A simple pain model for the evaluation of analgesic effects of NSAIDs in healthy subjects


Leopold Schmetterer PhD, Department of Clinical Pharmacology, Währinger Gürtel 18–20, A-1090 Vienna, Austria. Tel.: + 43 14 0400 2981; Fax: + 43 14 0400 2998; E-mail: leopold.schmetterer@univie.ac.at


Aims  Non-steroidal anti-inflammatory drugs (NSAIDs) are believed to counteract inflammation and inflammation-induced sensitization of nociceptors by inhibiting peripheral prostaglandin synthesis. We evaluated an experimental pain model for NSAIDs, that included an inflammatory component to mimic clinical inflammatory pain conditions.

Methods  The study was performed in a randomized, double-blind, placebo-controlled, two-way crossover design on 32 healthy volunteers. A small skin area of the proximal upper leg was irradiated with a UVB source using three times the individually estimated minimal erythema dose. Twenty hours after irradiation skin temperature, heat pain threshold and tolerance in sunburn spot were measured using a thermal sensory testing. These measurements were repeated 2 h after medication of either 800 mg ibuprofen as single oral dose or placebo capsules. Effects of ibuprofen on outcome parameters were assessed with analyses of covariance (ancova).

Results  Placebo did not affect heat pain threshold or tolerance. By contrast, ibuprofen increased heat pain threshold by 1.092 °C [confidence interval (CI) 0.498, 1.695; P = 0.0008) compared with placebo. Heat pain tolerance also increased significantly by 1.618 °C (CI 1.062, 2.175; P = 0.0001).

Conclusion  The pain model we evaluated was well tolerated in all subjects and the effects of ibuprofen were highly significant. This model is simple, sensitive to NSAIDs’ effects and therefore has potential for future experimental pain studies.


Non-steroidal anti-inflammatory drugs (NSAIDs) are most commonly prescribed for episodic and chronic pain alleviation and for long-term anti-inflammatory therapy. A large number of active compounds are available and the search goes on for new NSAIDs that would provide some advantages over those currently available such as increased efficacy or decreased risk of untoward effects.

In the last decades a number of experimental pain models involving healthy volunteers for the comparative evaluation of NSAID efficacy have been reported in medical literature [1]. Detectable effects were mainly demonstrated in complicated pain models including an inflammatory component. This is probably an important factor since NSAIDs are believed to counteract inflammation and sensitization of nociceptors by the inhibition of two enzymes, cyclooxigenase 1 and 2 (COX 1, COX 2), resulting in an inhibition of prostaglandins, particularly PGE1 and PGE2, which act locally by promoting inflammation and potentiating the effect of various algogenetic agents. In simple pain models, without inflammation and hyperalgesia, the results were inconsistent and unreliable.

None of the previously realized pain models fulfilled all of the six substantial demands for an ideal pain-induction method [2]: (i) it does not cause tissue damage; (ii) it does not cause psychological injury; (iii) it is simple to perform; (iv) the result is reliable; (v) there are no after-effects; (vi) the subject has control of cessation during the test.

In the present study we set out to modify a promising human pain model which was first applied by Bickel et al.[3]. This model, using UVB radiation to induce inflammation and radiant heat for the assessment of pain thresholds, demonstrated a significant analgesic effect of ibuprofen, but the setting was rather complicated.

In the present study we simplified this pain model with regard to UVB application and the algesimetric testing method and introduced pain tolerance as an additional outcome variable. The effect of 800 mg ibuprofen was tested vs. placebo. Special regard was given to potential gender differences in analgesic efficacy as described for ibuprofen in an electrically induced pain model [4].



The study was approved by the Ethics Committee of Vienna University/Medical School. After informed consent was obtained, 34 healthy, pain-free volunteers (16 male, 18 female) aged between 19 and 35 years participated in the study. Before entering the study, a full medical history was taken and a prestudy examination was performed. Further, subjects were asked to fill out a depression scale questionnaire [5]. Relevant abnormalities resulting from these examinations led to exclusion of the respective subject (no subjects were excluded at this time). All subjects had a body mass index (BMI) in the normal range (between 15th and 85th percentile). Exclusion criteria were as follows: regular use of medication, participation in a clinical trial in the preceding 2 weeks or during the study, history of hypersensitivity to the trial drug or drugs with similar chemical structure, history or presence of gastrointestinal, pulmonary, liver or kidney disease, or other conditions known to interfere with distribution, metabolism or excretion of the study drug, skin diseases that could alter the response to UVB irradiation in the test areas, pregnancy or breast feeding, abuse of alcohol, or addiction to nicotine or caffeine. Subjects agreed to abstain from alcohol, nicotine, and coffeinated drinks during the study period.

