Ultraviolet Radiation (UVR) from Fluorescent Lamps

In recent months, media articles have focused public interest on the issue of whether ultraviolet radiation (UVR) from fluorescent lamps poses a health hazard. As always when UVR is being considered, skin cancer is the health effect which is of most concern to people.

Fluorescent lamps are one of a series of products that produce UVR, which is emitted from low pressure mercury vapour. The mercury vapour emits UVR when an electrical discharge is passed through it - most of the energy emitted is at a wavelength of 254 nm. This lies in the UVC portion of the spectrum (180-280 nm). In the case of fluorescent lighting, the 254 nm radiation is used to excite a phosphor which coats the inside of the glass envelope of the lamp. The phosphor will re-emit at visible wavelengths (different phosphors produce different colours), and any UVC which is not absorbed by the phosphor will be absorbed by the glass wall of the lamp. However, the mercury discharge will also emit at other wavelengths - notably at 365 nm, which lies in the UVA (315-400 nm). This UVA radiation may not be absorbed by the phosphor, and much of it will pass out through the lamp walls into the environment.

In 1988 NRPB undertook a detailed series of measurements to determine the ultraviolet irradiance levels likely to be encountered in the workplace and the home where fluorescent lighting is used (see Table 1) ¹. Lamps of three colour types that were commonly available from manufacturers (denoted A, B and C in the study) were used, and were measured at the ages of 0, 100 and 2000 hours. All measurements were normalised to give the UVR irradiances at a lighting level (illuminance) of 500 lux, which is considered typical for office workers.

Table 1

		UVA as % of exposure limit		UVB as % of exposure limit
Lamp colour	Lamp age (hours)	1988 limit	1996 limit	1988 and 1996 limit
white	0	0.3-0.4	9.2-11.6	6.9-7.9
	2000	0.3	7.5-9.2	4.7-5.1
cool white	0	0.4-0.5	11.6-13.7	5.8-9.2
	2000	0.3-0.4	9.5-11.1	4.1-5.7
warm white	0	0.3	7.7-8.8	4.6-9.3
	2000	0.2-0.3	6.2-7.7	3.8-5.3

Assessments were made of the possible potential risk for the induction of acute effects (photokeratitis, erythema) by reference to the suggested UVR exposure limits then in use ². Assessments were also made of the possible potential for the induction of non-melanoma skin cancers (NMSC). The maximum measured UVB (280-315 nm) was 93 µW m ^-2 (ACGIH weighted), which represented only 9.2% of the 1988 limit for the avoidance of acute effects. The maximum measured UVA was 48 mW m ^-2, which represented only 0.5% of the 1988 limit. Expressed in terms of the current limits ³, this would represent 13.7% of the current UVA limit. There is no current limit for UVB in isolation, but 93 µW m ^-2 (weighted) is still only 9% of the current UVR limit ³. All of these levels would be reduced by the use of diffusers.

At the time of the 1988 study, fluorescent lamps were typically mounted on open battens with the option to attach a plastic diffuser over the lamp. These diffusers, if fitted, will reduce UVR emissions. Depending on the material used in the diffuser, UVA would be reduced by 17-99% and effective UVB by 19-100%. In recent years, the desire for more energy efficient lighting has led to increased use of so-called 'type II' luminaires. These incorporate a silvered reflector behind the lamps. As the reflector helps to produce a less directional light, type II luminaires are commonly used without plastic diffusers.

Earlier this year, in response to renewed public concern arising from use of type II luminaires, NRPB carried out a study similar to that of 1988. This study was based on a sample of readily available lamps, and was intended to allow comparison of the general range of lamps available to the public in 1988 to that available now. The study was not intended to identify changes in the output of any of the manufacturers' products. This time, the sample of lamps drew on four manufacturers, of which two also figured in the 1988 study (see Table 2). Not all lamp colours were represented for each manufacturer. For this study, the lamps were mounted on an open fronted type II luminaire.

Table 2

	Manufacturers
Lamp colour	In 1998 study	In 1997 study
white	A, B, C	D
cool white	A, B, C	A, B, E
warm white	A, B, C	E

The results indicate that fluorescent lamps still do not present an acute hazard (see Table 3). The apparent increase in UVB emitted by modern white and warm white lighting may be due to the smaller sample size of the 1997 study, only one type of each of these lamps being used as opposed to samples from all three manufacturers in 1988. The 1997 samples were also drawn from manufacturers not represented in the 1988 study. The cool white lamps, of which three types were used in both studies, show a reduction in the range of both UVA and UVB emissions. Two of the cool white lamps came from manufacturers represented in the 1988 study. For one of these manufacturers, UVB emissions had fallen and, for the other, emissions had increased since 1988.

