Personality and everyday functioning in older adults with and without HIV

Abstract

In a cross-sectional multi-method study of older adults living with and without HIV (n = 202; 69.8% HIV seropositive), we tested associations between personality traits and everyday functioning, and whether these associations differed depending on HIV serostatus. We found that higher levels of conscientiousness and lower levels of neuroticism were associated with higher odds of being clinically independent (vs. dependent) in everyday functioning. These findings replicated across self- and clinician-reports and persisted above and beyond relevant covariates. We found no evidence of interactions between personality and HIV serostatus, suggesting that personality was equally important for everyday functioning regardless of HIV serostatus. Given the present findings and the knowledge that personality is dynamic and amenable to intervention, we discuss two different possible pathways for intervention meant to improve everyday functioning and quality of life among older adults with and without HIV: personality change and personalized medicine.

Introduction

There were 36.9 million people living with HIV worldwide as of 2017 (U.S. Department of Health and Human Services). For those who have access to modern combination antiretroviral therapy (cART), life expectancy measured at age 20 is 74.9 years, which is remarkably 43 years higher than in the pre-cART era (Gueler et al., 2017). Accordingly, the prevalence of older people living with HIV has risen considerably (Centers for Disease Control and Prevention, 2018), with some projections indicating that over 70% of people living with HIV will be aged 50 years and older in the coming decade (Smit et al., 2015). However, the decrease in mortality associated with HIV disease has simultaneously introduced a new series of challenges for an aging population of people living with HIV. For example, older people living with HIV are at heightened risk for a variety of age-related medical comorbidities (Rodriguez-Penney, 2013), including cardiovascular disease, cancer, neurocognitive disorders, and frailty (Guaraldi & Palella, 2017). Such challenges are of considerable clinical relevance, as they often accelerate declines in the independent performance of both basic and instrumental activities of everyday living (Fazeli, Woods, & Vance, 2019; Kordovski, Woods, Verduzco, & Beltran, 2017; Tierney, Woods, Sheppard, & Ellis, 2019) and in quality of life (Moore et al., 2014; Rodriguez-Penney, 2013).

Indeed, HIV disease is reliably associated with moderate declines in functional capacity and manifest everyday functioning, including automobile driving (Marcotte et al., 2004), work (Kordovski, et al., 2017), and medication management (Heaton et al., 2004). Approximately two-thirds to three quarters of people living with HIV show clinically remarkable declines in everyday functioning (Blackstone et al., 2013). Among people living with HIV, older age is a robust predictor of more severe functional declines (Morgan et al., 2012; Kordovski, et al., 2017). Other sociodemographic factors, including lower education and minority racial/ethnic status are also associated with worse everyday functioning in the context of HIV disease (Arentoft et al., 2015). As noted above, higher medical comorbidity burden is strongly linked to poorer everyday functioning in HIV (Rodriguez-Penney, 2013), with particularly impactful comorbidities being neuropsychiatric factors (e.g., depression and anxiety; Uthman, Magidson, Safren, & Nachega, 2014), substance use (Gonzalez, Barinas, & O’Cleirigh, 2011), and neurocognitive impairment (Hinkin et al., 2002). Although prior research has identified the additive roles of socio-demographic and medical factors towards declines in everyday functioning and lower health-related quality of life experienced by people living with HIV (Fazeli, Woods, & Vance, 2019; Morgan et al., 2012), little research has focused on the role of more malleable behavioral factors. It is therefore critical to identify behavioral factors that may add to or even moderate the associations between HIV disease and poorer functional outcomes among older adults, with an eye towards factors that may be amenable to interventions aimed at improving health-related quality of life among older adults living with HIV.

To that end, research on personality traits(i.e., relatively stable patterns of thoughts, feelings, and behaviors; Roberts, 2009) has flourished in recent decades and has proven relevant in the context of health and everyday functioning. Indeed, Smith’s (2006) health behavior model, suggests that personality traits influence health behaviors, which in turn influence both subjective and objective health (Hampson et al., 2007; Smith, 2006). Empirical evidence has supported this theoretical model. Personality traits have been found to be reliably associated with health behaviors and outcomes across the lifespan, including in older adulthood (Chapman, Duberstein, & Lyness, 2007; Christensen et al., 2002; Goodwin & Friedman, 2006; Hampson, Edmonds, Goldberg, Dubanoski, & Hillier, 2013; Roberts, Kuncel, Shiner, Caspi, & Goldberg, 2007). Moreover personality traits are amenable to intervention and adaptive change, through psychotherapy (Roberts et al., 2017) and own volition (Hudson & Fraley, 2015), are malleable throughout the lifespan (Damian et al., 2019; Specht et al., 2014), and can be changed through intervention even in older adulthood (Jackson, Hill, Payne, Roberts, & Stine-Morrow, 2012).

The most commonly used personality trait framework is the Big Five (John, Naumann, & Soto, 2008), which includes the broad traits of openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism. Two of the Big Five personality traits are particularly relevant in the context of health: namely, conscientiousness (i.e., the tendency to be responsible, organized, and hard-working) and neuroticism (i.e., the tendency to be anxious, depressive, and moody). Numerous cross-sectional studies have found higher levels of conscientiousness to be associated with positive health behaviors (i.e., partaking in preventative habits and abstaining from risky habits; Chuah, Drasgow, & Roberts, 2006; Terracciano, Löckenhoff, Crum, Bienvenu, & Costa, 2008; for a review see: Bogg & Roberts, 2004), positive self-reported health status (Löckenhoff, Sutin, Ferrucci, & Costa, 2008), improved everyday functioning (Chapman, Duberstein, & Lyness, 2007), less chronic illnesses (Goodwin & Friedman, 2006; Sutin et al., 2010), and lower illness burden (Chapman et al., 2007). Conversely, higher levels of neuroticism have been associated with more negative health behaviors (e.g., tobacco smoking: Terracciano et al., 2008), negative self-report health status (Williams, O’Brien, & Colder, 2004; Duberstein et al., 2003; Löckenhoff et al., 2008), more chronic illnesses (Goodwin & Friedman, 2006; Sutin et al., 2010), and more physical impairment (Chapman et al., 2007). Of the other Big Five traits, some studies found that higher openness to experience (i.e., the tendency to engage with new ideas and experiences) was associated with greater physical functioning (Duberstein et al., 2003), but this effect did not replicate in other studies (Chapman et al., 2007; Löckenhoff et al., 2008). By way of comparison, agreeableness (i.e., the tendency to show empathy and cooperation) and extraversion (i.e., the tendency to be sociable and dominant) have shown little to no associations with health (Löckenhoff et al., 2008). Longitudinal research has largely supported cross-sectional findings, with conscientiousness and neuroticism having shown the most robust associations with health-related outcomes across the lifespan (Hampson, Goldberg, Vogt, & Dubanoski, 2007; Hampson et al., 2013; Israel et al., 2014).

