Journal of Immigrant and Minority Health

, Volume 17, Issue 5, pp 1444–1450

Does Integrated Care Affect Healthcare Utilization in Multi-problem Refugees?

  • Carol C. White
  • Craig A. Solid
  • James S. Hodges
  • Deborah H. Boehm
Original Paper

DOI: 10.1007/s10903-014-0088-6

Cite this article as:
White, C.C., Solid, C.A., Hodges, J.S. et al. J Immigrant Minority Health (2015) 17: 1444. doi:10.1007/s10903-014-0088-6


A history of trauma is common in refugee populations and appropriate treatment is frequently avoided. Using a convenience sample of 64 patients in a Somali primary care clinic, a culture and trauma specific intervention was developed to address retention into appropriate treatment. One goal of the intervention was to improve the rate of engagement in psychotherapy after a mental health referral and to test the effect of psychotherapy on health care utilization using a staged primary care clinical tool. Forty-eight percent of patients given a mental health referral engaged in psychotherapy. Patients engaging in psychotherapy had higher baseline utilization and over 12 months trended towards less emergency room use and more primary care. Our findings suggest that the intervention improved referral and retention in mental health therapy for East African refugee women.


Refugees Political trauma Integrated care Special populations Primary care 


US refugees have a high likelihood of a history of political trauma, having often experienced war, detainment, torture, forced migration, death or separation from family members, sexual violence and exile. The estimated prevalence of a history of torture in an East African Somali and Oromo (Ethiopian) population in Minnesota is between 29 and 69 % [1]. Traumatized refugees experience a high level of physical and psychological effects due to the trauma [1, 2] and trauma histories often go undetected. Refugees, if they seek Western healthcare services, are likely to present in urban medical clinics or emergency rooms with a large number of complaints, mostly somatic in nature. Patients may present with varied and complex symptoms including headaches, abdominal pains, sleep difficulties, traumatic brain injury, body aches and pains, and injuries to eyes, ears and mouth [3]. Primary Care Providers (PCPs) seldom include trauma in their assessment due to their lack of knowledge or comfort in addressing trauma histories [4, 5, 6]. Perceived or real-time constraints on PCPs may also act as a barrier to engaging in multi-dimensional assessment. As a result, the experience is frustrating to both patient and clinician. Commonly, a mental health referral may be made when medical causes of symptoms cannot be found, but many known barriers exist for a psychological referral to be successful such as stigma, perceived lack of mental health services in the native country, competing cultural practices, or lack of knowledge about services and language barriers [7]. Even when correctly identified and referred for trauma care, refugees often fail to follow through on the referral [1], resulting in recurrent urgent care and emergency rooms visits with recurring symptoms.

Inability to get mental health care for traumatized patients produces significant adverse effects. Studies have shown when traumatic events go undetected, medical care is often reduced to misdirected and ineffective treatments [8]. Compared with non-refugees, refugee patients receive less psychotherapy, have greater drop-out rates, and have poorer outcomes with mental health services [9]. The need for specialized language interpretation, differing cultural conceptions of mental disorder between the practitioner and the patient, and the individualistic orientation of psychotherapy are complicating issues in treatment [10].

If traumatized refugee patients were to receive appropriate mental health care, there is evidence that it would be beneficial. Research with traumatized veterans and rape and child abuse victims shows improvement with use of psychotherapy [11], psychotropic medications for PTSD [12], treatment of insomnia [13], and various types of physical rehabilitation, including massage, exercise and acupuncture [14]. Randomized controlled trials with torture survivors are scarce [11], but the few offer promising cross-cultural adaptations of evidence-based practices for trauma and PTSD [15, 16, 17].