One female volunteer withdrew her informed consent after the first study session because of cancerophobia, another female subject developed herpes zoster prior to the first UVB irradiation and was excluded. The two female subjects who did not complete the study were excluded from analysis and replaced by two other female volunteers.


Subjects were admitted to the Department of Clinical Pharmacology at the University of Vienna for a training session and two study sessions which were separated by an interval of 2 weeks.

The training session was scheduled to introduce the subjects to the algesimetric testing procedure and to make them familiar with the testing device. This was done in an effort to minimize potential training effects between the first and second study sessions. Data acquired in this training session were recorded but not included in analysis.

The study was performed in a randomized, double-blind balanced two-way crossover design. The sequence of study sessions was not only balanced for the overall study population and gender, but also – to avoid dextrality effects – for the subjects’ dominant and nondominant side. Each of the two study sessions consisted of two study days (Figure 1).

Figure 1.

Overall subject disposition.

On the first study day in the early afternoon, a small skin area on the subject's upper leg was irradiated using a solar simulating metal halogen lamp (Sellasol; Sellas Medizinische Geräte GMBH, Gevelsberg-Vogelsang, Germany). On the second study day 20 h later, as the cutaneous sunburn reaction had reached its maximum and was stable, pretreatment values for hyperalgesia and inflammation intensity were determined. Post-treatment measurements were done 2 h later. In the two study sessions phototoxic erythema was induced at the corresponding areas of the two body sides.

In both study sessions subjects were given a single oral dose (in two capsules) of 800 mg ibuprofen (2 × 400 mg Avallone®; Novartis Pharma, Vienna, Austria) or placebo in a randomized order. The placebo capsules were identical in appearance and were filled with maize-starch. Drug preparation and administration was done by staff members who did not participate in subject observation or data acquisition. The drug assignments for each subject were prepared by a scientist otherwise not involved in the present study using a randomization program (Statistica; Stat. Soft, Inc., Tulsa, OK, USA). Drug administration was blinded to both subjects and investigators. The algesimetric tests were performed by a single investigator. Data acquisition always took place at the same time in the morning in a quiet air-conditioned room to minimize circadian and air temperature-related influences on pain and perception. The acquired data were entered in a computer. After the study the drug assignments were unblinded for statistical analysis.

Pain model

The minimal erythema dose (MED) of UVB was determined according to the subjects’ skin type [6]. On the first study day of each of the two study sessions subjects were irradiated on a small skin area (diameter 5 cm) of the ventro-medial upper leg with 3 MEDs of UVB irradiation. Phototoxic erythema developed within a few hours and reached its maximum between 12 and 24 h after irradiation [7]. The interindividual variability in the intensity of the erythema response was small. In no case were blisters observed. The only longer lasting effect was tanning of the irradiated skin patches. As the study was performed in late spring and summer all subjects were instructed to avoid additional UV exposure during the study period.

On the second study day, 20 h after irradiation – when the erythema was fully developed and stable – subjects were asked to arrive after sleeping for 7–8 h.

A resting period in supine position of at least 15 min in a quiet air-conditioned room was observed prior to the measurement of skin temperature differences between inflammatory skin patches and the non-irradiated contralateral skin. Temperature was determined with a temperature measurement device (TEMP M1029A; Hewlett Packard) connected to a monitor (Model 66S, M1166A; Hewlett Packard, Palo Alto, CA, USA). Increased skin temperature in irradiated skin was taken as a surrogate for inflammation intensity.

Subsequently, measurements of heat pain threshold and heat pain tolerance were performed with a commercially available thermal sensory analyser (Medoc TSA-2001; Medoc Ltd, Ramat Yishai, Israel). For this purpose a Peltier-thermode with a size of 18 × 18 mm was attached to the irradiated skin area using elastic. In order to minimize variation of probe application pressure, the elastic was wrapped tightly around the upper leg. Then the elastic was stretched by 2 cm and the ends were adhered. Care was taken to consider upper leg curvature in placing the probe in order to achieve optimal contact between the probe and the leg surface. The adaptation temperature was 30 °C and the rate of temperature rise was 0.4 °C s−1. The upper range of the thermode was limited to 54 °C to exclude skin damage. However, none of the subjects reached this maximum temperature.

