Abstract
Background Boils and abscesses are common in primary care but the burden of recurrent infection is unknown.
Aim To investigate the incidence of and risk factors for recurrence of boil or abscess for individuals consulting primary care.
Design and setting Cohort study using electronic health records from primary care in the UK.
Method The Health Improvement Network (THIN) database was used to identify patients who had consulted their GP for a boil or abscess. Poisson regression was used to examine the relationship between age, sex, social deprivation, and consultation and to calculate the incidence of, and risk factors for, repeat consultation for a boil or abscess.
Results Overall, 164 461 individuals were identified who consulted their GP for a boil or abscess between 1995 and 2010. The incidence of first consultation for a boil or abscess was 512 (95% CI = 509 to 515) per 100 000 person-years in females and 387 (95% CI = 385 to 390) per 100 000 person-years in males. First consultations were most frequent in younger age groups (16–34 years) and those with the greatest levels of social deprivation. The rate of repeat consultation for a new infection during follow up was 107.5 (95% confidence interval [CI] = 105.6 to 109.4) per 1000 person-years. Obesity (relative risk [RR] 1.3, 95% CI = 1.2 to 1.3), diabetes (RR 1.3, 95% CI = 1.2 to 1.3), smoking (RR 1.3, 95% CI = 1.2 to 1.4), age <30 years (RR 1.2, 95% CI = 1.2 to 1.3), and prior antibiotic use (RR 1.4, 95% CI = 1.3–1.4) were all associated with repeat consultation for a boil or abscess.
Conclusion Ten percent of patients with a boil or abscess develop a repeat boil or abscess within 12 months. Obesity, diabetes, young age, smoking, and prescription of an antibiotic in the 6 months before initial presentation were independently associated with recurrent infection, and may represent options for prevention.
INTRODUCTION
Skin conditions are a major cause of consultations in primary care with approximately 2.4 million consultations per year in England and Wales.1 In 2010, based on an estimated UK population of 62 million, there were at least 280 000 primary care consultations for an abscess or boil.2
The epidemiology of boils and abscesses in primary care is poorly understood. Suppurative skin infections such as sycosis barbae,3 impetigo,4 and furunculosis5 are more common in patients who are colonised with Staphylococcus aureus and there is evidence that boils and abscesses are associated with social deprivation, overcrowding, the use of communal facilities, obesity, diabetes, and impaired immunity.6–9 Patients consulting primary care with a boil or abscess will be treated either with antibiotics or with surgical incision and drainage,10 but treatment options for those with recurrent boils is limited beyond addressing any underlying immune disorders and screening for diabetes.11 Patients can be tested for staphylococcal carriage and decolonised; in the UK this may include screening for the Panton-Valentine leukocidin toxin (PVL).12,13 However, the effectiveness of such interventions is questionable and they are resource intensive.14,15
There is a clear need for guidance on how to manage patients with recurrent disease in the community. The aim of this study was to determine the burden of recurrent disease and identify opportunities for secondary prevention in these patients.
METHOD
The THIN database is a source of anonymised clinical information about 11 million patients in primary care.16 In the UK, 98% of the population is registered with a GP who provides advice, treatment, prescriptions, and referrals and acts as a gatekeeper to specialist services.17 Practices that participate in the THIN scheme of data collection enter information on symptoms, diagnoses, treatments, laboratory investigations, and referrals every time a consultation takes place, using a hierarchical system of more than 103 000 Read Codes.18 Prescriptions are recorded using MULTILEX drug codes that link each formulation of each drug to the British National Formulary. Each patient in the database is allocated a Townsend score linked to their postcode (which is a composite measure of social deprivation based on levels of house ownership, overcrowding, car ownership, and unemployment). The practices are broadly representative of UK practices in terms of the age and sex of patients, the practice size and geographical location.19 Adequacy of death recording is assessed by determining the date at which the practice recorded mortality rates that were comparable to national age and sex standardised mortality rates (AMR date).20
How this fits in
Boils and abscesses are a common condition in primary care but there is little information on the burden of and risk factors for recurrent disease. In this study 10% of patients consulting for a boil or abscess developed a second boil or abscess during the following year. Consultations for boils and abscesses are most common in young, socially-deprived females and are associated with smoking, diabetes, obesity, and recent antibiotic use in the prior 6 months. Weight loss, smoking cessation, and reducing unnecessary antibiotic use could represent strategies for secondary prevention.
