Fewer Postoperative Fevers: An Unexpected Benefit of Multimodal Pain Management?

Abstract

Background

Elevated temperatures after total joint arthroplasty (TJA) are common and can be a source of anxiety both for the patient and the surgical team. Although such fevers rarely are caused by acute infection, many patients are subjected to extensive testing for elevated body temperature after surgery. We recently implemented a multimodal pain management regimen for TJA, which includes acetaminophen, pregabalin, and celecoxib or toradol, and because some of these medications have antipyrexic properties, it was speculated that this protocol might influence the frequency of postoperative pyrexia.

Questions/purposes

The purpose of this study was to determine whether patients treated under this protocol were less likely to exhibit postoperative fever after primary TJA, compared with a historical control group, and whether they were less likely to receive postoperative testing as part of a fever workup.

Methods

We compared 1484 primary TJAs in which pain was controlled primarily with opioid-based relief from July 2004 to December 2006 with 2417 procedures from July 2009 to December 2011 during which time multimodal agents were used. The same three surgeons were responsible for care in both of these cohorts. Oral temperature readings in the first 5 postoperative days (POD) were drawn from a review of medical records, which also were evaluated for fever workup tests, including urinalysis, urine culture, chest radiograph, and blood culture. Fever was defined by the presence of a temperature measurement over 38.5 °C. Patients having preoperative fever or postoperative fever starting later than POD 5 were excluded. Before surgery, there were no differences between the groups’ temperature measurements.

Results

Fewer patients developed fever in the multimodal analgesia group than in the control group (5% versus 25%, p < 0.001). Furthermore, fewer patients underwent workup for fever in the multimodal analgesia cohort (1.8% of patients undergoing 155 individual tests) compared with the control cohort (9.8% of patients undergoing 247 individual tests; p < 0.001).

Conclusions

In addition to fewer adverse effects and better pain control, the multimodal analgesia protocol has the hidden benefit of dampening the temperature response to the surgical insult of TJA. The decreased rate of postoperative fever avoids unnecessary anxiety for the patient and the treating team and reduces healthcare resource use occasioned by working up postoperative fever.

Level of Evidence

Level III, therapeutic study. See Guidelines for Authors for a complete description of levels of evidence.

Introduction

Recent advances in pain management strategies have led to improvements in perioperative pain control in patients undergoing total joint arthroplasty (TJA) [12]. These advances have included, among others, multimodal pain management, defined as the use of strategies targeting different steps in the pathways of pain generation, transmission, and perception, thus allowing more effective pain control with fewer side effects [8]. At this institution, a multimodal pain management protocol, including acetaminophen, pregabalin, and celecoxib or toradol after TJA, was implemented. This protocol has proven to be efficacious and safe [14].

Elevated temperatures after TJA are not uncommon and can occur in over one-third of patients [1, 7]. Although this represents a physiologic reaction to the surgical insult of TJA and is only rarely a result of acute infection, many patients are subjected to extensive workups [4, 18]. In addition, postoperative fevers can be a source of anxiety for both the patient and the surgical team.

Since the implementation of the multimodal analgesia protocol, which includes several agents with antipyrexic properties, we observed an anecdotal decrease in the occurrence of fever in patients on the floor after TJA, which led to this retrospective study. This study sought to quantify this effect and, if proven, disseminate the findings.

Therefore, this retrospective analysis was established to analyze whether patients treated under the new multimodal analgesia protocol, which includes several agents with antipyrexic properties, were less likely to exhibit postoperative fever after primary TJA compared with a historical control group and whether they were less likely to receive postoperative testing as part of a fever workup.

Patients and Methods

This was a retrospective study performed at a single institution. After obtaining institutional board review approval, two separate time periods (July 2004 to December 2006 and July 2009 to December 2011) were identified before and after complete implementation of the multimodal pain management protocol by three of our adult reconstruction specialists (MSA, WJH, JP). Patients treated by these three surgeons from December 2006 to July 2009 were excluded because this was a transition period during which there was not a uniform protocol for postoperative pain management.

