The increase in the number of carotid artery stenting (CAS) procedures over the last decade has necessitated critical appraisal of procedural outcomes and patterns of utilization including cost analysis. The main objectives of our study were to evaluate the postprocedural mortality and complications after CAS and the patterns of resource utilization in terms of length of stay (LOS) and cost of hospitalization. We queried the Healthcare Cost and Utilization Project’s Nationwide Inpatient Sample from 2006 to 2010 using the International Classification of Diseases, Ninth Revision, procedure code of 00.63 for CAS. Hierarchical mixed-effects models were generated to identify the independent multivariate predictors of in-hospital mortality, procedural complications, LOS, and cost of hospitalization. A total of 13,564 CAS procedures (weighted n = 67,344) were analyzed. The overall postprocedural mortality was low at 0.5%, whereas the complication rate was 8%, both of which remained relatively steady over the time frame of the study. Greater postoperative mortality and complications were noted in symptomatic patients, women, and those with greater burden of baseline co-morbidities. A greater operator volume was associated with a lower rate of postoperative mortality and complications, as well as shorter LOS and lesser hospitalization costs. In conclusion, the postprocedural mortality after CAS has remained low over the recent years. Operator volume is an important predictor of postprocedural outcomes and resource utilization.
The geriatric population in the United States is expected to double during 2000 to 2030 with the proportion of elderly individuals estimated to increase from 12.4% to 19.6%. With a steady increase in the older population, the burden of cardiovascular disease including carotid disease is bound to worsen. With stroke being the third most common cause of death in the United States and carotid stenosis accounting for 20% to 25% of the strokes, a focus on appropriate management strategies for carotid artery stenosis is becoming increasingly germane. Although carotid endarterectomy has long been the standard of care for carotid revascularization, carotid artery stenting (CAS) has been gaining popularity especially after randomized control trials demonstrating noninferiority within the last decade. The consequential increase in the number of CAS procedures has necessitated critical appraisal of procedural outcomes and patterns of utilization including cost analysis. The main objectives of our study were twofold: (1) to evaluate the postprocedural mortality and complications after CAS and (2) to study the patterns of resource utilization in terms of length of stay (LOS) and cost of hospitalization using the nation’s largest all-payer insurance inpatient database from recent years (2006 to 2010).
Methods
The study cohort was derived from the Nationwide Inpatient Sample (NIS) database, from 2006 to 2010, a subset of the Healthcare Cost and Utilization Project sponsored by the Agency for Healthcare Research and Quality. The NIS is the largest, publicly available, all-payer inpatient care database in the United States, including data on approximately 7 to 8 million discharges per year, and is a stratified sample designed to approximate a 20% sample of US community (nonfederal, short-term, general, and specialty) hospitals. National estimates are produced using sampling weights provided by the sponsor. Annual data quality assessments of the NIS are performed, which guarantee the internal validity of the database. Furthermore, comparisons against the following data sources strengthen the external validity of the NIS: the American Hospital Association Annual Survey Database, the National Hospital Discharge Survey from the National Center for Health Statistics, and the MedPAR inpatient data from the Centers for Medicare and Medicaid Services. We queried the NIS database from 2006 to 2010 using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) procedure code designating CAS (ICD-9: 00.63). Only adult patients (age ≥18 years) were selected. Any patients with concomitant carotid endarterectomy (ICD-9-CM: 38.12) were excluded. Patients were considered symptomatic if they had primary diagnosis of CAS with cerebral infarction (ICD-9-CM: 433.11) or stroke.
NIS variables were used to identify patients’ demographic characteristics including age, gender, and race. Census data were used to calculate the utilization rate of CAS for individual years using population data for adults aged ≥18 years. We defined severity of co-morbid conditions using Deyo modification of Charlson co-morbidity index (CCI), which contains 17 co-morbid conditions with differential weights. The score ranges from 0 to 33, with higher scores corresponding to greater burden of co-morbid diseases ( Supplementary Table 1 ). Preventable procedural complications were identified by patient safety indicators (PSIs), which were established by the Agency for Healthcare Research and Quality to monitor preventable adverse events during hospitalization. These indicators are based on ICD-9-CM codes and Medicare severity diagnosis-related groups, and each PSI has specific inclusion and exclusion criteria ( Supplementary Table 2 ). PSI’s individual technical specifications were used to identify and define preventable procedural complications, namely, postprocedural respiratory failure, postprocedural renal and metabolic derangements, postprocedural pulmonary embolism or deep vein thrombosis, procedural infectious complications, which included postprocedural sepsis and central venous catheter–related bloodstream infection, and pressure ulcers. Other procedure-related complications, which included hemorrhage requiring blood transfusion, iatrogenic cardiac complications, requirement of open heart surgery, procedural stroke, or transient ischemic attack, and vascular complications were identified using ICD-9-CM codes in any secondary diagnosis field. To prevent classification of a pre-existing condition (e.g., stroke or heart block) as a complication, cases with the ICD-9-CM code for a complication listed as the principal diagnosis (DX1) were excluded. Vascular complications were defined as a PSI code for accidental puncture or ICD-9-CM codes for injury to blood vessels, creation of arteriovenous fistula, injury to retroperitoneum, vascular complications requiring surgery, and other vascular complications not elsewhere classified. “Any complication” was defined as occurrence of ≥1 procedural complications listed in Supplementary Table 1 .
