Mediators and Moderators of Health-Related Quality of Life in People Living with HIV Gwang Suk Kim, PhD, RN Suhee Kim, PhD, RN* Jun Yong Choi, PhD, MD Jeong In Lee, BSN, RN Chang Gi Park, PhD Linda L. McCreary, PhD, RN, FAAN We examined whether social support moderated communication and self-management, and tested whether self-management mediated communication, instrumental and emotional social support (ISS, ESS), and health-related quality of life (HRQOL) in Korean people living with HIV (PLWH). A crosssectional research design using a self-reported survey questionnaire was conducted. Data for 205 PLWH were collected at the outpatient divisions of seven hospitals. HRQOL was positively associated with communication, ISS, ESS, and self-management. ESS moderated the relationship between communication and self-management through a significant interaction with communication. Johnson-Neyman analysis indicated that the interaction effect of ESS was significant at the range from 4 to 14.4. However, the ISS did not moderate the relationship between communication and self-management. Self-management mediated the relationship between communication, ESS, and HRQOL. The findings suggest that nursing interventions should be focused on providing interactive communication and ESS to improve self-management and HRQOL of PLWH. (Journal of the Association of Nurses in AIDS Care, -, 1-12) Copyright Ó 2018 Association of Nurses in AIDS Care Key words: communication, HIV, quality of life, self-management, social support
ecently, HIV infection has been considered as a chronic disease because the effective dissemination of antiretroviral therapy has lengthened life expectancy for people living with HIV (PLWH; Mitchell & Linsk, 2004; Webel et al., 2012). In 2016, about 36.7 million people were living with HIV around the world (World Health Organization, 2018). In Korea in 2016, 11,439 patients were living with HIV; 92.8% were male (Korean Centers for Disease Control and Prevention, 2017). Thus, consistent health care services are needed for PLWH to manage their chronic condition. Gwang Suk Kim, PhD, RN, is a Professor, Mo-Im Kim Nursing Research Institute, College of Nursing, Yonsei University, Seoul, Republic of Korea. Suhee Kim, PhD, RN, is an Assistant Professor, Division of Nursing and Research Institute of Nursing Science, Hallym University, Chuncheon, Gangwon-do, Republic of Korea. (*Correspondence to: [email protected]
). Jun Yong Choi, PhD, MD, is a Professor, Department of Internal Medicine, College of Medicine, Yonsei University, Seoul, Republic of Korea. Jeong In Lee, BSN, RN, is a Staff Nurse, Division of Nursing/Infectious Disease, Yonsei University Health System, Seoul, Republic of Korea. Chang Gi Park, PhD, is a Research Assistant Professor, College of Nursing, University of Illinois at Chicago, Chicago, Illinois, USA. Linda L. McCreary, PhD, RN, FAAN, is a Clinical Associate Professor, College of Nursing, University of Illinois at Chicago, Chicago, Illinois, USA.
JOURNAL OF THE ASSOCIATION OF NURSES IN AIDS CARE, Vol. -, No. -, -/- 2018, 1-12 https://doi.org/10.1016/j.jana.2018.02.001 Copyright Ó 2018 Association of Nurses in AIDS Care
2 JANAC Vol. -, No. -, -/- 2018
Health-related quality of life (HRQOL) is one of the optimal goals of health care in patients with chronic diseases, including HIV. HRQOL has been defined as the impact of illness on an individual’s function and well-being, and is considered as a parameter to decide whether a patient is doing well or is responding to the treatment (Revicki et al., 2000; Trask et al., 2009). Because PLWH are vulnerable to HIV-related symptoms, and their overall daily activities are influenced by their symptoms, their quality of life (QoL) is threatened (Sukati et al., 2005). Nursing interventions are aimed at helping patients attain the highest possible HRQOL. Therefore, it is of utmost importance that nurses understand the variables that affect HRQOL. In addition, moderator and mediator effects (variables) can help care providers plan more specific interventions and select priority targets. Moderator effects help to identify ‘‘when’’ or ‘‘for whom,’’ predicting or causing a dependent variable strongly (or weakly), and mediator effects help to explain ‘‘how’’ or ‘‘why,’’ predicting or causing a dependent variable such as HRQOL (Frazier et al., 2004). Self-management is an important antecedent of HRQOL. Chronic disease patient self-management has been considered a primary behavior to achieve health outcomes such as QoL, HRQOL, and disease-related physiological/functional status (Kawi, 2014; Whittemore et al., 2013). Selfmanagement by patients with chronic diseases was defined as performing a combination of lifelong work such as taking medication, eating well, performing the patient role, and managing emotional discomfort (Lorig & Holman, 2003). Researchers have explored the role of self-management in mediating certain health outcomes, including QoL. For example, Whittemore and colleagues (2013) found that self-management mediated the effects of family function and depressive symptoms on hemoglobin A1C and QoL in U.S. youth with diabetes. Communication between patients and physicians/ nurses has been found to be another important antecedent of self-management, with good communication related to high levels of self-management (Allen et al., 2008; Kruse et al., 2013). One characteristic of patient self-management is active participation in the care process. Active participation is realized through communications with providers.
