Table 5.
Adjusted Logistic Regression Analysis of Risk Factors Associated with Osteomyelitis in Wounded Military Personnel with Open Tibia Fractures (Odd Ratios; 95% Confidence Intervals)
| Factors | Full Population | Restricted Population:Excludes TTAs1 | ||
|---|---|---|---|---|
| GA Class Model2 | OTA OFC Model2 | GA Class Model2 | OTA OFC Model2 | |
| Time Period | ||||
| 2003–2006 | – | – | 4.09 (1.77–9.46) | 2.58 (1.05–6.32) |
| 2007–2009 | – | – | Reference | Reference |
| Blast injury mechanism | 2.38 (1.14–4.99) | – | 3.21 (1.37–7.52) | 2.68 (1.06–6.79) |
| Fracture classification3 | ||||
| GA-II | Reference | – | Reference | – |
| GA-IIIa | 3.06 (0.89–10.54)4 | – | 2.66 (0.75–9.39)4 | – |
| GA-IIIb | 4.81 (1.33–17.40) | – | 5.19 (1.39–19.38) | – |
| GA-IIIc | 5.68 (1.26–25.60) | – | 4.08 (0.90–18.50)4 | – |
| TTA | 15.10 (3.22–71.07) | – | NA | – |
| Use of antibiotic beads | 3.07 (1.59–5.94) | 2.86 (1.44–5.69) | 4.12 (1.86–9.11) | 3.19 (1.33–7.66) |
| OTA OFC: Muscle,5 | ||||
| Grade I | – | Reference | – | Reference |
| Grade II | – | 5.62 (2.21–14.25) | – | 3.99 (1.50–10.59) |
| Grade III | – | 8.46 (3.31–21.64) | – | 8.08 (2.16–30.20) |
| Type of foreign body at tibia site | ||||
| No foreign body | Reference | – | – | – |
| Fragment with hardware | 5.65 (1.36–23.38) | – | – | – |
| Fragment only | 4.04 (1.15–14.21) | – | – | – |
| Hardware only | 3.14 (0.83–11.90)4 | – | – | – |
TTA – transtibial amputation; GA – Gustilo-Anderson; OTA OFC – Orthopaedic Trauma Association’s Open Fracture Classification
Analysis excludes TTAs (population: 89 osteomyelitis cases and 72 controls)
GA fracture classification and the OTA OFC variables were highly correlated. The multivariable analysis was run separately after selecting for either the GA fracture classification (GA Class Model) or OTA OFC (OTA OFC Model) to be assessed in the model. Stepwise, backward, and forward model selections were conducted to choose the final multivariable risk factor model.
Using a modified GA classification of open fractures.28 There were no GA-I fractures. TTAs were prior to infection.
P-values for variables that were not significant in the model of the full population are: 0.077 for GA-IIIa and 0.092 for hardware. P-values for variables that were not significant in the model of the restricted population are: 0.129 for GA-IIIa and 0.069 for GA-IIIc.
Grade I - no muscle in area, no appreciable muscle necrosis, some muscle injury with intact function; Grade II - loss of muscle but function remains, some localized necrosis requiring excision, intact muscle-tendon unit; Grade III - dead muscle, loss of function, partial/complete excision, complete disruption of unit, defect does not approximated.