For determination of the heat pain threshold subjects were instructed to press a response button of the thermal sensory analyser when thermal sensation became unpleasant and aching. Pushing the button resulted in a quick cooling of the thermode until adaptation temperature was reached. The measurement of heat pain threshold was repeated three times in 30-s intervals so that overall four values were obtained. For determination of heat pain tolerance subjects were instructed to press the response button of the thermal sensory analyser when the heat pain sensation became unbearable. Overall three values for heat pain tolerance were obtained.

Heat pain threshold and heat pain tolerance were measured as estimates of the magnitude of hyperalgesia. The mean of heat pain thresholds and heat pain tolerance was calculated from the four and three repeated measurements, respectively.

Statistical analysis

For the analyses of heat pain threshold and heat pain tolerance the difference of the means of post-treatment and baseline values were calculated. To examine the effect of ibuprofen treatment by adjusting for period effect analyses of covariance (ancova) with the group factor treatment (two levels), day of treatment (two levels: 1st and 2nd session), sequence (placebo/ibuprofen and ibuprofen/placebo), sex, the random factor patient within sequence and sex (29 levels) and the covariable mean baseline value was performed for heat pain threshold, heat pain tolerance and skin temperature difference separately. Results are given as difference of means of post-treatment and baseline values ± SD. The ancova model was calculated using proc glm with test option within the SASâ computer package (Release 6.12, Cary, NC, USA).


Thirty-two subjects completed the study. No subject reported pain during the UVB irradiation or spontaneous pain associated with the phototoxic erythema which subsequently developed. In no case were phototoxic blisters observed. Further, no subject reported any side-effect after administration of a single oral dose of 800 mg ibuprofen or placebo.

Heat pain thresholds

The results of the ancova for the heat pain threshold model are shown in Table 1. Ibuprofen treatment increased heat pain threshold by 1.092 °C [confidence interval (CI) 0.498, 1.695; P = 0.0008] compared with placebo (see Figure 2).

Table 1. ancova results for the threshold model.
Variabled.f.Estimates95% CI of the estimateF-valueP-value
Day of treatment 11.128(0.577, 1.679)17.530.0002
Treatment 11.092(0.498, 1.695)14.160.0008
Sequence 10.044(− 0.472, 0.560) 0.01320.91
Patient29   2.3600.012
Sex 10.375(− 0.194, 0.944) 0.860.36
Baseline value 1− 0.435(− 0.754, − 0.117) 7.8090.0091
Figure 2.

Effects of 800 mg ibuprofen (▴) or placebo (▿) on heat pain threshold 2 h after oral intake separated for session 1 and session 2. Results are presented as means ± SD.

During the first study session the mean threshold decreased by − 0.18 ± 1.09 °C in the placebo group and increased by 1.48 ± 1.09 °C in the treatment group. In the second study session the differences of means for heat pain thresholds in the placebo group and the treatment group were 0.82 ± 1.20 °C and 2.17 ± 1.56 °C, respectively. Overall, in the second session, the mean temperature was 1.13 °C higher than in the first session, an effect which was highly significant (Table 1). The significant negative association with the baseline value might be due to slightly higher baseline values when no treatment was given. The sequence sex interaction significantly influenced heat pain thresholds. However, due to multiple tests performed a P-value of 0.037 may well arise by chance and the clinical relevance of this effect is questionable. No gender differences in response to placebo or active treatment were observed.

Heat pain tolerance

The results of the ancova for the heat pain tolerance model are shown in Table 2. Ibuprofen treatment significantly increased heat pain tolerance by 1.618 °C (CI 1.062, 2.175; P = 0.0001) compared with placebo (see Figure 3).

Table 2. ancova results for the tolerance model.
Variabled.f.Estimates95% CI of the estimateF-valueP-value
Day of treatment 10.873(0.340, 1.406)11.250.0022
Treatment 11.618(1.062, 2.175)35.350.0001
Sequence 1− 0.101(− 0.648, 0.446) 0.080.78
Patient29   2.000.0335
Sex 10.146(− 0.423, 0.715) 0.150.70
Baseline value 1− 0.084(− 0.402, 0.233) 0.290.59
Figure 3.

Effects of 800 mg ibuprofen (▴) or placebo (▿) on heat pain tolerance 2 h after oral intake separated for session 1 and session 2. Results are presented as means ± SD.

