Abstract
Objective
The impact of postoperative hyperglycemia in patients undergoing open and endovascular procedures on the lower extremities has not been fully characterized with regard to associated admission diagnoses, hospital complications, mortality and 30-day readmission. This study evaluated the relationship of postoperative hyperglycemia on outcomes after lower extremity (LE) vascular procedures for peripheral artery disease (PAD).
Methods
Patients with PAD admitted for elective LE procedures between September 2008 and March 2014 were selected from the Cerner Health Facts® database using ICD-9-CM diagnosis and procedure codes. Using chi-square analysis, we evaluated the relationship of post-operative hyperglycemia (> 180 mg/dL) with sociodemographic characteristics, acute and chronic diagnoses, infections, length of hospital stay, and 30-day readmission. An adjusted multivariable logistic model was used to examine the association of postoperative hyperglycemia with infection and length of stay (LOS).
Results
Of 3,586 patients, 2,812 (78%) had optimal post-operative glucose control, while 774 (22%) had suboptimal glucose control (hyperglycemia). On average, patients with postoperative hyperglycemia experienced: longer hospital stays (6.9 vs. 5.1 days, p < .0001); higher Charlson index scores (3.4 vs. 2.5, p < .0001); higher rates of infection (23% vs. 14%, p < .0001); more acute complications (i.e., fluid and electrolyte disorders, acute renal failure, postoperative respiratory complications); and chronic problems (i.e., anemia, CHD, CKD, and diabetes) than patients with optimal glucose control. Overall 30-day readmission was 10.9% and was similar between groups (10.9% for both, p = .93). Major amputations did not differ between groups (P=.21). After adjusting for other risk factors using multivariable logistic regression, patients with hyperglycemia have 1.3 times the odds to have an infectious complication compared to those with optimal glucose control (OR 1.34, 95% CI 1.06–1.69), and 1.7 times the odds to have a hospital length of stay > 10 days (OR 1.69, 95% CI 1.32–2.15). As well, patients with post-operative hyperglycemia have 8.4 times the odds of dying in the hospital (OR 8.40, 95% CI 3.95–17.9).
Conclusions
One in five patients undergoing vascular procedures had postoperative hyperglycemia. Postoperative hyperglycemia was associated with adverse events after LE vascular procedures in patients with and without diabetes, including infection, increased hospital utilization, and mortality. No difference was found with respect to hospital readmission. Post-procedure glucose management may represent an important quality marker for improving outcomes after lower extremity vascular procedures.
Introduction
Hospital hyperglycemia, in individuals with and without diabetes, has been identified as a marker for poor clinical outcomes in a variety of surgical patients.(1–7) Hyperglycemia has been associated with infectious complications, the need for re-operative interventions, anastomotic failure, increased length of stay (LOS), increased readmission, and even increased mortality.(1, 8, 9) As well, in-hospital hyperglycemia is a common finding and may represent an important marker of poor clinical outcome and mortality in patients with and without a history of diabetes.(10)
There are limited data evaluating the impact of postoperative hyperglycemia in vascular surgery patients undergoing lower extremity interventions, particularly endovascular procedures. The purpose of this study was to evaluate the relationship of postoperative hyperglycemia with outcomes, complications, and readmission in patients with peripheral artery disease undergoing elective lower extremity open and endovascular vascular procedures.
Methods
Data
This retrospective study included adult patients with a diagnosis of peripheral artery disease (PAD) who underwent a lower extremity (LE) procedure. Using ICD-9-CM diagnosis and procedure codes, data were extracted from the Cerner Health Facts® database for patients admitted between September 2008 and March 2014 who met inclusion criteria. The Cerner Health Facts® database is comprised of electronic clinical records from hospitals that use Cerner Corporation’s electronic health record. Hospitals choose whether to participate and what data elements to include (e.g., patient demographics, diagnosis and procedures, encounter data, diagnostic test results, medications, and billing data). The Cerner Health Facts® database includes rigorous validity checks and is de-identified before it is made available using statistical methods that are compliant with the Health Insurance Portability and Accountability Act (HIPAA). Diagnosis codes were used to group clinically relevant conditions with the Agency for Healthcare Research and Quality’s (AHRQ) Clinical Classifications Software.(11) The study was exempted by the Health Sciences Institutional Review Board at the University of Missouri. Informed patient consent was not required as the Cerner Health Facts® database is completely de-identified using statistical methods that are compliant with the Health Insurance Portability and Accountability Act (HIPAA).
Study Population
The study population (N = 3,586) included patients who had a diagnosis of peripheral artery disease (ICD-9-CM diagnosis codes: rest pain, 440.22; gangrene, 440.24; claudication, 440.21; ulceration, 440.23; or unspecified severity, 440.20, 440.29, 440.4 or 443.9) and who had an open (procedure codes: 38.08, 38.18, 38.38, 38.48, 38.88, 39.29, 39.56, 39.57, or 39.58) or an endovascular (procedure codes: 39.50 or 39.90) lower extremity procedure performed.
Patients were excluded from the sample if they were less than 21 years old, had both an open and endovascular procedure performed, had an admission coded as urgent or emergent, had a hospital stay of 30 days or more, had an encounter where there was another encounter within 3 hours of discharge or admission, had no post-operative laboratory or medication data in Health Facts®, or had all post-operative blood glucose levels below 80 mg/dL.
Covariates
We extracted patients’ sociodemographic characteristics (age, gender, and race), hospital characteristics (size, whether or not it was a teaching facility), and acute/chronic problems (fluid and electrolyte disorders, CKD, diabetes, etc.). Hospitals with fewer than 100 open-heart procedures over the study period were defined as lacking an open heart cardiac program. Additionally, the Charlson Comorbidity Index (CCI), a measure that predicts one-year mortality based on patients’ comorbidities, was calculated. Using ICD-9 diagnosis codes to identify comorbidities, each was weighted and summed to derive a single comorbidity index for each patient.(12)
Using the ICD-9-CM diagnosis and procedure codes detailed above, categorical variables were created for procedure type (open vs. endovascular) and PAD severity (rest pain, gangrene, claudication, ulceration, or unspecified severity). If multiple diagnoses were coded, the most severe diagnosis code associated with the encounter was used to denote PAD severity.
Hyperglycemia
The Cerner Health Facts® database includes hospital laboratory data, including blood glucose levels. According to the American Association of Clinical Endocrinologists and American Diabetes Association Consensus Statement, serum blood glucose levels for the majority of non-critically ill patients should aim for blood glucose values <180 mg/dl as long as this target can be safely achieved.(13, 14) Using the patient’s highest random post-operative blood glucose measurement within 7 days following the procedure, patients’ glucose control was classified as being optimal (glucose levels between 80–180 mg/dL) or suboptimal (hyperglycemic – glucose levels above 180 mg/dL). Patients whose blood glucose levels were all below 80 mg/dL (n=125) were excluded.
