bs_bs_banner
Australasian Journal of Dermatology (2014) 55, 255–259
doi: 10.1111/ajd.12130
ORIGINAL RESEARCH
Which patients with dermatological conditions are admitted via the emergency department? Julia E Lai-Kwon,1 Tracey J Weiland,2,3 George A Jelinek2,3 and Alvin H Chong1 1
Department of Medicine (Dermatology), 2Emergency Practice Innovation Centre, St Vincent’s Hospital Melbourne, and 3Faculty of Medicine Dentistry and Health Sciences, University of Melbourne, Melbourne, Australia
ABSTRACT Background and Objectives: Most dermatology admissions to tertiary hospitals in Australia are initially assessed in the emergency department (ED). This 3-year retrospective study examined the types of dermatological conditions that necessitated admission, the factors that predicted admission and the implications for dermatological resource allocation. Methods: The ED database was searched using the International Statistical Classification of Diseases, 10th revision (ICD-10) diagnosis codes and keywords in the presenting complaint and triage notes fields. The two lists were then merged and duplicates removed. All admissions were analysed and the medical records of admissions to the dermatology unit were reviewed to determine their final diagnosis. Results: In total, 4817 patients with dermatological conditions presented over the 3-year period. Of these, 937 (20%) required admission, of whom 108 (12%) were admitted under the dermatology unit. The most common conditions requiring admission were cellulitis (n = 534, 56%), boils, furuncles and pilonidal sinuses (n = 183, 19%), and non-specific skin infections (n = 32, 3%). The most common conditions admitted under dermatology were psoriasis (n = 27, 25%), eczema (n = 25, 23%), and cellulitis (n = 16, 15%). Key predictors of admission were Australasian triage code, referral by a health-care professional or corrections officer and arrival by ambulance. Conclusion: Approximately one-fifth of dermatological presentations required admission, mostly for
Correspondence: Dr Alvin Chong, Department of Medicine (Dermatology), St Vincent’s Hospital Melbourne, 41 Victoria Parade, Fitzroy, Melbourne, VIC 3065, Australia. Email: alvin.chong@svhm .org.au Julia E Lai-Kwon, MBBS; Tracey J Weiland, PhD; George A Jelinek, FACEM; Alvin H Chong, FACD. Conflict of interest: none Submitted 31 July 2013; accepted 20 November 2013. © 2013 The Australasian College of Dermatologists
infective processes that did not require specific dermatological care. The predictors of admission reflect the severity of the condition and patients demonstrating these predictors should be referred to the Dermatology unit for admission. Key words: emergency service, epidemiology, hospital, retrospective, skin disease.
INTRODUCTION Dermatology is considered to be a primarily outpatientbased speciality in Australia. However, it has been reported that approximately 5–8% of all emergency department (ED) presentations are for skin complaints.1 Several studies have identified the types of dermatological conditions with which people present to the ED, as well as characterising their management pathway through the ED.2–5 It has been reported that the admission rate for dermatology patients via the ED is 4–18%.2,4,6,7 However, few studies have attempted to determine the types of condition that require admission or to quantify the burden of dermatological conditions on inpatient units, which would have important implications for dermatological resource allocation. We conducted a 3-year retrospective study on patients with dermatological conditions presenting to the St Vincent’s Hospital ED in Melbourne, Australia, in order to obtain demographic and clinical data on dermatological admissions through the ED, as well as to determine what factors predict admission.
Abbreviations: ED ICD-10 LOS
emergency department International Statistical Classification of Diseases, 10th revision length of stay
256
JE Lai-Kwon et al.
METHODS
Table 1
Significant predictors of admission under any unit
Data selection
Variable
P
OR
95% CI
St Vincent’s Hospital Melbourne is a tertiary hospital with an ED managing approximately 40 000 presentations per year. The hospital is situated on the fringe of the city’s central business district and serves a population that is culturally diverse and socially disadvantaged. All ED visits between 1 January 2009 and 31 December 2011 were extracted from the hospital database. Dermatological presentations to the ED were identified by two methods: (i) a keyword search in the presenting complaint and triage notes fields of the Victorian Emergency Minimum Dataset at St Vincent’s Hospital Melbourne, and (ii) an International Statistical Classification of Diseases, 10th revision (ICD-10) diagnosis code search in the primary, secondary and tertiary diagnosis fields. Keywords and relevant ED ICD-10 diagnoses were selected by consultation among the researchers. The ICD-10 codes were selected from a list of available ICD-10 diagnoses in the ED and therefore did not include all the possible dermatological diagnoses listed in the complete ICD-10. In cases where there was any doubt as to whether the presentation constituted a primary skin complaint, the cases were reviewed by the consultant dermatologist (AC). These lists were then merged, and traumatic skin injuries, burns (excluding sunburn), postoperative wound complications and repeat presentations were excluded. In order to handle the data more efficiently, the 91 ICD-10 codes were grouped into 55 diagnostically related categories. The grouping of diagnoses was determined by consensus among the researchers. The cases were then de-identified and demographic (patient age, sex) and clinical data (triage code, ED time to doctor, ED length of stay, initial ED ICD-10 diagnosis) were analysed using Microsoft Excel 2010. The data of all patients admitted as inpatients were then extracted. Patients admitted under the dermatology unit specifically were identified and their medical histories reviewed to determine their primary dermatological diagnosis and length of inpatient stay. This study received approval from the Hospital Research Ethics Committee at St Vincent’s Hospital Melbourne (LRR 146/12, 10/10/2012).
