns regarding housing and housing instability, work and financial security are also related to the health of SMM.

Despite the abundance of evidence documenting the numerous challenges that SMM experiences, no single measure has sought to assess these worries in a comprehensive manner. Moreover, while not traditionally examined as a psychosocial challenge (i.e., worry) in a syndemic framework, financial- and work-related security are two key factors that need to be included in the development of a worries scale for this new generation of young SMM. This is especially important in light of the poor financial conditions created by Great Recession of 2008 (United Nations, 2017) both in the United States and globally (Scarpetta, Sonnet, & Manfredi, 2010).

To date, most studies that include some examination of the worries of SMM only examined those that were limited to HIV-related worries. These HIV-related worries are understood as comprising a facet of the construct ‘perceived threat.’ As described by Crosby et al. (2001) they constitute a “product of perceived risk and perceived severity relative to a given disease or event,” or, more specifically, worries that relate to discrepancies between risk behavior and perceived likelihood of contracting the HIV virus” (p. 208). Despite this conceptualization, attempts to rigorously and comprehensively measure HIV-related worries remain limited. At present, only one psychometrically tested measure was found among studies assessing multidimensional aspects of worries in connection with HIV. The Worry About Sexual Outcomes Scale (WASO) is a 10-item measure designed by Sales et al. (2009) for use heterosexual adolescent females. However, in the extant literature, it is not uncommon for studies to assess worry about HIV trough a single item e.g., Dolcini, Catania, Choi, Fullilove, & Coates, 1996.)

While understanding the drivers of HIV in SMM has dominated the literature for some three decades, recent advances in HIV prevention and treatment and the emergence of a new generation of young men in this era of effective treatments (Crosby et al., 2001) requires that our understanding of the current HIV epidemic be nested within a broader framework. Specifically, a framework that acknowledges that the concerns of gay men does not revolve solely around HIV is warranted. To this point, studies focusing on HIV-related worries perpetuate a myopic understanding of the broader life experiences of this vulnerable population. Moreover, the life experience of this new generation, the Queer Generation (Halkitis, 2016), who came of age in the last two decades differ from generations that preceded them, namely the Stonewall and AIDS Generations (Halkitis, 2013), who came of age in the latter half of the 20^th century. As a result of these generational differences, an understanding of issues that impact this generation of young SMM warrants understanding the broader constellation of worries in their lives in the current socio-political and economic context. This, in turn, will allow for a more meaningful examination of the associations between worries and overall health and well-being. But to assess the impact of such worries, we must first be able to effectively measure the construct.

There is no validated measure that comprehensively captures the challenges and life worries of a new generation of young SMM. Furthermore, no research studies have examined how these life worries may vary by sociodemographic characteristics, such as race, ethnicity, sexual attraction, and socioeconomic status, within SMM.

In this paper, we (1) describe the Life Worries Scale (LWS), a measure designed to capture the concerns of a new generation of SMM; (2) provide information on the psychometric properties of the LWS; (3) examine differences in life worries by demographic states; and (4) assess the associations between life worries and mental health indicators. Documenting and delineating the life worries of a new generation of SMM through a psychometrically tested instrument provides us with knowledge to more fully understand the life stressors of a new generation of SMM, as well as the impact of these burdens on health and well-being. In turn, this provides critical knowledge that may be used to develop programming that addresses these stressors and their impact on the health of new generation of SMM.

Method

Design

The P18 Cohort Study, launched in 2009, is an on-going prospective cohort study of the health of SMM (Halkitis, Moeller, et al., 2013). The study seeks to examine the development of health conditions in the population of gay, bisexual and other SMM as they emerge into adulthood (Arnett, 2000), as well as the biological, psychological, and social factors that drive these health states both within and across time. The objectives of the study are informed by syndemic theory (Halkitis, Wolitski, et al., 2013).

Initial recruitment for the P18 Cohort study began in 2009, and a total of n=600 SMM, between 18–19 years at screening, were recruited and completed semi-annual follow-up visits over a 3 year period. In 2014, the cohort was opened to allow for additional recruitment into the study. During this open recruitment period, n=274 participants (41.2% of the original cohort) were retained and an additional n=391 participants were newly enrolled into the cohort, yielding a sample of n=665 study participants.

Recruitment into the study, at both time points, was conducted via active and passive methodologies. For the 2014 recruitment phase, the majority of the study sample (73.9%) was recruited via social networking and other websites as well as social and dating apps. Approximately 5% were recruited via respondent driven sampling, and the remaining participants were recruited at community events and other social venues. Eligibility criteria for new recruits into the cohort included being between the ages of 22–23 years old, in order to ensure the same age at study entry as participants continuing in the study from the original. Across both recruitment periods, eligible participants had to report engaging in sex with a man in the six months preceding baseline assessment and report an HIV-seronegative status (which was verified via baseline testing).

At the baseline study visit, participants completed an audio computer assisted survey instrument (ACASI) that collects information on sociodemographic characteristics, behavioral factors, attitudes, psychosocial factors, and psychological measures.

All participants provided written, informed consent to take part in study related activities. This study protocol was approved by the University Committee on Activities Involving Human Subjects at New York University and holds a federal certificate of confidentiality issued by the Department of Health and Human Services.

