Rosacea is a common skin disease, characterised by facial flushing, telangiectasia, papules and pustules. It is generally regarded as inflammatory in nature. We believe that a contributory role of bacteria in this condition needs to be revisited.   


Rosacea is one of the most common office dermatology presentations. It is a chronic cutaneous disorder primarily affecting the central facial convexities (cheeks, chin, nose and central forehead). There is a clinical spectrum of primary and secondary manifestations of rosacea which includes flushing, persistent facial erythema, telangiectasia, papules, pustules, plaques, dryness, oedema, ocular lesions and phymatous change. The psychosocial consequences of rosacea can be devastating. The standard classification system of rosacea divides the disease into erythematotelangectatic, papulopustular, phymatous and ocular subtypes. Although separate, they are not necessarily progressive stages and are not mutually exclusive due to the varied presentations and the remitting and relapsing nature of the disorder.1,2,3,4 

The exact pathogenesis of rosacea remains elusive and is most probably multifactorial. Theories include genetic predisposition, environmental factors, inflammation and vascular dysfunction.  The role of micro-organisms and skin commensals have also been raised in the literature.4  The hair follicle mite Demodex folliculorum is often present in increased numbers within the sebaceous ducts and follicles in rosacea and its role is debated.5,6 Helicobacter pylori may contribute to rosacea by synthesising gastrins which stimulate flushing.7,8  More recently the role of the innate immunity involving cathelicidins and other antimicrobial peptides (AMPs) has been postulated.9   

Bacteria as a possible cause 

To date, a bacterial cause for the papules and pustules seen in rosacea has not been reported in the literature. The idea of an infectious cause for rosacea was first explored by Marks in 1968, whereby he examined bacterial cultures from swabs and biopsies of patients with rosacea and did not find any clear microbial cause in his mode of analysis.10

One of the main difficulties in isolating a causative organism is that many of the potential organisms considered are common members of commensal skin flora. We believe that the contributory role of bacteria needs to be revisited, particularly in light of the recognition of the role Propionibacterium acnes in acne vulgaris and Pityrosporum ovale in seborrhoeic dermatitis. It is interesting to note that proven rosacea therapy includes oral and topical antibiotics postulated to be due to their anti-inflammatory rather than antimicrobial properties.11 

The facial skin temperature is higher in patients with rosacea as a consequence of the persistent facial erythema and flushing.12  This increased blood flow leads to a chronic dilatation of blood vessels (and visible telangiectasia), which is thought to initiate an inflammatory cascade resulting in the formation of erythematous papules and pustules.13 This perhaps provides an altered microbial environment for the commensals of the skin and their protein products.14 In the literature, Dahl et al also raise the possibility of differences in the nature and behaviour of bacteria isolated from patients with rosacea compared with control subjects, with reported differences in the coagulase-negative staphylococci (CoNS) on the skin or follicles of patients with rosacea.14 

In order to more comprehensively review the role of bacteria in rosacea, we compare the bacterial isolates of rosacea subjects and controls.  We first revisit Marks’ experiment of comparing bacterial isolates from the facial skin of rosacea subjects and controls.  Next Dahl’s technique of incising a pustule of rosacea subjects to determine the presence or absence of bacterial isolate was replicated. This pustule swab was however then compared to the surrounding facial skin swab isolate of rosacea subjects, thus acting as an internal (self) control.  In addition a swab was taken from the eyelid margin of rosacea subjects and normal controls, which has previously not been considered.  If there is a bacterial presence in rosacea, the swab culture isolates in rosacea subjects may be different.  


Ethics approval 

This trial was approved by the St Vincent’s Human Research Ethics Committee (ref no: H04/017) on 22 August 2004, revised in 2006. Patients participating in the trial, with rosacea or as controls, were seen through the Skin and Cancer Foundation and St Vincent’s Hospital Outpatients Clinic in Darlinghurst.   

Subjects, Inclusion and Exclusion Criteria

Healthy men and women between the ages of 18 and 70 with and without rosacea were eligible to participate in the trial.  Rosacea subjects were identified in the clinic by an experienced dermatologist.  Rosacea subjects had to demonstrate persistent erythema, and telangiectasia as well as at least one pustule located on the face (nose, chin, forehead or nose). Control subjects were age (+/- 5 years) and sex matched. Study subjects were excluded if they had any other presenting serious facial dermatoses.  Participants were excluded if they had received treatment with any form of systemic or topical antibiotics within 14 days prior to the date of consultation. 

