Poor Treatment Outcomes Following Admission for Advanced Lower Extremity Osteomyelitis in an Urban Setting

Introduction

Osteomyelitis is a progressive infection that results in inflammatory destruction and necrosis of the bone. Healthy bone is relatively resistant to infection, but certain conditions may pre-dispose individuals to osteomyelitis including diabetes mellitus, diabetic ulcers, surgery, trauma, and intravenous drug use. ¹ In recent years, the prevalence of osteomyelitis appears to have increased secondary to an increase in the prevalence of diabetic foot infections. ²

Early diagnosis of osteomyelitis is thought to improve prognosis but is difficult to make. There are several methods for the detection of osteomyelitis including bedside probe-to-bone test, plain bone radiography, bone gallium scan, computed tomography (CT) scan, magnetic resonance imaging (MRI), and bone biopsy, with MRI thought to be most accurate. ^{2 3} Positive culture upon bone biopsy is often considered the gold standard for confirmation of osteomyelitis; however, bone biopsies are not routinely performed at all institutions and may not always be positive due to sampling issues. ⁴

Osteomyelitis remains challenging to treat despite advances in medical and surgical treatment. Antibiotic treatment is the mainstay for the treatment of acute osteomyelitis; however, if an infection becomes chronic, the infective nidus harbors matrix-protected organisms and can become impermeable to antibiotics. ⁵ Therefore, surgical intervention and radical debridement are often essential for treatment, with antibiotics playing an adjunctive role. In patients with infection in bones that are critical for stability, amputation or chronic suppressive therapy may be the only remaining course of treatment. ⁵

Some studies suggest that treatment with antibiotics alone in the absence of surgical debridement is more likely to result in treatment failure and the rate of recurrence, especially in diabetic patients, is high. ^{6 7} In contrast, other studies have found rates of cure approaching 80% with antibiotic treatment alone. ⁸ This study was undertaken to evaluate treatment outcomes in patients admitted to an urban tertiary care hospital for osteomyelitis based on the initial treatment received: antibiotic therapy alone, antibiotic therapy and surgical debridement, or amputation. We found that resolution was associated with amputation, likely due to a preponderance of underlying comorbidities and advanced presentation of disease in this patient population.

Methods

A retrospective analysis of patients with a discharge diagnosis of osteomyelitis between January 1, 2006 and December 31, 2007 at St. Barnabas Hospital in Bronx, New York was performed following approval by the institutional review board. Patients were included if they had a diagnosis of lower extremity osteomyelitis by positive bone radiograph, bone gallium scan, probe-to-bone test, or MRI, and a follow-up period of at least 6 months. Pediatric patients and patients with a prosthetic device were excluded. Patients were grouped based on the treatment received per the treating physician’s initial impression on first admission for osteomyelitis. Group 1 included those who had received a course of antibiotics only; Group 2 included those who had received antibiotics and surgical debridement; and Group 3 included those who had undergone amputation. The infection was considered resolved at last follow-up if the ulcer was healed and the patient was no longer receiving treatment for osteomyelitis. ⁹ A Chi-squared test was employed to test for differences in the proportion of osteomyelitic patients experiencing resolution in each treatment group.

Results

Forty-four patients were admitted to St. Barnabas Hospital with osteomyelitis during the 2-year period between January 2006 and December 2007. The average follow-up time from initial osteomyelitis diagnosis was 31 months (range 6 to 74 months). Twenty-seven patients (61%) were male, and the average age was 57 years (range 32 to 87 years). Seven patients (16%) were from nursing homes, and seven patients (16%) were bed-bound. Underlying conditions were common amongst study subjects; more than half were hypertensive, one-third had peripheral vascular disease, and more than three-fourths were diabetic (Table 1). The nine remaining non-diabetic patients had other risk factors for osteomyelitis including burn, rhabdomyolysis, paraplegia, bed-bound, active alcohol user with stroke, and one patient with human immunodeficiency virus and active drug use. Half of the study subjects had a prior history of osteomyelitis in a different area.

