Content uploaded by Kapil Sahnan
Author content
All content in this area was uploaded by Kapil Sahnan on Sep 14, 2017
Content may be subject to copyright.
Content uploaded by Kapil Sahnan
Author content
All content in this area was uploaded by Kapil Sahnan on Sep 02, 2017
Content may be subject to copyright.
Original article
Natural history of anorectal sepsis
K. Sahnan1,2,3 , A. Askari1,3,S.O.Adegbola
1,2,3,P.J.Tozer
2,3, R. K. S. Phillips2,3 , A. Hart2,3 and
O. D. Faiz1,2,3
1Surgical Epidemiology, Trials and Outcome Centre and 2Fistula Research Unit, St Mark’s Hospital and Academic Institute, Harrow, and 3Department
of Surgery and Cancer, Imperial College, St Mary’s Hospital, London, UK
Correspondence to: Mr K. Sahnan, Department of Surgery and Cancer, Imperial College, St Mary’s Hospital, Praed Street, London W2 1NY, UK (e-mail:
kapil.sahnan@nhs.net; @KSahnan, @alan_askari, @StMarksHospital, @StMarksFRU, @philtozer1, @OmarFaiz_SETOC).
Background: Progression from anorectal abscess to stula is poorly described and it remains unclear
which patients develop a stula following an abscess. The aim was to assess the burden of anorectal
abscess and to identify risk factors for subsequent stula formation.
Methods: The Hospital Episode Statistics database was used to identify all patients presenting with
new anorectal abscesses. Cox regression analysis was undertaken to identify factors predictive of stula
formation.
Results: A total of 165 536 patients were identied in the database as having attended a hospital in
England with an abscess for the rst time between 1997 and 2012. Of these, 158 713 (95⋅9 per cent)
had complete data for all variables and were included in this study, the remaining 6823 (4⋅1 per cent) with
incomplete data were excluded from the study. The overall incidence rate of abscess was 20⋅2 per 100 000.
The rate of subsequent stula formation following an abscess was 15⋅5 per cent (23 012 of 148 286) in
idiopathic cases and 41⋅6 per cent (4337 of 10 427 in patients with inammatory bowel disease (IBD)
(26⋅7 per cent coded concurrently as ulcerative colitis; 47⋅2 per cent coded as Crohn’s disease). Of all
patients who developed a stula, 67⋅5 per cent did so within the rst year. Independent predictors of
stula formation were: IBD, in particular Crohn’s disease (hazard ratio (HR) 3⋅51; P<0⋅001), ulcerative
colitis (HR 1⋅82; P<0⋅001), female sex (HR 1⋅18; P<0⋅001), age at time of rst abscess 41– 60 years (HR
1⋅85 versus less than 20 years; P<0⋅001), and intersphincteric (HR 1⋅53; P<0⋅001) or ischiorectal (HR
1⋅48; P<0⋅001) abscess location compared with perianal. Some 2⋅9 per cent of all patients presenting
with a new abscess were subsequently diagnosed with Crohn’s disease; the median time to diagnosis was
14 months.
Conclusion: The burden of anorectal sepsis is high, with subsequent stula formation nearly three times
more common in Crohn’s disease than idiopathic disease, and female sex is an independent predictor of
stula formation following abscess drainage. Most stulas form within the rst year of presentation with
an abscess.
Presented to the European Crohn’s and Colitis Organization Conference, Amsterdam, The Netherlands, March 2016,
and a meeting of the Royal Society of Medicine, London, UK, October 2015; published in abstract form as J Crohns
Colitis 2016; 10(Suppl 1): S75
Paper accepted 11 May 2017
Published online in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.10614
Introduction
Anorectal abscesses are common, with 18 000 patients
affected each year in England1. The condition is treated
primarily by incision and drainage in the operating the-
atre by a general surgeon. A proportion of patients sub-
sequently develop a stula; however, estimating the inci-
dence of anal stula is difcult and current evidence is
inconsistent. Sainio2derived the incidence of anal stula
from operating room data in a single institution in Helsinki
between 1969 and 1978; the incidence was 8⋅6 per 100 000
population per year (12⋅6 per 100 000 for males and 5⋅6per
100 000 for females)2. Zanotti and colleagues3reported a
higher incidence, but this was estimated from data from a
single year; based on a review of European databases the
incidence was 1⋅84 per 10 000 in the UK (2003–2004) and
2⋅32 per 10 000 in Italy (2002).
© 2017 BJS Society Ltd BJS
Published by John Wiley & Sons Ltd
K. Sahnan, A. Askari, S. O. Adegbola, P. J. Tozer, R. K. S. Phillips, A. Hart and O. D. Faiz
Table 1 ICD-10 and OPCS codes used to identify patients with abscess, stula, inammatory bowel disease and diabetes mellitus
Abscess Fistula IBD Diabetes
ICD-10 codes K61.0 Perianal abscess K60.3 Anal stula K50 Crohn’s disease E10 Type 1 diabetes mellitus
K61.1 Perirectal abscess K60.4 Rectal stula K51 Ulcerative colitis E11 Type 2 diabetes mellitus
K61.2 Anorectal abscess K60.5 Anorectal stula
K61.3 Ischiorectal abscess
K61.4 Intrasphincteric abscess
OPCS codes H58.1 Drainage of ischiorectal abscess H55.1 Laying open of low anal stula
H58.2 Drainage of perianal abscess H55.2 Laying open of high anal stula
H58.3 Drainage of perirectal abscess H55.3 Laying open of anal stula NEC
H55.4 Insertion of seton into high anal
stula and partial laying open of track
HFQ
H55.5 Fistulography of anal stula
H55.6 Probing of perineal stula
H55.7 Repair of anal stula using plug
H31.1 Image-guided percutaneous
occlusion of colorectal stula
H31.2 Image-guided transluminal
occlusion of colorectal stula
H41.4 Perianal mucosal proctectomy and
endoanal anastomosis
IBD, inammatory bowel disease; NEC, note elsewhere classied; HFQ, however further qualied.
