Use of Health Care Services Before and After a Natural Disaster Among Survivors With and Without PTSD

Posttraumatic stress disorder (PTSD) is associated with increased utilization of mental health care and general medical care services (1,2). The greater use of treatment among persons with PTSD may be explained by the need to treat the disorder, but it may be due to associations between PTSD and other mental and general medical disorders (3–5). For example, PTSD is strongly associated with depression, anxiety disorders, somatization, and substance use disorder (6,7). Also, PTSD is associated with hypertension, hyperlipidemia, obesity, and coronary heart disease (8–11).

Most posttraumatic stress research uses cross-sectional or quasiprospective study designs. Indeed, often data are collected after people have been exposed to trauma and sometimes even after people have developed PTSD (12). This is problematic because neither trauma experiences nor the magnitude of posttraumatic stress reactions is randomly distributed in a population. Thus it is often difficult to determine whether associations between PTSD and other mental or general medical disorders or between PTSD and increased health care service utilization are due to trauma exposure, subsequent PTSD symptoms, or preexisting personal, social, and behavioral factors (1,2,12).

This study used both a prospective- and a retrospective-cohort design to investigate associations between PTSD symptoms and the utilization of health care services outside hospitals before and after the 2004 tsunami in Southeast Asia. Our data consisted of the Danish National Health Service Register (NHSR) (13) as well as questionnaires completed ten months after the disaster by a community sample of Danish tourists who experienced the tsunami (14). We examined whether health care service utilization before and after the disaster differed among survivors with PTSD and those without PTSD. We also examined whether the severity of PTSD symptoms ten months after the disaster was related to more frequent use of health care services after the disaster.

Methods

Participants and questionnaire

The participants in this study were Danish tourists who were repatriated from regions affected by the 2004 Southeast Asian tsunami within one or two weeks of the disaster. Upon the survivors’ arrival in Denmark, the police registered their names and personal identification numbers. In total, 2,271 Danes above the age of 18 who potentially had been exposed to the disaster were registered. In November 2005, ten months after the disaster, our questionnaire was mailed to the homes of all the registered survivors who were traceable in the Danish civil registration system (N=2,178) (15). The study was approved by the Danish Data Protection Agency. Ethical approval is not required for this kind of study in Denmark.

The questionnaire asked about demographic variables, exposure to the tsunami, and current symptoms of posttraumatic stress. We asked about specific details regarding disaster exposure, such as whether the participants were caught, touched, or chased by the waves and whether they had witnessed the death, serious injury, or suffering of others (15).

The response rate was 48% (N=1,051). Because of geographic differences in the impact of the tsunami, individuals who were present in Southeast Asia experienced a wide range of exposure (14,16). Therefore, our sample included individuals who had no direct exposure to the tsunami (15). Thus we limited our analysis to those who fulfilled DSM-IV PTSD criterion A1 (17), meaning being caught, touched, or chased by the waves or having witnessed the death, serious injury, or suffering of others. We excluded 362 individuals who did not meet criterion A1 and 54 participants with missing data on outcome variables, leaving 635 participants eligible for the analysis.

The mean±SD age of the study participants was 43.2±12.7 years (range 18–82 years), 51% (N=322) were women, 68% (N=429) were married or cohabiting, 35% (N=223) had completed more than 15 years of education, and 73% (N=465) were employed. The participants were similar to the age-adjusted Danish population in levels of employment (www.statbank.dk/AKU22) and education (www.oecd.org/edu/eag2009). Still, they were more likely to be married or cohabiting (www.statbank.dk/FAM100N).