Better trauma treatment strategies could influence healthcare utilization and cost. In general, somatizing patients and those with co-occurring psychological complaints tend to have higher health care utilization rates than other patients. Medical costs, estimated at $250 billion per year, are attributed to somatization, with two times the outpatient and inpatient utilization as non-somatizing patients [18]. This results in high levels of role impairment, more days spent in bed per month than other chronic disease groups, elevated rates of sick leave and higher rates of unemployment. Medically unexplained symptoms are also more chronic and more refractory to treatment than many symptoms with a demonstrable organic basis [19]. Somatizing refugees frequently require interpretation during visits, an additional cost in personnel and time. However, studies of healthcare utilization by refugees are few. Given the high costs of somatizing patients, the correlation between somatic symptoms and psychological problems and the high degree of somatization among refugees with trauma histories, the same high utilization rates can be assumed. Clearly, strategies and better interventions need to be developed and tested to address multi-problem traumatized refugee patients from both an efficacy and efficiency standpoint.


This intervention study investigated two areas of concern in treating a group of East African refugee women with a high likelihood of previous war trauma and torture and multiple physical complaints. We explored (1) the willingness to engage in psychotherapy after a mental health referral from primary care and (2) the effect of engagement in psychotherapy on healthcare utilization.

The clinical intervention was developed in an urban primary care clinic designed to treat the large Somali population in Hennepin County. Because of regional and cultural similarities, Oromo patients were also seen if they presented at the Somali Clinic. The intervention components reflect considerations of culture, trauma, and somatization and facilitate ongoing co-management based in the somatic presentation. We also sought to compare patients who adhered versus did not adhere to therapy on factors that might be predictive of therapy engagement. We hypothesized that therapy adherence might decrease utilization of other types of healthcare, such as urgent and emergency care and primary care. Authorization was granted to this study by the Human Subjects Research Committee, Minneapolis Medical Research Foundation.

Population Studied

We designed a retrospective study of patients from the Somali Clinic at Hennepin County Medical Center, Minneapolis, Minnesota, USA. Hennepin County is home to an estimated 40,000 East African refugees from Somalia and Ethiopia. A convenience sample of 64 primary care patients was selected from East African women seen for at least one visit by the co-located psychologist between June 2000 and March 2003. Chart abstraction from electronic medical records was completed in December 2004 to assure that all patients had at least 12 months of visit data following their first psychological visit. Data elements included visit dates and types, extensive descriptive data and information abstracted from narrative notes.

The 64 women sampled averaged 43.5 years in age (from 19 to 71), had an average of 5.2 children, and 26 % were currently married. 63 were Somali and 1 was Ethiopian of Oromo ethnicity. They presented with a mean of 3.8 complaints: 2.1 psychological and 1.7 somatic. Table 1 shows the frequency of chief complaints. 65.6 % received a dual psychological diagnosis and 34.4 % a single psychological diagnosis. 56.3 % were diagnosed with PTSD; 40.6 % with major depression; 39.1 % with depressive disorder and 6.3 % with anxiety disorder.
Table 1

Frequency of chief complaints




Headaches/cervical extra thoracic



Nightmares/sleep disturbance/night-time wandering









All over body pain/joints



Memory problems/forgetfulness



Gastrointestinal pain/abdominal pain






Limb pain (specific)



Panic attacks/anxiety with no clear precipitant






Criteria were developed prior to chart abstraction to identify the 64 patients as either agreeing to engage in psychotherapy or not for comparison. To control for treatment-seeking as a confounding variable in the decision to engage in therapy, a new variable was created called therapy adherence, which combined the number of psychotherapy visits with one of two possible therapy milestones achieved and noted in the electronic health record. The visit cut-off for therapy adherence was four, based on completion of the early engagement process [20] and the effect of mental health treatment beginning at the fourth treatment session with a positive patient treatment fit [21]. The milestone options were either a return to discussion of traumatic events or a statement of benefit of psychotherapy by the patient and were abstracted and coded by the investigators. The four possible statements of benefit abstracted were (1) Patient states that the therapy or therapist is valued; (2) Attributes symptom reduction or improvements in functional behavior to therapy; (3) Notes symptom improvement; (4) Notes goal attainment.

The Intervention

The intervention consisted of five components: (1) integrated physical and mental health services; (2) trained interpreters and bicultural health workers; (3) staged but flexible four visit protocol for addressing physical and psychological complaints by the primary care provider; (4) co-management of patients receiving physical and mental health services; and (5) trauma-informed psychotherapy available in the primary care setting. The components address barriers to effective care for immigrants/refugees such as language, mental health stigma, somatization, co-morbidity and reluctance to divulge trauma.