Outcomes
The study examined outcomes including infections, length of stay, and 30-day readmission. Hospital infections included any patients who had a diagnosis code for lower extremity cellulitis (681.10, 682.6, or 682.7), pneumonia (031.0, 482, 482.2, 482.41, 482.42, 482.83, 482.9, 480.8, 483.8, 485, 481, or 486), sepsis (038.42, 038.40, 038.44, 038.49, 038.10, 038.11, 038.12, 038.19, 038.0, 038.8, 038.9, 449, 790.7, 995.91, or 995.92), surgical site infection (686.8, 686.9, 996.62, 997.62, 998.51, or 998.59), urinary tract infection (599.0 or 996.64), or another infection (999.31, 003.29, 008.01, 008.43, 008.45, 008.8, 009.0, 021–024, 032.89, 040, 040.00, 041.02, 041.04, 041.09–041.12, 041.19, 041.2–041.4, 041.49, 041.6, 041.7, 041.82, 041.84–041.86, 041.89, 041.9, V09.80, V09.81, V09.90, V09.91, 682, 682.2, 682.5, 682.8, 682.9, 686.01, 032.85, 680.2, 996.65, 996.66, 996.67, or 996.69). We also determined whether the hospital stay was longer than 10 days and whether patients were readmitted within 30 days of their discharge from the index admission.
Statistical Analysis
All analyses were performed in SAS version 9.4 (SAS Institute, Cary, NC). Using χ2 analysis, we evaluated the relationship of post-operative hyperglycemia with sociodemographic characteristics, acute and chronic problems, infections, hospital characteristics, stay characteristics, and 30-day readmission. Unadjusted relative risks (RR) and 95% confidence intervals (CI) were calculated. Additionally, multivariable logistic regression models were used to examine the association of postoperative hyperglycemia with infection, in-hospital mortality, and length of stay > 10 days, after adjusting for other patient and hospital factors. Odds ratios (OR) and 95% CIs were calculated. Model discrimination was assessed with the c-statistic, or area under the curve (1.0 indicates perfect fit and 0.5 is no better than a coin toss). The Hosmer-Lemeshow goodness-of-fit χ2 test was used to determine model calibration over the range of risk (p > .05 indicates adequate fit). To determine whether glucose levels were associated with outcomes independent of a diagnosis of diabetes, we performed a sub-analysis, removing subjects with a diabetes diagnosis from the multivariable logistic model.
Results
Overall
Characteristics of the 3,586 patients in the study population are shown in Table 1. Seventy-eight percent of patients (n = 2812) had optimal glucose control, while 22% had post-operative hyperglycemia (n = 774). Only 1373 patients (38%) of the study’s sample had pre-operative glucose lab data. Of these, 69.5% (954) had optimal glucose levels both pre- and post-surgery; 11% (152) had suboptimal levels both pre- and post-surgery; 12% (171) of patients had optimal levels pre-surgery but suboptimal levels post-surgery; and the remaining 7% (96) of patients had suboptimal levels pre-surgery but optimal levels post-surgery. Patients with optimal glucose control had a mean age of 68.3 years and patients with hyperglycemia were younger, with a mean age of 64.9 years (p < .0001). A majority of the sample was male (59%) or Caucasian (72%). There were no gender difference noted between optimal and suboptimal glucose control groups (p = .78), but fewer hyperglycemic patients were Caucasian (67%) compared to patients who had optimal glucose control (74%). Fifty-two percent of the study patients underwent open procedures, with no between-procedure differences in glucose control (p = .61).
Table 1.
Patient, hospital, and procedural characteristics for patients with peripheral artery disease who had lower extremity revascularization, by post-operative blood glucose levels [frequency (column %)].
| Post-operative Blood Glucose Levels | |||||||
|---|---|---|---|---|---|---|---|
| Total (N = 3586) |
Optimal1 (n = 2812) |
Suboptimal2 (n = 774) |
p-value3 | ||||
| Patient characteristics | |||||||
| Age (mean, SD) | 3586 | 67.6 (11.2) | 2812 | 68.3 (11.1) | 774 | 64.9 (11.4) | <.0001 |
| 21–59 | 897 | (25.0) | 634 | (22.6) | 263 | (34.0) | |
| 60–69 | 1102 | (30.7) | 859 | (30.6) | 243 | (31.4) | |
| 70–79 | 1013 | (28.3) | 832 | (29.6) | 181 | (23.4) | |
| 80 or older | 574 | (16.0) | 487 | (17.3) | 87 | (11.2) | |
| Gender (male) | 2100 | (58.6) | 1650 | (58.7) | 450 | (58.1) | .78 |
| PAD Severity | <.0001 | ||||||
| Gangrene | 474 | (13.2) | 312 | (11.1) | 162 | (20.9) | |
| Ulceration | 501 | (14.0) | 385 | (13.7) | 116 | (15.0) | |
| Rest pain | 517 | (14.4) | 433 | (15.4) | 84 | (10.9) | |
| Claudication | 967 | (27.0) | 788 | (28.0) | 179 | (23.1) | |
| Unknown | 1127 | (31.4) | 894 | (31.8) | 233 | (30.1) | |
| Race/ethnicity | <.0001 | ||||||
| African-American | 818 | (22.8) | 623 | (22.2) | 195 | (25.2) | |
| Caucasian | 2594 | (72.3) | 2074 | (73.8) | 520 | (67.2) | |
| Other/Unknown | 174 | (4.9) | 115 | (4.1) | 59 | (7.6) | |
| Hospital characteristics | |||||||
| Bed size | .02 | ||||||
| <200 | 287 | (8.0) | 236 | (8.4) | 51 | (6.6) | |
| 200–299 | 636 | (17.7) | 519 | (18.5) | 117 | (15.1) | |
| 300–499 | 950 | (26.5) | 746 | (26.5) | 204 | (26.4) | |
| 500 or more | 1713 | (47.8) | 1311 | (46.6) | 402 | (51.9) | |
| No cardiac program | 133 | (3.7) | 110 | (3.9) | 23 | (3.0) | .22 |
| Teaching facility | 2838 | (79.1) | 2177 | (77.4) | 661 | (85.4) | <.0001 |
| Procedural/stay characteristics | |||||||
| Admission Source | <.0001 | ||||||
| Physician | 2644 | (73.7) | 2105 | (74.9) | 539 | (69.6) | |
| Clinic | 391 | (10.9) | 318 | (11.3) | 73 | (9.4) | |
| Transfer | 90 | (2.5) | 64 | (2.3) | 26 | (3.4) | |
| Unknown | 461 | (12.9) | 325 | (11.6) | 136 | (17.6) | |
| Charlson index, mean (SD) | 3586 | 2.7 (1.7) | 2812 | 2.5 (1.6) | 774 | 3.4 (1.7) | <.0001 |
| 1 | 973 | (27.1) | 899 | (32.0) | 74 | (9.6) | |
| 2 | 1008 | (28.1) | 795 | (28.3) | 213 | (27.5) | |
| 3+ | 1605 | (44.8) | 1118 | (39.8) | 487 | (62.9) | |
| Discharge disposition | <.0001 | ||||||
| Home | 2748 | (76.6) | 2208 | (78.5) | 540 | (69.8) | |
| Other | 87 | (2.4) | 59 | (2.1) | 28 | (3.6) | |
| Transfer to SNF/Rehabilitation | 519 | (14.5) | 375 | (13.3) | 144 | (18.6) | |
| Died | 39 | (1.1) | 19 | (0.7) | 20 | (2.6) | |
| Unknown | 193 | (5.4) | 151 | (5.4) | 42 | (5.4) | |
| Length of stay, mean (SD) | 3586 | 5.5 (5.15) | 2812 | 5.1 (4.8) | 774 | 6.9 (5.9) | <.0001 |
| > 10 days | 534 | (14.9) | 352 | (12.5) | 182 | (23.5) | <.0001 |
| Procedure Type | .61 | ||||||
| Endovascular | 1711 | (47.7) | 1348 | (47.9) | 363 | (46.9) | |
| Open | 1875 | (52.3) | 1464 | (52.1) | 411 | (53.1) | |
| Readmission (within 30 days) | 392 | (10.9) | 308 | (10.9) | 84 | (10.9) | .93 |
Optimal glycemic control (highest post-operative blood glucose fell between 80–180 mg/dl).