Conscientiousness and neuroticism are also relevant in the context of HIV disease. For example, lower levels of conscientiousness and higher levels of neuroticism among seronegative people have been associated with increased engagement in HIV risk behaviors and incident HIV infection (Trobst et al., 2000; Trobst, Herbst, Masters, & Costa, 2002; Srivastava, Singh, & Chaudhury, 2016; Ruiz-Palomino, Ballester-Arnal, & Gil-Llario, 2017). Furthermore, among people living with HIV, low conscientiousness and high neuroticism have been associated with poorer engagement in health care (e.g., lower adherence; O’Cleirigh, Ironson, Weiss, & Costa, 2007), more rapid disease progression (Ironson, O’Cleirigh, Schneiderman, Weiss, & Costa, 2008; O’Cleirigh et al., 2007), and lower health-related quality of life (Penedo et al., 2003; Talukdar et al., 2013). These findings suggest that conscientiousness and neuroticism may play a role in the everyday functioning of people living with HIV.

Although ample research has shown that personality traits have unique contributions to health outcomes and everyday functioning (see Hampson & Friedman, 2008) and that they are relevant in the context of HIV (e.g., Penedo et al., 2003), we are not aware of any previous research that has investigated the associations between personality traits and everyday functioning among older adults living with HIV. This is an important oversight, because the main effects of conscientiousness and neuroticism on health-related outcomes are somewhat complicated in the context of aging, especially when older adults with chronic illness are considered. For example, prior research has suggested that the links between personality traits and perceived health may be stronger among older adults; specifically, when considering neuroticism, aging-related losses could increase the general tendency to worry about one’s health and evaluate it negatively, rendering a stronger association between neuroticism and self-reported health (Duberstein et al., 2003). Moreover, another study showed that associations between personality traits and self-reported physical health among older adults were stronger in healthier (as opposed to more medically-challenged) older adults, and age differences alone could not account for the observed discrepancies (Löckenhoff et al., 2008). Both these studies highlighted the need for further investigations into the role of personality traits on health-related outcomes among older adults with and without chronic illness, emphasizing the need for a multi-method approach, where both self- and informant-reports of health-related outcomes are employed in an effort to test whether effects are generalizable beyond perceived health.

Present Study

In this study, we evaluated the associations between personality traits (particularly, conscientiousness and neuroticism) and everyday functioning among older adults, and whether these associations differed depending on HIV serostatus. We adopted a multi-method approach where we used both self- and informant-reports of everyday functioning. The independent effects model presupposes no interaction between HIV serostatus and personality traits. To the extent that this is the only pattern to be found, we would expect personality traits to have the same associations with everyday functioning for all older adults, regardless of HIV serostatus.

Of course, the relation between personality traits, HIV serostatus, and everyday functioning may not be so straightforward. Indeed, prior theory makes two competing predictions. One possibility is the so-called “resource substitution hypothesis,” which states that resources will have more beneficial effects among people with fewer alternative resources (Mirowsky & Ross, 2003). Along these lines, traits like conscientiousness have been found to be more beneficial for future attainment for people who came from more disadvantaged homes (Damian, Su, Shanahan, Trautwein, & Roberts, 2015). . We might expect a similar moderating pattern with respect to personality resources (higher conscientiousness or lower neuroticism, respectively) becoming increasingly salient at lower levels of another critical resource (in this case, health). In other words, personality traits might matter more in predicting everyday functioning for people with (vs. without) HIV, because people living with HIV are at a disadvantage when it comes to everyday functioning (Blackstone et al., 2013).

Another possibility, the polar opposite of the above, is that the “strong situation” of living with HIV prevails over personality traits, and that personality traits might matter less in predicting everyday functioning in people with (vs. without) a strong situational challenge, in this case HIV (see Benjamin & Simpson, 2009). As described earlier, there is some empirical evidence to support the idea that “strong situations” (such as worse general health) can weaken associations between personality traits and health-related outcomes, but this effect was only tested for perceived health (Löckenhoff et al., 2008). Furthermore, a recent study that looked at associations between personality traits and health in adults with and without cancer (where cancer was considered the “strong situation”) found no personality by situation interaction effects, although they did find a positive main effect of conscientiousness and a negative main effect of neuroticism (Rochefort et al., 2019).Given the above competing predictions, the present study first tested unique associations between personality traits and everyday functioning among older adults, above and beyond HIV serostatus and relevant controls (i.e., independent effects model). We predicted that higher levels of conscientiousness (and lower levels of neuroticism, respectively) would be associated with higher odds of being clinically independent (vs. dependent) in terms of everyday functioning.

Second, we tested the role of the interaction between personality traits and HIV serostatus in everyday functioning (i.e., interaction effects model). Prior theory makes two competing predictions. According to the “resource substitution” hypothesis, older adults who are at a higher “disadvantage” (in this case living with vs. without HIV) would benefit more from higher levels of conscientiousness (or lower levels of neuroticism, respectively) when it comes to everyday functioning. According to the “strong situation” hypothesis, stronger situational challenges (in this case living with vs. without HIV) would render the personality traits of older adults less relevant in predicting everyday functioning.

Notably, throughout the analyses we adopted a multi-method approach, seeking to replicate our findings across self- and clinician-reported everyday functioning impairment, and we statistically controlled for theoretically relevant variables (i.e., gender, age, race/ethnicity, socio-economic status, substance use disorder, and medical comorbidities).

Methods

Participants

A total of 217 participants aged 50 and older were recruited, as part of a parent study on memory (see Woods et al., 2020), from greater San Diego county community-based organizations, infectious disease clinics, and the general community. The following participants were excluded from data collection and/or prior to any data analysis: participants with histories of severe psychiatric disorders (e.g., psychosis), neurological disease not due to HIV infection (e.g., Parkinson’s disease, seizure disorders, non-HIV-associated dementia), estimated verbal IQ scores < 70 (n=2), head injury with loss of consciousness > 30 min (n=2), color blindness, current substance dependence, positive breathalyzer for alcohol, or urine screens for illicit substances (n=2). Nine of the 211 otherwise eligible participants (4.3%) did not complete the primary measures for this study. Thus, the final sample for the present analyses consisted of 202 participants. We could not conduct an a priori power analysis, because we analyzed secondary data for this study. However a sensitivity power analysis, which is recommended in this case (Lakens, 2014), showed that we had 80% power, at alpha = .05, to detect effects as large as a correlation of r=.17, which is comparable with (or lower than) meta-analytic estimates of cross-sectional links between personality traits and health outcomes (see Bogg & Roberts, 2004). Mean age was 57.9 (SD = 6.7), 20.8% were women, 56.9% were White/European American, 69.8% were HIV seropositive (i.e., the sample included 141 persons with HIV disease and 61 HIV- comparison participants). Table 1 includes means and standard deviations for all variables in the study by HIV serostatus; for dichotomous variables, we also included frequency percentages. Table 2 includes total sample descriptive statistics and intercorrelations between all variables in the study.

Table 1.

Descriptive statistics split by group (people living with vs. without HIV) and independent samples t-test results.