We assessed posttraumatic stress symptoms with the PTSD Checklist (PCL) (18,19), a well-validated and reliable measure of the 17 PTSD symptoms listed in DSM-IV. The PCL assesses the intensity of PTSD symptoms during the previous month on a 5-point Likert scale, from 1, not at all, to 5, extremely. We used the PCL Stressor Specific Version with instructions to endorse only symptoms linked to tsunami experiences. We considered an item score of 3 or higher to indicate the presence of a particular symptom. This cutoff score is recommended by the National Center for PTSD (www.ptsd.va.gov/professional/pages/assessments/ptsd-checklist.asp). According to DSM-IV criteria, the diagnosis of PTSD requires at least one of five intrusion symptoms from criterion B, three of seven avoidance or numbing symptoms from criterion C, and two of five hyperarousal symptoms from criterion D. We allocated participants into three groups: PTSD, which consisted of participants whose symptoms met DSM-IV criteria for symptom clusters B, C, and D; partial PTSD, which consisted of participants with at least one symptom each from symptom clusters B, C, and D (20,21); and no PTSD, which comprised the remaining participants, who served as the comparison group. PTSD symptom severity was derived from the PCL sum score (possible range 17–85).

National register data

The tsunami occurred on December 26, 2004. We used the NHSR (13) to provide data on the participants’ annual number of contacts with health care services outside hospitals during the three calendar years before the disaster (2002–2004) and during the three calendar years after the disaster (2005–2007). The NHSR was established by Danish authorities, who use it to administer national health insurance for health care services offered outside hospital inpatient and outpatient clinics. All Danish citizens except conscripts, individuals living in institutions, and individuals living abroad are insured by the national health insurance system, and the registry is assumed to be a valid tool for health care research (13). The registry includes all contacts by Danish citizens with general practitioners (GPs), psychologists, psychiatrists, and members of other medical specialties, including physiotherapists, chiropractors, and podiatrists, who offer health care services in the health care sector outside Denmark’s hospitals. The contact dates are specified by week and year. Health care service contacts included consultations, group consultations, e-mail consultations, phone consultations, home visits, individual psychotherapy, and group psychotherapy. Contacts with a preventive purpose and contacts concerning pregnancy care, birth control, and general vaccination programs were excluded from our analysis.

Statistical analysis

We performed chi square tests (linear by linear association) to compare gender, education, marital status, and disaster exposure among participants with PTSD, partial PTSD, and no PTSD. We performed one-way analyses of variance (ANOVAs) with Bonferroni-corrected post hoc t tests to compare health care service utilization before and after the disaster, change in health care service utilization, and age among participants with PTSD, partial PTSD, and no PTSD. The distribution of these variables was found to be sufficiently close to the normal distribution for such analyses (22). Further, we used paired t tests to compare use of health care service before and after the disaster.

We examined the association between predisaster use of health care and postdisaster PTSD by comparing the prevalence of PTSD and partial PTSD and the severity of posttraumatic stress symptoms among participants grouped into quartiles of predisaster health care service contacts. Prevalence was compared by chi square test (linear by linear), and symptom severity was compared by one-way ANOVA.

Finally, the association between severity of posttraumatic stress symptoms ten months after the disaster (November 2005) and the subsequent annual number of contacts with health care services in 2006–2007 was analyzed by linear regression analysis, adjusted for demographic variables and predisaster health care service use. We used the Statistical Package for Social Sciences, version 14.0, for all analyses.

Results

According to their PCL scores, 79 (12%) of the 635 participants fulfilled the DSM-IV symptom criteria B, C, and D for PTSD. In addition, 97 (15%) participants fulfilled our criteria for partial PTSD. Table 1 shows demographic characteristics and tsunami exposure among participants with PTSD, partial PTSD, and no PTSD. Participants with PTSD and partial PTSD included a higher proportion of women, were less likely to be married or cohabiting, and reported more severe exposure to the tsunami than participants without PTSD.

Figure 1 shows the annual number of contacts with health care services before and after the tsunami among participants with PTSD, partial PTSD, or no PTSD. Compared with participants with no PTSD, participants with PTSD or partial PTSD had significantly more annual contacts during the three-year periods before and after the tsunami (Table 2). Annual contacts before and after the tsunami among participants with PTSD and partial PTSD did not differ significantly. Overall, the annual number of contacts with health care services increased .76 (9%) during the three years after the disaster compared with the three years before the disaster (t=4.19, p<.001). The increase in annual contacts for all health care services did not differ among the groups (Table 2).

