Jerusalem has been subjected to numerous terrorist attacks over the last 60 years, even before modern Israel reached independence (1948). From 1967 through September 2003, there were 486 attacks, with 510 persons killed and 2,636 injured (1, 2). During the present armed intifada (Arabic for insurrection), terrorism has escalated steeply, often including the suicide/homicide bomber modality targeting civilians (3). To illustrate, while five large-scale terrorist attacks occurred in Jerusalem between 1995–1997, and four civilians were killed in isolated attacks in 1998, since the beginning of the intifada in late September 2000 through December 2002 the number of attacks rose to 21. These later attacks caused 146 deaths, both of adults and children, untold number of wounded persons, and considerable property damage (2). Including the surrounding areas of Jerusalem, the number of terrorist episodes during this period reached 68, with 182 dead and 1,160 wounded.

Terrorist attacks leave in their wake personal and community sequelae of different duration and intensity (4–17). The psychological and psychopathological effects of terrorism on urban populations (e.g., Northern Ireland [8, 9] and Algeria [10]) have been raised anew in the literature following the 2001 events in the United States. The effects may be examined with regard to their psychiatric impact (14), the population-level behavior (17), and the human suffering that falls outside the psychopathological domain (11). In the United States, the reactions are a matter of ongoing research, with results (12–16, 18) suggesting that the psychiatric impact of the September 11 events may be similar to findings made in other disasters (19) in that exposure is a risk factor for anxiety, depression, and posttraumatic stress disorder (PTSD).

Prior to the current intifada, several studies were conducted in Israel on the effects of the Iraqi missile attacks during the 1991 Gulf War. The results indicated early elevated cardiovascular mortality (20) as well as worsened self-appraisal of physical health status, increased smoking, diminished physical activity, changed diet habits, and increased psychological distress (21). One study explored reactions among the elderly and found an increased vulnerability among Holocaust survivors (22). Nakar et al. (23) studied general practitioner burden in a highly exposed area and found a decreased total rate of visits. In addition, a laboratory study exploring the Gulf War impact on anxiety and cortisol and growth hormone levels showed that biological measures were within normal limits, presumably as a result of adaptation (24).

How does terrorism affect psychiatric services utilization by the general population? American reports suggest that psychiatric services were used sparingly after September 11 (15, 25–32) but that there were increases in prescriptions of anxiolytic drugs by non-psychiatrists (27) as well as by existing users (28). A similar pattern of changes in service utilization had been noted in Northern Ireland, where terrorism has been long-lasting (8). Despite reporting relatively elevated rates of traumatic stress-related symptoms, a low proportion (5.3%) of Israelis sought professional help according to a national telephone survey conducted during the current intifada (6).

Thus, the New York, Northern Ireland, and Israel studies suggest that under the stress of terrorism there are changes in some, but not all, health behaviors. In the study reported here we examined a comprehensive mosaic of routinely collected mental health and general health service indicators to ascertain intifada-related health utilization demands of specialized and nonspecialized services among Jerusalem residents.

Data Sources

Governmental and nongovernmental service organizations provided monthly data on general health and mental health services indicators, covering the Jerusalem Jewish population (N=261,400) (Table 1).

Psychiatric Services

Citywide, free, and easily accessible governmental psychiatric clinics that offer care for all types of psychiatric disorders provided information on the number of contacts by gender, age group (18–64 or 65 years and older), diagnoses, and administrative status (new contacts, renewed contacts after a 6-month break, and ongoing care). We also obtained data on the number of new patients who contacted these clinics before and during the current intifada and their respective diagnoses, as well as data on the number of adults that sought help for the first time in the city’s two clinics for substance abuse disorders.

The National Psychiatric Case Registry, which records all admissions to and discharges from inpatient services, provided the number of first and repeat hospitalizations of Jerusalem residents.

The national emergency (personal crisis) hotline submitted the number of telephone calls originating in Jerusalem.

General Health Services

Two of the four national health maintenance organizations, estimated to cover over 40% of our target population, provided information on the monthly rate of adult visitors to their primary care physicians.

Magen David Adom (local Red Cross) provided the number of home visits for ambulances equipped to handle cardio- and cerebrovascular emergencies, dispatched by the duty officer according to the problem described during the telephone call.

