Recalled Neighborhood Environments, Parental Control, and Cytokine-Mediated Response to Viral Challenge

Abstract

Objective:

Neighborhood risk in childhood is associated with poor health across the lifespan. However, many people who are reared in risky neighborhoods remain healthy in adulthood. In the context of high-risk neighborhoods, controlling parenting practices might promote better physical health outcomes later in life. The current study utilized a viral challenge paradigm to examine whether parental control throughout childhood moderated the association between recalled neighborhood risk and cytokine-mediated cold susceptibility.

Methods:

A sample of 209 healthy adults completed questionnaires to assess recalled neighborhood risk and parental control over the first 15 years of life, were exposed to a common cold virus, and were quarantined for six days. Researchers assessed nasal proinflammatory cytokine production and objective markers of illness. Participants were diagnosed with a clinical cold if they met infection and objective illness criteria.

Results:

A significant Neighborhood Risk × Parental Control interaction emerged to predict proinflammatory cytokine production. Further, parental control moderated the cytokine-mediated association between neighborhood risk and cold diagnosis (index = −.073, 95% CI [−.170, −.016]), likelihood of infection (index = −.071, 95% CI [−.172, −.015]), and meeting objective symptom criteria (index = −.074, 95% CI [−.195, −.005]). Specifically, there was a negative association between neighborhood risk and objective cold diagnosis and infection status at higher levels of parental control, but a nonsignificant association at lower levels of parental control.

Conclusion:

Findings suggest that the degree to which recalled neighborhood risk is related to adult health varies as a function of parental control throughout childhood.

Risky neighborhoods in childhood, broadly characterized by poor-quality physical and social characteristics (e.g., uncleanliness, pollution, high traffic volume, noisiness, limited access to public services and amenities, and compromised safety due to high rates of crime and violence) have been linked to poor lifespan physical health outcomes (1–4). For example, census tract measures of neighborhood-level socioeconomic disadvantage are associated with larger body mass indexes (BMI) and lower basal cortisol levels among adolescents (5) and poor neighborhood conditions predict harmful inflammatory outcomes among children with asthma (6,7). Further, recalled neighborhood physical and safety conditions throughout childhood are associated with increased latent virus reactivation in adulthood (2). Characteristics associated with neighborhood poverty, such as pollution and housing disrepair, are associated with allostatic load (8–10) and exposure to neighborhood crime and violence in childhood predicts engagement in poor health behaviors (1). Collectively, this evidence suggests neighborhood physical and social characteristics, assessed via self-reports and community data, are associated with health outcomes.

Early neighborhood risk may influence physical health through disruptions to the neuroendocrine, immune, and stress-response systems (1,8–11). In addition, individuals reared in unsafe neighborhoods often engage in risky health behaviors (e.g., drug use; (1)). In turn, alterations to biological systems and engagement in poor health behaviors may give rise to heightened susceptibility to acute and chronic illnesses (12–14). However, despite robust associations among indicators of neighborhood risk and health, many individuals who grow up in risky neighborhoods remain healthy. Protective factors, such as high-quality parenting, have been shown to mitigate negative consequences of adverse childhood environments (15–17).

While early research focused on the benefits of supportive and autonomy-promoting parenting (18–20), recent work suggests that social contexts shape the extent to which parenting behaviors are beneficial for child development (21–23). For example, in risky neighborhoods, greater parental control (i.e., attempts to govern children’s whereabouts and activities) and undermining of youths’ autonomy (i.e., restricting children from making decisions) may reflect parents’ intentions to shield children from risky neighborhood and social conditions (24,25). For youth in high-risk contexts, parental control may be positively associated with social functioning and academic achievement (22,26). However, in low-risk settings, greater parental control and undermining of children’s autonomy is negatively associated with children’s development (22,27), suggesting the influence of control-oriented parenting behaviors is context-dependent.

Importantly, in addition to proximal influences, parental control has long-term implications for individuals’ psychosocial outcomes. For example, among adolescents with greater delinquent peer affiliation, parental control protects against future engagement in problematic behaviors (28,29). Additionally, parental monitoring (i.e., supervision and knowledge of children’s day-to-day experiences)—a parenting construct widely recognized as separate but related to parental control (30)—has lasting ties to positive adjustment for individuals reared in risky environments (31–33). Conversely, prospective reports of high parental control (particularly in the absence of parental warmth and care) is associated with ongoing depressive symptomatology among low-risk samples (27,34). Taken together, these findings suggest that in risky environments, parents’ efforts to prevent youth from engaging in delinquent behaviors have persistent benefits, whereas parental control in low-risk environments may engender emotional distress. However, less is known about long-term physical health implications associated with parental control in different childhood contexts.

