figure a

Introduction

Fireworks associated with celebrations, such as Diwali, Chinese New Year, and New Year’s Eve (NYE), are a well-described cause of ocular trauma and preventable vision loss [1,2,3,4,5]. The rate of severe vision loss from fireworks-related ocular trauma has been reported at up to 1 in 6 fireworks injuries, with enucleation rates of 3.9% reported in a worldwide systematic review in 2010 [6]. Males, bystanders, and young children were the groups most affected by fireworks-related ocular trauma. Ocular trauma and the resulting severe vision loss have significant long-term social, economic, psychological, and physical consequences for the patient and the community [7,8,9]. Up to 90% of ocular trauma is preventable [10], with effective interventions including education, policies, legislation, and the use of eye protection.

Eye injury registries have effectively collected data from a large number of centers [11, 12]. This has facilitated international comparisons and informed the development of global consensus guidelines and/or regional co-ordination of efforts to prevent ocular and adnexal trauma [13]. For example, data from the “European Registration Fireworks Eye Injuries” was used to inform legislation such as the ban on bottle rocket fireworks in Norway in 2007, which was based on a high incidence of vision loss from these fireworks [14]. Legislation requiring a complete ban on the personal use fireworks has been shown to reduce fireworks-related injuries [6]. The International Globe and Adnexal Trauma Epidemiological Study (IGATES) is an ongoing, large multi-center study that has provided a methodology for data collection for ocular trauma. IGATES and the European Fireworks Registry have the potential to provide data on factors associated with severe vision loss to help inform the development of preventive strategies, such as legislation for fireworks-related ocular trauma.

Previous studies have identified the nature and incidence of ocular trauma associated with fireworks for countries including the Netherlands [15], China [3], India [1], Northern Ireland [16], the UK [17], the USA [18, 19], and New Zealand [20]. However, these studies are limited by their relatively small size, limiting their ability to understand severe vision loss from fireworks-related ocular trauma in different countries.

More data is required to understand the factors associated with severe vision loss to help provide targeted strategies to prevent vision loss from fireworks-related injuries. The objective of this study was to report on the factors associated with severe vision loss, from fireworks-related ocular trauma during celebrations, for a large dataset from four countries.

Methods

To capture data from a large number of patients, centers and countries, information from two sources was combined. The IGATES group and The Dutch Ophthalmic Society (Netherlands Oogheelkundig Gezerschap, NOG) have both developed systems for prospective data collection for fireworks-related ocular trauma. Fireworks-related ocular trauma resulting in hospital emergency department or ophthalmic clinic presentations were identified by attending physicians based on the patient or their carers’ description of the mechanism of injury for both groups.

Members of the Asia Pacific Ophthalmic Trauma Society (APOTS), which includes ophthalmologists with a specific interest in ocular trauma from the Asia-Pacific and a number of other countries, including the USA and Argentina, were invited to volunteer to participate in the study. A total of 12 tertiary eye care centers agreed to participate, from 98 centres approached, with each given a unique set of login details to enter patient data. Members of the NOG included 598 active ophthalmologists across the Netherlands from eye clinics and hospitals, who used the European Registration Fireworks form. Data from 12 tertiary eye care centers including 8 from India, 2 from Nepal, and 2 from Argentina were collected using the IGATES Fireworks online form (Appendix 1). The tertiary eye care centers comprised 3 government and 9 private institutions (Appendix 2). Participating centers were asked to fill in the form for all patients meeting the inclusion criteria specified below, with the treating ophthalmologist responsible for data entry. This study was conducted in accordance with the tenets of the Declaration of Helsinki. Ethics approval for the study was obtained from the NOG for the Netherlands study and from the relevant local or regional institutional ethics committee for each of the IGATES centres (see IGATES Fireworks study participants).

IGATES Fireworks data collection

A subset of IGATES—IGATES Fireworks—was developed to collect data on the circumstances and outcomes from fireworks-related eye and adnexal injuries. IGATES Fireworks provided a Web-based “smart form” on a secure encrypted platform with definitions for each category [21].

All patients meeting the inclusion criteria were given a unique code, with identifiers including date of birth and name omitted for patient anonymity. Data collected at presentation and at follow-up (between 3 and 5 weeks) included patient demographics, information about the circumstances including the type of fireworks and any precautions such as eye protection and supervision, the type of injury, treatment received, and visual outcomes. Clinical images where available were uploaded (Appendix 1). Data collection was not mandatory for the participating centers but clinicians were encouraged to include data for all cases that meet the criteria on the registry.

