Rheumatic Diseases – What Influence do they
Really Have on the Incidence of Periprothethic
Joint Infections after Total Knee Arthroplasty?
Stella Oberberg*, Leonard Ulbrich, Jan Nottenkamper, Jan Krapp and Roland E Willburger
Department of Orthopaedic Surgery, Katholisches Klinikum Bochum, Martin-Luther-Hospital, Germany
Submission:April 14, 2021; Published: April 23, 2021
*Corresponding author: Stella Oberberg, Department of Orthopaedic Surgery, Katholisches Klinikum Bochum, Martin-Luther-Hospital, Voedestrasse 79, 44866 Bochum, Germany
How to cite this article: Stella O, Leonard U, Jan N, Jan K, Roland E W. Rheumatic Diseases – What Influence do they Really Have on the Incidence of
Periprothethic Joint Infections after Total Knee Arthroplasty?. Ortho & Rheum Open Access J. 2021; 18(2): 555981. DOI: 10.19080/OROAJ.2021.18.555981
Introduction: Periprosthetic joint infections (PJI) are serious complications after total knee arthroplasty (TKA). According to literature predisposing diseases, operation-specific and postoperative factors are important in development of PJIs. There are still divergent positions in the literature regarding the significance of individual risk factors. The aim of this study was to determine the incidence of PJI in our institution from 2008 to 2018, to identify risk factors by comparing affected patients with a normal population.
Method: Fifteen PJIs were detected during research period. Data of 501 consecutive patients were collected retrospectively for comparison. A matched-pair analysis was performed in addition to an analysis of categorical and metric characteristics of the total population (n=516) to adjust the unequal strengths of both groups.
Results: The incidence of PJI was 0.8% in our institution. Analysis of all patients showed significant correlations of PJI regarding blood transfusion, hematoma formation and postoperative urinary tract infection. The presence of renal insufficiency, nicotine consumption, and prolonged duration of surgery were identified as risk factors. After matched-pairs analysis (MPA), prolonged surgical duration, blood transfusion and preoperatively decreased haemoglobin (HB) were confirmed as independent risk factors. There was no significant difference regarding the presence of rheumatic diseases.
Discussion: Determining risk factors were preoperative anaemia, blood transfusion and prolonged operation times. The presence of rheumatic disease did not appear to be a risk factor. These findings should be incorporated into surgical preparation. For infection prophylaxis, preoperative HB elevation and rapid surgical procedures with low intraoperative blood loss should be discussed.
Keywords: Periprosthetic infections; Total knee arthroplasty; Prophylaxis; Risk factors; Anaemia, Duration of surgery; Blood transfusion; Rheumatoid arthritis
Abbreviations: ASA: American Society of Anaesthesiologists; HB: Haemoglobin; BMI: Body-Mass-Index; DMARD: Disease-modifying Anti-Rheumatic Drug; IP: Infection-Population; MPA: Matched-Pairs Analysis; NP: Normal-Population; PJI: Periprosthetic Joint Infections; TKA: Total knee Arthroplasty
Implantation of a total knee arthroplasty (TKA) is one of the most common operations in orthopaedics. The number of TKA is steadily increasing due to demographic changes in society . The majority of patients (over 85 percent) present good to very good results after TKA . However, a significant decrease in revision surgeries has not yet been observed despite improved surgical techniques and innovative implants with long prosthesis service lives. Schwartz et al.  even predict an increasing number of complications with an increased need for second operations of 170 percent by the year 2030. Early complications include
periprosthetic infections in addition to aseptic loosening and dislocation and periprosthetic fractures . The rates of infection after primary TKA are inconsistent in the literature. Despite increasing prevention strategies, the incidence of infection is reported to be 1-2 percent in recent publications [5,6].
Various patient-dependent and independent risk factors are discussed in the literature. Patient-dependent factors include pre-existing conditions such as diabetes mellitus, obesity, rheumatoid arthritis, and nicotine and alcohol abuse. Patients with diabetes mellitus show worse outcomes after surgical intervention with prolonged inpatient stays and increased mortality .
