Osteomyelitis Risk in Patients With Transfemoral Amputations Treated With Osseointegration Prostheses
Percutaneous anchoring of femoral amputation prostheses using osseointegrating titanium implants has been in use for more than 25 years. The method offers considerable advantages in daily life compared with conventional socket prostheses, however long-term success might be jeopardized by implant-associated infection, especially osteomyelitis, but the long-term risk of this complication is unknown.
(1) To quantify the risk of osteomyelitis, (2) to characterize the clinical effect of osteomyelitis (including risk of implant extraction and impairments to function), and (3) to determine whether common patient factors (age, sex, body weight, diabetes, and implant component replacements) are associated with osteomyelitis in patients with transfemoral amputations treated with osseointegrated titanium implants.
We retrospectively analyzed our first 96 patients receiving femoral implants (102 implants; mean implant time, 95 months) treated at our center between 1990 and 2010 for osteomyelitis. Six patients were lost to followup. The reason for amputation was tumor, trauma, or ischemia in 97 limbs and infection in five. All patients were referred from other orthopaedic centers owing to difficulty with use or to be fitted with socket prostheses. If found ineligible for this implant procedure no other treatment was offered at our center. Osteomyelitis was diagnosed by medical chart review of clinical signs, tissue culture results, and plain radiographic findings. Proportion of daily prosthetic use when osteomyelitis was diagnosed was semiquantitatively graded as 1 to 3. Survivorship free from implant- associated osteomyelitis and extraction attributable to osteomyelitis respectively was calculated using the Kaplan-Meier estimator. Indication for extraction was infection not responsive to conservative treatment with or without minor débridement or loosening of implant.
Implant-associated osteomyelitis was diagnosed in 16 patients corresponding to a 10-year cumulative risk of 20% (95% CI 0.12–0.33). Ten implants were extracted owing to osteomyelitis, with a 10-year cumulative risk of 9% (95% CI 0.04–0.20). Prosthetic use was temporarily impaired in four of the six patients with infection who did not undergo implant extraction. With the numbers available, we did not identify any association between age, BMI, or diabetes with osteomyelitis; however, this study was underpowered on this endpoint.
The increased risk of infection with time calls for numerous measures. First, patients should be made aware of the long-term risks, and the surgical team should have a heightened suspicion in patients with method-specific presentation of possible infection. Second, several research questions have been raised. Will the surgical procedure, rehabilitation, and general care standardization since the start of the program result in lower infection rates? Will improved diagnostics and early treatment resolve infection and prevent subsequent extraction? Although not supported in this study, it is important to know if most infections arise as continuous bacterial invasion from the skin and implant interface and if so, how this can be prevented?
Level of Evidence
Level IV, therapeutic study.
Overall outcome for the first 100 patients with a transfemoral amputation indicated a higher success rate with newer treatment protocols and greater surgical experience, but failures relating to infections were not systematically evaluated . Branemark et al.  reported that four of 51 patients experienced a deep infection during a 2-year period. Previous prospective data from our center in a similar but smaller cohort suggested that 18% of patients have an implant-associated deep infection during a 3-year period . Many of these infections showed little inflammatory activity and involved only limited and temporary loss of function for the patient . S aureus was slightly more common than coagulase-negative staphylococci followed by Enterococcus spp. in diagnostic cultures.
This method, intended for life-long prosthetic limb support, is potentially vulnerable to infection limiting its usefulness. Therefore, we aimed to (1) quantify the risk of osteomyelitis, (2) characterize the clinical effect of osteomyelitis (including the risk of prosthesis extraction and impairments to function), and (3) determine whether common patient factors (age, sex, body weight, diabetes, and implant component replacements) are associated with osteomyelitis in patients with transfemoral amputations treated with osseointegrated titanium implants.