Compelling evidence points to inflammatory processes as key mediators that link early life experiences to many future health problems (13). As such, study designs that leverage quasi-experimental immunologic challenges, such as viral challenge paradigms, can be informative for understanding potential mechanisms through which childhood factors may influence adult health. In a common cold viral challenge paradigm, healthy adults report on various psychosocial factors and are then exposed to a relatively benign cold virus. Volunteers are subsequently quarantined for five to six days to determine who becomes infected with the virus, and who among those infected develops a clinical illness (35). Infection occurs when the virus replicates within the host, prompting both an innate and an adaptive immune response to eradicate the virus. Only a fraction of individuals who become infected with a cold virus develop clinical signs of a cold (e.g., increased mucus, nasal congestion). These signs of illness have been shown to be driven primarily by an over-exaggerated local inflammatory response to viral infection, rather than by the cold virus itself (36,37).

In the case of a common cold, an over-exaggerated local inflammatory response to infection results in unpleasant symptoms but is otherwise generally benign. However, tendencies toward excessive inflammation are known to contribute to the pathogenesis and progression of many chronic illnesses (38). Thus, viral challenge experiments provide a low-risk, highly controlled framework for determining how psychosocial stressors relate to inflammatory-linked processes and outcomes in a disease context. Psychosocial factors may modulate clinical illness risk by increasing the likelihood of becoming infected with the virus (suggesting impaired host immunity) or by promoting severe signs of illness in response to viral infection (suggesting an overly exaggerated host proinflammatory response to infection), or both. When examined together, these outcomes can provide important insights into the mechanisms through which psychosocial variables relate to risk for more serious illnesses (39,40).

The current study utilized open-access viral challenge data from the Common Cold Project (www.commoncoldproject.com) to examine whether perceived parental control in childhood and adolescence moderated the association between recalled neighborhood risk throughout childhood and inflammation-mediated response to viral exposure in adulthood (Figure 1; (41)). We hypothesized that the extent to which recalled neighborhood risk was related to inflammation-mediated cold susceptibility would be dependent on participants’ receipt of parental control. We predicted that individuals who grew up in a risky neighborhood would be less likely to meet criteria for viral infection and clinical illness if they reported greater levels of parental control.

Figure 1.

Conceptual model.

Method

Participants

The data for this study (Pittsburgh Cold Study 3; PCS3) were collected by the Laboratory for the Study of Stress, Immunity, and Disease at Carnegie Mellon University under the directorship of Sheldon Cohen, PhD, and were accessed via the Common Cold Project website (www.commoncoldproject.com; grant number NCCIH AT006694) (41). Participants included 213 adults ages 18 to 55 recruited between 2007 and 2011 from the metropolitan Pittsburgh, Pennsylvania area through newspaper advertisements. To participate, individuals were required to be in good health (as determined by medical history and physician examination). For full, detailed information on inclusion and exclusion criteria, please see https://www.cmu.edu/common-cold-project/human-subjects/index.html. Participants received $1000 in compensation for completing the study. The Institutional Review Boards at Carnegie Mellon University and the University of Pittsburgh approved the study, and all participants provided informed consent. For examples of previously published analyses using PCS3 data, please see references 2,42–45.

Procedure

Participants completed pre-study questionnaires to assess demographic information, psychosocial constructs, and health behaviors. Two to three days prior to the study, participants provided blood samples to assess pre-challenge antibody titers to rhinovirus 39 (RV39). Participants were subsequently quarantined in a hotel room for six days. On the first day, they were exposed to a standardized dose of RV39 via nasal drops and were monitored for the following five days to assess viral infection and objective markers of illness. Participants attended a follow-up visit approximately 28 days after initial viral exposure to evaluate post-challenge antibody titers used for determining RV39 infection status.

