Persistent postoperative wound drainage has been shown to be a predictor of
infection after total joint
arthroplasty1,2.
Estimates of the risk of postoperative infection range from 1.3% to 50% among
patients with persistent wound
drainage3,4.
Additionally, prolonged drainage may lead to a longer hospital stay and
subsequent surgical procedures, thus increasing the health-care economic
burden5. To our
knowledge, there are no publications on the risk factors for prolonged wound
drainage following a primary total joint arthroplasty.
There have been anecdotal observations that different methods of
prophylaxis against deep venous thrombosis may be associated with differences
in the degree and duration of postoperative wound drainage. Currently used
regimens for prophylaxis against deep venous thrombosis include the
administration of Coumadin (warfarin), low-molecular-weight heparin, and
aspirin in conjunction with foot or leg compression devices. Because these
drugs and devices have different mechanisms of action and are used throughout
the early postoperative period, they may affect wound drainage differently.
Other factors, such as body mass index, surgical time, and estimated blood
loss, also have been anecdotally implicated. The purpose of this study was
twofold: to identify pharmacologic factors (prophylaxis against deep venous
thrombosis), surgical factors (estimated blood loss and surgical time), and
patient-specific factors (body mass index and drain output) that are
associated with the time until the postoperative wound is dry following
primary total joint arthroplasty, and to determine whether prolonged wound
drainage results in a longer hospital stay or increases the risk of early
postoperative wound infection.
We performed a retrospective observational study with use of prospectively
collected data on 2437 consecutive primary unilateral total hip and knee joint
arthroplasties performed at our institution, a tertiary care teaching
hospital, between January 1997 and July 2004. We limited our analysis to
patients in whom a closed suction drain (Hemovac; Zimmer, Warsaw, Indiana) had
been utilized and who had a normal coagulation profile. Drains were removed at
thirty-six hours or whenever drain output during any eight-hour shift fell
below 25 mL, whichever came first. All patients received routine preoperative
and postoperative prophylactic antibiotics (1 g of Ancef [cefazolin] if the
patient was not allergic to penicillin, or 600 mg of clindamycin), including
at least thirty minutes before the initial operative incision. Our
institutional review board approved this study.
Patients were grouped according to the type of prophylaxis against deep
venous thrombosis that they had received. The method of prophylaxis was
determined by the surgeon's preference and consisted of either warfarin
(Coumadin), with a target international normalized ratio of 2.0, begun on the
day of surgery; low-molecular-weight heparin (Lovenox [enoxaparin]) started
between twelve and twenty-four hours postoperatively; or aspirin (325 mg) in
conjunction with pneumatic compression devices (Kendall AV Impulse System,
Mansfield, Massachusetts) started on the morning after surgery. The measured
surgical variables included intraoperative estimated blood loss, surgical
time, and drain output. The body mass index was calculated for each patient,
with the patient classified as being of normal weight (body mass index of
=24.9 kg/m2), overweight (body mass index of 25.0 to 29.9
kg/m2), obese (body mass index of 30.0 to 39.9 kg/m2),
or morbidly obese (body mass index of =40 kg/m2). Secondary
patient and surgical factors that could potentially confound associations
(age, gender, and anesthesia type [general compared with regional]) were also
recorded. The length of hospital stay was defined as the time/date of
admission to the time/date of discharge (as determined by the surgeon) either
to home or to a rehabilitation setting. In order to consider all possible
operative characteristics, we also evaluated surgical approach, operative
time, and cemented compared with noncemented fixation in both knees and
hips.
Drain output was recorded at the end of each nursing shift, three times
daily, by measuring the volume of discharge from the closed suction drain
reservoir. In addition, the surgical wound was inspected daily, with notation
made of the day when the wound appeared dry according to the definition of
Weiss and Krackow6.
The wound was considered to be actively draining (i.e., not dry) if a =2
× 2-cm area of gauze covering the wound was wet or if fluid was noted to
be originating from the surgical site. Wounds with simple spotting of the
dressing due to poorly approximated wound edges or due to ulceration were not
considered to be actively
draining6.
Fifteen patients (fifteen hips) treated with total hip arthroplasty and ten
patients (ten knees) treated with total knee arthroplasty in whom an acute
postoperative wound infection subsequently developed were compared with 1196
patients with a total hip arthroplasty and 1216 patients with a total knee
arthroplasty in whom an acute postoperative wound infection did not develop.
In neither the total hip arthroplasty nor the total knee arthroplasty cohort
was there a significant difference between the infected and noninfected groups
in terms of mean age (59.1 compared with 60.1 years, respectively, in the
total hip arthroplasty cohort [p = 0.79] and 63.1 compared with 66.5 years in
the total knee arthroplasty cohort [p = 0.34]) or gender proportion (an
infection developed in 1.2% of the women compared with 0.6% of the men in the
total hip arthroplasty cohort [p = 0.37] and in 1.0% of the women compared
with 0.3% of the men in the total knee arthroplasty cohort [p = 0.47]).
