The Explanatory Role of Insomnia in the Relationship between Pain Intensity and Posttraumatic Stress Symptom Severity among Trauma Exposed Latinos in a Federally Qualified Health Center

Abstract

Latinos, one of the fastest growing populations in the United States, suffer from high rates of post-traumatic stress symptoms (PTS) and its clinical correlates (e.g., disability). Although research suggests the experience of pain is closely related to PTS among trauma-exposed groups, there has been little exploration of the processes that may link pain intensity to greater PTS among trauma-exposed Latinos. The current study explored insomnia, a common problem associated with both pain intensity and PTS, as a mechanism in the association between pain intensity and PTS among trauma-exposed Latinos (N = 208, M_age = 39.39 years, SD = 11.48) attending a Federally Qualified Health Center. Results indicated that insomnia partially explained the relationship between pain intensity and PTS total score (B = 0.25, 95% CI [0.12, 0.43]), as well as re-experiencing (B = 0.09, 95% CI [0.04, 0.17]), avoidance (B = 0.09, 95% CI [0.04, 0.17]), and arousal symptoms (B = 0.10, 95% CI [0.04, 0.17]). Future work is needed to explore the extent to which insomnia accounts for relations between pain and PTS using longitudinal designs to further clarify theoretical health disparity models involving these comorbid conditions.

Latinos represent a large and fast-growing population in the United States (US), and it is estimated that Latinos will make up 30% of the population by 2060 [1]. However, little scientific attention has focused on mental health problems among Latinos, despite evidence that this group experiences significant mental health concerns [2,3]. Research suggests that Latinos report elevated rates of trauma exposure, posttraumatic stress symptoms (PTS), and greater conditional risk of developing posttraumatic stress disorder (PTSD), relative to other racial/ethnic groups [4,5]. Such elevated rates of PTS and PTSD are of public health importance given documented associations between PTS and an array of negative health outcomes in the general population [6], and in the Latino community specifically [7].

The experience of pain is a clinically-relevant factor in the context of trauma-related mental health problems, and pain is among the most common health problems endorsed by Latinos [8]. Latinos also are more apt to demonstrate poor pain management, experience longer wait times for pain care [9], and seek fewer medical visits for pain [10,11]. Additionally, within the Latino community, some work suggests that pain may be viewed as a “common problem” that does not (typically) warrant treatment [12]. Initial research has documented associations between pain intensity and more severe mental health problems among Latinos [13]. In terms of PTS specifically, one study showed that Latinos living in the US who were exposed to political violence reported greater pain intensity than those who were not exposed to political violence [7]. Another study found that chronic pain was the most common health complaint among Latinas living in a domestic violence shelter [14]. Although extant research on pain experience among Latinos in the context of trauma is presently limited in scope, available studies suggest that there is scientific and clinical merit in examining the association between pain intensity and PTS symptoms in this population. Specifically, it is clinically-important to identify processes that may mutually reinforce the relations between pain intensity and PTS symptoms, to ultimately inform the development of tailored interventions among this understudied group.

Insomnia is one promising integrative health-based construct that may bridge gaps in our understanding of relations between pain intensity and PTS. Insomnia reflects difficulty initiating or maintaining sleep, and/or unsatisfying sleep, which is related to later distress and functional impairment [15]. Previous research suggests Latinos are at increased risk for sleep problems, compared to non-Latino whites, after controlling for demographic factors [16]. Further, examining a range of populations there are robust and consistent associations between pain intensity and insomnia [17]. Among chronic pain samples, insomnia rates are as high as 70% [18–21]. There is also broad-based evidence that sleep influences emotional experience [22]. Indeed, sleep loss increases the occurrence of negative emotions, reduces the occurrence of positive emotions, and alters the ways in which individuals understand, express, and modify their emotions [23,24]. At a neurobiological level, decreased connectivity between frontal brain regions (e.g., medial pre frontal cortex) and emotion-based structures (e.g., amygdala) following periods of sleep deprivation is suggestive of broad-based problems with regulatory control and emotional processing [25,26]. In terms of PTS, the decreased connectivity between the frontal brain regions and the amygdala observed following insomnia may prevent individuals from decreasing the emotional content of traumatic memories, thereby contributing to amygdala hyperactivity and more PTS [27].

