Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
April 2004

Photokeratitis and UV-Radiation Burns Associated With Damaged Metal Halide Lamps

Author Affiliations

From the Epidemic Intelligence Service, Epidemiology Program Office (Drs Kirschke and Smith) and the National Center for Environmental Health (Dr Smith), Centers for Disease Control and Prevention, Atlanta, Ga; the Tennessee Department of Health, Nashville (Drs Kirschke and Jones); and Vanderbilt University School of Medicine, Nashville (Drs Jones and Schaffner). Dr Kirschke is now with the Department of Internal Medicine, James H. Quillen College of Medicine, East Tennessee State University, the James H. Quillen Veteran Administration Medical Center, and the Tennessee Department of Health, Northeast Tennessee Regional Health Office, Johnson City.


Copyright 2004 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2004

Arch Pediatr Adolesc Med. 2004;158(4):372-376. doi:10.1001/archpedi.158.4.372

Background  Damaged metal halide lamps are known to cause outbreaks of photokeratitis and UV-radiation burns among children and adults, which has prompted the Food and Drug Administration, Rockville, Md, to publish consumer recommendations to prevent such injuries. We investigated 3 outbreaks of photokeratitis and UV-radiation burns in gymnasiums associated with failure to heed these recommendations.

Objective  To determine the cause of the outbreaks and promote interventions to prevent further injuries.

Design and Setting  A cohort study of persons exposed to damaged metal halide lamps during the index outbreak in a community gymnasium and a descriptive epidemiologic study of 2 subsequently identified outbreaks in other gymnasiums.

Participants  A total of 273 persons potentially exposed during events in 3 gymnasiums.

Main Outcome Measure  Photokeratitis with onset within 12 hours of the event. The intensity of UV radiation was measured, and an occupational exposure standard applied.

Results  Investigation of the index outbreak identified 18 (approximately 3%) persons who met our case definition for photokeratitis. The median incubation period was 7 hours, and health care visits were reported by 11 persons (61%). Of the 18 patients, 17 (94%) were seated in the back of the gymnasium. Among 37 persons sitting in this high-risk area, the attack rate was 46%. Only 1 (9%) of 11 persons wearing glasses or contact lenses with UV-radiation protection in the high-risk area developed photokeratitis (relative risk, 0.15; P = .01). The safe occupational exposure limit in the high-risk area was 10 to 15 minutes, but exposures of 1 to 3 hours were reported. Prevention recommendations had not been instituted at any of the 3 facilities.

Conclusions  Injuries from metal halide lamps are avoidable, but prevention recommendations may not be widely observed. All facilities using metal halide lamps in areas where children may be exposed should follow the Food and Drug Administration recommendations; amending the National Electric Code may be warranted.

Photokeratitis is damage to the corneal epithelium from UV radiation (UVR) from natural or artificial sources.1 Well-recognized examples include snow blindness and welders' arc burns. Peak symptoms typically occur 6 to 12 hours after exposure and include photophobia, foreign-body sensation, tearing, and blurry vision. Characteristic clinical findings include conjunctival injection and punctuate erosions of the corneal epithelium with an intrapalpebral distribution. Another characteristic finding is UVR burns of the exposed skin around the eyes. Photokeratitis usually resolves spontaneously in 24 to 48 hours with symptomatic treatment.1 Clusters of photokeratitis among children and adults associated with damaged mercury-vapor or metal halide lamps, which are commonly used to light gymnasiums, have been reported.210

Metal halide lamps contain an inner arc tube, analogous to a welder's arc, that emits intense UVR along with visible light (Figure 1). The UVR is normally attenuated by the outer glass envelope so that UVR exposure is negligible. However, a UVR hazard can occur when the outer glass envelope is broken, as by a flying object, and the inner arc tube continues to function and emit UVR that is no longer attenuated by the glass envelope. Nevertheless, injuries from metal halide lamps are preventable. Because they are radiation-emitting devices, the Food and Drug Administration (FDA), Rockville, Md, has issued consumer recommendations for safe operation, which must be displayed on lamp packaging.11,12 However, nonadherence to these prevention measures may be widespread.

Figure 1.
Image not available

Photograph of a 400-W metal halide lamp. The inner arc tube, which contains mercury vapor and other gases, emits intense UV radiation along with visible light. The outer glass envelope of an intact lamp attenuates the UV radiation.