The rosacea subjects had their skin examined and assessed using the Clinician’s Global Severity Score, an accepted clinical standardized guideline created to determine the severity of a patient’s rosacea.2  In addition, a questionnaire was administered to ascertain the history of flushing, its triggers, the duration, and the patient’s opinion as to whether they had persistent facial erythema, telangiectasia, papules, pustules and ocular symptoms (Table I).

Study design

This study is composed of 2 parts. 

a) The first part compares the bacteria isolated from the face and eyelid margin of 15 subjects with papulopustular rosacea with the bacteria isolated from the face and eyelid margin in 15 age and sex-matched control subjects without rosacea.  Rosacea subjects had a swab taken from their cheek at least 2 cm away from any pustules and prior to cleansing of the skin. The control subjects had a swab taken from the skin of their cheeks. Another swab was then gently passed along the inferior eyelid margin of the eye on the same side of the face from which the pustule swab and control swab were taken.   

b) The second part of this pilot study compares the bacteria isolated from the pustules of patients with rosacea with the bacteria isolated from the surrounding facial skin in (a)  The surrounding skin was chosen to be at least 2 cm away from the pustule chosen for incision.

Based on clinical examination, an appropriate pustule for incision was identified and then cleansed with a 70% alcohol swab.  The pustule was incised with an 11- or 15-gauge scalpel blade as appropriate and the purulent content was extruded gently on to a sterile transport swab containing Copan® Amies Agar gel without charcoal.  

Aseptic technique was used for all swabs taken.

To isolate bacteria, swabs were then taken to the laboratory and directly streaked on to culture medium and incubated.   


For each swab taken, two plates were cultured aerobically on Chocolate agar at 37OC and 42OC and one Brain Heart Infusion agar (BHI) plate was cultured anaerobically. The aerobic plates were then incubated at 37oC and 42OC for at least 48 hours and no more than 72 hours. The anaerobic plates were incubated in a GasPak jar for the same period of time.   

After the incubation classical morphological and biochemical tests were used for isolate identification. Colonial appearance, catalase, coagulase production and then speciation were determined using the API typing kit. Once a CoNS result was confirmed, speciation was performed using the API STAPH identification kit (BioMerieux, Inc, Hazelwood Mo).15  API STAPH is an identification system that is capable of differentiating different species of staphylococci, micrococci and related genera.  Testing for beta-haemolysis was performed on Columbia Blood Agar Base. 

Once speciation was completed, antibiotic susceptibility tests were performed using the Clinical Laboratory Standards.16  Bacterial isolates were tested against the following antibiotics; penicillin, cefoxitin, cephalexin, erythromycin, tetracycline, gentamicin and vancomycin.   

Statistical Analysis 

Descriptive statistics were completed for both cases and controls in the following areas: patient demographics (age and gender), microbiology cultures, antibiotic sensitivity results, clinical examination results and questionnaire answers.   

Statistical testing was performed to answer questions posed in this paper:

1. a) Is there any difference in the bacteria isolated from the facial skin of subjects with rosacea compared to those without?

b) Is there any difference in the bacteria isolated from the eyelid skin of subjects with rosacea compared to those without?

2.  Is there any difference in the bacteria isolated from the pustules and the facial skin of subjects with rosacea?  

This was determined by using a two-tailed Fisher’s Exact Test, in which p values less than 0.05 were considered significant (p<0.05).   



This pilot study involved 15 patients who were identified in clinic with papulopustular rosacea (n = 15). Age- and sex-matched control subjects were identified for each case. 

Demographic data was very similar for both groups. In both the rosacea and control groups (n = 15), 9 subjects (60%) were female and 6 subjects (40%) were male. Of the rosacea cases, the age of patients ranged from 24 to 81 years, with a mean age of 45.6 years (+/-STD 14.76). Of the control group, ages ranged from 21 to 78 years, with a mean age of 45.13 years (+/- STD 15.45).

Microbiology (Table 2a&2b)

a)We compared the skin swabs of the rosacea subjects to the swabs taken from the control subjects (Table IIa &IIb).  In the rosacea group 11 cases (73%) cultured mixed growths, and 4 (27%) had no growth. For the controls, 13 (87%) had mixed growths and 2 (13%) no growth (Table III). 

