Immediate breast reconstruction rates have risen dramatically in the United States during the past decade.1 Despite studies demonstrating higher complication rates, immediate reconstruction has been demonstrated to be safe and cost-effective when compared with delayed reconstruction.2–4 Patients with diabetes mellitus are increasing in prevalence and pose a significant challenge for surgery.5 Diabetes mellitus can increase the risk of infection and delay wound healing, adding to postoperative morbidity.6–8 Among breast reconstruction patients, diabetics experience a higher incidence of overall complications compared with nondiabetics, particularly with autologous reconstruction.9,10 However, the literature lacks studies examining how the timing of breast reconstruction affects postoperative outcomes in the diabetic population.
Early postoperative outcomes are becoming increasingly relevant because of their association with health care costs, health care quality, Medicare reimbursements, and patient satisfaction.11,12 The American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database is a large, prospective national data set encompassing data from over 200 hospitals. The limitations of the NSQIP have been documented, including follow-up limited to 30 days and lack of pertinent outcomes for breast reconstruction (ie, fat necrosis, hematoma, capsular contracture).9 However, to our knowledge, no one has scrutinized the NSQIP database by analyzing it alongside a single institution’s experience of specific breast reconstruction cohorts.
This study aimed to determine the effect of timing on outcomes in diabetic patients undergoing breast reconstruction using a large multi-institutional database, the NSQIP, to identify the optimal timing for least postoperative morbidity. A second objective was to validate the NSQIP database for breast reconstruction outcomes using a comparison with results from The Johns Hopkins Hospital (JHH) experience.
In the first part of this study, we performed a retrospective analysis of patients in the NSQIP database. Deidentified patient data files from January 1, 2005, to December 31, 2012, were reviewed on all female diabetics undergoing breast reconstruction. Diabetic patients undergoing breast reconstruction with and without mastectomy were identified using the following Current Procedural Terminology (CPT) codes: 19340, 19342, and 19357 for prosthetic breast reconstructions; 19361, 19364, 19367, 19368, and 19369 for autologous breast reconstructions; and 19160, 19162, 19180, 19182, 19200, 19240, 19260, 19301, 19302, 19303, 19304, 19305, 19306, and 19307 for mastectomies. Immediate reconstruction included all events in which a reconstruction CPT code accompanied a mastectomy CPT code in the same encounter. Delayed reconstruction included all events in which a reconstruction CPT code was found without a concurrent mastectomy CPT code.
The NSQIP-defined demographic and preoperative variables were compared between the immediate and delayed reconstruction groups. They included demographic information (race, age, admission year, and body mass index class), preoperative risk factors (operation time, smoking status, inpatient status, alcohol intake, wound infection, corticosteroid use, and operation within the previous 30 days), and comorbidities (cardiovascular, respiratory, hepatobiliary, renal, neurologic, and hemato-oncologic). Outcomes were categorized as overall complications, surgical complications, medical complications, reconstruction failure, and reoperation. Surgical complications were defined as superficial infection, deep infection, organ space infection, wound dehiscence, and flap/prosthesis failure. Medical complications included pneumonia, unplanned reintubation, pulmonary embolism, ventilator dependence for greater than 48 hours, renal insufficiency, acute renal failure, urinary tract infection, stroke, coma, cardiac arrest, and myocardial infarction. Overall complications were defined as a sum of all surgical and medical complications.
Stata version 13.1 (StataCorp LP, College Station, Tex.) was used to analyze differences in the immediate and delayed cohorts. Significance was defined as P < 0.05. As a preliminary analysis, associations between variables and outcomes were examined using crosstabulation and unadjusted univariate analysis to inspect for missing data and data errors and to identify potential confounders and effect modifiers. Chi-square test and Fisher’s exact test were used, as appropriate, to compare categorical outcomes. t Test and Wilcoxon rank-sum test were used to compare continuous outcomes. A multivariable logistic regression model was fitted to estimate the association between timing of breast reconstruction and 30-day overall complications. Variables were selected using clinical relevance and change-in-estimate methods and included in a stepwise forward manner.13 Likelihood ratio tests were used to examine effect modification and departures from linearity.