In February 2003, a local health department notified the Tennessee Department of Health, Nashville, of 8 persons who reported severe eye symptoms after attending a fund-raising event in a nonprofit youth center gymnasium. An epidemiologic and environmental investigation was performed. As a result of publicity from this outbreak, 2 other outbreaks of photokeratitis were reported and investigated. One occurred 2 weeks later at a high school that had hosted a 3-day wrestling tournament in its gymnasium; the other was among volleyball players at a municipal gymnasium. This article describes the findings from these 3 outbreak investigations.


We performed a cohort study of persons who attended the fund-raising event at the youth center gymnasium. A case was defined as acute eye symptoms (including ≥3 of the following: conjunctival injection, burning or itching, photophobia, foreign-body sensation, tearing, blurry vision, periorbital edema, and skin erythema) with onset within 12 hours of the event occurring in persons who attended it and who had no preexisting eye symptoms. Case finding consisted of identifying and calling at least 1 representative from each table and all persons in the area where most injuries occurred (defined as the high-risk area). A questionnaire was administered to all persons contacted to determine the location of their table, whether they had any of the above symptoms, and whether they wore eyeglasses or contact lenses with UVR protection. Consent was obtained, and the medical records of those who sought medical attention were reviewed.

To determine the dose of UVR to which persons were exposed, we measured the effective irradiance, or biological effectiveness of UVR, from a damaged metal halide lamp in the high school gymnasium. The outer glass envelope, which normally attenuates UVR, was broken and completely missing. Evaluation of effective irradiance could not be performed at the youth center gymnasium because the damaged lamp was unavailable. However, all of the gymnasiums had high-bay lighting fixtures with aluminum reflectors and used 400-W, R-type metal halide lamps, which were 6.2 m above the floor. Therefore, data collected at the high school gymnasium were considered to be representative of effective irradiance during the index outbreak. Measurements were taken with a radiometer/photometer (IL model 1400A; International Light Inc; Newburyport, Mass), which spectrally matches the UVR hazard bandwidth (190-400 nm) to generate effective irradiance in microwatts per centimeter squared. Measurements were taken at 1.3 m above the floor (to simulate the height of a seated person's head) at a point directly below the damaged lamp (center point) and at 1-m intervals off the center point. Measurements were taken with the radiometer probe oriented horizontally (0°) toward the center point to simulate line of sight. To estimate the hazard to children and adults, effective irradiance was converted to occupational exposure limits by using a table published by the American Conference of Governmental Industrial Hygienists, Cincinnati, Ohio.13 Exposure time required to reach minimal erythemal dose (MED) was calculated by using the formula 5.83 µW/cm2 = 1 MED/h. Statistical analyses were performed using χ2 tests calculated with Epi Info 2002 software (Centers for Disease Control and Prevention, Atlanta, Ga).14


An estimated 600 persons attended the evening fund-raising event, which lasted 2 to 3 hours (6-9 PM), in the youth center gymnasium. Attendees were seated at tables arranged throughout the gymnasium for a catered meal and presentations by several speakers, including a nationally recognized sports figure. Of the approximately 600 persons who attended, we contacted 119 ( approximately 20%), including at least 1 person from each of the 52 prepaid tables and 10 general admission tables along the back of the gymnasium. All persons contacted completed the questionnaire.

Eighteen persons (approximately 3%) met our case definition for photokeratitis (Figure 2). Among the 18 patients, 13 (72%) also had UVR burns on their faces, most commonly on the forehead or eyelids. In addition, we identified 2 persons who were wearing UVR protective eyeglasses who had UVR burns on their faces without eye symptoms. The duration of exposure for most persons with photokeratitis (94%) was 2 to 3 hours; 1 person was present for only 1 hour. The median incubation period was 7 hours (range, 4-11 hours). Health care visits were reported by 11 persons (61%); 4 were awakened from sleep and visited an emergency department before 4 AM after the event, and 7 visited an ophthalmologist the next day, resulting in missed work. Punctate keratitis was noted in the medical records of 8 patients. Further characteristics of the persons with photokeratitis are listed in Table 1.

