Back to Annual Meeting Program
A Multivariate Logistic Regression Analysis of Complications Following Microsurgical Breast Reconstruction
Samir Rao, MD1, Ellen C. Stolle, BS1, Sara Sher, MD1, Lin Chun Wang, BS1, Bahram Momen, PhD2, Maurice Y. Nahabedian, MD1.
1Georgetown University Hospital, Washington, DC, USA, 2Univeristy of Maryland, College Park, MD, USA.
PURPOSE: Complications following microsurgical breast reconstruction have been examined for years. Most studies examining these complications are univariate analyses focusing on a specific risk factor and its impact on outcomes. A multivariate analysis takes into account several predictive variables simultaneously, offering the potential to model the risk factors and complications with more accuracy.
METHODS: We performed a retrospective review of all patients who underwent microsurgical breast reconstructions by the senior author (M.Y.N) between July 2005-July 2010. Patient records were analyzed for risk factors (age, BMI, smoking history, medical history, adjunct therapies, timing of reconstruction, type of reconstruction), and complications (hematoma, seroma, infection, wound dehiscence, PE, DVT, pneumonia, fat necrosis, leech use, partial flap loss, total flap loss.)
We then used the LOGISTIC Procedure of the SAS System (SAS 9.2, Carry, North Carolina) to relate the complications (dependent variables) and risk factors (independent variables). We used the STEPWISE option of the LOGISTIC procedure to select the explanatory variables with significant effects. The FREQ procedure of the SAS System was used to generate contingency tables and exact p-values if a cell within a contingency table was smaller than 5.
RESULTS: During the study perioid, 352 patients underwent 490 microvascular breast reconstructions. Average patient age was 49.6 years (range 27-74 years) and average BMI was 27.7 (range 19.0-47.0). The risk factor data is presented in Table 1 and complications data is presented in Table 2. Statisitically significant associations were identified between the risk factors and complications listed in Table 3.
Table 1. Risk factors & Reconstruction data (n=490)
|Connective Tissue Disease||10|
|Timing of Reconstruction|
|Side of Reconstruction|
|1 perforator DIEP||223|
|2 perforator DIEP||86|
|3+ perforator DIEP||12|
Table 2. Complications data (n=490)
|Complication||# of occurences||%|
|Partial flap loss||4||0.8%|
|Total flap loss||22||4.5%|
Table 3. Statistically significant associations
|Risk Factor||Complication||p-value||Odds ratio||95% Confidence Interval|
|Unilateral Reconstruction||Fat Necrosis||0.0083||4||1.4-11.4|
BMI is a continuous variable with a range of values, so odds ratios could not be calculated. However, a positive coefficient of 0.1617 for BMI indicates that as BMI increases, the chance of infection increases (p < .0001).
No other risk factors or reconstructive parameters listed in Table 1 were found to have statistically significant associations with complications listed in Table 2.
CONCLUSIONS: This study identified a statistically significant increase in seroma, infection, and pneumonia in active smokers undergoing microsurgical breast reconstructions. Additionally, fat necrosis was found to be more likely in unilateral reconstructions than bilateral reconstructions, likely from the use of Zone 3 and sometimes Zone 4 of the abdominal flaps, whereas bilateral reconstructions are restricted to a hemi-abdomen. Finally, the chance of infection was found to increase as BMI increases.
Back to Annual Meeting Program