|Home | About | Journals | Submit | Contact Us | Français|
The objective of this study was to identify salient parent and adolescent psychosocial factors related to somatic symptoms in adolescents. As part of a larger intervention study conducted in New York, 409 adolescents were recruited from 269 parents with HIV. A longitudinal model predicted adolescent somatization scores six years after baseline assessment. Adolescent somatic symptoms were assessed at baseline and at 3-month intervals for the first two years and then at 6-month intervals using the Brief Symptom Inventory. Baseline data from adolescents and parents were used to predict adolescent somatic symptoms. Variables related to increased adolescent somatic symptoms over six years included being younger and female; an increased number of adolescent medical hospitalizations; more stressful life events; adolescent perception of a highly rejecting parenting style; more parent-youth conflict; no experience of parental death; and parental distress over their own pain symptoms. Our findings extend the literature by virtue of the longitudinal design; inclusion of both parent and child variables in one statistical model; identification of study participants by their potentially stressful living condition rather than by disease or somatic symptom status; and inclusion of serious parental illness and death in the study.
Persistent somatic symptoms not explained by illness or tissue pathology are common and are often referred to as functional (Campo & Fritsch, 1994). Such symptoms are caused by altered physiological function rather than by structural abnormality (Hyman, 1999). Functional disorders are diagnosed with symptom-based criteria, not traditional medical tests. Examples of functional disorders include headaches, irritable bowel syndrome and non-epileptic seizures. There is interest in understanding early risk and protective factors for functional symptoms (Di Lorenzo et al., 2005).
The biopsychosocial model, positing that illness is the product of interacting biological, psychological and social systems, has been used to explain diseases and functional symptoms (Hyams & Hyman, 1998; Hyman, 1999; Zeltzer, Bursch, & Walco, 1997). For example, biological factors associated with functional abdominal pain include changes in intestinal wall sensory receptors, modulation of sensory transmissions, cortical perceptions, family history of pain and pain memories (Hyams & Hyman, 1998). Related psychological factors include temperament, increased focus on pain-related stimuli, emotional responses to pain, pain memories and efforts to cope with pain (Zeltzer et al., 1997). Social factors include family history of pain, parental responses to a child's pain and stressful life events (Walker, 1999). Central to this model is the role of stressful events and their impact on functioning (Tan, Tillisch, & Mayer, 2004). Ongoing daily stressors, such as a chronic family illness, have been found to be more important than single major stressors in provoking pain (Walker, Garber, Smith, Van Slyke, & Claar, 2001). While general and symptom-specific biological mechanisms are involved with the development of specific somatic symptoms, research suggests that important psychosocial variables are consistent across symptoms. For example, anxiety disorders and family stress are both commonly present in childhood abdominal pain and non-epileptic seizures (Walker & Greene, 1989; Wyllie, Glazer, Benbadis, Kotagal, & Wolgamuth, 1999). Most research has been conducted with individuals with a common symptom (e.g. headaches) rather than with those living in similar high stress situations (Walker et al., 2001; Walker & Greene, 1989; Wyllie et al., 1999).
Adolescents and young adults living with parents with HIV (PWH), provide an opportunity to examine risk factors for the development and maintenance of somatic symptoms. Families living with PWH have often endured stressors including poverty, residential instability, substance abuse, physical deterioration, emotional distress or early parental death (Rotheram-Borus, Stein, & Lin, 2001). In a previous study (Lester, Stein, & Bursch, 2003) we found that symptoms during adolescence persisted over 12 months and were predicted by female gender. This study also revealed that: (1) parental distress over pain predicted adolescent somatic symptoms at baseline and the 12 month follow-up, (2) adolescents who experienced their parents as highly rejecting and uncaring were more likely to report somatic symptoms at follow-up and (3) school problems and parental rejection correlated with somatic symptoms at baseline.
The current study re-examines these psychosocial predictors of somatic symptoms over six years and extends our research by examining additional predictors expected to increase symptoms due their potentially stressful nature, including: adolescent medical illness and hospitalizations (Pilowsky, Bassett, Begg, & Thomas, 1982), stressful life events (Walker, Garber, & Greene, 1993; Wyllie et al., 1999), adolescent sexual abuse (Wyllie et al., 1999), adolescent substance use, parent-child conflict (Walker et al., 1993), parental illness status and death (Wyllie et al., 1999), parental emotional distress (Garber, Zeman, & Walker, 1990; Wyllie et al., 1999) and substance use, and poor parental coping style (Garber et al., 1990). Our aim was to identify parent and adolescent psychosocial factors related to somatic symptoms within a longitudinal model among adolescents living with a seriously ill parent. Institutional Review Board approval was obtained.
The New York City Division of AIDS Services provides case management services to 95% of people with HIV/AIDS who qualify for assistance. Parents with HIV were selected between August 1993 and March 1995 from a consecutive series of clients dependent on governmental financial support. Of the 429 eligible PWH, 307 were recruited, 65 (15.1%) were untraceable, 46 (10.7%) refused and 11 (3%) were not recruited due to severe illness or incarceration. Thus, 71.6% (307) of eligible PWH were recruited.
