Introduction

Despite the enormous progress in surgical treatment for reducing the risk of infections, periprosthetic joint infections (PJI) remain a major challenge and risk for the affected patients, the health system, and attending personnel [1,2,3,4]. Considering the grave effects of PJI on patients and the society, the optimization of patient-related risk factors is vital [4]. The effects of nutritional status and protein balance are considered to be significant prognostic factors for the development of PJI [5, 6]. Malnutrition is reported to be associated with complications, which range from prolonged hospital stay to impaired wound healing, after the hip and knee endoprosthesis implantation [5, 7,8,9,10,11]; however, this association seems to receive too little attention in current clinical practice. Jensen et al. reported that 50% of the patients who received elective total hip arthroplasty (THA) already had a clinical or subclinical nutrient deficiency before surgery [12]. The increasing supernutrition and the growing number of obese persons in industrial countries, which have caused an increase in metabolic diseases, support this finding. In contrast, Kaidar-Person et al. reported that obese persons can also be malnourished because their diets are often low in vitamins, proteins, and nutrients [13, 14]. In particular, low serum albumin level and protein deficiency are reportedly negative predictive factors with respect to the rates for perioperative and postoperative complications, such as PJI, associated with primary joint endoprostheses [15, 16]. Accordingly, a seven-fold increased risk of arthroplasty-related infections was observed at a preoperative albumin level of <35 g/L [9]; however, proteins are essential for adequate wound healing and serve as carriers for most vitamins (such as vitamin D and its metabolites), hormones, and mediators. The importance of vitamin D in orthopedic infections and particularly in septic disease progression has been proven [17,18,19,20]. It is also known that 99% of the vitamin D present in the human body exists in a protein-bound form (e.g. vitamin D-binding protein and albumin) and that it is fully effective only when the protein level is balanced [21, 22]. It is striking that none of the cited studies took the importance of the nutritional status, particularly the protein balance, into account.

This retrospective analysis aimed to examine the nutritional status of and the protein balance in patients with periprosthetic hip and knee joint infections. Patients with primary endoprosthesis implantation and aseptic loosening constituted the comparison groups.

Material and methods

This study was approved by the ethics committee of the University of Leipzig (approval number: 025-16-01022016). The study was designed and the required sample size was estimated in advance using a power analysis program (G*Power; version 3.1.9.2., Axel Buchner, Edgar Erdfelder, Franz Faul, Albert-Georg Lang, Department of Psychology, Heinrich-Heine-University Düsseldorf, Germany). An effect size of medium value (d = 0.5) and a statistical test power of 80% (1 − β = 0.8) were assumed for detecting significant differences (α = 0.05).

This study was conducted at a university maximum care hospital in Europe and involved an analysis with patients of German nationality. The inclusion criteria for the study group (SG) were the presence of periprosthetic hip or knee joint infection, written consent for participation, and minimum age of 18 years. In the present study, periprosthetic infection was defined based on the criteria of the International Consensus Meeting on Periprosthetic Joint Infection (ICMPJI), 2013 [23, 24]. Patients with primary endoprosthesis implantation and revision surgery due to aseptic loosening served as the control groups I (CG I) and II (CG II), respectively. According to the matching procedure both the CGI and II were adapted to the SG with respect to age, body mass index (BMI), and implant location (THA or total knee arthroplasty, TKA). Aseptic loosening was defined as radiologically determined implant loosening with appropriate symptoms.

From 1 January 2015 to 29 September 2017, a total of 61 patients were included in the SG with 16/61 (26.2%) acute PJI and 45/61 (73.8%) low-grade PJI. Detailed information is given in Table 1. Acquisition of suitable patients for CGI and II (nCG I = 78 and nCG II = 55) was started at the same time but lasted until 3 November 2017. Total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, albumin, C‑reactive protein (CRP), hemoglobin, creatinine, and alanine aminotransferase levels as well as total protein in serum of all patients were determined via analysis of peripheral venous blood samples; all these levels were ascertained during inpatient admission prior to surgery. Furthermore, all patients completed a nutrition questionnaire comprising questions regarding eating habits, sports activities, and weight fluctuations. The nutritional risk screening (NRS) 2002 score and BMI were also calculated for all study participants [25].

Table 1 Comparison of the examination groups separated into acute and low-grade PJI with respect to laboratory tests

Data were statistically evaluated using Excel 2013 (Microsoft, Redmond, WA, USA) and SPSS v. 24.0 (IBM, Armonk, NY, USA). Data were reviewed for normal distribution using the Shapiro-Wilk test. The non-parametric Mann-Whitney U‑test was applied to compare metric scaled variables. Nominal and ordinal scaled variables were analyzed using the χ2-test or Fisherʼs test and p < 0.05 was considered to be statistically significant. Data were expressed as median and standard deviation (SD).