Patients were eligible for inclusion in the study if they were registered with a participating practice that met acceptable mortality recording (AMR) standards and was fully computerised between 1 January 1995 and 31 December 2011.21 Individuals were identified if they sought care for a boil, abscess, carbuncle, or furuncle, identified by a Read Code list (Appendix 1). Patients were excluded from the cohort if they were registered with a participating practice for <6 months before the date of first consultation (with the exception of infants aged <1 year), or if they had <1 year of follow-up data from that date. For the multivariate analysis patients with a diagnosis of hidradenitis suppurativa (a condition characterised by repeat boils) were excluded as these individuals were thought to represent a distinct patient group. Patients entered the cohort on the date of their first consultation for a boil or abscess and exited on the first of the following dates: the date of second (repeat) consultation for a boil or abscess; the date of death; the date the patient left the practice; or 31 December 2011.
Outcome was defined as any consultation that resulted in a record of a boil or abscess occurring between 3 weeks and 12 months after the date of first consultation for an abscess or boil. Recurrent boil or abscess was defined as a second consultation for a boil or abscess that occurred a minimum of 21 days after the previous consultation.
Exposure variables were grouped into past medical history, prescription history including antibiotics, treatment at baseline, and health indicators. Patients with a history of skin disease were identified if they had a record that included any of the following terms: ‘eczema’, ‘eczematous’, ‘atopic eczema’, ‘psoriasis’, ‘psoriatic’, and ‘dermatitis’, using the methods described by Davé et al.22 Finally, individuals diagnosed with hidradenitis suppurativa were identified. To identify patients with type 1 or type 2 diabetes a combination of the medical, prescription, and additional health data records was used. Patients with any record corresponding to diabetes were classified as diabetic, even if diagnosis was after the date of first consultation for a boil or abscess, on the assumption that these patients were diabetic at the time of infection.
Therapy records were used to identify patients prescribed oral corticosteroids or antibiotics in the 6 months prior to the date of index consultation. The British National Formulary was used for classification. Antibiotic prescriptions in the 30 days before first consultation for a boil or abscess were disregarded in case they represented treatment without a corresponding medical record for a boil or abscess.
Antibiotic treatment at baseline was defined as prescription of any antibiotic within 2 weeks of the date of the first consultation for a boil or abscess.
To define smoking status the record closest to the date of consultation for a boil or abscess was used in the context of the patients’ longitudinal record. Body mass index was defined using the record in the additional health data file that was closest to the date of consultation for a boil or abscess.
First, Poisson regression was used to investigate the relationship between initial consultations for a boil or abscess, age, sex, and social deprivation. Subsequently, a cohort was selected from the full dataset to study factors associated with recurrent boil or abscess. The baseline characteristics of this cohort were summarised using descriptive statistics. For variables where >5% of patients lacked a record (for body mass index [BMI] and smoking status), patients with missing data were compared to those with a complete case record. Provided those with a missing record were not systematically different from those with a complete record, missing records were reclassified as within the normal range on the assumption that this was why they lacked a record, for example smoking status is rarely recorded in non-smokers. Poisson regression was used to calculate the incidence of repeat consultation for a boil or abscess taking the denominator as the total number of person-years contributed by patients in the sample population. Univariate analyses were performed to identify factors associated with repeat consultation for a boil or abscess. The multivariable analysis was restricted to adults aged >16 years with a first consultation after 1995 because factors such as BMI were more likely to be recorded in adults and data were sparse before 1995. Multivariable Poisson regression was used to identify factors associated with recurrent boil accounting for age and sex. Variables identified from the univariate analysis and from knowledge of the literature were sequentially added to the model, assessing for interaction with age. Antibiotic treatment at diagnosis was added as a binary variable to the final model. However, antibiotic treatment was not considered as a potential risk factor, but rather as a marker of disease severity, because individuals with a severe infection are more likely to be treated with an antibiotic. Lastly, the final model was compared with and without a random effects term to assess whether there was evidence of clustering and to adjust precision estimates accordingly.