After recruiting and consenting the PWH, recruitment of their adolescents required both parental and adolescent informed consent. Some PWH (n = 38) did not have custody of their children or did not allow them to participate. A total of 409 adolescents/young adults were recruited from 269 PWH (1.5 per family, range 1–5). The term ‘adolescent’ will be used to refer to the teens and young adult children of the parent with HIV.
Baseline measures were collected from parents. Adolescents were followed longitudinally starting between 1993 and 1998, with the final interview completed in 2002 (mean = 63.5 months). They were reassessed at 3-month intervals for two years and then at 6-month intervals. An ethnically diverse and bilingual team of interviewers conducted 2-hour individual assessments with participants in their homes. Retention was high, with an annual re-evaluation rate ranging from 79% to 94%.
The purpose of the larger study was to evaluate an intervention designed to reduce risk in children of PWH (see Rotheram-Borus et al., 2003 for results of the larger study). While the analyses described in this paper reflect a secondary investigation focused on factors influencing somatic symptoms over time, the intervention is described here for context. The intervention was not designed to reduce somatic symptoms.
At baseline, PWH and adolescents were randomly assigned to the intervention (n = 132 PWH; 203 youth) or standard care (n = 137 PWH; 206 youth). Participants received the intervention the first year after the baseline measures were administered. The intervention was delivered in three modules at a central location, with two sessions each day. Eight sessions were for parents and addressed illness disclosure, adjustment to disease status and parenting skills. Twelve sessions were for parents and adolescents, focused on custody planning, reduction of risk and parent-youth communication. Eight sessions were for bereaved youth and new caregivers after parental death, focused on grieving, setting life goals and establishing a positive relationship with the guardian. Similar numbers of PWH died in the intervention (n = 63) and standard care (n = 70) conditions. Intervention manuals are available at: http://chipts.ucla.edu/interventions/manuals/index.html
We fit longitudinal models with SAS proc mixed (SAS Institute, Cary, NC, US) to the BSI somatization scale, which was log transformed after adding a small constant to adjust for skewness. The analysis generally followed that described by Weiss (2005). We used the autoregressive moving average (ARMA, 1,1) covariance model for the correlation of observations from the same adolescent and included a random parent effect to model the correlation between adolescents with the same parent.
We allowed for a bent line time trend that was linear with changes in slope at months 18 and 36. We first fit separate models with the adolescent or parent predictors one at a time, always including the bent line time trend and adolescent demographic characteristics. For the model with intervention as a predictor, we also allowed for separate bent lines in the intervention and control groups. We then fit a bent line time trend model, with adolescent demographic variables, and all other developmental and family predictors with significant associations with the outcome from the separate models. To produce a more parsimonious final model, non-significant predictors were deleted.
Descriptive statistics are summarized in Table 1. At baseline, the adolescents were 11–21 years old. Their primary caretakers (PWH) were 25–71 years old, including mothers (71.7%), fathers (16.7%), grandparents (3.3%), aunts (1.9%), uncles (0.4%) and step/ foster parents (0.4%). Relationship information was missing for 5.6%. The preponderance of mothers in the sample reflects social norms for mothers to be the custodial parent or primary parental caretaker. Correlations of all continuous variables are summarized in Table 2.
Baseline somatic symptom levels were not significantly different between the adolescents in intervention and standard care conditions (p = 0.11), mean baseline of 0.42 (SD = 0.57; range, 0–3.43). Lower than a standardization sample (Derogatis, 1993), only 2% of boys and 11% of girls were in the clinical range. In the first 18 months, symptoms of adolescents in the intervention condition decreased faster than the adolescents receiving standard care (slope difference = −0.013; p = 0.03). After 18 months, there were no differences in the rates of change of symptoms between the two conditions. The mean symptom score ranged from 0.25–0.28 (SD = 0.48–0.55; ranges = 0–3.71) at each of six subsequent assessment points.
As shown in Table 3, adolescent variables predictive of increased somatic symptoms over six years included: being female (p < 0.0001), being Latino (p = 0.03), increased medical diagnoses (p = 0.002), hospitalizations (p = 0.0003), school problems (p = 0.02), parent-youth conflict (p = 0.0008) and stressful life events (p < 0.0001). Additionally, adolescent substance use including alcohol (p = 0.0003), marijuana (p = 0.010) and hard drugs (p = 0.0003), was related to increased symptoms over time. Variables related to fewer symptoms over a 6-year period included: adolescent perception of parental caring (p < 0.0001) and parental death (p = 0.03). Adolescent age, sexual abuse and parental overprotection were not related to symptoms.