The increase in annual contacts with health care services was most prominent during 2005, the year after the disaster (Figure 1). We performed post hoc tests to examine whether there were differences among the groups in the extent to which health care service contacts increased from the three years before the disaster to 2005. The annual number of health care service contacts increased by 2.55 (26%) in the group with PTSD (t=3.54, p<.01), by 1.43 (17%) in the group with partial PTSD, and by 1.09 (17%) in the group with no PTSD (t=4.13, p<.001). The increase in health care service utilization did not differ significantly among the groups.

Table 2 shows that GP contacts constituted a majority of health care service contacts among all groups. Contacts with GPs varied more among the groups than visits for any other health care service. Contacts with psychologists, psychiatrists, and members of other medical specialties constituted only a minor part of the total use of health care services. Overall, contacts with GPs (t=3.34, p<.001), psychologists (t=5.03, p<.001), and members of other medical specialties (t=2.28, p<.05) showed a significant increase after the disaster (Table 2). Use of a psychologist was the only health care service that increased significantly more after the disaster among participants with PTSD and partial PTSD than among participants with no PTSD (F=10.96, p=.001), but even among participants with PTSD and partial PTSD the increase was rather small (.2±.6 contacts).

Table 3 shows the association between predisaster use of health care services and prevalence of postdisaster PTSD and partial PTSD and severity of posttraumatic stress symptoms. The prevalence of PTSD was considerably higher in the quartile with the highest number of predisaster health care service contacts (19%) compared with the quartile with the lowest number of health care service contacts (6%). Also, the severity of posttraumatic stress symptoms was significantly associated with predisaster use of health care services.

Table 4 shows the association between severity of posttraumatic stress symptoms ten months after the disaster and the annual number of contacts with health care services in 2006–2007. In bivariate analysis, symptom severity was significantly associated with subsequent use of health care services (model 1). The association remained significant when the analysis was adjusted for sociodemographic variables (model 2). When adjusted for predisaster use of health care services, however, the association between severity of posttraumatic stress symptoms and subsequent use of health care services disappeared (model 3).

Discussion

In our study of a sample of Danish tourists who experienced the Southeast Asian tsunami in 2004, unadjusted analyses showed a relatively strong association between PTSD, partial PTSD, and severity of posttraumatic stress symptoms and use of health care services outside hospitals after the disaster. However, participants with PTSD and partial PTSD had used health care services before the tsunami more frequently than participants with no PTSD. Also, the increase in health care service use before and after the disaster was not significantly higher among participants with PTSD or partial PTSD compared with those with no PTSD. When predisaster use of health care services was taken into consideration, the association between severity of posttraumatic stress symptoms and subsequent use of health care services disappeared.

The ten-month prevalence of PTSD (12%) among survivors who fulfilled DSM-IV PTSD criterion A1 (severely disaster exposed) is consistent with similar reports in the literature (23). The level of health care service utilization by all participants was consistent with rates of utilization in the general Danish population (www.sst.dk/publ/tidsskrifter/nyetal/pdf/2005/02_05.pdf [in Danish]). Also, the finding that PTSD, partial PTSD, and severity of posttraumatic stress symptoms were associated with higher use of health care services is in concert with previous research involving community samples (24), combat veterans (3,5), residents of New York City after the World Trade Center disaster (25), survivors of other man-made (26) and natural disasters (27), and even Scandinavian survivors of the 2004 tsunami (28).

To our knowledge, the association between PTSD and subsequent use of health care services has not previously been adjusted for preexisting use of health care services. Our finding that postdisaster health care service utilization was largely predicted by predisaster health care service utilization and that it was hardly affected by the onset of posttraumatic stress symptoms itself raises important questions about the consequences of traumatic stress, especially from a long-term perspective.