Limited work has focused on the relationship between insomnia and PTS among trauma-exposed Latinos. Theoretically, insomnia may be one potential mechanism linking the experience of pain intensity and PTS among trauma-exposed Latinos. Although research examining insomnia in the relationship between pain intensity and PTS is limited, of the available work among non-Latinos, insomnia was found in one study to mediate the relationship between PTS and pain intensity among Veterans [28]. However, research has not yet applied this model to Latinos, a population that reports among the highest levels of pain and PTS of all racial/ethnic minority groups [4,8]. A formative next step is to evaluate whether insomnia explains the association between pain intensity and PTSD symptoms among a Latino sample exposed to trauma.

Together, the current study aimed to examine insomnia in the relation between pain intensity and PTS in a sample of trauma-exposed Latinos attending a Federally Qualified Health Center (FQHC). FQHC’s serve the majority of primary healthcare needs of predominantly low-income Latinos and other groups [29]. Additionally, FQHCs represent an optimal portal for mental health research (Vega & Lopez, 2001), especially given the stigma among the Latino population associated with attending mental health-specific clinics (Jimenez, Bartels, Cardenas, & Alegria, 2013). We hypothesized that among trauma-exposed Latinos, insomnia would significantly explain the relationship between pain intensity and PTS symptoms.

Method

Participants

Participants (N = 208) were trauma exposed Latinos recruited from an FQHC located in an urban southwestern community, reporting mild to severe current pain, as determined by the Graded Chronic Pain Scale [30]. The inclusion criteria for the current study consisted of the ability to read, write and communicate in Spanish, and being between the ages of 18 and 64. Participants were excluded if they exhibited limited mental competency and/or inability to provide informed, voluntary, written consent, or if they endorsed current or past psychotic-spectrum symptoms via structured interview screening.

The sample was primarily female (88.0%) with an average age of 39.39 years (SD = 11.48). Nearly all participants (92.8%) reported Spanish as their first language, and self-identified with the following ethnic subgroups: 52.2% “Mexican American,” 30.5% “Central American,” 6.9% “American/Born in the U.S.,” 6.9% “South American,” 1.5% “Cuban American,” 0.5% “Puerto Rican,” and 1.5% as “Other.”

In terms of marital status, 38.1% of participants were married, 31.2% were single, 16.8% were living with a partner, 11.4% were divorced, and 2.4% were widowed. Thirty-eight percent of the sample was unemployed with 26.8% employed full-time, 12.3% working half-time (~20 hours per week), and 11.2% working part-time (less than 20 hours per week). Over half (58.8%) reported earning less than $14,999, and 42.2% reported earning less than $34,999 annually. Participants indicated visiting the primary care facility for the following reasons: family medicine (28.4%), dental (10.6%), psychiatric/psychological (7.7%), lab test or physical exam (7.2%), or accompanying someone/other reasons (34.1%); and 12.0% did not disclose a reason. In terms of psychiatric diagnoses, the most commonly endorsed disorders were Major Depressive Disorder (30.3%) and PTSD (17.8%). For a full list of psychiatric diagnoses, see Table 1. See Table 2 for the types of traumas endorsed by the participants.

Table 1.

Psychiatric Diagnoses

Psychiatric Diagnosis	% (n)
Major Depressive Disorder	30.3% (63)
Post-Traumatic Stress Disorder	17.8% (37)
Agoraphobia	13.0% (27)
Panic Disorder	7.7% (16)
Social Anxiety Disorder	7.7% (16)
Dysthymia	6.3% (14)
Generalized Anxiety Disorder	6.3% (13)
Obsessive-Compulsive Disorder	5.3% (11)
Bipolar Disorder	1.9% (4)

All diagnoses determined by clinical interview with the MINI [31]

Table 2.

Traumatic Event Exposure [as listed on the PDS; ,33]

Potentially Traumatic Event	% (n)
Natural disaster	43.8% (91)
Serious accident, fire, or explosion	35.1% (73)
Sexual contact in childhood/adolescence with someone 5 years older	26.9% (56)
Non-sexual assault by someone familiar	26.9% (56)
Non-sexual assault by a stranger	24.5% (51)
Sexual assault by someone familiar	22.6% (47)
Other traumatic event	17.8% (37)
Sexual assault by a stranger	11.5% (24)
Life-threatening illness	7.7% (16)
Imprisonment	7.2% (15)
Military combat or war zone	5.8% (12)
Torture	5.3% (11)

^*All traumas as listed on the PDS [33]

Measures

Demographics.

