HEA-HEA2-Brooks20121219-RRRR

SUPPLEMENTAL MATERIALS –
TO BE INCLUDED AS ONLINE SUPPLEMENT
System-Specific Analyses
In order to ensure that the observed associations between social variables at the AL
summary score were not being driven by one or two physiological systems, we investigated the
associations between the social variables and each of the physiological system risk scores which
contributed the to AL summary score. With age and demographic factors included as covariates,
the coefficients for the system risk scores were generally consistent with those reported earlier
for the AL summary score, although they tended to be smaller in magnitude and many did not
reach statistical significance. The most consistent associations were observed for negativity
variables: family negativity was associated with greater risk on the parasympathetic (B = 0.06, p
< .05), inflammatory (B = 0.06, p < .05), and metabolic/glucose (B = 0.06, p < .05) sub-scales,
friend negativity was associated with greater inflammatory risk (B = 0.05, p < .05), and spouse
negativity was associated with greater sympathetic (B = 0.06, p < .05) and inflammatory (B =
0.04, p < .05) risk. Averaged across sources, network negativity was associated with greater
parasympathetic (B = 0.07, p < .05) and inflammatory (B = 0.08, p < .001) risk and with
marginally greater metabolic/glucose risk (B = 0.05, p < .10). There were no associations
between family or network support and AL sub-scales, but friend support was marginally
associated with higher cardiovascular risk (B = 0.03, p < .10), and spouse support was associated
with lower inflammatory (B = -0.06, p < .01), cardiovascular (B = -0.05, p < .05), and
metabolic/lipids (B = -0.04, p < .05) risk. Family contact was associated with higher
metabolic/lipids risk (B = 0.02, p < .05), friend contact was related to higher cardiovascular risk
(B = 0.02, p < .05), and network contact was associated with higher cardiovascular (B = 0.01, p
< .05) and metabolic/lipids (B = 0.01, p < .05) risk. These results provide support for the
hypothesis that social influences affect multiple physiological regulatory systems, resulting in
their contributions to overall AL.
Change Over Time: Sensitivity Analyses
Although aggregate levels of change in support, negativity and contact were low, some
individuals may have experienced larger changes. In order to address this concern, we conducted
sensitivity analyses to examine whether those reporting higher levels of change differed from
those who remained stable. We explored this question by categorizing participants into groups
based on changes in social variables from M1 to M2 (i.e., increase, decrease, stable low, stable
moderate, stable high). We designated individuals as increasing if they reported an increase of
1+ units in support or negativity or 2+ units in contact (we adjusted the criterion so as to account
for the range of the scale and the standard deviation). We categorized participants into groups for
every social variable and tested for group differences in AL using ANCOVA with age as a
control. Analyses revealed no group differences in AL for any social variable except family
contact (F (4, 943) = 5.57, p < .05).
Post hoc tests indicated that participants reporting a decrease in family contact from M1 to
M2 had higher AL than all other groups except those reporting high levels of contact at both time
points. We considered whether the higher AL in the decreased contact group might reflect
worsening health with resulting reductions in social activities and interactions by examining
whether changes in family contact were related to changes in health status, as measured by the
difference between major chronic conditions at M1 and M2. ANOVA revealed group differences
in health status change (F (4, 943) = 2.41, p < .05), and inspection of the marginal means
indicated that participants reporting a decrease in family contact had the largest average increase
in health problems, although post-hoc tests indicated that only the comparisons with those
reporting consistently low or moderate levels of contact were statistically significant. Although
longitudinal data with AL are necessary to formally test this interpretation, these findings are
consistent with the idea that higher AL observed in the group reporting decreases in family
contact might reflect illness-related reductions in social activities.”
Controlling for BMI/WHR
As WHR and BMI are consequences of unhealthy behaviors in addition to being markers
of physiological dysregulation, we examined whether the effects of social variables were
maintained when WHR and BMI were adjusted for. To do so, we computed a new version of our
AL summary score that did not include WHR/BMI. Previously, WHR and BMI (measured at
Time 2, along with all other biomarkers) were included in the lipid/fat metabolism sub-scale, so
we computed a new version of this sub-scale score using the remaining indicators (high density
lipoprotein cholesterol, low density lipoprotein cholesterol, and triglycerides) and computed a
new AL summary score accordingly. We ran our final models using the adjusted AL measure,
and these analyses revealed that controlling for WHR and BMI did not alter the findings.
Supplemental Table 1
Inter-Item Correlations among Family and Friend Social Variables (N = 949)
Friend Support
Family
Negativity
Friend
Negativity
Family
Contact
Friend
Contact
Family Support
0.45***
-0.38***
-0.18***
0.38***
0.18***
Friend Support
--
-0.14***
-0.17***
0.17***
0.51***
--
0.55***
0.04
-0.04
--
0.01
0.11**
--
0.24***
Family Negativity
Friend Negativity
Family Contact
* p < .05, ** p < .01, *** p < .001, † p < .10
Supplemental Table 2
Inter-Item Correlations among Family, Friend, and Spouse Social Variables (N = 660)
Family
Support
Friend
Support
Spouse
Support
Family
Negativity
Friend
Negativity
Spouse
Negativity
Family
Contact
Friend
Support
Spouse
Support
Family
Negativity
Friend
Negativity
Spouse
Negativity
Family
Contact
Friend
Contact
0.47***
0.32***
-0.36***
-0.13**
-0.23***
0.40***
0.24***
--
0.21***
-0.16***
-0.17***
-0.14***
0.17***
0.52***
--
-0.22***
-0.14***
-0.67***
0.12**
0.06
--
0.53***
0.38***
0.03
-0.09*
--
0.31***
0.03
0.08*
--
-0.10*
-0.07†
--
0.30***
* p < .05, ** p < .01, *** p < .001, † p < .10