Suboptimal glycemic control (highest post-operative blood glucose was > 180 mg/dl).
Chi-square (t-test for continuous) comparison of having optimal vs. suboptimal post-operative glucose levels.
SD=standard deviation, SNF = Skilled Nursing Facility
There were 67 hospitals represented in the study sample. All were in an urban location, most were teaching facilities (79%), and 74% had 300 or more beds. All but three had cardiac programs. A higher proportion of hyperglycemic patients (85%) were in teaching hospitals (p < .0001) than patients with optimal glucose control (77%). On average, patients with postoperative hyperglycemia experienced longer hospital stays than those with optimal glucose control (6.9 vs. 5.1 days, respectively; p < .0001) and had higher Charlson Comorbidity Index scores (3.4 vs. 2.5, respectively; p < .0001). Overall 30-day readmission was 10.9% and was similar between groups (10.9% optimal vs. 10.9% suboptimal, p = .93).
In unadjusted analyses, patients who had post-operative hyperglycemia were 1.55 times more likely to also have an infection (95% CI 1.35–1.79) than patients who had optimal post-operative glucose control (Table 2). Diagnoses and conditions associated with suboptimal post-operative glucose levels included fluid and electrolyte disorders (RR 1.45, 95% CI 1.24–1.71), acute renal failure (RR 1.89, 95% CI 1.60–2.23), acute respiratory conditions (RR 1.50, 95% CI 1.23–1.84), chronic anemia (RR 1.28, 95% CI 1.10–1.49), chronic heart disease (RR 1.26, 95% CI 1.11–1.43), coronary artery disease (RR 1.20, 95% CI 1.01–1.42), chronic kidney disease (RR 1.56, 95% CI 1.37–1.78), and diabetes (RR 5.15, 95% CI 4.36–6.10) While just under half of patients (46%) had a diagnosis of diabetes, the majority of those with post-procedural hyperglycemia (81.3%) were diabetics.
Table 2.
Unadjusted association of selected diagnoses during the index hospital encounter with post-operative blood glucose levels [frequency (column %)].
| Postoperative blood glucose level | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total (N = 3586) |
Optimal1 (n = 2812) |
Suboptimal2 (n = 774) |
RR (95% CI) | p-value3 | ||||
| Acute problems | ||||||||
| Fluid and electrolyte disorders | 429 | (12.0) | 301 | (10.7) | 128 | (16.5) | 1.45 (1.24–1.71) | <.0001 |
| Acute renal failure | 277 | (7.7) | 171 | (6.1) | 106 | (13.7) | 1.89 (1.60–2.23) | <.0001 |
| Respiratory complications | 235 | (6.6) | 161 | (5.7) | 74 | (9.6) | 1.50 (1.23–1.84) | <.0001 |
| Chronic problems | ||||||||
| Anemia | 601 | (16.8) | 442 | (15.7) | 159 | (20.5) | 1.28 (1.10–1.49) | .001 |
| Chronic heart disease | 1698 | (47.4) | 1286 | (45.7) | 412 | (53.2) | 1.26 (1.11–1.43) | .0002 |
| Coronary artery disease | 455 | (12.7) | 340 | (12.1) | 115 | (14.9) | 1.20 (1.01–1.42) | .04 |
| Chronic kidney disease | 744 | (20.8) | 519 | (18.5) | 225 | (29.1) | 1.56 (1.37–1.78) | <.0001 |
| Diabetes | 1638 | (45.7) | 1009 | (35.9) | 629 | (81.3) | 5.15 (4.36–6.10) | <.0001 |
| Infections | ||||||||
| Any infection4 | 573 | (16.0) | 396 | (14.1) | 177 | (22.9) | 1.55 (1.35–1.79) | <.0001 |
| Lower extremity cellulitis | 273 | (7.6) | 192 | (6.8) | 81 | (10.5) | 1.41 (1.16–1.72) | .0007 |
| Pneumonia | 70 | (2.0) | 46 | (1.6) | 24 | (3.1) | 1.60 (1.15–2.23) | .009 |
| Sepsis | 67 | (1.9) | 38 | (1.4) | 29 | (3.8) | 2.04 (1.54–2.70) | <.0001 |
| Surgical site infection | 82 | (2.3) | 55 | (2.0) | 27 | (3.5) | 1.54 (1.12–2.11) | .01 |
| Urinary tract infection | 97 | (2.7) | 66 | (2.4) | 31 | (4.0) | 1.50 (1.11–2.02) | .01 |
| Other infection | 211 | (5.9) | 148 | (5.3) | 63 | (8.1) | 1.41 (1.14–1.76) | .002 |
| Other complications | ||||||||
| Posthemorrhagic anemia | 261 | (7.3) | 194 | (6.9) | 67 | (8.7) | 1.20 (0.97–1.49) | .09 |
| Amputation (above/below knee) | 81 | (2.3) | 59 | (2.1) | 22 | (2.8) | 1.26 (0.88–1.81) | .21 |
| Post-operative medications | ||||||||
| Steroids | 258 | (7.2) | 172 | (6.1) | 86 | (11.1) | 1.61 (1.34–1.94) | <.0001 |
RR = relative risk; CI = confidence interval.
