Prolonged Catheter Use and Infection in Regional Anesthesia: A Retrospective Registry Analysis

CONTINUOUS regional anesthesia improves perioperative analgesia^1,2  and may reduce morbidity and mortality.^3–5  However, long-term catheter use increases the risk of catheter-related infections,^6–9  which are painful, increase morbidity, and prolong hospitalization.^10–12  Depending on the catheter insertion site, the incidence of infection reportedly ranges from 0 to 7% for peripheral catheters.^9,13–15  For epidural catheters, reported risk ranges from 0.8 to 4.9%.^6,11,16,17 

The extent to which the risk of catheter-related infection increases with catheter duration remains unclear—in part because previous influential studies do not clearly define prolonged catheter use,^6–9  and perhaps by the fact that the duration of peripheral nerve and epidural catheter use differs considerably by country: in the United States, typical maximum catheter duration is reported to be 1 to 4 days^18,19 ; in Switzerland, 1.5 to 5 days^14,20 ; in Australia, 1 to 13 days⁶ ; and in Germany, 1 to 36 days.^8,9,16,17  Which approach is preferable remains unknown. We therefore evaluated the extent to which peripheral nerve and epidural catheter–related infections increase over time in adults using a prospective voluntary national multicenter registry in Germany. We hypothesized that each additional day of catheter use is associated with an increased risk of catheter-related infection.

Approval for this study was provided by the Ethics Committee of the Saarland Medical Chamber, Saarbrücken, Germany (Chairperson, Sanitätsrat Prof. Dr. Hermann Schieffer) on March 22, 2011, with the identification No. Ha50/11. Written consent was waived as the data were anonymous (regulatory proof of protection of data privacy, Saarland commissioner, March 12, 2014).

In 2007, the German Society for Anesthesiology and Intensive Care Medicine and the Professional Association of German Anesthesiologists (Nuremberg, Germany) established a network for safety in regional anesthesia. The German Network for Regional Anesthesia database collects preoperative, intraoperative, and postoperative data from treating physicians at 25 German centers who completed a standard form (appendix 1).²¹  Data from patients having regional anesthesia included detailed information about their medical conditions along with the procedure and postoperative course. These data were collected by pain nurses or treating physicians concurrently with patient care. The data were entered contemporaneous with standard documentation and were collected electronically or on paper.

This registry includes 114,543 cases acquired between September 2007 and December 2014. The study protocol is reported in appendix 2. Data integrity was evaluated according to specific rules to delete erroneously entered data as well as cases with missing information (proof of plausibility, appendix 2). The body mass index (BMI) was calculated as weight in kg/(size in m)² and defined from 17 to 70 kg/m². All participating centers were aware of the German guidelines to reduce catheter-related infection.²²  These include hand cleaning and disinfection, use of surgical mask, sterile gloves and gown, cap covering hair, shaving the insertion site, skin disinfection, aseptic sheeting, aseptic drugs, and sterile bandaging. The definition of multiple skin puncture was more than one skin puncture during a particular block procedure.

We included patients 0 to 100 yr old who had peripheral or epidural catheters inserted for surgical procedures, information about catheter in situ time, and information about catheter-related infection. Patients were excluded from our analysis when catheters were in place for more than 15 days—a rare event. Catheters used for obstetric analgesia were also excluded because they are rarely used for more than 48 h (fig. 1).²³ 

Among the prospectively recorded details was the catheter duration. Signs of infections were reported by pain nurses or physicians during postoperative ward rounds. Infections at the catheter insertion site were prospectively defined as previously described^24,25 : (1) mild infections were defined by at least two of three infection signs (redness, swelling, or local pain); (2) moderate infections were defined as mild in addition to at least one of the following findings: increased C-reactive protein, leucocytosis, fever, or pus at the punctured site; and (3) severe infections were defined by the need for a surgical incision or revision. Infection status was evaluated at least daily during surgical ward rounds. Data collection ended when catheters were removed.

The primary endpoint was a composite of the presence of a mild, moderate, or severe catheter-related infection up to 15 days. The secondary endpoint was progression of low-grade (mild/moderate) infection of catheters left in situ to higher-grade (moderate/severe) infections.