Results

Table 2 presents the characteristics of SG and CGs. A striking significant difference was observed in the NRS 2002 score between SG and CG I (SG: median = 1.0, SD = 0.9; CG I: median = 0.0, SD = 0.7, p < 0.001). A significant difference was also observed in the number of patients who exhibited a critical nutritional status (NRS 2002 score ≥3) between SG and both CGs (SG: 5/61, 8.2%; CG I: 1/78, 1.3%; CG II: 0/55, 0.0%). Fig. 1 shows the distribution of NRS 2002 scores within the individual groups; however, the differences found in NRS 2002 scores could not be confirmed using BMI, which showed no significant differences between the groups. Nevertheless, in all groups, the majority of the patients had an overweight BMI (Table 2; Fig. 2). The results of all laboratory tests as well as the laboratory-specific standard levels for each parameter are summarized in Table 3. The SG and both CGs showed significant differences in the hemoglobin (SG: 6.15 ± 2.05mmol/l, CG I: 8.56 ± 0.86mmol/l, p < 0.001, CG II: 8.18 ± 1.03mmol/l, p < 0.001) and CRP (SG: 86.21 ± 98.72mg/l, CG I: 4.55 ± 1.3 mg/l, p < 0.001, CG II: 13.69 ± 1.11mg/l, p < 0.001) levels. Significant differences were also found in total cholesterol levels, particularly HDL-cholesterol (SG: 1.04 ± 0.39mg/l, CG I: 1.62 ± 0.47mg/l, p < 0.001, CG II: 1.45 ± 0.46mg/l, p < 0.001) and LDL-cholesterol (SG: 2.65 ± 0.93mmol/l, CG I: 3.64 ± 1.18mmol/l, p < 0.001, CG II: 3.37 ± 0.94mmol/l, p < 0.001). The albumin levels (SG: 36.23 ± 7.34g/l, CG I: 44.37 ± 3.32g/l, p < 0.001, CG II: 44.06 ± 4.42g/l, p < 0.001) and total protein in serum (SG: 65.42 ± 8.66g/l, CG I: 70.80 ± 5.33g/l, p = 0.004, CG II: 71.22 ± 5.21g/l, p = 0.004) also significantly differed between CGs. A subanalysis of the available laboratory parameters of acute and low-grade PJI is shown in Table 1. In this context, it is noticeable that acute PJI showed significantly lower values concerning the albumin level and total protein in serum than low-grade PJI. With respect to the nutrition questionnaire responses, SG exhibited a significantly lower level of the daily time spent outdoors (SG: median = 1.0 h, SD = 1.15 h; CG I: median = 2.0 h, SD = 1.45 h, p = 0.006; CG II: median = 2.5 h, SD = 1.41 h, p < 0.001) or on the move (SG: median = 2.0 h, SD = 2.61 h; CG I: median = 3.0 h, SD = 5.35 h, p = 0.003 and CG II: median = 5.0 h, SD = 2.94 h, p < 0.001), i.e., for activities such as shopping or housework. Responses to the individual questions are presented in Table 4.

Fig. 1
figure 1

Nutritional risk screening (NRS) 2002 in the 3 groups: >3 nutritional risk points, preparation of a nutrition plan <3 points weekly repeated screening. If, for example, a large operation is planned for the patient a preventive nutrition plan should be pursued risk for PJI and wound healing disorders

Fig. 2
figure 2

Body mass index (BMI) values in the 3 groups

Table 2 Comparison of the examination groups (SG, CG I primary arthroplasty, CG II aseptic revision) with respect to the number of subjects, median age (min–max), number (absolute, %) of male subjects, THA or TKA BMI and NRS 2002
Table 3 Comparison of the examination groups (SG, CG I primary arthroplasty, CG II aseptic revision) with respect to laboratory evaluations
Table 4 Comparison of the examination groups (SG, CG I primary arthroplasty, CG II aseptic revision) with respect to questionnaire evaluation

Discussion

The current context of social, economic, and political changes has led to a modification in the eating habits of individuals in industrialized countries and has raised discussions about the impact of these changes. Thus, overnutrition and increasing obesity in particular, pose a serious problem. The proportions of overweight (BMI ≥25.0 kg/m2) adult women increased from 22.2% (1974–1975) to 39.1% (1989) and 47.0% (1995–1996), corresponding to an increase of approximately 112% [25]. In this context, associated metabolic diseases, such as diabetes mellitus and heart diseases, and degenerative joint diseases, such as gonarthrosis and coxarthrosis, have gained in importance [26]. This study also demonstrated an increase in severity of obese nutritional status (BMI >30 kg/m2) (SG: 41.0%, CG I: 43.6%, and CG II: 45.5%) in all groups. If patients with a BMI of >25 kg/m2 are included, as much as 68.9%, 79.5%, and 78.2% of the patients in SG, CG I, and CG II, respectively, had a pre-obese or obese nutritional status (Fig. 2).