Optimal glycemic control (post-operative blood glucose levels between 80–180 mg/dl).
Suboptimal glycemic control (post-operative blood glucose > 180 mg/dl).
Chi-square comparison of having optimal vs. suboptimal post-operative glucose levels.
Any infection includes (LE cellulitis, pneumonia, sepsis, surgical site infection, urinary tract infection, and other infection).
Models
Several variables were also related to post-operative hyperglycemia in a multivariable regression model (Table 3), including age (OR 0.97, 95% CI 0.97–0.98), PAD severity of gangrene (OR 1.39, 95% CI 1.03–1.88), admission to a teaching facility (OR 2.03, 95% CI 1.51–2.74), fluid and electrolyte disorders (OR 1.34, 95% CI 1.02–1.76), acute renal failure (OR 1.72, 95% CI 1.25–2.35), sepsis (OR 1.98, 95% CI 1.08–3.64), and post-operative steroid use (OR 2.32, 95% CI 1.69–3.20). With the exception of diabetes (OR 6.97, 95% CI 5.48–8.86), none of the chronic problems associated with suboptimal glucose control in unadjusted analyses remained statistically significant in the multivariable model. The presence of a cardiac program was also unrelated to post-operative hyperglycemia.
Table 3.
Multivariable logistic regression model for risk factors for post-operative hyperglycemia
| Post-operative hyperglycemia Full Sample (n = 774) |
|||
|---|---|---|---|
| OR | (95% CI) | p-value | |
| Age (years) | 0.97 | (0.97 – 0.98) | <.0001 |
| Female | 1.11 | (0.92 – 1.34) | .28 |
| Race (reference = Caucasian) | |||
| African-American | 0.81 | (0.65 – 1.01) | .06 |
| Other | 1.84 | (1.26 – 2.70) | .001 |
| Charlson Index (reference = 1) | |||
| 2 | 1.30 | (0.93 – 1.80) | .12 |
| 3+ | 1.25 | (0.88 – 1.78) | .22 |
| Procedure type (endovascular) | 0.80 | (0.66 – 0.97) | .02 |
| Severity (reference = Claudication) | |||
| Gangrene | 1.39 | (1.03 – 1.88) | .02 |
| Rest pain | 0.72 | (0.53 – 0.99) | .04 |
| Ulcer | 1.00 | (0.74 – 1.36) | .97 |
| Unknown | 1.01 | (0.79 – 1.29) | .95 |
| Hospital bed size (reference = < 200) | |||
| 200–299 | 0.62 | (0.41 – 0.94) | .02 |
| 300–499 | 0.56 | (0.37 – 0.84) | .005 |
| 500 or more | 0.63 | (0.42 – 0.96) | .03 |
| Teaching facility | 2.03 | (1.51 – 2.74) | <.0001 |
| Acute problems | |||
| Fluid and electrolyte disorders | 1.34 | (1.02 – 1.76) | .03 |
| Acute renal failure | 1.72 | (1.25 – 2.35) | .0008 |
| Respiratory complications | 1.29 | (0.91 – 1.84) | .15 |
| Chronic problems | |||
| Anemia | 0.98 | (0.77 – 1.25) | .87 |
| Diabetes | 6.97 | (5.48 – 8.86) | <.0001 |
| Chronic heart disease | 1.14 | (0.94 – 1.38) | .17 |
| Coronary artery disease | 0.81 | (0.61 – 1.06) | .12 |
| Chronic kidney disease | 1.12 | (0.87 – 1.44) | .37 |
| Infections | |||
| Lower extremity cellulitis | 1.00 | (0.72 – 1.39) | .98 |
| Surgical site infection | 1.36 | (0.79 – 2.35) | .27 |
| Pneumonia | 1.03 | (0.56 – 1.88) | .93 |
| Sepsis | 1.98 | (1.08 – 3.64) | .02 |
| Urinary tract infection | 1.51 | (0.89 – 2.55) | .12 |
| Other infection | 0.92 | (0.63 – 1.34) | .66 |
| Other complications | |||
| Posthemorrhagic anemia | 1.05 | (0.75 – 1.48) | .77 |
| Post-operative medications | |||
| Steroids | 2.32 | (1.69 – 3.20) | <.0001 |
OR = odds ratio; CI = confidence interval.
Model statistics (full sample, n = 3586, c = .80, χ2 = 4.88, p = .76).
Multivariable logistic models were used to examine the association of postoperative hyperglycemia with infection, length of stay (> 10 days), and in-hospital mortality (Table 4). Because we found no association between 30-day readmission and post-operative glucose levels in unadjusted analyses (p=.93), we did not develop a readmission model. Models were adjusted for patient characteristics (age, gender, race, and diabetes diagnosis), hospital characteristics (bed size and whether it was a teaching facility), Charlson Comorbidity Index, lower extremity procedure type, post-operative steroid use, and PAD severity.
Table 4.
Multivariable logistic regression models for risk factors for infections, length of stay > 10 days, and in-hospital mortality
| Infection (n = 573) |
Length of stay > 10 days (n = 534) |
In-hospital mortality (n = 39) |
|||||||
|---|---|---|---|---|---|---|---|---|---|
| OR | (95% CI) | p-value | OR | (95% CI) | p-value | OR | (95% CI) | p-value | |
| Post-operative hyperglycemia | 1.34 | (1.06 – 1.69) | .01 | 1.69 | (1.32 – 2.15) | <.0001 | 8.40 | (3.95 – 17.9) | <.0001 |
| Diabetes | 0.88 | (0.69 – 1.11) | .27 | 0.69 | (0.54 – 0.88) | .003 | 0.09 | (0.04 – 0.21) | <.0001 |
| Age (years) | 1.00 | (1.00 – 1.01) | .41 | 1.01 | (1.00 – 1.02) | .20 | 1.03 | (1.00 – 1.06) | .08 |
| Charlson Index (reference = 1) | |||||||||
| 2 | 1.25 | (0.90 – 1.72) | .18 | 1.10 | (0.78 – 1.55) | .59 | 1.77 | (0.55 – 5.74) | .34 |
| 3+ | 2.38 | (1.76 – 3.21) | <.0001 | 2.74 | (2.01 – 3.75) | <.0001 | 4.70 | (1.68 – 13.1) | .003 |
| Endovascular procedure | 1.33 | (1.09 – 1.63) | .004 | 0.92 | (0.74 – 1.13) | .42 | 1.43 | (0.72 – 2.88) | .31 |
| Female | 1.28 | (1.05 – 1.56) | .01 | 1.10 | (0.90 – 1.36) | .35 | 1.10 | (0.56 – 2.15) | .78 |
| Hospital bed size (reference = <200)1 | |||||||||
| 200–299 | 1.51 | (0.94 – 2.43) | .08 | 1.94 | (1.12 – 3.36) | .01 | |||
| 300–499 | 1.57 | (0.97 – 2.54) | .06 | 1.67 | (0.96 – 2.90) | .07 | |||
| 500 or more | 1.05 | (0.64 – 1.71) | .85 | 1.60 | (0.92 – 2.79) | .09 | |||
| Post-operative steroids | 1.73 | (1.26 – 2.39) | .0008 | 1.85 | (1.33 – 2.58) | .0003 | 0.41 | (0.09 – 1.79) | .23 |
| Race (reference = Caucasian) | |||||||||
| African-American | 1.11 | (0.88 – 1.39) | .39 | 1.41 | (1.12 – 1.78) | .003 | 0.87 | (0.38 – 1.95) | .72 |
| Other | 0.74 | (0.46 – 1.19) | .21 | 0.63 | (0.37 – 1.08) | .09 | 0.34 | (0.04 – 2.86) | .31 |
| Severity (reference = claudication) | |||||||||
| Rest pain | 1.56 | (1.02 – 2.39) | .03 | 2.36 | (1.54 – 3.61) | <.0001 | 0.29 | (0.04 – 2.44) | .25 |
| Ulcer | 4.45 | (3.12 – 6.36) | <.0001 | 3.99 | (2.69 – 5.91) | <.0001 | 0.19 | (0.02 – 1.60) | .12 |
| Gangrene | 7.98 | (5.63 – 11.3) | <.0001 | 12.12 | (8.36 – 17.6) | <.0001 | 3.14 | (1.14 – 8.63) | .02 |
| Unknown | 2.71 | (1.95 – 3.76) | <.0001 | 2.85 | (1.99 – 4.09) | <.0001 | 1.73 | (0.69 – 4.33) | .24 |
| Teaching facility | 1.62 | (1.18 – 2.21) | .002 | 1.77 | (1.24 – 2.51) | .001 | 6.85 | (0.92 – 50.8) | .05 |
OR = odds ratio; CI = confidence interval.