Australasian triage code 1 Australasian triage code 2 Australasian triage code 3 Australasian triage code 4 Referred by health professional or corrections Ambulance arrival Age ED management time Not married/de facto Accommodation unknown or undocumented
0.007 < 0.001 < 0.001 < 0.001 < 0.001
22.604 21.787 5.279 2.703 2.028
2.33–219.62 13.313–35.656 3.617–7.705 1.925–3.795 1.645–2.499
< 0.001 < 0.001 < 0.001 < 0.014 < 0.019
2.011 1.013 1.012 1.300 0.801
1.487–2.719 1.008–1.017 1.011–1.013 1.055–1.603 0.665–0.965
Data analysis Epidemiological data were summarised using frequencies and percentages for nominal data, as well as medians and interquartile ranges (IQR) for continuous data. A binary logistic regression was used to determine the odds ratios of the predictors of admission under any inpatient unit and the goodness of fit of the model. Preliminary tests of the assumptions of logistic regression were performed, including an examination of multicollinearity to ensure that continuous independent variables were not closely correlated (being a bivariate correlation > 0.70). Variables included in the model were: patients’ Australasian triage code (1–5, with 5 set as the reference category), age, sex, ambulance arrival (yes/no), marital status (married, defacto, unknown © 2013 The Australasian College of Dermatologists
ED, emergency department.
or not married, including single, widowed, never married, divorced, separated), interpreter required (yes/no), Aboriginal or Torres Strait Islander (yes/no), source of referral (health professional/corrections officer/other), usual accommodation: lives alone (private accommodation, hostel, homeless in public place) or lives with others (private or community residence, nursing home, prison, hospital or homeless shelter, unknown), and ED management time.
RESULTS Over the 3-year period there were 123 345 presentations to the St Vincent’s Hospital Melbourne ED. Of these, 4817 patients had a primarily dermatological complaint (4%), of whom 937 (20%) required admission. Patients were admitted under general medicine (n = 265, 28%), plastic surgery (n = 194, 21%), general surgery (n = 131, 14%) and dermatology (n = 108, 12%) followed by speciality surgical units (n = 107, 11%), speciality medical units (n = 68, 7%), ambulatory medical units (including hospital in the home) (n = 60, 6%) and psychiatry (n = 4, 0%). Patients with conditions that were most likely to require admission under any unit by their ICD-10 diagnosis code were those with cellulitis (n = 534, 56%), boils, furuncles or pilonidal sinuses (n = 183, 19%), non-specific skin infections (n = 32, 3%), psoriasis (n = 31, 3%), eczema/dermatitis (28, 3%), impetigo (n = 20, 2%), non-specific rashes (n = 15, 2%), varicella zoster (n = 14, 2%), allergy with skin involvement (n = 12, 1%) and unlisted diagnoses (n = 10, 1%). Logistic regression revealed a model that accounted for between 39 and 53% of the variance in admissions, with adequate goodness of fit (Hosmer–Lemeshow: γ2 = 10.13, P = 0.256). Several significant predictors of admission under any inpatient unit emerged, the strongest being Australasian triage code, which showed a strong linear pattern with increasing acuity, then referral by a health-care professional (including a dermatologist) or corrections officer and arriving by ambulance (Table 1). Of the patients admitted under the dermatology unit, 12 were admitted on more than one occasion. There was a female predominance, with women accounting for 64
Skin conditions admitted via the ED Table 2 ED International Statistical Classification of Diseases, 10th revision (ICD-10) diagnosis codes for admissions under the dermatology unit
Category
Relevant ICD-10 codes
n
% of total cases
Psoriasis Eczema
L409 L209, L 219, L22, L258, L309 L301, L302, L311, L039 L089 L989, R21 B000, B001 L519 L270, L 271 L029, L050, L059 L899 B019, B029 D485 L539 L272, L509, T780, T781, T782, T783, T784, T805, T886, T887
27 25
25 23
16 10 9 5 4 3 2 2 2 1 1 1
15 9 8 5 4 3 2 2 2 1 1 1
Cellulitis Non-specific skin infection Non-specific rash HSV infections Erythema multiforme Drug rash Boils/abscess Decubitus ulcers/bedsore Varicella zoster infection Neoplasm Erythema Allergy
admissions (59%) and men for 44 admissions (41%). The average age of the patients admitted under dermatology was 49.2 years (median: 50, range: 18–92). The patients were referred to the ED from a variety of sources including dermatology outpatients (n = 38, 35%), self-referrals (n = 31, 29%), private dermatologists (n = 18, 17%), local medical officers (n = 8, 7%) and other hospitals (n = 8, 7%). The modal Australasian triage category for patients requiring admission under dermatology was 4, which was the same as other dermatological presentations to the ED. Patients admitted under dermatology were generally seen more quickly by an ED doctor (median time to doctor: 27 min, IQR: 9–61.8 min) than those with other dermatological presentations (median time to doctor: 57 min, IQR: 19–119 min). However, their length of stay (LOS) in the ED was significantly longer (median emergency LOS: 319 min, IQR: 170.5–467.3 min) than for patients with other dermatological presentations (median emergency LOS: 186 min, IQR: 108–303 min). The most common ICD-10 diagnoses for admissions under dermatology are shown in Table 2. The most common diagnoses were psoriasis (n = 27, 25%), eczema (n = 25, 23%), cellulitis (n = 16, 15%), non-specific skin infections (n = 10, 9%) and non-specific rashes (n = 9, 8%). The primary diagnoses made on discharge from the dermatology unit are shown in Table 3. Eczema flare, rather than psoriasis, was the most common diagnosis (n = 29, 27%). Flares of atopic eczema accounted for 20 cases (69%), followed by eczema herpeticum (n = 5, 17%), erythrodermic eczema (n = 2, 7%), allergic contact dermatitis (n = 1, 4%) and sub-erythrodermic eczema (n = 1, 4%). Psoriasis flare was the second most common diagnosis on discharge (n = 28, 26%). Chronic plaque psoriasis was the most common subtype (n = 15, 54%), followed by generalised pustular psoriasis (n = 6, 21%), erythrodermic psoriasis