Clinically relevant vitamin or mineral deficiency is reported to be rare in western countries owing to the low cost and unlimited variety of available foods; however, many people consume enough food that is either unhealthy or of low nutritional value, i.e., food with a lack of proteins, vitamins, minerals, and fiber. In a series of publications, Kaidar-Person et al. reported that the prevalence of symptoms of deficiency of vitamins, proteins, etc. in the morbidly obese population was significantly higher than expected; [13, 14]; therefore, a high BMI with pre-obesity does not completely rule out the possibility of malnutrition. With respect to overweight patients, the survey results also revealed vitamin D and protein deficiencies in all groups (Table 3); in particular, patients in the SG exhibited significantly lower protein levels (i.e., albumin level and total protein in serum) than those in both CGs (Table 3). This finding can be attributed to the increased consumption during the acute phase reaction and is particularly evident in acute as well as in chronic PJI [27]. Whether a balancing of the proteins, which are consumed in the acute phase reaction is meaningful, cannot be proven based on this study. An isolated substitution of albumin in the course of infection, especially in sepsis, is controversially discussed [28]; however, studies have shown that reduced albumin levels are associated with an increased risk of orthopedic wound infections [29]. A further prospective study is planned for this purpose. Furthermore, optimal blood glucose control should be ensured to assess the effects of overnutrition and metabolic changes, such as prediabetes mellitus. In the present study, compared with those in the CGs a significantly higher number of patients in the SG showed abnormal blood glucose levels (Table 3). It has been proven that an increased level of glycated hemoglobin (HbA1c), in particular, is associated with an increased risk of PJI [30]. Determination and optimization of the level of HbA1c prior to elective endoprosthesis implantation is also advisable. Although a level of <7% is generally recommended, there is no international consensus on the ideal target level of HbA1c; however, in the short term this level can only be achieved to a limited extent in patients with severe diabetes [31, 32]. Therefore, optimal postoperative blood glucose control until adequate wound healing seems to be essential to avoid PJI [31].

A tendency with respect to conspicuous BMI could not be confirmed in the individual groups because there were no significant differences among them. In contrast, studies have reported that obesity is one of the most important modifiable patient factors in predicting PJI after THA or TKA; this is evidenced by data from a total of 10,690 and 9481 cases of primary THA and TKA, respectively, retrieved from the New Zealand Surgical Site Infection Improvement Program between 2013 and 2015 [33]. The negative influence of associated metabolic diseases, such as diabetes mellitus and circulatory disorders, on the development of PJI has been proven [34, 35]. Although a high BMI is reportedly associated with higher rates of wound healing disorders and PJI, an altered or decreased BMI is not a useful marker for malnutrition [4].

In the present study, NRS proved to be good predictive factor for PJI. Approximately 75.4% of the patients in the SG showed conspicuous NRS 2002 scores (NRS 2002 score ≥1). The exact distribution within the individual groups is shown in Fig. 1. Particularly in the field of spinal surgery and arthroplasty, the importance of NRS with respect to wound infections has been demonstrated [5, 27, 36]. A combination of NRS and preoperative analysis of the albumin levels and total protein in serum seems to be essential in this aspect [9, 27].

Limitations

Within the scope of this study the consensus meeting criteria for defining PJI were used. This aspect could result in a lower detection of low-grade infections. Therefore, it would be sensible to use a more sensitive classification for future follow-up studies. A group-specific determination of HbA1c was not performed. In addition, the data are only a snapshot of the levels without a follow-up examination of the patients. For this reason, it is not possible to make statements about the previous nutritional and protein balance of the patients, especially before symptoms of PJI occurred. Therefore, it cannot be said whether they have contributed to the emergence of PJI. Finally, the outcome cannot be assessed according to the study parameters; however, a follow-up examination of the patients has been planned.

Conclusion

To predict the risk for periprosthetic infections selected protein levels (e.g., albumin level and total protein in serum) should be preoperatively determined. In addition, NRS can be performed to exclude nutritional deficiencies. In contrast, BMI is not suitable for assessing nutritional deficiency because malnutrition can also be present in a pre-obese nutritional condition. In the case of an existing PJI, compensation for the nutritional deficiency with dietary supplements is recommended, particularly for protein and vitamin deficiencies. An algorithm for dealing with malnutrition in elective arthroplasty but especially in PJI should be followed. A procedure-related proposal for both scenarios is presented in Figs. 3 and 4, respectively.

Fig. 3
figure 3

Diagnostic algorithm for assessing the nutritional status before elective arthroplasty

Fig. 4
figure 4

Diagnostic algorithm for assessing the nutritional status of periprosthetic joint infection (PJI)