In the second part of this study, we performed a retrospective analysis of all diabetic patients at JHH undergoing immediate and delayed breast reconstruction. Institutional Review Board approval was sought and received. Electronic medical records for all patients that had undergone breast reconstruction at our institution from January 1, 2005, to July 31, 2014, were hand-searched to identify those with a diagnosis of diabetes and taking oral medication or insulin. Patients were included in the immediate reconstruction cohort if they received any sort of reconstructive modality (prosthetic or autologous) in the same operation as their mastectomy and in the delayed reconstruction cohort if they received any sort of reconstructive modality (prosthetic or autologous) on a separate operation from their mastectomy. Patients undergoing a 2-staged reconstruction (ie, those who received a tissue expander on the day of their mastectomy and a final prosthetic or autologous reconstruction at a later date) were included in both immediate and delayed groups. This was done to ensure that our data were comparable to those available in the NSQIP.
Demographic and preoperative variables collected from JHH patient medical records mirrored data available in the NSQIP. Outcomes were similarly categorized as overall complications, surgical complications, medical complications, reconstruction failure, and reoperation. However, surgical complications included all those specified in the NSQIP in addition to seroma, fat necrosis, hematoma, mastectomy skin necrosis, and abdominal hernia. Differences in demographics, preoperative risk factors, and outcomes between immediate and delayed reconstructions were compared.
Finally, overall morbidity, surgical morbidity, medical morbidity, reconstruction failure, and reoperation rates were compared between the NSQIP and JHH diabetic cohorts.
NSQIP Patient Demographics and Outcomes
A total of 1,408 patients met the inclusion criteria, of which 958 (68%) received immediate reconstructions and 450 (32%) received delayed reconstructions (Fig. 1). Within the immediate reconstruction cohort, 802 (84%) received prosthetic reconstructions and 156 (16%) received autologous reconstructions. Within the delayed reconstruction cohort, 332 (74%) received prosthetic reconstructions and 118 (26%) received autologous reconstructions. Demographic information for the NSQIP database population is presented in Table 1. The demographic information and preoperative risk factors of the patients in the immediate and delayed reconstruction cohorts are compared in Table 2. Significant differences existed among the immediate and delayed reconstruction groups in median operation time (138.5 ± 161 min vs. 205 ± 116 min; P < 0.001) and inpatient status (78.91% vs. 44.22%; P < 0.001).
The unadjusted postoperative outcomes for the immediate and delayed reconstruction cohorts are presented in Table 3. Reconstruction failure rates (1.57% vs. 0.22%; P = 0.027) were significantly higher in the immediate reconstruction group, whereas the rate of acute renal failure (0% vs. 0.44%; P = 0.039) was significantly higher in the delayed group. Rates of overall morbidity (10.75% vs. 7.78%; P = 0.080) and surgical morbidity (10.02% vs. 6.89%; P = 0.056) were higher in the immediate group, although these only trended toward statistical significance.
We fitted a multivariable logistic regression model to adjust for age, smoking status, preoperative anemia, inpatient status, type of reconstruction, and operation year. We tested this model for departures from linearity using likelihood ratio tests, showing no statistical evidence for such departures. Likewise, we tested for effect modification from variables identified with Mantel-Haenszel analysis as potential effect modifiers, finding no evidence for effect modification from type of reconstruction (P = 0.76), smoking status (P = 0.26), or preoperative anemia (P = 0.32).
Unadjusted and adjusted odds ratios (ORs) from this model are presented in Table 4. Notably, diabetic women undergoing immediate reconstructions had significantly higher odds of 30-day overall complications, as compared with the delayed reconstruction group [adjusted OR = 1.68; 95% confidence interval (CI) = 1.04–2.72, P = 0.033]. Additionally, autologous reconstructions were associated with significantly higher 30-day overall complications than prosthetic reconstructions (adjusted OR = 2.50; 95% CI = 1.63–3.85; P < 0.001).