RESULTS
Between 1 January 1995 and 31 December 2010, 164 461 people had at least one consultation for a boil or abscess. The overall incidence of consultation for a boil or abscess was higher in females compared to males at 512 (95% confidence interval [CI] = 509 to 515) versus 387 (95% CI = 385 to 390) per 100 000 person-years respectively. However this relationship was age-dependent and males aged ≥65 years had higher rates of consultation compared to females (Figure 1). The rate of consultation increased rapidly from adolescence, peaking in male 20–24-year-olds and female 30–34-year-olds, and declining thereafter (Figure 1). Consultations were most frequent in individuals with the highest levels of social deprivation and this relationship was more marked for females than males (Figure 2).
The relationship between age and consultation rate for a boil or abscess in male and female patients.
Relationship between social deprivation and consultation for a boil or abscess by sex.
In total, 122 473 had a consultation for a boil or abscess at least 6 months after they registered and with at least 1 year of follow-up. The mean age at infection was comparable for both males and females at 41 years (Table 1). Twelve per cent (14 388/122 473) of the cohort had a recorded diagnosis of diabetes and less than one-fifth (22 804/122 473) had a record of an underlying skin condition such as psoriasis, eczema, or dermatitis. Diagnosis of hidradenitis suppurativa was rare, recorded in <1% (429/51 986) of males and 2.5% (1751/70 487) of females consulting for a boil or abscess. Twenty-eight per cent (34 316/122 473) had been prescribed at least one antibiotic in the 6 months before they were diagnosed with their first abscess or boil and approximately one-third were current smokers (35 754/122 473). At diagnosis 71% (86 427/122 473) of patients consulting for a boil or abscess were prescribed an antibiotic and 6% (7609/122 473) had a record of surgical incision and drainage (Table 2). Flucloxacillin was the most commonly-prescribed antimicrobial, prescribed in over half of all cases (66 879/122 473). Mupirocin and chlorhexidine were rarely prescribed to eradicate S. aureus carriage (<2%).
Baseline characteristics of cohort
Treatment of patients with first presentation for a boil or abscess
Nine per cent (4762/51 986) of males and 11% (7550/70 487) of females had at least one repeat consultation for a boil or abscess in the 3 weeks to 12 months following their first consultation and 3% (3622/122 473) had two or more repeat consultations, separated by at least 3 weeks. The overall rate of repeat consultation for a boil or abscess during follow-up was 107.5 (95% CI = 105.6 to 109.4) per 1000 person-years.
In the multivariable analysis there was no evidence of a clinically-relevant interaction between age and sex (Table 3). Obesity (relative risk [RR] 1.3, 95% CI = 1.2 to 1.3), diabetes (RR 1.3, 95% CI = 1.2 to 1.3), smoking (RR 1.3, 95% CI = 1.2 to 1.4) and an antibiotic prescription in the prior 6 months (RR 1.4, 95% CI = 1.3 to 1.4) were all independently associated with a repeat consultation for an abscess or boil. The rate of repeat consultation was highest in individuals aged 16–34 years and declined with increasing age. Antibiotic treatment and surgical treatment were not assessed as independent risk factors for recurrence because of the risk of confounding by disease severity (those at greatest risk of recurrent severe disease are more likely to be treated). Furthermore, the estimated rate ratios were not affected by inclusion of antibiotic treatment in the model as a binary variable, as a marker of disease severity. Inclusion of social deprivation (Townsend score) or year of first consultation did not impact on the estimated rate ratios.