Measures

Sociodemographic characteristics

Information on Race/ethnicity was collected via participant self-report. Based on participant responses, race/ethnicity was categorized as Hispanic, Asian/non-Hispanic, Black/non-Hispanic, White/non-Hispanic, Mixed Race/non-Hispanic or Other/non-Hispanic. Sexual attraction was determined using the Kinsey scale with values ranging from 0 (exclusively heterosexual) to 6 (exclusively homosexual) (Kinsey, Pomeroy, Martin, & Sloan, 1948). For the present study, participants who selected a ranking of 6 on the Kinsey scale were categorized as exclusively homosexual, while those who selected a ranking between 0 – 5 comprised the not exclusively homosexual group. Current school enrollment was dichotomized as yes vs. no. Housing status was assessed using a single question asking, “Where do you currently live?.” Responses were categorized as: “on my own,” “with family”, “with roommates,” “in a dorm/at school,” and “unstably housed.” Information on individual annual income was obtained and collapsed into the following three categories:, “less than $5,000,” “between $5,000 and $25,000,” and “more than $25,000.” Participants self-related health (SRH) was ascertained using the single measure item “In general, how is your health?” with standard responses including “Excellent,” “Very Good,” “Good,” “ Fair,” or “Poor.” As with prior studies including a SRH measure, we dichotomized the responses “fair” and “poor” into one group and compared them to “Excellent,” “very good,” and “good.”

Life Worries Scale

The development of the LWS was conducted as part of the P18 Cohort Study. Between April through June 2014, a subset of n=35 study participants were selected to complete an additional survey component at the conclusion of their ACASI assessments. Participants were selected to ensure diversity with respect to race/ethnicity (n=3 were Asian/non-Hispanic, n=5 were Black/non-Hispanic, n=14 were White/non-Hispanic, n=9 were Hispanic, n=2 were of mixed race/ethnicity, and n=2 identified as being of “other” race/ethnicity. At the time of this additional assessment component, participants were on average 21.4 years old (SD = 0.56). As part of this component, participants were asked to answer the following question: “Thinking about what worries you most in life, please list worries in your life ordered from what worries you the most (#1) to what worries you the least (#10).” Participants were asked to write down 1 – 10 of their most important current worries.

Responses were thematically coded and tallied as per Miles and Huberman (1994) and data collection was discontinued when we reached saturation. Responses were reviewed in real-time as per techniques outlined in Fusch and Ness (2015). Theme saturation was determined to be the point at which no new themes (worries) emerged for 5 consecutive participants. The themes that did emerge during real-time review revolved around finances, well-being of family, friends, relationships, social, physical well-being, appearance, health, identity, career/employment, schooling, and housing. These themes as well as participants actual response items were employed to create the current 40-item LWS. For each of the 40 scale items, participants were asked to indicate their level of agreement with each worry on a scale ranging from 1 =Strongly Disagree to 5 =Strongly Agree. Each scale item was in the form of a declarative sentence beginning with the stem “I worry about…”. The LWS was administered to all n=665 participants following open recruitment into P18 Cohort Study.

Mental health

We ascertained information on experiences of depression symptoms for the two weeks preceding assessment using the Beck Depression Inventory, 2nd edition (BDI-II) (Beck, Steer, & Brown, 1996). The BDI-II is a 21-item self-report depression inventory with responses to items ranging from 0 (None) to 3 (Very Much). The BDI-II has been shown to have high construct validity for the depressive symptoms it measures, these include cognitive, affective, and somatic symptoms (Harris & Joyce, 2008). Finally, the BDI-II has demonstrated high internal consistency in our sample (α = .93). Experiences of post-traumatic stress disorder (PTSD) symptoms during the two weeks preceding assessment were measured using the 17-item PTSD Checklist (PCL) (Blanchard, Jones-Alexander, Buckley, & Forneris, 1996). Responses options ranged from “not at all” to “extremely” and were measured on a five point Likert scale. Internal consistency of the PCL was also high in this sample(α = .93). Finally, suicidal ideation and attempts were examined using two items also used in The National Longitudinal Study of Adolescent Health (Harris & Udry, 2001). Suicidal ideation (yes/no) was assessed using the question “During the past 12 months, did you ever seriously think about committing suicide?” If the individual reported suicidal ideation, we then ascertained information on the number of suicide attempts in the previous 12 months, “During the past 12 months, how many times did you actually attempt suicide?” These two questions were part of the suicidal ideation questionnaire that measures suicidal cognition in adolescents and young adults (Reynolds, 1991).

Participants

Table 1 provides a summary of the sociodemographic characteristics of the sample. The sample was racially/ethnically diverse, with approximately three-quarters of participants identifying as members of racial/ethnic minority groups. Approximately half of the sample identified an exclusively homosexual sexual attraction. Only 33.4% (n = 222) reported currently being in school. The majority of participants reported either living with family (38%, n = 253), or with roommates (34%, n = 226). Self-reported annual, individual incomes of $25,000 or less were also reported by a majority of this sample. Finally,, nearly three-quarters (73.4%, n= 488) of the sample reported their health as being excellent or very good.

Table 1.

Demographic characteristics of the sample (N = 665).