Kruse and colleagues (2013) reported that the median encounter time with patients with diabetes in primary care clinics was 22.3 minutes, and 23.5% of this time (5.2 minutes) was spent discussing self-care. Researchers and clinicians should seek ways to promote communication between patients and providers to improve self-management. Researchers have also examined the role of social support as a moderator in the context of health behavior and health outcomes (Baek et al., 2014; Oetzel et al., 2014). Although social support is known to be a multidimensional construct, it has been measured by single-dimensional and multidimensional instruments (Breet et al., 2014; Lee et al., 2013; Moser et al., 2012). When multidimensional social support is measured, the moderating effects should be tested not only for the total level of social support, but also for each dimension of social support separately, in order to plan specific and effective strategies. Studies that have examined associations between communication, social support, and selfmanagement in patients with chronic diseases can guide effective intervention strategies. However, few studies have examined how these variables are associated with PLWH; studies identifying moderator or mediator effects to get in-depth information about HRQOL in PLWH are scarce. The aim of our study was to verify factors associated with HRQOL in PLWH. We hypothesized that: (a) communication, social support, and selfmanagement would have direct associations with HRQOL; (b) social support would moderate relationships between communication and self-management; and (c) self-management would mediate relationships between communication, social support, and HRQOL.
Methods Study Design and Participants Our study was cross-sectional, descriptive correlational research. Data were taken from a previous study in which the self-management instrument was validated and a number of other variables examined (Kim et al., 2015). The participants were recruited
Kim et al. / Mediators and Moderators of Health-Related Quality of Life in People Living with HIV
from seven Korean general hospitals using a convenience sampling method. Eligibility criteria included patients who were (a) diagnosed with HIV; (b) receiving health care service at the outpatient department of infectious diseases; and (c) able to read, understand, and answer the questionnaire. Data Collection Procedures The institutional review board (IRB) of Yonsei University College of Nursing approved the study (IRB No. 2017-0022). Data were collected by a nurse in charge of patient education, treatment, and care for PLWH at each outpatient department. The principal investigator described the aim and process of the study to nurses in a total of 16 general hospitals, which were the only hospitals in Korea that provided HIV counseling services at the time of data collection, and asked them to participate in the study. A total of seven nurses, one nurse in each of seven general hospitals, agreed to participate in the study and were trained regarding data collection methods and procedures. The nurses initially assessed PLWH eligibility criteria for the study and introduced the PLWH to the aim of the study. PLWH were asked to sign an informed consent document and complete the survey voluntarily. Questionnaires (n 5 250) were distributed by the nurses and 219 completed questionnaires were received. Because of missing data on primary variables, 14 questionnaires were abandoned. Thus, data from 205 subjects were analyzed as the final sample. When participants finished the survey questionnaire, nurses provided individual health counseling in accordance with each participant’s questions and gave a token of appreciation as compensation for their time ($10 USD gift card).
Health-related quality of life (HRQOL, SF-12). HRQOL was measured using the Korean version of the Short Form Health Survey-12 (SF-12, www.sf36.org/tools/sf12.shtml). This scale is a common and widely used instrument to evaluate QoL, not only for healthy people but also for those with chronic diseases. The 12-item HRQOL is composed of eight domains: general health perception (1 item), physical function (2), role limitations due to physical health (2), role limitations due to emotional health (2), social function (1), mental health (2), vitality (1), and pain (1). The eight domains were distributed into a physical component summary (PCS, 6 items) and a mental component summary (MCS, 6 items), following the technique used by Chariyalertsak and colleagues (2011). The scale is used frequently to measure HRQOL in PLWH (Chariyalertsak et al., 2011; Viswanathan et al., 2005). The Cronbach’s a value for our study was strong at 0.89. Self-management. Self-management was measured using the HIV Self-Management Scale developed and revised by Webel and colleagues (2012). The 20-item scale is composed of three domains: daily self-management health practices (12 items), social support of HIV self-management (3), and chronic nature of HIV self-management (5). Each item was assessed with a 4-point Likert scale (0 5 not applicable, 3 5 all of the time). Webel and colleagues (2012) developed the scale to reflect the day-to-day decisions that individuals make to manage their diseases. Possible total scores ranged from 0 to 60, with higher scores indicating a higher level of self-management. Our research team tested the validity and reliability of the Korean version of the HIV Self-Management Scale following a translation process (Kim et al., 2015). We found high reliability (Cronbach’s alpha .89 for the total scale) and adequate content, concurrent, and construct validity.
Measures Demographic and clinical characteristics. We collected participant demographic characteristics, including gender, age, education level, job status, marital status, and type of household. Clinical characteristics were measured by length of time since HIV diagnosis, participation in self-help groups, and awareness of current viral load.