Immediate reconstruction among diabetics was also associated with significantly higher 30-day surgical complications compared with delayed reconstructions (adjusted OR = 1.75, 95% CI = 1.05–2.9, P = 0.029). In contrast, no statistically significant association was found for medical complications between the 2 groups (adjusted OR = 0.93, 95% CI = 0.25–3.42, P = 0.914).
The odds of reconstruction failure within immediate reconstruction were significantly higher compared with delayed reconstruction within diabetics (unadjusted OR = 7.14; 95% CI = 0.94–54.47; P = 0.027). After adjustment for confounders, immediate reconstruction had higher odds of reconstruction failure than delayed, but without statistical significance (adjusted OR = 5.61; 95% CI = 0.67–47.25; P = 0.112). There was no significant difference in reconstruction failures between reconstruction modalities. We also did not find any statistically significant differences between timing and reconstructive modality in regards to reoperation rates.
JHH Patient Demographics and Outcomes
A total of 52 documented diabetic women underwent 114 breast reconstructions at JHH between January 1, 2005 and July 31, 2014. Thirty-nine patients underwent 19 unilateral and 20 bilateral immediate reconstructions (59 total immediate reconstructions) and 36 patients underwent 17 unilateral and 19 bilateral delayed reconstructions (55 total delayed reconstructions). The majority of patients underwent 2-staged reconstruction using tissue expanders in the first stage followed by either implants or autologous reconstruction (n = 23). General demographic information of diabetic women undergoing breast reconstruction at JHH is presented in Table 1. The demographic information and preoperative risk factors of the patients in the immediate and delayed reconstruction cohorts were not significantly different (Table 5).
The unadjusted postoperative outcomes for the immediate and delayed reconstruction cohorts are presented in Table 6. At 30-day follow-up, reoperation rates were significantly higher in the immediate cohort (12.8% vs. 0%; P = 0.028); however, the rates of overall and surgical morbidity did not significantly differ between the 2 groups. The rates of reoperation at the 60-day postoperative time point remained significantly different between the 2 cohorts (15.4% vs. 0%; P = 0.014). On long-term follow-up (median = 16.5 months, range = 0.5–150 months), the rates of reconstruction failure (25.6% vs. 0%; P = 0.001) and reoperation (30.8% vs. 2.8%; P = 0.001) differed significantly between the 2 cohorts. At long-term follow-up, all reconstruction failures and reoperations occurred within the prosthetic group. The overall, surgical, and medical morbidity was not statistically significant between the 2 groups at the 60-day postoperative time point and on long-term follow-up.
Comparison of NSQIP and JHH Outcomes
A comparison of the 30-day postoperative outcomes using data from the immediate reconstruction cohorts in the NSQIP and JHH differed significantly. Compared with the NSQIP, JHH data using only the outcome variables available in the NSQIP demonstrated a significantly higher percentage of medical morbidities (P = 0.008), reconstruction failures (P = 0.005), and reoperations (P = 0.005). Comparing the NSQIP with JHH 30-day morbidity and JHH long-term morbidity data yielded statistically significant (P < 0.05) differences in overall morbidity, surgical morbidity, medical morbidity, reconstruction failure, and reoperation rates. In the delayed cohort, a comparison of 30-day postoperative outcomes in the NSQIP and JHH data using only the outcome variables available in the NSQIP did not yield any significant differences in morbidity. This trend continued when the NSQIP data were compared with JHH 30-day morbidity; however, when 30-day postoperative outcomes in the NSQIP were compared with JHH long-term follow-up data, overall morbidity (P < 0.001) and surgical morbidity (P < 0.001) were significantly higher in the JHH patient population.