Measures

Recalled Neighborhood Risk

The Places You’ve Lived Interview (PLI; (2)) is a 13-item questionnaire designed to assess recalled safety, physical, and social characteristics of the neighborhood environment based separately on when participants were 5, 10, and 15 years of age. The 3-item Neighborhood Safety scale asked participants to reflect on whether their childhood neighborhood was safe; items were “Was your street considered safe?”, “How often did you observe violent acts on your street?”, and “How often did you see people using drugs or drinking alcohol on your street?” (α = 0.64–0.67; M = 1.53, SD = 0.58). The four-item Social Environment scale evaluated the quality of interactions with neighbors, friends, and adults; example items include “Did you have friends in the neighborhood?” and “Were there any adult neighbors who might watch out for you?” (α = 0.71–0.76; M = 1.82, SD = 0.59). The 6-item Physical Environment scale included items about litter, noisiness, and other conditions of the neighborhood; example items include “Was the condition of the street very poor?” and “Was the street very noisy?” (α = 0.76–0.82; M = .24, SD = 0.22). Response options for the neighborhood safety and social environment scales ranged from all the time (1) to never (4). For the physical environment items, response options were yes (1)/no (0). Participants’ average responses across all three ages were z-scored for each scale. The z-scores for the three scales across all ages were significantly interrelated (r range: .24 – .78, p’s < .001; α = 0.70) and averaged to form a composite for recalled neighborhood risk. Higher scores indicated riskier childhood neighborhood environments.

Parental Control

The Overprotection Dimension of the Parental Bonding Instrument (PBI; (46)) assessed recalled parental control throughout the first 15 years of life. This subscale included 6 items with response options ranging from very like (1) to very unlike (4). Participants were prompted with the following statement: “During the first 15 years of my life, my parents…” for the following 6 items related to parental control: “were overprotective of me,” “tried to control everything I did,” “tended to baby me,” “let me do things I liked doing,” “liked me to make my own decisions,” and “let me decide things on my own.” The latter three items were reverse scored, and participants’ responses were averaged so that higher scores indicated greater perceived parental control (α = 0.68; M = 2.13, SD = 0.56).

Objective Disease Outcomes

Infection Status.

Infection was present if there was recovery of the challenge virus in nasal secretions on any of the post-exposure quarantine days, or at least a 4-fold increase in antibody titers from the pre-challenge to 28-day post-challenge assessments.

Objective Symptom Criteria.

Clinical cold diagnoses were determined using previously validated criteria (35) utilizing objective measures of the following two symptoms throughout the five post-exposure days of quarantine: 1) mucus production and 2) nasal mucociliary clearance function. Total baseline-adjusted mucus weight (in grams) was measured by asking participants to use pre-weighed tissues (in grams) to collect their mucus secretions, which they sealed in pre-weighed bags (in grams). The total mucus weight was determined by subtracting the bag and tissue weight from the total weight (M = 10.6, SD = 18.2; (47)). Average nasal mucociliary clearance function was determined by the amount of time (in minutes) it took for 20 μL of a sweetened dye solution placed in participants’ nasal turbinate to produce a taste in their mouth (M = 3.82; SD = 3.88; (48)) across each post-exposure day of quarantine.

Objective Cold Diagnosis.

A clinical cold diagnosis was made if a participant (a) showed evidence of viral infection, and (b) showed sufficiently severe illness signs defined as either (i) a total baseline-adjusted mucus weight of at least 10 g or (ii) an average baseline-adjusted nasal mucociliary clearance time of at least 7 minutes (35).

Inflammatory Markers

Nasal Proinflammatory Cytokine Production.

Due to previous evidence that the development of clinical illness following infection with a cold-causing virus is a cytokine-mediated process (36,40,44), we created a composite of the nasal proinflammatory cytokines interleukin-(IL-)6, IL-1β, and tumor necrosis factor alpha (TNF-α). Participants’ daily quarantine nasal wash fluid was assayed for IL-6 (M = 58.3, SD = 80.6), IL-1β (M = 31.6, SD = 44.0), and TNF-α (M = 6.00, SD = 9.72) in pg/mL via commercially available enzyme-linked immunosorbent assays (Endogen) performed using the manufacturer’s instructions. Nasal cytokine response to viral exposure was determined by log-transforming the area under the curve (AUC) for the five post-exposure days in quarantine, adjusted for the baseline cytokines. Nasal secretion levels of IL-6, IL-1β, TNF-α were highly interrelated (r range: .55–.66, ps < .001), and therefore a composite measure was created by summing standardized IL-6, IL-1β, and TNF-α scores (see references 44,49).