Statistical Analysis

Time series analysis was applied given the temporal nature of the data (33). We used an autoregression approach, a variant of a linear regression permitting autocorrelation of residuals. All health indicators were examined monthly. The period of observation was divided into two time intervals: the 21 months preceding the current intifada (Jan. 1, 1999–Sept. 30, 2000) and the 27 months after the intifada onset (Oct. 1, 2000–Dec. 31, 2002). The models included 1) a periodic seasonal component to account for the seasonality in the use of health services; 2) a general linear trend, since there is a steady increasing trend in service utilization resulting from widespread health education and greater access, among other factors; and 3) two intifada-related components, short- and long-term impact. Short-term impact included effects that resulted from an initial reaction to the intifada outbreak, defined as including a peak at the third month and ending about 1 year thereafter. Long-term impact included effects that followed a linear trend starting at the intifada outbreak that may represent a more stable behavior of a population living under continuous threat.

The full model we applied included all components regardless of their statistical significance. While preserving the general linear trend, we also performed a backward elimination of components to reach a parsimonious model in which all remaining components were statistically significant (p<0.05).

To eliminate the seasonal component we performed median smoothing. Smoothed values were calculated for the period June 1999 to June 2002. A smoothed value at a specific point was defined as the median of a 12-month period starting 5 months before that point and ending 6 months thereafter. With regard to clinic diagnoses, we compared the proportions of patients for each of the eight diagnostic groups between the periods before and following the intifada outbreak using chi-square statistics. The correction for multiple comparisons was performed using the step-down Bonferroni procedure (SAS procedure Multitest). All analyses were performed with SAS 8.12, using the SAS ETS AUTOREG procedure (34).

Psychiatric Services Utilization

Following the intifada outbreak, adult psychiatric outpatient visits (new, renewed, or ongoing) did not increase (Figure 1, smoothed line). In contrast, the number of visits by the elderly (i.e., those 65 and older) increased on three of six short- and long-term impact parameters: increases in visits among ongoing cases (short- and long-term) and short-term increases in new contact visits (Table 2).

The diagnoses of new cases at the psychiatric clinics did not significantly change during the two study periods with regard to the following categories: organic, mood (affective), personality, and unspecified disorders. There was a decrease with regard to behavioral syndromes associated with physiological disturbances and an increase in psychoactive substance use (p<0.01), schizophrenia (p<0.01), and neurotic stress-related and somatoform disorders (p<0.03). However, when Bonferroni correction for multiple comparisons was applied, no significant changes remained.

Similarly, we found no changes over time in the number of first contacts at the city’s two substance abuse clinics (1998: N=168; 1999: N=161; 2000: N=171; 2001: N=164; 2002: N=160; 2003: N=128).

There were no short-term changes in the number of first admissions to psychiatric hospitals but there was a significant long-term decrease (Table 2). There was a significant short-term increase in the number of psychiatric readmissions but a long-term (albeit nonsignificant) decrease. There was also a long-term increase in the number of hotline calls (Table 2).

General Health Services Utilization

Following the intifada outbreak, the rate of monthly general practitioner visits reported by the two health insurers increased for both men and women (Table 2 and Figure 2, smoothed line; data are shown for both genders combined). There were also short- and long-term increases in the number of monthly ambulance calls (Figure 3).

Our inquiry focused on a single behavioral dimension, health service use, among the general population in response to terrorism. Analogously, Stecklov and Goldstein (17) focused on driving behavior during part of the period we covered. Neither that study nor ours targeted people that survived the attacks or were bereaved by the loss of a significant other.

Our results show that, with few exceptions, the Jerusalem residents did not seek increased help from psychiatric services during the study period. Thus there was no increased influx of adult patients to governmental psychiatric clinics, except for elderly patients, a subgroup that may be vulnerable to terrorism-related stress due to earlier life exposure to the Holocaust and past wars (22). Last, the changes in the relative proportion of diagnoses over time among new patients failed to remain significant following statistical correction. These almost general negative findings are reinforced by the reduction in the long-term first psychiatric hospitalizations, or, stating it more conservatively, by the nonreversal of a negligible descending trend that characterizes first inpatient admissions in the country (35). The readmission of patients who had been previously hospitalized showed mixed results, a short-term increase (a possible effect of stress on this vulnerable group [36]) but no long-term changes, perhaps an effect of adaptation. Consistent with studies in the United States (37) and in Israel (6), emergency personal crisis telephone hotline service use increased. This is not surprising, since this service is easily accessible from home, it operates around the clock, it is free, and it is anonymous. It is important to note in order to better judge the results that Jerusalem did not lose more residents during the period studied compared with previous years (www.jils.org.il/shnaton), although life in the city was more dangerous.

This report has some limitations. Information was collected and recorded by agencies outside our immediate supervision, except those from the government (35). Also, data from the university clinic specializing in stress reactions following all types of traumatic events were not collected nor were data from private practice psychologists funded by the National Insurance Institute. Those settings, which charge a fee for their services, mostly or solely provide care for survivors of the attacks and their families as well as for bereaved families that may be referred by the National Insurance Institute rehabilitation officers upon need. Recall here that our focus was the general Jerusalem population and not persons directly involved in the attacks.