Patients and Methods
Basic demographics at implant surgery
Number of patients (men/women)
Number of implants (bilateral implants)
Reasons for amputation: tumor/trauma/ischemia/infection/other
Time since amputation, mean (range)
11.5 (< 1–44) years
Age, mean (range)
43.5 (19–65) years
BMI, mean (range)*
26 (16-43) kg/m2
Number of smokers
Number of patients with diabetes (insulin dependent)
Residual limb lengths, short/normal/long
Definitions of osteomyelitis around the implant system
Type of infection
Signs and symptoms*
Positive tissue cultures
Definite implant infection
Probable implant infection
Possible implant infection
Furthermore, the total number of short courses of oral antibiotics for superficial infection in each patient was recorded. There was concern that pressure-induced bone marrow contamination could occur during the exchange of abutments and all exchanges therefore were recorded. Previously identified risk factors for impaired bone-implant healing and infection common in this cohort (ie, smoking, diabetes, [high] BMI, and age), were compiled from chart review and analyzed for association with osteomyelitis [4, 5, 19, 29, 32, 37].
Statistical end-points were first implant osteomyelitis and first implant extraction owing to infection. We used the Kaplan-Meier estimator to calculate the risk of osteomyelitis and extraction with time. Based on data at the time of implant insertion, risk factor correlation was performed with the Cox proportional hazard model. A hazard ratio (HR) for cumulative abutment replacements was obtained by time-modified Cox analysis. Differences were considered significant at a probability less than 0.05. Data were computed with the GraphPad Prism 6 (GraphPad Software Inc, San Diego, CA, USA) and SAS (SAS Institute Inc, Cary, NC, USA) statistical software.
Long-term Risk of Osteomyelitis
Clinical Effect of Osteomyelitis
Clinical outcome for patients classified as having osteomyelitis during the study period
Recovery after antibiotics with or without minor débridement (number of patients)
Recovery and later relapse (number of patients)
Successful reimplantation (number of patients)
Recovery after extraction (number of patients)
Chronic with fistula (number of patients)
Bacterial yield by intraoperative bone or marrow cultures in patients with osteomyelitis
Number of isolates*
Staphylococcus aureus, including one case of methicillin-resistant S aureus
Coagulase-negative staphylococci, including one case of Staphylococcus lugdunensis
Patient Factors Examined for Association With Osteomyelitis
With the numbers available, we found no association between selected patient factors and osteomyelitis. An increase in replacement of abutments was not related to implant infection (HR, 1.13; p = 0.16), and no risk increase was seen in patients who were overweight (BMI > 25 kg/m2) (HR, 0.99; 95% CI, 0.90–1.09; p = 0.97), elderly (HR, 0.98; 95% CI, 0.96–1.02; p = 0.75), or who smoked (HR, 1.8; 95% CI, 0.69–4.73; p = 0.22) during the time of implant surgery. Furthermore, there was no sex difference (HR, 0.92; 95% CI, 0.39–2.17; p = 0.85) or risk increase in patients with uncomplicated diabetes at the time of implant insertion (HR, 2.70; 95% CI, 0.78–9.39; p = 0.11).
In individuals with transfemoral amputations, especially with a short or malformed residual limb, socket prostheses cause substantial disadvantages for many including discomfort, skin problems, and poor function. These problems are reduced substantially when the prosthetic limb is attached to a percutaneous osseointegrated implant. However, the long-term risk of implant-associated infection and its clinical effect are not known with this method. In the first 96 patients treated with this novel method, we estimated that the cumulative risk of implant-associated osteomyelitis was 20% and the risk of implant extraction attributable to infection was 9% during a 10-year implant period. In four of the six patients for whom infection did not lead to extraction, prosthetic use was impaired. Resolution of osteomyelitis was attainable with prolonged combined antibiotics in one-quarter of the patients.
Our study has several limitations. Comorbidities associated with risk of poor healing and infection (arterial disease and complicated diabetes) are underrepresented in the patient cohort owing to preoperative selection. BMI, uncomplicated diabetes, and smoking were recorded only at the time of implantation, and no adjustments were made owing to changes in these variables with time which would only permit detection of a very strong relationship. The data were extracted retrospectively from medical charts by one author (JT) and reexamined by all authors before being added to the final dataset. We acknowledge the difficulty of defining method-typical osteomyelitis. Our definitions, although not validated, avoid excluding patients with culture-negative results with a high overall likelihood of osteomyelitis. Any overestimation is likely small as osteomyelitis was definite in 75% of the patients. Furthermore, the reliability of the method-specific tissue sampling is not known. Although less invasive than percutaneous bone tissue biopsies, bone marrow aspirations through the fixture possibly are more vulnerable to sample contamination, and it is uncertain if multiple samples add to diagnostic precision, as contaminants and true pathogens alike disperse in the liquid phase. Functional impairment could be graded only approximately owing to the 20-year retrospective span. Six patients were lost to followup including one patient with method-unrelated death, but all patients from the early phase (1990–1998) of the method development (custom-design period) have been included. Since then, several modifications have been made leading to standardized implants, surgical technique, and a strict rehabilitation protocol. This may result in reduced infectious complications. Outcome improvement also can be expected with standardized diagnostic protocols and early treatment with biofilm-effective antibiotics [6, 33, 36].
