Introduction

Electrical injuries occur in homes and at workplaces, despite increased technical safety and rules and regulations for handling electricity. Exposure to electrical current may result in a short pain and scare, but may also cause an injury with physiological damage such as tissue damage, burns or even heart failure or death [1]. The extent of the physical damage is related to voltage, point of current entry and pathway, duration and if the person was involuntarily stuck to the power source (no-let-go) [2]. The incidence rate of electrical injuries in Denmark is difficult to estimate as non-fatal electrical injuries are not systematically reported. This is also the case in other countries. Mortality from electrical injuries is rare in Denmark, with less than two cases on average per year from 1996 to 2005 [3].

Besides the physiological damages, electrical injuries have been reported to cause psychological problems, such as post-traumatic stress disorder(PTSD) [4–7], depression [5–9], anxiety [6–8, 10, 11], sleep disorders [5–7, 10, 11], cognitive problems [6, 8, 11, 12] and sexual dysfunction [9, 11]. An American retrospective study evaluated psychiatric examinations of 73 referred electrical injured patients and found that having experienced “no-let-go” or having lost or impaired consciousness was associated with psychiatric diagnosis [13]. The same research team repeated the study with 86 new post-acute electrical injured patients referred for treatment and supplemented with neuropsychological evaluations. They found that 78% of the subjects had one or more psychiatric diagnoses and that those with diagnosis also had poorer cognitive performance [14]. Follow-up time varied tremendously in both studies, from acute up to nine years [13, 14].

A French register-study on electrical injuries report that neuropsychiatric sequelae is the second most common type of sequelae after those directly related to burns [4]. In a small long-term follow-up study from the US, the neuropsychological problems were persistent more than a decade after the initial electrical injury [15]. In a follow-up study of 40 electrical injured patients, early emotional sequelae was found to predict poor outcomes 4 years after the injury in terms of adjustment to injury, psychological distress and return to work [16]. A larger multicenter study from Canada with 114 patients, who were interviewed shortly after the injury and again one year later, found that several neuropsychological complaints were persistent over time or even increased [17]. However, a recent Swedish study found no long-term cognitive dysfunctions [12].

Andrews et al. suggest that these neuropsychological symptoms relate to a specific syndrome related to electrical injuries, in line with post-commotional syndrome [18] and expand this in a recent review with suggestions that several types of cognitive problems are present following an electrical injury [19]. A study in an outpatient burn unit found that electrical injured patients are often referred to specialty consultation in various specialties, with psychology as the most frequent, and often the consultations does not lead to relevant findings [20].

A review of the literature from 2013 concluded that neuropsychological sequelae are common, including behavioral and cognitive changes, irritability, frustration as well as depression and PTSD. The sequelae were similar to symptoms resulting from traumatic brain injury [21]. A common problem in the literature was that the pre-injury mental health was not taken into account, since the studies use retrospective methods or are case studies. Another problem is that persons with electrical injuries are rarely compared to any control group. We identified one exception from this; a study from 1998 that compared electrical injury patients with electricians without history of electrical injury and found that the injured group had more physical, cognitive and emotional complaints [9]. We were not able to find any previous studies that had dementia as outcome, as previous studies focused on cognitive symptoms in terms of e.g. memory loss and concentration problems.

Swedish researchers recently interviewed 23 male electricians who reported different problems after an electrical injury, and found that the respondents reported both acute emotional reactions as well as long-term consequences in terms of anxiety, fear of performing the same task, feeling of guilt or anger and cognitive impairments. The study included only electricians with complaints, some even several years after the injury [22].

The existing evidence on mental and psychological problems following electrical injuries is based on either retrospective studies, studies without comparison groups or small case studies often conducted several years after the injury. We aim to improve this to a study using diagnoses from a national register and not self-reported symptoms as outcomes in a matched design.

The aim of this study was to investigate if persons with an electrical injury suffer from mental disorders in the following years in a prospective matched cohort study. This is based on the hypothesis, that electrical injuries more often result in mental health problems compared to other injuries.