257
Table 3 Primary dermatological diagnosis on the discharge summary for admissions to dermatology
Primary dermatological diagnosis
Number of cases
% of total cases
Eczema Psoriasis Pyoderma gangrenosum Cellulitis Hidradenitis suppurativa Bullous pemphigoid DRESS Pemphigus vulgaris Pityriasis rubra pilaris Urticarial vasculitis Venous ulceration Baboon syndrome Cutaneous Crohn’s disease Cutaneous polyarteritis nodosa Darier’s disease Erythema multiforme Erythema nodosum Herpes zoster infection Infected sebaceous cyst Rash not otherwise specified Bacterial skin infection Sweet’s syndrome
29 28 15 6 5 4 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1
27 26 14 6 5 4 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1
DRESS, drug rash with eosinophilia and systemic symptoms.
(n = 3, 11%), palmoplantar pustular psoriasis (n = 2, 7%), palmoplantar psoriasis (n = 1, 4%) and sub-erythrodermic psoriasis (n = 1, 4%). Pyoderma gangrenosum flares were the third most common primary dermatological diagnosis (n = 15, 14%), followed by cellulitis (n = 6, 6%) and hidradenitis suppurativa (n = 5, 5%). On analysis of the 19 cases initially diagnosed as nonspecific skin infections or non- specific rash in the ED and admitted under dermatology, nine (47%) were later diagnosed as pyoderma gangrenosum, three (16%) as hidradenitis suppurativa, three (16%) as atopic dermatitis, and one (5%) as each of Sweet’s syndrome, drug rash with eosinophilia and systemic symptoms, pemphigus vulgaris and bullous pemphigoid. The median length of stay for admissions to dermatology via the ED was 8 days (IQR: 5–10 days).
DISCUSSION A substantial proportion (20%) of patients presenting to the ED with a dermatological complaint required admission. This is somewhat higher than previously reported.2–4,7 Cellulitis was the most common condition requiring admission, which is consistent with similar studies from South Australia4 and South Korea.3 The socially disadvantaged patient population at St Vincent’s Hospital Melbourne, including a high proportion of homeless patients and i.v. drug users, may also have contributed to the number of cellulitis cases. Boils, furuncles and pilonidal sinuses represented the second most common condition, followed by non-specific © 2013 The Australasian College of Dermatologists
258
JE Lai-Kwon et al.
skin infections. Most patients were admitted to general medicine and plastic surgery rather than dermatology, which is traditional for conditions such as cellulitis and boils or furuncles at this institution, highlighting the importance of a dermatology consulting service to provide management support to these units. For example, a consulting service may assist general medical units in managing recurrent episodes of cellulitis that may be caused by modifiable risk factors such as tinea pedis, which may not be considered by the home team. There were several predictors of admission identified in the study, including a more urgent Australasian triage code, referral by a health professional or corrections officer and arrival to hospital by ambulance. While these predictors are quite non-specific, referral from a health professional in particular remains an important predictor of admission. This is further supported by the high proportion of patients referred to the ED by a dermatologist (from private rooms or outpatients) who required admission to the dermatology unit. This, combined with the other predictors of admission, provides a useful method of screening for potential candidates for admission under any unit, expediting their progress through the ED. In all, 12% of patients required admission specifically under the dermatology unit. This is a significantly higher percentage than previously reported3,4 which may reflect the inclusion of cases referred directly from dermatology outpatients and private dermatologists, which were excluded in previous studies. Patients requiring admission to dermatology waited significantly longer in the ED than other dermatological presentations for two reasons. Firstly, under the prevailing model of care at the time of the study, these patients required a specialist registrar review before being transferred to the ward; a task that may have been delayed by other responsibilities. This model has since changed to meet current emergency access targets in Australia, allowing most ED to make these admission decisions without the need for a speciality review if one is delayed. Secondly, the lack of availability of dermatology inpatient beds may have resulted in patients waiting longer in the ED, a problem that is not unique to dermatology and will become increasingly significant as emergency presentations increase and inpatient bed numbers continue to decrease relative to population demands. The most common conditions necessitating admission under the dermatology unit were psoriasis, eczema, cellulitis, non-specific skin infections and non-specific rashes. This reflects the high prevalence of conditions such as psoriasis and eczema in the general community, coupled with the tendency of these conditions to flare and require admission for intensive topical therapy. Non-specific dermatological diagnoses were also frequently made in the ED, which in most cases were eventually diagnosed as rare and often serious dermatological conditions such as pyoderma gangrenosum. This reflects the insufficient time and perhaps expertise available to make definitive dermatological diagnoses in the ED; an issue raised in several previous studies.4,6,8 © 2013 The Australasian College of Dermatologists