The cryptoglandular hypothesis proposes that infection
of an intersphincteric gland via an internal opening leads
to stulation out to the perianal skin. However, not all
anorectal abscesses lead to persistent stulas. Oliver and
co-workers4reported that 83 per cent of 200 abscesses
had an internal opening when examined by an experienced
proctologist. Treating stulas in the acute phase reduced
the recurrence rate (from 29 to 5 per cent), demonstrat-
ing that not all patients who had abscesses with an internal
opening subsequently presented with a stula. Read and
Abcarian5similarly found that 40 per cent of abscesses pro-
gressed to stula formation, but other studies6–11 reported
rates ranging from 26 to 87 per cent. This was even the case
in studies in which abscesses in all patients had an internal
opening (acute stula)12,13. This means that some patients
with abscesses associated with an internal opening would
never present with a persistent stula14. Treating the stula
in the acute phase also risks iatrogenic injury, either to the
sphincter or by the creation of new tracts.
It would be helpful to stratify patients with an abscess
according to risk of presenting with a persistent stula15.
Other than Crohn’s disease, there are few reliable clinical
factors that predict which abscesses will develop into
stulas16. It is important to be able to predict which
abscesses are likely to stulate, so clinicians can make
informed follow-up plans. Follow-up after abscess is
clinician-dependent; most simple abscesses are not fol-
lowed up. Understanding which patients are more likely
to develop stulation will allow clinicians to follow up
those at high risk of stula development and prevent
complications15.
Between 28 and 38 per cent of patients with Crohn’s
disease suffer from perianal disease17,18. Perianal Crohn’s
disease is associated with a higher rate of stula recur-
rence following treatment and a shorter median time
to recurrence2,3. It represents a distinct and aggressive
phenotype19 and is one of the most disabling manifestations
of Crohn’s disease20; 12–39 per cent of patients with peri-
anal Crohn’s disease undergo proctectomy for intractable
proctitis21 –24. Early identication of patients with perianal
Crohn’s disease is important to avoid complications.
The aim of this population-based study was to assess the
burden of new diagnosis of anorectal abscess and identify
risk factors that lead to subsequent stula formation.
Methods
Ethical approval for the study was obtained through the
National Research Ethics Service (project ID 134045,
Research Ethics Committee reference 13/LO/1235) and
local Research and Development permission was sought
(reference 13.051) from London North West Hospitals
NHS Trust.
Hospital Episode Statistics (HES) is an administrative
data set with almost complete capture of all hospital
episodes in England since its inception in 1987. Within
the data set, diagnoses are coded according to the WHO
ICD-10 classication. The diagnostic codes and opera-
tive codes used to identify abscess, stula and inamma-
tory bowel disease (IBD) are shown in Ta b l e 1 .Patients
were identied according to these codes, capturing inpa-
tient, outpatient and emergency attendances. The HES
© 2017 BJS Society Ltd www.bjs.co.uk BJS
Published by John Wiley & Sons Ltd
Natural history of anorectal sepsis
Table 2 Baseline demographics by abscess type
Perianal Perirectal Anorectal Ischiorectal Intersphincteric Total
(n=136 021) (n=3221) (n=805) (n=17 267) (n=1399) (n=158 713) P†
Age at time of rst abscess (years) <0⋅001
<20 12 687 (9⋅3) 165 (5⋅1) 56 (7⋅0) 995 (5⋅8) 39 (2⋅8) 13 942 (8⋅8)
21– 40 59 798 (44⋅0) 749 (23⋅3) 270 (33⋅5) 6383 (37⋅0) 588 (42⋅0) 67 788 (42⋅7)