	People living with HIV		People living without HIV		t-test results
Variable	M1 (SD)	%	M2 (SD)	%	da (p-value)
1. Femaleb	.15 (.36)	14.9	.34 (.48)	34.4	−.46 (.005)
2. Age	56.69 (5.85)	-	60.57 (7.38)	-	−.58 (.000)
3. POCc	.44 (.50)	44	.41 (.50)	41	.06 (.695)
4. SES	44.11 (11.38)	-	44.74 (10.73)	-	-.06 (.707)
5. Substance Use Disorderd	.71 (.46)	70.9	.56 (.50)	55.7	.32 (.045)
6. Medical Comorbidities	2.18 (1.59)	-	1.64 (1.37)	-	.37 (.015)
7. Conscientiousness	3.77 (.51)	-	3.96 (.55)	-	−.35 (.027)
8. Neuroticism	2.76 (.71)	-	2.46 (.80)	-	.40 (.012)
9. Extraversion	3.25 (.67)	-	3.57 (.79)	-	−.44 (.007)
10. Agreeableness	3.92 (.53)	-	3.97 (.52)	-	-.09 (.558)
11. Openness	3.70 (.59)	-	3.75 (.58)	-	-.08 (.585)
12. ADL Dependente (Self-report)	.37 (.48)	36.9	.16 (.37)	16.4	.47 (.001)
13. ADL Dependente (Clinician-rated)	.31 (.47)	31.2	.08 (.28)	8.3	.60 (.000)

Note:

^aCohen’s d computed using the following formula: [Cohen’s d = ((M₁ –M₂)/SD _pooled); SD pooled = √((SD²₁ + SD²₂)/2)].

^bFemale: 0=male, 1=female.

^cPOC: 0=White, 1=Person of Color.

^dSubstance Use Disorder: 0=no, 1=yes.

^eADL Dependent: 0=independent, 1=dependent. People living without HIV (HIV-) n = 60–61, people living with HIV (HIV+) n = 138–141. Bold effects were statistically significant at p< .05, two-tailed.

Table 2.

Intercorrelations among all variables in the study

Predictor	M (SD)	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. HIV Statusa	.70 (.46)	-	-	-	-	-	-	-	-	-	-	-	-	-	-
2. Femaleb	.21 (.41)	−.22	-	-	-	-	-	-	-	-	-	-	-	-	-
3. Age	57.86 (6.58)	−.27	.12	-	-	-	-	-	-	-	-	-	-	-	-
4. POCc	.43 (.50)	.03	.12	−.22	-	-	-	-	-	-	-	-	-	-	-
5. SES	44.3 (11.16)	−.03	−.19	.25	−.23	-	-	-	-	-	-	-	-	-	-
6. Substance Use Disorderd	.66 (.47)	.15	−.07	−.17	.03	−.07	-	-	-	-	-	-	-	-	-
7. Medical Comorbidities	2.02 (1.55)	.16	−.09	.20	.04	.01	.10	-	-	-	-	-	-	-	-
8. Conscientiousness	3.83 (.53)	−.16	−.02	.08	.08	.05	−.26	−.17	-	-	-	-	-	-	-
9. Neuroticism	2.67 (.75)	.19	.04	−.03	−.15	−.05	.21	.11	−.55	-	-	-	-	-	-
10. Extraversion	3.35 (.72)	−.20	.07	−.04	.04	.09	−.10	−.04	.36	−.45	-	-	-	-	-
11. Agreeableness	3.94 (.53)	−.04	.06	.01	.05	−.01	−.13	.02	.51	−.42	.39	-	-	-	-
12. Openness	3.71 (.59)	−.04	−.14	−.02	.01	.29	.03	−.06	.26	−.17	.31	.17	-	-	-
13. AIDS Statuse	.69 (.46)	N/A	.02	−.02	.13	−.17	.11	.24	−.10	.003	.01	−.06	−.08	-	-
14. ADL Dependentf (Self-report)	.31 (.46)	.20	−.05	−.08	.05	−.03	.18	.19	−.27	.35	−.26	−.08	−.14	.07	-
15. ADL Dependentf (Clinician-rated)	.24 (.43)	.25	.002	−.06	.03	−.12	.22	.33	−.29	.36	−.10	−.15	−.05	.02	.36

Note: n = 198–202;

^aHIV Status: 0=HIV-, 1=HIV+,

^bFemale: 0=male, 1=female,

^cPOC: 0=White, 1=Person of Color,

^dSubstance Use Disorder:0=no, 1=yes,

^eAIDS Status: 0=no AIDS, 1=AIDS ,

^fADL Dependent:0=independent, 1=dependent. AIDS Status was only relevant among the HIV+ sub-sample, hence correlations with AIDS status were only reported for the HIV+ sample. Bold coefficients were significant at p< .05, two-tailed.

Materials and Procedure

All participants provided written informed consent prior to completing an IRB-approved medical, psychiatric, and neuropsychological research evaluation, for which they received nominal financial compensation.

HIV Serostatus.

HIV serostatus was confirmed with Medmira rapid tests and was coded as (0= HIV seronegative, 1= HIV seropositive). As mentioned before, 69.8% of our participants were HIV seropositive.

Personality Traits.

We measured personality traits with the BFI-44 (John, Donahue, & Kentle, 1991). The BFI-44 has suitable validity and psychometric properties (Soto & John, 2009), as well as high convergent validity with the NEO-FFI (John & Srivastava, 1999). The BFI consists of 44 items designed to measure the Big Five dimensions of personality: Agreeableness (e.g., I see myself as someone who is considerate and kind to almost everyone), Neuroticism (e.g., I see myself as someone who is emotionally stable, not easily upset, reverse scored), Conscientiousness (e.g., I see myself as someone who is a reliable worker), Extraversion (e.g., I see myself as someone who is outgoing, sociable), and Openness (e.g., I see myself as someone who is curious about many different things). Participants were asked to rate how well each of the characteristics applied to them on a 5-point scale ranging from 1 (Disagree strongly) to 5 (Agree strongly). We averaged the responses across 8–10 items corresponding to each Big Five personality trait, to form the five scale scores: extraversion (α = .85), agreeableness (α = .79), conscientiousness (α = .77), neuroticism (α = .86), and openness (α = .83).

Covariates.

The analyses included six relevant covariates, which were determined a priori from the literature on everyday functioning in HIV disease (Blackstone et al., 2013; Morgan et al., 2012). Namely, we included gender (0 = male, 1=female), age (in years), race/ethnicity (0 = White/European American, 1 = Person of Color), socio-economic status (SES), substance use disorder, and medical comorbidities as covariates.

Race/ethnicity originally included several categories, as follows: White/European American (n=115), Black/African American (n=34), Asian/Asian American (n=1), Native American (n=11), Hispanic (n=32), and Multiracial (n=9). However, given the small samples across the various groups (other than White/European American), we recoded this variable into White/European American and Person of Color.

Socio-economic status was measured using the two-factor Hollingshead index (Adams, Weakliem, & August, 2011), which combines an individual’s level of education and highest occupational attainment to derive a summary SES score (range was 8 to 66, with higher scores reflecting higher SES).

Substance use disorder was determined using the Composite International Diagnostic Interview (CIDI version 2.128; World Health Organization, 1998). The CIDI is a semi-structured interview that was administered by certified research assistants and yields lifetime diagnoses of substance abuse and dependence according to the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 1994). For this study, participants who met criteria for either abuse or dependence on alcohol, marijuana, stimulant, opioid, inhalant, sedative, or any other illicit drugs were classified as having a lifetime substance use disorder (0 = no lifetime substance use disorder, 1= lifetime substance use disorder). Of all participants, 66.3% had a lifetime diagnosis of any substance use disorder.