Standard Control Variables

Based on standard analytic procedures utilizing Common Cold Project viral challenge data published over the past three decades (e.g., 35,40), seven “standard covariates” were included in all analyses: age, sex (male = 0; female = 1), race (White = 0; non-White = 1), socioeconomic status (represented by years of education), BMI, season (spring was the reference category), and pre-challenge antibody titers to RV39 (coded as < 4 = 0 and ≥ 4 = 1).¹

Psychological Control Variables

Previous evidence suggests that contemporaneous perceived stress, negative and positive affectivity, and personality characteristics (e.g., neuroticism) can bias recalled experiences (50). We ran follow-up analyses including these factors as covariates. Perceived stress was assessed via the 10-item Perceived Stress Scale (PSS), on which participants rated how often they felt or thought a certain way in relation to stressful/upsetting events over the past month (never [0] to very often [4]). Responses were summed and higher scores indicated greater perceived stress (M = 12.0, SD = 5.67). Trait positive and negative affect were assessed via the Positive and Negative Affect Schedule (PANAS) – Expanded Form (51). Participants rated how they generally felt on a scale from very slightly or not at all (1) to extremely (5). The sum of 10 items assessed positive affect (M = 34.9, SD = 7.08), and the sum of an additional 10 items assessed negative affect (M = 16.4, SD = 5.61). Neuroticism was assessed via the 10-item ‘emotional stability’ subscale of the International Personality Item Pool (IPIP) Big-Five Factor Markers (52); items were scored on a scale ranging from very inaccurate (1) to very accurate (5). Responses were summed so that higher scores indicated greater neuroticism (M = 25.2, SD = 7.57). Because trait negative affect and neuroticism were highly correlated (r = .71, p < .001), they were standardized and summed to create a composite measure of negative affectivity.

Data Analytic Plan

Data were analyzed using IBM SPSS Statistics Version 25 and PROCESS Version 2.16 (53). First, a moderation model assessed the Recalled Neighborhood Risk × Parental Control interaction to predict nasal inflammatory cytokine production (i.e., the first leg of the Figure 1 model). Next, percentile bootstrapping techniques based on 50,000 resamples (seed value = 3276459) were used to test moderated mediation models. These models assessed the indirect effect of recalled neighborhood risk on cold susceptibility through cytokine production as a function of perceived parental control (see Figure 1).

The index of moderated mediation is used to determine whether an indirect effect (i.e., the indirect effect of neighborhood risk on cold susceptibility through cytokine production) depends on the level of the moderator (i.e., parental control). If the confidence interval associated with the index of moderated mediation does not include 0, there is evidence that the mediation is moderated (53). For models with a significant index of moderated mediation, conditional indirect effects for the moderator were probed at the mean and ±1 SD.

Results

Of the original 213 PCS3 participants, complete data were available for 209 individuals. Of these individuals, 156 (74.6%) met criteria for infection, and 63 (30.1%) met criteria for a clinical illness. Table 1 presents sample characteristics and demographic information for the entire sample as well as just those who were infected and just those who met illness criteria. These rates are nearly identical to rates reported in previous publications from different subsets of these data; see references 2,42–45.

Table 1.

Sample characteristics

	Total Sample (n = 209)	Met Criteria for Infection (n = 156)	Met Criteria for Objective Cold Diagnosis (n = 63)
Age	30.2 (10.9)	30.3 (11.2)	32.8 (12.1)
Sex
Male	120 (57.4)	92 (59.0)	31 (49.2)
Female	89 (42.6)	64 (41.0)	32 (50.8)
Race/Ethnicity
White	139 (66.5)	107 (68.6)	39 (61.9)
Non-White	70 (33.5)	49 (31.4)	24 (38.1)
Years of Education	14.1 (1.9)	14.1 (2.0)	14.1 (2.1)
Season of Trial
Spring	67 (32.1)	51 (32.7)	11 (17.5)
Summer	85 (40.7)	68 (43.6)	31 (50.8)
Winter	57 (27.3)	37 (23.7)	20 (31.7)

Values are means (SD) or numbers (%).

Intercorrelations and descriptive statistics for the principal variables are presented in Table 2. Neighborhood risk and parental control were not correlated with proinflammatory cytokine production or any of the objective illness outcomes (r range: −.01 to −.10, ps > .10). However, cytokine production was significantly correlated with all illness outcomes.

Table 2.