The mean tolerance in the first study session was − 0.45 ± 0.98 °C in the placebo group and 1.07 ± 1.24 °C in the treatment group. In the second study session mean tolerance increased by 0.22 ± 0.91 °C after placebo and 2.06 ± 1.53 °C after ibuprofen. In the second session the mean increase during ibuprofen was 0.87 °C higher than in the first session. As with heat pain threshold, the day of treatment had a significant impact on the results obtained, but no gender differences were observed.

Temperature differences

Skin temperature differences between irradiated skin areas and unirradiated skin were determined prior to the algesimetric measurements. The treatment, the day of treatment and the baseline value had significant influence on the skin temperature differences (Table 3). The mean temperature difference in the first study session was 0.14 ± 0.83 °C in the placebo group and − 1.04 ± 1.05 °C in the treatment group (Figure 4). In the second study session the mean temperature difference was − 0.51 ± 1.02 °C in the placebo group and − 0.94 ± 0.79 °C in the treatment group. In the second session the mean decrease was 0.58 °C higher than in the first session, an effect which was significant. A negative association between the baseline value and the skin temperature effect was observed. In addition, we observed sex differences for this parameter, which are depicted in Figure 4. It is evident that for skin temperature differences men were more sensitive to ibuprofen treatment than women.

Table 3. ancova results for the skin temperature differences model.
Variabled.f.Estimates95% CI of the estimateF-valueP-value
Day of treatment 1− 0.579(− 1.000, − 0.158) 7.9000.0088
Treatment 1− 0.674(− 1.054, − 0.293)13.120.0011
Sequence 1− 0.251(− 0.629, 0.127) 1.780.19
Patient29   1.040.46
Sex 1− 0.499(− 0.935, − 0.062) 5.320.0256
Baseline value 1− 0.595(− 0.987, − 0.204) 9.670.0042
Figure 4.

Effects of 800 mg ibuprofen (▴) or placebo (▵, ▿) on skin temperature differences between normal skin and phototoxic erythema 2 h after oral intake separated for session 1 and session 2. Up triangles (▵) represent means of male subjects, down triangles (▿) represent means of female subjects. Results are presented as means ± SD.


A variety of experimental pain models involving healthy human volunteers for the comparative evaluation of NSAID efficacy have been reported in medical literature. In cold-induced pain models [8–11], several NSAIDs have not demonstrated consistent analgesic effects. Electrically induced pain models were able to discriminate the effects of diclofenac from placebo [12] and to describe gender differences in response to ibuprofen [4]. Dental pain models inducing pain by removal of impacted or semi-impacted third molars did clearly demonstrate analgesic efficacy of NSAIDs compared with placebo [13–15]. In a laser-induced pain model [16] paracetamol showed significantly higher analgesic potency compared with placebo but lower compared with a combination of paracetamol and codeine. Further, mechanically induced pain models [1, 3, 17], chemically induced pain models [17], freezing injury pain models [17], heat-burn injury pain models [1], and UV pain models using UVA or UVB radiation [3] were used to study the effects of NSAIDs.

In the present study we addressed the anti-inflammatory and analgesic effects of ibuprofen in a modified and simple UVB pain model. UVB-induced erythema served as model for hyperalgesia in healthy subjects and the effect of ibuprofen 800 mg on primary hyperalgesia could be easily detected using a commercially available sensory testing device. Ibuprofen significantly increased heat pain threshold and heat pain tolerance. However, a large interindividual variability, as in most studies focusing on subjective or semiobjective outcome parameters, was observed. In addition, ibuprofen significantly decreased skin temperature differences between inflammatory and normal skin.

It is striking that the day of treatment had significant influence on all three outcome parameters. For heat pain threshold and heat pain tolerance this fact may be interpreted as a training effect that occurred between the first and second study sessions. A study that addressed experimental pain measurement in multiple sessions with the same thermal sensory analyser used in our study reported bias between the first and successive sessions of heat pain threshold tests [18]. However, no bias was found between sessions other than the first. Although we have tried to avoid such effects by making all subjects familiar with the thermal sensory testing procedure prior to the study, it must be noted that this training was carried out before application of the UVB light and the subjects participating in the present study reported that heat pain sensation differed in inflammatory skin compared with normal skin.

We have no explanation for the influence of day of treatment on the temperature-lowering effect of ibuprofen, since in both study sessions UVB irradiation was performed in an identical manner and with the same UVB dose. Moreover, we used the same setting for determination of skin temperature differences. The impact of baseline values on the reduction of temperature differences is expected, because for larger baseline temperature differences the potential of reduction by NSAIDs should also be larger.