Each patient with prolonged catheter use was included only with the first observed infection. Population characteristics are reported as absolute standardized differences (absolute value of means [infection-free catheter use minus catheter-related infection] divided by the pooled SD).

A Kaplan–Meier survival curve was plotted to examine the relationship between catheter duration and probability of infection-free catheter use. Cox regression analysis was performed, and an adjusted survival curve was plotted. Cox regression survival curves were estimated using the default setting of SPSS Statistics 19 (IBM, USA): a patient with the mean of all covariates. This analysis was used in the final study population: patients with a complete set of covariates, which are specified in tables 1 and 2. Potential confounders were sex, age, BMI, American Society of Anesthesiologists physical status, diabetes, multiple skin puncture, surgical specialty, catheter site, year of surgery, and hospital. Age and BMI were included as continuous variables; all other covariates were included as categorical variables. Variables with a positive or negative correlation greater than 0.3 and less than or equal to –0.3 were evaluated for interactions. The assumption of proportional hazard was checked for all included variables. An omnibus test was performed to calculate P value from likelihood ratio test statistics. Additionally, the –2 log likelihood of final covariate model is given.

As a sensitivity analysis, we also estimated a Kaplan–Meier survival curve in a larger study population that contained details about catheter duration and infection, but incomplete covariables (fig. 1).

Data analysis was performed using IBM SPSS Statistics for Windows, version 19 (IBM). Continuous variables are expressed as means and SDs. Categorical variables are presented as absolute and relative frequencies, respectively. Two-sided P values 0.05 or less were considered statistically significant.

There were 65,291 patients with continuous nerve blocks and information about catheter duration and infection (sensitivity analysis). A total of 44,555 of these patients had complete covariable data (fig. 1), which are specified in tables 1 and 2 (final study population). Our analysis is based on a total of 693 peripheral nerve catheter infections and 804 epidural catheter infections.

Among 24,103 patients with peripheral catheters, 941 were less than 18 yr old (6 cases with infection), and 153 were less than 12 yr old (1 case with infection). Among 20,452 patients with epidural catheters, 387 were less than 18 yr old (16 cases with infection) and 90 were less than 12 yr old (5 cases with infection).

Characteristics of the patients with complete data about covariables are presented in table 1. Patients without infection were younger, more likely to be female, and less likely to have comorbidities. Patients without infection were also more likely to have surgery in traumatology and orthopedics and more likely to receive psoas blocks. Clear differences with absolute standardized differences (> 0.2) between the groups were found for BMI (absolute standardized differences, 0.33), diabetes (absolute standardized differences, 0.22), surgical department (absolute standardized differences, 0.68), femoral site (absolute standardized differences, 0.27), and psoas site (absolute standardized differences, 0.46).

Before data exclusion, the incidence of infection in a larger patient cohort with incomplete information about catheter duration was 3.0% in 36,300 peripheral catheters. In the final study population, the incidence of peripheral catheter–related infections was 2.9% in 24,103 patients. The infection grade was mild in 593 cases (2.5%), moderate in 83 cases (0.3%), and severe in 17 cases (0.07%). The probability of peripheral infection–free catheter use in patients with complete covariables was 98% at day 4 of catheter duration, 91% at day 7, and 57% at day 15 (fig. 2). With adjusted Cox regression analysis, the probability of peripheral infection–free catheter use was 99% at day 4 of catheter duration, 96% at day 7, and 73% at day 15 (fig. 3). Detailed information about the model is reported in Supplemental Digital Content 1A (https://links.lww.com/ALN/B615). The Kaplan–Meier survival curve calculated from a larger population lacking covariable details was generally similar (fig. 2, sensitivity analysis). In this larger population of 32,172 patients, 975 catheters were infected (3.0%).

Characteristics of the patients with complete information about covariables are presented in table 2. Patients without infection were more likely to be female and less likely to have comorbidities. Patients without infection were also more likely to have lumbar epidural catheters and more likely to have surgery in traumatology and orthopedics. They were also less likely to require multiple skin puncture. Clear differences with absolute standardized differences > 0.2 between the groups were found for traumatology and orthopedics (absolute standardized differences, 0.44), other surgical department (absolute standardized differences, 0.51), and thoracic epidural site (absolute standardized differences, 0.44).