A total of 3613 patients (1946 females and 1667 males) undergoing 3901 primary TJAs at a mean age of 61.4 years (range, 16.9–97.1 years) were included in the study. This represented all patients treated by these three surgeons during each respective time period. When the multimodal protocol was in place, patients with any contraindications (ie, renal insufficiency, known medication allergy, hepatic failure, etc) to administration of any component protocol were managed with alternative medications including opioid pain control regimens. For patients undergoing more than one procedure, each admission was treated separately in the analysis. The procedures consisted of 2386 primary THAs, of which 137 (5.7%) were simultaneous bilateral cases, and 1515 primary TKAs, of which 239 (15.8%) were simultaneous bilateral cases. A total of 1484 procedures (939 primary THAs and 545 primary TKAs) were performed with a primarily opioid-based pain management protocol, whereas 2417 (1447 primary THAs and 970 primary TKAs) procedures were performed with the multimodal pain management protocol. Patients in the control group had a mean inpatient length of stay of 3.6 days (95% confidence interval [CI], 3.5–3.7 days) compared with 2.9 days (95% CI, 2.8–2.9 days) for the multimodal cohort (p < 0.001).

Patients in the control group received an opioid-based perioperative pain management protocol consisting of a postoperative intravenous patient-controlled analgesia (PCA) pump with fentanyl followed by oral opioid medication (tramadol, hydrocodone/acetaminophen, or codeine/acetaminophen) once the PCA was discontinued, usually on postoperative day (POD) 1.

Patients in the multimodal group received a nonopioid-based pain management protocol consisting of 1000 mg acetaminophen, 400 mg celecoxib, and 75 mg pregabalin orally 1 hour before the incision. Only 200 mg celecoxib was given if patients were on celecoxib before surgery. Postoperatively, patients were systematically administered 1000 mg acetaminophen orally every 6 hours in addition to 75 mg pregabalin and 200 mg celecoxib orally every 12 hours. Patients undergoing TJA and presenting with a sulfa allergy received ketorolac (30 mg intravenously) in place of celecoxib. NSAIDs were used regardless of whether cemented or cementless fixation was used. Insufficient pain control with these agents alone was addressed with oral opioid medications (tramadol or oxycodone) on an as-needed basis.

All patients received antibiotic prophylaxis per institutional protocol with intravenous cefazolin within 1 hour of incision unless the patient was allergic, in which case vancomycin was administered (vancomycin in addition to cefazolin was also administered in patients at risk of methicillin-resistant Staphylococcus aureus infection per protocol). All patients were prescribed incentive spirometry in the immediate postoperative period. All procedures were performed under spinal anesthesia with bupivacaine unless contraindicated; no changes in the protocol regarding anesthesia were made between the two time periods.

Oral temperature readings in the first 5 PODs were collected for these patients and charts were reviewed for febrile patients to record fever evaluation tests, including urinalysis, urine culture, chest radiograph, and blood culture. Fever was defined by the presence of a temperature measurement over 38.5 °C. Patients having preoperative fever or postoperative fever starting later than POD 5 were excluded from the analysis. This led to the exclusion of four patients based on the presence of preoperative fever, three in the control group and one in the multimodal group. Three patients in the control group and two patients in the multimodal group were further excluded because of fever appearing after POD 5. These were patients who presented complications in their postoperative course, necessitating a prolonged hospital stay. There were no differences between the groups in terms of preoperative temperature: 36.2 °C in control patients and 36.2 °C in multimodal patients, average postoperative temperature readings in the 2 groups are graphically represented (p = 0.53; Fig. 1).

Fig. 1.

The average postoperative temperatures after TJA for patients on the control group versus patients on the multimodal group are depicted.

A positive workup for fever was defined as urinalysis with a positive leukocyte esterase and/or nitrite test or with a white blood cell count > 5 cells per high-power field, urine culture results with growth > 100,000 colonies, chest radiographs with infiltrates or opacities related to conditions that may explain the fever, or blood cultures with positive growth. Chest radiographs performed postoperatively in febrile patients but not related to a workup of isolated fever (eg, fever with presence of shortness of breath or desaturation) were excluded from the analysis.