Social support. Social support was measured using a modified Medical Outcomes Study Social Support Survey (Moser et al., 2012). The instrument was originally developed as a 19-item, fourdimension survey by Sherbourne and Stewart (1991) to measure social support perceived by PLWH, and was shortened to eight items assessing two dimensions: instrumental social support (ISS) and emotional
4 JANAC Vol. -, No. -, -/- 2018
social support (ESS; Moser et al., 2012). Each item was rated on a 5-point scale from never 5 1 to always 5 5. Possible scores ranged from 8 to 40, and higher scores were indicative of higher levels of social support. Moser and colleagues (2012) reported high internal consistency reliability of the total social support (TSS) scale, with Cronbach’s alpha from 0.88 to 0.93 in three different populations. To our knowledge, ours was the first study to use the tool with a Korean sample. Two investigators who were experts in English and health terminology translated the tool from English to Korean. They then reviewed both Korean translations, and discussed item meanings, comparing them to the original forms of the scale. After minor revisions, the Korean version of the scale was confirmed. For our study, Cronbach’s alpha coefficient for TSS was .94, for ISS it was .95, and for ESS it was .90. Communication. Communication items were constructed by our research team to assess patient– health care provider communication in Korea. The authors generated the original item pool from existing instruments that measured communication between patients and physicians/nurses (Kim, 2009; Park, 2007). The authors also conducted interviews with five PLWH and seven experienced nurses, and identified essential aspects of communication between PLWH and their physicians/nurses that should be measured. Seven questions were adapted from existing instruments to assess the levels of communication between PLWH and providers: asking questions comfortably to my physician/nurse, telling my physician/nurse everything about my symptoms, communicating to my physician/nurse with eye contact, my physician/nurse listening carefully to my words, answering honestly to my physician/nurse’s questions, having a suitable place to communicate with my physician/nurse, and having enough time to communicate with them. Each item was rated on a 5-point scale, and higher scores indicated higher levels of communication. The Cronbach’s a value for this instrument was 0.92. Statistical Analysis Descriptive statistical analyses such as t-test and one-way analyses of variance were used to examine
the differences in self-management and HRQOL according to demographic and clinical characteristics. The correlations between length of time since HIV diagnosis, communication, social support, selfmanagement, and HRQOL were examined using Pearson’s correlation coefficient. Structural equation modeling (SEM)-based path analysis was used to test moderation effects and mediation effects on HRQOL. Time since HIV diagnosis, job status, and awareness of own recent viral load were controlled in path analysis. The evaluation of the model fit was based on (a) the p value of X2 . .05, indicating an adequate model fit to accept the null hypothesis of nondiscrepancy between fitted and sample covariances; (b) the comparative fit index $ .90 indicating an adequate fit; and (c) the root mean square error of approximation , .08, indicating an adequate model fit (Hu & Bentler, 1999; Tabachnick & Fidell, 2012). To identify the statistically significant region, or range of scores in which social support acted as a moderator, the Johnson-Neyman (J-N) method was used. The J-N method has been adapted for cases in which moderators were continuous variables (Hayes, 2013; Lazar et al., 2013). The significance level was set at p , .05. The descriptive analyses, Pearson’s correlation coefficient, and the J-N method were carried out using IBM SPSS version 19 (SPSS Inc., Chicago, IL, USA), and SEM was carried out using STATA 13.0 (StataCorp LLC, College Station, TX, USA).
Results Description of Participants and Main Variables Characteristics of the participants are described in Table 1. Most participants were male (92.6%), and the mean age was 41.6 years (SD 5 11.04). Most participants (85.9%) had a high school or college-level education; only 14.1% were married; and 63.7% were employed. The mean time length since HIV diagnosis was 6.7 years (SD 5 5.63). Two-thirds (63.4%) of participants were aware of their own recent viral load. Communication scores ranged from 15 to 35, with a mean of 28.47 (64.78). The mean scores for TSS, ISS, and ESS were 25.34 (68.05), 12.66 (64.55),
Kim et al. / Mediators and Moderators of Health-Related Quality of Life in People Living with HIV Table 1.