Participants reported on demographic factors including age, gender, ethnic origin, education level, years in the US, income, employment status (coded 0 “unemployed” and 1 “employed”), marital status (coded 0 “no partner” and 1 “partner”), and reason for visiting the primary care clinic.

M.I.N.I. International Neuropsychiatric Interview 6.0 [MINI; 31].

Diagnostic assessments were performed using the MINI. Interviews were administered by trained, Spanish-speaking staff, under the supervision of an independent doctoral-level rater. The MINI has demonstrated satisfactory inter-rater reliability, test-retest reliability and validity (Sheehan et al., 1997), and has been deemed applicable for use in research settings (Lecrubier et al., 1997). Approximately 12% of randomly selected interviews were checked for accuracy, with no cases of diagnostic disagreement noted.

Graded Chronic Pain Scale [GCPS; 30].

The GCPS is an 8-item self-report measure assessing the impact of chronic pain on daily life. Participants are asked to rate the number of days they were disabled by pain in the last 3 months, as well as the intensity of their pain and the disability due to pain over the last 3 months, using a 0 (No pain) to 10 (Pain as bad as it can be) scale. The measure yields 2 subscales, pain disability and pain intensity. The GCPS has been used among Spanish-speaking samples in past work [32]. The pain intensity subscale was used as the predictor variable and showed adequate internal consistency (α = .76).

Posttraumatic Diagnostic Scale [PDS; 33].

The PDS is a 49-item self-report measure designed to assess the severity of posttraumatic stress disorder (PTSD) symptoms. The scale is comprised of 4 sections: 1) a checklist of traumatic events the individual has experienced, 2) identification of the most distressing traumatic event, 3) an assessment of PTSD symptoms, and 4) an assessment of the interference of symptoms. The assessment of PTSD symptoms yields a total score and three subscales (re-experiencing, avoidance, and arousal). The PDS has been used among Spanish-speaking samples in past work [34]. The total score and three subscales were examined as dependent variables, and showed good internal consistency (re-experiencing α = 0.90, Avoidance α = 0.88 and Arousalα = 0.88).

Inventory of Depression and Anxiety Symptoms [IDAS; 35].

The IDAS is a 64-item self-report measure designed to assess distinct affect symptom dimensions within the past two weeks, using a 5-point Likert scale ranging from “not at all” to “extremely”. The IDAS contains 10 symptom subscales for suicidality, lassitude, ill temper, well-being, insomnia, appetite loss, appetite gain, panic, social anxiety, and traumatic intrusions, and two broad subscales for general depression and dysphoria. The IDAS subscales show strong internal consistency, convergent and discriminant validity with psychiatric diagnoses and self-report measures; and short-term retest reliability with both community, and psychiatric patient samples [35,36]. The IDAS has been used among Spanish-speaking samples in past work [37]. The present study examined the insomnia subscale (e.g. “I slept less than usual”) as the explanatory variable in the statistical model, and it evinced excellent internal consistency (α= 0.90).

Procedure

Participants for the study attended an FQHC primary health care clinic. Individuals provided informed written consent (in Spanish) prior to participation in study procedures. After providing consent, participants completed a diagnostic interview (conducted in Spanish by staff under the supervision of a licensed clinical psychologist) and completed self-report measures. Participants were compensated $20. The University of Houston Institutional Review Board approved the study protocol.

Data Analytic Plan

Analyses were conducted using SPSS version 24. First, bivariate correlations among the variables were examined. Then, analyses were conducted using the PROCESS macro [38]. Pain intensity was included as the predictor variable, IDAS insomnia was included as the indirect variable, and PDS total and subscale scores as dependent measures. Covariates included number of traumatic events, age, gender, years in the US, education level, marital status, and employment status, as these are commonly used covariates in the sampled population [e.g. 39]. Both direct and total effects for each model were reported. To detect the significance of the indirect effects, bootstrapping with 10,000 bootstrap re-samplings was conducted. Bootstrapping estimates the sampling distribution of an estimator based on re-sampling with replacement from the dataset, which creates an empirically generated sampling distribution [40]. A bootstrapped confidence interval that does not include zero indicates a statistically significant indirect effect [41]. Additionally, to test the specificity of the models for each dependent variable, alternative models were run, where the predictor (X) and indirect explanatory variables (M) were reversed [41]. Effect sizes were calculated using percent mediation (PM), which indicates the percent of the total effect that can be accounted for by the indirect effect [42,43].