Poisson regression model of the relationship between clinical and demographic risk factors and consultation for recurrent boil or abscess, 1995–2011
DISCUSSION
Summary
In this large study of patients attending primary care, 10% reconsulted for a boil or abscess at least once within 1 year of diagnosis. Infections were more common in female patients but followed a marked age-distribution in both sexes, with the greatest burden of disease in socially-deprived young adults. Factors independently associated with repeat consultation for boil or abscess were obesity, diabetes, chronic skin disease, current smoking, and antibiotic use in the 6 months prior to the initial consultation.
Strengths and limitations
This study’s strengths lie in its scale and the fact that the dataset is broadly nationally representative, containing the medical records of approximately 6% of the UK population.17 The limitations relate to the fact that data were recorded for patient management and not research. Read Codes were used to identify patients consulting for a boil or abscess. However, some GPs may prescribe an antibiotic without recording a diagnostic code, thereby underestimating incidence.
The actual incidence of recurrent disease in the community may have been higher because patients with a boil or abscess could attend an urgent care centre or go directly to accident and emergency for treatment. First recurrence was measured only and some patients will have multiple episodes of recurrent boil. By using an arbitrary cut-off value of 3 weeks to distinguish between people with a second (recurrent) infection and those with a prolonged first infection, some patients may have been misclassified, potentially biasing estimates of the recurrence rate.
Comparison with existing literature
Boils and abscesses are generally thought to be more common in males, although there is little published data to support this.10 This study found that females aged <65 years were much more likely to consult for a boil or abscess than males, with a reversal of this trend in older age groups. This may be partly explained by the increased tendency of female patients to consult their GP,23 but it is striking that the increased rate of consultation starts in puberty and declines with increasing age and decreasing fertility. This suggest hormonal changes may be important, not only in adolescence but throughout a female’s reproductive years.7 Alternatively behavioural factors such as hair removal through waxing or shaving of non-facial sites could underlie the increased rate of consultation in females. In both sexes increased social deprivation was associated with an increased rate of consultation, and this may indicate that predisposing behaviours and other risk factors such as obesity are socially stratified.
This study found that a substantial proportion of patients re-consult their GP for a further episode of boil or abscess within 1 year of first diagnosis. This is unlikely to represent ongoing treatment of a single infection because consultations were separated by a minimum of 3 weeks. In the UK, the mainstay of treatment for patients with recurrent boils or abscesses if they are colonised with S. aureus is decolonisation therapy, based on guidance from the National Institute for Health and Care Excellence (NICE).10 This treatment to eradicate S. aureus carriage has been used effectively in settings such as surgical wards to reduce the risk of postoperative infection.24 More recently, community studies from the US have shown decolonisation may reduce the risk of recurrent boils and abscesses, particularly when an entire household is treated simultaneously.14,15 However there is little consensus on how decolonisation should be undertaken in the community or how to manage patients with recurrent boils who remain persistently colonised despite treatment.
The rate of recurrent infection was strongly associated with an antibiotic prescription in the 6 months before the patient’s first consultation for a boil or abscess, which may reflect that individuals who are immunosuppressed are more likely to be prescribed an antibiotic. Alternatively, prior antibiotic treatment may actually increase an individuals’ subsequent risk of infection. Previous research has linked recent antibiotic use to community-MRSA (Methicillin-resistant Staphylococcus aureus),25 and it is plausible that antibiotics may encourage carriage of drug-resistant or virulent strains because antibiotics place selective pressure on antibiotic-susceptible bacteria. A further explanation is that individuals with prior antibiotic treatment have a lower threshold for consultation in general, and are therefore more likely to reconsult for a further infection, potentially introducing a spurious association between antibiotic treatment and the rate of repeat consultation. However, the fact that this analysis is restricted to those who had consulted at least once for a boil or abscess decreases the likelihood of this being a major effect.
Implications for research and practice
Boils and abscesses are a common problem and the management of recurrent disease is difficult. There is a clear need for treatment guidance in this area. Modifiable risk factors such as obesity, smoking, and antibiotic overuse are common in the general population and may represent targets for the secondary prevention of boils and abscesses in primary care.