Comparisons of Self-Management and HRQOL by Demographic and Clinical Characteristics of Participants
Characteristics Age (years) 18-40 41-66 Gender Male Female Education level Middle school or less High school College or more Marital status Not married Married Divorced or separated Job status Employed Unemployed Type of household Living alone Living with family Living with friend(s) Participation in self-help group Participate Do not participate Time length of HIV diagnosis (years) #1 .1 Awareness of own recent viral load Know Do not know
Self-Management M (SD) p-Value
HRQOL M (SD)
73 (46.8) 81 (53.2)
38.90 (10.41) 40.51 (9.59)
87.17 (10.32) 86.93 (9.81)
189 (92.6) 15 (7.4)
39.78 (10.00) 43.93 (6.31)
87.00 (10.30) 88.40 (10.01)
29 (14.1) 82 (40.0) 94 (45.9)
38.38 (12.05) 40.73 (8.16) 39.97 (10.41)
83.64 (13.98) 88.03 (10.17) 87.31 (8.74)
151 (73.7) 29 (14.1) 25 (12.2)
39.61 (9.79) 40.00 (10.66) 42.76 (8.85)
86.84 (10.04) 87.66 (11.31) 87.81 (10.49)
130 (63.7) 74 (36.3)
40.42 (9.87) 39.50 (9.79)
88.55 (9.17) 84.55 (11.55)
77 (39.3) 94 (48.0) 25 (12.7)
40.57 (10.96) 39.91 (8.99) 40.96 (9.64)
86.31 (10.71) 87.83 (9.90) 87.37 (9.33)
25 (12.2) 180 (87.8)
42.88 (9.72) 39.66 (9.79)
87.90 (10.66) 86.96 (10.20)
24 (15.6) 130 (84.4)
37.71 (10.32) 40.12 (9.92)
86.79 (9.80) 87.09 (10.10)
130 (63.4) 73 (35.6)
41.58 (9.00) 37.14 (10.46)
87.86 (10.92) 85.56 (9.57)
Note. HRQOL 5 health related quality of life. a. Total of each variable varies because of the exclusion of nonresponse or missing data.
and 12.68 (64.04), respectively (range 5 8-40, 4-20, and 4-20, respectively). Self-management scores ranged from 15 to 60, with a mean of 40.05 (69.82). For the full HRQOL scale, the mean was 87.08 (610.24), with the range from 55.91 to 110.41. For the two summary HRQOL scores, the mean PCS was 41.73 (65.84) and the mean MCS was 45.34 (612.33). Differences of HRQOL and Self-Management by Characteristics HRQOL varied significantly according to job status. Participants who were employed had higher
scores for SF-12 than those who were not (t 5 22.56, p 5 .012). The difference of the level of self-management was reported according to awareness of own recent viral load. A higher level of selfmanagement was reported in the participants who knew their recent viral loads compared to those who did not (t 5 23.18, p 5 .002). See Table 1 for these findings. HRQOL was significantly correlated to communication, social support, and self-management (see Table 2). HRQOL was positively associated with communication (r 5 .174, p , .05), ISS (r 5 .367, p , .001), ESS (r 5 .361, p , .001), and selfmanagement (r 5 .320, p , .001). Additionally,
6 JANAC Vol. -, No. -, -/- 2018 Table 2.
Correlations Between Variables
1. Time length of HIV diagnosis 2. Communication 3. TSS 4. ISS 5. ESS 6. Self-management 7. HRQOL
2.028 .296*** .944***
2.025 .271** .929*** .755***
.182* .330*** .409*** .332*** .441***
2.031 .174* .389*** .367*** .361*** .320***
Note. TSS 5 total social support; ISS 5 instrumental social support; ESS 5 emotional social support; HRQOL 5 health-related quality of life. *p , .05, **p , .01, ***p , .001.
self-management was positively associated with the length of time since HIV diagnosis (r 5 .182, p , .05).
Dura on of Dx .24**
Moderation Effect of Social Support SEM-based path analysis was conducted to test the moderation effect of social support and mediation effect of self-management. We tested moderation and mediation effects with three models; model 1 included TSS, model 2 included ISS, and model 3 included ESS (Figure 1). To test moderation effects, three interaction terms of ‘‘communication 3 TSS,’’ ‘‘communication 3 ISS,’’ and ‘‘communication 3 ESS’’ were specified. Moderation effects of ESS and interaction with communication on selfmanagement were observed (coefficient 5 21.08). On the other hand, no moderation effects of interactions of ‘‘communication 3 TSS’’ and ‘‘communication 3 ISS’’ were found. In other words, this significant negative interaction effect meant that the positive effect of independent variable (communication) on dependent variable (self-management) will increase as the moderate variable (ESS) decreases. To probe the interaction between communication and ESS and to identify the ranges of ESS for which the effect on self-management was significant, the J-N method was used. This analysis showed that the range of statistical significance was in the lower range of ESS scores, from 4 to 14.4 (Figure 2). Mediation Effect of Self-Management Mediation effects of self-management between communication/social support and HRQOL were
Social support (total)
.91*** Communica on x SS (total)
Quality of life
Dura on of Dx .23** Selfmanagement
.22** Social support (instrumental)
.92*** Communica on x SS (instrumental)
Quality of life
Dura on of Dx Communica on
.30*** Social support (emo onal)
Communica on x SS (emo onal)
.18 * -1.08* Quality of life
Figure 1. Moderation and mediation of health-related quality of life in people living with HIV: the role of emotional social support and self-management. Note. *p , .05, **p , .01, ***p , .001. SS 5 social support; Dx 5 diagnosis.