	N (%)
Race/Ethnicity
Hispanic/Latino	214 (32.2)
Asian Non-Hispanic	51 (7.7)
Black Non-Hispanic	182 (27.4)
White Non-Hispanic	167 (25.1)
Other Non-Hispanic	11 (1.7)
Mixed Race Non-Hispanic	40 (56.0)
Conformed HIV Serostatus
HIV-positive	33 (5.0)
HIV-negative	630 (94.7)
Test not undertaken	2 (0.3)
Sexual Attraction
Exclusively. homosexual	333 (50.1)
Not exclusively homosexual	332 (49.9)
School Enrollment
Currently enrolled	222 (33.4)
Not enrolled	443 (66.6)
Housing Status
On my own	101 (15.2)
With family	253 (38.0)
With roommates	226 (34.0)
Dorm/school	20 (3.0)
Unstably housed	59 (8.9)
Other	6 (0.9)
Annual Income
< 5K	219 (32.9)
≥5K and < 25K	274 (41.2)
≥ 25K	172 (25.9)
Self-rated Health
Excellent	210 (31.6)
Very Good	278 (41.8)
Good	142 (21.4)
Fair or Poor	32 (4.8)
Missing	3 (0.5)

There was also strong association between the demographic variables. Race/ethnicity was found to be associated with HIV status (χ2 (5) = 11.18, p < .05), school enrollment status (χ2 (5) = 16.74, p < .01), and total annual income (χ2 (10) = 19.09, p < .05). A greater proportion of participants who tested HIV-positive identified as Black (8.3%), Hispanic/Latino (5.1%), or Mixed Race (7.5%) than those who identified as White (1.2%) or Asian (2.0%). A smaller proportion of White participants (24.6%) were currently enrolled in school, as compared to 34.0% of Hispanic/Latino, 39.0% of Black, and 41.2% of Asian participants. Race/ethnicity was also found to be significantly associated with housing status (χ2 (20) = 102.66, p < .001), and self-rated health (χ2 (15) = 35.35, p < .01). A higher proportion of participants identifying as Hispanic or Black reported currently living with family (53.8% and 43.3%, respectively), as compared to 18.1% of Whites. Not surprisingly, annual personal income was related to housing status (χ2 (8) = 59.23, p < .001). Approximately half of those with incomes of $5,000 or less live with their families as compared to 28% of those who have an annual income of $25,000 or more.

Statistical Analyses

First, we undertook an analysis of all 40 items of the LWS for content adequacy, a key step in scale development (Hinkin, Tracey, & Enz, 1997), to determine which of the 40 items aligned with the theory of syndemics, which was our overarching paradigm for the psychometric analysis of the scale. Five independent reviewers assessed each items, and then through a consensus building discussion, items that were deemed not to be related to any of the theoretical components of syndemics theory were eliminated prior to undertaking exploratory factor analysis (EFA). This initial step resulted in a final set of 24 items. These 24-items and the distribution of responses of the 5-Likert scale categories are shown in Table 2. Using the initial 24-items from the content analysis we conducted an exploratory factor analysis (EFA). Due to the ordered nature of the original 24-items, we treated them as categorical. In addition, we utilized the weighted least-squares with mean and variance adjustment (WLSMV) as our estimation procedure, as it has been shown to be superior to the maximum likelihood estimation procedure for ordered-categorical variables (Muthén & Muthén, 2005). Initially, we requested factor solutions for 1- to 6-factors. Goodness-of-fit was evaluated with the root mean square error of approximation (RMSE), comparative fit index (CFI), and the Tuker-Lewis index (TLI) (Hu & Bentler, 1999; Marsh, Hau, & Wen, 2004). We dropped ambiguous items (items that loaded on to more than one factor) and items with factor loading scores of less than .03 initially, and repeated this process until we arrived at an acceptable factor solution. All analyses were conducted in MPLUS v7.31 (Muthén & Muthén, 2012). We then conducted exploratory structural equation modeling (ESEM), a relatively new estimation procedure created to address some of the limitations of confirmatory factor analysis (CFA) (Asparouhov & Muthén, 2009; Marsh et al., 2009; Marsh, Liem, Martin, Morin, & Nagengast, 2011) to confirm the results of the EFA. Alpha coefficients were computed for each of the identified subscales, as well as total score, and subscale intercorrelations were computed. The total score and subscales were examined in relation to key demographic states using bivariable procedures. Finally, we considered the extent to which the scale was associated with key mental health indicators as a further step in establishing evidence for validity.

Table 2.

Descriptive statistics of items of the Life Worries Scale (LWS)

Item/Worry	Strongly Disagree % (n)	Disagree % (n)	Neither % (n)	Agree % (n)	Strongly Agree % (n)
Finances/having enough money	10.5 (68)	10.5 (68)	9.9 (64)	41.6 (274)	27.6 (179)
Having a career	7.6 (49)	6.5 (42)	11.0 (71)	40.8 (264)	34.2 (221)
Being successful in career	7.3 (47)	6.8 (44)	10.2 (66)	37.5 (242)	38.2 (247)
Not enjoying job	11.1 (72)	15.0 (97)	16.8 (109)	34.0 (220)	23.0 (249)
Having a job	10.6 (69)	13.7 (89)	14.3 (93)	36.6 (238)	24.8 (161)
Well-being of family	10.3 (67)	13.9 (90)	14.0 (91)	43.1 (280)	18.6 (121)
Losing touch with family	11.6 (75)	22.0 (143)	20.8 (135)	31.3 (203)	14.3 (93)
Well-being of friends	7.4 (48)	10.5 (68)	14.8 (96)	51.0 (331)	16.3 (106)
Losing touch with friends	8.7 (56)	14.4 (93)	16.6 (107)	43.0 (278)	17.3 (112)
Well-being of romantic partners	8.5 (53)	9.5 (59)	19.6 (122)	42.4 (264)	20.1 (125)
Not being popular	23.3(150)	30.7 (198)	22.3 (144)	17.1 (110)	6.7 (43)
Not having enough friends	22.2(143)	32.1 (207)	18.9 (222)	19.1 (123)	7.8 (50)
Being alone	16.6(107)	23.4 (151)	16.4 (106)	28.8 (186)	14.9 (96)
Not being happy	14.6 (94)	18.6 (120)	14.7 (95)	33.0 (213)	19.2 (124)
How good my body looks	9.7 (63)	11.7 (76)	16.4 (104)	42.3 (274)	20.1 (130)
Gaining weight	13.1 (85)	18.7 (121)	15.8 (102)	35.9 (232)	15.3 (99)
Not being good-looking	12.5 (81)	17.8 (115)	18.1 (117)	36.3 (235)	15.3 (99)
Not being sexy	13.8 (89)	18.2 (118)	18.7 (121)	33.7 (218)	15.6 (101)
How smart I am	15.3 (99)	25.2 (163)	18.1 (117)	28.7 (186)	12.8 (83)
HIV	12.0 (78)	17.7 (115)	19.3 (125)	37.0 (240)	14.0 (91)
STIs other than HIV	11.4 (74)	14.6 (95)	19.4 (126)	41.4 (269)	13.2 (86)
Health issues other than HIV/STIs	10.5 (68)	14.6 (95)	18.6 (121)	42.4 (275)	13.9 (90_
How much alcohol I drink	25.3 (163)	35.9 (231)	20.7 (133)	12.7 (82)	5.4 (35)
My drug use	35.5 (223)	29.5 (185)	191. (120)	10.8 (68)	5.1 (32)