Hospital bed size was not included in the in-hospital mortality model due to no deaths occurring in one or more of the bed size categories.
Model statistics (n = 3586; infection: c = .76, χ2 = 5.11, p = .74; length of stay > 10 days: c = .79, χ2 = 14.7, p = .07; in-hospital mortality: c = .88, χ2 = 6.88, p = .54).
After adjusting for patient, procedure, and hospital characteristics, patients with post-operative hyperglycemia have a higher odds of infectious complications (OR 1.34, 95% CI 1.06–1.69) than patients with optimal post-operative glucose control. Women were more likely to have an infection than men (OR 1.28, 95% CI 1.05–1.56). While a diagnosis of diabetes was not associated with infection (p=.27), higher Charlson scores were associated with greater odds of infection. Compared to patients with claudication, those with LE ulcers (OR 4.45, 95% CI 3.12–6.36), gangrene (OR 7.98, 95% CI 5.63–11.3), or rest pain (OR 1.56, 95% CI 1.02–2.39) had substantially higher odds of infection. Patients in teaching hospitals and those who were on post-operative steroids were also at higher risk of infection. Model calibration was adequate (χ2 = 5.1, p = .74) and discrimination was good (c = 0.76).
Similarly, patients with post-operative hyperglycemia were more likely to have a hospital stay of more than 10 days (OR 1.69, 95% CI 1.32–2.15) compared to patients with optimal post-operative glucose control (Table 4). African-American patients also had greater odds of prolonged hospitalization (OR 1.41, 95% CI 1.12–1.78), while a diagnosis of diabetes was associated with lower odds (OR 0.69, 95% CI 0.54–0.88). Charlson scores of 3 or more (OR 2.74, 95% CI 2.01–3.75), greater PAD severity, post-operative steroid use, and being in a teaching facility were also associated with greater odds of longer hospital stays. Model discrimination was good, with a c-statistic of 0.79; model calibration was adequate (χ2 = 14.7, p = .07).
In-hospital death was relatively rare (n = 39, 1%). In the multivariable logistic model, post-operative hyperglycemia was strongly associated with mortality (OR 8.40, 95% CI 3.95–17.9), while a diagnosis of diabetes was associated with lower odds (OR 0.09, 95% CI 0.04–0.21). A Charlson Index of 3 or more, and PAD severity of gangrene were also associated with higher odds of in-hospital mortality.
Sub-Analysis Excluding Diabetics
The non-diabetic sample included 1,948 subjects, 7.4% of whom had post-operative hyperglycemia. After adjusting for patient, procedure, and hospital characteristics, patients with post-operative hyperglycemia without a diagnosis of diabetes had 1.88 times the odds to have an infectious complication (OR 1.88, 95% CI 1.19–2.97), 2.13 times the odds to have a hospital stay of more than 10 days (OR 2.13, 95% CI 1.35–3.37), and higher odds of in-hospital mortality (OR 11.4, 95% CI 4.85–26.6), than patients with optimal post-operative glucose control (Table 5). Each model had adequate fit and discrimination.
Table 5.
Multivariable logistic regression models for risk factors for infections, length of stay > 10 days, and in-hospital mortality, excluding diabetics from sample
| Infection (n = 244) |
Length of stay > 10 days (n = 232) |
In-hospital mortality (n = 29) |
|||||||
|---|---|---|---|---|---|---|---|---|---|
| OR | (95% CI) | p-value | OR | (95% CI) | p-value | OR | (95% CI) | p-value | |
| Post-operative hyperglycemia | 1.88 | (1.19 – 2.97) | .007 | 2.13 | (1.35 – 3.37) | .001 | 11.4 | (4.85 – 26.6) | <.0001 |
| Age (years) | 1.01 | (0.99 – 1.02) | .40 | 1.00 | (0.99 – 1.02) | .65 | 1.03 | (0.99 – 1.07) | .19 |
| Charlson Index (reference = 1) | |||||||||
| 2 | 1.20 | (0.82 – 1.75) | .34 | 1.03 | (0.68 – 1.54) | .90 | 1.45 | (0.41 – 5.20) | .56 |
| 3+ | 2.11 | (1.51 – 2.96) | <.0001 | 2.86 | (2.02 – 4.03) | <.0001 | 4.76 | (1.67 – 13.6) | .003 |
| Endovascular procedure | 1.12 | (0.83 – 1.52) | .45 | 0.84 | (0.61 – 1.15) | .26 | 1.60 | (0.70 – 3.67) | .26 |
| Female | 1.57 | (1.17 – 2.11) | .002 | 0.98 | (0.72 – 1.33) | .89 | 0.80 | (0.36 – 1.79) | .59 |
| Hospital bed size (reference = <200)1 | |||||||||
| 200–299 | 1.66 | (0.76 – 3.60) | .20 | 2.46 | (0.98 – 6.16) | .05 | |||
| 300–499 | 1.58 | (0.72 – 3.44) | .25 | 1.77 | (0.70 – 4.44) | .22 | |||
| 500 or more | 1.10 | (0.50 – 2.41) | .81 | 2.15 | (0.86 – 5.39) | .10 | |||
| Post-operative Steroids | 2.30 | (1.47 – 3.61) | .0003 | 1.90 | (1.17 – 3.09) | .009 | 0.51 | (0.11 – 2.35) | .38 |
| Race (reference = Caucasian) | |||||||||
| African-American | 1.20 | (0.83 – 1.74) | .33 | 1.40 | (0.97 – 2.01) | .07 | 0.61 | (0.21 – 1.80) | .37 |
| Other | 0.86 | (0.37 – 1.97) | .71 | 0.79 | (0.32 – 1.97) | .60 | 0.28 | (0.03 – 2.95) | .28 |
| Severity (reference = claudication) | |||||||||
| Rest Pain | 1.63 | (0.93 – 2.86) | .08 | 1.96 | (1.15 – 3.36) | .01 | 0.46 | (0.05 – 4.13) | .49 |
| Ulcer | 5.39 | (3.28 – 8.86) | <.0001 | 3.76 | (2.24 – 6.31) | <.0001 | 0.27 | (0.03 – 2.49) | .24 |
| Gangrene | 7.90 | (4.70 – 13.3) | <.0001 | 10.1 | (6.01 – 16.9) | <.0001 | 4.93 | (1.45 – 16.7) | .01 |
| Unknown | 2.28 | (1.46 – 3.57) | .0003 | 1.94 | (1.23 – 3.07) | .004 | 1.90 | (0.64 – 5.63) | .24 |
| Teaching facility | 2.52 | (1.46 – 4.35) | .0009 | 2.65 | (1.43 – 4.92) | .001 | 4.76 | (0.63 – 36.2) | .13 |
OR = odds ratio; CI = confidence interval.