Appendix 1. Read Code lists for boils and abscesses
Read Code lists were developed to identify a patient with an abscess or boil using a three-stage process. First a word-search was performed on the Read Code dictionary using the terms ‘abscess’, ‘boil’, ‘furuncle’, ‘carbuncle’ and ‘staphylococcal skin’. Next Read Codes identified in the word-search were included as search terms to identify additional codes. Finally, unwanted codes were manually excluded and reviewed and the code list was finalised with all research collaborators including a GP familiar with Vision. Codes were excluded where the diagnosis of boil or abscess was mixed with a diagnosis of cellulitis, as were codes for parionychias and perionychias because patients with these conditions tended to be older and did not reflect the target population of young and healthy individuals with community-onset disease. Read Code lists for boils and abscesses are included.
| Read Code | Description |
|---|---|
| 7303100 | Drainage of abscess of external ear |
| 7G25011 | Drainage of abscess of head or neck |
| 7G25012 | Drainage of boil of skin of head or neck |
| 7G25111 | Drainage of abscess NEC |
| 7G25112 | Drainage of boil of skin NEC |
| 7G25211 | Incision of boil of skin of head or neck |
| 7G25311 | Incision of boil of skin NEC |
| 7G25700 | Incision and drainage of abscess |
| F4G0200 | Orbital abscess |
| F501111 | Abscess external ear |
| F506.00 | Abscess of external ear |
| H1y1000 | Nasal septum abscess |
| H1y1011 | Boil in nose |
| K310.11 | Abscess breast non puerperal |
| K310400 | Acute nonpuerperal breast abscess |
| K310500 | Chronic nonpuerperal breast abscess |
| K310600 | Chronic subareolar nonpuerperal abscess |
| K318.00 | Breast abscess |
| L450.11 | Abscess of nipple — obstetric |
| L451.00 | Obstetric breast abscess |
| L451000 | Obstetric breast abscess unspecified |
| L451100 | Obstetric breast abscess — delivered |
| L451200 | Obstetric breast abscess — delivery with postnatal complication |
| L451300 | Obstetric breast abscess with antenatal complication |
| L451400 | Obstetric breast abscess with postnatal complication |
| L451z00 | Obstetric breast abscess NOS (not otherwise specified) |
| M00..00 | Carbuncle |
| M000.00 | Carbuncle of face |
| M000000 | Carbuncle of ear |
| M000100 | Carbuncle of face (excluding eye) |
| M000200 | Carbuncle of nasal septum |
| M000300 | Carbuncle of temple region |
| M000z00 | Carbuncle of face NOS |
| M001.00 | Carbuncle of neck |
| M002.00 | Carbuncle of trunk |
| M002000 | Carbuncle of chest wall |
| M002100 | Carbuncle of breast |
| M002200 | Carbuncle of back |
| M002400 | Carbuncle of umbilicus |
| M002500 | Carbuncle of flank |
| M002600 | Carbuncle of groin |
| M002700 | Carbuncle of perineum |
| M002z00 | Carbuncle of trunk NOS |
| M003.00 | Carbuncle of upper arm and forearm |
| M003000 | Carbuncle of shoulder |
| M003100 | Carbuncle of axilla |
| M003200 | Carbuncle of upper arm |
| M003300 | Carbuncle of elbow |
| M003400 | Carbuncle of forearm |
| M003z00 | Carbuncle of upper arm and forearm NOS |
| M004.00 | Carbuncle of hand |
| M004000 | Carbuncle of wrist |
| M004100 | Carbuncle of thumb |
| M004200 | Carbuncle of finger |
| M004z00 | Carbuncle of hand NOS |
| M005.00 | Carbuncle of buttock |
| M005000 | Carbuncle of anus |
| M005100 | Carbuncle of gluteal region |
| M005z00 | Carbuncle of buttock NOS |
| M006.00 | Carbuncle of leg (excluding foot) |
| M006000 | Carbuncle of hip |