Results

Exploratory Factor Analysis

A series of initial exploratory factor analysis of the 24 items were undertaken (handling the items as categorical variables). The 6-factor model, with the retention of 19 of the 24 items yielded best fit (RMSEA = 0.069; CFI = 0.992; TLI = 0.982). Missing data were minimal, with only 17 missing data patterns. The loadings and standard errors derived from the EFA are shown in Table 3, with each set of the worries grouped by factor name, specifically worries associated Financial Stability (n = 5), Social Stability (n = 4), Self Esteem (n = 2), Loneliness (n = 2), Physical Appearance (n= 3) and Physical Health (n = 3). Given that we assumed correlated factors, the loading shown in Table 3 are regression coefficients (Deegan, 1978). Factor loadings were significant at the p = .05 level and ranged from a minimum of .475 to a maximum loading of .920

Table 3.

EFA geomin rotated factor loadings (standard errors) of the Life Worries Scale (LWS)

Factor: WORRY	Factor 1	Factor 2	Factor 3	Factor 4	Factor 5	Factor 6
Factor 1: FINANCIAL STABILITY (n = 5)
Finances/having enough money	.475 (.029)
Having a career	.851 (.021)
Being successful in career	.802 (.019)
Not enjoying job	.480 (.028)
Having a job	.481 (.025)
Factor 2: SOCIAL STABILITY (n = 4)
Well-being of family		.715(.025)
Losing touch with family		.710(.023)
Well-being of friends		.756(.020)
Losing touch with friends		.643(.025)
Factor 3: SELF ESTEEM (n = 2)
Not being popular			.772 (.046)
Not having enough friends			.820 (.054)
Factor 4: LONELINESS (n = 2)
Being alone				.645(.045)
Not being happy				.659(.046)
Factor 5: PHYSICAL APPEARANCE (n = 3)
Gaining weight					.503(.026)
Not being good-looking					.801(.023)
Not being sexy					.811(.021)
Factor 6: PHYSICAL HEALTH (n = 3)
HIV						.805(.013)
STIs other than HIV						.921(.012)
Health issues other than HIV/STIs						.748(.014)

Exploratory Structural Equation Model (ESEM)

The model results for the CFA and ESEM are provided in Table 4. The goodness-of-fit indices for the CFA indicated marginally adequate fit (RMSEA = 0.093; CFI = 0.974; TLI = 0.670), whereas the ESEM was a better fit for the data (RMSEA = 0.069; CFI = 0.992; TLI = 0.982). Table 5 provides the correlations for the CFA and ESEM. Correlations between factors were lower for the ESEM (.145 to .486) compared to the CFA (.385 to .846). Factor loadings were slightly lower for the ESEM (.470 to.921) than the CFA (.705 to .948), however, the factors for both the CFA and ESEM confirmed the factor structure specified in the EFA.

Table 4.

Standardized factor correlation for the CFA and the ESEM of the Life Worries Scale (LWS)

	F1	F2	F3	F4	F5	F6
F1 - Financial Stability	-	.686	.442	.675	.591	.548
F2 - Social Stability	.368	-	.423	.599	.466	.591
F3 – Self Esteem	.145	.169	-	.846	.685	.385
F4 - Loneliness	.281	.335	.186	-	.768	.518
F5 – Physical Appearance	.308	.178	.442	.286	-	.510
F6 – Physical Health	.416	.359	.560	.299	.486	-

Note. All correlations significant at the p < .0001 level. CFA (above the diagonal) and the ESEM (below the diagonal)

Table 5.

CFA and ESEM standardized geomin rotated factor loadings and standard errors of the Life Worries Scale (LWS)