Kim et al. / Mediators and Moderators of Health-Related Quality of Life in People Living with HIV
LLCI ULCI 20
Effect of communication on self-management
Emotional social support
Figure 2. Significant region of interaction between communication and emotional social support. Note. Johnson-Neyman (J-N) technique shows the differing interaction effect of communication and emotional social support on self-management (thick line with square). The area between the LLCI and the ULCI represents the 95% confidence interval. When the zero line of the x-axis is included in the confidence space, the effect at the emotional social support is not significant. The interaction effect was significantly moderated at the range of emotional social support from 4 to 14.4. LLCI 5 lower-limit confidence interval; ULCI 5 upper-limit confidence interval.
observed in model 1 and model 3 (Table 3 & Figure 1). In model 1, the standardized path coefficient between communication and selfmanagement was .45, the path between TSS and self-management was .74, and the path between self-management and quality of life was .18. All of these path coefficients were significant, but no indirect effects of communication, TSS, and ‘‘communication 3 TSS’’ to HRQOL were found. In model 3, the standardized path coefficient between communication and self-management was .67, the path between ESS and self-management was 1.24, and the path between self-management and HRQOL was .18. All path coefficients were significant as well. Indirect effects of communication to HRQOL (p 5 .043) and ESS to HRQOL (p 5 .045) were found, while no indirect effect of interaction of ‘‘communication 3 ESS’’ to HRQOL was found. Additionally, time since HIV diagnosis was associated with self-management in the three models, and there were indirect effects on HRQOL in the three models.
Only model 3, including ESS, demonstrated both moderation effect of ESS and mediation effect of self-management together. The structural model based on the hypothesized relationships between communication, ESS, self-management, and HRQOL (model 3) was a good model fit: X2(15) 5 .998 (p value 5 1), root mean square error of approximation 5 .000, comparative fit index 5 1.000.
Discussion We examined the moderating effect of social support on the relationships between communication and self-management, and the effect of self-management as a pathway to mediate the relationships between communication, social support, and HRQOL in PLWH. ESS moderated the relationship between communication and self-management through the significant interaction with communication. The range from 4 to 14.4 of ESS was identified as the
8 JANAC Vol. -, No. -, -/- 2018 Table 3.
Mediating Effect of Self-management: Standardized Direct, Indirect, and Total Effects in HRQOL Variables
Model 1 SM
Model 2 SM
Model 3 SM
CC TSS CC 3 TSS DDx CC TSS CC 3 TSS SM DDx
0.917 (.020) 0.895 (.045) 20.016 (.291) 0.035 (.001) 0.340 (.433) 0.715 (.147) 20.011 (.500) 0.191 (.011) –
– – – –
CC ISS CC 3 ISS DDx CC ISS CC 3 ISS SM DDx
0.535 (.124) 0.457 (.564) 0.005 (.854) 0.034 (.001) 0.235 (.528) 1.001 (.237) 20.012 (.671) 0.220 (.002) –
– – – –
CC ESS CC x ESS DDx CC ESS CC 3 ESS SM DDx
1.384 (,.001) 3.009 (.001) 20.072 (.017) 0.036 (,.001) 0.445 (.327) 1.554 (.126) 20.030 (.379) 0.191 (.013) –
– – – –
0.175 (.085) 0.171 (.115) 20.003 (.329) – 0.007 (.041)
0.118 (.170) 0.101 (.571) 0.001 (.854) – 0.007 (.027)
0.264 (.043) 0.574 (.045) 20.014 (.085) – 0.007 (.041)
TE (p) 0.917 (.020) 0.895 (.045) 20.016 (.291) 0.035 (.001) 0.515 (.237) 0.885 (.074) 20.014 (.396) 0.191 (.011) 0.007 (.041) 0.535 (.124) 0.457 (.564) 0.005 (.854) 0.034 (.001) 0.353 (.351) 1.102 (.202) 20.011 (.705) 0.220 (.002) 0.007 (.027) 1.384 (,.001) 3.009 (.001) 20.072 (.017) 0.036 (,.001) 0.709 (.115) 2.128 (.034) 20.044 (.202) 0.191 (.013) 0.007 (.041)
Note. DE 5 direct effect; IE 5 indirect effect through self-management; TE 5 total effect; SMC 5 squared multiple correlation; SM 5 self-management; HRQOL 5 health-related quality of life; CC 5 communication; TSS 5 total social support; ISS 5 instrumental social support; ESS 5 emotional social support; DDx 5 duration of diagnosis.
statistically significant region in which emotional support served as a moderator. Thus, PLWH with low ESS should be considered priority targets for participating in intervention programs to improve communication skills, and physicians/nurses should listen carefully and have enough time to communicate with patients because the effect of communication on self-management will be increased in PLWH who have lower levels of ESS (,14.4). Self-management mediated the relationship between communication, ESS, and HRQOL. In other words, communication and ESS have indirect effects of HRQOL through self-management, although these did not have direct effects on HRQOL. Thus, communication and ESS should be increased to enhance HRQOL as well as self-management.