This method, intended for life-long prosthetic limb support, is potentially vulnerable to infection limiting its usefulness. However, we had no data with respect to very long-term risk of osteomyelitis and the clinical effects thereof. Previous outcome results not centering on infection indicate higher success rates with newer treatment protocols and greater surgical experience [8, 16]. In the current study, we found a high long-term risk of osteomyelitis in a young (mean age at first surgery, 43.5 years) patient group with good health compared with the average patient undergoing arthroplasty . This raises concerns regarding increasing risk of infection attributable to aging and related morbidities. Lower infection rates are needed before the indication can be widened. Currently, this treatment is only offered when socket prostheses are not an option, preexisting conditions likely to increase risk for failure are ruled out, and patients are made fully aware of the elevated risk of infectious complications. The method has similarities to percutaneous bone fixation and joint arthroplasty. In the latter, infection rates have decreased dramatically since its introduction , with a 10-year cumulative incidence of approximately 1.5% to 2%, using standardized infection control measures . Frequent but limited osteitis in external pin fixation  underscores how readily a percutaneous device promotes infection. Beyond this however, differences in surgical technique and tissue-material interaction do not allow for meaningful comparison. Some investigators aim for a microbial barrier by skin attachment to the percutaneous component [22, 35]. Conversely, our method aims to minimize skin mobility and secondary inflammation , in addition to allowing easy component replacements for the patient and surgeon. S aureus was twice as common as coagulase-negative staphylococci, likely because the skin stoma in this method favors the invasion-prone S aureus and not owing to a higher proportion of blood borne inoculation .
Full or moderate prosthetic use was maintained in more than half of the patients with infections not subjected to later implant extraction. Our interpretation is that load-bearing ossointegration can be preserved even in local osteomyelitis, justifying attempts of conservative treatment.
In a few patients, the infectious complications were acute, involving bone and marrow, indicating blood seeding from distant loci. That no patient had severe sepsis is important. However, a method resulting in frequent use of antibiotics contributes to bacterial resistance and increases adverse drug events. A strict policy in perioperative antibiotic prophylaxis and skin infection treatment will partly address this issue. Based on time of diagnosis, osteomyelitis develops later, and fewer implants are removed owing to early infection with this method compared with prosthetic joint infections [21, 28]. This likely stresses the importance of good primary osseointegration, translating to painless performance and possibly protection against early infectious development . The robustness of the osseointegration and antiinfective properties of the titanium (oxide) surface [2, 13] might explain the lack of infectious problems for many patients despite the intimacy between the skin microflora and the foreign material. However, it is conceivable that diagnostic delay in part explains the above, especially during rehabilitation, when diffuse pain indirectly related to increased prosthetic loading appears difficult to distinguish from pain caused by osteomyelitis. From these observations, deep infection should be suspected in rehabilitation that is delayed because of pain, even in absence of other signs of infection. Furthermore, low-grade infection can be very difficult to distinguish from aseptic loosening, much like in arthroplasties . Although not validated for our method, we suggest that diagnostic algorithms based on clinical signs, radiology, histopathology, and multiple tissue cultures [3, 33] be used to aid in decision making. Patients with distal osteitis were difficult to define as having implant-associated infection or not. This was partly because tissue culturing had been avoided owing to concern that it would interfere negatively with tissue integration, and partly because biomechanical bone wear was expected in this region. Frequent antibiotic use in this group however, suggests infectious bone degradation and supports appropriate culturing. It is not yet known whether distal osteitis progresses to osteomyelitis involving proximal parts of the implant system in the long term, but no such observation was made in the current study. Whether distal osteitis should be treated differently or at all therefore is not clear. Prospective investigations are warranted.
