Introduction

Despite advances in interventions to reduce the risk of surgical site infections [1, 7, 15, 27, 40], periprosthetic joint infection (PJI) remains a devastating problem, one that causes a severe burden to patients, the healthcare system, and the economy [2, 3, 8, 25, 26]. In fact, PJI accounts for approximately 15% of all revision THAs [5] and 25% of all revision TKAs [4] with an absolute number of revisions related to PJI that is projected to increase severalfold over the next two decades [25]. Given the tremendous impact of PJI on patients and society alike, optimizing patient-related risk factors for PJI is of paramount importance.

Malnutrition, as defined by abnormal laboratory parameters or anthropometric measurements, has been linked in the orthopaedic literature to postoperative complications after hip and knee arthroplasty, ranging from increased length of stay to impaired wound healing [10, 11, 13, 14, 18, 19, 21, 32, 33]. Furthermore, the reported prevalence of malnutrition among patients undergoing total joint arthroplasty (TJA) has been reported to be as high as 50% [22, 34]. As the population continues to grow increasingly obese, malnutrition may be of even greater importance for patients undergoing TJA and the surgeons who take care of them, because obese patients can be paradoxically malnourished [23, 24, 36] secondary to the consumption of calorie-dense but protein- and nutrient-poor foods [36]. Although malnutrition has been identified as a risk factor for the development of PJI after irrigation and débridement (I&D) to treat persistent postoperative wound drainage [19], previous studies suggesting a link between malnutrition and PJI after a primary or revision TJA [12, 18, 21, 32] have been limited by small sample sizes [12, 21], heterogeneous study populations consisting of both primary and revision TJAs [18, 32], and vague definitions of malnutrition [32]; thus, strong evidence linking PJI after revision TJA remains lacking.

The purposes of our study were to determine (1) if one or more laboratory values suggestive of malnutrition is independently associated with being revised for infected joint arthroplasty failure as opposed to being revised for aseptic failure; (2) the relationship between laboratory parameters suggestive of malnutrition and obesity; and (3) if laboratory parameters suggestive of malnutrition are independently associated with acute PJI complicating an aseptic revision procedure.

Patients and Methods

After institutional review board approval, we retrospectively reviewed data on all nonsecond-stage reimplantation revision procedures (600 total in 547 patients; 325 knees and 275 hips) performed by one fellowship-trained orthopaedic surgeon (CJDV) between November 2002 and December 2010 for potential study inclusion. During the study period, patients had a total lymphocyte count, serum albumin, and serum transferrin performed as part of their routine preoperative care. Laboratory parameters suggestive of malnutrition were defined as total lymphocyte count < 1500/mm3, serum albumin < 3.5 g/dL, or serum transferring < 200 mg/dL [10, 19]. Although less robust and precise than that suggested by the Academy of Nutrition and Dietetics and American Society for Parenteral and Enteral Nutrition [38], the most commonly used definition of malnutrition in the orthopaedic literature has been at least one of three abnormally low laboratory values (serum albumin, transferring, or total lymphocyte count) [10]; therefore, for the purposes of this study, we use this same definition for potential malnutrition. Patients were excluded from our analysis if they had an early revision performed within the first 6 weeks after a primary or revision arthroplasty (in the early postoperative period, the laboratory values used to diagnose malnutrition are decreased secondary to postoperative stress and are therefore unreliable for nutritional status assessment) [14, 21, 22] or if no nutrition laboratory values were available for review.

After excluding 99 revision procedures performed in 84 patients, 501 of the original 600 revision procedures (84%) performed in 463 patients were included in our final analysis (Fig. 1): 375 failures were secondary to an aseptic mode of failure (200 knees and 175 hips) and 126 failures were secondary to a deep periprosthetic joint infection (70 knees and 56 hips). The most common reason for revision of the 375 aseptic revisions was aseptic loosening of components (Table 1). The mean age at the time of revision surgery was 65 ± 12 years (range, 31–95 years), and in the study group, 310 revisions were performed in women (62%) and 191 in men (38%).