Inclusion and exclusion criteria

Patients with the following clinical features were included:

  1. (a)

    Open-/closed-globe injury, and/or

  2. (b)

    Adnexal/orbital involvement, and/or

  3. (c)

    Intraocular/intraorbital foreign body

  4. (d)

    Chemical/thermal injury to the eye or adnexa

Patients presenting for management of long-term sequela of fireworks injuries, such as traumatic cataract, were excluded.

NOG data collection

Nederlands Oogheelkundig Gezerschap (NOG) collected data for patients treated in the Netherlands for eye or adnexal fireworks-related injury the 4 days before and 4 days after NYE since 2008.This data has been reported in the literature [22]. All ophthalmologists in the Netherlands were asked to fill in the form for all patients meeting the inclusion criteria during the specified period, and share it with NOG who collated the data. The European Registration Fireworks Eye Injuries form was used for data collection. Variables collected included patient date of birth and gender, incident details, type of injury and treatment (Appendix 2).

Definitions

Vision loss was defined in line with World Health Organization (WHO) with mild vision impairment as best corrected visual acuity (BCVA) worse than 6/12, moderate vision impairment as worse than 6/18, severe visual impairment as worse than 6/60, and blindness as worse than 3/60. Patients defined as having “severe vision loss” in this study were those with severe visual impairment or blindness, that is vision 6/60 or worse at 4-week follow-up. [23]

Data analysis

The demographic and clinical data were summarized, and the distribution of patients across centers tested for homogeneity using Pearson’s Chi-square test for categorical data and Student’s t-test for continuous data. The statistical significance of factors associated with severe vision loss was assessed using Marginal homogeneity test [24].

The effect of demographic parameters like age, gender, setting, location, and clinical parameters, i.e., grade of corneal burn, presence of penetrating, or lens injury on vision status, i.e., severe vision loss or not, at the time of presentation, was determined using multiple logistic regression. The association of demographic and clinical parameters on VA at presentation was assessed through bivariate and multiple logistic regression analysis.

Vision change was regarded as a dichotomous outcome (improvement or no improvement/deterioration) and modelled against the baseline status of clinical parameters using binary logistic regression. All the analyses were performed using SPSS version 20.0 (IBM Corp, Armonk, USA), and the statistical significance was tested at 5% level.

Results

From APOTS, twelve members volunteered to participate in the study and from NOG 96% of members participated in the survey. The participating ophthalmologists completed the form for all patients meeting the inclusion criteria over the study period.

A total of 388 patients from the IGATES and NOG databases met the inclusion criteria, including patients from the Netherlands (n = 132), India (215), Argentina (44), and Nepal (18). Vision at presentation was available for 229 patients and at 4-week follow-up for 151 patients, with complete data on clinical parameters available for 122 patients (Fig. 1).

Fig. 1
figure 1

Sample data distribution for the study

Full size image

Visual outcomes

From 388 patients, 71 (18.3%) patients had severe vision loss after fireworks-related ocular trauma at follow-up (Table 1). The proportion of patients with severe vision loss due to fireworks-related ocular trauma at 4 weeks was highest in India (n = 46, 21.4%), followed by the Netherlands (21, 15.9%), Nepal (3, 16.7%), and Argentina 1 (4.4%).

Table. 1 Comparison of presence and absence of severe vision loss at final presentation (4 weeks) across countries (n = 388)
Full size table

For all patients with VA data at presentation and 4-week follow-up, a statistically significant improvement in vision at follow-up was observed (p value < 0.0001). Of patients, 83.5% (126/151) had an improvement in VA at follow-up, suggesting a significant overall improvement in vision after treatment. However, 2 patients who had moderate vision loss at presentation had severe vision loss at follow-up.

Comparing vision at presentation to final visual outcomes, of 15 patients with vision worse than 6/60 at presentation, 10 (66.7%) had an improvement in their vision (Table 2). For 61 patients with vision worse than 3/60 at presentation, 32 (52.5%) had improvement. Of 15 patients with mild vision loss at presentation, 11 (73.3%) had no vision loss at 4 weeks, while of the 47 patients with moderate vision loss at presentation, 15 (31.9%) had no vision loss, and 14 (29.8%) had mild vision loss at 4 weeks. Phthisis bulbi was recorded in 3 patients (3/93, (0.03%).