To assess medical comorbidities, participants underwent a brief medical evaluation led by a research nurse, which included a review of systems, medications, history, urine toxicology, and a blood draw. Common medical comorbidities (e.g., cardiovascular disease, hepatitis C, diabetes, cancer) were summed (sample range = 1–7 medical conditions).

In analyses that only included people living with HIV (see “Data Analysis” section below), in addition to the above six covariates, we also included AIDS status (0=non-AIDS, 1 = AIDS), which has been previously shown to adversely impact everyday functioning in people with HIV disease (Van Gorp et al., 2007). AIDS status was derived from the semi-structured interview with the research nurse. Of the participants with HIV disease, 68.8% had AIDS.

Assessment of Everyday Functioning (Self-report).

Self-reported declines in activities of daily living (ADL) were assessed with the widely used Heaton revision of the Lawton and Brody (Lawton & Brody, 1969) ADL Scale (see Heaton et al., 2004 and Woods et al., 2004). Participants rated their “best” and “current” level of independence in performing 16 total instrumental (e.g., work, finances, shopping) and basic (e.g., bathing, grooming) ADLs. For each of the 16 ADL items, current ratings that represented a decline from participants’ best level of functioning were coded as an ADL decline and summed (α = .81). For example, an ADL decline was coded in the cooking domain if an individual was able to “…plan, prepare, and serve many of [their] own meals” at their best, but currently endorsed that they “…prepare meals if someone else provides…the right ingredients”. Due to the highly skewed distribution of this variable (median = 0, interquartile range was 0 to 3) and considerations of clinical significance in the context of current diagnostic formulations for functional declines in HIV (Antinori et al., 2007; American Psychiatric Association, 2013), participants with 2 or more ADL domain declines (n = 62; 30.7%) were classified as being “ADL Dependent” for analyses.

Assessment of Everyday Functioning (Clinician-rated).

Clinician-rated ADL functioning was operationalized using the Karnofsky Performance Status Scale (KPSS; Karnofsky & Burchenal, 1949), which shows high inter-rater reliability (Chow et al., 2016) and is widely used in studies of HIV disease (Morgan et al., 2012; O’Dell, Lubeck, O’Driscoll, & Matsuno, 1995). A trained and certified research nurse blinded to the results of the psychological evaluation assigned participants a rating from zero (mortality) to 100 (normal, no evidence of disease). As was observed with the self-report ADL variable detailed above, the distribution of the KPSS was highly skewed (median = 100, interquartile range was 90 to 100). Therefore, consistent with prior research in HIV (Doyle et al., 2013), participants with a KPSS of < 90 were classified as having “ADL Difficulties” for analyses (n = 48; 24.2%). The overall agreement between patient classification based on self- and clinician-reports of everyday functioning was 74% (κ = .36, 95% CI [.22; .50]).

Data Analysis

All continuous variables were standardized prior to being included in models and prior to computing the interaction terms. We used listwise deletion to deal with missing data, though notably, only three participants had missing data. Critical alpha level was set at .05 and the statistical package used was IBM SPSS Statistics version 25. To test our hypotheses, we conducted a series of logistic regressions (separately for each outcome: self- vs. clinician-reported everyday functioning; and separately for each personality predictor of interest: conscientiousness vs. neuroticism; see Tables 3–6). Across the two outcomes, Models A were our baseline models, including HIV serostatus and relevant control variables (i.e., gender, age, race/ethnicity, socio-economic status, substance use disorder, and medical comorbidities). Models B included conscientiousness and neuroticism, respectively, and Models C included interaction factors between HIV serostatus and conscientiousness and neuroticism, respectively. To further test the robustness of our findings, Models D included all remaining Big Five personality traits.

Table 3.

Logistic regressions testing the baseline (Model A), independent effects model for Conscientiousness (Model B), and interaction model for the moderating role of Conscientiousness on HIV Status (Model C) in predicting daily activities dependence (ADL Dependence) (Self-reported).

Predictor	Model A ORa (p-value)	Model B ORa (p-value)	Model C ORa (p-value)	Model D ORa (p-value)
HIV Status b	2.30 (.045)	2.20 (.066)	2.33 (.061)	1.78 (.211)
Femalec	1.05 (.904)	.96 (.935)	.95 (.912)	.83 (.705)
Age	.88 (.504)	.91 (.618)	.91 (.616)	.86 (.456)
POCd	1.10 (.787)	1.22 (.568)	1.23 (.549)	1.50 (.280)
SES	1.01 (.944)	1.01 (.940)	1.02 (.903)	1.18 (.401)
Substance Use Disordere	2.07 (.052)	1.64 (.201)	1.62 (.216)	1.52 (.315)
Medical Comorbidities	1.47 (.024)	1.39 (.058)	1.37 (.072)	1.38 (.088)
Conscientiousness	-	.59 (.004)	.49 (.082)	.78 (.284)
HIV Status * Conscientiousness	-	-	1.28 (.588)	-
Neuroticism	-	-	-	1.98 (.003)
Extraversion	-	-	-	.69 (.085)
Agreeableness	-	-	-	1.46 (.078)
Openness	-	-	-	.81 (.273)
R2 (PSEUDO)f	.13	.19	.19	.29

Note:

^aOR = Odds Ratio (Exp(B)).

^bHIV Status: 0=HIV-, 1=HIV+.

^cFemale: 0=male, 1=female.

^dPOC: 0=White, 1=Person of Color.

^eSubstance Use Disorder: 0=no, 1=yes.

^fPSEUDO-R²= Nagelkerke R². n = 202. All continuous predictors were standardized prior to the analysis. Bold coefficients were statistically significant at p< .05, two-tailed.

Table 6.

Logistic regressions testing the baseline (Model A), independent effects model for Neuroticism (Model B), and interaction model for the moderating role of Neuroticism on HIV Status (Model C) in predicting daily activities dependence (ADL Dependence) (Clinician-rated).

Predictor	Model A ORa (p-value)	Model B ORa (p-value)	Model C ORa (p-value)	Model D ORa (p-value)
HIV Status b	3.83 (.013)	4.21 (.014)	4.65 (.036)	5.11 (.007)
Femalec	1.52 (.404)	1.36 (.562)	1.34 (.587)	1.27 (.674)
Age	.85 (.490)	.86 (.542)	.87 (.564)	.91 (.686)
POCd	.84 (.645)	.97 (.936)	.97 (.937)	.96 (.921)
SES	.81 (.280)	.81 (.330)	.81 (.335)	.71 (.149)
Substance Use Disordere	3.25 (.016)	2.50 (.077)	2.51 (.076)	2.30 (.121)
Medical Comorbidities	2.23 (.000)	2.32 (.000)	2.30 (.000)	2.39 (.000)
Neuroticism	-	2.40 (.000)	2.72 (.080)	2.38 (.001)
HIV Status * Neuroticism	-	-	.86 (.809)	-
Conscientiousness	-	-	-	.76 (.296)
Extraversion	-	-	-	1.52 (.107)
Agreeableness	-	-	-	.82 (.431)
Openness	-	-	-	1.17 (.493)
R2 (PSEUDO)f	.28	.39	.39	.41

Note:

^aOR = Odds Ratio (Exp(B)).