The total duration of hospital stay in days was estimated for all patients, after excluding those who died in the hospital, using the information on LOS provided in the NIS data set. Disposition was classified into 3 categories: those who were discharged home or with home care services were classified as home-based discharge, those who were discharged to short- or long-term nursing home or transferred to another facility were classified as discharge to another facility, and those who died in hospital were classified as in-hospital mortality. Annual hospital volume was determined on a year-to-year basis using the unique hospital identification number to calculate the total number of procedures performed by a particular institution in a given year. Similarly, operator volume was computed using operator identification number. Annual hospital and operator volumes in tertiles were used for further analysis and to create hierarchical multivariate models. Adjustment term was added to multivariate models to adjust for interaction effect between hospital and operator volumes. Hospital ID was incorporated as a random effect in the model to account for the effect of hospital clustering (meaning that patients treated at the same hospital may experience similar outcomes as a result of other processes of care).
The Healthcare Cost and Utilization Project NIS contains data on total charges for each hospital in the databases. This charge information represents the amount that hospitals billed for services but does not reflect how much the hospital’s services actually cost or the specific amounts that hospitals received in payment. To calculate estimated cost of hospitalizations, the NIS data were merged with cost-to-charge ratios available from the Healthcare Cost and Utilization Project. Using the merged data elements from the cost-to-charge ratio files and the total charges reported in the NIS database, we converted the hospital’s total charge data to cost estimates by simply multiplying total charges with the appropriate cost-to-charge ratio. These costs are in essence standardized and can be measured across hospitals and are used in the remainder of this report. Adjusted cost for each year was calculated in terms of the 2010 cost, after adjusting for inflation according to the latest consumer price index data released by US government on January 16, 2013. The similar method has been used for previous studies.
Stata IC 11.0 (StataCorp, College Station, Texas) and SAS 9.3 (SAS Institute Inc., Cary, North Carolina) were used for analyses. Weighted values of patient-level observations were generated to produce a nationally representative estimate of the entire US population of hospitalized patients. Differences between categorical variables were tested using the chi-square test, and differences between continuous variables were tested using the Student t test. p Value <0.05 was considered significant. The NIS data set is inherently hierarchical, namely, the data have group-specific attributes (i.e., hospital), and within each group (i.e., hospital), there are patients, which contributes patient-specific attributes to the data. Hierarchical models take into consideration the effect of nesting (e.g., patient-level effects nested within hospital-level effects). Hence, it is superior to simple regression modeling for the available data set. Two-level hierarchical models (with patient-level factors nested within hospital-level factors) were created with the unique hospital identification number incorporated as random effects within the model. Hierarchical mixed-effects logistic regression models were used for categorical dependent variables such as in-hospital mortality and procedural complications, and hierarchical mixed-effects linear regression models were used for continuous dependent variables such as cost of hospitalization and LOS. In all multivariate models, we included hospital-level variables such as hospital region and teaching versus nonteaching hospitals and patient-level variables such as age, gender, Deyo modification of CCI, use of intracardiac echocardiography, occurrence of procedural complications, admission over the weekend, and primary payer (with Medicare and/or Medicaid considered as referent) in addition to hospital and/or operator procedure volume tertile. All interactions were thoroughly tested. Multicollinearity, defined as a perfect linear relation or very high correlation between ≥2 predictor variables (independent), was assessed using variance inflation factor, with variance inflation factor >20 suggestive of multicollinearity.
Results
A total of 13,564 CAS procedures (weighted n = 67,344) were analyzed. Table 1 lists baseline characteristics of the study population. The mean age was 71 ± 10 years, with majority of the cohort being men (60.6%) and white (68.9%). The average CCI score for the cohort was 2.23 ± 0.01, with hypertension being the most common co-morbidity present in 75% of the patients who underwent CAS. A significant proportion of these procedures were done in large (69.4%) and teaching hospitals (60.3%) after an elective admission (69.8%). The most common primary payer for these procedures was Medicare and/or Medicaid (74.6%).
| Actual (Projected) Number of Percutaneous Carotid Stent Placement Procedure in the US | 13,564 (67,344) |
|---|---|
| Age (years) | 71 ± 10 |
| Men | 60.6% |
| White ∗ | 68.9% |
| Non-white ∗ | 10.5% |
| Charlson/Deyo comorbidity index † | 2.23 ± 0.01 |
| Obesity | 5.4% |
| Hypertension | 75% |
| Diabetes mellitus | 31% |
| Congestive heart failure | 10.5% |
| Chronic pulmonary disease | 20.9% |
| Peripheral vascular disease | 27.4% |
| Pulmonary circulatory disease | 1.2% |
| Renal failure | 13.4% |
| Neurological disorder/paralysis | 0.4% |
| Anemia/coagulopathy | 8.1% |
| Hematological/oncological malignancy | 1.8% |
| Weight loss/cachexia | 0.5% |
| Rheumatoid arthritis/collagen vascular disease | 1.4% |
| Depression/substance abuse | 6.8% |
| Median household income category for patient’s zip code ‡ (percentile) | |
| 0–25th | 26.2% |
| 26–50th | 27.4% |
| 51–75th | 24.5% |
| 76–100th | 19.7% |
| Primary payer | |
| Medicare/Medicaid | 74.6% |
| Private including HMO | 21.2% |
| Self-pay/no charge/other | 3.9% |
| Hospital bed size | |
| Small | 10.5% |
| Medium | 19.2% |
| Large | 69.4% |
| Hospital region | |
| