Should Insulin-dependent Diabetic Patients be Screened for Malnutrition Before Total Joint Arthroplasty?

A Cohort at Risk

Andrew M. Schneider, MD; Nicholas M. Brown, MD


J Am Acad Orthop Surg. 2021;29(15):673-680. 

In This Article


Data for this study were acquired from the American College of Surgeon's National Surgical Quality Improvement Program's database, a validated collection of deidentified patient information from over 700 hospitals across the United States. This database was searched over an 8-year period from 2011 to 2018. The National Surgical Quality Improvement Program (NSQIP) database is rigorously maintained by trained surgical clinical reviewers and has been found to be extremely accurate, with rates of discrepancy <2% on random audits.[20,21] Patients for inclusion were selected using Current Procedural Terminology codes: primary Current Procedural Terminology codes of 27447 and 27130 were used to identify patients receiving primary total knee arthroplasty (TKA) and total hip arthroplasty, respectively. Only patients with a diagnosis of osteoarthritis were included. Exclusion criteria included patients with missing height, weight, sex, American Society of Anesthesiologists (ASAs) class 5, concurrent or bilateral procedures, sepsis or surgical site infection (superficial or deep) present at the time of surgery, and patients without preoperative albumin values. Using these criteria, 203,277 patients remained for the analysis (Figure 1). This study was exempt from institutional review board approval because of the deidentified nature of the database.

Figure 1.

Flow diagram depicting the construction of cohorts. ACS-NSQIP = American College of Surgeon's National Surgical Quality Improvement Program's, ASA = American Society of Anesthesiologist, BMI = body mass index, IDDM = insulin-dependent diabetes mellitus, NIDDM = noninsulin-dependent diabetes mellitus

NSQIP provides data for the 30-day period after surgery. Data collected included preoperative demographic information, risk factors for postoperative complications, and complications occurring in the 30-day postoperative period. Patients were stratified as morbidly obese (BMI ≥ 40 kg/m2) or nonmorbidly obese (BMI ≥18.5 kg/m2 and <40 kg/m2) and as malnourished (serum albumin level <3.5 mg/dL) or not malnourished (serum albumin level ≥3.5 mg/dL). In addition, patients were grouped as having no DM, noninsulin-dependent diabetes mellitus (NIDDM), or insulin-dependent diabetes mellitus (IDDM).

Outcomes in this study were reported as a composite group of complications, as provided by the NSQIP database, that includes death, myocardial infarction, cardiac arrest, stroke, renal failure, renal insufficiency, reintubation, failure to wean from ventilator, pulmonary embolism, deep space surgical site infection, deep incisional wound infection, sepsis, septic shock, superficial wound infection, wound dehiscence, pneumonia, urinary tract infection, and deep vein thrombosis. In addition, a separate composite outcome category was created to represent the risk of any infection and included superficial wound infection, deep incisional wound infection, deep space surgical site infection, and surgical wound dehiscence.

The data analysis was completed using SPSS (V26) to investigate the relationship between diabetes status (no diabetes versus NIDDM versus IDDM) and malnutrition on 30-day complication and infection risk. The univariate analysis was conducted using descriptive statistics and a Chi-Squared test. For each outcome (complication and infection), a multivariate logistic regression model was created to evaluate the impact of DM and albumin level, although controlling for potential confounders including age, sex, BMI, smoking, and ASA class. The models were built in a hierarchical fashion with independent variables incorporated with consideration of known risk factors for infection, statistical significance on univariate analysis, and model fit statistics. The results of the multivariate analysis were reported as adjusted odds ratios (ORs) with 95% confidence intervals. Statistical significance was set at a P value less than 0.05. 189,394 of 393,996 patients (48%) were excluded because of the absence of a recorded albumin value. To understand the bias this may introduce, the patients with recorded albumin values were compared with the patients without recorded albumin values using the Student t-test for continuous variables and the chi-squared for categorical variables.