Results

Bivariate Correlations

For full correlation results, see Table 3. Insomnia symptoms were significantly correlated with both pain intensity, as well as each of the PTSD dependent measures (r’s 0.41–0.62). Additionally, pain intensity was significantly correlated with each of the PTSD dependent measures (r’s 0.30–0.38). Number of traumas was significantly positively related to insomnia and PTSD symptoms (r = 0.29).

Table 3.

Bivariate Correlations and Descriptive Statistics

	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.	11.	12.	Mean (SD) or %
1. N-traumas	1												2.35 (1.84)
2. Marital Status	.140*	1											45 %
3. Employment	−0.030	−0.062	1										49.7 %
4. Age	0.103	0.054	0.001	1									39.39 (11.48)
5. Gender (female)	0.010	0.029	.186*	−0.104	1								88 %
6. Years in US	0.042	0.073	−0.052	.414**	−.188*	1							15.36 (8.79)
7. Education Years	−0.079	−0.041	−0.099	−.159*	−0.018	−0.116	1						10.41 (3.68)
8. Insomnia	.293**	.280**	0.147	.201**	0.027	−0.006	−0.092	1					13.71 (7.03)
9. Pain Intensity	0.071	0.069	0.095	0.123	0.057	0.085	−.242**	.411**	1				11.70 (6.92)
10. Re-experiencing	.333**	0.128	0.039	−0.001	0.113	0.016	−0.011	.551**	.300**	1			8.49 (4.28)
11. Avoidance	.455**	0.142	0.053	−0.045	0.090	−0.010	−0.089	.567**	.378**	.754**	1		11.65 (5.70)
12. Arousal	.360**	0.135	0.064	−0.001	0.056	−0.003	−0.081	.577**	.331**	.748**	.818**	1	8.73 (4.45)
13. PTS Total	.436**	.159*	0.057	−0.020	0.092	0.008	−0.064	.622**	.370**	.893**	.944**	.924**	25.42 (11.73)

* indicates a p values less than 0.05, and ** indicates a p value less than 0.01.

Insomnia as measured with the IDAS [35], pain intensity as measured on the Graded Chronic Pain Scale [30], and re-experiencing, avoidance, arousal, and PTS total score as measured on the PDS [33]

Primary Analyses

PDS Total Score.

For PDS total score, there was a significant total effect of pain intensity (B = 0.56, SE = 0.13, p = 0.001). Additionally, there was a significant indirect effect of pain intensity, through insomnia symptom severity, on PDS total score (B = 0.25, SE = 0.08, Bootstrapped 95% CI [0.12, 0.43], PM = 0.45). After controlling for the effect of insomnia, there remained a significant direct effect of pain intensity on PDS total score (B = 0.31, SE = 0.13, p = 0.02). The reverse model (i.e. switching the predictor and the mediator variables), also yielded a significant indirect effect of insomnia symptoms, through pain intensity, on PDS total score (B = 0.12, SE = 0.06, Bootstrapped 95% CI [0.04, 0.27], PM = 0.15).

Re-Experiencing Symptoms.

In predicting re-experiencing symptoms, there was a significant total effect of pain intensity (B = 0.17, SE = 0.14, p = 0.003). Additionally, there was a significant indirect effect of pain intensity, through insomnia symptoms, on re-experiencing symptoms (B = 0.09, SE = 0.03, Bootstrapped 95% CI [0.04, 0.17], PM = 0.54). After accounting for the effect of insomnia symptoms, there was no significant direct effect of pain intensity on re-experiencing symptoms (B = 0.07, SE = 0.05, p > 0.05). There was no significant indirect effect of insomnia symptoms, through pain intensity, on re-experiencing symptoms in the reverse model (B = 0.03, SE = 0.02, Bootstrapped 95% CI [−0.001, 0.08]; see Table 2).

Avoidance Symptoms.