^bHIV Status: 0=HIV-, 1=HIV+.

^cFemale: 0=male, 1=female.

^dPOC: 0=White, 1=Person of Color.

^eSubstance Use Disorder: 0=no, 1=yes.

^fPSEUDO-R²= Nagelkerke R². n = 198. All continuous predictors were standardized prior to the analysis. Bold coefficients were statistically significant at p< .05, two-tailed.

Exploratory Analyses.

In addition to the above analyses, we conducted three sets of exploratory analyses. First, because we wanted to test the magnitude of the association between personality and everyday functioning in HIV disease, we conducted additional exploratory analyses on the sample of people living with HIV (see Table 7). Specifically, we ran three kinds of models for each of the two outcomes (self- vs. clinician-reported dependency). Models A’ tested the unique effect of conscientiousness over and above relevant covariates; Models B’ tested the unique effect of neuroticism over and above relevant covariates; Models C’ included all other personality traits as covariates in the analyses. The covariates were the same as in the previous models presented in Tables 3–6, the only difference being that we replaced HIV seropositive status, which was no longer relevant given that we were only focusing on HIV seropositive people in these analyses, with AIDS status, which reliably impacts everyday functioning in HIV (Van Gorp et al., 2007).

Table 7.

Exploratory Analyses (people living with HIV). Logistic regressions testing the incremental role of personality traits in predicting ADL Dependence, both self- and clinician-reported).

Predictor	ADL Dependence (Self-reported) ORa (p-value)			ADL Dependence (Clinician-reported) ORa (p-value)
	Model A’	Model B’	Model C’	Model A’	Model B’	Model C’
Femaleb	.91 (.871)	.89 (.840)	.86 (.800)	.97 (.964)	.98 (.979)	.96 (.949)
Age	.84 (.460)	.79 (.339)	.75 (.267)	.86 (.580)	.81 (.450)	.81 (.494)
POCc	.91 (.815)	1.02 (.960)	.98 (.962)	.93 (.873)	1.04 (.940)	1.08 (.870)
SES	1.03 (.886)	1.08 (.693)	1.29 (.260)	.69 (.117)	.71 (.150)	.58 (.046)
Substance Use Disorderd	2.16 (.089)	2.01 (.136)	2.20 (.110)	3.76 (.020)	3.83 (.025)	3.67 (.036)
Medical Comorbidities	1.42 (.076)	1.50 (.050)	1.42 (.112)	2.21 (.001)	2.34 (.001)	2.65 (.001)
AIDS Statuse	1.05 (.919)	1.10 (.837)	1.32 (.556)	.51 (.176)	.50 (.172)	.37 (.076)
Conscientiousness	.65 (.032)	-	.89 (.652)	.59 (.020)	-	.84 (.548)
Neuroticism	-	2.32 (.000)	2.13 (.005)	-	2.34 (.000)	2.43 (.003)
Extraversion	-	-	.60 (.046)	-	-	1.42 (.225)
Agreeableness	-	-	1.46 (.126)	-	-	.77 (.338)
Openness	-	-	.71 (.134)	-	-	1.39 (.223)
R2 (PSEUDO)f	.14	.23	.30	.27	.34	.37

Note:

^aOR = Odds Ratio (Exp(B)).

^bFemale: 0=male, 1=female.

^cPOC: 0=White, 1=Person of Color.

^dSubstance Use Disorder: 0=no, 1=yes.

^eAIDS Status: 0=no AIDS, 1=AIDS.

^fPSEUDO-R²= Nagelkerke R². ADL Dependence (Self-reported) n = 141, ADL Dependence (Clinician-reported) n = 138. All continuous predictors were standardized prior to the analysis. Bold coefficients were statistically significant at p< .05, two-tailed.

Second, our a priori-selected list of covariates did not include mood disorder diagnosis because previous research has shown neuroticism to be the personality core of mood disorders, with high correlations between neuroticism and mood disorder diagnosis (Kotov, Gamez, Schmidt, & Watson, 2010). Thus, we were concerned that including mood disorder diagnosis among the covariates would have posed a threat of multicollinearity in the regression model. Nevertheless, to test the robustness of our findings, we re-ran all our analyses (from Tables 2–6) with one change: we included mood disorder (anxiety or depression) diagnosis as a covariate. The results of these analyses can be found in Tables S1–S5 of our Online Supplemental Material available for masked peer-review at this link: https://osf.io/tyu7d/?view_only=11ec23daa2d4490c97f579bed42320e0 (this document will be made publicly available once the review process is complete).

Third, our main analyses (Tables 3–6) were only concerned with conscientiousness and neuroticism, because, as explained above, these are the personality traits that have been most reliably associated with health outcomes. Hence, we only had a priori predictions for these two traits. However, for the sake of completeness, we conducted exploratory analyses on the remaining three traits; the results can be found in Tables S6 to S11 of our Online Supplemental Material available at this link: https://osf.io/tyu7d/?view_only=11ec23daa2d4490c97f579bed42320e0.

Effect Sizes and Clinical Relevance.

The clinical relevance of any effect size depends on the constructs being studied (Cumming, 2014), which, in this case, are everyday functioning outcomes (dependent vs. independent) among persons with and without HIV disease. Therefore, even small effects possess direct relevance to daily life and clinical care. Notably, we had 80% power to detect a minimum effect of r = .17. One way to better understand the clinical relevance of such an effect, in the context of our study, is to use the Binomial Effect Size Display (BESD; Rosenthal & Rubin, 1982). Given our outcomes of interest (dependence vs. independence in everyday functioning) and personality trait predictors, such as conscientiousness, for which we predicted more independence at higher levels of the trait, using the BESD, a correlation of .17 would have been the equivalent of saying that people who were above (vs. below) the median in conscientiousness had 59:41 odds of being independent (vs. dependent). Given the importance of the outcomes studied, this effect size magnitude would have arguably been clinically relevant; nevertheless, our observed effects were much larger than that. To better highlight the clinical relevance of our findings, throughout the Results section, we presented effect size interpretations using odds ratios.

Results

Table 1 includes descriptive statistics split by HIV serostatus (seropositive vs. seronegative), results from independent samples t-tests, and effect sizes transformed in Cohen’s d of the differences between the two groups. Notably, the HIV seropositive (vs. the HIV seronegative) sample had more men (85% vs. 66%, respectively) and was younger (57 vs. 61 years old, respectively), lower in conscientiousness and extraversion, and higher in neuroticism. The racial distribution and socio-economic status were similar across the two groups. In line with prior evidence (Rodriguez-Penney et al., 2013; Lyketsos, Hanson, Fishman, McHugh, & Treisman, 1994), people living with HIV were more likely to have had a substance use disorder and had more medical comorbidities. To address these differences, we statistically controlled for demographics, substance use disorder, and medical comorbidities in all analyses. Regarding the main outcomes of interest, as predicted, people living with HIV were more likely to be dependent in the context of everyday functioning, and the effect replicated across self- and clinician-reports (d = .47 and .60, respectively).