Existing obesity is not only a risk factor for the development
of osteoarthritis, but also a risk factor for postoperative
complications. Thus, obese patients tend to have poorer wound
healing, wound dehiscence, prolonged secretion, and increased
hematoma formation . According to literature, regular nicotine
use causes peripheral vasoconstriction and associated tissue
hypoxia due to activation of the sympathetic nervous system .
This results, among other things, in slower wound healing with
an increased incidence of periprosthetic complications . In
their retrospective work, Crowe et al. identified nicotine abuse
within one month before surgery as an independent risk factor
for periprosthetic infections after primary knee arthroplasty .
Some authors also describe an association between the presence
of renal insufficiency and an increased risk of PJI [12, 13]. Regular
consumption of alcohol is also associated with an increase in
complications after prosthetic implantation [14,15]. Increased
complication rates and infection rates have also been repeatedly
described in patients with rheumatoid arthritis . It is assumed
that, in addition to the influence of immunomodulatory therapy,
T-cell dysfunction is partly responsible .
The aim of the present study was to determine the frequency
of periprosthetic infections in our institution over a period of 10
years. The focus was on detecting risk factors for periprosthetic
infection after TKA. The supposed risk factors were compared with
the results of a control group without periprosthetic infection.
Supposed risk factors were: Elevated Score of the Physical
Status System “American Society of Anaesthesiologists” (ASA),
diabetes mellitus, obesity, history of malignancy, preoperative
anaemia, and comorbidities such as rheumatoid arthritis or renal
insufficiency. In addition, the influence of surgical time, surgeon
experience, blood loss during surgery, anaesthetic procedure,
and administration of foreign blood were observed. Other
abnormalities such as prolonged wound secretion, electrolyte
imbalances, and postoperative urinary tract infections were also
considered in the study.
The data collection for this study was performed retrospectively
by evaluating patient data from our department. For the research,
the respective patient files as well as digital documents in the
information management system „Orbis“ were accessed. All
existing documents such as physician‘s reports, admission notes,
operation reports, anaesthesia protocols, laboratory findings,
microbiological findings, and medical course documentation were
considered. The period under investigation was 2008 to 2018. In
the years mentioned, a total of 15 patients with periprosthetic
infection after implantation of primary knee arthroplasty were
registered in our department. These patients constitute the
„infection-population“ (IP) in the present study. The date of the
primary implantation of the first included patient of this group was
in May 2008 and that of the last in April 2018. The control-group
consists of 501 patients who also received a primary TKA in the
same department and did not develop a periprosthetic infection
(“normal-population”; NP). The surgeries of this population were
performed between January 2016 and December 2018, with a
consecutive data collection backwards from 2018.
The data collection was performed with “Microsoft Excel”,
“Stata/IC 16.1 for Unix” was used for statistical analysis.
For descriptive analysis, patient data were categorized into
categorical (such as prior surgery) and metric characteristics
(such as patient age). Fisher‘s exact test was used to test whether
group membership and the corresponding characteristic were
independent. Whether the groups differed with respect to the
distribution of a characteristic was tested with the Mann Whitney
U test. To eliminate the influence of the large difference in the
number of cases in the two groups (infection group and normal
group) and to detect the significance of the results, an additional
matched-pairs analysis (MPA) was performed. Matching criteria
were gender, presence of diabetes, body-mass-index (BMI)
(<40/≥40), ASA (≤2/>2), and age (± 2 years). The comparison
of the groups was performed by McNemar‘s exact test for the
categorical characteristics and the Wilcoxon signed-rank test
for connected samples. All statistical tests were performed at a
significance level of 0.05.
For the evaluation, general data such as patient name, age, and
gender as well as previous operations on the affected knee joint
were recorded. Furthermore, patient-related data were collected.