Fig. 1
figure 1

A flowchart shows patients who underwent revision TJA included in the study.

Full size image
Table 1 Reasons for revision
Full size table

Age, sex, insurance type, race, Charlson Comorbidity Index (CCI), and body mass index (BMI) at the time of revision were also reviewed for each patient. Race was coded as “white” or “nonwhite” and insurance was coded as “private” or “nonprivate.” All data were collected by review of a prospective database and patient charts by someone other than the operating surgeon. Each patient was categorized as underweight (BMI < 18.5 kg/m2), normal weight (BMI ≥ 18.5 kg/m2 and < 25 kg/m2), overweight (BMI ≥ 25 kg/m2 and < 30 kg/m2), or obese (BMI ≥ 30 kg/m2) [39].

The patients who underwent an aseptic revision were subsequently reviewed for the presence of PJI presenting within the first 90 days postoperatively. Infection was diagnosed using Musculoskeletal Infection Society criteria [31]. All patients were followed for 90 days postoperatively. No patients were recalled specifically for this study; all data were obtained from medical records.

Statistical Analysis

The prevalence of having one or more laboratory parameters suggestive of malnutrition and demographic variables was compared between the patients who underwent the 375 aseptic revisions and the patients who underwent the 126 septic revisions and between patients who underwent aseptic revision who did and did not have an acute postoperative infection complicating their procedure using Fisher’s exact test and Student’s t-test, as appropriate. The pooled variance t-test was performed unless there was a violation of the homogeneity of variance assumption, in which case the Satterthwaite adjustment method was used. Multivariate logistic regression modeling was used to evaluate independently associated factors for (1) septic as opposed to an aseptic mode of failure; and (2) acute postoperative infection complicating an aseptic revision. Covariates of age, sex, BMI, CCI, race, and insurance type were included in the model to adjust for these effects. A power analysis was performed a priori indicating that a minimum of 36 patients per group would be required to detect a difference in prevalence of having one or more laboratory parameters suggestive of malnutrition of 35% versus 5% with an α value of 0.05 and power of 0.90. These values were chosen based on prior work [19] that showed that the prevalence of potential malnutrition (using the same definition as in our study) at the time of I&D after a THA or TKA was 35% in those who developed subsequent PJI compared with 5% in those who did not. All analysis was performed using SAS® Version 9.1.3 software (SAS Institute Inc, Cary, NC, USA).

Results

The overall prevalence of malnutrition in our study population was 38% with patients in 190 of 501 procedures having one or more laboratory parameters suggestive of malnutrition. Patients in 67 of 126 (53%) revisions for PJI had laboratory parameters suggestive of malnutrition compared with 123 of 375 (33%) undergoing revision for a noninfectious etiology (odds ratio [OR], 2.3; 95% confidence interval [CI], 1.5–3.5; p < 0.001). Demographics of the two groups are presented (Table 2). Multivariate regression analysis revealed that having one or more laboratory parameters suggestive of malnutrition is independently associated with septic revision (p = 0.003; OR, 2.1, 95% CI, 1.294–3.512) along with younger age (p = 0.0445; OR, 0.979; 95% CI, 0.957–1.001), male sex (p = 0.0250; OR, 1.776; 95% CI, 1.075–2.936), and having nonprivate insurance (p = 0.0143; OR, 2.311; 95% CI, 1.182–4.516) (Table 3).

Table 2 Demographics of patients undergoing revision arthroplasty
Full size table
Table 3 Potentially associated factors in multivariate analysis for chronic septic failure
Full size table

The presence of one or more laboratory parameters suggestive of malnutrition was most common among patients of normal weight at the time of revision compared with overweight patients (p = 0.012; OR, 2.8; 95% CI, 1.5–5.2) and obese patients (p = 0.0023; OR, 2.2; 95% CI, 1.3–3.7) (Table 4). Paradoxical “malnutrition” (obese but having laboratory parameters suggestive of malnutrition) was common in both the aseptic and septic failure groups (41% and 29%; p = 0.1494). Demographic characteristics of patients with laboratory parameters suggestive of malnutrition are presented (Table 5).