Table 2 includes inter-correlations between all variables in the study (all correlations that include dichotomous variables are point biserial correlations). The largest and most robust associations between personality traits and everyday functioning were those found for conscientiousness and neuroticism. As predicted, higher levels of conscientiousness were associated with less dependence, whereas higher levels of neuroticism were associated with more dependence. These associations replicated across self- and clinician-reports of everyday functioning (for conscientiousness, r = −.27 and −.29, respectively; for neuroticism, r = .35 and .36). Thus, we proceeded with the planned logistic regressions.

Tables 3 and 4 show the logistic regression Models A through D for self- and clinician-reports of everyday functioning respectively, when conscientiousness and its interaction with HIV serostatus were the main predictors of interest. Models A (baseline; see Tables 3 and 4) showed that HIV serostatus, above and beyond relevant covariates, was statistically significantly associated with a 130% increase in the odds of dependency, when everyday functioning was self-reported, and a 283% increase in the odds of dependency, when everyday functioning was clinician-reported. Models B (independent effects model; see Tables 3 and 4) showed that conscientiousness was a statistically significant predictor of everyday functioning above and beyond HIV serostatus and relevant controls. Specifically, a one standard deviation increase in conscientiousness was associated with a 41% decrease in the odds of dependency when everyday functioning was self-reported, and a 45% decrease in the odds of dependency when everyday functioning was clinician-reported. Thus, the independent effects model, whereby conscientiousness shows incremental validity in predicting everyday functioning over and above HIV serostatus and relevant controls, was supported. Models C (interaction effect model; see Tables 3 and 4) showed that conscientiousness was not a statistically significant moderator of HIV serostatus when predicting everyday functioning. Thus, the interaction effects model was not supported, that is, neither the “resource substitution”, nor the “strong situation” hypothesis received support. Models D tested the robustness of the independent effect of conscientiousness when including the remaining personality traits as control variables. The independent effect of conscientiousness ceased to be statistically significant; this result, however, should be taken with caution due to issues of multicollinearity; because the Big Five personality traits are inter-correlated, as can also be seen in Table 2 (correlations ranging −.55 to .51), some might suggest not to include all personality traits together as covariates in the same regression models, but to evaluate them separately; nevertheless, others might argue that all traits should be included in the analyses as a robustness check. For the sake of transparency and to accommodate both points of view, we presented Models A through D in progression.

Table 4.

Logistic regressions testing the baseline (Model A), independent effects model for Conscientiousness (Model B), and interaction model for the moderating role of Conscientiousness on HIV Status (Model C) in predicting daily activities dependence (ADL Dependence) (Clinician-rated).

Predictor	Model A ORa (p-value)	Model B ORa (p-value)	Model C ORa (p-value)	Model D ORa (p-value)
HIV Status b	3.83 (.013)	4.19 (.011)	6.57 (.025)	5.11 (.007)
Femalec	1.52 (.404)	1.43 (.489)	1.41 (.514)	1.27 (.674)
Age	.85 (.490)	.88 (.596)	.89 (.613)	.91 (.686)
POCd	.84 (.645)	.89 (.768)	.90 (.800)	.96 (.921)
SES	.81 (.280)	.78 (.239)	.80 (.273)	.71 (.149)
Substance Use Disordere	3.25 (.016)	2.55 (.064)	2.41 (.084)	2.30 (.121)
Medical Comorbidities	2.23 (.000)	2.21 (.000)	2.12 (.001)	2.39 (.000)
Conscientiousness	-	.55 (.005)	.29 (.098)	.76 (.296)
HIV Status * Conscientiousness	-	-	2.03 (.361)	-
Neuroticism	-	-	-	2.38 (.001)
Extraversion	-	-	-	1.52 (.107)
Agreeableness	-	-	-	.82 (.431)
Openness	-	-	-	1.17 (.493)
R2 (PSEUDO)f	.28	.33	.34	.41

Note:

^aOR = Odds Ratio (Exp(B)).

^bHIV Status: 0=HIV-, 1=HIV+.

^cFemale: 0=male, 1=female.

^dPOC: 0=White, 1=Person of Color.

^eSubstance Use Disorder: 0=no, 1=yes.

^fPSEUDO-R²= Nagelkerke R². n = 198. All continuous predictors were standardized prior to the analysis. Bold coefficients were statistically significant at p< .05, two-tailed.

Tables 5 and 6 show the logistic regression Models A through D for self- and clinician-reports of everyday functioning respectively, when neuroticism and its interaction with HIV serostatus were the main predictors of interest. Models A and D from Tables 5 and 6 are identical to Models A and D from Tables 3 and 4, respectively; we only included them in Tables 5 and 6 to make direct comparisons across Models A through D easier within each table. Models B (independent effects model; see Tables 5 and 6) showed that neuroticism was a statistically significant predictor of everyday functioning above and beyond HIV serostatus and relevant controls. Specifically, a one standard deviation increase in neuroticism was associated with a 121% increase in the odds of dependency, when everyday functioning was self-reported, and a 140% increase in the odds of dependency, when everyday functioning was clinician-reported. Thus, the independent effects model, whereby neuroticism shows incremental validity in predicting everyday functioning over and above HIV serostatus and relevant controls, was supported. Models C (interaction effect model; see Tables 5 and 6) showed that neuroticism was not a statistically significant moderator of HIV serostatus when predicting everyday functioning. Thus, the interaction effects model was not supported, that is, neither the “resource substitution”, nor the “strong situation” hypothesis received support. Models D tested the robustness of the independent effect of neuroticism when including the remaining personality traits as control variables. The independent effects of neuroticism remained statistically significant across both self- and clinician-reported everyday functioning.

Table 5.

Logistic regressions testing the baseline (Model A), independent effects model for Neuroticism (Model B), and interaction model for the moderating role of Neuroticism on HIV Status (Model C) in predicting daily activities dependence (ADL Dependence) (Self-reported).

Predictor	Model A ORa (p-value)	Model B ORa (p-value)	Model C ORa (p-value)	Model D ORa (p-value)
HIV Status b	2.30 (.045)	2.02 (.114)	1.89 (.153)	1.78 (.211)
Femalec	1.05 (.904)	.89 (.794)	.91 (.842)	.83 (.705)
Age	.88 (.504)	.88 (.517)	.87 (.492)	.86 (.456)
POCd	1.10 (.787)	1.40 (.352)	1.39 (.364)	1.50 (.280)
SES	1.01 (.944)	1.07 (.726)	1.06 (.738)	1.18 (.401)
Substance Use Disordere	2.07 (.052)	1.56 (.261)	1.56 (.261)	1.52 (.315)
Medical Comorbidities	1.47 (.024)	1.44 (.043)	1.47 (.036)	1.38 (.088)
Neuroticism	-	2.21 (.000)	1.80 (.098)	1.98 (.003)
HIV Status * Neuroticism	-	-	1.32 (.506)	-
Conscientiousness	-	-	-	.78 (.284)
Extraversion	-	-	-	.69 (.085)
Agreeableness	-	-	-	1.46 (.078)
Openness	-	-	-	.81 (.273)
R2 (PSEUDO)f	.13	.25	.25	.29

Note:

^aOR = Odds Ratio (Exp(B)).

^bHIV Status: 0=HIV-, 1=HIV+.

^cFemale: 0=male, 1=female.

^dPOC: 0=White, 1=Person of Color.

^eSubstance Use Disorder: 0=no, 1=yes.