41– 60 46 076 (33⋅9) 1116 (34⋅6) 272 (33⋅8) 6587 (38⋅1) 561 (40⋅1) 54 612 (34⋅4)
>60 17 460 (12⋅8) 1191 (37⋅0) 207 (25⋅7) 3302 (19⋅1) 211 (15⋅1) 22 371 (14⋅1)
Sex <0⋅001
M 95 629 (70⋅3) 2011 (62⋅4) 518 (64⋅3) 10 837 (62⋅8) 1017 (72⋅7) 110 012 (69⋅3)
F 40 392 (29⋅7) 1210 (37⋅6) 287 (35⋅7) 6430 (37⋅2) 382 (27⋅3) 48 701 (30⋅7)
Diabetes <0⋅001
No 126 085 (92⋅7) 2913 (90⋅4) 739 (91⋅8) 15 492 (89⋅7) 1326 (94⋅8) 146 555 (92⋅3)
Type 1 2326 (1⋅7) 50 (1⋅6) 6 (0⋅7) 379 (2⋅2) 10 (0⋅7) 2771 (1⋅7)
Type 2 7610 (5⋅6) 258 (8⋅0) 60 (7⋅5) 1396 (8⋅1) 63 (4⋅5) 9387 (5⋅9)
IBD <0⋅001
No 127 487 (93⋅7) 2792 (86⋅7) 715 (88⋅8) 16 040 (92⋅9) 1252 (89⋅5) 148 286 (93⋅4)
Crohn’s disease 6252 (4⋅6) 258 (8⋅0) 58 (7⋅2) 895 (5⋅2) 108 (7⋅7) 7571 (4⋅8)
Ulcerative colitis 2282 (1⋅7) 171 (5⋅3) 32 (4⋅0) 332 (1⋅9) 39 (2⋅8) 2856 (1⋅8)
Social deprivation* <0⋅001
Q1 (afuent) 57 639 (42⋅4) 1515 (47⋅0) 357 (44⋅3) 7450 (43⋅1) 661 (47⋅2) 67 622 (42⋅6)
Q2 42 792 (31⋅5) 978 (30⋅4) 251 (31⋅2) 5341 (30⋅9) 422 (30⋅2) 49 784 (31⋅4)
Q3 24 111 (17⋅7) 503 (15⋅6) 130 (16⋅1) 3064 (17⋅7) 208 (14⋅9) 28 016 (17⋅7)
Q4 9608 (7⋅1) 179 (5⋅6) 55 (6⋅8) 1193 (6⋅9) 87 (6⋅2) 11 122 (7⋅0)
Q5 (deprived) 1871 (1⋅4) 46 (1⋅4) 12 (1⋅5) 219 (1⋅3) 21 (1⋅5) 2169 (1⋅4)
Charlson Co-morbidity Index score <0⋅001
0 117 174 (86⋅1) 2366 (73⋅5) 664 (82⋅5) 14 151 (82⋅0) 1221 (87⋅3) 135 576 (85⋅4)
1– 3 17 830 (13⋅1) 720 (22⋅4) 128 (15⋅9) 2883 (16⋅7) 169 (12⋅1) 21 730 (13⋅7)
>3 1017 (0⋅7) 135 (4⋅2) 13 (1⋅6) 233 (1⋅3) 9 (0⋅6) 1407 (0⋅9)
Abscess progressed to stula <0⋅001
No 113 760 (83⋅6) 2741 (85⋅1) 672 (83⋅5) 13 151 (76⋅2) 1040 (74⋅3) 131 364 (82⋅8)
Yes 22 261 (16⋅4) 480 (14⋅9) 133 (16⋅5) 4116 (23⋅8) 359 (25⋅7) 27 349 (17⋅2)
Values in parentheses are percentages. *Based on Indices of Multiple Deprivation score. IBD, inammatory bowel disease. †Kruskal– Wallis test.
data set from April 1997 to March 2012 was analysed. The
rst patient admission with an abscess code was taken as
the index abscess admission. Patients admitted to hospital
owing to an anal abscess or stula before or at the same
time as their index abscess admission were excluded from
the study.
Age and social deprivation data were also obtained.
Deprivation was assessed using the Indices of Multiple
Deprivation, a measure of relative levels of depriva-
tion, classied in quintiles (Q1, most afuent; Q5, most
deprived). Pre-existing co-morbidity was evaluated using
the Charlson Co-morbidity Index. The time from admis-
sion with an abscess to the development of a stula was
calculated. Censoring of patients was carried out as fol-
lows. For patients in whom stula formation following
abscess drainage occurred, the admission date for the rst
episode with a stula after an abscess episode was taken
as the censor date. Patients who did not have a stula
episode were censored at the date of death (if applicable)
or the last admission date for the data set (31 March
2012).
Statistical analysis
Groups were compared using Kruskal–Wallis analyses.
Binary logistic regression and Cox proportional hazard
regression models were developed to determine fac-
tors associated with progression from abscess to stula.
Variables were individually tested initially against the
outcome (progress to stula from abscess). Kaplan–Meier
methodology was used to assess stula rates over time.
Variables with P≤0⋅001 were considered signicant and
entered into a multivariable (adjusted) regression model.
In multivariable analyses, factors with P≤0⋅050 were
considered statistically signicant. All statistical analyses
were carried out using SPSS®version 21.0 (IBM, Armonk,
New York, USA).
Results
A total of 165 536 patients aged over 18 years had an
episode of new abscess formation between 1997 and 2012.