This included height, weight, body mass index, and alcohol
and nicotine abuse. Comorbidities such as malignant diseases,
relevant cardiovascular diseases, diabetes mellitus, diseases of the
respiratory tract, liver and kidney were recorded. Special attention
was paid to rheumatic diseases or other inflammatory systemic
diseases and associated medication with immunomodulatory
drugs or anticoagulants. The surgery-dependent-factors included
the indication-related diagnosis, whereby a differentiation
was made between primary and secondary arthrosis and
inflammatory diseases. In addition, the duration of the operation,
the side location, and whether the operation was performed by
an experienced surgeon or a surgeon in training were recorded.
In addition, the type of anaesthesia and antibiotic prophylaxis as
well as the ASA classification were documented.
The NP included more female than male patients (69.7%
versus 30.3%). In contrast, an almost equal gender distribution
could be observed in the IP. This difference between the groups
turned out to be statistically insignificant.
Regarding to pre-existing diseases or conditions, only the
presence of renal insufficiency (p=0.021) and nicotine abuse
(p=0.047) showed significant differences in the group comparison.
Renal insufficiency showed the largest percentage difference of all
surveyed pre-existing conditions. In the infection group, 27 percent
(4 of 15) reported having renal insufficiency. In comparison, only
7 percent (35 of 501) of patients in the normal group reported
renal insufficiency. Patients were considered smokers if they
were current smokers or reported having smoked regularly
for at least one year in the past. The type of rheumatic disease
included in the study was rheumatoid arthritis, spondylarthritis,
and polymyalgia rheumatica. In the infectious group, one of five
showed a rheumatoid disease, in the comparison group only one
in ten. However, the difference between the two groups regarding
the presence of a rheumatic disease was not found to be significant
(p=0.170). There were also no significant differences between the
two groups regarding previous operations on the affected knee
joint (p=1.000) and the presence of diabetes mellitus (p=0.158)
as well as other previous diseases (cardiovascular, respiratory
diseases, etc.) (Table 1).
In the MPA, no significant difference could be found for any
of the above-mentioned conditions (renal disease p=0.083;
rheumatic disease p=0.655; nicotine consumption p=0.083),
so that the preoperative patient findings in our collective are
considered negligible regarding the incidence of periprosthetic
infection (Figure 1).
A statistically highly significant difference (p< 0.001) was
provided by the duration of surgery, which was significantly
longer in the infection-population (mean 91.9 minutes) than in
the normal-population (mean 70.3 minutes). The IP showed a
minimum of 68 and a maximum of 114 minutes. In the NP, the
minimum surgery duration was 43 minutes, while the maximum
surgery duration was 115 minutes. The examination of the
surgery duration also showed the highest significance of all
factors examined in the MPA (p = 0.037) (Figure 2).
The groups differed most regarding the postoperative course.
There was a highly significant correlation of periprosthetic
infections with receipt of blood transfusions (p<0.001),
hematoma formation (p=0.001), and proven urinary tract
infection (p=0.008). In the IP, 40 percent of patients received a
blood transfusion. Comparatively, in the NP, transfusions occurred
at a strikingly low rate of 3.6 percent. Thus, in the infectionpopulation,
the rate of transfusions was three times higher than
in the normal-population (40% versus 3.6%). In the MPA, the
transfusion of blood also showed a significant difference between
the groups (p=0.046). Regarding to this, the preoperative HB
value also showed a significant difference between the two groups
in the MPA (p=0.049). The formation of hematomas of significant
size and other wound healing disorders was documented in seven
cases. Three of these cases were in the IP and four cases in the NP (20% of infections versus 0.8%). However, MPA revealed
no significant difference between the collectives in this regard
(p=0.83). Approximately one-quarter (26.7%) of patients in
the infection-group developed a urinary tract infection in the
postoperative setting. In the normal-collective, it affected only
about one in twenty (5.2%). However, the defined statistical
significance level could not be reached in the MPA in this case
either (Figure 3).
