Phototesting is used to evaluate the skin's response to light and attempt to establish the spectrum of radiation that will cause adverse reactions or inhibit them.1

A phototesting session can determine a) the minimal erythema dose (MED), which is to say the lowest dose of UV radiation able to cause visible erythema; b) abnormal responses to UV-A light; and c) abnormal responses to visible light. Although phototesting guidelines have recently become available,2 the UV irradiation doses able to induce erythema in healthy individuals and standardized MED doses for testing have not been well established in the literature.3–6 Physicians who undertake phototesting, therefore, may disagree about how to read and interpret the results. For practical purposes, the MED thresholds described by Fitzpatrick7,8 are usually considered reference values for assessing an individual's MED according to phototype.

Lack of homogeneity, both in the use of light sources and dose measurements, between the various members of the Spanish Photobiology Group (GEF) has prevented the development of a common phototesting protocol in the recent past.9

This study's aim was to standardize the phototesting protocol, analyze MED values in a healthy Spanish population by phototype, determine MED threshold values below which an individual could be considered photosensitive, and assess the doses that may trigger abnormal responses to UV-A radiation in this population.

Nine hospitals participated. Four of them used solar simulators (Solar Light S12 or S16), 3 used BBUVB lamps, and 2 used both types of light source (Table 1). The 9 centers recruited 232 volunteers in total.

Most volunteers were women (69%). There were no statistically significant differences (P>.05) in mean age between the sexes (women, 37.9 [14.3] years, range 21–76 years; men, 37.4 [13.8] years, range 21–74 years).

Phototypes assigned by dermatologists using their usual system were mostly II and III (89 and 108 volunteers, respectively) (Fig. 2). The dermatologists classified 4 volunteers as phototype I and 31 as phototype IV. The classifications based on self-reporting (questionnaire) were distributed as follows: I, 3 volunteers; II, 65; III, 124; and IV, 40. When phototype was based on self-assessment, fewer were assigned to phototype II and more to phototypes III and IV.

Figure 2.

Number of volunteers in each Fitzpatrick phototype, according to the dermatologists’ assignment of type or the type based on self-reporting of items on the Fitzpatrick questionnaire.

(0.1MB).

The dermatologist-assigned phototypes were the ones used to guide choice of a MED test series.

Figure 3A shows mean MEDs for each phototype as recorded at centers using solar simulators. MED increased significantly with phototype at all centers (P<.05, one-way ANOVA). No between-center differences were found on comparing the mean MED findings for phototypes II or III (P>.05, one-way ANOVA; MEDs for phototype II, 28[4] and for phototype III, 35 [4]mJ/cm2; Table 3). The MEDs for phototype IV were very similar in all centers (mean, 51 [6]mJ/cm2) although a higher value was seen in a center that recruited a single volunteer with this phototype. For phototype I, the MEDs were significantly different (P<.05) between the 2 centers that tested phototype-I volunteers; the mean MED was 23(8)mJ/cm2.

Figure 3.

A, Adjusted MED by center. B, Adjusted mean MED by phototype when solar simulators were used for phototesting. MED refers to minimal erythema dose; H, hospital; UMA, Universidad de Málaga. Complete hospital names are given in Table 1.

(0.2MB).

The average MED values are shown in Table 3 and displayed graphically in Fig. 3B by phototype for all the hospitals that used solar simulators. The MED thresholds below which positivity was considered to indicate photosensitivity are also included. The mean MED values were significantly different between phototypes (P<.05, one-way ANOVA), and there was a positive correlation between mean MED and phototype.

No between-center differences were found on comparing the mean findings of BBUVB phototesting for phototypes II or III (P>.05, one-way ANOVA), but for phototype IV, 1 center had significantly higher MED values (P<.05, one-way ANOVA) (Fig. 4A).

Figure 4.

A, Adjusted mean minimal erythema dose (MED) by center. B, Adjusted mean MED by phototype when broadband UV-B lamps were used for phototesting. MED refers to minimal erythema dose; H, hospital; UMA, Universidad de Málaga. Complete hospital names are given in Table 1.

(0.22MB).

The mean MED values by phototype for all the hospitals that tested with BBUVB lamps are shown in Table 4 and displayed graphically in Fig. 4B. The MED thresholds indicating photosensitivity in BBUVB phototesting are also included. No statistically significant differences in mean MED values or photosensitivity thresholds were found between phototypes II and III.

None of the volunteers in the study displayed abnormal reactions to UV-A exposure.

The phototesting protocol in this study could be followed regardless of whether a hospital uses a solar simulator or a BBUVB lamp as the light source. As a result, we are able to propose phototype-appropriate irradiation ranges that have been validated in different centers. The data can also be used to establish the MED values that, if positive, should be interpreted as an abnormal erythema response for each phototype.

One of the strengths of this study is that we took an important aspect—calibration—into consideration in the first phase. We compared and calibrated the equipment used in the different participating centers, following a suggestion made some years ago.14 The participating centers’ equipment was grouped into 2 categories in this study, even though the devices within each category differed by manufacturer and model. Xenon arc lamps (solar simulators) comprised one group and fluorescent lamps (emitting BBUVB light) comprised the other. Although the devices have evident spectral differences, both these UV light sources have been widely used in many studies and are accepted by the European dermatology forum.2,15

Spectral analyses of all the light sources used for calibration facilitated the normalization of values in the test series. By weighting each light source's emissions with values for the erythemal action spectrum adopted as a reference by the CIE,11 we were able to normalize incremental doses of exposure in terms of irradiance. Because measurements were normalized, phototests with solar simulators and BBUVB devices could be compared, even though the solar simulators used different dose intervals.

