The majority of symptoms reported by survivors of the fireworks disaster was not presented to the GP. Survivors with persistent symptoms (symptoms reported at both T1 and T2) were more likely to seek medical care for their symptoms than survivors with self-reported symptoms at only one timepoint. Presenting the individual self-reported symptoms to the GP was not consistently associated with a high level of functional impairment and distress. When presented to the GP, the majority of symptoms could not be related to a medical disorder in the episode of care.
To our knowledge, this is the first study that has compared self-reported symptoms among survivors of a disaster with the symptoms that were registered in the medical records. Despite this, some potential limitations of the present study deserve attention. Firstly, selective response and possible bias is of concern in this study. Shortly after the disaster, all affected residents were registered; this database was used to detect possible demographic differences between participants and non-participants at T1. Only an estimated 30% of all affected residents participated in the questionnaire health survey, and the EMRs were not available for all survivors who participated in the health survey. Compared to the total affected group, the study population in the present study (participants at both T1 and T2) were more likely to be women, living with a partner, aged between 45 and 64 years and were more often relocated as a result of the disaster. In addition, participants at both T1 and T2 for whom the EMR was available were less likely to have a paid job than participants at both T1 and T2 for whom the EMR was not available. In two previous studies, we examined whether selective response resulted in biased prevalence estimates of health problems among survivors [14, 23]. In these studies, multiple imputation, a statistical technique that uses a statistical model to fill in plausible values for each missing data point, was used [24, 25]. Because the missing values are drawn from a distribution, a range of values is imputed for each missing value (e.g. five values), with variation appropriately reflecting the uncertainty about that value. Using this technique, it can be estimated what the prevalence of the outcomes of interest would have been if there had been no (systematic) attrition in the longitudinal study. The results from multiple imputation indicated that, despite selective response, the imputed prevalence estimates did not significantly differ from the crude prevalence estimates of health problems, indicating that the selective response did not result in highly biased prevalence estimates.
Secondly, we compared self-reported symptoms with ICPC codes registered in the EMRs of survivors . Therefore, we have to consider the sensitivity and specificity of the ICPC codes corresponding to the self-reported symptoms. The GP might not register all symptoms that the patient presented. There may be some false negatives, and the percentage of survivors presenting self-reported symptoms may be an underestimation of the actual number of symptoms that are presented to the GP. In addition, table 2 shows percentages of survivors who had no self-reported symptoms at T1 or T2 but for whom symptoms were registered in the EMRs. These cases are not necessarily false positives, since the symptoms could have been correctly registered, if the symptoms occurred only between the two waves.
Compared to self-rating scales, relatively little is known about the validity and reliability of GP records. Despite this, a high level of agreement (mean 81%) between 161 GPs with regard to ICPC coding was found in the second Dutch national survey of general practice. This study indicated that health problems registered by GPs according to the ICPC provide a reliable overview of morbidity . Other studies have also shown that ICPC coding is a valid method for studying morbidity . In addition, the ICPC allows registration of symptoms with a high level of specificity, since the symptoms and complaints can be registered at the symptom level in the EMRs of patients. This is an important advantage compared to other classification systems. Despite this, it is possible that the GP did not register all symptoms presented by a patient. Indeed, most survivors have multiple symptoms and it is likely that the survivor presents, or the GP registers, only the most important symptom(s). This could have resulted in an underestimation of the symptoms presented to the GP.
Thirdly, we could not compare the self-report data with GP data for a control group. Despite this, our findings are comparable to the findings from general population studies in which only a minority of symptoms was presented to the GP [2, 3]. For example, Green et al. have shown that in the US general population, adults visit a physician for about 25% of the symptoms they experience .