Table 3

		UVA as % of 1996 exposure limit		UVB (ACGIH weighted) as % of exposure limit
Lamp colour	Lamp age (hours)	1988 lamps	1997 lamps	1988 lamps	1997 lamps
white	0	9.2-11.6	8.5-9.1	6.9-7.9	9.9-12.4
	2000	7.5-9.2	7.1-10.3	4.7-5.1	6.2-8.5
cool white	0	11.6-13.7	9.0-11.0	5.8-9.2	3.4-12.8
	2000	9.5-11.1	7.8-10.2	4.1-5.7	1.0-6.8
warm white	0	7.7-8.8	6.9-9.9	4.6-9.3	10.5-11.4
	2000	6.2-7.7	6.2-8.8	3.8-5.3	6.3-7.7

In order to estimate the potential for chronic health effects, the data were weighted using the CIE reference erythemal action spectrum ^{4, 5}. Exposures from different sources could then be compared in terms of the number of some standardised dose quantity delivered. In the original study ¹, the standardised dose quantity was a so-called 'minimal erythemal dose' (MED), which was defined as 300 J m ^-2 (effective). The concept of MED was originally developed as a clinical measure of an individual's threshold for erythema, and avoided the need for the measurement of physical quantities. In order to relate the quantity of erythemally effective radiation to its potential for harmful effects in a population, average values of MED have since been used. At the time of the original study, 300 J m ^-2 (effective) was felt to be an appropriate value for a north European population, but since then a range of lower values - 150-200 J m ^-2 (effective) - has been used. Recently, it has been suggested ⁶ that the MED should be restored to purely clinical use, and a standardised erythemal dose (SED) equivalent to 100 J m ^-2 (effective) should be used to quantify the likely health effects to population groups.

The number of MEDs delivered by the lamps in the 1997 study over 2000 hours seems to show a greater range of results, with perhaps more MEDs delivered by unaged lamps in particular (see Table 4). However, if only data from aged lamps are considered (most lamps in service will not be new), the number of MEDs delivered has risen from an average of 4.6 to 6.0. This should be compared to a typical range of exposure from solar radiation for indoor workers in the UK population of about 300 SED annually ⁷.

Table 4

		MEDs * delivered (all manufacturers)
Lamp colour	Lamp age (hours)	1988	1998
white	0	6.5-7.4	8.7-10.6
	2000	4.6-5.0	5.7-8.0
cool white	0	6.0-8.2	9.0-9.9
	2000	4.1-5.3	1.1-6.4
warm white	0	4.6-7.4	3.6-11.1
	2000	3.6-5.0	5.9-8.7

* 1 MED = 300 J m ^-2.

For this risk assessment, it is assumed that a person is subject to 157 MED y ^-1 of solar UVR up to age 18, and then (as an indoor worker) is subject to 93 MED y ^-1 of solar radiation. These values have been used in a study of the impact of ozone depletion ⁶ using an MED of 200 J m ^-2. The exposure model assumes a two-week holiday in August in the UK ⁸. The indoor worker is also exposed to UVR from workplace lighting. The increased risk of contracting NMSC from an occupational exposure of 6.0 MED y ^-1 as opposed to 4.6 MED y ^-1 (at 300 J m ^-2 per MED) can be calculated, and is presented in Table 5.

Table 5

	Increase in relative risk from higher occupational exposure
Age at which risk calculated	BCC	SCC
at age 40	1.9%	2.4%
at age 60	2.3%	2.9%

The additional risk of NMSC induction is slight for any type of unfiltered luminaire. The amount by which this risk increases if unfiltered type II luminaires are used is not likely to exceed 2.9%. For people who take holidays in countries nearer to the equator than the UK, this additional risk will be an even smaller percentage of the risk from solar UVR.

In summary, under the conditions of this analysis, it is still possible to adhere to the conclusion of the 1988 study that:

at commonly used illumination levels the UVR emissions presented neither an acute nor a significant chronic hazard ¹.

Acknowledgement

The author wishes to thank Professor Brian Diffey for his advice in the preparation of this article.

Note

In this context of this article, one MED equals two SEDs, which are described in CIE Publication No. 125 (1997).

1 Whillock, M et al. Ultraviolet radiation levels associated with the use of fluorescent general lighting, UV-A and UV-B lamps in the workplace and home. Chilton, NRPB-R221 (1988).

2 ACGIH. Threshold limit values and biological exposure indices, 1986-87. Cincinnati, American Conference of Governmental Industrial Hygienists (1986).

3 ICNIRP. Guidelines on UV radiation exposure limits. Health Phys., 71, No. 6, 978 (1996).

4 McKinlay, A F and Diffey, B L. A reference action spectrum for ultraviolet induced erythema in human skin. CIE J., 6, 17-22 (1987).

5 NRPB. Health effects from ultraviolet radiation: Report of an Advisory Group on Non-Ionising Radiation. Doc. NRPB, 6, No. 2, 7-190 (1995).

6 Diffey, B L et al. The standard erythema dose: A new photobiological concept. Photodermatol. Photoimmunol. Photomed., 13, 64-6 (1997).

7 Diffey, B L. Human exposure to ultraviolet radiation. In Photodermatology (J I M Hawk, ed). London, Chapman and Hall (in press).

8 Diffey, B L. Stratospheric ozone depletion and the risk of non-melanoma skin cancer in a British population. Phys. Med. Biol., 37, No. 12, 2267-79 (1992).

Last reviewed: 4 September 2008