Figure 2.
Image not available

Time of onset of photokeratitis in 18 persons after the fund-raising event. The x-axis represents time in hours after the end of the fund-raising event (indicated by the arrow). The y-axis represents the number of cases. The clustering of cases suggests a point-source outbreak with an incubation period of 5 to 11 hours.

Image not available
Characteristics of 18 Persons With Photokeratitis From Index Outbreak

Of the 18 patients, 17 (94%) were seated at or standing near 1 of 5 adjacent tables in the back of the gymnasium. The other patient was seated at the front of the gymnasium looking toward the back. We contacted all 37 persons sitting in this high-risk area and performed a cohort analysis. They were all white individuals, and 28 (76%) were men. The attack rate for persons sitting in this high-risk area was 46%. Wearing glasses or contact lenses did not significantly reduce the risk of developing photokeratitis. However, only 1 (9%) of 11 persons in the high-risk area who were wearing glasses or contact lenses with UVR protection developed photokeratitis, compared with 16 (62%) of 26 persons who were not wearing them (relative risk, 0.15; P = .01) (Figure 3). The person who developed photokeratitis despite wearing glasses with UVR protection was affected only in the corner of his left eye in an area not covered by the glasses.

Figure 3.
Image not available

Schematic drawing of persons in the high-risk area at a fund-raising event. The circles represent persons seated or standing around tables in the high-risk area. Pink-shaded eyes represent photokeratitis, shaded circles represent UV-radiation (UVR) burns, and black-shaded eyes represent UVR-protective eyeglasses or contact lenses. Only 1 person wearing UVR-protective lenses developed photokeratitis (*) and only in the corner of 1 eye in an area likely not covered by the lenses.

The gymnasium was lit by 20 high-bay lighting fixtures with metal halide lamps and aluminum reflectors. When the investigators arrived at the gymnasium 2 days later, no damaged lamps (ie, broken outer glass envelopes) were identified. However, an employee verified that a lamp with a broken outer glass envelope had been replaced since the event. This lamp was located above and in the line of sight of persons at the 5 tables in the high-risk area as they looked toward the stage. The aluminum reflectors of several lighting fixtures throughout the gymnasium were dented or tilted, indicating possible damage from flying objects. The 5 high-risk tables were estimated to have been 3 to 6 m off center from a point on the floor directly below the implicated lamp.

The environmental investigation demonstrated that the peak effective irradiance, which is a measure of the biological effectiveness of UVR, was 5.2 µW/cm2 at 4.5 m off center with the sensor in a horizontal position (Figure 4). When an occupational standard is applied, that effective irradiance corresponds to an occupational exposure limit of 5 to 10 minutes according to the American Conference of Governmental Industrial Hygienists.15 The effective irradiance at 6 m off center was 3.8 µW/cm2, which corresponds to an occupational exposure limit of 10 to 15 minutes. When the effective irradiance measured during the environmental investigation was converted to MED, it was determined that MED would be reached by 0.7 hour at 3 m and 2.1 hours at 6 m in a person with fair skin.

Figure 4.
Image not available

Effective irradiance and occupational exposure limits associated with damaged metal halide lamps. The bars represent the occupational exposure limits15 in hours associated with the effective irradiance (line) measured directly under a damaged metal halide lamp at the gymnasium and at 1.5-m intervals off center. Persons in the high-risk area of the index outbreak were seated 3 to 6 m off center, which corresponds to 10- to 15-minute occupational exposure limits.

As a result of publicity from this outbreak, we were notified of and investigated 2 other outbreaks of photokeratitis. The first occurred 2 weeks later at a high school that hosted a 3-day wrestling tournament in its gymnasium. It was noted that 1 of the metal halide lamps in the gymnasium was damaged. We interviewed 149 (about 40%) of the approximately 300 to 400 persons who attended the tournament. Among the 149 persons, 46 (31%) met our photokeratitis case definition and had exposure times of up to 12 hours. Spectators were disproportionately affected. Among the 46 persons with photokeratitis, 24 (52%) also reported UVR burns to the face. The most common eye symptoms were conjunctival injection, burning, and photophobia, and 24 (52%) persons sought medical care.