Table 4 Laboratory parameters suggestive of malnutrition status by body mass index
Full size table
Table 5 Demographics based on nutritional status
Full size table

We found that having one or more laboratory parameters suggestive of malnutrition is independently associated with an acute postoperative infection complicating an aseptic revision arthroplasty (p = 0.02; OR, 5.9; 95% CI, 1.317–26.057) (Table 6). Of the 375 aseptic revisions, 12 developed an acute postoperative infection within the first 90 days (3%). The frequency of infection was nine of 123 in the group having laboratory parameters suggestive of malnutrition and three of 252 in the group having normal laboratory parameters (7% versus 1%; p = 0.003). Age was also independently associated with each additional year of age decreasing the odds of acute infection by 8%.

Table 6 Potentially associated factors in multivariate analysis for acute periprosthetic joint infection complicating aseptic revision arthroplasty
Full size table

Discussion

Although having one or more laboratory parameters suggestive of malnutrition is commonly cited as increasing susceptibility to PJI after hip and knee arthroplasty, direct data confirming this link have been lacking. Hence, we sought to assess whether having laboratory parameters suggestive of malnutrition was independently associated with septic as opposed to a noninfectious mode of failure in a series of patients undergoing revision TJA, the relationship of having laboratory parameters suggestive of malnutrition to obesity, and if having laboratory parameters suggestive of malnutrition predisposed to an acute postoperative infection complicating an aseptic revision procedure. Our results suggest that the presence of one or more laboratory parameters suggestive of malnutrition is common among both obese and normal-weight patients, and in our data set, it was independently associated with both chronic PJI and the development of an acute postoperative infection after an aseptic revision.

We note several limitations of our study. First, this is a retrospective study and selection bias is a concern, because we excluded several potential patients who did not have malnutrition laboratory values available for review; additionally, several patients lacked demographic data points such as BMI and insurance type and had to be dropped in the multivariate analysis. However, because the number of patients with missing data was a small proportion of the total study population and our prevalence of having one or more laboratory parameters suggestive of malnutrition is consistent with previous estimates in patients undergoing TJA [22, 34], we believe that these missing data did not substantially bias our results. Furthermore, although a prospective study is ideal, the rarity of PJI makes prospectively studying the topic difficult. Our series was from an urban tertiary referral center that specializes in the treatment of complex and infected revisions, calling into question the generalizability of our results and raising potential bias toward infected and complex cases that are more prone to infection. However, in our multivariate analysis, we attempted to adjust for these potential confounders such as age and medical comorbidities. Fourth, our sample size is relatively small with only 12 acute infections noted in our study. Nevertheless, our sample size was large enough to identify the association. Finally, although serum biochemical markers are good indicators of nutritional status, they can be affected by chronic and acute diseases associated with inflammation and stress [12, 16, 20, 28]. Although we attempted to account for such potential confounders by multivariate analysis taking into account patient comorbidities, the possibility remains that the presence of low malnutrition laboratory values does not indicate malnutrition in all patients (serum albumin has been shown to not consistently change with caloric deprivation or weight loss [38]) and may, in fact, be more reflective of an inflammatory response than poor nutritional status [38]. Furthermore, it is a possibility that chronic PJI could theoretically cause laboratory abnormalities suggestive of malnutrition; because our study was purely observational, we are unable to comment on the existence of a causal relationship between potential malnutrition and PJI nor on the direction of such a relationship if it exists. Our definition of potential malnutrition is thus limited and should be interpreted in light of these considerations.