A significant influence of gender on the skin temperature-lowering effect of ibuprofen was observed in the present study. Even though baseline differences in skin temperature were consistently higher in women than men, we observed that men were more sensitive to ibuprofen than women, indicating that for men the anti-inflammatory effect of ibuprofen is larger. In an electrically induced pain model gender differences in response to ibuprofen were reported previously. In this study male subjects had greater stimulus thresholds and a greater pain tolerance compared with women and only men exhibited a statistically significant analgesic response to ibuprofen [4].

Electrically induced pain techniques have demonstrated their capacity to distinguish NSAIDs from placebo in randomized placebo controlled trials [4, 12]. Moreover, for codeine a temporal resolution of its analgesic effects was reported [12]. However, electrical stimulation may evoke other sensations than nociceptive pain and is thus more likely to be influenced by drugs acting upon the central nervous system [10]. UVB pain models offer several advantages over other pain models in healthy volunteers. A major issue is the intentional induction of inflammation and hyperalgesia, which is thought to be essential for NSAID pain models [1, 19]. Further, inflammation and hyperalgesia are constant for several hours, representing an advantage not only over freezing, burning, or chemically induced inflammation, but also over painful clinical disorders which often have an episodic nature. UVB pain models are therefore able to test the effects of NSAIDs without high variability and, in the future, they might be used in pharmacodynamic studies on NSAIDs. Sunburn is a condition that nearly everyone experiences once in life. Generally, neither induction nor development or the stable condition cause spontaneous pain. Hence, compared with freezing, burning or chemically induced inflammation and hyperalgesia, UVB pain models should be better tolerated by the volunteers. This is of importance with regard not only to ethical considerations but also to the precision of algesimetric testing [20].

A previous study on heat pain thresholds in a UVB pain model for determining the anti-inflammatory and analgesic effects of ibuprofen reported results which are compatible with the results of the present trial [3]. In this study the individual MED was established by irradiating five circular skin areas with increasing dosages of UVB light prior to each session, a procedure which is time consuming. Heat pain threshold but not heat pain tolerance was used as surrogate parameter for hyperalgesia and determined with infrared radiation from a halogen bulb. The skin surface temperature was feedback controlled by a thermocouple gently attached to the skin. As a surrogate for inflammation intensity the relative blood fluxes in the irradiated skin were assessed with a laser Doppler probe.

Compared with this model the UVB pain model in our study was simplified in that we estimated the MED according to the volunteers’ skin type. This is a time-saving and widely use practice in clinical phototherapy and was found to be suitable for the purposes of our study. Our model is also applicable to non-European Caucasians, with the only difference that in brown and black skin (skin types V and VI) higher UVB doses are required to elicit a phototoxic erythema reaction [21]. In addition to heat pain threshold we measured heat pain tolerance. Further, all these measurements were done with a commercially available thermal sensory testing device. As a surrogate for the degree of inflammation the increased temperature of irradiated skin was compared with normal skin at the contralateral side with a thermal probe. Although we modified the pain model in these matters the sunburn of all subjects was in the intended range and the effects the of ibuprofen were highly significant. Effects of ibuprofen were slightly more pronounced in the previous UVB model [3] than in the present study. Nevertheless, these effects were highly significant in both studies and effects of NSAIDs can be assessed with both approaches.

In this study we focused on the effects of a high dose of ibuprofen to discover if the pain model deserves further investigation. However, it would be important to know whether our pain model is adequate to pick up the time course of the effects and to compare the potency of NSAIDs. Further studies investigating the dose dependence of NSAIDs in the present pain model are warranted to gain information on the sensitivity of the method.

The modified UVB pain model evaluated in this study is simple to perform and sensitive to ibuprofen effects (treatment effects were highly significant for all outcome variables). The well-controlled UVB-induced inflammation and hyperalgesia in small circumscribed areas of the skin is painless (none of the subjects under study reported discomfort due to UVB irradiation). Further, the assessments of heat pain threshold and tolerance were well tolerated by all subjects. The significant influence of the day of treatment on the analgesic effect of ibuprofen indicates a training effect that occurred between first and second study sessions. To avoid this effect it seems essential to train subjects prior to the study in inflamed skin. Our findings, together with physiological models of inflammation and pain, suggest that the peripheral anti-inflammatory activity of ibuprofen is at least partially responsible for the reduction in hyperalgesia.