Complete data sets were available for 158 713 (95⋅9per
cent); there were data missing for age, sex and indices of
© 2017 BJS Society Ltd www.bjs.co.uk BJS
Published by John Wiley & Sons Ltd
K. Sahnan, A. Askari, S. O. Adegbola, P. J. Tozer, R. K. S. Phillips, A. Hart and O. D. Faiz
12 240
0·0
0·1
0·2
Cumulative stula formation
0·3
0·4
0·5
0·6
48 6036 84 96
Follow-up (months)
No. at risk
Not IBD
Crohn’s disease
Ulcerative colitis
148286
7571
2856
29724
2747
824
13459
774
267
10876
494
204
10033
421
188
9231
379
156
8918
338
162
8551
331
131
8140
304
134
7645
260
142
7286
278
117
7040
271
125
6994
267
113
6923
226
113
6971
257
103
6456
222
76
72 120 132108 156 16 8144 180
Not IBD
Crohn’s disease
Ulcerative colitis
Fig. 1 Fistula formation over 15 years in cohorts with Crohn’s disease or ulcerative colitis and those without inammatory bowel disease
(IBD)
multiple deprivation. The 6823 patients (4⋅1 per cent) with
incomplete data were excluded from the study. The median
follow-up time for the whole cohort was 60 (i.q.r. 17– 115)
months. The majority of patients were men (110 012, 69⋅3
per cent) and the most commonly coded abscess type was
perianal (136 021, 85⋅7 per cent).
The median age at the time of rst presentation with an
abscess was 40 (i.q.r. 29–52) years. Abscesses were most
commonly identied in the 41–60-year age group. The
majority of stula episodes were in men (65⋅3 per cent).
Some 12 158 patients (7⋅7 per cent) had a diagnosis of
diabetes before their rst admission with an abscess, and
7571 (4⋅8 per cent) had a diagnosis code for Crohn’s disease
at some point in the study interval (Ta b l e 2 ).
Incidence of abscess
The crude incidence rate of rst admission to hospital with
an abscess in the study was 20⋅4 per 100 000 over 15 years
0
10
20
30
40
% of patients who developed a fistula
50
60
70
80
90
100
Not IBD Crohn’s disease Ulcerative colitis
Fig. 2 Of patients diagnosed with a stula this chart demonstrates
the proportion of patients with Crohn’s disease or ulcerative
colitis and those without inammatory bowel disease (IBD) who
had developed a stula by 1-year follow-up. Error bars represent
95 per cent condence intervals
in England. Census data from the Ofce of National
Statistics were used to establish population size and hence
an age standardized rate of 20⋅2 (95 per cent c.i. 20⋅1to
© 2017 BJS Society Ltd www.bjs.co.uk BJS
Published by John Wiley & Sons Ltd
Natural history of anorectal sepsis
Table 3 Predictors of stula formation identied by Cox regression analyses
Adjusted analysis
Unadjusted PHazard ratio P
Age at time of rst abscess (years) <0⋅001
<20 1⋅00 (reference)
21– 40 1⋅69 (1⋅61, 1⋅78) <0⋅001
41– 60 1⋅85 (1⋅76, 1⋅95) <0⋅001
>60 1⋅64 (1⋅54, 1⋅74) <0⋅001
Sex <0⋅001
M 1⋅00 (reference)
F 1⋅18 (1⋅15, 1⋅21) <0⋅001
Diabetes <0⋅001
No 1⋅00 (reference)
Type 1 0⋅42 (0⋅36, 0⋅48) <0⋅001
Type 2 0⋅76 (0⋅71, 0⋅82) <0⋅001
IBD <0⋅001
No 1⋅00 (reference)
Crohn’s disease 3⋅51 (3⋅38, 3⋅63) <0⋅001
Ulcerative colitis 1⋅82 (1⋅69, 1⋅96) <0⋅001
Social deprivation <0⋅001
Q1 (afuent) 1⋅00 (reference)
Q2 0⋅92 (0⋅90, 0⋅95) <0⋅001
Q3 0⋅84 (0⋅82, 0⋅87) <0⋅001
Q4 0⋅80 (0⋅76, 0⋅84) <0⋅001
Q5 (deprived) 0⋅73 (0⋅65, 0⋅82) <0⋅001
Charlson Co-morbidity Index score <0⋅001
0 1⋅00 (reference)
1–3 0⋅92 (0⋅88, 0⋅97) <0⋅001
>3 0⋅83 (0⋅68, 1⋅00) <0⋅001
Abscess type <0⋅001
Perianal 1⋅00 (reference)
Perirectal 0⋅87 (0⋅79, 0⋅95) <0⋅001
Anorectal 0⋅94 (0⋅80, 1⋅11) 0⋅499
Ischiorectal 1⋅48 (1⋅43, 1⋅53) <0⋅001
Intersphincteric 1⋅53 (1⋅38, 1⋅70) <0⋅001
Values in parentheses are 95 per cent condence intervals. IBD, inammatory bowel disease.
20⋅3) per 100 000. The age standardized rate does not
include the signicant number of patients with recurrent
abscesses or patients treated in the private sector.
Incidence and timing from abscess to stula
Over the 15 years, the overall rate of stula formation, as
dened by ICD-10 and OPCS codes, following anorec-
tal abscess was 17⋅2 per cent (27 349 of 158 713). Subclas-
sication by IBD status revealed that the rate of stula
formation following an abscess was 15⋅5 per cent (23 012
of 148 286) in idiopathic cases and 41⋅6 per cent (4337
of 10 427) among patients with IBD (26⋅7 per cent coded
concurrently as ulcerative colitis; 47⋅2 per cent coded as
Crohn’s disease).