The goals of the staged four visit protocol were: (1) to prioritize and treat the multiple complaints and start to assess their meaning; (2) to establish trust with the patient allowing for increasing candor, belief that the provider can be beneficial, and willingness to consider treatment options that are unfamiliar or threatening; and (3) to connect the patient’s experience and understanding of symptoms to treatment options available to the clinician. The framework shown in Table 2 illustrates the four visit protocol.
Table 2

Staged approach by primary care providers to address suspected trauma

Visit flow

Relationship building activities


Assessment/diagnostic activities

Patient education


Trust building

Introduction of roles, including interpreter

Include family that is with patient in discussion

“What brings you in?”

Are vague and may represent cultural beliefs

“Fire in abdomen,” “my liver is bad,” “blood is boiling.”

Immediate referral for suspected psychosis. Diagnostic laboratory testing. Limit physical touch

Suggest ongoing continuity: “I will see you in a few weeks. Bring all medications to appointments.”

Therapeutic listening

Non-narcotic pain drugs

Treat gastric reflux

Physical examination


Work around clothes for physical assessments

Usually pain and/or sleep are only slightly, or not at all, improved

Review labs in detail. “How is the pain with the medications?” “How is sleep?” Diagnostic testing of affected area: (X-ray)

Give relevant patient education for symptoms

Treat active infection, pain and insomnia with low-risk medications

Ongoing symptoms


Frequent visits at first help to establish rapport

Pain may have changed somewhat, sleep may be better or too much sleep

Review labs/X-ray(s). Ask about sleep and pain. Use positive framing: “so far there doesn’t seem to be a disease causing your symptoms, yet you don’t feel better.”

“Let us try some physical therapy.”

“Let us try a different medication to calm the body.”

Physical therapy referral. Introduce antidepressant and prescribe low-dose

Mental Health referral


Use of frequent visits enforces concern about patient’s distress

Continuing symptoms, with no apparent physical cause

If patient doesn’t mention trauma experiences by this visit, empathize, then relate pain to “memories of hard times” or use story telling: “sometimes people in this situation have experienced…”

Reinforce value of person’s spirituality in coping

Increase dose of SSRI

Recommend mental health referral

Follow-up to continue co-management with mental health

Statistical Methods

Baseline utilization was defined as including visits for primary care, urgent care, and emergency care going back 2 years from the initial therapy visit. Subsequent utilization was computed for two time intervals: 0–6 and 7–12 months after the initial therapy visit. Utilization rates are presented per month and are calculated as the number of visits divided by the number of months at risk in each period. Rates were compared between time periods using generalized estimating equations (GEE) with a Poisson error distribution log link, and offset equal to the log of the number of months (24 months for the baseline period and 6 months for each period after the initial therapy visit), allowing over dispersion. The working correlation structure was independent and standard errors are empirical. Adjusted average log rates were calculated as SAS’s least squares means using observed margins, and differences between time periods were tested using differences in these least squares means. P-values less than 0.05 were considered statistically significant. Estimates and confidence limits for rates per month were computed by exponentiating estimates and confidence limits of the least squares means. All analyses were executed using the GENMOD procedure in SAS version 9.3 (Cary, NC).


Group Comparison

Our criteria created two post hoc groups: therapy adherents (n = 31) and therapy non-adherents (n = 33). The two groups were examined to see if they differed other than by adherence or non-adherence to psychotherapy. More therapy adherents were separated from their minor children (25.8 vs. 9.1 %, 0.08) and carried a dual diagnosis of major depression and PTSD. Not all of the women in the sample voluntarily reported traumatic events. Patients were not given standardized assessment tools to elicit trauma history; therefore, the number of reported traumas is likely to be much less than actual. Among those that did report, the sample as a whole reported an average of 1.8 trauma events. A higher percentage of the therapy adherent group reported any trauma events (84 vs. 52 %, p = 0.006). The mean number of reported trauma events was higher for those therapy adherents reporting any trauma than for the non-adherents, and trending towards statistical significance (2.00 [SD = 0.87] vs. 1.53 [0.70], p = 0.0578). Table 3 shows the types of trauma events endorsed by the sample as a whole.
Table 3