The presence of rheumatologic disease has been described by
some authors as a possible risk factor for periprosthetic infection
[12,16,18], with continued “disease-modifying anti-rheumatic
drug” (DMARD) therapy in particular being discussed as a crucial
risk factor . In addition, there is competing literature that
does not assign an increased risk for PJI to rheumatoid arthritis
patients without immunosuppressive therapy [5,19]. In our
population, we did not find a statistically significant association
between the presence of rheumatic disease and an accumulation
of PJI. In relative terms, the infection-population included
two times more patients with rheumatic disease compared
with the normal-population, but without reaching statistical
significance. Regarding therapy with DMARD and biologics, the
recommendations for perioperative procedures of the „German
Society of Rheumatology“ were strictly applied in our population
. This might have been one reason why no significant
association could be found. All other preoperative conditions such
as nicotine abuse or the presence of renal insufficiency also failed
statistical significance in the MPA, so that we could not confirm
these conditions as risk factors in our population. On the other
hand, a prolonged operation time turned out to be an independent
risk factor with sufficient significance in our collective in the MPA.
Wang et al.  were also able to identify prolonged surgery time
as a risk factor in their study, showing that a 20-minute increase
in surgery time resulted in a 25% increase in the probability of
In our study, the administration of foreign blood also resulted
in significant differences between the two groups in both the
overall analysis and after pairing. In the literature, studies have
also demonstrated an increased incidence of infections after
blood transfusion [22,23]. Pulido et al.  calculated a 2.1-
fold increased risk for PJI following the transfusion of blood.
In summary, it can be stated for our collective that both a low
preoperative HB value as described by Müller et al.  and
the necessity of blood transfusion represent an independent
risk factor for a PJI, so that the securing of a sufficiently high
preoperative HB value should be given special importance in
the surgical preparations. Furthermore, a high intraoperative
blood loss can be avoided by a rapid and blood-saving surgical
procedure, if necessary, with the use of antifibrinolytic drugs.
While Pulido et al.  identified urinary tract infection as an
independent risk factor for PJIs, Koulouvaris et al. did not show an
increased rate of deep wound infections associated with urinary
tract infections . In our collective, postoperative urinary tract
infection increased the risk of PJI significantly. However, after
matching, statistical significance could not be confirmed, so we
cannot consider the presence of a urinary tract infection as an
independent risk factor in our study. According to the literature,
the formation of a hematoma is also associated with an increased
rate of wound infections , although sufficient significance
was not achieved here in some studies [6,27]. In our population,
this observation can be confirmed, as also here, after attenuating
the confounding effects, the significance value decreased below
the defined significance level. The main limiting factor of our
study might be the retrospective study design, so that a causal
association between the risk factors and the infection rate cannot
be proven. Some confounders could be attenuated in the study
using a matched-pair analysis; nevertheless, other confounders
may have influenced the results.
In our sample, the most important risk factors for the incidence
of periprosthetic infection were preoperative anaemia with
HB values < 11.5 mg/dl, the application of blood transfusions,
and prolonged surgery times. These are all factors that can be
considered during surgical preparation and argue for a possible
need to raise the preoperative HB level. Furthermore, surgery
should be as short as possible and thus probably less bloody.
In our collective, there was no evidence for an increased risk in
the presence of rheumatic disease under strict adherence to the
recommendations of the German Society of Rheumatology on the
perioperative management of DMARD and biologicals therapy in
inflammatory rheumatic diseases.
Bozic KJ, Lau E, Kurtz S, Ong K, Berry DJ (2012) Patient-related risk factors for postoperative mortality and periprosthetic joint infection in medicare patients undergoing TKA. Clin Orthop Relat Res 470: 130-137.
Bongartz T, Halligan CS, Osmon DR, Reinalda MS, Bamlet WR, et al. (2008) Incidence and risk factors of prosthetic joint infection after total hip or knee replacement in patients with rheumatoid arthritis. Arthritis Rheum 59: 1713-1720.