Across the study population, of the 27 349 patients in
whom a perianal abscess progressed to stula, it did so
within 12 months in 18 469 (67⋅5 per cent). In a strati-
ed analysis using Kaplan–Meier methodology, among the
non-IBD population, 15 623 of 29 724 patients (52⋅6(95
per cent c.i. 51⋅6to53⋅0) per cent) progressed to stula
from anorectal abscess within 12 months. This rate of pro-
gression from anorectal abscess to stula was higher in
those with ulcerative colitis, of whom 537 of 824 patients
(65⋅2(61⋅4to68⋅1) per cent) progressed, and higher still
in patients with Crohn’s disease, of whom 2309 of 2747
patients (84⋅1(82⋅6to85⋅9) per cent) progressed within
12 months (all P<0⋅001) (Figs 1 and 2).
Timing of stula diagnosis
The median time to rst episode (inpatient, outpatient,
or accident and emergency) of stula following abscess
drainage was 7⋅0(i.q.r.6⋅9–7⋅1) months for patients who
developed a stula. Of patients diagnosed with a stula
within 1 year, the median time to stulation was 7⋅0(95
per cent c.i. 6⋅9to7⋅1) months for the idiopathic cohort,
5⋅0(4⋅5to5⋅5) months for patients with ulcerative colitis
© 2017 BJS Society Ltd www.bjs.co.uk BJS
Published by John Wiley & Sons Ltd
K. Sahnan, A. Askari, S. O. Adegbola, P. J. Tozer, R. K. S. Phillips, A. Hart and O. D. Faiz
First presentation
with anorectal
abscess
Time
Median time to diagnosis 14 (i.q.r. 4–40) months
41·5% diagnosed with
Crohn’s disease after first
abscess
Crohn’s disease
diagnosis before fistula
Crohn’s disease
diagnosis at time of fistula
Crohn’s disease
diagnosis after fistula
16·0% diagnosed with
Crohn’s disease at
time of first abscess
42·5% known diagnosis
of Crohn’s disease
before first abscess
Fig. 3 Diagnosis of Crohn’s disease following presentation with abscess and subsequent stula formation
and 5⋅0(4⋅8to5⋅2) months among those with Crohn’s
disease.
Factors associated with stula
Cox proportional hazards regression analyses was under-
taken to identify factors associated with diagnosis of a
stula after rst admission with an abscess (Ta b l e 3 ). Inde-
pendent predictors of stulation were female sex (hazard
ratio (HR) 1⋅18, 95 per cent c.i. 1⋅15 to 1⋅21; P<0⋅001),
age 41 –60 years (HR 1⋅85, 1⋅76 to 1⋅95; P<0⋅001), and
ischiorectal (HR 1⋅48, 1⋅43 to 1⋅53; P<0⋅001) or inter-
sphincteric (HR 1⋅53, 1⋅38 to 1⋅70; P<0⋅001) location of
the initial abscess. The presence of IBD was particularly
strongly associated with progression to stula compared
with idiopathic disease (ulcerative colitis: HR 1⋅82, 1⋅69
to 1⋅96, P<0⋅001; Crohn’s disease: HR 3⋅51, 3⋅38 to 3⋅63,
P<0⋅001).
Diabetes, an existing co-morbidity or a social depriva-
tion quintile greater than Q1 were not associated with an
increased risk of stula formation.
Diagnosis of Crohn’s disease in relation
to anorectal sepsis
Some 7571 patients in the cohort (4⋅8 per cent) had Crohn’s
disease. An established diagnosis of Crohn’s disease before
the rst admission with an anorectal abscess was recorded
in 3218 patients (42⋅5 per cent). A further 1209 patients
(16⋅0 per cent) were diagnosed with Crohn’s disease within
their admission for the rst abscess. The remaining 3144
patients (41⋅5 per cent) had a diagnostic code for Crohn’s
disease at a median of 14 (i.q.r. 4–40) months after their
rst abscess (Fig. 3).
Of the 3144 patients with Crohn’s disease diagnosed after
their abscess admission, 1725 (54⋅9 per cent) went on to
develop a stula within the study period. The median time
to diagnosis of the 958 patients who were diagnosed with
Crohn’s disease after the stula admission was 5 (i.q.r.
2–20) months.
Of the 158 713 patients presenting with an anorectal
abscess, 155 495 were not known to have Crohn’s disease
before admission. Of these, 4353 (2⋅8 per cent) were sub-
sequently diagnosed with Crohn’s disease, either at the
time of admission or subsequent to their admission with
an anorectal abscess.
Discussion
In this observational population-based study, the incidence
(age-standardized rate) of rst admission with an abscess
was 20⋅2 per 100 000 in England. Anorectal abscesses were
common, with approximately one-fth of all patients with
© 2017 BJS Society Ltd www.bjs.co.uk BJS
Published by John Wiley & Sons Ltd
Natural history of anorectal sepsis
an abscess going on to develop a stula. The majority
of stulas occurred within the rst year of the preceding
abscess and this was most pronounced in patients with
Crohn’s disease. Nearly half of all patients with Crohn’s
abscesses developed a stula.