Before data exclusion, the incidence of infection in a larger patient cohort with incomplete information about catheter duration was 3.9% in 43,568 epidural catheters. In the final study population, the incidence of epidural catheter–related infections was 3.9% in 20,452 patients. The grade of infection was mild in 676 cases (3.3%), moderate in 114 cases (0.6%), and severe in 14 cases (0.07%). The probability of epidural infection–free catheter use in patients with complete covariables was 99% at day 4 of catheter duration, 93% at day 7, and 65% at day 15 (fig. 2). With adjusted Cox regression analysis, the probability of epidural infection–free catheter use was 99% at day 4 of catheter duration, 95% at day 7, and 73% at day 15 (fig. 3). Detailed information about the model is reported in Supplemental Digital Content 1B (https://links.lww.com/ALN/B615). The Kaplan–Meier survival curve calculated from a larger population lacking covariable details was generally similar (fig. 2, sensitivity analysis). In this larger population of 33,119 patients, 1,441 catheters were infected (4.4%).

In our final study population of 44,555 patients, 1,497 infected catheters were identified, of which 36 were left in situ (fig. 4). All 36 catheters progressed to higher infection grades. Twelve were peripheral and 19 were epidural; they initially showed mild signs of infection after 5.2 ± 2.4 (range, 1 to 10) days and progressed to moderate infection after an additional 1.7 ± 1.8 (0.5 to 8) days. The remaining five cases (three lumbar and two thoracic epidural catheters) initially showed mild or moderate signs of infection after 5.2 ± 2.8 (3 to 10) days and progressed to severe infection after an additional 1.6 ± 0.6 (1 to 2) days that prompted surgical intervention.

Each additional day of catheter use, starting on the fourth day after insertion, was strongly associated with an increased risk of catheter-related infection for both peripheral and epidural catheters. Previous studies also identify prolonged catheter use as a risk factor for infection,^6–9  but our multicenter results enhance current understanding by specifically evaluating infection risk as a function of catheter duration. The fact that infection risk increases over time is consistent with experience with central venous catheters.^26,27 

The overall incidence of peripheral catheter–related infections was 2.9% in our study, which is higher than previously reported.^9,13–15  The 3.9% incidence of nonobstetrical epidural catheter–related infections was higher than in previous reports.⁶  However, also, the probability of infection-free catheter use was 99% at day 4 for both peripheral and epidural catheters. It is likely that our incidence was higher because our maximum catheter duration was 15 days, which is longer than in previous studies.^6,9,13–15  In previous studies, an increased incidence of catheter-related infection in trauma patients was observed.^7,9,15  In these studies, trauma patients had a prolonged intensive care unit stay, which was identified as an independent risk factor for catheter-related infections. In contrast, our trauma/orthopedic patients mainly had elective surgery without intensive care unit stays. There were also differences in definitions of infection and inflammation, patient population, preventive hygiene measures, and probably many unknown factors. A strength of our study is a clear a priori definition of the criteria for infection, which was lacking in some previous investigations.

In most cases, only mild signs of infection (redness, swelling, or local pain) were observed—presumably because catheter insertion sites were inspected daily and catheters were usually removed when signs of infection were first detected. Infected catheters that were left in situ progressed to higher infection grades. Our results thus suggest that catheter insertion sites should be inspected daily because the time interval between the onset of symptoms and infection progression is usually less than 48 h. Among 36 patients with initial signs of mild or moderate infection in whom the catheter was left in situ, five developed severe infection requiring surgical exploration. Interestingly, in our final study population of 44,555 patients, we observed only 31 cases (0.07%) with severe infection and prompt surgical exploration. Catheters were removed with the first observed sign of local infection in 26 of these 31 cases. However, progression could presumably have been avoided in five cases in which the catheter was left in situ with initial mild or moderate signs of infection. This observation is consistent with previous studies about infection of body-foreign material in situ including cardiovascular implantable electronic devices.²⁸  We therefore recommend removing infected catheters immediately.