The two health insurers that provided data on visits to primary care physicians cover slightly over 40% of the Jewish city residents. We are confident, however, that the populations covered are not biased, since the insured belong to a heterogeneous population in terms of socioeconomic status, degree of religious observance, and residence, and their doctors’ clinics are spread across the city, including the neighborhoods in which the attacks occurred. Obviously, our study is neither covering later periods of the current intifada nor the time when terrorism will cease and peace and reason will prevail. Thus, we cannot claim that the mental health help-seeking pattern will remain as we found it here.

This study has several strengths. Methodologically, it relies on a composite mosaic built by using multiple sources of information that kept their system of data collection unchanged over the years under examination. Note here that, in contrast with the community surveys conducted in New York City (32) and Israel (6), purported to provide unbiased population rates, our data sources were not affected by the relatively low response rates obtained in those studies nor by respondent recall. With regard to the analysis, we utilized time-series analyses, controlling for long-term effects and seasonal fluctuations. Boscarino et al. (25), who did not use time-series analysis, found a small, yet significant increase in 1-month services utilization in Manhattan after September 11, 2001, but their finding may be questioned by lower services utilization in the month of August, when many New Yorkers (including service providers) are on vacation. Rosenheck and Fontana (31) studied six 6-month periods of services utilization and applied analysis of variance to assess the impact of the September 11 attacks in New York on Veteran Affairs mental health services utilization. Their study likely contains too few data points to distinguish between short-term and long-term effects and to ensure that identified effects (or the lack thereof) are not the result of a general trend. The importance of adjusting for a general linear trend, which reflects inherent dynamics of service utilization, is obvious since changes due to the greater use of the services by an increasingly health-educated population may be mistakenly attributed to the stress of terrorism.

The field of public mental health faces a certain paradox. Community-based surveys on terrorism highlight adverse psychopathological effects (4, 6, 12–16). Yet, our data and previous service-based inquiries (15, 23–27, 29) reveal a much less worrisome picture. Of late, Druss and Marcus reached a similar conclusion (28). We acknowledge that treatment gaps measured by true prevalence studies and services utilization rates are no novelty (38), but under the shadow of terrorism, the treatment gap with regard to psychiatric services looms very large. Indeed, a relatively small gap would have been expected, given that the social stigma of terrorism-induced mental disorder is conceivably less than that surrounding most other disorders, and that terrorism-induced psychopathology is likely to have a relatively sudden (19) and thus noticeable onset to the person and others compared with the insidious beginning of most mental disorders. In addition, one would expect that the public mental health information continuously available through the media during the current intifada, and the encouragement to seek consultation if needed, may have led to a lowering of the threshold for utilization of specialized services.

For the apparent paradox we offer two preliminary explanations. First, most trauma-exposed persons may consider their suffering as a normal psychological reaction and do not believe that psychiatric services are the adequate address to obtain the necessary relief. As noted, “In the aftermath of terrorist attacks, many Americans may have regarded their distress as a ‘normal’ reaction…than as a disorder needing [psychiatric] care” (28). As in regular times, people that may react with varied symptoms of terrorism-induced stress seem to turn to their general practitioner for help or to other sources (28). To better ascertain this we would have required the general practitioner diagnoses, assuming that they were reliable, but this is not yet feasible.

Second, some epidemiological studies may overstate the extent of trauma-related psychopathology (4, 6, 12–16); up to 14.3% of those living in proximity to the World Trade Center in Manhattan surveyed in October 2001 met criteria for PTSD or depression (32). In contrast to these results, Northern Ireland psychiatrists stated: “because someone experiences or witnesses an act of violence does not mean that…[he or she] will inevitably develop psychiatric morbidity” (10).

The time may be approaching to re-evaluate psychiatric predictions of the effects of terrorism and to make better empirical attempts to explain this paradox. In particular, research is needed on determinants of help-seeking both among trauma survivors and the general population indirectly affected. The pathway from the perception of distress to the utilization of psychiatric services involves attribution of lack of normalcy to the distress and, obviously, available effective care. These perceptions need to be studied in detail. Meanwhile, access to mental health care should be facilitated by buttressing the mental health component in primary health care in emergencies, as recommended by the World Health Organization (38, 39) and clearly showed in this inquiry. This strategy is most effective when the specialized mental health agents are community-based and collaborate with the general practitioners and nurses in dealing with the diverse trauma-related reactions, provided they avoid “medicalizing [complaints] without justification” (40).