Material and methods

Results

We identified 20,155 electrical injuries in DNPR and 1,810 in DWEA (Fig 1 and Table 1, previously published in Biering et al. [39] available from: https://oem.bmj.com/content/78/1/54). After exclusion of persons < 18 years, persons without a valid identification-number and persons who died within the first 2 days after the accident, there was an overlap of 817 persons from the two registers. Invalid identification-numbers could originate from tourists or migrant workers with a temporary number as well as mistyping in DWEA. When removing the overlap and keeping only the first electrical injury for each person we had 13,317 from DNPR and 795 from DWEA for the injury matches. For the occupation match we furthermore excluded 2,646 persons not in the workforce. We gave priority to DWEA registrations, if there was a double registration, and thus we had 10,764 injuries from DNPR and 702 injuries from DWEA available for the occupation match. A match with 10 match persons was possible for almost all electrical injuries.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1.

https://doi.org/10.1371/journal.pone.0247317.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Exclusions and outcomes for each of the three matches.

https://doi.org/10.1371/journal.pone.0247317.t001

The majority of the injuries happened to men, especially in DWEA, and younger persons were overrepresented, most evident in DNPR. The occupations with most injures were craft workers, but also service workers/sales workers were overrepresented, even when comparing to the distribution of occupations in Denmark. The length of hospitalization was in most cases less than one day. Over the study period, we noticed that the number of electrical injuries in DNPR increased, while the opposite is the case in DWEA (Biering et al. [39]). For each outcome, we excluded those who had experienced the outcome before the injury. The numbers of these exclusions are provided in Table 1. Frequencies of persons with the outcomes during 5 years follow up and during full follow-up are also provided in Table 1. The numbers reflect that some of the outcomes are very rare, which affects the precision of the estimates in the following tables.

We provide the risk estimates in separate tables for each match (Table 2). Due to some of the outcomes being rare, the estimates are not always possible to obtain or have wide confidence intervals. We provide the results for these underpowered analyses, but encourage cautious interpretation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Associations between electrical injuries and outcomes over the whole study and in intervals (electrical injuries matched with dislocation/sprain injured, eye injured and occupation controls).

https://doi.org/10.1371/journal.pone.0247317.t002

We did not find increased risk of Alzheimer’s or Dementia in any of the three matches, however, there was an exception for the analysis of only the working population in match 1 and 2, that showed increased risk of Alzheimer’s. In match 3, there were also a hazard ratio of 2.13, however, not statistically significant (Table 2).

We found increased risk of ‘mental disorders due to known physiological condition’ in all three matches and furthermore found that the risk was highest shortly after the injury in match 1, but highest after around 2 years in the other two matches. All estimates had wide confidence intervals, especially shortly after the injury.

We found only limited associations between electrical injuries and depression. Again, there was a tendency that the estimates were highest shortly after the injury and slightly higher for the working population compared to the full population.

In all three matches, there was an increased risk of ‘anxiety and adjustment disorders’, especially ‘Post Traumatic Stress Disorder (PTSD)’ that showed extremely high estimates shortly after the injury in match 1 and 2. In match 3, there were too few cases to estimate risk at 6 and 12 months, but after 2 years the risk was still extremely high.

Risk of ‘Somatoform disorders’ seemed to increase over time in the two first matches, while the same pattern was difficult to identify in match 3, although the associations also were high (OR and HR above 2.3 in all time points and in the time to event analysis).

Diagnosis of ‘other non-psychotic mental disorders’ seemed to be associated in the time to event analysis in the two first matches, but not in match 3, but also here the confidence intervals were wide.

Finally, we found ‘sleep disorders’ associated with exposure to electrical injuries, but the risk estimates were in general smaller and not consistent over time. The risk was higher in match 3 and in the subgroups of the working population. We did not identify increased risk of ‘sexual dysfunction’.

In general, all estimates were highest in match 3, where the match persons were controls without injury, but with the same occupation. In match 1 and 2, the risk estimates were often higher when restricting the analysis to the working population only.