A distinct difference between the ED diagnoses and the primary diagnoses for admissions to dermatology was also noted. Eczema, rather than psoriasis, was the most common primary diagnosis upon discharge. Pyoderma gangrenosum and hidradenitis suppurativa were also included in the top five primary diagnoses on discharge but did not feature in the initial ED diagnoses. However, on a retrospective analysis of the conditions initially admitted with non-specific diagnoses, pyoderma gangrenosum and hidradenitis suppurativa both featured prominently as the final dermatological diagnosis. This reflects the widely recognised gap between dermatologists and non-dermatologists in specific dermatological diagnoses due to differences in time available to diagnose skin complaints in the ED together with insufficient dermatological experience.9,10 Interestingly, classical dermatological emergencies such as Stevens–Johnson syndrome and toxic epidermal necrolysis did not feature as a major cause of admission to the dermatology unit. This is in contrast with other studies from Singapore2 (n = 26 over 1 year), the USA6 (n = 8 over 1 year) and India11 (n = 11 over 4 months). This may be due to hospital-specific factors, such as the lack of a burns unit at St Vincent’s Hospital Melbourne, which would be the most appropriate setting in which to manage such patients. Also, the high prevalence of admissions for less common conditions such as pyoderma gangrenosum and hidradenitis suppurativa may also reflect the hospital’s specialisation and may not be generalisable to other tertiary hospitals. This study was limited by incomplete demographic information on the ED database, which it was impractical to clarify due to the volume of dermatological presentations to the ED. This may have resulted in inaccuracies in the demographic data. The study was also limited by its reliance on the ICD-10 diagnosis codes entered by ED doctors at the time of the patient’s presentation. This involved searching for keywords pertaining to the patient’s problem, and the system suggesting the appropriate ICD-10 code. Coding inaccuracies may have resulted from inaccurate keyword searches. The lack of codes for specific dermatological conditions such as pyoderma gangrenosum may have also resulted in patients being inaccurately classified under less specific codes, concealing the true prevalence of these conditions.
CONCLUSION This is the first study to our knowledge that has examined the types of dermatological conditions requiring admission via the ED in a major Australian tertiary hospital and the implications for dermatological resource allocation arising from the data. We have identified factors that predict admission, including the Australasian triage code, referral by a health-care professional and arrival by ambulance. These factors correlate with the severity of the disease and patients in these categories should be recognised and referred to dermatology without delay. This may in turn improve the efficiency of the patients’ admission process and prevent patients with rarer skin diseases being coded as non-specific by ED staff.
Skin conditions admitted via the ED
REFERENCES 1.
2.
3.
4.
5.
Trott AT. Rash. In: Hamilton GC, Sanders AB, Strange G (eds). Emergency Medicine: An Approach to Clinical Problem Solving, 2nd edn. Philadelphia, PA: Saunders, 2003; 185–206. Wang E, Lim BL, Than KY. Dermatological conditions presenting at an emergency department in Singapore. Singapore Med. J. 2009; 50: 881–4. Kim JY, Cho HH, Hong JS et al. Skin conditions presenting in emergency room in Korea: an eight-year retrospective analysis. J. Eur. Acad. Dermatol. Venereol. 2013; 27: 479–85. Wallett A, Sidhu S. Management pathway of skin conditions presenting to an Australian tertiary hospital emergency department. Aust. J. Dermatol. 2012; 53: 307–10. Martinez-Martinez ML, Escario-Travesedo E, RodriguezVazquez M et al. Dermatology consultations in an emergency department prior to establishment of emergency dermatology cover. Actas Dermosifiliogr. 2011; 102: 39–47.
6.
7. 8.
9.
10.
11.
259
Jack AR, Spence AA, Nichols BJ et al. Cutaneous conditions leading to dermatology consultations in the emergency department. West. J. Emerg. Med. 2011; 12: 551–5. Baibergenova A, Shear NH. Skin conditions that bring patients to emergency departments. Arch. Dermatol. 2011; 147: 118–20. Dolan OM, Bingham EA, Glasgow JF et al. An audit of dermatology in a paediatric accident and emergency department. J. Accid. Emerg. Med. 1994; 11: 158–61. Wagner RF Jr, Wagner D, Tomich JM et al. Residents’ corner: diagnoses of skin disease: dermatologists vs. nondermatologists. J. Dermatol. Surg. Oncol. 1985; 11: 476–9. Feldman SR, Fleischer AB Jr, Young AC et al. Time-efficiency of nondermatologists compared with dermatologists in the care of skin disease. J. Am. Acad. Dermatol. 1999; 40 (2 Pt 1): 194–9. Gupta S, Sandhu K, Kumar B. Evaluation of emergency dermatological consultations in a tertiary care centre in North India. J. Eur. Acad. Dermatol. Venereol. 2003; 17: 303–5.