The diagnosis of osteomyelitis was made in several different ways including plain bone radiograph, bone gallium scan, probe-to-bone test, and MRI. Many patients had more than one positive test with bone radiography being the most common modality (Table 2). Nine of 44 patients had a negative bone radiograph. Of these, two had a positive MRI, five had a positive bone gallium scan, and three had a positive probe-to-bone test. Following diagnosis, patients were treated per the treating physician’s initial impression. Based on chart review, patients were assigned to one of three treatment groups. Group 1 (13 patients) had been treated with antibiotics alone. Most patients were treated empirically with intravenous antibiotics that were adjusted based on culture results if available. Group 2 (7 patients) had been treated similarly with antibiotics and had also undergone surgical debridement. Group 3, the largest group (24 patients), had undergone amputation (Table 3). Of the 24 patients who required amputation on first admission for osteomyelitis, 67% presented with gangrene. Sixteen patients underwent amputation of a digit, four patients had transmetatarsal amputations, and two patients had above-knee amputations. Of the 24 patients who underwent initial amputation, 10 (42%) had resolution of infection at follow-up (Table 3). In contrast, only 3 of 13 (23%) patients treated with antibiotics alone, and 2 of 7 (29%) patients treated with antibiotics and surgical debridement had resolution of infection at follow-up (Table 3). There was no statistically significant difference in treatment outcomes between the three groups or between initial amputation versus no amputation by Chi-squared analysis. At last follow-up (mean of 31 months), 5 patients (11%) experienced resolution without amputation, 8 (18%) were participating in ongoing treatment or suppression, and 31 (71%) had undergone amputation (Figure 1). Amputation was significantly associated with resolution at final follow-up by Chi-squared analysis (P <0.001).

The majority of bones involved were the toes and/or bones in the body of the foot. Two patients had involvement of the tibia/fibula. One was secondary to a burn followed by ulcer formation and osteomyelitis. The other was secondary to infected muscle tissue following an episode of myonecrosis. Both patients with tibia/fibula involvement were treated with antibiotics alone. One patient had no further evidence of infection and the other required multiple courses of antibiotics for recurring osteomyelitis. Bone involvement for the remaining 42 patients is shown in Table 4. Of 28 patients with involvement of the toes, 16 (57%) underwent amputation on first admission. Of nine patients with involvement of the body of the foot, three (33%) underwent amputation on first admission. All five patients who had involvement of both the toes and body of the foot underwent amputation upon first admission.

The organisms responsible for causing osteomyelitis were not always determined. Twelve (27%) patients had no or negative cultures. One organism was isolated from 14 (32%) patients, and in half of these patients the cultured organism was Staphylococcus aureus. Multiple organisms were cultured from 18 (41%) patients, with 11 (61%) having S. aureus as one of the organisms. The bacterium most frequently isolated with S. aureus was P. aeruginosa, which was present in 7 of 11 cultures.

The majority of patients experienced an ulcer for a protracted period of time before the admission for osteomyelitis, with 30 (68%) patients reporting the presence of ulcers for longer than one month. Thirteen (29%) patients had ulcers present for up to four weeks before admission. No information on the length of presence of the ulcer was available for one patient.

Discussion

At St. Barnabas Hospital, patients treated for osteomyelitis by amputation upon first admission were more likely to experience resolution than patients treated with antibiotics alone or treated with antibiotics and surgical debridement, although the difference was not statistically significant. Several factors suggest that this particular study population was at risk for poor prognosis. At the time of admission 79% of patients with osteomyelitis had a positive radiograph, suggesting that the osteomyelitis was present for at least 2 to 4 weeks. ^{10 11 12} While antibiotics are a mainstay of treatment for acute osteomyelitis, ⁵ the positive radiographic findings suggest that most cases were either sub-acute or chronic at the time of presentation. ^{10 11 12} Furthermore, the vast majority of patients had underlying conditions including diabetes, peripheral vascular disease, and hypertension, which are known to contribute to poor prognosis upon treatment for osteomyelitis. ¹³ Malizos et al. ¹³ observed that osteomyelitic patients with comorbidities had higher complication and recurrence rates, required more operative procedures, and had higher amputation rates than those without comorbidities. In the present study, poor prognosis due to advanced disease and underlying host conditions may account for the high rate of initial amputation as well as the finding that initial amputation resulted in the highest cure rate. Additionally, at final follow-up, only 11% of patients resolved the infection without amputation, again speaking to the advanced nature of disease in this study population.

Osteomyelitis is especially common in individuals with diabetes and usually occurs secondary to ulceration of the foot. It is thought that 3% to 4% of individuals with diabetes have foot ulcers or deep infections, and 15% of diabetic individuals develop foot ulcers in their lifetime. ¹⁴ Studies suggest that as many as 20% to 60% of diabetic patients with a foot wound will have an infection involving the underlying bone. ¹⁵ Ramsey et al ¹⁶ estimated that as many as 16% of diabetic patients with osteomyelitis will require amputation, and as many as 45% to 60% of non-traumatic lower extremity amputations occur in diabetic patients. ² Consistent with these findings, in the present study more than three-fourths of patients with osteomyelitis of the lower extremities were diabetic; however, the rate of amputation in the present study was much greater than that reported by Ramsey et al. ¹⁶ Foot complications are well-known as the most serious and costly complications of diabetes mellitus.. ¹⁷ The bones of the toes were involved in 67% of the patients and the plantar bones in 21%. In 12% of patients, both the toes and plantar bones were involved, all of whom underwent amputation. Two patients had involvement of the tibia/fibula, and bone involvement was consistent with disease etiology. In the two patients with tibia/fibula involvement, osteomyelitis was secondary to a burn and an episode of myonecrosis. The remaining patients experienced involvement of the toes and/or bones of the foot, consistent with the large percentage of patients with peripheral vascular disease and/or diabetes.

