Introduction

In the first week post-burn, the wound is colonized by gram-positives' from the hair follicles, but later gramnegative micro-organisms are predominant. Gram-negatives come mainly from the gastrointestinal (GI) tract` and they are usually able to infect distal body areas by going through the intestinal barrier by way of the lymphatics and veins as far as the skin .
Staphylococcus aureus (SA) colonization comes from an exogenous source, such as other patients' or the nose of the burn patient.
Neely et al." showed that Pseudomonas aeruginosa (PA) and Candida albicans (CA) are present in the patient before colonization; in the temporal evolution of colonization, PA is first, SA second, Staphylococcus epidermidis (SE) third, and CA last.
Nonabsorbent intestinal antibiotics reduce intestinal micro-organisms and are useful in the prophylaxis of GI surgery. They can also be used to reduce infection in burn patients 'although there is some disagreement on this point.'
The first intestinal decolonization',1 included oral aminoglycosides plus erythromycin plus antifungal drugs; we prefer to replace erythromycin with polymyxin B.
Nowadays selective intestinal decolonization (SID) is used in the same way in neutropenic patients" (polymixin B 800 mg + amphotericin B 1000 mg + tobramycin 300 mg p.o. per day);" however Mason" used trimethroprim-sulphamethoxazole instead of tobramycin. SID is used in traumatic patients and other intensive care patients.
There are various creams for skin prophylaxis in burn patients. The most popular are silver sulphadiazine (1%), mafedine acetate (10%), and povidonc-iodine. In the light of our experience 16 we prefer clorhexidine 0.5%.
This paper describes the temporal evolution of microbial colonization in burn patients and considers whether or not the aetiology of the burn infection was endogenous.

Material and methods

For a period of three years we studied all patients admitted for two or more days to the burn intensive care unit (BICU) at La Paz medical centre. Colonization studies were developed twice per patient from different parts of the body (pharynx, rectum, skin, body burn surface, etc.).
Our BICU is multidisciplinary, with 12 rooms (one per patient). We have written protocols on manipulations that are published every two years.
Our main prophylactic techniques are wound debridement and surgical coverage, i.v. antibiotic prophylaxis, microbiological patient flora monitorization, and intensive burn decolonization (IBD).
IBD is divided into four parts: gastrointestinal decolonization: oral tobramycin 300 mg/day + nystatin 100,000-150,000 u/kg/day + polymyxin E 400 mg/day

• nasal decolonization: fusidic acid cream or chlorhexetidine 5%
• pharyngeal decolonization: hexetidine spray
• burn surface decolonization: chlorhexidine 0.5% cream (Rovi Lab.)

IBD should continue until total burn coverage. We do not use i.v. prophylactic antibiotics other than surgical prophylaxis, except in electrical burns, when we use penicillin 20 ml/day for two days.
Active epidemiological surveillance was practised with all patients.
We consider as skin infection cases in which the patient presents clinical features (fever without any other sign of infection and negative blood cultures, defective scarring, etc.) and microbiological features. We do not use burn surface biopsy as we prefer the serniquantitative burn surface study," which gives basically the same results but is a less aggressive procedure.
We represent the colonization-like prevalencel" in every week of the study as follows: number of colonized patients x 100/number of patients admitted that week.
We perform microbiological monitorization in every patient once or twice a week. In these procedures we look for possible pathogenous micro-organisms (PPM)" in the pharynx, rectum, nose, and skin; we use agar blood, agar chocolate, and McConey with antibiotic disk (cephalothin) and trimethoprime-sulphametoxazole);'I when these cultures do not present PPM we diagnose as normal flora (NF).
We use the Crickett-Graph programme to calculate the colonization tendency lines and the polynomial regression equation.
SA isolated from any patients and sanitary personnel were submitted to the National Center of Microbiology in order to establish whether or not there was cross-transmission in our unit.
We calculated positive and negative predictive values" (any infection vs burn surface infection) when any PPM from the patient were isolated; these values enable us to know the efficacy of the predictive values as risk factors for infection in our patients.

Results

Two hundred and eight patients were studied (mean age, 40.4 ± 1.3 yr; mean body burn surface burned, 24.8 ± 1.5%; 46% on respiratory support; 100% on antibiotics at some time).

Colonization

1. Pharynx (Fig. 1a). NF decreased by 50% in a week and a half and by 70% in the fourth week. This reduction was due to SA colonization and subsequently to PA. CA and Enterobacteriaceae have a very low colonization percentage.

2. Rectum (Fig. lb). NF decreased 25% in the first month and continued decreasing to 40% in the following months. PA is the most important colonizing micro-organism.

Fig. la - Pharyngeal colonization in ICU burn patients. Fig. lb - Rectum colonization in ICU burn patients

3. Nose (Fig. 2a). In the first week NF was reduced by 40% and remained at 50% until the fifth week. NF was ducto PA and SA colonization. CA colonization was about 10% after one month.

4. Healthy skin. Normal flora showed many changes in this area, but usually decreased to around 40%. About 30% of the patients were colonized by SA. Non-methicillin-resistant SA were the most important in the first week but subsequently methicillin-resistant SA appeared.

5. Body burn surface (Fig. 2b). NF decreased to 60% and remained around this level. SA and PA were the most important micro-organisms. The SA were usually methicillin-resistant.

Fig. 2a - Nose colonization in ICU burn patients Fig. 2b - Burn surface colonization in ICU burn patients

6. Table I shows colonization positive and negative predictive values in different areas of the body. It is important to note that positive predictive values were low and that the high levels occurred between the second and third weeks (>45 % in PA and 25 % in SA). Mean positive predictive values were 17% in PA and SA and 11% in Enterobacteriaceae. Negative predictive values were very high (100% in many weeks).