© 2013 The Australasian College of Dermatologists
Australasian Journal of Dermatology (2014) 55, 255–259
doi: 10.1111/ajd.12130
ORIGINAL RESEARCH
Which patients with dermatological conditions are admitted via the emergency department? Julia E Lai-Kwon,1 Tracey J Weiland,2,3 George A Jelinek2,3 and Alvin H Chong1 1
Department of Medicine (Dermatology), 2Emergency Practice Innovation Centre, St Vincent’s Hospital Melbourne, and 3Faculty of Medicine Dentistry and Health Sciences, University of Melbourne, Melbourne, Australia
ABSTRACT Background and Objectives: Most dermatology admissions to tertiary hospitals in Australia are initially assessed in the emergency department (ED). This 3-year retrospective study examined the types of dermatological conditions that necessitated admission, the factors that predicted admission and the implications for dermatological resource allocation. Methods: The ED database was searched using the International Statistical Classification of Diseases, 10th revision (ICD-10) diagnosis codes and keywords in the presenting complaint and triage notes fields. The two lists were then merged and duplicates removed. All admissions were analysed and the medical records of admissions to the dermatology unit were reviewed to determine their final diagnosis. Results: In total, 4817 patients with dermatological conditions presented over the 3-year period. Of these, 937 (20%) required admission, of whom 108 (12%) were admitted under the dermatology unit. The most common conditions requiring admission were cellulitis (n = 534, 56%), boils, furuncles and pilonidal sinuses (n = 183, 19%), and non-specific skin infections (n = 32, 3%). The most common conditions admitted under dermatology were psoriasis (n = 27, 25%), eczema (n = 25, 23%), and cellulitis (n = 16, 15%). Key predictors of admission were Australasian triage code, referral by a health-care professional or corrections officer and arrival by ambulance. Conclusion: Approximately one-fifth of dermatological presentations required admission, mostly for
Correspondence: Dr Alvin Chong, Department of Medicine (Dermatology), St Vincent’s Hospital Melbourne, 41 Victoria Parade, Fitzroy, Melbourne, VIC 3065, Australia. Email: alvin.chong@svhm .org.au Julia E Lai-Kwon, MBBS; Tracey J Weiland, PhD; George A Jelinek, FACEM; Alvin H Chong, FACD. Conflict of interest: none Submitted 31 July 2013; accepted 20 November 2013. © 2013 The Australasian College of Dermatologists
infective processes that did not require specific dermatological care. The predictors of admission reflect the severity of the condition and patients demonstrating these predictors should be referred to the Dermatology unit for admission. Key words: emergency service, epidemiology, hospital, retrospective, skin disease.
INTRODUCTION Dermatology is considered to be a primarily outpatientbased speciality in Australia. However, it has been reported that approximately 5–8% of all emergency department (ED) presentations are for skin complaints.1 Several studies have identified the types of dermatological conditions with which people present to the ED, as well as characterising their management pathway through the ED.2–5 It has been reported that the admission rate for dermatology patients via the ED is 4–18%.2,4,6,7 However, few studies have attempted to determine the types of condition that require admission or to quantify the burden of dermatological conditions on inpatient units, which would have important implications for dermatological resource allocation. We conducted a 3-year retrospective study on patients with dermatological conditions presenting to the St Vincent’s Hospital ED in Melbourne, Australia, in order to obtain demographic and clinical data on dermatological admissions through the ED, as well as to determine what factors predict admission.
Abbreviations: ED ICD-10 LOS
emergency department International Statistical Classification of Diseases, 10th revision length of stay
256
JE Lai-Kwon et al.
METHODS
Table 1
Significant predictors of admission under any unit
Data selection
Variable
P
OR
95% CI
St Vincent’s Hospital Melbourne is a tertiary hospital with an ED managing approximately 40 000 presentations per year. The hospital is situated on the fringe of the city’s central business district and serves a population that is culturally diverse and socially disadvantaged. All ED visits between 1 January 2009 and 31 December 2011 were extracted from the hospital database. Dermatological presentations to the ED were identified by two methods: (i) a keyword search in the presenting complaint and triage notes fields of the Victorian Emergency Minimum Dataset at St Vincent’s Hospital Melbourne, and (ii) an International Statistical Classification of Diseases, 10th revision (ICD-10) diagnosis code search in the primary, secondary and tertiary diagnosis fields. Keywords and relevant ED ICD-10 diagnoses were selected by consultation among the researchers. The ICD-10 codes were selected from a list of available ICD-10 diagnoses in the ED and therefore did not include all the possible dermatological diagnoses listed in the complete ICD-10. In cases where there was any doubt as to whether the presentation constituted a primary skin complaint, the cases were reviewed by the consultant dermatologist (AC). These lists were then merged, and traumatic skin injuries, burns (excluding sunburn), postoperative wound complications and repeat presentations were excluded. In order to handle the data more efficiently, the 91 ICD-10 codes were grouped into 55 diagnostically related categories. The grouping of diagnoses was determined by consensus among the researchers. The cases were then de-identified and demographic (patient age, sex) and clinical data (triage code, ED time to doctor, ED length of stay, initial ED ICD-10 diagnosis) were analysed using Microsoft Excel 2010. The data of all patients admitted as inpatients were then extracted. Patients admitted under the dermatology unit specifically were identified and their medical histories reviewed to determine their primary dermatological diagnosis and length of inpatient stay. This study received approval from the Hospital Research Ethics Committee at St Vincent’s Hospital Melbourne (LRR 146/12, 10/10/2012).