For avoidance symptoms, there was a significant total effect of pain intensity (B = 0.29, SE = 0.06, p < 0.001). Additionally, there was a significant indirect effect of pain intensity, through insomnia symptoms, on avoidance symptoms (B = 0.09, SE = 0.03, Bootstrapped 95% CI [0.04, 0.17], PM = 0.31). After controlling for the effects of insomnia, there was a significant direct effect of pain intensity on avoidance symptoms (B = 0.20, SE = 0.07, p = 0.003). The reverse model also yielded a significant indirect effect of insomnia symptoms, through pain intensity, on avoidance symptoms (B = 0.08, SE = 0.03, Bootstrapped 95% CI [0.03, 0.16], PM = 0.24; see Table 2).

Arousal Symptoms.

In predicting arousal symptoms, there was a significant total effect of pain intensity (B = 0.18, SE = 0.05, p = 0.001). Further, there was a significant indirect effect of pain intensity, through insomnia symptoms, on arousal symptoms (B = 0.10, SE = 0.03, Bootstrapped 95% CI [0.04, 0.17], PM = 0.54). After accounting for the variance of insomnia symptoms, there was no significant direct effect of pain intensity on arousal symptoms (B = 0.08, SE = 0.05, p > 0.05). There was no significant indirect effect of insomnia symptoms, through pain intensity, on arousal symptoms in the reverse model (B = 0.03, SE = 0.02, Bootstrapped 95% CI [−0.002, 0.09]; see Table 2).

Discussion

The present study examined the role of insomnia in associations between pain intensity and PTS among trauma-exposed Latinos attending an FQHC. Results provide initial evidence that insomnia significantly explained, in part, the relationship between pain intensity and PDS total score, re-experiencing, avoidance, and arousal symptoms. Specifically, among trauma-exposed Latinos, higher levels of pain were related to greater insomnia, which in turn, was related to greater degrees of PTS. The observed indirect effects were evident after accounting for theoretically relevant covariates including number of traumatic events, age, gender, marital status, years of education, employment, and number of years in the US. These findings are broadly in line with previous non-Latino research suggesting that insomnia explains, in part, the relationship between PTSD symptoms and pain [28], and uniquely extends such work to Latinos in an FQHC [4,8]. The findings are also broadly consistent with the perspective that increasing pain intensity is related to more severe insomnia, which in turn is related to more severe PTS among trauma-exposed Latinos.

Notably, tests of the reverse mediation models indicated the potential for bi-directional relations between subsets of the studied variables. Specifically, the reverse mediation models were significant for PDS total score, as well as for avoidance symptoms, suggesting that a dynamic relationship may exist between pain and insomnia in terms of global and avoidance aspects of PTS [44,45]. To more fully explore the nature of relations among these variables over time, future prospective modeling of the temporal ordering of insomnia, pain, and PTSD is warranted. The current findings suggest that insomnia may increase pain intensity, which in turn, may exacerbate global and avoidance cluster PTS symptoms.

Although not a primary study objective, the rates and types of traumatic events and PTSD detected in the current sample are clinically noteworthy. Indeed, almost 50% of the sample reported experiencing a natural disaster, about a third reported experiencing a serious accident, fire, or explosion, and about a quarter of the sample reported experiencing assault, both non-sexual and sexual. It is possible that certain types of traumatic events may be more likely to be associated with pain and/or insomnia. Some previous research suggests that the type of trauma individuals experience may differentially impact emotional and cognitive problems [46,47]. Additionally, 17.8% of the sample met DSM criteria for PTSD. This is clinically important, as the clear majority of the present sample was not seeking treatment for mental health problems. Thus, trauma exposure and PTSD were highly prevalent among this sample of Latinos in a FQHC. This finding highlights the importance of assessing for traumatic event exposure and PTSD in routine medical care for the Latino population.

Results from the current study may also contribute to the development of specialized intervention strategies for trauma-exposed Latinos. Cognitive behavioral therapy for insomnia is one evidenced-based treatment for individuals with sleep difficulties [48]. Studies examining cognitive behavioral therapy for insomnia secondary to pain among non-Latino samples have found that individuals who receive cognitive-behavioral therapy for insomnia were able to improve their sleep, relative to controls [e.g., 49]. Other work has found that cognitive behavioral therapy for insomnia among chronic pain patients improves sleep and decreases pain [45]. Collectively, the current findings and the extant literature suggest that cognitive behavioral interventions may be useful for reducing PTS in the context of pain and insomnia among trauma-exposed Latinos. Future work should examine the feasibility of addressing sleep problems in the context of an integrated, culturally-informed behavioral framework in a primary care facility, among trauma-exposed Latinos with co-occurring pain.