For each skin swab taken, organisms were identified by visual speciation to genera. The mixed growth demonstrated the presence of typical commensal flora, which included Micrococci, Diphtheroids, Staphylococcus and Proprionibacterium. For these rosacea group of subjects the swabs taken from the skin of the face, 6 (40%) had Micrococci present, 4 (27%) cases had Diphtheroids, 9 (60%) cases had Staphylococcus, and 6 (40%) Proprionibacterium present. For the control group of subjects the swabs taken from the skin of the face , 13 demonstrated a mixed growth of which 10 (67%) had Micrococci present, 8 (53%) had Diphtheroids, 11 (73%) had Staphylococcus and 9 (60%) had Propionibacterium.  This represents typical commensal flora of the skin.  

In case of the eyelid swabs, in the rosacea subject group 4 patients (27%) grew pure growths of Staphyloccoccus epidermidis, 8 (53%) demonstrated a mixed growth and 3 (20%) had no growth. For the control subjects, 11 (73%) grew mixed growths and the remaining 4 (27%) had no growth (Table IV).   

Pure growth of Staphylococcus epidermidis was only isolated from the rosacea subjects.  Using a 2 tailed analysis there was no statistically significant difference regarding the presence or absence of Staphylococcus epidermidis as a pure growth between the rosacea subjects and normal subjects p= 0.100.  However, using a 1 tailed hypothesis to see if Staphylococcus epidermidis is isolated more commonly in people with rosacea, this becomes statistically significant p=0.049. 

Of the 4 patients who grew pure growths of Staphylococcus epidermidis from their eyelid swabs 3 reported ocular symptoms of grittiness, itchiness, soreness or burning.  A total of 8 of the rosacea subjects reported ocular symptoms.   

b) We then compared the growth of bacteria isolated from the pustules of the rosacea subject group, with skin swabs of their face (cheek) located in the same area as the pustules.

Of the cases in which growths of bacteria isolated from the swabs taken of pustules were considered, 9 (60%) swabs cultured pure growths of Staphylococcus epidermidis (Fig. 1),  2 (13%) were mixed growths and 4 (27%) no growth (Table V). Of the swabs taken from the surrounding non-pustular skin of rosacea subjects (see a), all 15 cases cultured mixed or no growth p = 0.0007.  Thus the finding of a pure growth of Staphylococcus epidermidis in the lesional pustules of rosacea patients is found to be highly statistically significant. 

Antibiotic sensitivities (Table VI)

Antibiotic sensitivities were performed on the 9 cases that produced pure growths of Staphylococcus epidermidis. For the 9 cases, 1 (11%) was sensitive to penicillin, 9 (100%) were sensitive to cefoxitin, 9 (100%) sensitive to cephalexin, 8 (89%) were sensitive to erythromycin, 8 (89%) were sensitive to tetracycline and 9 (100%) were sensitive to gentamycin and vancomycin.   

Discussion and Implications of Findings

This study revisits the subject of bacteria as a contributory factor to the formation of pustules seen within papulopustular rosacea, which has not been explored in the literature since Marks published on it in 1968.10  Marks’ methodology was to compare the bacteria isolated from the skin of persons with rosacea and those without. Marks found no change in the range of bacteria found on the skin when persons with rosacea were compared with persons without rosacea, and therefore concluded that there was no evidence to support the theory that bacteria were implicated as a factor in pustular rosacea. The first part of our study concurs with what Marks observed, with no statistically significant difference between the skin swabs of rosacea subjects and normal controls. Our findings also did not demonstrate a difference between subjects with pustular rosacea compared to controls based on the methodology that we employed.   

Marks did not perform swabs of the eyelid conjunctiva.  Our preliminary data suggests that Staphylococcus epidermidis may be involved in the eye symptoms of patients with pustular rosacea.  In particular it would be important to look at those patients with ocular rosacea and especially those with pustules along the conjunctival margin to see if the results that were shown with the cutaneous pustules could be replicated.   

The remarkable finding of this study, comes from its second part, which investigates the presence or absence of bacteria isolated after aseptic incision of the pustules and found the highly significant pure growth of Staphylococcus epidermidis.  Our study compares the bacteria present or absent in a rosacea pustules with the bacteria isolated from the skin of the same rosacea subject, with each subject acting as their own control. This subtle difference in methodology has shown that of the 15 rosacea subjects who had pustules swabbed, 9 (60%) cultured pure growths of Staphylococcus epidermidis (and 2 mixed, 4 no growth), p = 0.0007 with no pure growths of Staphylococcus epidermidis isolated from the surrounding skin. We believe that this indicates that Staphylococcus epidermidis is significantly more likely to be an integral part of the disease process, rather than an incidental contaminant.  