Another outbreak occurred among volleyball players at a municipal gymnasium. We interviewed 5 persons who played together who had symptoms consistent with photokeratitis or UVR burns on 3 occasions during a period of 3 months after playing volleyball at the gymnasium. The duration of exposure on each occasion was estimated to have been 1 to 3 hours. An investigation initiated after complaints by patrons revealed a damaged metal halide lamp over the volleyball court.


We investigated 3 outbreaks of photokeratitis and UVR burns associated with exposure to UVR from damaged metal halide lamps in gymnasiums used primarily by children. The intensity of the UVR measured during the environmental investigation of the second outbreak was such that the occupational exposure limit would have been 10 to 15 minutes in the high-risk area, which is the same limit as during the index outbreak at the youth center; the actual exposure times were 2 to 3 hours. The attack rate in the index outbreak was 46% for photokeratitis and 41% for UVR burns for persons in the high-risk area. Our cohort study demonstrated that UVR-attenuating eyeglasses or contact lenses were protective against photokeratitis.

Photokeratitis from acute UVR exposure during these outbreaks produced severe eye symptoms that resulted in emergency department and ophthalmologist visits and missed work. Although its exact role in the development of cataracts is debated, animal studies have demonstrated cataractogenesis after acute laboratory exposure to UVR.15 In addition, an association between lifetime UV-B exposure and cataract formation has been shown in a human epidemiologic study.16 Some persons exposed to UVR during the outbreaks we investigated also experienced UVR burns, often resulting in desquamation. Although the risk of cataracts or skin cancer from acute exposures such as described in these outbreaks is unknown, any preventable increased risk warrants attention. The Centers for Disease Control and Prevention, Atlanta, Ga, has recommended minimizing exposure to UVR, especially among children.17

The 3 outbreaks described here, combined with reports by the Centers for Disease Control and Prevention and the FDA,6,18 suggest that injuries and outbreaks related to damaged metal halide lamps may be more common than currently appreciated. Indeed, on a second visit to the high school that hosted the wrestling tournament, a damaged metal halide lamp was discovered that had not been recognized by the maintenance staff even with heightened awareness following the outbreak. Metal halide lamps that do not self-extinguish are commonly used in open fixtures in gymnasiums where they are at risk for damage by flying objects. Once damage has occurred, if the broken glass is not reported to the proper person, the problem can be difficult to detect because lamps may continue to function normally. Disincentives may exist for children to report damaged lamps, such as disciplinary action. Furthermore, many facilities do not have the proper equipment on site, such as a mechanical lift, to reach the 6-m-high fixtures to change a damaged lamp. Even after a damaged lamp is discovered, it might not be recognized as a health hazard and therefore might not be replaced until it fails, is replaced during scheduled maintenance, or causes injuries.

Injuries and outbreaks related to metal halide lamps are likely underreported, especially in children. Since the problem is not well recognized, when sporadic injuries occur, the possibility of exposure to a damaged metal halide lamp may not be as apparent as when an outbreak occurs. Sporadic injuries or small clusters might be attributed to infectious causes of conjunctivitis or rash illness in children.

To prevent these injuries, the FDA recommends that either enclosed fixtures or self-extinguishing lamps be used in areas where persons may be exposed for prolonged periods.12 Enclosed fixtures would protect the lamp from damage and, in the event of damage, the glass or plastic lens would attenuate the UVR. Self-extinguishing lamps are designed to stop functioning within 15 minutes if the outer glass envelope is damaged and may be used in open fixtures, but such lamps are more expensive than those that do not self-extinguish and so are not routinely used or widely available. The National Electrical Manufacturers Association, Rosslyn, Va, has also recently published recommendations for schools.19

The fact that outbreaks associated with metal halide and mercury-vapor lamps continue to occur indicates a failure of the current level of recommendations. Ultimately, a change to the National Electric Code might be required and has been proposed by the National Electrical Manufacturers Association. All facilities that use metal halide lamps where persons may be exposed for extended periods of time should follow the FDA or National Electrical Manufacturers Association recommendations.12,19 Injuries or outbreaks associated with metal halide lamps should be reported to the state or local health department and the FDA (Center for Devices and Radiological Health, Office of Compliance, MS HFZ-342, 2098 Gaither Rd, Rockville, MD 20850, telephone: 301-594-4654).