Table. 2 Comparison of BCVA at presentation and after 4 weeks (n = 151). Shaded cells highlight cases where there was improvement in vision at 4 weeks compared to presentation
Full size table

The analysis revealed that patients with a CGI had a significantly higher likelihood of their vision improving compared to those with an OGI [OR of 3.667; 95% CI 1.096–2.27; p = 0.04] (Table 3). None of the other clinical parameters were associated with a significant effect on the odds of vision improvement.

Table. 3 Effect of different demographic and clinical parameters on BCVA due to fireworks-related ocular trauma at presentation (n = 122)
Full size table

Demographics and circumstances

None of the circumstances of injury or patient demographics were significantly associated with a patient with final visual outcome of 6/60 or worse. Table 4 shows the demographic profile of patients with a fireworks-related ocular trauma by country. Demographic characteristics were available for 256 patients from India, Argentina, and Nepal from IGATES. These patients had a mean age at presentation of 20.6 years (range 2 to 83 years), with the lowest mean age reported in Nepal (12.4 ± 11.4 years, 0 to 38 years) and highest in India (21.1 ± 14.8 years, 2 to 63 years). The difference in the mean age across countries was statistically insignificant (p = 0.0653).

Table. 4 Demographic profile of patients and circumstances associated with fireworks related ocular trauma across different countries (n = 256)
Full size table

A male predominance of almost 4:1 was seen for the IGATES cohort. Nepal had the lowest (2.4:1), and India had the highest male/female (M:F) ratio (4.5:1); however, the difference was not significant (p = 0.3785). Gender was not associated with an increased risk of blindness at presentation (OR 1.385, 95% CI 0.549,3.495; p = 0.49).

When stratified by age, the 15- to 20-year age group had the highest M:F ratio (9:1). A significantly higher number of patients sustained their injury in urban versus rural setting (66%; p = 0.0013), and the highest proportion of injuries occurred at home (178, 69.5%).

The type of firework was identified for 95.7% patients and included atom bombs (32, 12.6%), loose crackers (30, 11.8%), sparkling fountain (flower pots) (16, 6.3%), and air bombs (12, 4.7%).

The majority of fireworks-related eye injuries in India and Nepal were recorded during Diwali festival (n = 203 and 16, respectively), while in Argentina (18), they occurred during NYE celebrations.

Eye protection — including spectacles, goggles, and face shields — were reported to be used in 7 (2.7%) of the IGATES patients and did not differ by setting (p value 0.684). For patients < 18 years (n = 151), adult supervision at the time of the firework injury occurred more often in Argentina (10/15), than India (30/122) and Nepal (3/14) (Table 4). This difference in supervision across countries was statistically significant (p value = 0.0001).

More than half (133, 52.2%) of the injuries were to a bystander, including 90 (35.3%) where the firecracker was lit by a relative or friend. The majority (98.4%) of injuries in IGATES patients were unintentional, with 2 injuries of uncertain intent and potentially self-harm.

Presenting vision and clinical features

In total, there were 219 (85%) CGI and 34 (13.2%) OGI (Fig. 2 a and b). Where clinical and demographic data were available in the IGATES cohort (n = 122) (Table 3), the nature of injuries were predominantly mechanical (32%) or thermal (25%). In bivariate analysis, CGI had a significantly lower associated risk compared to OGI [OR of 0.102; 95% CI 0.027–0.386; p = 0.0008]. Other clinical parameters including corneal injury, corneal burns, sclera injury, and conjunctival injury had associated OR of more than 1.0, although these were not statistically significant (Table 5).

Fig. 2
figure 2

a Type and distribution of closed globe injuries (%). b Type and distribution of open-globe injuries

Full size image
Table. 5 Odds of visual improvement associated with different clinical characteristics at 4 weeks after treatment using binary logistic regression (n = 120)
Full size table

The presence of a penetrating injury, lens injury, and vitreous injury were associated with a significantly higher risk of blindness at presentation (OR of 16.81 [95% CI 5.451–51.838; p < 0.0001], OR of 14.52 [95% CI 5.697–36.997; p < 0.0001], OR of 4.855 [95% CI 2.058–11.453; p = 0.0003], respectively). Retinal injury, such as macula damage suffered by the patient featured in Fig. 3a, had associated odds of 4.125 [95% CI 1.832–9.287; p = 0.0006] toward blindness. Parameters such as adnexal injury, eyelid injury, and orbital injury were associated with a reduced odds of severe vision loss, but this was statistically insignificant.