Our analysis of diabetic females in the NSQIP undergoing breast reconstruction demonstrates that those who undergo an immediate reconstruction have significantly higher odds of developing 30-day overall and surgical complications compared with those undergoing delayed reconstruction. Autologous reconstructions were associated with higher overall, surgical, and medical complications compared with prosthetic reconstructions. However, the odds of reconstruction failure and reoperation at 30 days postoperatively did not differ based on breast reconstruction timing or modality within the NSQIP data set. Analysis of JHH data compared with the NSQIP data showed significant discrepancies between the 2 data sources.
Previous studies have established that diabetic patients are at a higher risk of developing complications after surgery,14,15 and perioperative hyperglycemia may be the determining factor.16,17 High blood glucose levels negatively affect the mobilization and phagocytotic ability of neutrophils and monocytes, predisposing to infection.18,19 Diabetics may also have aberrant molecular and cellular signaling pathways, affecting keratinocyte and fibroblast migration, extracellular matrix remodeling, and angiogenesis, and thus causing wound-healing deficiencies.20–22 In addition, hyperglycemia increases inflammatory cytokines and reactive oxygen intermediates, resulting in endothelial cell dysfunction and a prothrombotic environment.23–25 Given these known hematological and vascular abnormalities, optimizing the surgical course of diabetic patients is especially important.
Recent analysis of diabetes with respect to postoperative complications after breast reconstruction has yielded mixed results. In a single-center study of 893 patients undergoing autologous reconstruction, Miller et al found no significant difference in flap and donor-site complications between diabetics and nondiabetics.26 Fischer et al analyzed the NSQIP database from 2005 to 2010 and showed increased postoperative surgical complications within diabetics, but found no statistical significance when adjusted for other confounding variables.2 They also found immediate reconstruction to be independently associated with increased postoperative surgical complications. Qin et al studied the NSQIP database from 2005 to 2012 and found that both insulin-dependent and non-insulin-dependent diabetics were associated with significantly higher surgical and overall complications after autologous reconstruction than nondiabetics.9 However, the study only examined immediate reconstructions. Our analysis corroborated previous findings of autologous-based reconstructions with increased overall and surgical morbidity in the early postoperative period and also elucidated that immediate reconstruction is associated with increased risk of early overall and surgical morbidity among diabetics.
Autologous and prosthetic reconstructions predispose patients to distinct postoperative complications. However, our multivariable model established immediate timing as an independent risk factor for overall and surgical morbidity among diabetics even after controlling for reconstruction modality. An explanation for this association is likely multifactorial, and may include contamination of the surgical field during mastectomy, poor skin flap viability, and increased inflammation due to the mastectomy. These problems are likely to be exacerbated in diabetic patients who are more susceptible to infection.3 Our NSQIP analysis identified superficial and deep surgical site infection as the predominant early surgical complications among the diabetic cohort. Additionally, postoperative radiation after immediate reconstruction has previously been shown to increase early surgical site complications,27 which may compound the already compromised wound-healing abilities in diabetic patients.
Although the NSQIP permitted high-powered statistical analysis, the database has several limitations that prompted our evaluation of the JHH cohort. The chief limitations include the limited follow-up period of 30 days, no delineation of unilateral versus bilateral reconstructions, and omission of pertinent complications after breast reconstruction procedures. In addition, inconsistent reporting of prior radiation exposure prevented us from adjusting for it in our multivariable model. Although these limitations have been cited previously, there has been no evaluation of their importance by comparing the NSQIP data with the institutional data.