Descriptive statistics and Pearson correlations among principal study variables

Variables	1	2	3	4	5	6
1. Neighborhood risk	--
2. Parental control	.159*	--
3. Proinflammatory cytokines	−.099	−.053	--
4. Objective cold diagnosis	−.026	−.036	.442***	--
5. Infection status	−.044	−.080	.257**	.383**	--
6. Illness signs	−.011	−.006	.399**	--	--	--
Mean (SD)	0.01 (0.79)	2.13 (0.56)	0.02 (2.58)	--	--	--
Range	−1.1 – 2.5	1 – 3.5	−2.1 – 11.5	0 – 1	0 – 1	0 – 1

^*p < .05,

^**p < .01,

^***p < .001.

Neighborhood Risk, Parental Control, and Nasal Proinflammatory Cytokine Production

Regression analyses (Table 3) revealed no conditional effects of neighborhood risk or parental control on cytokine production (see Table 3, Model 2). However, the Neighborhood Risk × Parental Control interaction predicted nasal proinflammatory cytokine production (see Table 3, Model 3). At values ≥ 2.48 on the parental control scale (represented by 29.7% of the sample, or 62 individuals), there was a significant negative association between recalled neighborhood risk and cytokine production. Figure 2 illustrates the simple slopes for mean ± 1 SD values of parental control. Please see supplemental materials for a scatterplot that displays the association between neighborhood risk and nasal pro-inflammatory cytokine production at high and low levels of parental control (Figure S1, Supplemental Digital Content).

Table 3.

Regression analyses predicting nasal proinflammatory cytokine production

Predictor	Model 1 b (SE)	Model 2 b (SE)	Model 3 b (SE)
Pre-Challenge RV39 Antibody Titers	−.546 (.436)	−.525 (.439)	−.501 (.433)
Age	−.020 (.017)	−.022 (.017)	−.020 (.017)
Sex	.277 (.358)	.262 (.360)	.151 (.358)
Race	.803 (.400)*	.725 (.416)	.690 (.410)
Education	.095 (.097)	.077 (.100)	.070 (.099)
BMI	.037 (.029)	.038 (.029)	.030 (.029)
Winter	.760 (.453)	.783 (.456)	.915 (.452)*
Summer	1.452 (.416)***	1.457 (.418)***	1.478 (.412)***
Neighborhood Risk		−.163 (.243)	1.934 (.848)*
Parental Control		−.010 (.054)	−.002 (.053)
Neighborhood Risk × Parental Control			−.168 (.065)*

^*p < .05,

^**p < .01,

^***p < .001.

Figure 2.

Neighborhood Risk × Parental Control interaction predicting nasal proinflammatory cytokine production. Simple slopes at the mean (b = −.204, p = .40, 95% CI [−.677, .972690]), + 1 SD (b = −.763, p = .023, 95% CI [−1.42, −.105]), and −1 SD (b = .354, p = .26, 95% CI [−.261, .970]) values of parental control are presented in the figure. The shaded regions represent the Johnson-Neyman regions of significance when the moderator is flipped (i.e., along the continuum of neighborhood risk).

Post-hoc exploration of the Johnson-Neyman regions of significance when the moderator is flipped from parental control to neighborhood risk indicated that the interaction was also significant at values below −1.07 and above 0.87 on the neighborhood risk composite, meaning that at very low (2.87% of the sample) and high (14.4% of the sample) levels of neighborhood risk, nasal proinflammatory cytokine production varied as a function of parental control. The interaction plot (Figure 2) indicates that at the lower end of neighborhood risk, individuals who reported greater (vs. lower) levels of parental control were more likely to mount a proinflammatory response to viral exposure. Conversely, at higher levels of neighborhood risk, individuals who reported greater (vs. lower) levels of parental control were less likely to mount a proinflammatory response.

Given evidence for a significant Neighborhood Risk × Parental Control interaction in the prediction of nasal proinflammatory cytokine production and evidence for associations between cytokine production and illness outcomes, we proceeded to conduct moderated mediation models to test the Recalled Neighborhood Risk × Parental Control interaction as a predictor of cytokine-mediated objective cold diagnosis, infection status, and illness criteria.

Neighborhood Risk × Parental Control to Predict Cytokine-Mediated Cold Outcomes

Clinical Cold Diagnosis

Analyses revealed a significant index of moderated mediation in the prediction of clinical cold diagnosis (Table 4). In other words, parental control moderated the indirect association between neighborhood risk and clinical cold diagnosis through proinflammatory cytokine production. At −1 SD and the mean of parental control, neighborhood risk was not associated with cytokine-mediated cold diagnoses. However, at +1 SD of parental control, neighborhood risk was negatively associated with cytokine-mediated cold diagnoses. Thus, individuals who reported greater parental control were more likely to be diagnosed with a clinical illness if they were from lower-risk neighborhoods than if they were from higher-risk neighborhoods.