The current study has some limitations that warrant comment. First, the data collected were cross-sectional, prohibiting causal or temporal claims to be made. Additionally, the study did not explicitly aim to recruit Latinos experiencing pain, and therefore, it is possible that the findings may not generalize to individuals with more severe or chronic pain. Future research should test similar models among individuals experiencing chronic or recurring pain. Second, the current sample was predominantly female. Future research should replicate and extend these findings to samples with a more equitable gender distribution. Third, there is evidence that rates of psychopathology differ by country of origin [50]. Future studies are needed to examine the potential moderating role of country of origin in the relationship between pain and PTS through insomnia. Fourth, as many of the key variables were assessed via self-report, there is a possibility that the observed relations were in part a function of shared method variance. Future research would benefit by employing a multi-method assessment approach to cross-index the nature of the relations observed in the current report. In particular, utilization of an objective measure of sleep (e.g., actigraphy) would be helpful to cross-index actual sleep parameters [51]. Fifth, we focused on PTS symptoms as primary dependent measures. It is also possible that more culturally-specific processes may be related to pain-insomnia relations. For example, future work may benefit from exploring the explanatory role of insomnia in the relation between pain intensity and acculturative stress, or discrimination, among trauma-exposed Latinos. Finally, the instrument used to assess PTS symptoms (i.e., PDS) does not capture the frequency of exposure to each type of potentially traumatic event, or stressors more frequently experienced by Latinos, such as those related to migration; family separation; and community and political violence in communities of origin. Future research may therefore benefit from using clinical interviews for PTSD to capture culturally-relevant aspects of trauma among this group.

Overall, the current study represents an initial investigation into the role of insomnia in the relationship between pain intensity and PTS among a sample of trauma exposed Latinos. Results suggest consistent associations between pain intensity, insomnia, and PTS, although bi-directional relations may also be evident. Future work is needed to explore the extent to which insomnia accounts for relations between pain and PTS using longitudinal designs to further clarify theoretical health disparity models involving these comorbid conditions.

Table 4.

Mediation models

Y	Model	B	SE	t	p	LLCI	ULCI
1	Pain Intensity→Insomnia (a1)	0.35	0.08	4.25	<0.001	0.19	0.51
	Insomnia→Re-experiencing (b1)	0.27	0.06	4.87	<0.001	0.16	0.38
	Pain intensity→Re-experiencing (c)	0.17	0.05	3.29	0.001	0.07	0.28
	Pain intensity→Re-experiencing (c’)	0.08	0.05	1.53	>0.05	−0.02	0.18
	Pain intensity→Insomnia→Re-experiencing (a1*b)	0.09	0.03	-	-	0.04	0.17
2	Insomnia→Avoidance (b1)	0.25	0.07	3.58	0.001	0.11	0.39
	Pain intensity→Avoidance (c)	0.29	0.06	4.47	<0.001	0.16	0.41
	Pain intensity→Avoidance (c’)	0.20	0.06	3.02	0.003	0.07	0.33
	Pain intensity→Insomnia→Avoidance (a1*b)	0.09	0.03	-	-	0.04	0.17
3	Insomnia→Arousal (b1)	0.29	0.06	5.03	<0.001	0.17	0.40
	Pain Intensity→ Arousal (c)	0.18	0.05	3.24	0.001	0.07	0.29
	Pain Intensity→ Arousal (c’)	0.08	0.05	1.53	>0.05	−0.02	0.19
	Pain Intensity→ Insomnia→Arousal (a1*b)	0.10	0.03	-	-	0.04	0.17

Note. a Effects of X on M; b effects of M on Y; c total effect of X on Y; c’ direct effect of X on Yi controlling for M; Path a is consistent in all models; therefore, it presented only in model 1. LLCI lower bound of a 95% confidence interval; ULCI upper bound; → affects. The indirect effect (a*b) is the product of path a and path b. Bolded values are statistically significant at p < 0.05. Age, education, race/ethnicity, sexual orientation, and relationship status were included as covariates