Since causal attributions and illness perceptions are strongly related to health care utilization for symptoms, it is possible that survivors are more likely to seek medical care for symptoms when they attribute their symptoms to exposure to toxic substances from a disaster [28, 29]. Following an aircraft disaster in Amsterdam, survivors attributed their symptoms to depleted uranium. After this disaster, 53 to 80% of reported symptoms were known to the GP . We have reason to believe that the survivors of the Enschede fireworks disaster did not attribute their symptoms to exposure to toxic substances. Three weeks after the disaster blood and urine samples were taken to examine possible exposure to toxic substances . The results did not indicate elevated body burden. In the aftermath of the fireworks disaster no conspiracy theories about health problems due to exposure to toxic substances developed
To date, most studies that have examined MUS have been performed in primary or secondary care and have examined symptoms for which patients sought medical attention. The present study examined whether self-reported symptoms among survivors of a man-made disaster were presented to the GP, and showed that the majority of self-reported symptoms was not presented to the GP. Symptoms that were reported at only T1 were less often presented to the GP than symptoms reported at only T2 or persistent symptoms (at both T1 and T2). It is possible that survivors were mostly impaired by psychological problems such as anxiety and depression shortly after the disaster, and that they only sought medical help for physical symptoms after a longer period of time, when these symptoms became persistent or disabling. It can also be speculated that symptoms at T1 were likely to be transient or were explained by the survivors as a normal reaction to the disaster. Indeed, cognitions about symptoms affect medical care-seeking decisions. For example, Sensky et al. found that frequent attenders of general practice had fewer normalizing explanations for their symptoms than the comparison group . In addition, Cameron et al., found that symptoms attributed to stress rather than to illness were less likely to be presented to the GP . Although survivors presented only a minority of their symptoms to the GP, it is important to note that most survivors have multiple symptoms. In a previous study we found that 33% of the survivors of the fireworks disaster reported 10 or more symptoms in the health survey ; it is unlikely that survivors present all these symptoms to the GP. Instead, it is more likely that survivors report only the most important or worrisome symptoms to the GP. Since the number of symptoms is strongly related to the degree of functional impairment [5, 33], it is important that GPs ask patients about additional symptoms.
In a recent study among survivors of the fireworks disaster, the authors found that self-reported symptoms were strongly related to a high level of functional impairment and psychological problems . For that reason, we examined whether survivors with a high level of functional impairment and psychological distress were more likely to seek medical care for their self-reported symptoms. Survivors who presented fatigue to the GP were significantly more likely to have a high level of impairment and distress. Survivors who presented pain in bones and muscles, pain in back and stomachache to the GP were more likely to report a poor physical health and a have a high level of physical symptoms at T2 compared to those who did not present these symptoms to the GP. This pattern was, however, not found for the other symptoms. Apparently, a high level of impairment and distress was not the major reason to seek medical care for individual symptoms. It is likely that the decision to consult a GP was based on other factors, such as perceived susceptibility to illness, perceived severity of the symptom or beliefs about the cause of the symptoms [28, 29, 31, 33].
Of the self-reported symptoms that were also presented to the GP, 56% to 91% remained medically unexplained in the episode of care. This finding is consistent with a study among survivors of an airplane crash in Amsterdam in which it was shown that 57% to 85% of symptoms presented to the GP remained unexplained . General population studies have found similar percentages [1, 4]. The studies following disasters suggest that, as in the general population, physical symptoms such as headache, stomachache and fatigue among survivors are likely to remain medically unexplained after clinical judgment of a physician.
In conclusion, the majority of self-reported symptoms among the survivors of the fireworks disaster were not presented to the GP. On the one hand, this indicates that some symptoms reported in epidemiologic studies after traumatic events are transient and not a reason to seek medical care. On the other hand, this study shows that the symptoms presented to the GP are only the tip of the iceberg of symptoms that are related to functional impairment among survivors. The results of the present study indicate that the survivors who present their symptoms to the GP are not always those who have a high level of functional impairment and distress. When presented to the GP, most symptoms could not be related to a medical disorder, and were labeled as MUS. This result suggest that in a post-disaster context, just as in routine clinical practice, most physical symptoms reported on a symptom questionnaire will be transient, medically unexplained, or both.