Several different risk factors for catheter-related infection have been identified: American Society of Anesthesiologists physical status, diabetes, type of surgery, catheter site, multiple skin puncture, and BMI.^{6,7,9,11,12,15,24,25,29–31}  All were included as confounders in our Cox regression analysis.

Our analysis was limited to the risk of catheter-related infection for peripheral and epidural catheters. We do not have sufficient information about the method of treatment, longer-term recovery, duration of hospitalization, or mortality. Consequently, we cannot determine whether the observed infections were linked to more serious outcomes. Moreover, some confounders with potential influence on catheter-related infection are missing in our analysis, including severity of diabetes, stage of cancer, grade of renal failure, redo surgery, and amount of steroid use or other immunosuppressive medication.

Our clinically routine documentation was electronically transferred into the registry. Since the registry design was pragmatic, the level of documentation varies from center to center. Many cases were thus excluded because of missing information about duration of the catheter and infection. The high number of excluded patients increases the risk of bias in our analysis. Nevertheless, we included 65,291 cases in our primary cohort, all with continuous nerve blocks and information about catheter duration and infection. This population was reduced to 44,555 cases, all with complete covariables. However, univariable results from all relevant patients (n = 65,291) were generally similar to both univariable and adjusted results using a multiple regression approach in patients with all available information (n = 44,555), suggesting data were missing completely at random and did not introduce substantial bias. As in any nonrandomized analysis, residual confounding may invoke error which cannot be eliminated in the framework of our sensitivity analysis.

During the seven-year observation period, there were presumably improvements in knowledge, skills, techniques, and disinfectant methods. However, our results were adjusted for the year of surgery. There was heterogeneity in the incidence of infection among the hospitals in our analysis, and this was added as confounder in a multiple model. Registries critically depend on the quality of data entry and handling; the validity of registry analyses thus always depends on the quality of the underlying data. Although our analysis was retrospective, infection data in our registry were specifically collected concurrent with patient care using an a priori definition.

In summary, the risk of peripheral and epidural catheter–related infection substantially increases over time. When catheters develop signs of infection, attention is needed to avoid progress of infection.

German Network for Regional Anesthesia investigators are all participating hospital centers that collected data about catheter duration and grade of infection for the current study: Berlin—Charité Campus Virchow Klinikum and Campus Mitte, Charité University Medicine Berlin, Germany, Claudia Spies, M.D.; Berlin—German Red Cross Hospital Westend, Berlin, Germany, Arnd Timmermann, M.D.; Bochum—Ruhr-University Bochum, University Hospital Knappschaftskrankenhaus Bochum, Germany, Michael Adamzik, M.D.; Bad Saarow—HELIOS Hospital Bad Saarow, Germany, Stefan Wirtz, M.D.; Dresden—University Hospital Carl Gustav Carus, Technische Universität Dresden, Germany, Thea Koch, M.D.; Erfurt—HELIOS Hospital Erfurt, Germany, Andreas Meier-Hellmann, M.D., and Simone Liebl-Biereige, M.D.; Frankfurt—Orthopedic University Hospital, Frankfurt, Germany, Paul Kessler, M.D.; Frankfurt—Berufsgenossenschaftliche Casualty Hospital, Frankfurt, Germany, Rolf Teßmann, M.D.; Hamburg—University Medical Centre Hamburg-Eppendorf, Germany, Alwin E. Goetz, M.D.; Hannover—Friederikenstift Hannover, Germany, André Gottschalk, M.D.; Jena—University Hospital, Jena, Germany, Winfried Meissner, M.D.; Lengerich—HELIOS Hospital Lengerich, Germany, Albrecht Pfeiff, M.D.; Ludwigsburg—Hospital Ludwigsburg, Germany, Götz Geldner, M.D.; Marburg—Philipps University Marburg, Germany, Hinnerk F. W. Wulf, M.D.; Memmingen—Hospital, Memmingen, Germany, Lars Fischer, M.D.; Siegen—St. Marien-Hospital, Siegen, Germany, Werner Hering, M.D.; Solingen—Academic Hospital Solingen, Germany, Thomas Standl, M.D.