Hospital bed size was not included in the in-hospital mortality model due to no deaths occurring in one or more of the bed size categories.
Model statistics (n = 1948; infection: c = .77, χ2 = 6.65, p = .57; length of stay > 10 days: c = .78, χ2 = 10.07, p = .25; in-hospital mortality: c = .89, χ2 = 4.42, p = .81).
Discussion
This analysis has demonstrated that postoperative hyperglycemia is a common occurrence after lower extremity procedures, affecting one in five patients. Poor glycemic control was associated with increased risk of hospital infections, longer hospital stays, and in-hospital mortality. After adjusting for patient characteristics, severity of disease, comorbidities, and hospital characteristics, post-operative hyperglycemia remained significantly associated with infection and prolonged hospital stays. Furthermore, post-operative hyperglycemia was strongly associated with hospital mortality – patients with elevated post-operative blood glucose had almost eight times the odds of in-hospital mortality compared to those with optimal glucose control.
Diabetes as a diagnosis, however, was not associated with increased infection or mortality, suggesting that post-procedure hyperglycemia is a key marker for inferior outcomes after lower extremity procedures in both non-diabetic and diabetic patients. This finding was confirmed in a sub-analysis of non-diabetic patients that demonstrated similar findings. While poor post-operative glycemic control was associated with poor outcomes, having a diagnosis of diabetes was associated with lower odds of both increased length of stay and in-hospital mortality. This suggests that when the risk of post-operative hyperglycemia is known for those with a diagnosis of diabetes, patients for whom hyperglycemia seems less likely might be at particular risk. After adjustment, specific patient level factors associated with suboptimal glucose control include younger age; increased severity of disease; and acute problems including fluid and electrolyte disorders, acute renal failure, and respiratory complications. Procedure type was not associated with post-procedure hyperglycemia.
Postoperative hyperglycemia has been associated with inferior outcomes in the general surgery population, non-cardiac surgery patients with diabetes, and cardiac surgery patients.(3, 15) Despite current guidelines,(16) 22% of patients undergoing elective vascular procedures had suboptimal glucose control in our study. While we found no differences in the occurrence of post-operative hyperglycemia with procedure type, we did find that severity of disease (primarily gangrene), diabetes, and several acute conditions were associated with developing post-operative hyperglycemia.
This importance of postoperative glucose control has been evaluated in other surgical populations. Kwon et al. reported on the importance perioperative glycemic control in general surgery, and after controlling for clinical factors, found that patients with hyperglycemia had a significantly increased risk of infection and increased risk of poor outcomes both for patients with and without diabetes.(4) Other authors, using The Michigan Surgical Quality Collaborative database, assessed the relationship between high blood glucose levels in the early postoperative period following colorectal operations and found that superficial surgical site infections, sepsis, and death were associated with postoperative serum hyperglycemia in patients without diabetes, and concluded that glucose monitoring is critical for all patients undergoing colorectal surgery.(17) Additionally, clinical outcomes associated with postoperative hyperglycemia in cardiac surgery patients who were stratified by diabetes status demonstrated that in patients without diabetes, hyperglycemia (≥180 mg/dL) was associated with an increase in infections and respiratory complications.(18) A systematic review of the effects of tight glycemic control (blood glucose levels <= 200 mg/dL) reported a reduction in the incidence of surgical site infections in adult patients with diabetes after cardiac surgery.(19) For vascular surgery patients undergoing open lower extremity bypass, Hirashima et al. evaluated the effect of strict glucose control after lower extremity bypass and compared 104 patients in an intravenous insulin study group to a historic control group of 189 patients. They concluded that strict glucose control with a postoperative insulin infusion protocol significantly decreased the incidence of postoperative in-hospital wound infection in the diabetic population, but did not lower other postoperative complications.(20)
An unexpected finding from this analysis was that after multivariable adjustment, diabetes was not associated with increased infection risk, but was associated with shorter hospital stays and decreased mortality. It has been previously reported that in-hospital hyperglycemia is a common finding and represents an important marker of poor clinical outcome and mortality in patients with and without a history of diabetes, and that patients with newly diagnosed hyperglycemia have worse outcomes.(10) Our results from both the entire cohort and the non-diabetic sub-analysis sample suggest that post-procedure hyperglycemia is independently associated with inferior outcomes and strengthens the importance of vigilance non-diabetic patient after lower extremity procedures.
Previous research has found similar results in non-vascular surgical cohorts.(1, 4) Kotagal et al. demonstrated that hyperglycemic diabetic patients did not have an increased risk of adverse events, and they suggested that hyperglycemia indicates higher levels of stress in in the non-diabetic patient.(1) As well, Kwon et al. reported that, after controlling for clinical factors, increased risk of poor outcomes was observed for patients both with and without diabetes.(4) These previous reports strengthen our finding that diabetes itself was not associated with increased infection, length of stay, or mortality after LE vascular procedures, but that the marker of hyperglycemia may represent the greatest risk after open vascular and endovascular procedures. As well, the diagnosis of diabetes is a known risk factor for hyperglycemia in the post-operative period and these patients are more likely to be placed on protocols and more carefully monitored during their hospitalization, explaining why diabetes is not associated with inferior outcomes.