Previous research has examined whether social support moderated in the health condition or health outcome contexts. Breet and colleagues (2014) reported that social support did not moderate the relationship between HIV-related stigma and depression. Oetzel and colleagues (2014) also found no evidence of a moderating effect of social support on the relationship between social undermining and HRQOL in PLWH. On the other hand, Baek and colleagues (2014) demonstrated that social support satisfaction and number of supports significantly moderated the relationship between diabetes complications and distress in adults with type 2 diabetes. We considered possible reasons for the inconsistent moderation effects of social support in these studies. One possible explanation was the
Kim et al. / Mediators and Moderators of Health-Related Quality of Life in People Living with HIV
differences in measurement of social support. The studies we reviewed measured social support using various dimensions, such as a single dimension, friend and family support, and instrumental and emotional support. Additionally, most studies examined moderation effects of social support by a total score, even when social support was measured by multiple dimensions. Furthermore, because the research questions of these studies differed, the variables that interacted with social support differed as well. Thus, researchers should deliberately consider the type of social support they intend to measure when they select a scale. If they choose a multidimensional measure, they should conduct analysis to identify moderating effects of each dimension of social support. In our study, we examined the moderating effects of TSS and each of its dimensions (ISS and ESS) on the relationship between self-management and communication. After analysis, we found that ESS moderated between communication and selfmanagement through a significant interaction with communication, whereas the moderation effect of ISS between communication and self-management was not significant. Although we found no studies that examined the moderation effect of ESS and ISS in PLWH, Lee and colleagues (2013) examined the moderation effect of ESS and ISS seeking on blood pressure in Chinese Americans. They reported that the interactions between instrumental support seeking and interdependent self-construal on diastolic blood pressure were significant, whereas emotional support seeking was not significant. According to that study and our study, we can assume that the significant dimension of social support as a moderator might depend on the health behavior and health outcome context and on the specific population. We suggest that health care providers evaluate levels of social support as moderators in relationships with communication for improving self-management of PLWH. Additionally, future studies should examine moderation effects of social support by each dimension separately to find the most significant dimension of social support that interacts with other determinants of health behaviors or health conditions. Analysis using the J-N method showed that the interaction effect was significant at the lower range
of ESS, from 4 to 14.4. Researchers have asserted that identification of patient baseline characteristics may or may not identify those who might benefit from a particular intervention to result in more effective health care (Garcia et al., 2013; Lazar et al., 2013). The J-N method has been used in other disciplines, such as dental research and marketing, as one approach to identify participant characteristics (Garcia et al., 2013; Lazar et al., 2013; Spiller et al., 2013). These findings suggest that an intervention program to improve selfmanagement by enhancing communication may be more effective for PLWH who have low ESS. We also investigated the mediation effect of selfmanagement in PLWH. Self-management significantly mediated the relationship between communication, ESS, and HRQOL. Thus, it appears that the level of HRQOL of the patient improved or worsened how communication and ESS affected the ability of the patient to manage his/her own disease. In line with this finding, Whittemore and colleagues (2013) demonstrated that self-management mediated the relationship between family function and depressive symptoms on hemoglobin A1C and QoL in youth with diabetes. These findings provided strong evidence regarding self-management as an antecedent of HRQOL in patients with chronic diseases. Additionally, in our study, self-management was positively correlated with the length of time since HIV diagnosis. Thus, a program to increase the selfmanagement of PLWH, including special components, such as communication with providers and screening for emotional support, may be more beneficial to enhancing the HRQOL of patients immediately after an HIV diagnosis. We also suggest a strategy to improve self-management abilities of PLWH. Participants in our study who were aware of their current viral loads showed higher levels of self-management compared to those who were not aware. Thus, health care providers should provide information about recent viral loads to encourage patient self-monitoring. The mean levels of HRQOL in our study were 41.73 for PCS and 45.34 for MCS. These were lower than general population norms and lower than found by Chariyalertsak and colleagues (2011): 49.0 for PCS and 45.6 for MCS in the United States. Another study conducted in the United States used the SF-12
10 JANAC Vol. -, No. -, -/- 2018
to examine HRQOL in PLWH. Viswanathan and colleagues (2005) reported mean scores of 41.0 for PCS and 41.9 for MCS. A conceivable explanation respecting the lower mean score of the HRQOL might be the high rate in those with an unemployed and/or low-income status. Only 64% of our study participants reported working for a salary, and patients who were employed had higher scores for total HRQOL than those who were not employed. Viswanathan and colleagues (2005) sampled lowincome PLWH, and 70% of them were unemployed. In contrast, according to chariyalertsak and colleagues (2011), only 13% of participants in their study were unemployed. These results highlight the necessity of a program or policy to stabilize PLWH employment or guaranteed income, which may help to enhance HRQOL. Limitations Our study has several limitations. First, because of the cross-sectional nature of the study, evaluating the moderating and mediating effects cannot result in causal inference. Second, generalization of these findings to the general population may not be assured, because (a) the majority of participants were male, (b) the self-assessment method may have overestimated participant levels of communication and selfmanagement, and (c) participant levels of communication, social support, and self-management were possibly higher than those of the general population of PLWH. In the future, male and female participants, subjective and objective assessments, and random sampling should be used to improve generalizability of the findings.