The true incidence of both abscess and stula is poorly
dened in the literature15. The incidence in the present
study is an accurate reection for new abscesses in Eng-
land, but does not include recurrent abscesses or those in
patients presenting to the private sector. Current under-
standing suggests that the male population has a higher
incidence of abscess and stula25, the male :female ratio is
in the order of 2 : 1. Here, of patients with abscesses, 69⋅3
per cent were men. Of those who went on to develop a
stula, two-thirds were men. Based on the present analysis,
the mean annual number of new diagnoses of abscess was
10 580 in a population of 53⋅0 million26. From the new
abscess diagnoses, an overall incidence of abscess of 298⋅9
per 100 000 can be derived. [Correction added on 13
September 2017, after rst online publication: ‘incidence
of stula’ has been corrected to ‘incidence of abscess’]
This, however, underestimates the true burden, as it does
not include recurrent presentations with abscesses.
Analysis according to type of IBD revealed that almost
one-half of patients with Crohn’s abscesses (47⋅2 per cent)
and one-quarter of those with ulcerative colitis (26⋅7per
cent) went on to be diagnosed with a stula. Following
an abscess, the majority of subsequent stula presentations
occurred within the rst year. The data do not allow cal-
culation of the incidence of anorectal disease in IBD, but
do conrm that abscesses in patients with IBD are more
likely to stulate. The exact pathogenesis of stulas is less
clearly dened in IBD. The cause is likely to be multifac-
torial, including genetic, immunological and microbiolog-
ical factors. In addition to suppuration from anal glands,
other aetiological theories have been proposed, such as s-
tulas arising as a result of a deep penetrating ulcer in the
anorectum20. Patients with Crohn’s disease had an acceler-
ated diagnosis of stula, with the median time to diagnosis
decreasing from 7 to 5 months.
Although a high risk of underlying stula may be
expected in the Crohn’s cohort, an interesting observation
is the rate of stula formation among patients with ulcer-
ative colitis. For the purpose of interpretation, if a patient
had a diagnostic code for Crohn’s disease at any point dur-
ing the study, this was the nal IBD diagnosis recorded, in
an effort to mitigate the argument that some patients actu-
ally had Crohn’s disease and were misdiagnosed as having
ulcerative colitis. Nevertheless, misdiagnosis remains a
possibility and the development of perianal disease in a
patient with a known diagnosis of ulcerative colitis should
prompt clinicians to review the IBD status, and investigate
the possibility that the true diagnosis is Crohn’s disease.
Concomitant perianal disease has been described in 5
per cent of patients with ulcerative colitis by Zabana and
colleagues27. Another possible hypothesis for the increased
rate of stula formation in patients with ulcerative colitis
is that a combination of the systemic nature of the disease
and immunosuppressant medications might lead to an
increase in the formation of cryptoglandular-type stulas.
The present data showed a higher incidence of stula in
men but, surprisingly, women were 18 per cent more likely
to develop a stula following an abscess. A similar associa-
tion was found in a study28 of 146 patients, which reported
a higher incidence of stula formation in females, particu-
larly for anterior abscesses. However, the study contained
only 44 female patients and regression analysis was not
applied. The authors also found that there was no differ-
ence in bacterial growth by sex28. They hypothesized that
anatomical differences might explain the higher incidence
of anterior abscesses. Other hypotheses include inadequate
incision and drainage in the rst instance or higher rates of
anal digitation in women, but these need validation.
The majority of the abscesses in the present study were
coded as perianal, but the anatomical classication of
abscesses was determined by the operating surgeon. This
is often a junior trainee, who may use ‘perianal’ as an
umbrella term in place of ‘anorectal’. The incidence of
stula development was higher in patients who presented
with ischiorectal and intersphincteric abscesses. Given the
signicant preponderance of abscesses coded as perianal in
the present study, this result must be interpreted with cau-
tion. However, Sözener and co-workers29 also found that
ischiorectal (odds ratio 7⋅82) and intersphincteric (odds
ratio 3⋅35) abscesses had a higher risk of progression to s-
tula than perianal abscesses.
The relationship between socioeconomic status and s-
tula formation followed an almost linear trend from Q1
to Q5, with patients in the Q1 (afuent) socioeconomic
group having both a higher proportion of perianal sepsis
and a greater tendency to stula formation. This raises the
question of how socioeconomic status might be related to
abscess/stula presentation. It is not possible to draw any
conclusions from the present data, but the greater tendency
reported could be related to behavioural patterns and/or
access to healthcare.
Perhaps unsurprisingly, diabetes was not associated with
stula formation, diabetes being a precursor to cutaneous
infections. Patients with type 1 diabetes were less likely
to present with a stula after drainage of an abscess. In a
univariable analysis, Hamadani and co-workers30 reported
that patients with diabetes had a twofold increased risk of
© 2017 BJS Society Ltd www.bjs.co.uk BJS
Published by John Wiley & Sons Ltd
K. Sahnan, A. Askari, S. O. Adegbola, P. J. Tozer, R. K. S. Phillips, A. Hart and O. D. Faiz
anorectal sepsis recurrence compared with those without
diabetes. Yano et al.31 noted that diabetes was not a risk
factor for abscess recurrence.