First, our findings prompt questions about previous assertions that the presence of PTSD is the causal factor leading to high health care service utilization (2,29), particularly after one or more years have passed since the trauma. Another alternative explanation is that survivors with PTSD have higher use of health care services even before symptom onset and that PTSD symptoms themselves are not responsible for health care service utilization. Survivors with PTSD may neglect seeking help for their posttraumatic stress (25,30), they may have limited access to adequate treatment in the health care system (31), or their help seeking may replace, or be integrated into, health care consultations they would otherwise carry out. However, neglect of symptoms or limited access to health care was not supported by self-reports of health seeking in the Danish tsunami cohort. For example, ten months after the disaster, 40% of the repatriated Danish tourists, including those who had not been exposed to the tsunami, reported having consulted with their general practitioner for disaster-related problems (28). This treatment-seeking behavior was strongly associated with posttraumatic stress. It is possible, however, that treatment seeking for PTSD may have been strongly limited in the number of consultations and primarily occurred in the close aftermath of the disaster.

Second, our findings raise questions about previous assertions that psychological trauma explains the association between PTSD and other mental and general medical conditions (32). An alternative explanation is that survivors who develop PTSD have a preexisting vulnerability for mental and general medical conditions (33), as suggested by the higher use of health care services before the tsunami. Evidence for this hypothesis includes prospective studies that have identified preexisting mental and general medical conditions (34), neuroticism (35,36) and low cognitive ability (37) as risk factors for PTSD. Thus the frequent use of health care services among individuals with PTSD, as well as their increased vulnerability for other mental and general medical disorders, may be primarily a result of preexisting health conditions, low cognitive functioning, and ineffective coping strategies.

Virtually the entire population of Danes who were in Southeast Asia during the 2004 tsunami was asked to participate in our study, reducing sample selection bias. Furthermore, the Danish population is serviced by national health care that provides relatively equal access to care. Therefore, it is possible to evaluate the relation between PTSD and health care service utilization without considering economic factors. In nations with multipayer health incurrence, based on a mixture of public and private insurance programs, the role of economic factors in use of health care services could not be ignored.

Limitations of the study included a relatively low response rate. However, according to a study of nonresponders in a comparable cohort of Norwegian tsunami survivors who had a similar response rate (38), nonresponders were less likely than responders to have been in a place directly affected by the tsunami and to have fulfilled DSM-IV PTSD criterion A1. Furthermore, in this study, responders and nonresponders were similar with regard to age (15) and predisaster use of health care outside hospitals. Also, participants were similar to the general Danish population with regard to education and employment and included women and men of all ages.

Whether participants fulfilled DSM-IV PTSD criterion A1 was assessed by retrospective self-report, which may have been distorted by memory bias (39). To minimize such bias, we defined exposure on the basis of what may be considered objective events. The use of a single, self-report questionnaire to measure participants’ symptoms ten months after the disaster was also a limitation. Although the PCL is based on the DSM-IV symptom criteria, it cannot substitute for more comprehensive clinical interviews.

Our use of prospective national register data provided a measure of health care service utilization outside hospitals that is unaffected by possible recall and information biases that typically influence self-reports. Also, the equal accessibility of commonly acknowledged treatment minimized possible selection biases. The use of GPs is a sensitive measure of individual use of both social and health care services, given that the GP is a gatekeeper to other health care services as well as the physician who bears responsibility for communicating to social authorities.

Another limitation was entering utilization of health care services in the second and third years after the disaster as a variable in our predictive regression model. During this time frame, some survivors may have recovered from PTSD, resulting in reduced health care use. However, epidemiological research suggests that one-third of individuals who develop acute PTSD remain symptomatic for six years or longer (6). Similarly, 43% of the Danish tsunami survivors who had PTSD or partial PTSD ten months after the disaster were still symptomatic during a follow-up survey conducted two-and-one-half years after the disaster, indicating that a majority still suffered from disaster-related health problems (40).

Conclusions

This study found that postdisaster health care service utilization was predicted by predisaster health care service utilization and was hardly affected by the onset of PTSD itself. Our findings underline the importance of considering antecedents to trauma exposure when addressing the association between PTSD and health care service utilization. Similar considerations may also have consequences for the interpretation of the relationship between PTSD and later onset of comorbid disorders. The absence of important historical or pretrauma information may be a serious limitation within posttraumatic stress research.