Conclusions We found that ESS moderated the relationships between communication and self-management. PLWH with lower levels of ESS should be especially considered as high-priority patients for intervention programs to improve communication abilities with health care providers. In addition, our study found that self-management mediated the relationship between communication, ESS, and HRQOL. Planned interventions that increase self-management and
focus on strengthening relationships between communication and ESS might be an important strategy to improve HRQOL in PLWH. Nurses should assess patient communication, social support, and selfmanagement levels; such appraisals might be especially needed in the early period after HIV diagnosis, when patients may not have learned to enhance the determinants of HRQOL. Future studies to expand our understanding of moderation and mediation effects on HRQOL in PLWH should be performed using longitudinal designs that include randomized samples.
Key Considerations Improving communication and social support may help increase self-management in people living with HIV (PLWH). Clinicians should evaluate communication, social support, and self-management levels to ensure the health-related quality of life (HRQOL) for PLWH. Such evaluation might be needed in the early period after HIV diagnosis. Clients living with HIV should be provided intervention focused on interactive communication and emotional social support to improve their self-management and HRQOL.
Disclosures The authors report no real or perceived vested interests that relate to this article that could be construed as a conflict of interest.
Acknowledgments This research was supported by the Korea Centers for Disease Control and Prevention (KCDC) and Hallym University (7-2012-0333, Principal Investigator: GS Kim), and the National Research Foundation of Korea (NRF-2015R1D1A1A01057423, Principal Investigator: GS Kim).
Kim et al. / Mediators and Moderators of Health-Related Quality of Life in People Living with HIV
References Allen M., Lezzoni L. I., Huang A., Huang L., Leveille S. G. (2008). Improving patient-clinician communication about chronic conditions: Description of an internet-based nurse e-coach intervention. Nursing Research, 57(2), 107-112. https://doi.org/10.1097/01.NNR.0000313478.47379.98 Baek R. N., Tanenbaum M. L., Gonzalez J. S. (2014). Diabetes burden and diabetes distress: The buffering effect of social support. Annals of Behavioral Medicine, 48(2), 145-155. https://doi.org/10.1007/s12160-013-9585-4 Breet E., Kagee A., Seedat S. (2014). HIV-related stigma and symptoms of post-traumatic stress disorder and depression in HIV-infected individuals: Does social support play a mediating or moderating role? AIDS Care, 26(8), 947-951. https:// doi.org/10.1080/09540121.2014.901486 Chariyalertsak S., Wansom T., Kawichai S., Ruangyuttikarna C., Kemerer V. F., Wu A. W. (2011). Reliability and validity of Thai versions of the MOS-HIV and SF-12 quality of life questionnaires in people living with HIV/AIDS. Health and Quality of Life Outcomes, 9, 15. https://doi.org/10.1186/ 1477-7525-9-15 Frazier P. A., Tix A. P., Barron K. E. (2004). Testing moderator and mediator effects in counseling psychology research. Journal of Counseling Psychology, 51(1), 115-134. https:// doi.org/10.1037/0022-0126.96.36.199 Garcia I., Kuska R., Somerman M. J. (2013). Expanding the foundation for personalized medicine: Implications and challenges for dentistry. Journal of Dental Research, 92(Suppl. 7), 3S-10S. https://doi.org/10.1177/0022034513487209 Hayes A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: The Guilford Press. Hu L., Bentler P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1-55. https://doi.org/10.1080/10705519909540118 Kawi J. (2014). Self-management and self-management support on functional ablement in chronic low back pain. Pain Management Nursing, 15(1), 41-50. https://doi.org/10.1016/j. pmn.2012.05.001 Kim G. S., Chu S. H., Park Y., Choi J. Y., Lee J. I., Park C. G., McCreary L. L. (2015). Psychometric properties of the Korean version of the HIV self-management scale in patients with HIV. Journal of Korean Academy of Nursing, 45(3), 439-448. Kim S. H. (2009). Factors affecting communication between cancer patients and nurses. Unpublished master’s thesis. Gyeongsan, South Korea: Yeungnam University. Korean Centers for Disease Control and Prevention. (2017). Annual report on the notified HIV/AID in Korea. Retrieved from http:// www.aids.or.kr/bbs/board.php?bo_table5sub04_03 Kruse R. L., Olsberg J. E., Oliver D. P., Shigaki C. L., VetterSmith M. J., LeMaster J. W. (2013). Patient-provider communication about diabetes self-care activities. Family Medicine, 45(5), 319-322.