^fPSEUDO-R²= Nagelkerke R².n = 202. All continuous predictors were standardized prior to the analysis.Bold coefficients were statistically significant at p< .05, two-tailed.

Exploratory Analyses Results

Table 7 shows results from exploratory analyses conducted on the sample of people living with HIV. In this sub-sample, again, when considered separately from other personality traits, both conscientiousness and neuroticism were statistically significantly associated with both self- and clinician-reported dependency, above and beyond relevant covariates. Nevertheless, when including all traits in the prediction model, neuroticism had the more robust associations across measures of everyday functioning, showing that, among people living with HIV, a standard deviation increase in neuroticism was statistically significantly associated with a 113% increase in the odds of dependency when everyday functioning was self-reported, and a 143% increase in the odds of dependency when everyday functioning was clinician-reported.

Tables S1 (see Online Supplemental Material available at this link: https://osf.io/tyu7d/?view_only=11ec23daa2d4490c97f579bed42320e0) includes mood disorder diagnosis (anxiety or depression) in the correlation matrix. Not surprisingly, HIV seropositive people were more likely to have been diagnosed with a mood disorder (r = .32, p< .001), and having a mood disorder was positively associated with neuroticism (r = .31, p< .001) and dependency (both self- and clinician-reported; r = .27 and .32, respectively, ps < .001), and negatively associated with conscientiousness (r = −.20, p = .004). Tables S2 to S5 show the results from the logistic regressions of interest when including mood disorder diagnosis as a covariate in all analyses (see Online Supplemental Material available at this link: https://osf.io/tyu7d/?view_only=11ec23daa2d4490c97f579bed42320e0). Interestingly, although the associations between HIV serostatus and the two dependency outcomes were somewhat lowered by the inclusion of this covariate (which makes sense given the positive correlation between HIV serostatus and mood disorder diagnosis), the associations between conscientiousness and neuroticism and the two dependency outcomes remained largely unaffected. Thus, conscientiousness was associated with a decrease in the odds of dependency, and neuroticism with an increase, and the findings held regardless of whether people had a mood disorder diagnosis or not, and across self- and clinician-reported dependency outcomes. Moreover, the associations between neuroticism and self- and clinician-reported dependency outcomes persisted when including all other personality traits and mood disorder diagnosis as covariates.

Although, based on prior literature, we only made predictions regarding conscientiousness and neuroticism, we conducted exploratory analyses for the other three traits (i.e., extraversion, agreeableness, and openness). The results can be found in Tables S6 to S11 in the Online Supplemental Material available at this link: https://osf.io/tyu7d/?view_only=11ec23daa2d4490c97f579bed42320e0. Extraversion and openness showed statistically significant associations with self-reported, but not clinician-reported dependency. Specifically, a one standard deviation increase in extraversion (/openness) was associated with a 46% (/30%) decrease in the odds of self-reported dependency (see Tables S6 and S7 for extraversion and Tables S10 and S11 for openness). Conversely, agreeableness showed a statistically significant association with clinician-reported, but not self-reported dependency. Specifically, a one standard deviation increase in agreeableness was associated with a 37% decrease in the odds of clinician-reported dependency (see Tables S8 and S9). Like conscientiousness and neuroticism, none of the remaining three traits showed statistically significant interactions with HIV serostatus. Notably, the results across these three traits should be considered with caution because they were part of exploratory analyses, as we could not make clear a priori predictions based on previous literature. Moreover, these findings seemed less robust since they did not replicate across self- and clinician-reported outcomes, unlike the findings reported for conscientiousness and neuroticism.

Across all analyses (both confirmatory and exploratory), the results supported the independent effects model of personality and health, but not the interactive effects model, and the associations between neuroticism (vs. conscientiousness) and everyday functioning appeared to be more robust.

Discussion

In this study we tested two alternative models of the role of personality in the everyday functioning of older adults, in the context of HIV disease. First, we tested unique associations of personality traits with everyday functioning, above and beyond HIV serostatus and relevant controls (independent effects model), and, based on previous literature on personality and health, we focused on conscientiousness and neuroticism. As per our predictions, we found that higher levels of conscientiousness (and lower levels of neuroticism, respectively) were associated with higher odds of being clinically independent (vs. dependent) in everyday functioning among older adults. The associations between neuroticism and everyday functioning persisted even when including all other personality traits as covariates, which is a relatively conservative robustness check. Exploratory analyses further supported the above findings, suggesting that the findings replicated among older adults living with HIV and persisted when mood disorder diagnosis was included as a covariate. Moreover, the associations between the other three personality traits (extraversion, agreeableness, and openness) and everyday functioning were less robust and consistent, which aligns with previous literature (Roberts et al., 2007).

But why would conscientiousness and neuroticism be associated with everyday functioning among older adults, in the context of HIV disease? According to extensive past theory and empirical evidence, the lifespan link observed between personality traits and health-related outcomes can be explained via health-related behaviors and other socio-cognitive mechanisms that indirectly affect health, as exemplified below (Hampson et al., 2007; Lodi-Smith et al., 2010; Mroczek et al., 2009; Smith, 2006). Personality traits are pervasive patterns of thoughts, feelings, and behaviors, which means they may impact all areas of life and functioning, across time and situations (Roberts, 2009). Someone who is higher in conscientiousness tends to act responsibly, be on time, organized, make clear plans, get things done, be thorough, and reliable. Thus, in the context of health, they will likely make more responsible, healthier decisions; for example, they will tend to adhere to medication better (Hill & Roberts, 2011), and they will eat better and exercise more (Bogg & Roberts, 2004). In the context of everyday functioning, they might compensate for memory problems by using a calendar or they will plan activities well ahead of time and in detail, considering any physical limitations. Other areas of life, which can have an indirect effect on health, can also be impacted. For example, people who are higher in conscientiousness tend to have better relationships (Claxton, O’Rourke, Smith, & DeLongis, 2012), which in turn may provide them with increased social support, which may help with their everyday functioning. People who are higher in conscientiousness are also more valued in the work context (Barrick, Mount, & Judge, 2001) and tend to make more money (Judge, Livingston, & Hurst, 2012), which can also provide them with increased social support and resources that can help with their everyday functioning. They also tend to have higher well-being because they experience less stress due to procrastination or other problems that arise from not acting responsibly (Steel, 2007).

In contrast, people who are higher in neuroticism, tend to be moody, irritable, angry, and spend a lot of time worrying about possible negative outcomes. Thus, their relationships tend to suffer (Watson, Hubbard, & Wiese, 2000), they tend to have poorer work performance (Barrick et al., 2001), lower well-being (Headey, Muffels, & Wagner, 2010), and they will often make poorer health choices, such as emotional eating, smoking, or drinking (Walker, Christopher, Wieth, & Buchanan, J., 2015; Terracciano & Costa, 2004). All these outcomes can cumulate and impact everyday functioning by robbing highly neurotic people of important social resources and essential coping mechanisms.

In sum, due to their pervasiveness across areas of life and because personality traits shape people’s identities (i.e., how people see themselves, which dictates their next patterns of behavior) and reputations (i.e., how other people see them, which dictates patterns of social interaction), personality traits tend to have large effects on important life outcomes, including health outcomes and everyday functioning (Roberts et al., 2007).