Pearson’s chi-square test was used for comparison of fever rates between the two groups (main outcome) as well as for all other qualitative variables. Independent Student’s t-test was used to compare quantitative variables. Nonparametric Mann Whitney U-test was used to compare the POD of appearance of fever and of maximal fever between both groups. There were no missing data in this analysis. SPSS Version 16.0 (SPSS Inc, Chicago, IL, USA) was used for all statistical analyses.

Results

Patients in the multimodal analgesia group were less likely to develop postoperative fevers. Over the 5-day time period, 377 patients (25.4%) in the control cohort developed a postoperative fever compared with 109 (4.5%) in the multimodal group (p < 0.001; Fig. 2). This resulted in a relative risk (RR) of 5.6 (95% CI, 4.6–6.9) for developing postoperative fever in the control cohort compared with the multimodal protocol. In both groups, fever appeared most frequently on POD 1 (p = 0.801). When breaking down patients by joint, fever occurred in 25.0% (235 of 939) of patients after THA in the control group and 4.4% (63 of 1447) in the multimodal cohort (p < 0.001) compared with 26.1% (142 of 545) of patients after TKA in control patients and 4.7% (46 of 970) in the multimodal cohort (p < 0.001; Table 1).

Fig. 2.

The percentage of patients that received temperature readings for each time period compared with the rate of patients with pyrexia in between the control and the multimodal groups is depicted.

Table 1.

Comparison of febrile response based on analgesic modality

Variables	Control group	Multimodal group	p value
Total number of admissions	1484	2417
Preoperative temperature (°C)	36.2	36.2	0.530
Average midnight temperature response (°C)	37.4	36.7	< 0.001
Total number of febrile admissions	377 (25.4%)	109 (4.5%)	< 0.001
Fever rates by joint	THA: 25.0%,	THA: 4.4%,	< 0.001
	TKA: 26.1%	TKA: 4.7%	< 0.001
Maximal temperature (°C)	38.9	38.8	0.018
Mean number of spikes over 38.5°C	1.9	1.5	0.013

Fewer patients in the multimodal analgesia cohort underwent workup for fever than did patients in the control group (1.8% versus 9.8%, p < 0.001; RR, 5.4 [95% CI, 3.9–7.5]). These numbers represented 44 patients undergoing 155 individual tests and 145 patients undergoing 247 individual tests, respectively (p < 0.001; Table 2). When looking at febrile patients only, 38.5% of control patients were worked up compared with 40.4% of multimodal patients (p = 0.72). The workup was positive in 34.5% of control cases and 40.9% of multimodal cases (p = 0.44). A notable driver for differences in workup of febrile patients between the two groups was the use of urinalysis. A subanalysis in which urinalysis was excluded as a workup measure showed that 3.0% versus 1.7% (p = 0.01) of patients were worked up in the control versus multimodal cohorts, respectively. Furthermore, the rate of positive febrile-related workup was 0.6% versus 0.4% (p = 0.66), respectively. The incidence of high-risk fevers (> 39º C, appearance of fever after POD 3 or fever lasting > 24 hours) varied significantly (p < 0.001) between the control (8.0% [118 of 1484]) and multimodal cohorts (1.0% [25 of 2417]). Eighty-four percent of the multimodal patients with high-risk fevers received a workup (21 of 25) compared with only 51% (60 of 118) in the control group (p = 0.003) with a positive result in 38% (eight of 21) and 30% (18 of 60), respectively (p = 0.59).

Table 2.

