The Red Leg Dilemma: A Scoping Review of the Challenges of Diagnosing Lower-limb Cellulitis

M. Patel; S.I. Lee; K.S. Thomas; J. Kai


The British Journal of Dermatology. 2019;180(5):993-1000. 

In This Article


From the 3926 initial search results, 2779 records were screened at the title and abstract stage after duplicates were removed. Next, 533 full-text articles were assessed for eligibility and 71 were included for data extraction (Figure 1).[9–79] Nine papers were foreign-language texts: six French, two Spanish and one Turkish.

Figure 1.

PRISMA flow diagram of the entire search.

The articles were first grouped into four themes: clinical cases of misdiagnosis, diagnostic aid, service development and aetiology. Clinical cases of misdiagnosis were studies where lower-limb cellulitis was the incorrect initial diagnosis or was initially misdiagnosed as another pathology. Service development reports were studies looking at how service set-up may reduce misdiagnosis. Diagnostic aids included studies that developed or tested tools to help diagnosis. Aetiology included studies that discussed microbiological causes of cellulitis.

Three themes were deemed to be of particular relevance and were explored further: clinical cases of misdiagnosis, service development and diagnostic aids. The aetiology theme, identifying the microbiological cause of cellulitis, is also an important research topic from the cellulitis Priority Setting Partnership.[6] We did not include this theme in this review as the papers identified highlighted treatment failure due to targeting the wrong organism, rather than a wrong diagnosis of cellulitis. For the themes service development and diagnostic aids, 11 papers were excluded as the site of cellulitis was not specified or the results of lower-limb cellulitis were not separated.[80–90]

Clinical Cases of Misdiagnosis

For the misdiagnosis theme, 66 papers were included, with three observational studies[9–11] and 63 case reports or series.[12–74]

Observational studies. One prospective study found that of the 635 patients referred with lower-limb cellulitis to a cellulitis clinic, 210 patients had 44 other diagnoses. Of these other diagnoses, the most common were eczema (118 patients), lymphoedema (14 patients) and lipodermatosclerosis (nine patients).[9] Another prospective study of children aged under 15 years found that 19 of 50 patients with osteomyelitis were initially misdiagnosed as having cellulitis.[10] One retrospective observational study showed that in 43 patients with an initial clinical suspicion of deep vein thrombosis, nine patients were diagnosed with cellulitis.[11]

Case report and case series. Overall 94 patients were included (43 male, mean age 41 years) (Table S2; see Supporting Information). In total, 47 different pathologies were misdiagnosed, with two initially diagnosed as another pathology before being correctly diagnosed as cellulitis.[6,39,64] The pathologies were grouped by specialty: vascular (nine pathologies) was the most common group.[13,21,22,24,39,45,52,54,55] Necrotizing fasciitis,[40,51,68,71] sarcoidosis,[19,32,42,72] lymphoma[33,53,56,59] and chemotherapy-related pathology[20,47,67,69] had the most case reports or series as a misdiagnosis.

Typical symptoms and signs of inflammation seen in cellulitis are erythema, pain, swelling, fever and warmth. Of the patients subsequently found to have been misdiagnosed, 74 (79%) had erythema of the skin, 73 (78%) experienced pain, 52 (55%) had swelling, 23 (24%) had fever and 19 (20%) had increased warmth of the skin. Unilateral features were present in 73 patients (78%) and bilateral features in 15 (16%). Prior antibiotics were given to 26 patients (28%).

Ten patients (11%) were later diagnosed with a malignancy,[17,18,23,29,33,35,53,56,59,60] including one case of metastatic malignant melanoma[35] and a neonatal case with kaposiform haemangioendothelioma.[23]

Key learning points suggested by the authors of the included case reports are shown in Table 1.

Service Development

Two studies had developed services to help reduce the rates of cellulitis misdiagnosis within both primary and secondary care.