Previous analyses have demonstrated an association between glucose control and increased hospital utilization. Evans at al. evaluated the relationships between hyperglycemia and length of hospital stay in an acute medical unit, finding that length of stay for patients with hyperglycemia on admission was significantly longer.(21) In the cardiac population, Greco et al. demonstrated that hyperglycemia was associated with an additional cost and higher hospital LOS.(18) Ables et al. also demonstrated that glycemic control shortens the length of stay in non-critically hospitalized patients.(8) Similar to these reports, lower extremity vascular procedures with optimal glucose control in the post-operative period demonstrated decreased LOS in our analysis. Patients with glucose levels greater than > 180 mg/dl had 1.7 times the odds to have stays longer than 10 days.
Postoperative glucose control after elective vascular procedures did not affect readmission in our analysis. Readmission after lower extremity procedures was 10.9% within 30 days, and did not differ significantly between patients with optimal and suboptimal glucose control. Lee et al. evaluated the association of mean and maximum blood glucose levels with 30-day hospital readmission among patients in the intensive care unit undergoing invasive cardiovascular surgery and found that higher mean and maximum blood glucose levels were not associated with 30-day readmission.(22) As well, Myint et al., using the Older Persons Surgical Outcomes Collaboration multicenter audit data, did not find that readmission within 30 days was significantly associated with hyperglycemia.(23) Contrary to these reports, Evans at al. evaluated the relationship between admission glucose levels and subsequent readmission in an acute medical unit, demonstrating that the 28-day readmission rates increased as admission glucose levels increased.(21)
We also evaluated hospital characteristics associated with suboptimal post-operative glucose control. Teaching facilities were significantly associated with suboptimal glucose control in this analysis and remained significant after multivariable adjustment. Reasons for this finding are unclear, but may be related to severity of illness or other confounders not clearly elucidated in the data that are seen and treated at teaching institutions. Other authors have reported that hospital characteristics should be considered when examining national inpatient glucose data, and that there can be a significant interaction between hospital characteristics and inpatient glucose control.(24) As well, an evaluation of high hospital infection-related readmissions in California acute care hospitals demonstrated that hospitals serving populations with high levels of comorbidities, prolonged length of stay, populations living in a federal poverty area, and academic hospitals had higher all-cause and infection-related readmission rates.(25) Therefore, suboptimal post-operative glucose control in teaching facilities may represent a disproportionate rate of high disease severity and comorbidities seen in academic institutions.
This study has an inherent limitation in that ICD-9 codes were used to identify procedures and diagnoses, and coding can vary between institutions. However, data undergo rigorous validity evaluation prior to being added to Health Facts. As well, of all the encounters contained in the data set, approximately half had blood glucose measurements, which may have introduced selection bias. We are unable to determine if patients had uncontrolled diabetes as a reason for postoperative hyperglycemia from inpatient laboratory data; few patients in the sample had hemoglobin A1c data, as this test is often performed in the ambulatory setting. All encounters in this study were coded as elective admissions, though there may have been errors in coding. Based on the data available, we were unable to determine the temporal timing of events such as diagnosis assignment during a hospitalization, and can only draw associations. Furthermore, areas of cost, quality markers such as Leapfrog scores, and hospital-acquired conditions cannot be evaluated in this data set. All of the procedures utilized for this analysis were performed in hospitals, as no outpatient facilities are contained within the data set. In addition, we are unable to determine readmissions to hospitals in different health systems. As Cerner Corporations' Health Facts is a proprietary database comprised of electronic clinical records from hospitals and hospital systems that use Cerner’s electronic health record, the ethnicity variable may not be completely reflective of the US population. Some urban hospitals, smaller hospitals, and rural hospitals may be underrepresented. Proportions are comparable to previously published evaluations of LE procedures utilizing Medicare data which is considered a representative sample of the elderly US population.(26)
In conclusion, this analysis has demonstrated that postoperative hyperglycemia was a frequent occurrence after both open and endovascular lower extremity procedures and was significantly associated with increased risk of hospital complications and death. Diabetes alone was not associated with increased infection, increased length of stay, or increased mortality, suggesting that post-procedure hyperglycemia is a significant risk factor for interior outcomes after lower extremity procedures. Hyperglycemia in diabetic and non-diabetic patients was associated with complications and death during the hospital stay in the post-procedural period, and may represent an important marker for future improvement in lower extremity vascular surgery procedures.
Acknowledgments
Funding Source: Support from the Agency for Healthcare Research and Quality (R24HS022140) was used to fund the research reported in this publication. The authors take sole responsibility in the content of this report, which does not necessarily represent the official views of the Agency for Healthcare Research and Quality.
Footnotes
Presented at the 40th Annual Midwestern Vascular Surgical Society, Plenary Session, Columbus, Ohio, September 8–10, 2016.