Table 4.

Moderated mediation (Hayes model 7) regression results predicting cytokine-mediated cold outcomes

		Conditional Indirect Effects
Outcome	Index of Moderated Mediation (95% CI)	−1 SD Parental Control (95% CI)	Mean Parental Control (95% CI)	+1 SD Parental Control (95% CI)
Clinical Illness (n = 209)	−.073 (−.170, −.016)	.155 (−.104, .534)	−.089 (−.313, .114)	−.334 (−.760, −.065)
Infection Status (n = 209)	−.071 (−.172, −.015)	.150 (−.105, .535)	−.086 (−.318, .112)	−.323 (−.774, −.063)
Illness Signs (among infected individuals; n = 156)	−.074 (−.195, −.005)	.185 (−.117, .677)	−.070 (−.337, .196)	−.325 (−.865, .019)

Bolded coefficients represent statistically significant indices and simple effects.

Infection Status.

There was a significant index of moderated mediation in predicting the likelihood of viral infection (Table 4). Again, the conditional indirect effects at −1 SD and at the mean of parental control were non-significant. However, the conditional indirect effect for individuals who were +1 SD in parental control was significant. Like the above findings for clinical cold diagnoses, there was a negative association between recalled neighborhood risk and cytokine-mediated infection status at greater levels of parental control.

Objective Illness Criteria.

There was a significant index of moderated mediation in predicting the likelihood of meeting objective illness criteria among individuals who met infection criteria (n = 156; Table 4). However, the conditional indirect effects of neighborhood risk on objective illness criteria through cytokine production did not meet statistical significance at ±1 SD of parental control.

Alternative Explanations

To rule out the possibility that contemporaneous positive and negative affectivity and perceived stress influenced the results explained above, we re-analyzed our models including perceived stress and positive and negative affectivity as covariates. After including these covariates, the Neighborhood Risk × Parental Control interaction continued to predict nasal proinflammatory cytokine production (b = −.17, SE = .07, p = .013). Likewise, the moderated mediation findings were unchanged (clinical cold diagnosis: index of moderated mediation = −.082, 95% CI [−.197, −.017]), (infection status: index of moderated mediation = −.074, 95% CI [−.185, −.016]), (objective illness criteria: index of moderated mediation = −.081, 95% CI [−.222, −.010]).

Discussion

In the present study, we examined the extent to which perceived parental control shaped the association between recalled neighborhood risk and cytokine-mediated cold susceptibility. Although neighborhood risk and parental control were not directly associated with cytokine production or clinical illness outcomes, a significant Neighborhood Risk × Parental Control interaction emerged to predict nasal proinflammatory cytokine production and cytokine-mediated cold outcomes. Specifically, at greater (but not lower) levels of parental control, individuals who grew up in a riskier neighborhood had lower cytokine production and a lower likelihood of cytokine-mediated cold susceptibility relative to individuals who grew up in a lower-risk neighborhood. Stated differently, in the context of greater parental control, individuals who grew up in riskier neighborhoods might have experienced a potential immune advantage relative to individuals who grew up in low-risk neighborhoods.

Post hoc examination of these analyses revealed that among individuals who grew up in high-risk neighborhoods (approximately 14% of the sample), those who reported greater parental control throughout childhood were less likely to mount a proinflammatory response to viral exposure relative to those who reported lower parental control. In addition, among individuals who grew up in low-risk neighborhoods, greater parental control throughout childhood was associated with a greater likelihood of mounting a proinflammatory response to viral exposure relative to individuals who reported lower parental control (however, this was only significant among approximately 3% of the sample). These findings are consistent with the growing awareness that many parenting behaviors (such as parental control) cannot be viewed in the absence of the broader social context in which children are living (22,26,28–31). As such, it is worthwhile to consider the potential processes by which parental control might play a role in previously identified links between neighborhood settings and physical health (1–4).