Cellulitis clinic. One study initiated a new care model with a 'cellulitis clinic' in a single hospital in the U.K., operated by nurses and junior doctors from 09·00 to 17·00 h on weekdays, with faxed or telephone referrals from clinicians for patients diagnosed with suspected cellulitis.[9] In total 635 patients were treated through the specialist service, of whom 425 (67%) had cellulitis. Overall 41% were given intravenous antibiotics in the community, with 512 patients avoiding admission for intravenous treatment in the hospital, with a bed day saving of £818 000 over 40 months. In total, 1470 days of antibiotic use were avoided in the patients without cellulitis.

Red legs service. In one hospital in the U.K., a retrospective audit of patients who were admitted with bilateral red legs found that 15 of 50 were misdiagnosed as having cellulitis.[75] This hospital subsequently commissioned a nurse-led 'red legs' service to manage patients with bilateral red legs. Diagnostic algorithms were developed with relevant clinicians. Clinical photographs were shared with the lead clinicians via the hospital computer system. Seventy-seven patients were seen by the service, of whom 58 (75%) were discharged and 19 (25%) required a follow-up appointment. The cost saving was estimated to be £100 000. From the feedback available, 23 patients (82%) were extremely satisfied with their level of care.

Diagnostic Aids to Help Diagnosis

Four papers looked at developing or using an existing tool to help differentiate lower-limb cellulitis from alternative pathologies (Table 2).[76–79] Raff et al. explored cellulitis as the main pathology.[79] Three studies included patients with lower-limb cellulitis as a comparison group, where cellulitis and other diagnoses were compared. All four studies were observational studies conducted in different healthcare specialties.

Predictive test. An ALT-70 model was designed that involved assessment of asymmetry (unilateral involvement), leucocytosis (white blood cell count ≥ 10 × 109 cells L−1), tachycardia (heart rate ≥ 90 b.p.m.) and age ≥ 70 years. A score below 3 had > 83·3% likelihood of indicating pseudocellulitis (an alternative diagnosis to cellulitis) and a score above 4 had > 82·2% likelihood of indicating cellulitis.[79]

Biochemical test. When cellulitis was compared with acute gout, delta neutrophil index > 1·7% was the only independent factor for predicting cellulitis (P = 0·002), compared with white blood cell count (P = 0·41), C-reactive protein (P = 0·28) and procalcitonin (P = 0·12).[78]

Imaging. In comparison with patients with Dercum disease, in cellulitis, attenuation was more linear, diffuse and nonmass-like on computed tomography and magnetic resonance imaging (MRI). In addition, there was postcontrast enhancement in all three cases of contrast provided to patients with cellulitis.[77]

Three-phase immunoscintigraphy using 99Tcm-labelled antigranulocyte monoclonal antibodies was used in patients with infectious diabetic foot, with six of nine cellulitis lesions showing significantly increased uptake.[76]

Excluded Studies

Service development. Looking at service development, four papers were excluded because the site of cellulitis was not specified. Three studies in the U.S.A. showed that dermatology consultation improves the accuracy of cellulitis diagnosis,[80–82] often done in a single consultation.[81] Jain et al. showed that input from an infectious disease specialist cellulitis clinic improved differentiation from pseudocellulitis and reduced rates of hospitalization and cellulitis recurrence.[83]

Diagnostic aids. Four studies did not state the site of cellulitis. Of these, David et al. used a visually based computerized diagnostic decision support system for patients admitted with cellulitis from the emergency department.[84] Pallin et al. looked at procalcitonin and HLA-DQA1 gene expression among cellulitis cases and mimickers.[85] Schmid et al. and Rosenthall et al. used MRI[86] and radiophosphate imaging,[87] respectively.

Three studies did not separate the results for lower-limb cellulitis: Borschitz et al. utilized a modified Laboratory Risk Indicator for Necrotizing Fasciitis score to differentiate cellulitis from necrotizing fasciitis,[88] Rahmouni et al. used MRI[89] and Sullivan et al. looked at nuclear scintigraphy.[90]