Most frequently reported trauma events




Witnessed family killed



Physically assaulted



Non-family shot/killed



Witnessed or had child killed/injured



Sexually assaulted



Witnessed sexual assault of family/abduction of girls



Experienced sadistic violence



Witnessed physical assault






Home (or other property) invaded/destroyed



Mean # of events by those voluntarily reporting any events (n = 44 reporting any events)

1.8 (SD = 0.82)


Healthcare Utilization

Table 4 shows the unadjusted rate of visits per month in each time period. Therapy adherents had more healthcare utilization of each type overall in all time periods. At baseline, (i.e. prior 2 years), rates of overall utilization (all types of visits) were 0.38 per month (95 % CI [0.31, 0.47]) for adherents and 0.17 (0.13, 0.21) for non-adherents (data not shown). PCP visit rates differed significantly for therapy adherents compared to non-adherents (p = 0.0006) as well as between time period (p < 0.0001). The small number of urgent care and ED visits in the non-adherent group did not allow for a similar analysis.
Table 4

Unadjusted utilization

Time period from initial therapy visit

Type of visit

Therapy adherents (N = 31)

Therapy non-adherents (N = 33)





Prior 2 years






Urgent Care










0–6 months






Urgent Care










7–12 months






Urgent Care










PCP primary care physician, ED emergency department

Figures 1 and 2 show utilization rates for PCP utilization in both groups (Fig. 1) and for urgent care and ER visits for the adherent group (Fig. 2). Rate of PCP utilization changed significantly over time (Fig. 1). For both adherents and non-adherents, PCP rates increased significantly from the prior 2 years to 0–6 months after the initial therapy visit, but then decreased in months 7–12 (although they remained significantly higher than in the 2 years prior to the initial therapy visit). PCP utilization rates in months 0–6 were 6.5 times higher (95 % CI [4.5, 9.3]) for adherents and 4.5 times higher (3.2, 6.3) for non-adherents compared to baseline. Rates in months 7–12 were almost twice as high as during baseline (2.1 [1.6, 2.8] for adherents, 1.7 [1.3, 2.4] as for non-adherents). For therapy adherents, rates of ER visits were lowest in months 7–12 (Fig. 2), but the change was not statistically significant. Similarly, rates of urgent care visits did not differ significantly between time periods.
Fig. 1

PCP utilization over time. Error bars represent 95 % confidence limits

Fig. 2

Urgent care and ED utilization over time (therapy adherents). Error bars represent 95 % confidence limits


This study investigated two areas of concern in treating East African refugee women with high likelihood of previous war trauma or torture and multiple physical complaints: (1) the willingness to engage in psychotherapy after a mental health referral from primary care and (2) the effect of engagement in psychotherapy on healthcare utilization.

This clinical intervention was associated with a high level of engagement and adherence to a mental health referral—48 %. The five components of the intervention described in the Methods section reflected a culture- and trauma-informed approach to multi-problem refugee patients. Three of these components (integrated physical and behavioral health, skilled interpretation, and co-management of patients with physical and psychological diagnoses), have been recommended previously to overcome barriers to mental health services for underserved minorities [22].

Primary care is thought to be the place where issues of somatization and physical distress are most likely to be first addressed and a mental health referral most likely to be effective. Integrating mental health into primary care shows promise in increasing access to and acceptance of mental health services [23]. Older primary care patients were found more likely to accept collaborative mental health treatment within primary care [24]. Integrated care was shown to improve access to mental health services among older African Americans [25].

The Veterans’ Affairs Medical Centers are showing positive effects of integrated care on veterans’ willingness to engage in mental health treatments [26].