Also displayed in Table 3, parent variables related to increased somatic symptoms over a 6-year period included parental distress over their own pain symptoms (p = 0.02), parental global emotional distress (p = 0.003) and negative coping (p = 0.03). Parental disease status and substance use were not related to youth symptoms.
In the final model (Table 4), adolescent variables related to increased somatic symptoms over a 6-year period included being younger and female (p = 0.04 and p < 0.0001, respectively). Ethnicity was a marginally significant predictor (p = 0.06), with Latino youth tending to report more symptoms. Increased symptoms were also predicted by more hospitalizations (p = 0.02), stressful life events (p < 0.0001), less parental care/highly rejecting (p = 0.0002), more parent-youth conflict (p = 0.03) and no experience of parental death (p = 0.04). The school problems variable was a marginally significant predictor for symptoms (p = 0.07). Predictors that dropped out in the final model included number of medical diagnoses and substance use.
Once both distress over pain and parental emotional distress were included in the longitudinal model simultaneously, neither of them was a significant predictor. We kept parental distress over pain symptoms in the final model (p = 0.03) because it can be more easily targeted for intervention. Also dropped in the final model was negative coping.
Rates of somatic symptoms for these high-risk youth were similar to a non-clinical sample (Derogatis, 1993), revealing that having an HIV-positive parent is not sufficient to cause high levels of symptoms. Other than the expected demographic variables, chronic stress, specifically family distress, predicted symptoms. Within the biopsychosocial model, persistent stressors result in illness and somatic symptoms due to a disturbance in the internal response systems, balance and homeostasis, such as the allostatic load and/or emotional motor system (McEwen & Seeman, 1999).
While it seems counter-intuitive that experiencing parental death predicts fewer somatic symptoms, exposure to medical problems and the anticipation of parental death can be more traumatizing than parental death (Saldinger, Cain, & Porterfield, 2003). The current sample of adolescents was previously found to report less emotional distress following the parental death compared to before the parental loss (Rotheram-Borus, Weiss, Alber, & Lester, 2005). Over half of PWH made custody plans, enabling a newcaretaker to care for the child (Rotheram-Borus, Lester, Wang, & Shen, 2004), often providing a more stable household than the PWH (Rotheram-Borus et al., 2002). The finding that parental distress over their own pain predicted increased symptoms better than severity of parental illness supports the idea that the parent's response to physical illness rather than the illness itself is central to the child's functioning.
The number of adolescent medical hospitalizations, but not the number of medical diagnoses, predicted symptoms. This is consistent with previous research that painful or traumatic childhood hospitalization can increase somatic sensitivity over time (Taddio, Katz, Ilersich, & Koren, 1997). Additionally, hospitalizations in childhood have previously been related to maternal healthcare utilization, family conflict and maternal worry, suggesting increased medical hospitalizations may be a proxy measure of increased family distress (Janicke, Finney, & Riley, 2001; Riley et al., 1993). Childhood hospitalizations may be related to severity of adolescent medical disorder; we do not have illness severity data to allow this analysis.
Adolescents who experienced their parents as highly rejecting/less caring reported increased symptoms over time. A perception of greater parental emotional responsiveness appeared to be a protective factor. Our findings are not consistent with clinical observations that parents of children with more symptoms are overprotective (Minuchin et al., 1975). It may be that those parents who are high utilizers of medical care are more overprotective than parents who normalize somatic complaints in their children.
School problems approached significance related to symptoms. The relationship between school problems and disability among those with chronic pain has been identified in the literature, with academic competence being an important factor (Claar, Walker, & Smith, 1999). Had our study measured self-competence and functioning, we expect the importance of school problems would have emerged.
While the intervention taught active coping strategies (e.g. problem solving, emotional expression, emotional modulation, decision making), resulting in a more rapid decline in somatic symptoms during the first 18 months, the intervention youth did not have fewer symptoms when compared with the control group. Research on childhood pain has revealed that active coping strategies (e.g. doing something to get rid of the pain, such as asking someone for help or trying to figure out howto make it go away) have been inconsistently related to pain severity, while accommodative coping (e.g. distraction, acceptance, positive thinking, cognitive restructuring) has been related to less pain (Compas & Thomsen, 1999). Other implications for clinical intervention include the potential value of targeting parental distress regarding pain and transmission of distress to adolescents via parent-child conflict and rejecting interactions.
Strengths of the study include the longitudinal design, collection of data from both adolescents and caregivers, identification of participants by living condition rather than by symptom and inclusion of serious parental illness and death as variables. Limitations of the study provide directions for future research. Future research would benefit from inclusion of a low-risk comparison group, wider variation in socioeconomic backgrounds, biological data, assessments of cognitive competence and functional status, measurement of time since parental death and longitudinal measurement of potentially key measures, such as substance use, hospitalizations and stressful life events.
This paper was completed with the support of National Institute of Mental Health (NIMH) grants K-23 MH02050-03; MH068194; P30MH-58107; and R01MH63779.
Publisher's Disclaimer: Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.