| M006100 | Carbuncle of thigh |
| M006200 | Carbuncle of knee |
| M006300 | Carbuncle of lower leg |
| M006400 | Carbuncle of ankle |
| M006z00 | Carbuncle of leg (excluding foot) NOS |
| M007.00 | Carbuncle of foot |
| M007000 | Carbuncle of foot unspecified |
| M007100 | Carbuncle of heel |
| M007200 | Carbuncle of toe |
| M007z00 | Carbuncle of foot NOS |
| M00y.00 | Carbuncle of other specified site |
| M00y000 | Carbuncle of head (excluding face) |
| M00yz00 | Carbuncle of other specified site NOS |
| M00z.00 | Carbuncle NOS |
| M01..00 | Furuncle — boil |
| M010.00 | Boil of face |
| M010000 | Boil of ear |
| M010100 | Boil of face (excluding eye) |
| M010200 | Boil of nasal septum |
| M010300 | Boil of temple region |
| M010400 | Boil of external nose |
| M010z00 | Boil of face NOS |
| M011.00 | Boil of neck |
| M012.00 | Boil of trunk |
| M012000 | Boil of chest wall |
| M012100 | Boil of breast |
| M012200 | Boil of back |
| M012300 | Boil of abdominal wall |
| M012400 | Boil of umbilicus |
| M012500 | Boil of flank |
| M012600 | Boil of groin |
| M012z00 | Boil of trunk NOS |
| M013.00 | Boil of upper arm and forearm |
| M013000 | Boil of shoulder |
| M013100 | Boil of axilla |
| M013200 | Boil of upper arm |
| M013300 | Boil of elbow |
| M013400 | Boil of forearm |
| M013z00 | Boil of upper arm and forearm NOS |
| M014.00 | Boil of hand |
| M014000 | Boil of wrist |
| M014100 | Boil of thumb |
| M014200 | Boil of finger |
| M014z00 | Boil of hand NOS |
| M015.00 | Boil of buttock |
| M015000 | Boil of anus |
| M015100 | Boil of gluteal region |
| M015z00 | Boil of buttock NOS |
| M016.00 | Boil of leg (excluding foot) |
| M016000 | Boil of hip |
| M016100 | Boil of thigh |
| M016200 | Boil of knee |
| M016300 | Boil of lower leg |
| M016400 | Boil of ankle |
| M016z00 | Boil of leg (excluding foot) NOS |
| M017.00 | Boil of foot |
| M017000 | Boil of foot unspecified |
| M017100 | Boil of heel |
| M017200 | Boil of toe |
| M017z00 | Boil of foot NOS |
| M01y.00 | Boil of other specified site |
| M01y000 | Boil of head (excluding face) |
| M01yz00 | Boil of other specified site NOS |
| M01z.00 | Boil NOS |
| M01z.11 | Recurrent boils |
| M01z.12 | Boils of multiple sites |
| M01z000 | Multiple boils |
| M020100 | Finger pulp abscess |
| M03..11 | Abscess of skin area excluding digits of hand or foot |
| M034011 | Abscess of dorsum of hand |
| M034012 | Abscess of palm of hand |
| M03y011 | Abscess of scalp |
| M03z100 | Abscess NOS |
| M09..00 | Cutaneous abscess |
| M090.00 | X]Abscess of face [ |
| M091.00 | X]Abscess of neck [ |
| M092.00 | X]Abscess of trunk [ |
| M092000 | X]Abscess of buttock [ |
| M093.00 | X]Abscess of buttock [ |
| M094.00 | X]Abscess of limb [ |
| M094000 | X]Abscess of axilla [ |
| M095.00 | Skin abscess |
Notes
Funding
No funding was obtained to support this research study. Laura J Shallcross was funded by a Medical Research Council training fellowship G0501879.
Ethical approval
Freely submitted; externally peer reviewed.
Provenance
The authors have declared no competing interests.
Competing interests
The authors have declared no competing interests.
Discuss this article
Contribute and read comments about this article: bjgp.org/letters
- Received January 12, 2015.
- Revision requested April 13, 2015.
- Accepted April 30, 2015.
- © British Journal of General Practice 2015
REFERENCES
In this issue
Jump to section
More in this TOC Section
Related Articles
Cited By...