The sensitivity analyses performed on match 1 (Table 3) showed that when we restricted the analysis for hospitalizations one day or longer, most of the risk estimates were slightly attenuated. Excluding persons who were given a diagnosis for concussion during the same stay as the electrical injury did not change the estimates. When we stratified the analyses on gender, we found that the risk of ‘mental disorders due to known physiological condition’ was higher for women compared to men, while the risk for ‘somatoform disorders’ was higher for men compared to women.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Sensitivity analyses (electrical injuries matched with dislocation/sprain controls).

https://doi.org/10.1371/journal.pone.0247317.t003

Discussion

This is, to our knowledge, the first study using a matched design based om population based registers to investigate the association between exposure to electrical injury and mental disorders, and thus it is difficult to compare directly to previous studies. We studied 14,112 electrical injuries identified over a period of 19 years. We found that several of the examined outcomes were associated with exposure to an electrical injury compared to the matched controls. Some of the outcomes showed the strongest associations shortly after the injury, namely ‘mental disorders due to known physiological condition’, ‘anxiety and adjustment disorders’, especially the ‘Post Traumatic Stress Disorder (PTSD)’ subgroup. ‘Depression’ showed weaker association and only shortly after the injury, while others took time to develop (somatoform disorders), or were only present in the time to event analysis (‘other non-psychotic mental disorders’ and ‘sleep disorders’). Although the relative risk of some of the outcomes was high, this does not necessarily reflect a high absolute risk. Many of the outcomes were rare, despite a large amount of exposed persons.

Restricting the analysis to persons from the workforce generally attenuated the estimates slightly in match 1 and 2 but with decreased precisions due to limited power. In match 3 with only persons from the workforce and not controls with another accident, the estimates were also higher compared to match 1 and 2. The gender differences in the exposure to electrical injuries are well-know, but the differences in risk for men and women in mental disorders due to known physiological condition and somatoform disorders point in different directions and further research in this topic is needed.

A limitation of the study was the lack of registration in DNPR in the early years of the study. Probably only a low proportion of electrical injuries were registered, especially during the first years of the study. If the types of electrical injuries that were registered differed in type, severity or duration from those not registered, this may cause bias in an unknown direction. Even today, underreporting may be a problem, since the detailed registration may be deprioritized in acute situations. If the electrical injuries not registered in DNPR are the more serious injuries, where the consequence of the injury, i.e. a burn, was registered instead, and not the code for the electrical injury itself, we may have overlooked the most severe injuries, and thus underestimated the associations. Misclassification of the outcomes is possible in both the electrical injured persons and their match persons. Especially the rare outcomes are vulnerable to misclassification, where only one more person with the outcome in either group could impact the risk estimates to a large extent. Some of the outcomes may be prone to risk of misclassification as the exposure may be interpreted as part of the outcome. This is relevant for the PTSD outcome, but also for ’somatoform disorders’ and ’mental disorders due to known physiological condition’ where physical or cognitive symptoms are reported from the person as a consequence of an electrical injury and other diagnoses are not relevant. In the part of the cohort derived from DWEA, we have no reason to think that there is any difference in the reporting of the exposure, although the number of injuries decreased over time, which is also the case in other types of working injuries. The severity and other characteristics of the electrical injury were not registered, since the definition of an electrical injury was based on the ICD-10 code in DNPR and type of injury in DWEA. Other previous studies have distinguished between high and low voltage injuries. Radulovic et al. found no differences in neuropsychological sequelae, although high voltage injuries causes more physical damage [7], while Rådman et al. found higher incidence of sleep disturbances, anxiety and fatigue in high voltage injuries and of concentration difficulties, sleep disturbances and anxiety in ‘no-let-go’ injuries [36]. We tried to accommodate this by restricting the analysis to persons with a length of hospitalization of at least one day as a proxy for severity, and found that most estimates attenuated, meaning that the risk of mental diseases were higher in those persons that were most affected by the injury.

The hospitalization was in most cases very short. A large proportion of electrical injuries registered in DWEA either did not cause hospitalization or caused only an outpatient visit. This is surprising because the definition of an occupational accident that should be reported to DWEA is an accident that has caused at least sick-leave the day after the accident, and thus of a certain severity. On the other hand, it is possible that some persons were seen by their general practitioner, and thus not registered in DNPR.