Anal sepsis has been found to be the initial manifesta-
tion and the presenting complaint in 10–25 per cent of
Crohn’s diagnoses17,32. A more recent population-based
cohort study33 noted that up to one-fth of patients
develop perianal disease at least 6 months before lumi-
nal disease. In the present study, approximately three in
every 100 patients presenting with an anorectal abscess
went on to a diagnosis of Crohn’s disease. The median
time to diagnosis was just over 1 year. More than half
of these patients without a Crohn’s diagnosis at the time
of admission for abscess subsequently developed a s-
tula. In luminal disease, there is a benet in initiating
earlier biological therapy in appropriate patients34,35;this
has been shown to increase the remission rate36.Inperi-
anal disease, the impact of a delay in diagnosis on out-
comes is not known, but it is likely to be detrimental
given that perianal Crohn’s disease represents a severe
phenotype.
Although previous studies have demonstrated the high
diagnostic accuracy of routinely collected clinical data37,
as with any retrospective administrative study there is a
potential for coding error. This may be a problem particu-
larly in determining the site of abscess. Furthermore, in this
study, the rst admission with an abscess was taken as an
index admission for abscess and it was assumed that neither
stula nor abscess had occurred previously. It is possible
that some patients (especially those with an index admission
at the beginning of the data set in 1997) might have had an
abscess/stula previously that would not be captured by the
present data. The HES database does not contain informa-
tion on other confounders (for example smoking status) or
the role of drugs such as antitumour necrosis factor agents
and immunosuppressants. This study has provided an esti-
mation of the incidence of stula arising from an abscess
in the English National Health Service, but this will be
different from that in countries where there is no univer-
sal healthcare access or where the management of perianal
abscess differs.
Although HES data have inherent limitations, there are
important messages to be taken from this analysis. This
large population-based study reports a large number of
observations over 15 years with almost complete data cap-
ture. The results indicate that approximately one-fth of
all patients develop a stula following an abscess; women
have the greatest risk, although both abscess and stula
are more common in men. Nearly half of all patients with
Crohn’s abscesses develop a stula, the majority occurring
within 12 months. Some 2⋅8 per cent of patients without
a diagnosis of Crohn’s disease at the time of admission for
anorectal abscess will be found to have the disease after a
median of 14 months.
Disclosure
The authors declare no conict of interest.
References
1 Pearce L, Newton K, Smith SR, Barrow P, Smith J,
Hancock L et al.; North West Research Collaborative.
Multicentre observational study of outcomes after drainage
of acute perianal abscess. Br J Surg 2016; 103: 1063–1068.
2 Sainio P. Fistula-in-ano in a dened population. Incidence
and epidemiological aspects. Ann Chir Gynaecol 1984; 73:
219–224.
3 Zanotti C, Martinez-Puente C, Pascual I, Pascual M,
Herreros D, García-Olmo D. An assessment of the
incidence of stula-in-ano in four countries of the European
Union. Int J Colorectal Dis 2007; 22: 1459–1462.
4 Oliver I, Lacueva FJ, Pérez Vicente F, Arroyo A, Ferrer R,
Cansado P et al. Randomized clinical trial comparing simple
drainage of anorectal abscess with and without stula track
treatment. Int J Colorectal Dis 2003; 18: 107–110.
5 Read DR, Abcarian H. A prospective survey of 474 patients
with anorectal abscess. Dis Colon Rectum 1979; 22: 566–568.
6 Vasilevsky CA, Gordon PH. The incidence of recurrent
abscesses or stula-in-ano following anorectal suppuration.
Dis Colon Rectum 1984; 27: 126–130.
7 Raghavaiah NV. Anal stula in India. Int Surg 1976; 61:
243–245.
8 Hyman N. Anorectal abscess and stula. Prim Care 1999;
26: 69–80.
9 Marcus RH, Stine RJ, Cohen MA. Perirectal abscess. Ann
Emerg Med 1995; 25: 597–603.
10 Turra G, Gherardi GM, Mangiarotti S, Arrighi E.
[Recurrent anorectal abscesses.] Chir Ital 1984; 36:
266–271.
11 Fucini C. One stage treatment of anal abscesses and stulas.
A clinical appraisal on the basis of two different
classications. Int J Colorectal Dis 1991; 6: 12–16.
12 Ho YH, Tan M, Chui CH, Leong A, Eu KW, Seow-Choen
F. Randomized controlled trial of primary stulotomy with
drainage alone for perianal abscesses. Dis Colon Rectum 1997;
40: 1435–1438.
13 Tang CL, Chew SP, Seow-Choen F. Prospective
randomized trial of drainage alone vs. drainage and
stulotomy for acute perianal abscesses with proven internal
opening. Dis Colon Rectum 1996; 39: 1415–1417.
14 Schouten WR, van Vroonhoven TJ. Treatment of anorectal
abscess with or without primary stulectomy. Results of a
prospective randomized trial. Dis Colon Rectum 1991; 34:
60–63.
15 Sahnan K, Adegbola SO, Tozer PJ, Watfah J, Phillips RK.
Perianal abscess. BMJ 2017; 356: j475.
© 2017 BJS Society Ltd www.bjs.co.uk BJS
Published by John Wiley & Sons Ltd
Natural history of anorectal sepsis
16 Pigot F. Treatment of anal stula and abscess. JViscSurg
2015; 152(Suppl 2): S23–S29.
17 Hellers G, Bergstrand O, Ewerth S, Holmström B.
Occurrence and outcome after primary treatment of anal
stulae in Crohn’s disease. Gut 1980; 21: 525 –527.
