The study was approved by the local ethics committee. Written informed consent was obtained from each patient or next of kin. The setting was a 12-bed medical-surgical intensive care unit (ICU) at the Oulu University Hospital. Patients were treated according to the normal ICU protocol and current severe sepsis guidelines 14, including hydrocortisone supplementation in septic shock refractory to vasopressor therapy.

All patients admitted from 9 May 2005 to 15 December 2006 with sepsis were considered eligible for the study. Standard definitions were used to define sepsis, severe sepsis, and septic shock 15. Exclusion criteria included age under 18, any bleeding disorder, immunosuppression therapy, surgery not related to the sepsis, surgery during the preceding 6 months, malignancy, chronic hepatic failure, chronic kidney failure, and steroid treatment not related to sepsis.

The following information was collected from all study patients: age, gender, reason for admission to the ICU, focus of infection, severity of underlying diseases on admission as assessed by the Acute Physiology and Chronic Health Evaluation II (APACHE II) and Simplified Acute Physiology Score II (SAPS II) scores, evolution of daily organ dysfunctions assessed by daily sepsis-related organ failure assessment (SOFA) scores, presence of ischaemic heart disease, chronic obstructive pulmonary disease (COPD), diabetes mellitus, and asthma. The lengths of stay in the ICU and in hospital were recorded as well as the ICU, hospital, and 30-day mortalities. For controls, we used 15 healthy Caucasian age-matched volunteers (7 men and 8 women).

We used the suction blister device to create experimental wounds of standard size as described earlier 716. The study outline is presented in Table 1. The first set of experimental wounds was induced within 48 hours from the first sepsis-induced organ failure (early wound). A suction blister device (Mucel Ky, Nummela, Finland) with five 8-mm-diameter bores was applied to the intact abdominal skin and connected to the vacuum pump, which created a negative pressure on the area. First we used a higher vacuum (about 60 to 70 kPa) and after 20 to 30 minutes a lower one (40 to 50 kPa). During blister induction, the warming of the skin accelerates blister formation. The blister roofs were removed after induction. We used the same device for both patients and controls. A second set of wounds (late wound) was induced 4 days after the first set of wounds (Table 1). One set of suction blisters was induced in the controls.

The restoration of epidermal barrier function was followed up by measuring water loss from the blister wound. Since the epidermal barrier is a tightly regulated gateway to a percutaneous passage, TEWL decreases when the epidermal barrier is restored 17. After blister induction (when there is no epidermis), the water evaporation is 15- to 20-fold higher than in the intact skin. During the healing process, evaporation decreases, enabling the non-invasive follow-up of epidermal healing 781819. In this study, TEWL was measured using a VapoMeter (Delfin Technologies Ltd, Kuopio, Finland) 20, which measures the amount of water loss in grams per square metre. There is a cylindrical chamber in the head of the VapoMeter where the sensors for humidity and temperature are located. The VapoMeter forms a closed chamber on the skin in which the system automatically calculates the evaporation rate from the increase in relative humidity. The VapoMeter has been shown to be a reliable device to measure barrier function 2122. All five blister wounds were measured and the mean value was calculated and reported. We also measured TEWL from the intact abdominal skin while simultaneously measuring that of the blister wounds.

A laser Doppler flow meter (Periflux Pf1; Perimed KB, Stockholm, Sweden) was used to measure the blood flow in the blister wounds and also in the intact abdominal skin 2324. The laser beam penetrates about 1 mm into the skin. The vasculature of the skin contains two plexuses but the laser Doppler reaches only the superficial one, which lies just beneath the dermo-epidermal junction 1725. All five blister wounds were measured and the mean was calculated and reported. The measurements are expressed as perfusion units, which is arbitrary. Measurements of TEWL and skin blood flow in the blister wounds were taken twice on each set of wounds: the first measurements following the induction of the wound and the second measurements on the fourth day of healing (Table 1). All blister wounds were covered with an air- and water vapour-permeable, self-adhesive dressing between the study days (Mepore; Mölnlycke Health Care AB, Göteborg, Sweden). All blister inductions and measurements were performed by MK and FG under the same circumstances, such as the same air temperature.

The data were entered into an SPSS database (SPSS Data Entry, version 3.0; SPSS Inc., Chicago, IL, USA). Summary statistics are expressed as median with 25th and 75th percentiles or as mean with standard deviation (SD), and the analysis between the groups was done using the Kruskal-Wallis test. The Mann-Whitney U test was applied to analyse the differences between the two groups. The categorical variables were analysed by Fischer exact test. Spearman correlation was calculated. Two-tailed P values are reported and the analyses were performed by the SPSS software (version 15.0; SPSS Inc.). The differences were considered significant at P values of less than 0.05.

Two hundred sixty-three patients with sepsis were screened during the study period. Consent was obtained from 44 patients who fulfilled the inclusion criteria of severe sepsis. The first set of blisters could be induced in 35 patients within 48 hours of detection of the first organ failure and these 35 patients were included in the final analysis. The later blister wound was induced in the patients who were in hospital and alive on day four (27 patients, 77%). The last measurements were taken when the patients were still hospitalised, on the eighth day of the study. Table 2 summarises the clinical characteristics and the co-morbidities of the patients. There was no difference in average age between the septic patients and controls.

During epidermal healing, the TEWL from the blister wound decreases. The decrease reflects the restoration of the epidermal barrier function. The decrease of TEWL from day 0 to day 4 in the early wound was lower in the septic group than in the control one (Figure 1, Table 3). The mean decreases were 56 g/m² per hour (SD 91) in the septic group and 124 g/m² per hour (SD 31) in the control group. The same trend was seen in the late wound, for which the decreases were 77 g/m² per hour (SD 63) in the septic group and 124 g/m² per hour in the control group (P = 0.091) (Table 3). This suggests that the restoration of the epidermal barrier function is diminished in severe sepsis.

There were no differences in the TEWL of the intact skin at any point in time between the septic patients and the control subjects. On day 0, the mean TEWL values from the intact abdominal skin were 14 g/m² per hour (SD 17) in the septic group and 10 g/m² per hour (SD 7) in the control group. Neither the wound nor the intact skin showed any difference in TEWL between patients who received or did not receive steroid treatment due to sepsis.

In the early wound, there were no differences in blood flow after blister induction between groups (Table 4). This suggests that the initial inflammation does not show any alteration in the early phase of sepsis. On the contrary, the mean blood flow on the fourth day of healing in the early wound was significantly higher in the septic group, which suggests that sepsis aggravates the healing-related induction of inflammation (Figure 2 and Table 4). In the late wound, the mean blood flow after blister induction was significantly higher in septic patients than in the control subjects (Figure 3). It was also higher on the fourth day of healing (eighth day of the study) (Table 4). This suggests that both initial and induced wound inflammation are intensified in patients with established septic disease, which is possibly the result of systemic inflammation. The blood flow values did not differ between patients who received or did not receive steroid treatment. On the first day, there were no differences in mean blood flow from the intact abdominal skin in the septic group (15 units, SD 12) and in the control group (14 units, SD 9). However, on the fourth day, the mean blood flow from intact skin was higher in the septic group (24 units, SD 18) compared with the controls (6 units, P = 0.000) (Figure 4).