Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Non semper ea sunt quae videntur (Things are not always what they seem)
Rubor (redness), tumour (swelling), dolor (pain) and calor (heat) represent the four classical signs of inflammation, as described by the Roman physician Celsus in the first century AD.1 When observed in the skin in the acute setting, these signs are often considered synonymous with bacterial cellulitis, a reasonable assumption given the incidence and importance of early treatment of skin and soft tissue infections.2 Unfortunately, an incorrect initial diagnosis of infection can continue to follow the patient, resulting in unnecessary treatments, prolonged discomfort, and delayed diagnosis. In one study, some 13.6% of patients referred to a specialty centre for cellulitis were found to have alternative diagnoses, suggesting that this is not a rare occurrence.3
Many entities may mimic cellulitis and, compounding the confusion, many of these processes may be complicated by secondary or concomitant bacterial infection.3–5 For this review, however, I have selected three relatively common cellulitis mimics to discuss in detail: allergic contact dermatitis, erythema nodosum and insect bite-induced hypersensitivity (papular urticaria). By closely examining these cases, I hope to identify some of the pitfalls in the diagnosis of cellulitis and highlight some helpful pearls that may be applied more broadly.
Cellulitis is an acute bacterial infection of the skin and subcutis with secondary inflammation, usually characterised by warmth, tenderness, erythema and swelling.6 It is an extremely common clinical problem, representing 2.2% of all office visits for general practitioners in one study.7 Most cases are caused by Streptococci and Staphylococcus aureus, although many other pathogens may be implicated.6 The commonness of cellulitis appears to exploit the bias of the availability heuristic — the decision-making shortcut that places weight on things that are easily called to mind resulting in its mis- and overdiagnosis.8
Allergic contact dermatitis (ACD) generally presents as a pruritic eczematous rash, sometimes in linear or geometric configurations. Unlike type I hypersensitivity reactions that are IgE mediated and result in urticarial reactions, ACD is mediated by type IV hypersensitivity, in which T cells migrate to the skin in a more delayed and persistent fashion.9 Though there are literally thousands of well-known contact allergens, metals, topical antibiotics and urushiol from plants of the Toxicodendron genus (eg, poison ivy) remain very common culprits.9 10
Erythema nodosum (EN) is an inflammatory panniculitis that presents clinically with painful erythematous plaques and nodules frequently in the pretibial area. Systemic symptoms such as fever, malaise and arthralgias may also be present.11 EN is generally thought of as a non-specific reactive process with numerous disease associations including streptococcal pharyngitis, inflammatory bowel disease and sarcoidosis.12 13
Papular urticaria, sometimes known as insect bite-induced hypersensitivity reaction, is distinct from anaphylactic reactions which may occur in sensitive individuals from stings and much less commonly, insect bites.14 It is a local reaction characterised by pruritic erythematous papules sometimes with small crusted puncta where the actual bite (frequently mosquito or flea) occurred.15 There is a range of severity with the “normal” reaction being fairly mild and more severe reactions occurring with significant swelling, heat, erythema, pain and, very rarely, fever.16
Few situations in dermatology highlight the importance of gathering a careful history as when trying to differentiate between these entities. As all may share redness, swelling, pain and heat, the time course and the inciting factors can be tremendously helpful.
Knowing that a scratch on the leg slowly became increasingly red and tender over several days would point towards bacterial cellulitis with a portal of entry. A presentation mid-week of an itchy-then-sore rash on the legs after a weekend in the woods might point towards ACD from poison ivy or poison oak. Multiple tender nodules on the legs with fever and arthralgias would strongly suggest EN, while the recollection that a mosquito was swatted away from the arm just hours before the extremely itchy red blotch appeared would make papular urticaria guilty until proven innocent.16 In table 1, four useful history questions to help separate each disease are summarised.
While we are somewhat limited in terms of the basic morphological descriptors for these entities — all can present as erythematous patches or plaques — other features help to differentiate between them.
Perhaps the single most important physical finding is multifocality. This extremely uncommon and distinctive presentation of cellulitis is almost exclusively seen in immunocompromised individuals with fever, and in this setting is generally associated with Helicobacter cinaedi infection.17 Thus, for practical purposes, cellulitis should not be bilateral or multifocal.i ACD, on the other hand, is frequently multifocal, as is EN and, frequently, the papular urticaria of insect bites.
In fig 1A we see the left medial lower extremity of a teenage patient who had been in the woods several days before developing a widespread itchy rash. While it was initially pruritic, as the swelling and redness increased, it became somewhat sore but not exquisitely tender to palpation. She was admitted to the hospital for presumed bacterial cellulitis and received 3 days of intravenous antibiotic therapy without improvement before dermatology was consulted. Importantly, she was afebrile and had a normal white blood cell count throughout the duration and otherwise felt well.
Figure 1B reveals a finding that had not been noted initially: a similar lesion on the other leg. In this case, the bilaterality above all else suggested that this was not bacterial cellulitis. The history of recently being in the woods and walking through many plants while wearing shorts, the failure of response to antibiotics, the lack of fever or elevated white blood cell count all supported the diagnosis of ACD to poison ivy. Antibiotics were stopped and potent topical corticosteroids were initiated with improvement noted within 24 h and almost complete resolution within 7 days.