The authors suggest that ongoing co-management, in addition to treating multiple problems, aids in maintaining adherence to psychotherapy during difficult times when traumatic events are discussed for the purpose of reframing, desensitizing, and “moving on” [27]. The patient’s trust in the primary care provider can be leveraged to continue with psychotherapy, as other medical treatments are also available.

The four-visit primary care protocol was newly designed to improve assessment of trauma by PCPs as multiple complaints are triaged and treated, and to increase the likelihood of a successful referral to mental health services where indicated. The details described in Table 2 operationalize the process of assessment, trust-building, and ongoing co-management that can challenge the most diligent provider.

Other factors could have influenced the decision to engage in psychotherapy. The therapy adherents could simply have been more prone to utilize care or could have had a higher level of distress than therapy non-adherents. Higher distress may have played a role in overcoming the barriers to mental health treatment. The Jaranson et al. study of torture prevalence among East Africans [1] found that only 1 % of traumatized refugees with a likely PTSD diagnosis reported seeking Western treatment, despite high reported levels of distress. The co-managed care model described in this paper engaged traumatized patients in a shared decision for mental health therapy while continuing a robust primary care relationship.

In general, those who engaged in psychotherapy showed a significant increase in primary care utilization from baseline, while utilization of emergency and urgent care decreased slightly (not significant). Non-adherents showed a similar pattern in primary care utilization. The change from baseline did not differ significantly between groups, (rate ratios = 6.5 vs. 4.5 in period 0–6 months, 2.1 vs. 1.7 in period 7–12 months). Among non-adherents, utilization of emergency and urgent care was minimal in all periods. The finding that more appropriate primary care increased over time is encouraging. The slight (not significant) increase in ED utilization by therapy adherents during the first 6 months of psychotherapy is not surprising. Processing traumatic events during treatment can temporarily increase symptoms such as acute anxiety or flashbacks. Nevertheless, evidence-based trauma therapies recommend reprocessing the trauma narrative as a critical aspect of successful treatment [28, 29].

This quasi-experimental retrospective study had several important limitations. There was no a priori control group for testing the effects of psychotherapy on healthcare utilization. Patients were not randomly assigned to mental health services and they self-selected appointment attendance. The therapy adherent and non-adherent groups differed substantially in their baseline utilization rates. Therefore, higher service utilization may have been a confounding variable reflecting greater general treatment-seeking among therapy adherents. The formation of the independent variable therapy adherence was done post hoc and therefore had no validity testing. The two groups also differed in their willingness to divulge traumatic events. This study did not try to assess whether patients with more reported trauma events felt a greater need for mental health services. The type of psychotherapy provided was not tested in this study, nor were formal measures taken of symptom reduction in the two groups. The use of chart abstraction and the coding of patient “milestone” achievements via provider chart notations, rather than objective measure, were subject to provider bias as well as coder bias and errors in the abstraction process.

This is the first study of refugees to test integrated mental health and primary care and its association with health care utilization. The results trended in the direction of more appropriate and cost effective care. Forty-eight percent of the group of 64 female refugee patients in this study chose to engage in psychotherapy for an average of 22 visits, demonstrating that integrated and collaborative care can be achieved. Overcoming cultural barriers using a staged intervention led to an effective referral from primary care. This promising technique suggests referral adaptations may be needed with this population. It will be important to develop and validate a reliable measure for therapy adherence in refugees with trauma histories. Multidisciplinary clinics in hospitals may be uniquely positioned to address engagement in psychotherapy by East African refugee women with mental disorders using a collaborative and informed approach.


Authors received Grant support from Minneapolis Medical Research Foundation.

Copyright information

© Springer Science+Business Media New York 2014

Authors and Affiliations

  • Carol C. White
    • 1
  • Craig A. Solid
    • 2
  • James S. Hodges
    • 3
  • Deborah H. Boehm
    • 4
  1. 1.Center for Victims of TortureSt. PaulUSA
  2. 2.Minneapolis Medical Research FoundationMinneapolisUSA
  3. 3.Division of BiostatisticsUniversity of MinnesotaMinneapolisUSA
  4. 4.Division of General Internal MedicineHennepin County Medical CenterMinneapolisUSA

Personalised recommendations