	CFA	ESEM
	Factors	Factor 1	Factor 2	Factor 3	Factor 4	Factor 5	Factor 6
Factor 1: FINANCIAL STABILITY
Finances/having enough money	.705 (.023)	.475(.029)	.164(.028)	−.053(.047)	.119(.052)	.143(.035)	.070(.031)
Having a career	.948(.009)	.850(.021)	.029(.014)	.006(.021)	.126(.022)	.069(.018)	.072(.017)
Being successful in career	.964(.009)	.805(.019)	.050(.016)	.027(.027)	.108(.029)	.101(.025)	.094(.019)
Not enjoying job	.787(.017)	.480(.028)	.151(.027)	.170(.038)	.193(.041)	.044(.031)	.019(.027)
Having a job	.809(.017)	.480(.025)	.197(.023)	.083(.040)	.163(.046)	.087(.032)	.076(.025)
Factor 2: SOCIAL STABILITY
Well-being of family	.860(.012)	.175(.024)	.715(.024)	.127(.031)	−.045(.034)	.039(.022)	.114(.021)
Losing touch with family	.803(.013)	.012(.027)	.710(.023)	.084(.030)	.075(.037)	.054(.024)	.089(.020)
Well-being of friends	.878(.011)	.091(.020)	.757(.020)	−.024(.028)	.137(.033)	.023(.022)	.108(.018)
Losing touch with friends	.890(.011)	.010(.022)	.644(.025)	.015(.030)	.342(.035)	.036(.021)	.060(.018)
Factor 3: SELF ESTEEM
Not being popular	.899(.011)	−.019(.023)	.062(.019)	.778(.047)	.055(.031)	.133(.021)	.034(.015)
Not having friends	.947(.010)	−.001(.021)	−.033(.017)	.812(.055)	.192(.050)	.051(.016)	.017(.014)
Factor 4: LONELINESS
Being alone	.880(.010)	−.004(.02)	−0.03(.017)	.291(.034)	.649(.049)	.123(.019)	.002(.084)
Not being happy	.920(.009)	.088(.024)	0.024(.02)	.158(.023)	.656(.045)	.127(.018)	.084(.016)
Factor 5: PHYSICAL APPEARANCE
Gaining weight	.776(.018)	.102(.029)	.073(.027)	.095(.033)	.089(.036)	.503(.026)	.134(.026)
Not being good-looking	.954(.007)	.038(.014)	−0.01(.013)	.130(.018)	.125(.020)	.800(.023)	.020(.012)
Not being sexy	.929(.009)	.011(.015)	0.003(.012)	.109(.017)	.101(.020)	.811(.021)	.043(.012)
Factor 6: PHYSICAL HEALTH
HIV	.895(.008)	.044(.017)	.041(.016)	.063(.024)	.025(.028)	.082(.017)	.805(.013)
STI’s other than HIV	.959(.005)	.035(.012)	.049(.010)	.055(.015)	.054(.018)	.017(.012)	.921(.012)
Health issues other than HIV/STIs	911(.007)	055(.018))	.09(.015)	.002(.026)	.123(.030)	.091(.018)	.748(.014)

Internal Consistency and Associations of Subscales

Based on the factor structure that was confirmed, the internal consistency of each of the 6 worries subscales was estimated using Cronbach’s α and examining the effect of individual item removal on the α of each subscale. All subscales indicted a high level of internal consistency: Financial Well Stability (α = .88; n = 5); Social Stability (α = .89; n = 4); Self Esteem (α = .89; n = 2); Loneliness (α = .86; n = 2); and Physical Appearance (α = .86; n = 3); and Health (α = .92; n = 3). The α for the 19-items total score was estimated at .94. In no case did the removal of any individual item form any of the subscales result in on increase in the α.

The six subscales of the LWS were found to be highly positively correlated ranging from a high of r = .69 (p < .001) for the association between Self Esteem and Loneliness related worries to r = .32 (p < .001) for the association between Self Esteem and Health related worries.

Associations of the Life Worries Scale with Demographic States

We examined the difference in the six subscale scores and total score for the LWS across the key demographic states. These are shown in Table 6. No differences were noted for race/ethnicity and school enrollment. Differences were observed in relation to confirmed HIV serostatus with HIV-negative men reporting higher levels across all classes of worries than HIV-positive participants. Participants who reported not being exclusively homosexual also reported higher levels of worries across all scales other than physical health related worries, than those participants identifying as exclusively homosexual. With regard to income, YMSM indicating an annual personal income of $25,000 or greater reported lower levels of worries about physical health (p = .002) than those reporting an annual income of less than $5,000. Those reporting an income of less than $5,000 also reported higher levels of worries about health than those reporting between $5,000 and $25,000 annual income (p = .02). Differences were also noted for worries along housing status on all subscales other than worries about social stability. Participants living on their own reported lower levels of financial stability worries than those living with a roommate (p =.011) and in a dorm or at school (p = .004). Those living in a dorm or at school reported more loneliness related worries than any other groups (p’s ranging from .003 to .022) Finally, across all five classes of worries, those who rated their health as excellent expressed less concern than those reporting their health as poor or fair (p ranging from < .001 to .008); increased levels of worries across all five subscales were noted as self-rated health decreased from excellent, to very good, to good, and finally to fair or poor.

Table 6.

Subscale and overall scale scores for the Life Worries Scale (LWS) by demographic characteristics