Second, we tested the interacting role of personality traits and HIV serostatus in everyday functioning (interaction effects model). We found no evidence of interactions and thus no support for the “resource substitution” hypothesis or the “strong situation” hypothesis in this sample, suggesting that personality was equally important for everyday functioning in older adults with and without HIV disease. There are several potential explanations for these null findings. First, it is possible that our sample was too high-functioning overall (as Table 1 suggests), which may have posed a range-restriction problem and minimized any interactive effects indicative of resource substitution. Second, it is possible that other factors in the lives of older adults living with HIV had already fulfilled the roles of “resource substitutes.” Indeed, previous studies suggested that factors such as emotional support might play a role in the positive aging of people living with HIV (Moore et al., 2018). Third, according to a sensitivity power analysis we conducted (as reported in the “Methods” section), we had 80% power (at alpha = .05) to detect effects as larger or larger than a correlation of .17 (d = .34); thus, if the interaction effect in the population were smaller than that, we were underpowered to detect it. Thus, although the present findings suggest that large interaction effects are unlikely, they cannot rule out the possibility that smaller interaction effects might occur between personality traits and HIV status when predicting functioning. However, the smaller the effects, the less likely they are to be of clinical significance. The present findings reinforce the conclusion of prior studies (Damian et al., 2015; Shanahan, Hill, Roberts, Eccles, & Friedman, 2014), namely that the independent effects model might be most appropriate, unless very small interaction effects are of interest. The present findings are also in line with prior research that tested the “strong situation” hypothesis and found no interaction effects between personality and cancer status in predicting health outcomes (Rochefort et al., 2019).

This study has several advantages: (a) we targeted a vulnerable and often overlooked population in the aging literature, that is, older adults living with HIV; this is important because the prevalence of older people living with HIV has risen considerably (Centers for Disease Control and Prevention, 2018) and recent calls have been made for the integration of gerontology, geriatrics, and HIV medicine, to improve functioning and quality of life among older adults living with HIV (Singh et al., 2017); (b) we used both self- and clinician-reports for the outcome of interest; this multi-method approach served as a within-study replication, reduced method variance, and showed the cross-method robustness of the findings, thus, addressing limitations with past personality-health research in older adults (e.g., Duberstein et al., 2003; Löckenhoff et al., 2008); (c) we used a multitude of relevant control variables, and (d) this is the first study to test the associations between personality traits and everyday functioning, and whether these associations differed depending on HIV serostatus.

Despite its many advantages, this study also had several limitations. First, we used a cross-sectional sample, which precludes inferences regarding within-person developmental processes and cannot establish causality. Second, we had a small sample, especially for the HIV seronegative group. Third, it is possible our sample suffered from survival bias, which is a common problem in aging research and especially in aging samples who have chronic diseases; this may have contributed to the range-restriction problem we mentioned above. Fourth, although the present study did not enroll anyone who was participating in an intervention study at the same center, we did not formally exclude people who might have been participating in intervention studies at external agencies. Thus, although unlikely, it is possible that observed correlational effects between personality and everyday functioning might have been partly explained by participation in intervention studies at external agencies; to be clear, we have no evidence of this and no possibility of testing this hypothesis, but even if this were the case, and people who were higher in conscientiousness, for instance, were more likely to enroll in intervention programs, and that, in turn increased their independence, that would have been in line with our proposed indirect link between personality and health outcomes via health-oriented behaviors (Hampson et al., 2007; Smith, 2006); future studies should test these proposed mechanisms empirically. Fifth, the HIV seropositive and HIV seronegative groups were mismatched on some demographics, although we attempted to address this issue by including relevant covariates in all the analyses. Sixth, we did not include any performance-based measures of everyday functioning capacity. Finally, it is possible that everyday functioning impairments shaped personality, resulting in the observed cross-sectional associations. Given these limitations, future studies should focus on cross-sectional replications of the present findings and longitudinal extensions, both within this population and within other chronic illness communities.

Nevertheless, the present findings are important because they suggest that personality traits, and particularly neuroticism and conscientiousness, are meaningful sources of variance in everyday functioning, they may have equally strong associations with everyday functioning across HIV seropositive and seronegative older adults, and they show unique predictive validity across self- and clinician-rated outcomes, and above and beyond other physical and mental health factors that are known to impact everyday functioning. These findings are in line with past theory and empirical evidence on the role of personality traits in health-related outcomes (e.g., Hampson et al., 2007; Lodi-Smith et al., 2010; Mroczek et al., 2009; Smith, 2006), and together, they suggest two different possible pathways for intervention: (a) intervening to change relevant personality trait levels and (b) personalizing medical interventions to cater to people’s diverse personality profiles.

Regarding personality interventions, a key finding of the past decades of personality research is that personality traits are not static, but dynamic and malleable constructs. Indeed, personality traits have been found to change naturally with age, following normative transitions and social investment (Damian et al., 2019; Roberts, Walton, & Viechtbauer, 2006; Lodi-Smith & Roberts, 2007), but also following life experiences (Bleidorn, Hopwood, & Lucas, 2018), volitional goals (Hudson & Fraley, 2015), or psychological therapy intervention (for a meta-analysis, see Roberts et al., 2017). In fact, the latter meta-analysis of clinical interventions has found that all types of psychological therapy are highly effective in changing personality after as little as eight weeks, with effects lasting well beyond the end of the intervention. Therapy was most effective at reducing neuroticism, but changes in the other Big Five traits, including increases in conscientiousness, were also observed. Furthermore, past research has shown that interventions can change personality traits among older adults as well (Jackson et al., 2012). More importantly, personality change in adulthood has been associated with health outcomes; specifically, decreases in neuroticism and increases in conscientiousness have been linked to increases in self-reported health (Letzring, Edmonds, & Hampson, 2014). Thus, the findings of the present study, combined with the knowledge that personality traits are dynamic and amenable to intervention, suggest that a promising avenue for developing new behavioral interventions meant to improve everyday functioning and quality of life among older adults living with HIV might be to focus on personality interventions. In calling for potential interventions at the personality-level, however, it is important to keep in mind that for effective and lasting personality change to occur, individuals must want to change their personality (Hudson & Fraley, 2015) and must follow up on these change goals (Hudson et al., 2018).

Beyond attempting to intervene with personality change, another option might be to tailor health interventions to better fit each patient’s unique personality profile, particularly their neuroticism and conscientiousness (e.g. Hirsh et al., 2012; Rochefort et al., 2019). For example, because people higher in neuroticism are more susceptible to poor health-related behaviors due to their difficulty regulating emotions, future studies could examine whether patients high in neuroticism might benefit from introducing stress management training into health interventions (Chapman et al., 2011). For people low in conscientiousness, because they are more susceptible to poor health-related behaviors due to their difficulty following instructions from their healthcare providers, future studies could examine whether patients low in conscientiousness might benefit from more reminders, care management, and/or more detailed medical instructions (Bogg & Roberts, 2013).

In conclusion, to our knowledge, this is the first study to test the associations between personality traits and everyday functioning among older adults, and whether these associations differed depending on HIV serostatus. We found that personality traits, and especially neuroticism and conscientiousness might be important sources of variance in explaining everyday functioning among older adults, and they appeared to be equally important regardless of HIV serostatus.