Safety of Outpatient Single-Level Cervical Total Disc Replacement

A Propensity-Matched Multi-Institutional Study

Patawut Bovonratwet, BS; Michael C. Fu, MD, MHS; Vineet Tyagi, MD; Nathaniel T. Ondeck, MD, MHS; Todd J. Albert, MD; Jonathan N. Grauer, MD

Disclosures

Spine. 2019;44(9):E530-E538. 

In This Article

Materials and Methods

Patient Population

The NSQIP database contains data from over 500 participating institutions in the United States. Trained clinical reviewers prospectively record over 150 variables, such as patient baseline characteristics, intraoperative variables, and 30-day perioperative complications, using medical records, operative reports, and patient interviews.[7] Inter-rater disagreement rates for all recorded variables are reported to be below 2%.[8] The use of the NSQIP database in orthopedic research has become increasingly common and accepted.[9,10] Our institutional review board granted an exemption for studies using this dataset, as all data are de-identified.

Patients who underwent single-level CTDR procedures between 2005 and 2016 were identified from the NSQIP database using the Current Procedural Terminology (CPT) code 22856. Concurrent ACDF, posterior cervical fusion, and multilevel CTDR procedures were excluded using CPT codes. In addition, cases involving fracture, trauma, infectious diseases, or neoplasms were also excluded using International Classification of Diseases (ICD) diagnosis codes. Further, patients who had fatal complications occur on the day of surgery were excluded as well, as hospital length of stay was no longer relevant. Lastly, patients with missing baseline characteristics assessed in the current study were excluded (n = 11). On the basis of these inclusion and exclusion criteria, 1985 patients were included for further analysis.

Patient baseline characteristics were directly extracted from the NSQIP database and included variables such as age, gender, height, weight, functional status before surgery, American Society of Anesthesiologists (ASA) classification, smoking status (current smoker within 1 year), diabetes mellitus type, hypertension, and dyspnea on exertion. Body mass index, defined as weight (kg)/height (m)2, was calculated from height and weight. Following previous literature, the ASA classification was used as a measure of global comorbidity burden in the current study.[9,11] In addition, a previous published study has shown the ASA classification to be a superior predictor for postoperative outcomes than the modified Charlson Comorbidity Index and the modified Frailty Index.[12]

Definition of Outpatient Status

The NSQIP database records the total length of hospital stay (LOS), which is defined as the length of hospital stay from hospital admission to discharge, for each patient.[8] In the current study, patients with a LOS of 0 days were discharged on the same day of surgery and were considered to have undergone outpatient procedures. Patients with a LOS greater than 0 days stayed at least one night at the hospital and were considered to have undergone inpatient procedures.

On the basis of this categorization, patients who were kept overnight on 23-hour observation following surgery were considered to be part of the inpatient group. The maximum LOS in the current study is limited to 30 days.

Perioperative Complications and 30-day Readmission

The NSQIP database records the occurrence of individual perioperative complications through the 30th perioperative day for each patient regardless of discharge status. These individual complications were used to generate three groups of aggregated adverse events.

The occurrence of a minor adverse event (MAE) was defined as the occurrence of any of the following: wound complications (superficial wound infection, deep surgical site infection, organ space surgical site infection, or dehiscence), urinary tract infection, and blood transfusion. The occurrence of a serious adverse event (SAE) was defined as the occurrence of any of the following: death, return to the operating room, cardiac complications (myocardial infarction, cardiac arrest), thromboembolic event, pulmonary complications (on ventilator > 48 hours, unplanned intubation, and pneumonia), and sepsis/septic shock. The occurrence of any adverse event (AAE) was defined as the occurrence of a MAE or SAE.[13]

The NSQIP database also records the number of days after the principal procedure that each complication occurred for each patient. This information, along with hospital length of stay, was used to determine which complications occurred following discharge.

The occurrence of 30-day readmissions is also recorded in the NSQIP database for each patient. However, this variable is only recorded for cases that occurred in 2011 to 2016 but not for earlier cases. Therefore, the analysis of 30-day readmissions includes only 1827 of 1985 cases, but this represents approximately 92% of all cases included in the current study.

Data Analysis

Unadjusted Analysis. The first set of statistical analyses entailed unadjusted comparisons of patient baseline characteristics, individual perioperative complications, 30-day readmissions, and aggregated adverse events between the outpatient and inpatient CTDR cohorts. Comparisons were performed for both complications that occurred anytime during the 30-day perioperative period and following hospital discharge. Pearson Chi-squared test and Fisher exact tests were used for categorical variables.

Propensity Score Matched Analysis. For the second set of statistical analyses, propensity score matching was utilized in order to control for variances in patient baseline characteristics, which could be potential confounding factors between the two treatment groups.[14] Propensity score matching uses a propensity score, which is a single score that is calculated based on available covariates, to match patients between different treatment groups. Propensity score matching accounts for the nonrandom assignment of patients between the outpatient and inpatient cohorts.[14] In the current study, each outpatient CTDR case was matched with three inpatient CTDR cases by age, gender, BMI, functional status before surgery, ASA, smoking status, diabetes mellitus type, hypertension, and dyspnea on exertion. Propensity score matching is a commonly used statistical method in orthopedic surgical research.[15,16]

After the outpatient and inpatient CTDR cohorts were matched with propensity scores, patient baseline characteristics, individual perioperative complications, 30-day readmissions, and aggregated adverse events were again compared between the outpatient and matched inpatient cohorts. Comparisons were done for both complications that occurred anytime during the 30-day perioperative period and those following hospital discharge. Pearson Chi-squared test and Fisher exact tests were again used for categorical variables. For comparisons of patient baseline characteristics, statistical significance was set at P = 0.05. However, for comparisons of perioperative complications, as 13 tests were performed on both individual complications and aggregated adverse events, statistical significance was set at P = 0.004 with Bonferroni correction.[17] Therefore, 99.60% confidence intervals are reported.

For the third set of statistical analyses, Poisson regressions with robust error variance were used to calculate the relative risks (RRs) of AAEs and SAEs that occurred anytime during the 30-day perioperative period between the outpatient and matched inpatient CTDR groups. The same type of regression was used to compare AAEs and SAEs that occurred postdischarge between the two cohorts. RRs of MAEs were not calculated, as no MAE occurred in the outpatient cohort.

Subanalysis Comparing Perioperative Complications of Outpatient Single-level CTDR Versus Outpatient Single-Level ACDF. Outpatient, again defined as LOS=0 days, single-level ACDF was extracted from the 2005 to 2016 NSQIP database using CPT code 22551. Cases involving fracture, trauma, infectious diseases, neoplasms, multiple levels, or fatal complications on day of surgery, were again excluded. Each outpatient CTDR case was then matched with three outpatient single-level ACDF cases by age, gender, BMI, functional status before surgery, ASA, smoking status, diabetes mellitus type, hypertension, and dyspnea on exertion. Perioperative complications that occurred anytime during the 30-day perioperative period were then compared between outpatient CTDR and matched outpatient single-level ACDF.

All statistical analyses were performed using STATA version 13 (StataCorp LP, College Station, TX).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....