Fig. 3
figure 3

a Right eye contusion and blunt ocular trauma with macular damage after being hit by Roman Candle firework. Final VA 20/200. b Surgical removal of gunpowder (as a result of fireworks exposure) from conjunctival fornix under general anaesthesia to prevent chemical injury of a right eye. Final visual outcome 20/200. Images provided by Jan de Faber. Figure 3 a appeared on the front page of NRC Handelsblad, the daily evening newspaper published in the Netherlands by NRC Media, on the 30th of December 2013 with patient consent obtained

Full size image

Significant clinical parameters were included in the multiple logistic regression model. CGI had significantly reduced odds of severe vision loss [OR 0.192; 95% CI 0.04–0.916; p = 0.04] after adjustment. The presence of penetrating injury or lens injury continued to be significantly associated with an increased risk of blindness with OR of 4.874 [95% CI 1.298–18.304; p = 0.02] and 7.023 [95% CI 2.378–20.736; p = 0.0004] respectively in the multivariate model

Surgical procedures

A total of 93 surgeries were performed on 62 patients from the IGATES cohort (average 1, range 0–3). Surgeries included corneal, scleral, or conjunctival repairs (21/93, 22.6%) and cataract surgery (18/93, 19.4%). Removal of gunpowder residue was also conducted for one patient (see Fig. 3b).

Discussion

Severe vision loss at 4-week follow-up due to fireworks-related ocular trauma was recorded for 18.3% patients, with the highest rate of severe vision loss in our cohort recorded in India (21.4%). Severe vision loss from fireworks-related ocular trauma during celebrations was found to be associated with a range of factors including worse presenting BCVA, an OGI penetrating injury or lens injury. No association was identified with severe visual loss and reported use of eye protection nor supervision of patients < 18 years old by adults.

The proportion of patients with severe vision loss was dissimilar to prior studies. Full visual recovery following fireworks-related eye injury in 61.5% patients has been reported in the Netherlands [15] and vision of 6/18 or better in 77.5% of children in India [1]. A lower rate of full visual recovery in our study may be related to delays in treatment, with children more represented in our study. Children may not report injuries for fear of repercussions, but no evidence is available to support this conclusion. Further, our follow-up period was shorter than other studies, averaging 6 to 26 weeks; there may be further improvement after 4-week follow-up. It may also be that the clinicians in our study were more inclined to record severe cases of ocular trauma in the registries.

Gender has been previously identified as a risk factor for fireworks-related ocular injury, with a male preponderance of 6.8:1 to 8.7:1 [15, 25]. Our study, however, found a lower rate of male preponderance, 4.1:1. The lower rate of males injured may relate to participation in celebrations being family-related and more likely to be gender balanced. The average age of patients from India, Argentina, and Nepal was 20.6 years, which is similar to previous studies where average age reported was 22 [15] to 25 years [3]. In our cohort, 15- to 20-year-olds were more likely to present with vision < 6/60 but age was not significantly associated with the final visual outcome.

Prevention measures, including education about the hazards associated with fireworks, use of eye protection and legislation have been used with the aim of reducing fireworks- related ocular trauma. Programs in the Netherlands have promoted the use of eye protection whilst using fireworks, with eye protection provided during peak fireworks use periods [15] resulting in a reduction in eye injuries [22]. However, eye protection use was not commonly reported in this study and it is possible that that those wearing eye protection were not injured therefore our dataset maybe biased. Low numbers of patients reporting use of eye protection may have been related to many injuries occurring at home or during leisure, where awareness and access to eye protection is limited. The low rate of use could also be as the history of eye protection was not taken and/or recorded in the medical records by the clinician. More data needs to be gathered to fully understand the influence of preventive measures on incidence of eye injuries from fireworks.

In patients aged 18 years and under, a low number of presentations were associated with adult supervision (17.8%). This finding is in contrast to previous studies that have reported higher numbers of fireworks-related ocular trauma in children [3] despite the presence supervision in more than 50% of cases [26]. Further studies are needed to establish the effect of adult supervision on reducing fireworks-related injury in children. This could become an important message in public awareness campaigns to reduce fireworks-related ocular injury in children.