In risky neighborhoods, parental control may represent greater vigilance on the part of parents to prevent dangerous exposures and behaviors among their children (24,25). In this context, parental control may shelter children from stressful and unsafe situations. Further, when parents practice greater levels of control in high-risk environments, youth might be less likely to engage in delinquent behaviors associated with neighborhood risk, such as smoking and drinking (1), that serve as gateways to substance use in adulthood. Thus, greater parental vigilance and control in high-risk neighborhoods may curb hazardous exposures and behaviors in childhood and adolescence, which in turn could shield individuals from vulnerabilities in defending against illnesses in adulthood.

In contrast, in low-risk settings, parental control may not serve to protect youths from danger because they are unlikely to encounter real threats in their environment. As a result, in these low-risk settings, greater control may be perceived as needlessly inhibiting children’s independence and autonomy (21–23). Not surprisingly, parental control in low-risk contexts has been linked to immediate and long-term psychological distress (27,34), both of which have implications for physical well-being. Extensive research has demonstrated that poor mental health and psychological distress serve as precursors for poorer self-rated health (54) and immune functioning (55) in adulthood, as well as a greater likelihood of engaging in poor health behaviors (55) and developing chronic physical disease (56).

This study expands knowledge of the ways in which healthy adults’ perceptions of childhood contexts and recalled parenting behaviors relate to illness risk in a disease model. More importantly, these findings pose broader conceptual implications for adult physical health. Although neighborhood risk and compromised immune function have been linked (10,12), the current study’s findings suggest that parental control may influence inflammatory processes associated with dangerous childhood environments. As such, parental control may also protect adults who were raised in high-risk neighborhoods from other diseases modulated by excessive inflammation, such as cardiovascular disease and diabetes (39,40).

In support of the hypothesis that parental control protects against inflammation-mediated health problems for individuals from risky backgrounds, previous research suggests that high-quality parenting is associated with better trajectories of inflammatory-mediated health outcomes for young adults from high-risk backgrounds (57,58). Furthermore, recalled nurturant parenting mitigates the association between low-SES environments and proinflammatory processes for young and middle-aged adults (59,60). Future research should explore whether there are unique benefits of parental control that extend beyond previously identified protective parenting characteristics. We note, however, there are likely limits to the extent parenting behaviors can mitigate the progression of poor health (e.g., parental control likely does not affect health outcomes that have a strong genetic basis).

Several study limitations will be important to address in future research. Childhood experiences were measured retrospectively, and biases in reports of recalled parental control and neighborhood risk are possible (e.g., reference 61). Although prior research has shown that individual differences in personality and current mood and life circumstances can lead to biased reconstructions of memories (61–64), including these factors as covariates did not attenuate our findings. Additionally, the measure of parental control asked participants to recall their parents’ behaviors over the first 15 years of life; as such, we were unable to address questions about whether parents’ controlling behaviors during particular periods in development (e.g., early childhood vs. adolescence) had an outsized influence on cold outcomes. One intriguing opportunity to address these limitations would be to leverage ongoing longitudinal studies for use in a new viral challenge investigation, which would allow for an examination of how prospective and retrospective assessments of childhood experiences and their precise developmental timing are related to exaggerated inflammatory processes in the context of a disease model.

Another potential limitation of the current study is the wide range of participant ages, which causes variability in the number of years participants had to reflect back on. This variability may have led to differences in memory accuracy for older versus younger participants. We note, however, that age was not associated with reports of parental control or neighborhood qualities (r range: .04–.10, p range: .15–.57). Therefore, our findings are not likely to be the result of generational differences in recall.

In addition, the analyses in the present study were conducted using an open-access dataset that has been used to address other research questions about psychosocial predictors of cold susceptibility. Long-term confidence in the reported results will require replicating the current pattern of findings in additional studies. Lastly, one final limitation to note is that although our analytic sample was well-powered to detect medium-sized conditional indirect effects, it was was only moderately powered to detect small-sized conditional indirect effects (65). As such, the null findings are potentially at risk of being Type II errors.

In summary, findings from the current study suggest the association between recalled neighborhood risk and cytokine-mediated cold susceptibility varies as a function of perceived parental control throughout childhood. As such, controlling parenting behaviors may potentially reduce subsequent health risk sequelae associated with high-risk neighborhoods in childhood and amplify health risk for individuals from low-risk settings. Future research should evaluate whether parental control protects against other cytokine-mediated illnesses for adults who were raised in high-risk contexts and begin to consider the mechanisms by which controlling parenting behaviors affect long-term physical health for individauals from low-risk contexts.

Glossary

BMI

body mass index

RV39

rhinovirus 39