Australasian triage code 1 Australasian triage code 2 Australasian triage code 3 Australasian triage code 4 Referred by health professional or corrections Ambulance arrival Age ED management time Not married/de facto Accommodation unknown or undocumented
0.007 < 0.001 < 0.001 < 0.001 < 0.001
22.604 21.787 5.279 2.703 2.028
2.33–219.62 13.313–35.656 3.617–7.705 1.925–3.795 1.645–2.499
< 0.001 < 0.001 < 0.001 < 0.014 < 0.019
2.011 1.013 1.012 1.300 0.801
1.487–2.719 1.008–1.017 1.011–1.013 1.055–1.603 0.665–0.965
Data analysis Epidemiological data were summarised using frequencies and percentages for nominal data, as well as medians and interquartile ranges (IQR) for continuous data. A binary logistic regression was used to determine the odds ratios of the predictors of admission under any inpatient unit and the goodness of fit of the model. Preliminary tests of the assumptions of logistic regression were performed, including an examination of multicollinearity to ensure that continuous independent variables were not closely correlated (being a bivariate correlation > 0.70). Variables included in the model were: patients’ Australasian triage code (1–5, with 5 set as the reference category), age, sex, ambulance arrival (yes/no), marital status (married, defacto, unknown © 2013 The Australasian College of Dermatologists
ED, emergency department.
or not married, including single, widowed, never married, divorced, separated), interpreter required (yes/no), Aboriginal or Torres Strait Islander (yes/no), source of referral (health professional/corrections officer/other), usual accommodation: lives alone (private accommodation, hostel, homeless in public place) or lives with others (private or community residence, nursing home, prison, hospital or homeless shelter, unknown), and ED management time.
RESULTS Over the 3-year period there were 123 345 presentations to the St Vincent’s Hospital Melbourne ED. Of these, 4817 patients had a primarily dermatological complaint (4%), of whom 937 (20%) required admission. Patients were admitted under general medicine (n = 265, 28%), plastic surgery (n = 194, 21%), general surgery (n = 131, 14%) and dermatology (n = 108, 12%) followed by speciality surgical units (n = 107, 11%), speciality medical units (n = 68, 7%), ambulatory medical units (including hospital in the home) (n = 60, 6%) and psychiatry (n = 4, 0%). Patients with conditions that were most likely to require admission under any unit by their ICD-10 diagnosis code were those with cellulitis (n = 534, 56%), boils, furuncles or pilonidal sinuses (n = 183, 19%), non-specific skin infections (n = 32, 3%), psoriasis (n = 31, 3%), eczema/dermatitis (28, 3%), impetigo (n = 20, 2%), non-specific rashes (n = 15, 2%), varicella zoster (n = 14, 2%), allergy with skin involvement (n = 12, 1%) and unlisted diagnoses (n = 10, 1%). Logistic regression revealed a model that accounted for between 39 and 53% of the variance in admissions, with adequate goodness of fit (Hosmer–Lemeshow: γ2 = 10.13, P = 0.256). Several significant predictors of admission under any inpatient unit emerged, the strongest being Australasian triage code, which showed a strong linear pattern with increasing acuity, then referral by a health-care professional (including a dermatologist) or corrections officer and arriving by ambulance (Table 1). Of the patients admitted under the dermatology unit, 12 were admitted on more than one occasion. There was a female predominance, with women accounting for 64
Skin conditions admitted via the ED Table 2 ED International Statistical Classification of Diseases, 10th revision (ICD-10) diagnosis codes for admissions under the dermatology unit
Category
Relevant ICD-10 codes
n
% of total cases
Psoriasis Eczema
L409 L209, L 219, L22, L258, L309 L301, L302, L311, L039 L089 L989, R21 B000, B001 L519 L270, L 271 L029, L050, L059 L899 B019, B029 D485 L539 L272, L509, T780, T781, T782, T783, T784, T805, T886, T887
27 25
25 23
16 10 9 5 4 3 2 2 2 1 1 1
15 9 8 5 4 3 2 2 2 1 1 1
Cellulitis Non-specific skin infection Non-specific rash HSV infections Erythema multiforme Drug rash Boils/abscess Decubitus ulcers/bedsore Varicella zoster infection Neoplasm Erythema Allergy
admissions (59%) and men for 44 admissions (41%). The average age of the patients admitted under dermatology was 49.2 years (median: 50, range: 18–92). The patients were referred to the ED from a variety of sources including dermatology outpatients (n = 38, 35%), self-referrals (n = 31, 29%), private dermatologists (n = 18, 17%), local medical officers (n = 8, 7%) and other hospitals (n = 8, 7%). The modal Australasian triage category for patients requiring admission under dermatology was 4, which was the same as other dermatological presentations to the ED. Patients admitted under dermatology were generally seen more quickly by an ED doctor (median time to doctor: 27 min, IQR: 9–61.8 min) than those with other dermatological presentations (median time to doctor: 57 min, IQR: 19–119 min). However, their length of stay (LOS) in the ED was significantly longer (median emergency LOS: 319 min, IQR: 170.5–467.3 min) than for patients with other dermatological presentations (median emergency LOS: 186 min, IQR: 108–303 min). The most common ICD-10 diagnoses for admissions under dermatology are shown in Table 2. The most common diagnoses were psoriasis (n = 27, 25%), eczema (n = 25, 23%), cellulitis (n = 16, 15%), non-specific skin infections (n = 10, 9%) and non-specific rashes (n = 9, 8%). The primary diagnoses made on discharge from the dermatology unit are shown in Table 3. Eczema flare, rather than psoriasis, was the most common diagnosis (n = 29, 27%). Flares of atopic eczema accounted for 20 cases (69%), followed by eczema herpeticum (n = 5, 17%), erythrodermic eczema (n = 2, 7%), allergic contact dermatitis (n = 1, 4%) and sub-erythrodermic eczema (n = 1, 4%). Psoriasis flare was the second most common diagnosis on discharge (n = 28, 26%). Chronic plaque psoriasis was the most common subtype (n = 15, 54%), followed by generalised pustular psoriasis (n = 6, 21%), erythrodermic psoriasis