Statistical Analysis
Multiple linear regression analysis was used to model the effects of
prophylaxis against deep venous thrombosis, body mass index, drain output,
surgical time, estimated blood loss, patient age, and type of anesthesia on
the time to a dry wound. The Spearman correlation coefficient was used to
assess the association between prolonged wound drainage and the number of
hospital days. Patients in whom an acute postoperative infection developed
were compared with their uninfected counterparts with use of multiple logistic
regression analysis to calculate the odds ratios between postoperative
infection and the time to a dry wound while controlling for the type of
prophylaxis against deep venous thrombosis, body mass index, surgical time,
estimated blood loss, and drain output. Because of the low number of
infections, our logistic model included the body mass index in lieu of the
body mass categories defined above (normal, overweight, obese, and morbidly
obese), as those categories overstratified the model and thus created unstable
estimates of body mass.
Kaplan-Meier survivorship curves for the combined hip and knee cohorts were
generated for each anticoagulant subgroup with an end point of achievement of
a dry wound. A p value of <0.05 was considered to be significant.
The study group consisted of 1211 total hip arthroplasties and 1226 total
knee arthroplasties. In the total hip arthroplasty group, the only variables
that showed an independent association with prolonged drainage after the other
factors were controlled for were drain output, prophylaxis against deep venous
thrombosis with low-molecular-weight heparin, and a body mass index defined as
obese (borderline significance) or morbidly obese
(Table I).
Increased drain output was an independent risk factor for prolonged wound
drainage in both the total hip arthroplasty group (p < 0.001) and the total
knee arthroplasty group (p = 0.023), and it was the only significant risk
factor in the total knee arthroplasty group
(Table II). The mean
postoperative drain output was 283 mL (range, 0 to 1100 mL) in the total hip
arthroplasty group and 307 mL (range, 0 to 1160 mL) in the total knee
arthroplasty group. In the total hip arthroplasty group, every 100-mL increase
in postoperative drain output resulted in an additional 0.20 day of wound
drainage time. In the total knee arthroplasty group, each 100-mL increase in
postoperative drain output resulted in an additional 0.03 day of wound
drainage.
Association of Prophylaxis Against Deep Venous Thrombosis with
Prolonged Wound Drainage
In the total hip arthroplasty group, prophylaxis against deep venous
thrombosis consisted of Coumadin for eighty-nine patients (7.3%),
low-molecular-weight heparin for 731 patients (60.4%), and aspirin and
mechanical compression for 325 patients (26.8%); the remaining sixty-six
patients (5.5%) were treated with a combination of these methods. In the total
knee arthroplasty group, prophylaxis against deep venous thrombosis consisted
of Coumadin for 131 patients (10.7%), low-molecular-weight heparin for 571
patients (46.6%), and aspirin and mechanical compression for 399 patients
(32.5%); the remaining 125 patients (10.2%) were treated with a combination of
these methods. Prophylaxis with low-molecular-weight heparin was associated
with a significant increase in wound drainage after total hip arthroplasty (p
= 0.027) but not after total knee arthroplasty (Tables
I and
II).
Kaplan-Meier survivorship curves for each anticoagulant group are presented
in Figure 1. The time to a dry
surgical wound was longer for the group treated with low-molecular-weight
heparin than it was for the groups treated with Coumadin or with aspirin and
mechanical compression. At five days postoperatively, the surgical wound was
dry in 87% of the patients receiving Coumadin, in 83% of those treated with
aspirin and mechanical compression, and in 77% of those receiving
low-molecular-weight heparin. A Wilcoxon test showed that the group treated
with low-molecular-weight heparin was moderately, yet significantly,
distinguishable from the aspirin group (p = 0.003) on the fifth postoperative
day, but this difference largely disappeared by the eighth postoperative day.
There was no significant association between the mean postoperative drain
output volume and the type of prophylaxis against deep venous thrombosis.
Association of Body Mass Index with Postoperative Wound Drainage
The mean body mass index in the total hip arthroplasty group was 29.0
kg/m2 (range, 17.4 to 50.0 kg/m2); 22.6% of the patients
were of normal weight, 28.4% were overweight, 23.4% were obese, and 25.7% were
morbidly obese. The mean body mass index in the total knee arthroplasty group
was 31.5 kg/m2 (range, 19.1 to 55.2 kg/m2); 13.4% of the
patients were of normal weight, 24.6% were overweight, 30.3% were obese, and
31.7% were morbidly obese. After we controlled for patient age, anticoagulant
type, and drain output, morbid obesity was an independent risk factor for
prolonged wound drainage following total hip arthroplasty (p = 0.001) but not
following total knee arthroplasty (p = 0.590) (Tables
I and
II). Obesity was also of
borderline significance (p = 0.054) in the total hip arthroplasty group.