	Financial	Social Stability	Self Esteem Stability	Loneliness	Physical Appearance	Physical Health	Total Score
	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)
	(n = 643)	(n = 646)	(n = 644)	(n = 645)	(n = 646)	(n = 648)	(n = 635)
	18.40 (5.10)	13.65 (4.12)	5.12 (2.32)	6.25 (2.51)	9.67 (3.41)	9.88 (3.37)	63.01 (16.06)
Race/Eth
Hisp/Lat	18.00 (5.32)	13.79 (4.30)	5.09 (2.41)	6.10 (2.63)	9.59 (3.59)	9.91 (3.56)	62.58 (16.70)
Asian	18.83 (4.53)	13.13 (3.46)	5.77 (2.19)	6.73 (2.11)	10.35 (3.12)	9.98 (2.71)	64.79 (13.86)
Black	18.01 (5.70)	13.66 (4.67)	4.85 (2.38)	5.98 (2.63)	9.20 (3.71)	10.06 (3.74)	61.73 (18.47)
White	19.06 (4.15)	13.51 (3.56)	5.32 (2.15)	6.46 (2.25)	10.05 (3.02)	9.59 (2.92)	64.03 (13.00)
Other	17.72 (6.15)	13.50 (4.48)	5.06 (2.41)	6.67 (2.83)	9.44 (3.18)	10.00 (3.63)	62.39 (19.09)
Mix. Race	19.33 (4.63)	14.22 (3.42)	4.78 (2.23)	6.69 (2.48)	9.75 (2.81)	9.92 (3.01)	64.69 (14.03)
HIV Status	***	*	**	**	***	**	***
Positive	15.10 (6.73)	11.93 (5.20)	4.03 (1.99)	5.03 (2.68)	6.52 (3.04)	8.17 (4.27)	50.79 (19.18)
Negative	18.55 (4.96)	13.71 (4.05)	5.16 (2.31)	6.31 (2.48)	9.82 (3.35)	9.95 (3.30)	63.56 (15.64)
Sexual Attract.	*	**	*	***	+		**
Ex. homo.	17.93 (5.37)	13.19 (4.38)	4.90 (2.36)	5.94 (2.59)	9.42 (3.51)	9.70 (3.56)	61.08 (16.82)
Not	18.88 (4.76)	14.13 (3.79)	5.34 (2.26)	6.58 (2.38)	9.92 (3.29)	10.06 (3.16)	65.01 (14.98)
School Enrollment
Enrolled	18.45 (5.36)	13.86 (4.32)	5.18 (2.39)	6.26 (2.65)	9.86 (3.50)	10.05 (3.56)	63.73 (17.31)
Not	18.37 (4.97)	13.54 (4.02)	5.08 (2.28)	6.25 (2.43)	9.57 (3.36)	9.79 (3.27)	62.64 (15.39)
Housing Status	**		*	+	**	***	**
On own	17.46 (5.45)	12.70 (4.36)	5.14 (2.23)	6.22 (2.49)	9.90 (3.48)	8.78 (3.51)	60.16 (17.57)
Family	18.16 (5.35)	13.86 (4.27)	4.99 (2.21)	6.21 (2.63)	9.48 (3.48)	10.02 (3.38)	62.78 (16.67)
Roommate	19.05 (4.37)	13.72 (3.67)	5.26 (2.34)	6.36 (2.29)	9.99 (3.11)	10.00 (3.11)	64.41 (13.57)
School	21.05 (3.07)	14.80 (4.11)	6.40 (2.52)	7.70 (2.30)	11.37 (2.73)	11.90 (2.36)	72.95 (11.03)
Unstably	17.88 (6.06)	13.83 (4.67)	4.72 (2.65)	5.79 (2.69)	8.55 (3.88)	9.92 (3.96)	61.07 (19.51)
Annual Income						**	+
<5K	18.73 (5.26)	13.98 (4.12)	5.34 (2.32)	6.42 (2.53)	9.70 (3.42)	10.45 (3.31)	64.67 (16.42)
5K &< 25K	18.62 (4.99)	13.74 (4.12)	5.03 (2.33)	6.23 (2.50)	9.70 (3.48)	9.74 (3.43)	63.10 (16.36)
≥25K	17.60 (5.00)	13.08 (4.10)	4.97 (2.30)	6.07 (2.50)	9.57 (3.29)	9.35 (3.25)	60.70 (14.86)
Self-rated Health	***	***	**	***	***	**	***
Exc.	16.93 (5.87)	12.82 (4.74)	4.86 (2.27)	5.67 (2.50)	8.69 (3.50)	9.28 (3.68)	58.31 (18.19)
V. good	18.59 (4.65)	13.73 (3.70)	4.98 (2.22)	6.14 (2.42)	9.69 (3.30)	9.99 (3.19)	63.19 (14.29)
Good	19.69 (4.34)	14.17 (3.78)	5.54 (2.38)	7.04 (2.40)	10.67 (3.27)	10.28 (3.14)	67.40 (14.18)
Fair/poor	20.68 (4.04)	15.94 (3.60)	6.19 (2.70)	7.72 (2.37)	11.47 (3.22)	10.97 (3.27)	73.03 (13.71)

^*p ≤ .05;

^**p ≤.01;

^***p ≤.001

⁺p = .06

Associations of the Life Worries Scale and Mental Health Indicators

As a final step, we examined the extent to which each of the six subscales of the LWS was related to the following four mental health indicator variables: depression, PTSD, suicidal ideation in the last month, and suicide attempts in the last month. Depression was highly associated with each of the six subscales ranging from r = .27 (p < .001) for Social Stability related worries to .42 (p < .001) for Loneliness related worries, as well as with the total LWS score (r = .41, p < .001). Similarly, significant associations were detected for the subscales with PTSD ranging from .26 (p < .001) for Self Esteem related worries to r = .39 (p < .001) for Loneliness; PTSD was also related to total LWS score (r = .36, p < .001). Suicidal ideation and was correlated with Loneliness (r = .11, p < .01), while number of suicide attempts in the year prior to assessment was associated with both Self Esteem (r = .09, p <.05) and Loneliness (r = .11, p < .01).