257
Table 3 Primary dermatological diagnosis on the discharge summary for admissions to dermatology
Primary dermatological diagnosis
Number of cases
% of total cases
Eczema Psoriasis Pyoderma gangrenosum Cellulitis Hidradenitis suppurativa Bullous pemphigoid DRESS Pemphigus vulgaris Pityriasis rubra pilaris Urticarial vasculitis Venous ulceration Baboon syndrome Cutaneous Crohn’s disease Cutaneous polyarteritis nodosa Darier’s disease Erythema multiforme Erythema nodosum Herpes zoster infection Infected sebaceous cyst Rash not otherwise specified Bacterial skin infection Sweet’s syndrome
29 28 15 6 5 4 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1
27 26 14 6 5 4 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1
DRESS, drug rash with eosinophilia and systemic symptoms.
(n = 3, 11%), palmoplantar pustular psoriasis (n = 2, 7%), palmoplantar psoriasis (n = 1, 4%) and sub-erythrodermic psoriasis (n = 1, 4%). Pyoderma gangrenosum flares were the third most common primary dermatological diagnosis (n = 15, 14%), followed by cellulitis (n = 6, 6%) and hidradenitis suppurativa (n = 5, 5%). On analysis of the 19 cases initially diagnosed as nonspecific skin infections or non- specific rash in the ED and admitted under dermatology, nine (47%) were later diagnosed as pyoderma gangrenosum, three (16%) as hidradenitis suppurativa, three (16%) as atopic dermatitis, and one (5%) as each of Sweet’s syndrome, drug rash with eosinophilia and systemic symptoms, pemphigus vulgaris and bullous pemphigoid. The median length of stay for admissions to dermatology via the ED was 8 days (IQR: 5–10 days).
DISCUSSION A substantial proportion (20%) of patients presenting to the ED with a dermatological complaint required admission. This is somewhat higher than previously reported.2–4,7 Cellulitis was the most common condition requiring admission, which is consistent with similar studies from South Australia4 and South Korea.3 The socially disadvantaged patient population at St Vincent’s Hospital Melbourne, including a high proportion of homeless patients and i.v. drug users, may also have contributed to the number of cellulitis cases. Boils, furuncles and pilonidal sinuses represented the second most common condition, followed by non-specific © 2013 The Australasian College of Dermatologists
258
JE Lai-Kwon et al.
skin infections. Most patients were admitted to general medicine and plastic surgery rather than dermatology, which is traditional for conditions such as cellulitis and boils or furuncles at this institution, highlighting the importance of a dermatology consulting service to provide management support to these units. For example, a consulting service may assist general medical units in managing recurrent episodes of cellulitis that may be caused by modifiable risk factors such as tinea pedis, which may not be considered by the home team. There were several predictors of admission identified in the study, including a more urgent Australasian triage code, referral by a health professional or corrections officer and arrival to hospital by ambulance. While these predictors are quite non-specific, referral from a health professional in particular remains an important predictor of admission. This is further supported by the high proportion of patients referred to the ED by a dermatologist (from private rooms or outpatients) who required admission to the dermatology unit. This, combined with the other predictors of admission, provides a useful method of screening for potential candidates for admission under any unit, expediting their progress through the ED. In all, 12% of patients required admission specifically under the dermatology unit. This is a significantly higher percentage than previously reported3,4 which may reflect the inclusion of cases referred directly from dermatology outpatients and private dermatologists, which were excluded in previous studies. Patients requiring admission to dermatology waited significantly longer in the ED than other dermatological presentations for two reasons. Firstly, under the prevailing model of care at the time of the study, these patients required a specialist registrar review before being transferred to the ward; a task that may have been delayed by other responsibilities. This model has since changed to meet current emergency access targets in Australia, allowing most ED to make these admission decisions without the need for a speciality review if one is delayed. Secondly, the lack of availability of dermatology inpatient beds may have resulted in patients waiting longer in the ED, a problem that is not unique to dermatology and will become increasingly significant as emergency presentations increase and inpatient bed numbers continue to decrease relative to population demands. The most common conditions necessitating admission under the dermatology unit were psoriasis, eczema, cellulitis, non-specific skin infections and non-specific rashes. This reflects the high prevalence of conditions such as psoriasis and eczema in the general community, coupled with the tendency of these conditions to flare and require admission for intensive topical therapy. Non-specific dermatological diagnoses were also frequently made in the ED, which in most cases were eventually diagnosed as rare and often serious dermatological conditions such as pyoderma gangrenosum. This reflects the insufficient time and perhaps expertise available to make definitive dermatological diagnoses in the ED; an issue raised in several previous studies.4,6,8 © 2013 The Australasian College of Dermatologists