Antibiotics are a mainstay in the treatment of papulopustular rosacea, but are considered to work by their anti-inflammatory effects, rather than anti-microbial activity. However, topical steroids, which are also anti-inflammatory, are not as consistently effective compared to the antibiotics routinely used. Dahl et al also detail other reasons why this belief maybe doubtful, and in particular notes that antibiotics are highly successful in abolishing lesions completely rather than blunting their inflammatory effects .14  

The antibiotic sensitivities of the Staphylococcus epidermidis shown in our study reflect the appropriateness of the antibiotics used in the treatment for papulopustular rosacea. 10 of the 11 (91%) samples were sensitive to tetracycline, and only one  was sensitive to penicillin.  It is interesting to note that in the 1970s penicillin and tetracycline were both used as effective for the treatment of rosacea. Penicillin is no longer   recommended as a treatment for rosacea in the overview of therapy published by Gupta in 2005.11  

Antibiotic therapy responsiveness is consistent with the hypothesis that Staphylococcus epidermidis is an integral part of pustule development. The pustule development appears to be a complex process mediated by a number of factors that are ongoing areas of research interest.  The symptoms of rosacea are exacerbated by factors that trigger the innate immune response, which include the release of cathelicidin antimicrobial peptides (AMP).9  Yamasaki et al have shown that expression of cathelicidin, an AMP and its proteolytic processing enzyme stratum corneum tryptic enzyme (SCTE) is different to in rosacea subjects compared to normal subjects.  These substances can cause inflammation and increased blood vessel growth with abnormal processing.9,17  Staphylococcus aureus has been shown to be a potent trigger for cathelicidin release.18  It is possible that Staphylococcus epidermidis is also a potent trigger for the induction of cathelicidin release.19  

The temperature of the skin is known to be increased in persons with rosacea.14  It is likely that this is due to the long history of flushing and the development of altered dermal vasculature, which increases the blood flow to the areas affected by rosacea.20,21   Dahl et al investigated the possible role of elevated facial skin temperature causing a change in bacterial protein production in rosacea patients compared to controls. In this experiment Dahl et al reported isolating coagulase-negative Staphylococcus epidermidis from all 4 patients with rosacea pustules.14  The Staphylococcus epidermidis isolated were consistently beta-haemolytic.  From 4 of our 15 rosacea subjects we were able to test their Staphylococcus epidermidis cultures for beta-haemolysis.  All staphylococci from patients with rosacea were beta-haemolytic when cultured on Columbia Blood agar base. Although our entire case series was not tested, this supports Dahl’s findings.   

Differences in the Staphylococcal epidermidis beta-haemolysis profiles, raises the possibility that the strains of Staphylococcus epidermidis isolated from patients with rosacea may be different from those of normal controls and this may have both treatment implications as well as pre-operative considerations for those rosacea patients having elective surgery.  In the broader medical context, our knowledge of CoNS, such as Staphylococcal epidermidis, being implicated in systemic infections is important. The role of staphylococcal strains isolated from the skin may have particular relevance in nosocomial infections such as on indwelling central venous lines, in which over two thirds of isolates are CoNS.22  The period of flushing (and/or the development of telangiectasia) preceding the development of the pustules raises the possibility that it is the elevated temperature in rosacea patients which may create an environment for the normally ubiquitous Staphylococcus epidermidis to behave in a pathogenic manner, or for certain strains of Staphylococcus  epidermidis  to become more dominant.  The possible role of Staphylococcus epidermidis and its variable strains in the pustules of rosacea warrants further examination. 


In summary, rosacea is a complex common dermatological condition, whose exact pathogenesis has long baffled the medical condition. In light of our results, we believe that Staphylococcus epidermidis may play a significant role in the formation of pustules in papulopustular rosacea and hence an integral part of the inflammatory cascade which is involved in this disease. Understanding cutaneous rosacea more comprehensively requires an understanding of ocular rosacea.  This study sheds more light onto one of the first theories on the pathogenesis looking to explain both conditions. Thus, we suggest that Staphylococcus epidermidis does play an role in the formation of the pustules of rosacea, and may also explain some of the features of ocular rosacea.   


We would like to thank Associate Professor Jock Harkness and St Vincent’s Hospital microbiology laboratory staff, in particular, Ms Sanna Hawkins, for their help and support.




































Abbreviations and acronyms 

Coagulase-negative staphylococcus (CoNS)

Brain Heart Infusion agar (BHI)

antimicrobial peptide (AMP)

stratum corneum tryptic enzyme (SCTE)







The contents of this manuscript have not been previously published and are not currently submitted elsewhere.





260003316 Rev A BA Photos BBL binder.pdf© 2011 St Leonards Dermatology & Laser            
Website by Magicdust