Discussion

As part of our ongoing cohort study of SMM, we sought to understand the life worries experienced by a new generation of YMSM. The impetus for understanding these stressors or worries was to differentiate the life experiences of the young men in the Millenlial Generation from the previous generations of gay men whose lives were shaped and defined, in great part, by the AIDS epidemic (Halkitis, 2013). For Millennials, who came of age at the onset of the 21^st century, the experience of HIV is radically different from those who came of age in the 1980’s and 1990’s. Since biomedical advances have enhanced HIV treatment, HIV has been transformed into a chronic disease, if managed correctly (Deeks, Lewin, & Havlir, 2013). In addition, HIV prevention in the form of pre-exposure prophylaxis (PrEP) and post-exposure prophylaxis (PEP) has empowered HIV-negative men with another tool to avert infection (Mansergh et al., 2010). Moreover, treatment of HIV-positive gay men has also contributed to the HIV prevention toolkit in so much as the likelihood of HIV transmission to a seronegative person is extremely unlikely, and potentially near zero, if the HIV-positive individual has a sustained undetectable viral load (Grulich et al., 2015). This is not to suggest that HIV does not remain a worry for many young SMM in light of the biomedical advances as previously demonstrated (Kalichman, 1998), but the relative intensity of the worry may be tempered in light of increasingly successful HIV prevention and treatment (Vanable, Ostrow, McKirnan, Taywaditep, & Hope, 2000).

While these advances in HIV treatment and prevention have created a sense of hope in the gay community, the life experiences of young gay and bisexual men coming of age in the last two decades, however, have been burdened by the economic downturn and ensuing economic conditions in the United States (Lee Badgett, Durso, &, Schneebaum, 2013).. In recent years the unemployment rate of young and emerging adults has dramatically increased particularly among those with less than a college education (Fry & Passel, 2014). These economic conditions have impeded the path to fiscal independence for many young adults, affecting all aspects of their lives including the ability to live independently (Fry & Passel, 2014). In our sample, approximately 38% lived with their families, a proportion that was significantly different for those reporting an annual income of $5,000 or less (45%) when compared to those reporting an income of $25,000 or more (28%); only 9% of those in the lowest income group and 13% of those in the middle income group reported living on their own.

Regardless of the period of coming out and coming of age, young gay men experience the social stressors of transitioning into adulthood and finding their place in the gay community where social norms and conceptions of hypermasculinity abound (Hamilton & Mahalik, 2009; Kimmel & Mahalik, 2005) and where such expectations may have even more deleterious effects on those who are HIV-positive (Halkitis et al., 2004). Taken together, these factors are likely to shape the worries that burden the lives of young gay men. Importantly, HIV is not the only aspect of life that worries young gay men, as there are numerous challenges that young gay men have faced across time and continue to face (D’Augelli & Hershberger, 1993).

It is with this background that we sought to develop the LWS. Our goal was to develop a tool that could effectively assess the stressors experienced by young gay men—stressors that might contribute to the burdens in their lives. To date, no other tool of this type has been developed and psychometrically assessed, and much of the work undertaken regarding the worries of gay men have focused primarily, if not exclusively, on HIV (Bermúdez, Castro, & Buela-Casal, 2011; Nemeroff, Hoyt, Huebner, & Proescholdbell, 2008; Stephenson, White, Darbes, Hoff, & Sullivan, 2015). However, for gay men of all ages - their health is defined by more than just HIV (Halkitis, Kapadia, Ompad, & Perez-Figueroa, 2015), and for a new generation coming of age in the era of effective treatments, the disease likely has differential salience as compared to earlier generations of gay men. To this end, it was our hope that the development of the LWS would provide us with a powerful tool to asses the construct “worries” and ultimately to examine the extent to which this construct contributes to the health and well-being of young gay men. Use of the syndemic theory (Halkitis, 2010; Halkitis, Wolitski, et al., 2013) to guide this study allowed us to also consider how worries may be defined as psychosocial burdens which in turn may function to diminish the health of the individual and population. Utilizing this perspective, those experiencing higher levels of worries (i.e., burdens) are also more likely to demonstrate a myriad of interlocking health problems including, but not limited to, substance use and abuse, mental health problems, poverty, violence, as well the acquisition of sexually transmitted diseases (Halkitis, Wolitski, et al., 2013; Wilson et al., 2014).

Our work has resulted in a robust and psychometrically sound scale, which is highly applicable to the population of SMM. While the scale that we developed and tested was conceived for use with SMM, who are at much higher risk for health disparities than their heterosexual peers (Institute of Medicine, 2011), there is potential applicability of this work to a broader population, specifically for a broader population of young adults. However, it must be noted that the LWS was developed based on qualitative data collection undertaken with emerging adult SMM. Our initial work led to the creation of a 40-item scale which was administered to 665 racially, ethnically, and economically diverse SMM ages 22 to 23 as part of the baseline assessment of an ongoing cohort study. Using syndemic theory as a framework, a content analysis of the items led to the retention of 24-Likert items, which were examined through both exploratory and confirmatory factor analyses, handling the items as categorical variables and ESEM. These analyses resulted in a final scale consisting of 19 items and 6 subscales tapping into six domains of worry, namely concerns about financial stability, social stability, self esteem, loneliness, physical appearance, and physical health, which includes but is not limited to HIV. All six subscales demonstrate high levels of internal consistency. These analyses support robust psychometric properties of the LWS.