The dose thresholds at the upper and lower limits of the test series were defined for each phototype, and 6 geometrically increasing dose steps were set, as in protocols used in other studies.16,17 This 6-step approach, with upper and lower limits set according to phototype, favors the central MED values and increases the likelihood of accurately detecting photosensitivity (by an abnormally low MED). By focusing the series on the central MED thresholds, the 2 highest points are generally positive for erythema and the individual's MED usually falls between the third and fourth step, as the 2 lowest steps do not usually produce erythema in normal skin. The threshold dose for indicating photosensitivity would be found between these 2 lower points.

In the second phase of the study the volunteers were tested in all the participating centers using the study protocol; no field work of this type had ever been undertaken prior to this study.

Most volunteers had phototype II (28.1%) or III (53.3%) skin, and their results were highly similar in all the centers. Volunteers with phototype I skin made up less than 2% of the study population, while 17.1% had phototype IV skin. No previously published data are available to indicate the prevalence of these phototypes in the Spanish population, but these frequencies are clearly valid for the population studied. Given the similarity of phototype distributions and MED values recorded in centers in different geographic locations within Spain, we can probably conclude that the data are representative of the entire Spanish population, although they might well differ considerably from data for populations outside Spain.3,4

Phototype assessment criteria are far from objective. We therefore applied 2 different methods for classifying volunteers by phototype. One was the dermatologist's usual method. The other was the self-report questionnaire leading to classification on the Fitzpatrick scale. Classifications were consistent between the 2 methods for phototypes II and III, although the responses to the questionnaire tended to suggest a slightly higher classification. This difference shows that phototype may be open to question when indicating a phototesting protocol to follow or a therapy to apply.

MED values varied greatly for individuals with the same phototype classification, consistent with many other reports in the literature3–5 and possibly attributable to a variety of factors, such as subjectivity in reading the response or the influence of ambient light. Other confounding factors might be the choice of different anatomical sites for testing and the history of recent sun exposure, but these are sources of bias we tried to minimize when designing the present study.

The results for volunteers tested with solar simulators were very similar to the MEDs established by Fitzpatrick for phototypes I through IV. However, while subjects tested with BBUVB lamps also had mean MED findings within the same ranges, their values differed less markedly between phototypes. It is noteworthy that very similar MEDs were recorded for individuals classified within phototypes III and IV who were tested with BBUVB lamps, suggesting that solar simulator MED testing is more sensitive to change of phototype, probably attributable to the size of dose increases between steps in the series. Larger intervals between steps will lead to lower sensitivity for establishing an individual's MED, and in fact the difference between the lower and upper threshold doses in the BBUVB protocol are larger, as shown in Table 2.

Variability in MED results within the same phototype must lead us to reflect seriously when trying to objectively assess clinical phototest findings. The first question is whether a MED reading bears a clear relationship to phototype. Phototesting gives a result that is said to be measurable and reproducible, but phototype classification itself is based on questions posed by the dermatologist to a healthy subject or a patient, or on self-reporting by the individual; hence it is subject to a higher index of subjectivity and error. We found a lack of correspondence between certain dose intervals and phototype in MED testing. Stated in other words, the MED values found for different phototypes overlapped, consistent with observations from an earlier study.5 The lack of correspondence may be attributable to accumulated error in the assignment of phototype (because of subjectivity and reliance on questionnaires). Alternatively, the cause may simply be that the dose at which erythema develops in response to UV-B radiation, which is tested objectively, is unrelated to the factors used to assess phototype. Nonetheless, the mean values for pooled data from all the centers showed a clear relationship to phototype assignment by the dermatologists, and there was little overlap in our population, in which data showed little dispersion.

A second goal of phototesting is that it identify a photosensitive subject whose MED is abnormally low in cases in which we do not already know what that individual's MED would have been under normal circumstances. In fact, the literature does not establish MED thresholds below which photosensitivity can be diagnosed. It must also be remembered that a large percentage of individuals with photodermatoses (nearly 70% in some series) have phototest results interpreted as normal.18 One of the most important aspects of the present study, therefore, is our proposal for erythemal threshold values below which an observed MED would indicate photosensitivity within a 95% CI; in other words, the probability of observing an abnormal response within the range of normal values would be less than 5% if this test protocol is followed. Although the small number of volunteers with phototypes I and IV in this study might be considered problematic, the consistency of results from all the different centers (Fig. 3B) argues against this being a limitation. In any case, the real clinical application of the values in the test series we propose (Tables 3 and 4) will tell us if these data are in fact reliable.

Finally, none of the volunteers we tested had an abnormal response to UV-A. This exploratory series therefore seems to have been appropriate and gave results that are consistent with those reported by other authors,18 who set the cutoff for an abnormal response at a level of irradiance of less than 20J/cm2.

In conclusion, this study used normalized units of measurement and effective erythema dose step increases across the different light sources used by the participating centers. The results recorded at the centers were pooled to give standardized MED values according to phototypes that were consistent with those of Fitzpatrick. We also proposed a MED threshold indicative of pathologic response for each phototype, defining phototesting series for detecting abnormal reactions to UV-A irradiation. Finally, we note the need for further study to improve the testing series used for BBUVB lamps in order to increase the sensitivity of MED measurements between phototypes.

The authors declare that they have no conflicts of interest.