In our cohort (age range, 19–65 years), increased age did not appear to be a risk factor for infection, whereas time since implantation clearly was. We were concerned that exchanges of abutments after component failure might cause bacterial contamination, but we found no correlation between frequent abutment changes and infection. However, in four patients, there was a temporal relationship between abutment change and the start of infectious complications, which calls for prospective studies and strict aseptic conditions while performing this procedure which now have been implemented in our protocol. Only three patients had insulin-dependent diabetes, and none of them was severely obese. Therefore, it is not to be interpreted that diabetes is a negligible risk factor in this method. It might be expected that patients with a short residual limb are more prone to infection owing to the rich bacterial flora in the groin and a higher risk of implant instability and poor tissue integration, but our data do not support these assumptions.
The bone-anchored prosthesis using the osseointegration technique substantially improves quality of life after a transfemoral amputation . The key question is whether these advantages outweigh the risk of deep or recurring superficial implant-associated infection.
We showed that the overall risk of implant osteomyelitis in patients who receive percutaneous osseointegrated implants after transfemoral amputation increases with time. This is a major problem, as the method is intended to be a lifetime solution for prosthetic support. Infections which do not lead to implant removal only moderately reduce prosthetic function, and with more than 20 years’ experience with the method, we believe that improved daily living outweighs the risks and inconvenience of treatment for most patients  in this respect. As in all other medical interventions, it is important to inform patients regarding possible infectious complications and the risk of implant extraction if severe infectious complications occur. Will the surgical procedure, rehabilitation, and general care standardization since the start of the program result in lower infection rates? Prospective studies are warranted. Will improved diagnostics guiding early treatment and better handling of system components reduce infectious complications? Although not directly supported in this study, continuous bacterial invasion from the skin-implant interface as a cause for osteomyelitis cannot be ruled out. Further studies of method-specific diagnostics and treatment, bacterial properties, and characteristics of the skin-penetration area should be done to address these issues.
We thank Prof Peter Thomsen (Institution of Biomedicine, Department of Biomaterials); team coordinator nurse Eva Borg Pearce for assistance in retrieval of medical records; consultant statistician Mattias Molin BS (Statistiska Konsultgruppen, Gothenburg, Sweden) for support; and Cecilia Berlin PhD (Chalmers University of Technology, Gothenburg Sweden) for the schematic drawing of the implant system.
- 3.Atkins BL, Athanasou N, Deeks JJ, Crook DW, Simpson H, Peto TE, McLardy-Smith P, Berendt AR. Prospective evaluation of criteria for microbiological diagnosis of prosthetic-joint infection at revision arthroplasty: the OSIRIS Collaborative Study Group. J Clin Microbiol. 1998;36:2932–2939.PubMedPubMedCentralGoogle Scholar
- 12.Dorland’s Illustrated Medical Dictionary. 32nd ed. Philadelphia, PA: Elsevier; 2012.Google Scholar
- 29.Mellado-Valero A, Ferrer Garcia JC, Herrera Ballester A, Labaig Rueda C. Effects of diabetes on the osseointegration of dental implants. Med Oral Patol Oral Cir Bucal. 2007;12:E38–43.Google Scholar
- 31.Moojen DJ, van Hellemondt G, Vogely HC, Burger BJ, Walenkamp GH, Tulp NJ, Schreurs BW, de Meulemeester FR, Schot CS, van de Pol I, Fujishiro T, Schouls LM, Bauer TW, Dhert WJ. Incidence of low-grade infection in aseptic loosening of total hip arthroplasty. Acta Orthop. 2010;81:667–673.CrossRefPubMedPubMedCentralGoogle Scholar
- 33.Osmon DR, Berbari EF, Berendt AR, Lew D, Zimmerli W, Steckelberg JM, Rao N, Hanssen A, Wilson WR; Infectious Diseases Society of America. Diagnosis and management of prosthetic joint infection: clinical practice guidelines by the Infectious Diseases Society of America. Clin Infect Dis. 2013;56:e1–e25.CrossRefPubMedGoogle Scholar
- 41.Tsaras G, Osmon DR, Mabry T, Lahr B, St Sauveur J, Yawn B, Kurland R, Berbari EF. Incidence, secular trends, and outcomes of prosthetic joint infection: a population-based study, Olmsted county, Minnesota, 1969-2007. Infect Control Hosp Epidemiol. 2012;33:1207–1212.CrossRefPubMedPubMedCentralGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.