Moreover, the factors manifested in the LWS align with an extant literature regarding the challenges faced by SMM. These worries, which have been documented extensively in the literature, encompass the domains of self esteem (Rosario, Rotheram-Borus, & Reid, 1996; Ryan, Russell, Huebner, Diaz, & Sanchez, 2010), loneliness and social well-being (Hubach, DiStefano, & Wood, 2012; Mereish & Poteat, 2015), dissatisfaction with physical appearance and poor body image (Allensworth-Davies, Welles, Hellerstedt, & Ross, 2008; Siconolfi, Halkitis, Allomong, & Burton, 2009), and health inclusive but not solely centered on HIV (Dean et al., 2000; Gruskin, Greenwood, Matevia, Pollack, & Bye, 2007; Halkitis et al., 2015; Mayer et al., 2008). These are in addition to concerns about financial well-being experienced by sexual minority populations which intersect with the aforementioned and also function to diminish overall well-being (Billies, Johnson, Muringi, & Pugh, 2009; Redman, 2010). Taken together, these worries or minority stressors are psychosocial challenges that may diminish the overall health of SMM (Mereish & Poteat, 2015) as understood within syndemic theory (Singer & Clair, 2003) and illustrate the potential drivers of health dipartites faced by LGBT populations (Institute of Medicine, 2011). The LWS is, in effect, an instrument which can effectively assess the level of challenges experienced by SMM, the population for which this scale was developed, more broadly to all SMM and women, and with modification to heterosexual populations.

Interestingly, given the experiences of many LGBT individuals and young gay and bisexual men in particular (Institute of Medicine, 2011; Halkitis, Wolitski, et al., 2013), discrimination did not emerge as a factor. We understand this finding in several ways. First, the factors of social stability, self esteem, and loneliness may function as proxy indicators of discrimination, and thus likely capture a set of those sexual identity related experiences. Second, as we have previously stressed, this scale was developed specifically to measure the experience of you gay and bisexual men, who came of age in the era of marriage equality, greatly differentiating their own experiences form those of previous generations. Finally, as we reiterate in the limitations, the data were collected on a New York City based sample.

Our findings indicate that there are no differences by race/ethnicity across the six classes of worries. While some of the life experiences and psychosocial challenges of racial and ethnic minority gay and bisexual men may be shaped by their racial and ethnic identities (Hightow-Weidman et al., 2011), the instrument that was developed did not assess these burdens, which potentially explains why we identified no differences along race/ethnicity. In addition, no differences emerged with regard to current school enrollment. Not surprisingly, income differentiated the worries that these young men expressed specifically with regard to the physical health subscale such that those indicating lower annual incomes also indicated higher levels of worry about their overall health. This conception aligns with the abundance of research over the last century indicating the impact on income inequality on the health of marginalized populations (Subramanian, Blakely, & Kawachi, 2003; Wilkinson & Pickett, 2006). The association with overall health also likely explains the pattern noted for housing status, which is a proxy for socioeconomic status. In our sample, income and housing status were highly associated and must be understood in relation to the report by Fry and Passel (2014) on multigenerational housing. Across all six categories of worries as well as total score, HIV-negative men reported higher levels of concerns than HIV-positive men. To date, there is no research to indicate why our findings support this difference. Perhaps anxiety about seroconversion may be associated with more heightened generalized anxiety for HIV-negative men. This idea is purely speculative and needs to be explicated further and substantiated with additional research. Finally, those men who indicated non-exclusive homosexual attraction reported higher levels of worries than those who reported exclusive homosexual attraction. This finding may be considered in light of higher rates of psychopathology documented in bisexual men (Cochran et al., 2003).

As a final step, we considered each of the six life worries in relation to indicators of mental health; namely depression, PTSD, and suicidal ideation and attempts. Both depression and PTSD were highly associated with each of the life worries, and suicidal ideation was found to be associated with self esteem and loneliness related worries. These findings begin to provide evidence for the validity of our measure. Moreover, these findings support the notion that life worries may in fact be viewed as psychosocial burdens, which may influence health outcomes; in this case, mental health. (Halkitis, Wolitski, et al., 2013). However, these findings regarding life worries should not be conflated with psychopathology nor should they be interpreted as indicative of elevated psychopathology in this sample.

There are some limitations that require noting. First, results of the CFA provide less than ideal fit statistics, although better fit statistics are achieved with the application of ESEM. The fit of the ESEM, when considered along with the findings of the EFA and assessments of internal consistency, provide robust sufficient evidence to support the 6-factor structure of the LWS. Second, we note that two of the subscales (self esteem and loneliness) each consist of two items. While this may seem like a small number of items to assess a construct, the validity of these two factors are supported by their associations with the other four subscales of the LWS as well as with the mental health indicators of depression and PTSD. Furthermore, there is ample support for the overall scale, which includes these two subscales. We were also limited in our ability to utilize test-rest approaches to analyzing the data. The scale is included in the baseline and Month 36 assessment of the cohort study, which is still ongoing. To this point, the Month 36 data are not yet available. Also, the 3 year period between testing may present too large of a window for effectively conducting examination of test-retest reliability, especially given the large developmental changes that accompany emerging adulthood (Arnett, 2000). In addition, while syndemic theory may not perfectly align with the structure of the LWS, the basic tenets of this conceptual framework provide an overarching paradigm for understanding the psychosocial burdens or worries of SMM. Finally, we note that these data are collected on a sample of young men coming of age in New York City, which potentially creates a different life experience for them as compared to other young gay and bisexual men emerging into adulthood in non-metropolitan areas or in environments that are not socially and politically liberal enclaves. In future, testing of the LWS could and should be undertaken with other segments of the sexual minority population including but not limited to those living in rural areas, as well as older SMM, among others.

In this paper, we provide support for the LWS. This tool provides a psychometrically sound instrument to assess the life stressors experienced by young SMM, which may be helpful to clinicians working with this at-risk population. The tool also holds potential applicability for the use with emerging adults regardless of sexual orientation. Moreover, the tool provides a system for assessing worries as a construct and expands the conception of psychosocial burdens as determinants of negative health sequelae as informed by syndemic theory (Panchakis, 2015). A critical next step will be to determine the extent to which these life worries are associated with health and well-being and the manner through which these effects are mediated.