Fireworks-related ocular traumas are rarely associated with public fireworks displays [17], rather are more commonly seen in and around either the home or at a friend or relatives place. In our study, more than three quarters of the injuries occurred in the patient’s or a relative’s home. Legislation to reduce or prevent the private sale and use of fireworks has reduced the rates of fireworks-related ocular traumas in many countries and regions, including Australia and the USA [6]. In the Netherlands, consumer fireworks (category 1, 2, and 3) are allowed to be sold 3 working days before NYE to persons older than 16 years for use from December 31, 6.00 PM, to January 1, 2.00 AM. Roman candles, little bottle rockets and launching tubes for rockets are banned in the Netherlands. In India, where concerns include pollution and noise, a ban on fireworks use outside of 8 pm and 10 pm during Diwali and 11.45 pm and 12.15 am during Christmas and NYE was called by the Supreme Court in 2018 [27]. In the state of Tamil Nadu, India, fireworks are restricted for use 6 am to 7 am and 7 pm to 8 pm during Diwali. Australia has almost entirely eliminated personal use of fireworks, except for in the Northern Territory where sale and use is legal for a limited period around Territory Day [28]. What is clear from the Australian example is that any relaxation of fireworks legislation relating to private use even for a short period of time results in injuries including to the eyes with resulting vision loss [28].

Argentina’s legislation varies by territory, and in Buenos Aires where most of our data was collected, penalties include fines and arrest for manufacture, transport, storage, and/or sale of pyrotechnic devices without proper authorization. Efforts to ban all fireworks in Argentina, especially in the greater urban area surrounding the capital, were resisted by pyrotechnic manufacturers on the basis that the ban was unconstitutional (Suprema Corte de Justicia Provincia de Buenos Aires) [29]. Our data have shown that fireworks-related ocular trauma still occurred during festival times despite legislation in these countries.

Categorization of fireworks, whilst not harmonized internationally, is generally based on the level of hazard and noise. In Europe, fireworks are classified into four categories based on the danger to the user. Categories range from F1, which pose very little danger, e.g., sparklers, to F4 which pose grave danger and are intended for use by professionals with specialized knowledge. This system could help purchasers of fireworks for personal use assess the risk. Not enough information was available in our data set to determine the classification of fireworks.

In addition to legislating and restricting the sale and personal use of fireworks, alternatives, e.g., drones for lighting displays and support for professional fireworks, should be encouraged. Several organizations including the World Health Organization, National Fire Protection Association and American Academy of Pediatrics, International Fire Marshalls Association have supported bans on fireworks for private use. Despite a complete ban on fireworks they are illegally imported into Nepal resulting in ocular trauma. Banning specific products related to a higher incidence of ocular trauma has been successfully implemented in some countries, e.g., bottle rockets are banned in Norway and the Netherlands.

Limitations

With no active follow-up conducted, a large proportion (37.8%, n = 99/256) of patients did not present after initial treatment and were lost to follow-up. Further, visual potential information was missing for 85 patients (33.2%) from the baseline visit and VA at follow-up missing for 107 (41.8%) patients. Further, follow-up was relatively short at an average of 4 weeks, whereas further vision improvement may occur after this time. Information regarding the circumstances of injuries relies on self-reported data from patients and their carers, which may have introduced reporting or selection bias.

As patients with severe clinical presentations are more likely to present and return for further follow-up, this may have also resulted in reporting bias, leading to an over representation of severe injuries. Information bias could have arisen with potential differences in data collection methods; this was likely reduced as ophthalmologists recorded and entered the data. Nonetheless, as data entry into the registry was not mandatory there may have been a bias toward reporting more severe cases. Data not missing at random is likely to include the presence or absence of safety measures (such as use of eye protection which was not recorded for 2 patients ) and presenting VA, where VA could not be measured, e.g., in very young patients, both of which should be incorporated into our conclusions.

Our data was collected around the time of the festivals, which varied between the IGATES study and the NOG data. For example, the NOG data included patients within the period 4 days before or after NYE with the assumption that the use of fireworks during this period was associated with NYE celebrations. The geographic locations associated with our data collected are vastly different, which could have led to a potential bias in access to healthcare, legislation, and enforcement. Any reported difference in prevention measures may have been biased by a difference in access to eye protection in each of the countries.

Conclusion

Fireworks-related ocular and adnexal trauma associated with celebrations in a variety of countries can lead to severe vision loss, with factors including penetrating injury and lens injury increasing the risk. Countries should adopt prevention measures, including education, on the risks associated with the use of fireworks, improved supervision of children and legislation to limit their use.