Influence of Prolonged Drainage on Length of Hospital Stay
Overall, there was a strong positive correlation between the length of
hospital stay and the number of days until the surgical wound was dry (r =
0.29, p <0.001). Although the correlation was stronger in the total hip
arthroplasty group (r = 0.34, p < 0.001), it was also strong and
significant in the total knee arthroplasty group (r = 0.26, p < 0.001).
Influence of Prolonged Drainage on Wound Infection
Five of the fifteen patients in whom an infection developed following a
total hip arthroplasty were considered to have cellulitis, which resolved with
antibiotics only. The other ten patients had persistent drainage despite the
administration of intravenous antibiotics for at least three days, and they
underwent operative irrigation and débridement. Seven of the ten
patients who had an infection after a total knee arthroplasty were considered
to have cellulitis, which resolved with antibiotics; two patients eventually
required component removal because of persistent infection despite early
operative intervention, and one patient underwent operative irrigation and
débridement because of an infected hematoma but the total joint
components were retained.
A comparison of the patients who had an infection with their uninfected
counterparts with respect to wound drainage, body mass index, drain output,
and prophylaxis against deep venous thrombosis revealed that prolonged wound
drainage was a significant predictor of wound infection after total hip
arthroplasty (odds ratio = 1.42, 95% confidence interval = 1.18 to 1.71, p
< 0.001), and each day of prolonged drainage was associated with a 42%
increase in the risk of wound infection; this result was observed even after
we controlled for body mass index. There was no association between the type
of prophylaxis against deep venous thrombosis, patient age, or drain output
and postoperative wound infection (Table
III). In the total knee arthroplasty cohort, increased body mass
index, but not prolonged wound drainage, was a significant risk factor for
postoperative wound infection (Table
IV); each day of prolonged drainage was associated with a 29%
increase in the risk of postoperative infection, a result that was not
significant after we controlled for body mass index. For every 1
kg/m2 increase in body mass index, there was an 8% increase in the
risk of postoperative infection after total knee arthroplasty (p = 0.018). The
type of prophylaxis against deep venous thrombosis, patient age, and drain
output showed no significant association with infection after we controlled
for wound drainage and body mass index
(Table IV).
Prolonged postoperative wound drainage following total hip arthroplasty has
been shown to be associated with an increase in the risk of
infection1. Because
we could find no reports that assessed the risk factors associated with
prolonged wound drainage, we could not estimate the necessary parameters for
sample-size calculation for our particular study population. Although this is
an observational retrospective study, we believe that it provides the
necessary data for designing clinical trials relevant to the pharmacologic,
surgical, and patient-specific factors that influence the duration of
postoperative wound drainage and the effects of such drainage on wound
infection.
It is our impression that many surgeons believe, on the basis of only
anecdotal evidence, that one or another type of prophylaxis against deep
venous thrombosis results in prolonged wound drainage and more bruising. The
results of the present study show that the method of prophylaxis against deep
venous thrombosis does influence the time to a dry surgical wound. A greater
proportion of patients receiving low-molecular-weight heparin had active
drainage from the surgical wound in the early postoperative period, and the
drainage persisted slightly but significantly longer than that in patients
treated with aspirin and mechanical compression or with Coumadin (p = 0.003).
By the eighth postoperative day, however, this difference had disappeared. One
possible explanation for this observation is that low-molecular-weight heparin
has a faster onset of activity than Coumadin, which may take up to several
days to reach a therapeutic level.
Obesity has been shown to increase the risk of postoperative
infection7. We found
obesity to be an independent risk factor for prolonged postoperative wound
drainage following total hip arthroplasty regardless of the type of
prophylaxis against deep venous thrombosis. This finding is not surprising
since one expects to use a larger surgical incision that consequently results
in more fat necrosis in obese patients. Also, the volume of postoperative
drain output was shown to be an independent risk factor for prolonging the
time to a dry wound in both the total hip arthroplasty and the total knee
arthroplasty group. The clinical relevance of this finding is unclear,
however, since studies have failed to show a difference in the rate of wound
infection, wound hematoma, or reoperations for wound complications between
wounds treated with and those treated without a closed suction
drain8.
A prolonged time until a surgical wound is dry is clinically relevant. Our
results demonstrated that patients with prolonged wound drainage after total
hip or total knee arthroplasty had a significantly longer hospital stay. The
risk of postoperative wound infection was significantly higher for patients
who had prolonged wound drainage after total hip arthroplasty, and this risk
was independent of the type of prophylaxis against deep venous thrombosis. The
same was not true for the total knee arthroplasty group, after we controlled
for body mass index, which may reflect the preponderance of obesity in this
group compared with the total hip arthroplasty group. Additional study is
required to clarify the different effects of drainage time on wound infection.
?