To date, no randomized controlled trials of treatment regimens for osteomyelitis have been performed. Such trials would be complicated by the heterogeneity of osteomyelitis in terms of etiology, bone involvement, advancement of disease, and underlying host conditions. ^{1 5 18} Additionally, many centers do not recommend surgery until non-surgical options have failed, and in such centers, randomization to early surgery would be difficult to justify. ⁴ At present, treatment is guided largely by physician experience and expertise. Despite the absence of definitive treatment guidelines, several studies have attempted to determine the cure rate for various treatment protocols. Some studies report a cure rate for osteomyelitis treated with antibiotics and some “limited” debridement of 53% to 88%; however, inconsistency in the definition of “cured” and “limited” complicate the comparison of these studies. ^{19 20 21 22 23 24} Arágon-Sánchez et al ²⁵ are of the opinion that antibiotics alone are not sufficient for the cure of osteomyelitis in most circumstances. They advocate a conservative surgical approach whereby patients are first treated empirically with antibiotics, then, based on culture and susceptibility testing results, infected soft tissue and bone is removed surgically. ²⁵ Such an approach resulted in cure without amputation in nearly half of diabetic patients with foot ulcer and bone involvement. The other 50% of their patients did require some form of amputation for cure. ²⁵ While fewer patients in our study experienced resolution without amputation, the population of diabetic patients with foot ulcers in the study by Arágon-Sánchez et al ²⁵ does not lend itself well to direct comparison with our population in light of the particularly advanced nature of disease in the present study. Furthermore, many studies evaluating the utility of antibiotic therapy alone exclude patients with gangrene or other indications of severe disease, ²⁶ making generalizability to the population at hand difficult.

We did not observe an association between the duration of ulcers, underlying medical conditions, length of antibiotic therapy, age, or the initial management of osteomyelitis (antibiotics alone, antibiotics with debridement, or amputation) with outcome. This could be due to the small sample size and to the small number of patients that did not eventually require amputation (5 patients). However, we did observe a significant association between amputation and resolution at final follow-up.

Several limitations of this study must be acknowledged, most notably those associated with the retrospective nature of the study. Because the study was retrospective and based on chart review, we were unable to question diagnoses or treatment decisions and were limited by the decisions of the treating physician. Additionally, the sample size was small and few patients were treated without eventual amputation, limiting comparisons made between groups. Finally, the high rate of comorbidities and advanced disease presentation in our patient population likely accounted for poor outcomes in the absence of amputation. While this limits the generalizability of our findings to other patient populations, it underscores the challenges in developing standardized treatment guidelines for osteomyelitis through randomized controlled trials. The patient population examined here has limited access to health care and a high rate of underlying conditions, which must influence physician treatment decisions to some extent.

At St. Barnabas Hospital in Bronx, New York having just one admission for treatment of osteomyelitis predicts a poor prognosis. Only 5 (11%) patients had infection resolution without amputation; 24 patients (54%) required amputation during first admission and an additional 12 patients had amputations before final follow-up. Overall, 31 (71%) patients underwent amputation and 16% of patients were receiving ongoing treatment or suppression at final follow-up. The data presented suggest that patients in this study population had advanced disease based on the high percentage of patients with gangrene (41%) and positive bone radiography (80%) upon admission. Furthermore, the vast majority of patients had underlying conditions including diabetes, peripheral vascular disease, and hypertension, which are associated with a poor prognosis in osteomyelitis. ¹³ Taken together, these results suggest that in a highly at-risk, urban population with advanced disease, osteomyelitis may have a poor prognosis. In light of these findings, it seems unlikely that all institutions can be expected to conform to a single set of standard treatment guidelines, regardless of the outcomes of randomized clinical trials, but rather will always depend to at least some extent on population-specific attributes and physician experience.

Acknowledgments

The authors thank the Marshfield Clinic Research Foundation Office of Scientific Writing and Publication for assistance in the preparation of this manuscript.