This analysis was supported by institutional funds. The German Network for Regional Anesthesia was supported by the German Society of Anesthesiology and Intensive Care Medicine (Nuremberg, Germany), the Association of German Anesthesiologists (Nuremberg, Germany), and the German Research Foundation (Bonn, Germany).

The authors declare no competing interests.

Clinical department

Ward

American Society of Anesthesiologists physical status

Sex

Age

Weight

Height

Operation and procedures code (German modification of the International Classification of Procedures in Medicine)

Type of surgery

Chronic pain patient

Opioid use longer than 1 month

Preoperative pain level at rest and on exertion

Renal function

Diabetes

Peripheral arterial occlusive disease

Rheumatoid arthritis

Alcohol abuse

Drug abuse

Liver insufficiency (greater than Child–Pugh score B)^32–34 

Immune deficiency

Steroid use

Other immunosuppressive drugs

Transplanted organs

Sepsis or presence of systemic infection

Antibiotic therapy or prophylaxis

Anticoagulant drugs according to guideline

Patient status awake, sedated (opioid, benzodiazepine, other), general anesthesia, block in an anesthetized area

Multiple skin penetration

Sterile standards (gown, tunneling, filter, suture)

Bloody tap, unintended dura puncture, pneumothorax, intoxication

Block abandoned for anatomical reasons or patient’s wish

Use of sonography (quality of visualization and of local anesthetic distribution)

Needle type

Drugs given

Unplanned additional analgesia necessary (including general anesthesia if block fails)

Loss of resistance with air, sodium chloride solution, or hanging drop

Unintended paresthesia

Patient identification, date, time, duration

Regular or irregular catheter removal

Anticoagulation status at the time of removal according to guidelines

Mobilization scale, sedation scale

Satisfaction with pain therapy

Presence of transient neurologic symptoms, headache after dural puncture hematoma, neuropathic pain, blood patch

Pain levels (Numeric Pain Rating Scale [NRS]*) at rest and on exertion within the expected area of effective regional analgesia

Pain level of whole body (NRS)

Muscle strength (six-point system according to Janda with 0 = -inability of a contraction, 1 = detectable contraction, 2 = movements possible but not against gravity, 3 = against gravity, 4 = against moderate resistance, 5 = full muscle strength)

Hypoesthesia, paresthesia

Pain interference with mobilization, respiration, sleep

Coanalgesics

Treatment necessity for urinary retention, respiratory depression, nausea, vomiting, pruritus

Catheter-associated hypotension

Catheter manipulations, site-specific alterations like occlusions, leaks, technical problems

Grade of catheter-related infection

Infusion rate, boli, lockout times

*NRS is a numeric version of the visual analog scale in which a respondent selects a whole number (0 to 10). The scale ranges from 0, “no pain,” to 10, “worst pain.”^35,36 

We hypothesized that each additional day of catheter use is associated with an increased risk of catheter-related infection.

The primary outcome was a composite of the presence of a mild, moderate, or severe catheter-related infection up to 15 days. The secondary outcome was progression of low-grade (mild/moderate) infection of catheters left in situ to higher-grade (moderate/severe) infections.

Inclusion criteria for the final study population were patients 0 to 100 yr old who had peripheral or epidural catheters inserted for surgical procedures, information about catheter in situ time, information about catheter-related infection, sex, BMI, American Society of Anesthesiologists physical status, diabetes, surgical specialty, multiple skin puncture, year of surgery, and hospital center. Exclusion criteria were defined as catheter duration for more than 15 days, catheter used for obstetric analgesia, and implausible data.

Table A2.1

A sensitivity analysis was planned in a larger cohort of patients with continuous nerve block, information about catheter duration, catheter-related infection, and incomplete covariables (age, sex, BMI, American Society of Anesthesiologists physical status, diabetes, surgical specialty, multiple skin puncture, year of surgery, and hospital center).

Post hoc sensitivity analysis was performed in a larger patient cohort with information about catheter-related infection and catheter site only.

Absolute standardized differences to describe population characteristics between the groups were defined a priori. Kaplan–Meier survival curves and Cox regression analyses were planned post hoc.