References
- 1.Kotagal M, Symons RG, Hirsh IB, Umpierrez GE, Dellinger EP, Farrokhi ET, Flum DR. Perioperative hyperglycemia and risks of adverse events among patients with and without diabetes. Ann Surg. 2015;261:97–103. doi: 10.1097/SLA.0000000000000688. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Frisch A, Chandra P, Smiley D, Peng L, Rizzo M, Gatcliffe C, Hudson M, et al. Prevalence and clinical outcome of hyperglyxemia in the perioperative period in noncardiac surgery. Diabetes Care. 2010;33:1783–8. doi: 10.2337/dc10-0304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.King JT, Jr, Goulet JL, Perkal MF, Rosenthal RA. Glycemic control and infections in patients with diabetes undergoing noncardiac surgery. Ann Surg. 2011;253(1):158–65. doi: 10.1097/SLA.0b013e3181f9bb3a. [DOI] [PubMed] [Google Scholar]
- 4.Kwon S, Thompson R, Dellinger P, Yanez D, Farrohki E, Flum D. Importance of perioperative glycemic control in general surgery: a report from the Surgical Care and Outcomes Assessment Program. Ann Surg. 2013;257(1):8–14. doi: 10.1097/SLA.0b013e31827b6bbc. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Buehler L, Fayfman M, Alexopoulos AS, Zhao L, Farrokhi F, Weaver J, Smiley-Bryd D, et al. The impact of hyperglycemia obesity on hospitalization costs and clinical outcomes in general surgery patients. Journal of Diabetes and Its Complications. 2015;29:1177–82. doi: 10.1016/j.jdiacomp.2015.07.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Fiorillo C, Rosa F, Quero G, Menghi R, Doglietto GB, Alfieri S. Postoperative hyperglycemia in nondiabetic patients after gastric surgery for cancer: perioperative outcomes. Gastric Cancer. 2017;20:536–542. doi: 10.1007/s10120-016-0621-5. [DOI] [PubMed] [Google Scholar]
- 7.Vriesendorp TM, Morelis QJ, Devries JH, Legemate DA, Hoekstra JB. Early post-operative glucose levels are an independent risk factor for infection after peripheral vascular surgery. A retrospective study. Eur J Vasc Endovasc Surg. 2004;28(5):520–5. doi: 10.1016/j.ejvs.2004.08.006. [DOI] [PubMed] [Google Scholar]
- 8.Ables AZ, Bouknight PJ, Bendyk H, Beagle R, Alsip R, Williams J. Blood glucose control in noncritically ill patients is associated with a decreased length of stay, readmission rate, and hospital mortality. J Healthc Qual. 2016;38:e89–e96. doi: 10.1097/JHQ.0000000000000023. [DOI] [PubMed] [Google Scholar]
- 9.Van Kuijk JP, Dunkelgrun M, Schreiner F, Flu WJ, Galal W, van Domburg RT, Hoeks SE, et al. Preoperative oral glucose tolerance testing in vascular surgery patients: long-term cardiovascular outcome. Am Heart J. 2009;157(5):919–25. doi: 10.1016/j.ahj.2009.02.011. [DOI] [PubMed] [Google Scholar]
- 10.Umpierrez GE, Isaacs SD, Bazargan N, You X, Thaler LM, Kitabchi AE. Hyperglycemia: An independent marker of in-hospital mortality in patients with undiagnosed diabetes. J Clin Endocrinol Metab. 2002;87(3):978–82. doi: 10.1210/jcem.87.3.8341. [DOI] [PubMed] [Google Scholar]
- 11.Clinical Classifications Software (CCS) for ICD-9-CM. Agency for Healthcare Research and Quality; Rockville, MD: 2015. ( http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp) [Google Scholar]
- 12.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–83. doi: 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
- 13.Moghissi ES, Korytkowski MT, DiNardo M, Einhorn D, Hellman R, Hirsch IB, et al. American Association of Clinical Endocrinologists and American Diabetes Association consensus statement on inpatient glycemic control. Endocr Pract. 2009;15(4):353–69. doi: 10.4158/EP09102.RA. [DOI] [PubMed] [Google Scholar]
- 14.Umpierrez GE, Hellman R, Korytkowski MT, Kosiborod M, Maynard GA, Montori VM, et al. Management of hyperglycemia in hospitalized patients in non-critical care setting: an endocrine society clinical practice guideline. J Clin Endocrinol Metab. 2012;97(1):16–38. doi: 10.1210/jc.2011-2098. [DOI] [PubMed] [Google Scholar]
- 15.Ramos M, Khalpey Z, Lipsitz S, Steinberg J, Panizales MT, Zinner M, et al. Relationship of perioperative hyperglycemia and postoperative infections in patients who undergo general and vascular surgery. Ann Surg. 2008;248(4):585–91. doi: 10.1097/SLA.0b013e31818990d1. [DOI] [PubMed] [Google Scholar]
- 16.Moghissi ES, Korytkowski MT, DiNardo M, Einhorn D, Hellman R, Hirsch IB, et al. American Association of Clinical Endocrinologists and American Diabetes Association consensus statement on inpatient glycemic control. Diabetes Care. 2009;32(6):1119–31. doi: 10.2337/dc09-9029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mohan S, Kaoutzanis C, Welch KB, Vandewarker JF, Winter S, Krapohl G, et al. Postoperative hyperglycemia and adverse outcomes in patients undergoing colorectal surgery: results from the Michigan surgical quality collaborative database. Int J Colorectal Dis. 2015;30(11):1515–23. doi: 10.1007/s00384-015-2322-7. [DOI] [PubMed] [Google Scholar]
- 18.Greco G, Ferket BS, D'Alessandro DA, Shi W, Horvath KA, Rosen A, et al. Diabetes and the association of postoperative hyperglycemia with clinical and economic outcomes in cardiac surgery. Diabetes Care. 2016;39(3):408–17. doi: 10.2337/dc15-1817. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Boreland L, Scott-Hudson M, Hetherington K, Frussinetty A, Slyer JT. The effectiveness of tight glycemic control on decreasing surgical site infections and readmission rates in adult patients with diabetes undergoing cardiac surgery: A systematic review. Heart Lung. 2015;44(5):430–40. doi: 10.1016/j.hrtlng.2015.06.004. [DOI] [PubMed] [Google Scholar]
- 20.Hirashima F, Patel RB, Adams JE, Bertges DJ, Callas PW, Steinthorsson G, et al. Use of a postoperative insulin protocol decreases wound infection in diabetics undergoing lower extremity bypass. J Vasc Surg. 2012;56(2):396–402. doi: 10.1016/j.jvs.2012.01.026. [DOI] [PubMed] [Google Scholar]
- 21.Evans NR, Dhatariya KK. Assessing the relationship between admission glucose levels, subsequent length of hospital stay, readmission and mortality. Clin Med (Lond) 2012;12(2):137–9. doi: 10.7861/clinmedicine.12-2-137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Lee LJ, Emons MF, Martin SA, Faries D, Bae J, Nathanson BH, et al. Association of blood glucose levels with in-hospital mortality and 30-day readmission in patients undergoing invasive cardiovascular surgery. Curr Med Res Opin. 2012;28(10):1657–65. doi: 10.1185/03007995.2012.718268. [DOI] [PubMed] [Google Scholar]
- 23.Myint PK, Owen S, Pearce L, Greig MF, Tay HS, McCormack C, et al. The prevalence of hyperglycaemia and its relationship with mortality, readmissions and length of stay in an older acute surgical population: a multicentre study. Postgrad Med J. 2016;92:514–519. doi: 10.1136/postgradmedj-2015-133777. [DOI] [PubMed] [Google Scholar]
- 24.Bersoux S, Cook CB, Kongable GL, Shu J. Trends in glycemic control over a 2-year period in 126 US hospitals. J Hosp Med. 2013;8(3):121–5. doi: 10.1002/jhm.1997. [DOI] [PubMed] [Google Scholar]
- 25.Gohil SK, Datta R, Cao C, Phelan MJ, Nguyen V, Rowther AA, et al. Impact of Hospital Population Case-Mix, Including Poverty, on Hospital All-Cause and Infection-Related 30-Day Readmission Rates. Clin Infect Dis. 2015;61(8):1235–43. doi: 10.1093/cid/civ539. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Vogel TR, Dombrovskiy VY, Carson JL, Graham AM. In-hospital and 30-day outcomes after tibioperoneal interventions in the US Medicare population with critical limb ischemia. J Vasc Surg. 2011;54(1):109–15. doi: 10.1016/j.jvs.2010.12.055. [DOI] [PubMed] [Google Scholar]