Lazar A. A., Gansky S. A., Halstead D. D., Slajs A., Weintraub J. A. (2013). Improving patient care using the Johnson-Neyman analysis of heterogeneity of treatment effects according to individuals’ baseline characteristics. Journal of Dental, Oral and Craniofacial Epidemiology, 1(3), 19-33. Lee Y. S. C., Suchday S., Wylie-Rosett J. (2013). Social support and networks: Cardiovascular responses following recall on immigration stress among Chinese Americans. Journal of Immigrant and Minority Health, 17(2), 543-552. https://doi. org/10.1007/s10903-013-9955-9 Lorig K. R., Holman H. (2003). Self-management education: History, definition, outcomes, and mechanisms. Annals of Behavioral Medicine, 26(1), 1-7. https://doi.org/10.1207/ S15324796ABM2601_01 Mitchell C. G., Linsk N. L. (2004). A multidimensional conceptual framework for understanding HIV/AIDS as a chronic long-term illness. Social Work, 49(3), 469-477. https://doi. org/10.1093/sw/49.3.469 Moser A., Stuck A. E., Silliman R. A., Ganz P. A., CloughGorr K. M. (2012). The eight-item modified Medical Outcomes Study Social Support Survey: Psychometric evaluation showed excellent performance. Journal of Clinical Epidemiology, 65(10), 1107-1116. https://doi.org/10.1016/j. jclinepi.2012.04.007 Oetzel J., Wilcox B., Archiopoli A., Avila M., Hell C., Hill R., Muhammad M. (2014). Social support and social undermining as explanatory factors for health-related quality of life in people living with HIV/AIDS. Journal of Health Communication, 19(6), 660-675. https://doi.org/10.1080/10810730. 2013.837555 Park W. P. (2007). The structural relationships of the factors related to patient-doctor communication in general hospitals. Unpublished doctoral dissertation. Seongnam, South Korea: Kyungwon University. Revicki D. A., Osoba D., Fairclough D., Barofsky I., Berzon R., Leidy N. K., Rothman M. (2000). Recommendations on health-related quality of life research to support labeling and promotional claims in the United States. Quality of Life Research, 9(8), 887-900. https://doi.org/10.1023/A: 1008996223999 Sherbourne C. D., Stewart A. L. (1991). The MOS social support survey. Social Science & Medicine, 32(6), 705-714. https:// doi.org/10.1016/0277-9536(91)90150-B Spiller S. A., Fitzsimons G. J., Lynch J. G., McClelland G. H. (2013). Spotlights, floodlights, and the magic number zero: Simple effects tests in moderated regression. Journal of Marketing Research, 50(2), 277-288. https://doi.org/10.1509/jmr. 12.0420 Sukati N. A., Mndebele S. C., Makoa E. T., Ramukumba T. S., Makoae L. N., Seboni N. M., . Holzemer W. L. (2005). HIV/AIDS symptom management in Southern Africa. Journal of Pain and Symptom Management, 29(2), 185-192. https://doi.org/10.1016/j.jpainsymman.2004.05.007 Tabachnick B. G., Fidell L. S. (2012). Using multivariate statistics (6th ed.). Old Tappan, NJ: Allyn & Bacon.
12 JANAC Vol. -, No. -, -/- 2018 Trask P. C., Hsu M. A., McQuellon R. (2009). Other paradigm: Health-related quality of life as a measure in cancer treatment: Its importance and relevance. Cancer Journal, 15(5), 435-440. https://doi.org/10.1097/PPO.0b013e3181b9c5b9 Viswanathan H., Anderson R., Thomas J. (2005). Nature and correlates of SF-12 physical and mental quality of life components among low-income HIV adults using an HIV service center. Quality of Life Research, 14(4), 935-944. https://doi. org/10.1007/s11136-004-3507-7 Webel A. R., Asher A., Cuca Y., Okonsky J. G., Kaihura A., Rose C. D., . Burant C. J. (2012). Measuring HIV selfmanagement in women living with HIV/AIDS: A psychometric evaluation study of the HIV Self-Management Scale. Journal of Acquired Immune Deficiency Syndromes, 60(3), e72-e81. https://doi.org/10.1097/QAI.0b013e318256623d Whittemore R., Liberti L., Jeon S., Chao A., Jaser S. S., Grey M. (2013). Self-management as a mediator of family functioning and depressive symptoms with health outcomes in youth with type I diabetes. Western Journal of Nursing Research, 36(9), 1254-1271. https://doi.org/10.1177/0193945913516546 World Health Organization. (2018). HIV/AIDS. Retrieved from http://www.who.int/gho/hiv/en/