Colonization tendency
If we group all micro-organisms coming from any area of the body, we can see how these bacteria colonized NE Fig. 3 shows a continuous decrease in NF from the first week of admittance (50% in the first month and 40% in the second). PA increased every week from the beginning up to 30% in the second month.

Fig. 3 - Mean colonization by Saprophyte flora and P. aeruginosa.

Pharyngeal, nose, and healthy skin vs body burn surface colonization relationship
SA - Fig. 4a shows that SA colonization was similar in body burn surface, healthy skin, and the nose, but was totally different in the pharynx.
PA - Body burn surface and rectum colonization by PA were similar (Fig. 4b).

Fig. 4a - S. aureus colonization tendency in nose, burn, and safe skins. Fig. 4b - P. aeruginosa colonization tendency in rectum and burn skins.

Phagotyping
We compared SA isolated from different body areas in patients, nurses, and doctors. We were unable to prove cross-transmission (patient-patient; patient-staff) except in one case where we found the same SA in the nose of one nurse and the burn surface of one patient. The nurse's nasal colonization was treated with clortexidine 0.5%.

The temporal evolution of burn colonization and the aetiology of certain micro-organisms like SA have not been well studied.
Sawhney et al.,11 in a five-year study, examined 342 patients with more than 20% BSA burns. In the first two weeks SA colonization was the most important, while in the third week PA was the predominant micro-organism. The burn was colonized by PA in 38% of cases, Enterobacteriaceac in 14%, and SA in 37%.
We followed up our patients for eight weeks. In the first month we observed 15% colonization by PA and SA, while in the second month SA increased and PA decreased. Enterobacteriaceae and yeast were very unusual (5-8%) in the first month, after which they disappeared.
Colonization by PA increased in other body areas up to 30%. If we compare the studies of Deutsch' and Mason" with our own we can see that Deutsch found 93% burn colonization by micro-organisms with an enteric source (PA 87%, Proteus 27%) and 57% SA. In the Mason study 21% of burn colonization was by PA, 13% by Enterobacteriaceae (no Proteus), 9.5% by yeast, and 51% by SA. In our study, 18% of burn colonization was by PA, 12% by SA, 4.5% by Enterobacteriaceae, and 2% by yeast; this proves that SID, as used by Mason and ourselves, very effectively prevents colonization by gram-negative micro-organisms of enteric origin.
We also believe that our SID could be even more effective than that of Mason to prevent Enterobacteriaceae and yeast colonization. However, our results were better for SA because we also developed nasal, skin, pharyngeal, and especially nasal colonizatioff with chlorhexidine 0.5%, which is superior to silver sulphadiazine .15, 11 These differences were not related to the gravity of the patient because although the mean BSA was lower in our study, patient colonization with more than 30% BSA was similar.
In patients with 30% BSA, SID was more aggressive and as a result we found in these patients 9% colonization by PA, 3% by Enterobacteriaceae, and 16% by SA.
Patient colonization had a very low predictive value for burn infection. In the second and third week the relationship was 1: 2 for PA and less than I% for other microorganisms. The negative predictive value was very high, i.e. there was no infection without previous colonization.
Figs. 4a and 4b show that the micro-organisms usually came from the patient. In our study, environmental bacteria were always different from the micro-organisms in our unit.
SID also prevents respiratory and urinary infection, 15,24,25 and we therefore think this should be a routine procedure.

Fig. 5a - Pharyngeal colonization in polyLraumatic ICU patients. Fig. 5b - Rectum colonization in polytraumatic ICU patients Fig. 5c - Nose colonization in polytraumatic ICU patients. Fig. 6 - Mean colonization by saprophyte flora and P. aeruginosa

Figs. 5a and 5b show pharyngeal, nasal, and rectal colonization. Comparing Figs. I and 2 we can see that there was an important decrease in normal flora in the first two weeks and that usually PA was the main colonizing micro-organism.
If we consider mean colonization in any body area, it can be seen, for instance, that in polytraumatic patients (Fig. 6) there was an important decrease in normal flora in the second week (70% colonization) followed by a slow increase (80% colonization) in two months.

RESUME. La décolonisation intestinale sélective constitue une des techniques plus controverses pour la réduction de l'infection dans les unités des soins intensifs des brûlures. Dans les patients brûlés cette technique peut être utilisée avec la décolonisation nasale et la décolonisation intensive de la peau. Notre but était de contrôler l'évolution dans le temps de la flore de la peau chez les patients brûlés en condition critique et d'établir si l'infection était endogène. Deux cent huit patients brûlés provenant de notre unité de brûlures intensive ont été étudiés. La colonisation de zones différentes (pharynx, nez, gastrointestin, peau brûlée et saine) a été étudiée toutes les semaines. Nous avons obtenu la valeur colonisation contre valeur prédictive de l'infection. Nous avons contrôlé le Staphylococcus aureus (PA) chez les patients, les médecins et les infirmiers pour établir la route possible de la transmission de ce micro-organisme. Nous avons trouvé que chez les patients traités avec la décolonisation intensive de la peau la flore normale microbiologique a été mieux conservée. Quand la flore microbiologique normale disparaît la colonisation par Pseudomonas aeruginosa (PA) est très fréquente. La valeur de la prédiction de la colonisation contre l'infection par la brûlure etait basse mais pendant la troisième et la quatrième semaine de l'hospitalisation cette valeur a gagné les valeurs maximales (45% pour PA). La colonisation de la brûlure etait similaire dans les zones différentes du corps. La phagotypie de SA montrait que tous les patients étaient colonisés par un micro-organisme différent. Les auteurs concluent que la décolonisation intensive du patient brûlé est utile pour la préservation de la flore microbiologique et pour prévenir la colonisation et l'infection par d'autres micro-organismes, pour la plupart par voie endogène.