Details of fever workup in both groups

Variables	Control group (n = 1484)	Multimodal group (n = 2417)	p value
Admissions including any fever workup	145 (9.8%)	44 (1.8%)	< 0.001
Admissions with positive workup	50 (34.5%)	18 (40.9%)	0.44
Admissions with urinalysis	143 (9.6%)	43 (1.8%)	< 0.001
Admissions with positive urinalysis	48 (3.2%)	12 (0.5%)	< 0.001
Admissions with urine culture	37 (2.5%)	39 (1.6%)	0.06
Admissions with positive urine culture	7 (0.5%)	5 (0.2%)	0.23
Admissions with chest radiograph	17 (1.1%)	30 (1.2%)	0.88
Admissions with positive chest radiograph	3 (0.2%)	7 (0.3%)	0.75
Admissions with blood cultures	8 (0.5%)	7 (0.3%)	0.29
Admissions with positive blood cultures	0 (0%)	0 (0%)	1.0

Discussion

Pain management protocols based on intravenous opioid administration through PCA have constituted the traditional protocols for postoperative pain management in patients undergoing TJA [8]. In addition to posing a challenge for perioperative pain control in the many patients taking chronic opioids preoperatively, these protocols are associated with a number of adverse effects, delays in recovery, and prolongation of hospital stays [16]. Because of these limitations, we developed a multimodal pain management regimen for TJA, which includes acetaminophen, pregabalin, and celocoxib or toradol. These drugs, used alone or in combination, have been shown to be effective in managing postoperative pain and serve as opioid-sparing adjuncts in various other perioperative pain management protocols as well [2, 3, 5, 6, 9, 11]. This protocol’s pain control was shown to be efficacious with fewer side effects in a previous analysis [15]. Subsequently, we anecdotally noticed a decrease in the frequency of postoperative fevers, perhaps because several of the agents in our multimodal analgesia protocol have antipyrexic properties. We sought to validate and quantify this finding; accordingly, the purpose of this study was to determine whether patients treated under the multimodal protocol were less likely to experience postoperative fever and whether they were less likely to undergo a medical workup for fever.

This study had a number of limitations. This is a retrospective review in which patient management was not uniform. This perhaps makes the study more consistent with real-life practice but no doubt allowed discrepancies in approaches to postoperative fever workups among providers. This was mitigated by including the same surgeons before and after the introduction of the multimodal pain protocol and similar staff. The length of inpatient stay for these two groups was disparate. This likely was a result of the broader trends toward shorter hospitalization after TJA during the study period, and it may have caused some inflation in the frequency of fevers in the control cohort, whose length of stay was greater, compared with the multimodal cohort simply because of the prolonged opportunity for temperature readings. It seems to us, however, that this is unlikely to account for the over fivefold increase in pyrexia between the study groups, especially because the most frequent day for pyrexia was the first day after surgery. Finally, we do not have anesthesia data accessible for all patients in the opioid protocol period; however, the general approaches and anesthesia protocols were similar during these two time periods.

When observing the frequency of “positive” workup between the two groups, no difference was noted. This is surprising considering that the multimodal analgesia group had fewer overall workups. Unfortunately, this analysis did not investigate the validity of these positive workups. The fact that there is not a difference in the rate of workups that were positive suggests and is supported by previous analyses [10] that a positive workup in the postoperative setting frequently does not represent true pathology. Urinalysis was the major driver of differences between the cohorts with respect to the incidence and positive rate of workup. However, in a subanalysis in which urinalysis was not considered, we found no difference in the overall rate of positive febrile-related workups. This analysis compared two pain management protocols and their associations with postoperative fever; unfortunately, it did not allow for isolated comparisons of specific antipyretic pain medications. There was occasional overlap in the two protocols where antipyretics like Tylenol were used, and this likely did result in the control group patients’ occasional receiving antipyretic medication such as Tylenol. However, despite this, there remained substantial differences favoring the multimodal group in terms of the frequency of fever and workups for it. The present study did not investigate the safety of the multimodal protocol, and it does introduce some element of “polypharmacy” in a population that included older patients; however, the safety of the multimodal pain management protocol has been studied previously, and it was reported to be safe [13].