18 Schwartz DA, Loftus EV, Tremaine WJ, Panaccione R,
Harmsen WS, Zinsmeister AR et al. The natural history of
stulizing Crohn’s disease in Olmsted County, Minnesota.
Gastroenterology 2002; 122: 875–880.
19 Lakatos PL, Czegledi Z, Szamosi T, Banai J, David G,
Zsigmond F et al. Perianal disease, small bowel disease,
smoking, prior steroid or early azathioprine/biological
therapy are predictors of disease behavior change in patients
with Crohn’s disease. World J Gastroenterol 2009; 15:
3504–3510.
20 Marzo M, Felice C, Pugliese D, Andrisani G, Mocci G,
Armuzzi A et al. Management of perianal stulas in Crohn’s
disease: an up-to-date review. World J Gastroenterol 2015; 21:
1394–1403.
21 Michelassi F, Melis M, Rubin M, Hurst RD. Surgical
treatment of anorectal complications in Crohn’s disease.
Surgery 2000; 128: 597–603.
22 Faucheron JL, Saint-Marc O, Guibert L, Parc R.
Long-term seton drainage for high anal stulas in Crohn’s
disease – a sphincter-saving operation? Dis Colon Rectum
1996; 39: 208–211.
23 Sangwan YP, Schoetz DJ, Murray JJ, Roberts PL, Coller JA.
Perianal Crohn’s disease. Results of local surgical treatment.
Dis Colon Rectum 1996; 39: 529–535.
24 Williams JG, Rothenberger DA, Nemer FD, Goldberg SM.
Fistula-in-ano in Crohn’s disease. Results of aggressive
surgical treatment. Dis Colon Rectum 1991; 34: 378–384.
25 McColl I. The comparative anatomy and pathology of anal
glands. Arris and Gale lecture delivered at the Royal College
of Surgeons of England on 25th February 1965. Ann R Coll
Surg Engl 1967; 40: 36 –67.
26 Ofce for National Statistics. 2011 UK Censuses; 2011.
http://ons.gov.uk/ons/guide-method/census/2011/uk-
census/index.html [accessed 20 February 2016].
27 Zabana Y, Van Domselaar M, Garcia-Planella E, Mañosa M,
San Román AL, Gordillo J et al. Perianal disease in patients
with ulcerative colitis: a case–control study. J Crohns Colitis
2011; 5: 338–341.
28 Hämäläinen KP, Sainio AP. Incidence of stulas after
drainage of acute anorectal abscesses. Dis Colon Rectum 1998;
41: 1357–1361.
29 Sözener U, Gedik E, Kessaf Aslar A, Ergun H, Halil Elhan
A, Memiko˘
glu O et al. Does adjuvant antibiotic treatment
after drainage of anorectal abscess prevent development of
anal stulas? A randomized, placebo-controlled,
double-blind, multicenter study. Dis Colon Rectum 2011; 54:
923–929.
30 Hamadani A, Haigh PI, Liu I-LA, Abbas MA. Who is at risk
for developing chronic anal stula or recurrent anal sepsis
after initial perianal abscess? Dis Colon Rectum 2009; 52:
217–221.
31 Yano T, Asano M, Matsuda Y, Kawakami K, Nakai K,
Nonaka M. Prognostic factors for recurrence following the
initial drainage of an anorectal abscess. Int J Colorectal Dis
2010; 25: 1495–1498.
32 Castaño-Milla C, Chaparro M, Saro C, Barreiro-de Acosta
M, García-Albert AM, Bujanda L et al. Effectiveness of
adalimumab in perianal stulas in Crohn’s disease patients
naive to anti-TNF therapy. J Clin Gastroenterol 2015; 49:
34–40.
33 Eglinton TW, Barclay ML, Gearry RB, Frizelle FA. The
spectrum of perianal Crohn’s disease in a population-based
cohort. Dis Colon Rectum 2012; 55: 773–777.
34 Vavricka SR, Spigaglia SM, Rogler G, Pittet V, Michetti P,
Felley C et al.; Swiss IBD Cohort Study Group. Systematic
evaluation of risk factors for diagnostic delay in inammatory
bowel disease. Inamm Bowel Dis 2012; 18: 496–505.
35 Schoepfer AM, Dehlavi M-A, Fournier N, Safroneeva E,
Straumann A, Pittet V et al.; IBD Cohort Study Group.
Diagnostic delay in Crohn’s disease is associated with a
complicated disease course and increased operation rate. Am
J Gastroenterol 2013; 108: 1744 –1753.
36 D’Haens G, Baert F, van Assche G, Caenepeel P, Vergauwe
P, Tuynman H et al.; Belgian Inammatory Bowel Disease
Research Group; North-Holland Gut Club. Early combined
immunosuppression or conventional management in
patients with newly diagnosed Crohn’s disease: an open
randomised trial. Lancet 2008; 371: 660–667.
37 Burns EM, Rigby E, Mamidanna R, Bottle A, Aylin P,
Ziprin P et al. Systematic review of discharge coding
accuracy. J Public Health (Bangkok) 2012; 34: 138– 148.
© 2017 BJS Society Ltd www.bjs.co.uk BJS
Published by John Wiley & Sons Ltd