When comparing results from JHH data with the NSQIP on immediate reconstructions, we found significantly higher 30-day medical morbidity, reconstruction failure, and reoperation rates than those captured in the NSQIP even after restricting these data to the variables available in the NSQIP. Furthermore, these differences were amplified when taking into account additional variables in the JHH chart review and when comparing with 60-day and long-term follow-up complication rates. This was likely due to our inclusion of additional surgical complications (ie, seroma, fat necrosis, hematoma, mastectomy skin necrosis, and capsular contracture) unavailable in the NSQIP. In contrast, our analysis on delayed reconstruction data from the 2 data sources found that differences only became significant when comparing long-term follow-up rates from JHH data with 30-day NSQIP rates. This second analysis was underpowered as compared with the one on immediate reconstructions, potentially explaining the fewer differences detected.
The findings from our JHH patients at 60-day and long-term follow-up indicate that a substantial number of diabetic patients undergoing breast reconstruction develop postoperative complications outside of the 30-day window captured by the NSQIP. A significant number of patients continue to develop complications even after the 60-day follow-up, particularly in implant-based reconstructions. Most of the complications that occurred after the 60-day time point were associated with infection and reconstruction failure. Notably, the number of deep infections in the immediate group tripled from the 60-day time point to long-term follow-up, flap/prosthetic failures quadrupled, and seromas nearly doubled. Reconstruction failure rates at long-term follow-up tripled when compared with the 60-day time point and reoperation rates doubled. It is noted that the low number of patients in the JHH diabetic cohort poses a limitation to draw powerful statistic conclusions.
Another notable limitation of using the NSQIP database for breast reconstruction studies is the inability to differentiate 2-staged reconstructions (ie, delayed immediate) from purely immediate or delayed reconstructions. As the popularity of a 2-stage reconstructive procedure increases, particularly in patients receiving postmastectomy radiotherapy, it is becoming increasingly clear that the complication profiles of these procedures may be different from patients receiving an immediate reconstruction with permanent implants or flaps, or a truly delayed reconstruction without any involvement from a plastic surgeon at the time of mastectomy. Therefore, staged procedures must be evaluated as a separate cohort. The JHH portion of this study classified staged reconstruction patients in the immediate and delayed cohorts, which could limit our interpretation of outcomes due to potential within-patient variations that were not taken into account in the analysis due to a small sample size. However, even though the institutional practice of JHH favors the two-staged procedure, a separate group was not delineated for staged procedures in an effort to mirror the NSQIP database.
Using a large, multicenter database, immediate breast reconstruction was found to be associated with higher risk of early postoperative complications in diabetic women. Additionally, autologous breast reconstruction is independently associated with higher risk of complications in this population, which is in agreement with previous studies. Our evidence suggests that performing delayed breast reconstruction decreases postoperative complications in diabetic patients; however, this must be balanced with the known improvements in quality of life for immediate reconstruction. Future studies may further delineate staged from true immediate and delayed reconstruction patients.
Although composed of robust multi-institutional data, NSQIP-based analyses should be interpreted with caution with regards to breast reconstruction outcomes. Our study showed that the NSQIP data might not entirely reflect the complication profile of diabetic patients undergoing breast reconstruction as evidenced by our institutional data. These differences warrant further exploration for other patient groups and surgical outcomes, particularly for those procedures whose complications are not recorded by the NSQIP.
The Johns Hopkins Medicine Institutional Review Board approved this retrospective study of Johns Hopkins Hospital breast reconstruction patients. This study conforms to the Declaration of Helsinki ethical principles for medical research.
1. Albornoz CR, Bach PB, Mehrara BJ, et al. A paradigm shift in U.S. breast reconstruction: increasing implant rates. Plast Reconstr Surg. 2013;131:1523.
2. Fischer JP, Nelson JA, Au A, et al. Complications and morbidity following breast reconstruction—a review of 16,063 cases from the 2005-2010 NSQIP datasets. J Plast Surg Hand Surg. 2014;48:104114.
3. Sullivan SR, Fletcher DR, Isom CD, et al. True incidence of all complications following immediate and delayed breast reconstruction. Plast Reconstr Surg. 2008;122:1928.
4. Khoo A, Kroll SS, Reece GP, et al. A comparison of resource costs of immediate and delayed breast reconstruction. Plast Reconstr Surg. 1998;101:964968; discussion 969.