In fig 2A, a similar case is presented. An adolescent female with a history of multiple congenital anomalies and medical problems was noted to develop an erythematous eruption over the site of a prior hip replacement scar (fig 2A). The surgery had been performed 2 years prior and she had healed without complication. She was non-verbal but did not appear to be in pain. She had no fever and a normal white blood cell count. On examination, there was a large, fairly well-circumscribed erythematous patch with vesiculation in the centre. She was admitted with a presumed diagnosis of bacterial cellulitis, possibly related to the prior surgical site. She had radiographic imaging of the area which was unremarkable. She was started on broad-spectrum intravenous antibiotics and after 2 days, the rash continued to spread. Her antibiotics were tailored to cover methacillin-resistant S aureus and the spectrum further broadened. Multiple blood cultures were negative.
On day 5 of hospitalisation, dermatology was consulted. In reviewing the history, it was revealed that the patient’s mother had recently started to use a scar treatment cream that contained the topical antibiotic neomycin. Neomycin is a well-known contact allergen and her mother admitted to have continued to use the neomycin during the hospitalisation, which explained the continued worsening of the rash. A skin biopsy was performed and confirmed the diagnosis of ACD. Antibiotics were stopped and treatment with a potent topical corticosteroid was initiated. Figure 2B shows the resolution at the 1-week follow-up appointment.
In EN, the multifocal nature of the eruption is sometimes the only distinguishing feature from cellulitis. While the lesions tend to be deeper-seated nodules reflecting the fact that the primary process is in the fat, more plaque-like lesions are also common.18 In fig 3, multiple exquisitely tender erythematous plaques are shown on the bilateral lower extremities of a 9-year-old boy. He had low-grade fevers, malaise, arthralgias and a slightly elevated white blood cell count. EN was suspected but clinically it was difficult to exclude cellulitis; a skin biopsy revealed septal panniculitis which secured the diagnosis of EN. He made full recovery without specific treatment but several months later developed inflammatory bowel disease.
Finally, a 16-year-old girl had been on vacation in a warm climate. After an evening outside wearing only a bikini, she developed several itchy bites. One bite in particular in the right subscapular region swelled and became intensely itchy the next morning. She saw her primary care physician the next day seeking relief from the itch. Cellulitis was diagnosed and systemic antibiotics as well as topical bacitracin were prescribed. The patient dutifully applied the bacitracin but after 2 days developed a second, more widespread rash in the area she was applying it and was even more itchy (fig 4). She was afebrile and otherwise felt well. The diagnosis of papular urticaria was made, with a secondary diagnosis of ACD to bacitracin, another common contact allergen. The bacitracin was stopped and a potent topical steroid was started with excellent effect.
Dr Harley Haynes, a celebrated teacher and mentor with many years of experience, has distilled the essence of this dilemma into what I have dubbed “The Triad of Haynes” (H Haynes, personal communication, 2001). The triad consists of fever, elevated white blood cell count, and tenderness to palpation. It is important to place this triad in its proper context: for patients with widespread (meaning multifocal, bilateral or even simply involving large areas) “cellulitis”, particularly those that appear to be failing conventional antibiotic therapy, each characteristic of the triad that they lack should raise suspicion that they do not have cellulitis at all (fig 5). In other words, for the patients who have skin findings that would suggest significant bacterial infection, one would expect them to also be more likely to have fever, elevated white blood cell count and tenderness of the lesions; lacking these features should prompt a search for why. Because bacterial cellulitis, particularly at an early stage, can lack all of these features, the triad is by no means an exclusionary tool; rather it serves as a reminder to consider other diagnoses when things are not fitting together properly.
Table 2 summarises the key physical exam and laboratory findings to help distinguish between these causes of inflammation.
Once the diagnostic quagmire has been negotiated, appropriate therapy can be selected. For true bacterial cellulitis and for those cases where it cannot yet be excluded but must not be overlooked, empiric antibiotic therapy is typically chosen. Culture of the skin by various methods is unreliable and identifies a pathogen only in some 40% of cases.19 As the majority of cases are caused by Streptococci and S aureus, however, beta-lactam antibiotics with activity against S aureus are the drugs of choice and are generally highly effective.6 Elevation and immobilisation of the affected limb may also be helpful adjuncts and improvement can be expected in 1–3 days.7
ACD severe enough to mimic cellulitis tends to warrant more aggressive therapy. In such cases, systemic corticosteroids for 2–3 weeks with a taper and/or ultra-potent topical corticosteroids are very effective.9 Allergen avoidance is paramount and patch testing may need to be undertaken to identify the allergen in chronic or recurrent cases.10
The painful panniculitis of EN can be disabling for some patients, but will resolve in approximately 4 weeks even if left untreated, and it may be sooner than this in children.4 18 Treatment should be aimed at the underlying problem which can be found in some 77% of patients according to one study, but this leaves a significant proportion with idiopathic disease.18 For symptom control, aspirin and non-steroidal anti-inflammatory drugs may be used, though both intralesional corticosteroid injection and systemic corticosteroids are very effective if no co-existing infection is found.12
For acute papular urticaria, systemic antihistamines such as hydroxyzine and diphenhydramine are helpful for pruritus, and ultra-potent topical steroids as well as intralesional steroids often provide rapid relief.20 21 In more chronic or recurrent disease, eradication of the offending arthropod is critical (table 3).22
Cellulitis is so common a clinical entity and its mimics so relatively rare that we can be lulled into thinking anything inflammatory on the skin is bacterial infection. The vast majority of the time this assumption will be correct, or if not correct at least prudent. However, for the few times when it is not the case, to have considered other possibilities can mean the difference between calling out a mimic or missing it.
Competing interests: None.
↵i Of course, there are exceptions and by no means am I suggesting to exclude infection solely on the grounds that it is bilateral or multifocal; on the contrary, in such cases extra attention and suspicion are warranted for such an uncharacteristic presentation.