Another limitation was the choice of match persons. We found it difficult to identify the perfect type of injury to match an electrical injury. Since electrical injuries are heterogeneous in severity, we should optimally use a similar heterogenetic group and at the same time it should be a rather frequent type of injury to be able to find a sufficient amount of suitable match persons. Furthermore, we should use only matching diagnoses that are not suspected to cause the outcomes of interest. Our solution was to use three different types of match persons: persons with dislocation/sprain injury, persons with eye injury and finally persons with the same type of work as the injured persons. The first two matches had the disadvantage that the injuries were not in themselves life-threatening or disabling as an electrical injury can be. This would cause us to overestimate the frequency of our outcomes, especially regarding PTSD since been exposed to a traumatic event is per definition part of the outcome. On the other hand, those match persons shared the characteristic that they also visited a hospital for an injury the same year and thus shared the same health care seeking preferences as the injured persons.

Match 3 using persons with the same type of job had the disadvantage that the match persons did not have any registered injury, and thus were most likely not in the health care system at the time of the match. This implies that the estimates in this match are most likely overestimated, if injured persons had other habits in relation to health care seeking. In this case, we could not solve this possible bias by adjustment for length of hospitalization. The same approach was used in another Danish cohort study that matched electrical injured persons with random controls from the general population using age and gender with cardiac diseases and mortality as outcomes [37]. We aimed to match with other injured persons (match 1 and 2) to avoid using too healthy controls, but also to take socioeconomic position into account when matching with occupation controls (match 3).

The size of the study was the largest possible using Danish data. Even though DNPR was established in 1977, information about electrical injuries was not sufficiently registered before the initiation of ICD-10 in 1994. Since we chose to use two years of observation time before the accident to exclude persons with the outcome of interest and two years of observation time for new outcomes to occur, we were limited to 19 years from 1996 to 2014. Despite the long period and the considerable amount of injuries, we still had problems with limited power in some analyses, especially with rare diagnoses and when the outcome had occurred before the injury. The latter could cause bias if the outcome was related to a previous unregistered electrical injury. However, since the matching was performed using the injury year, both electrical injured persons and controls were at the same risk of overlooking previous exposure and outcome and thus the bias in most likely non-differential.

Our findings were in line with several previous studies with more retrospective or casuistic design, and the reviews concluding on these. We were not able to compare the size of the risk estimates to previous studies, since the control groups and outcomes from previous studies are not comparable to our design. However, we were able to confirm the findings of both cognitive and psychological problems described in previous studies [4, 8, 11, 22].

The findings related to Alzheimer’s Dementia were surprising, especially that increased risk was found when restricting the analysis to the working population. A possible explanation is that the disease was already in development at the time of the injury and other examinations following the injury caused the disease to be diagnosed before it would have been diagnosed without the health contact caused by the injury. On the other hand, only very few diagnoses were given during the first 5 years following the injury, and thus not related to health contacts following the injury.

Some of the chosen outcomes could be related to the electrical injury itself, i.e. PTSD where the diagnosis is based on a specific traumatic event and ‘mental disorders due to known physiological condition’ could possibly be a diagnosis coded as the physician knows about the exposure to an electrical injury and thus codes cognitive problems according to this.

The findings from this study can be generalized to populations with the same access to hospital treatment and/or a similar system for registration of working injuries. Moreover, the excess risk for mental disorders is most likely also present in other kinds of populations, due to the matched design that reduce confounding [38]. As already suggested by Radulovic et al, more focus on this topic as well as on rehabilitation is warranted [7]. A study in patients from an outpatient burn department found that patients were to a large extend referred to other specialist departments, but often with no clear conclusion or diagnosis [20]. This was in line with our finding from this cohort study, were we found that electrical injures patients had more contact with general practitioners as well as more sick leave and poorer work participation [39].

Conclusion

This study confirms that exposure to an electrical injury increase the risk of mental disorders both at long and short term. Some of the outcomes were rare, thus the high relative risk is not reflected in the absolute risk which is low. To our knowledge, this is the first study to examine mental disorders in a controlled design.