What This Study Adds

Several outbreaks of photokeratitis and sunburn associated with damaged metal halide lamps were reported in the late 1970s, and as a result, the FDA required warning labels on these lamps. However, our investigation demonstrates that outbreaks of photokeratitis and rash continue to occur and may be more common than previously appreciated. Because damaged metal halide lamps are often recognized only after they are linked to a health complaint, increased awareness of this hazard is necessary to correctly diagnosis this preventable cause of keratitis and rash in school-aged children, which can be mistaken as an infectious process. The FDA prevention recommendations are infrequently implemented in facilities used by children, which has prompted the National Electrical Manufacturers Association to petition that these recommendations be reflected in the National Electric Code.

Back to top
Article Information

Corresponding author and reprints: David L. Kirschke, MD, Tennessee Department of Health, Northeast Tennessee Regional Health Office, 1233 Southwest Ave Extension, Johnson City, TN 37604-6519 (e-mail:

Accepted for publication November 17, 2003.

We thank Beth Collier, RN, Middle Cumberland Regional Health Department, Nashville, Tenn, and Mary Ellen Chesser, RN, Williamson County Health Department, Franklin, Tenn.

Parrish  CMChandler  JW Corneal trauma. Kaufman  HEBarron  BAMcDonald  MBeds.The Cornea Boston, Mass Butterworth-Heinemann2003;633- 671
Rose III  RCParker  RL Erythema and conjunctivitis: outbreak caused by inadvertent exposure to ultraviolet light. JAMA. 1979;2421155- 1156
Betts  LSSmilkstein  MJRumack  BHCwinn  A Unrecognized ultraviolet-induced keratoconjunctivitis. Ann Emerg Med. 1986;151378
Thun  MJAltman  REllingson  OMills  LFTalansky  ML Ocular complications of malfunctioning mercury vapor lamps. Ann Ophthalmol. 1982;141017- 1020
Halperin  WAltman  RBlack  KMarshall  FJGoldfield  M Conjunctivitis and skin erythema: outbreak caused by a damaged high-intensity lamp. JAMA. 1978;2401980- 1981
Centers for Disease Control, Conjunctivitis caused by malfunctioning mercury vapor lamps: Michigan, New Jersey. MMWR Morb Mortal Wkly Rep. 1979;28121- 123
Asbury  JF Sore eyes due to accidental exposure to ultraviolet B. Lancet. 1984;21163;
Elliott  BSilverman  PR Keratoconjunctivitis caused by a broken high-intensity mercury-vapor lamp. Del Med J. 1986;58665- 667
Przyjemski  CJ Keratoconjunctivitis from ultraviolet light. JAMA. 1980;243732- 733
Telles  NCCoakley  JMKluger  JA Mercury vapor lamp injuries. JAMA. 1975;234149
Mercury vapor lamps may be dangerous, says the FDA, JAMA. 1979;2421837
Not Available, High-intensity Mercury Vapor Discharge Lamps. 21CFR §1040.301997;
American Conference of Governmental Industrial Hygienists (ACGIH), Ultraviolet radiation. TLVs and BEIs: Based on Threshold Limit Values for Chemical Substances and Physical Agents and Biological Exposure Indices. Cincinnati, Ohio ACGIH2003;157- 160
Not Available, Epi Info 2002 [computer program]. Version 1 Atlanta, Ga Centers for Disease Control and Prevention2002;
Sliney  DH UV radiation ocular exposure dosimetry. J Photochem Photobiol B. 1995;3169- 77
Taylor  HRWest  SKRosenthal  FS  et al.  Effect of ultraviolet radiation on cataract formation. N Engl J Med. 1988;3191429- 1433
Saraiya  MGlanz  KBriss  PNichols  PWhite  CDas  Dfor the Task Force on Community Preventive Services on Reducing Exposure to Ultraviolet Light, Preventing skin cancer: findings of the Task Force on Community Preventive Services on Reducing Exposure to Ultraviolet Light. MMWR Recomm Rep. 2003;521- 12
Not Available, Hazards of mercury vapor lamps. JAMA. 1978;2402144Article
National Electrical Manufacturers Association, Recommendations for the care and maintenance of high intensity metal halide and mercury vapor lighting in schools. Available at: June 2, 2003