In our series of 501 revision TJAs, the prevalence of having one or more laboratory parameters suggestive of malnutrition was 38%, which is consistent with previous reports’ prevalence of malnutrition ranging from 8.5% to 50% for patients undergoing both primary and revision TJA [14, 18, 21, 22, 34, 35]. Our findings may be reflective of society as a whole; the general hospital and elderly populations have estimated malnutrition prevalence rates of 11% to 44% and 29% to 61%, respectively [9, 17]. Patients undergoing septic revision for chronic PJI were more likely to have laboratory parameters suggestive of malnutrition than those undergoing aseptic revision. Furthermore, after accounting for demographic variables and comorbidity index, the presence of laboratory parameters suggestive of malnutrition remained independently associated with septic revision, suggesting a potential link between the presence of laboratory parameters suggestive of malnutrition and the development of chronic PJI. In both a retrospective study of 6489 primary and revision TKAs [32] and a prospective study of 213 TKAs [12], preoperative malnutrition increased the risk of postoperative infection. However, the results of these studies are difficult to interpret, because both studies grouped deep and superficial infections as the infectious outcome variable. Furthermore, the study of 6489 TKAs [32] did not define what constituted malnutrition. The prospective study of 213 TKAs [12], on the other hand, defined malnutrition as decreased triceps skin fold, which is an indirect measure of malnutrition. Recently, a prospective study of 2161 patients undergoing elective primary or revision TJA (grouped together) found that malnutrition was more common among patients who developed PJI compared with those who did not [18]; however, in multivariate analysis, malnutrition was not shown to be an independent risk factor for PJI, suggesting that other confounding variables may have influenced their finding. We also found that patients undergoing septic revision were more likely to be male and to have nonprivate insurance, which is consistent with previous studies’ findings that male sex [6, 29] and lower socioeconomic status are risk factors for PJI [30, 37].

Interestingly, many patients were both obese and had one or more laboratory parameters suggestive of malnutrition in both the septic and aseptic groups, demonstrating a paradoxical “malnutrition” (laboratory parameters suggestive of malnutrition in an obese patient). Although it may seem logical that obese patients, by virtue of having caloric excess, would be adequately nourished (or even overnourished), our data show that the relationship between laboratory parameters suggestive of malnutrition and obesity is not so straightforward. In fact, our results suggest the opposite, that obese patients frequently have laboratory parameters suggestive of malnutrition, perhaps secondary to the consumption of calorie-dense, nutrient-poor foods [36]. This finding is consistent with previous observations that obese patients are frequently paradoxically malnourished [23, 24, 36] and confirms previous findings from a recently published study on the impact of malnutrition on TJA outcomes [18]. Nevertheless, normal-weight patients were more likely to have laboratory parameters suggestive of malnutrition than those who were overweight or obese; hence, we believe this indicates that laboratory parameters suggestive of malnutrition are a potential problem in patients of any size or weight.

The most concerning finding in our study, however, was the strong relationship between the presence of one or more laboratory parameters suggestive of malnutrition and the incidence of an acute postoperative infection complicating an aseptic revision with an OR of almost 6. In a somewhat more complex clinical scenario, one retrospective study showed that malnutrition, using the same definition as the one used in this study, is a risk factor for the failure (ie, subsequent progression to PJI) of a persistently draining wound treated with I&D [19]. This suggests that surgeons may wish to consider delaying an elective revision if potential malnutrition is identified by way of one or more laboratory parameters. However, we must again highlight that our study was observational in nature and, as such, unfortunately cannot provide insight into causality between potential malnutrition and infection risk; such a relationship remains unknown. Furthermore, whether infection may cause these laboratory abnormalities or vice versa remains indeterminate as well.

In summary, our study suggests that having one or more laboratory parameters suggestive of malnutrition is common among patients undergoing revision arthroplasty and is independently associated with both chronic septic failure and acute postoperative infection complicating revisions performed for a noninfectious etiology. Nevertheless, a direct causal effect of malnutrition on PJI remains unclear. A prospective study implementing a standardized protocol of malnutrition screening and subsequent correction as needed would be ideal to assess such a potential relationship. Encouragingly, efforts for such a study are reportedly already underway [18].