Fever is a known event to occur after TJA and multiple authors have concluded that this febrile reaction mostly represents a normal physiological response to the stress of surgery and should not warrant evaluation in the early postoperative period, especially with negative patient history and physical examination [1, 4, 7]. Indeed, Buvanendran et al. [2] demonstrated a significant increase in cytokines, especially interleukin-6, in the serum and wound drain fluid of patients who underwent TKA, partly explaining the high incidence of fever observed in those patients. Fewer patients in the multimodal analgesia group than in the control group experienced postoperative fevers. Reported rates of postoperative fever after TJA have varied extensively in the literature between different authors based on the definition of fever, including numerical thresholds, methods of measurement, and the studied populations. Ghosh et al, for example, reported a rate of 36.5% after TKA when looking at temperatures over 38 °C for 5 days using axillary measurements [7]. Similar rates were found in a study by Athanassious et al. with 36% of patients undergoing THA and 31% undergoing TKA exhibiting postoperative fever [1]. In another study on patients undergoing TKA, 45.2% of patients had temperatures over 38 °C, whereas only 10.2% had readings > 38.5 °C [17]. Most studies have reported average peak temperatures on POD 1, similar to what was demonstrated in this cohort of patients [1, 4, 15]. Kennedy et al. reported that pyrexia, as defined by an axillary temperature greater than 37 °C, occurred in all patients undergoing TKA at their institution [10]. In their series, 17% of patients had temperature readings over 39 °C with none of the patients developing subsequent infection. Interestingly, patients with a diagnosis of urinary tract infection in the latter series all had temperature readings < 38 °C. When defining fever as > 39 °C, the incidence of postoperative fever was 19% in another study [10]. The detected rate of fever in 25.5% of our patients undergoing TJA falls within these reported ranges in the literature.

Fewer patients in the multimodal analgesia group than in the control group underwent pyrexia workup after surgery. Because the proportion of febrile patients undergoing a workup was similar between the two groups, the decreased rate of workup is likely attributable only to the decreased incidence of fever in the multimodal analgesia group. In this study, urinalysis (a historically nonspecific test) was the major driver of differences between the cohorts with respect to the incidence and positive rate of workup. However, when not considering urinalysis, there was no difference noted in the overall rate of positive febrile-related workup, yet the decreased rate of workup in the multimodal group compared with the control group persisted. Consistent with previous literature, it appears that workup of fever in the postoperative period does not usually lead to identification of a serious infection in the majority of patients [4, 18]. In one study, in which fever was defined as temperature readings > 40.2 °C, only 28.6% of patients undergoing TKA and 20.5% of patients undergoing THA had a positive workup for fever [4]. In another study, workup was performed in 42.9% of febrile patients, similar to the proportion in this series, with an average cost for fever evaluation per patient amounting to USD 959.45 when excluding costs associated with the incurred increase in the hospital stay [18]. The cost per health effect, that is the cost per change in treatment management, was USD 8208. Because of the associated cost and lack of an association between postoperative fever and a serious infection, most studies caution against routine workup of postoperative fever [7, 10, 15]. One study has revealed that fever occurring after POD 3, fevers lasting > 24 hours, and fevers with temperature of 39 °C or higher were more likely to be important and would benefit from full workup [18]. When identifying such patients in this cohort, no differences compared with the general febrile cohort were identified. The yield of workup in “high-risk” febrile patients was nearly identical to the yield in all febrile patients for both the multimodal and the control groups. Despite the presence of evidence suggesting that postoperative fever is a relatively common event and routine workup may not be justified, the anxiety that this event provokes for patients and the treating medical team as well as the fear of potential litigation leads many physicians to routinely administer fever workup [4, 7, 18].

Although this study did not further address the safety or efficacy of postarthroplasty multimodal pain management as established in a previous study [15], it did demonstrate that the administration of multimodal analgesia lowers the rate of postoperative fever in this group of patients undergoing TJA. The study revealed that benefits of multimodal analgesia extend beyond its efficacy and opioid-sparing effects to include a reduction in the frequency of postoperative fever and the frequency of fever workup.