6. Terranova A. The effects of diabetes mellitus on wound healing. Plast Surg Nurs. 1991;11:2025.
7. Rosenberg CS. Wound healing in the patient with diabetes mellitus. Nurs Clin North Am. 1990;25:247261.
8. Greenhalgh DG. Wound healing and diabetes mellitus. Clin Plast Surg. 2003;30:3745.
9. Qin C, Vaca E, Lovecchio F, et al. Differential impact of non-insulin-dependent diabetes mellitus and insulin-dependent diabetes mellitus on breast reconstruction outcomes. Breast Cancer Res Treat. 2014;146:429438.
10. Fischer JP, Tuggle CT, Au A, et al. A 30-day risk assessment of mastectomy alone compared to immediate breast reconstruction (IBR). J Plast Surg Hand Surg. 2014;48:209215.
11. Knechtle WS, Perez SD, Medbery RL, et al. The Association between hospital finances and complications after complex abdominal surgery: deficiencies in the current health care reimbursement system and implications for the future. Ann Surg. 2015;262:273279.
12. Danforth RM, Pitt HA, Flanagan ME, et al. Surgical inpatient satisfaction: what are the real drivers? Surgery. 2014;156:328335.
13. Greenland S. Modeling and variable selection in epidemiologic analysis. Am J Public Health 1989;79:340349.
14. Malone DL, Genuit T, Tracy JK, et al. Surgical site infections: reanalysis of risk factors. J Surg Res. 2002;103:8995.
15. Lange CP, Ploeg AJ, Lardenoye JW, et al. Patient- and procedure-specific risk factors for postoperative complications in peripheral vascular surgery. Qual Saf Health Care 2009;18:131136.
16. van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. N Engl J Med. 2001;345:13591367.
17. Jones KW, Cain AS, Mitchell JH, et al. Hyperglycemia predicts mortality after CABG: postoperative hyperglycemia predicts dramatic increases in mortality after coronary artery bypass graft surgery. J Diabetes Complications 2008;22:365370.
18. Peleg AY, Weerarathna T, McCarthy JS, et al. Common infections in diabetes: pathogenesis, management and relationship to glycaemic control. Diabetes Metab Res Rev. 2007;23:313.
19. Vardakas KZ, Siempos II, Falagas ME. Diabetes mellitus as a risk factor for nosocomial pneumonia and associated mortality. Diabet Med. 2007;24:11681171.
20. Brem H, Tomic-Canic M. Cellular and molecular basis of wound healing in diabetes. J Clin Invest. 2007;117:12191222.
21. Falanga V. Wound healing and its impairment in the diabetic foot. Lancet 2005;366:17361743.
22. Galiano RD, Tepper OM, Pelo CR, et al. Topical vascular endothelial growth factor accelerates diabetic wound healing through increased angiogenesis and by mobilizing and recruiting bone marrow-derived cells. Am J Pathol. 2004;164:19351947.
23. Kaneto H, Katakami N, Matsuhisa M, et al. Role of reactive oxygen species in the progression of type 2 diabetes and atherosclerosis. Mediators Inflamm. 2010;2010:453892.
24. Maritim AC, Sanders RA, Watkins JB 3rd. Diabetes, oxidative stress, and antioxidants: a review. J Biochem Mol Toxicol. 2003;17:2438.
25. King GL. The role of inflammatory cytokines in diabetes and its complications. J Periodontol. 2008;79(8 Suppl):15271534.
26. Miller RB, Reece G, Kroll SS, et al. Microvascular breast reconstruction in the diabetic patient. Plast Reconstr Surg. 2007;119:3845; discussion 46.
27. Tran NV, Chang DW, Gupta A, et al. Comparison of immediate and delayed free TRAM flap breast reconstruction in patients receiving postmastectomy radiation therapy. Plast Reconstr Surg. 2001;108:7882.