A distinct difference between the ED diagnoses and the primary diagnoses for admissions to dermatology was also noted. Eczema, rather than psoriasis, was the most common primary diagnosis upon discharge. Pyoderma gangrenosum and hidradenitis suppurativa were also included in the top five primary diagnoses on discharge but did not feature in the initial ED diagnoses. However, on a retrospective analysis of the conditions initially admitted with non-specific diagnoses, pyoderma gangrenosum and hidradenitis suppurativa both featured prominently as the final dermatological diagnosis. This reflects the widely recognised gap between dermatologists and non-dermatologists in specific dermatological diagnoses due to differences in time available to diagnose skin complaints in the ED together with insufficient dermatological experience.9,10 Interestingly, classical dermatological emergencies such as Stevens–Johnson syndrome and toxic epidermal necrolysis did not feature as a major cause of admission to the dermatology unit. This is in contrast with other studies from Singapore2 (n = 26 over 1 year), the USA6 (n = 8 over 1 year) and India11 (n = 11 over 4 months). This may be due to hospital-specific factors, such as the lack of a burns unit at St Vincent’s Hospital Melbourne, which would be the most appropriate setting in which to manage such patients. Also, the high prevalence of admissions for less common conditions such as pyoderma gangrenosum and hidradenitis suppurativa may also reflect the hospital’s specialisation and may not be generalisable to other tertiary hospitals. This study was limited by incomplete demographic information on the ED database, which it was impractical to clarify due to the volume of dermatological presentations to the ED. This may have resulted in inaccuracies in the demographic data. The study was also limited by its reliance on the ICD-10 diagnosis codes entered by ED doctors at the time of the patient’s presentation. This involved searching for keywords pertaining to the patient’s problem, and the system suggesting the appropriate ICD-10 code. Coding inaccuracies may have resulted from inaccurate keyword searches. The lack of codes for specific dermatological conditions such as pyoderma gangrenosum may have also resulted in patients being inaccurately classified under less specific codes, concealing the true prevalence of these conditions.
CONCLUSION This is the first study to our knowledge that has examined the types of dermatological conditions requiring admission via the ED in a major Australian tertiary hospital and the implications for dermatological resource allocation arising from the data. We have identified factors that predict admission, including the Australasian triage code, referral by a health-care professional and arrival by ambulance. These factors correlate with the severity of the disease and patients in these categories should be recognised and referred to dermatology without delay. This may in turn improve the efficiency of the patients’ admission process and prevent patients with rarer skin diseases being coded as non-specific by ED staff.
Skin conditions admitted via the ED
REFERENCES 1.
2.
3.
4.
5.
Trott AT. Rash. In: Hamilton GC, Sanders AB, Strange G (eds). Emergency Medicine: An Approach to Clinical Problem Solving, 2nd edn. Philadelphia, PA: Saunders, 2003; 185–206. Wang E, Lim BL, Than KY. Dermatological conditions presenting at an emergency department in Singapore. Singapore Med. J. 2009; 50: 881–4. Kim JY, Cho HH, Hong JS et al. Skin conditions presenting in emergency room in Korea: an eight-year retrospective analysis. J. Eur. Acad. Dermatol. Venereol. 2013; 27: 479–85. Wallett A, Sidhu S. Management pathway of skin conditions presenting to an Australian tertiary hospital emergency department. Aust. J. Dermatol. 2012; 53: 307–10. Martinez-Martinez ML, Escario-Travesedo E, RodriguezVazquez M et al. Dermatology consultations in an emergency department prior to establishment of emergency dermatology cover. Actas Dermosifiliogr. 2011; 102: 39–47.
6.
7. 8.
9.
10.
11.
259
Jack AR, Spence AA, Nichols BJ et al. Cutaneous conditions leading to dermatology consultations in the emergency department. West. J. Emerg. Med. 2011; 12: 551–5. Baibergenova A, Shear NH. Skin conditions that bring patients to emergency departments. Arch. Dermatol. 2011; 147: 118–20. Dolan OM, Bingham EA, Glasgow JF et al. An audit of dermatology in a paediatric accident and emergency department. J. Accid. Emerg. Med. 1994; 11: 158–61. Wagner RF Jr, Wagner D, Tomich JM et al. Residents’ corner: diagnoses of skin disease: dermatologists vs. nondermatologists. J. Dermatol. Surg. Oncol. 1985; 11: 476–9. Feldman SR, Fleischer AB Jr, Young AC et al. Time-efficiency of nondermatologists compared with dermatologists in the care of skin disease. J. Am. Acad. Dermatol. 1999; 40 (2 Pt 1): 194–9. Gupta S, Sandhu K, Kumar B. Evaluation of emergency dermatological consultations in a tertiary care centre in North India. J. Eur. Acad. Dermatol. Venereol. 2003; 17: 303–5.
© 2013 The Australasian College of Dermatologists
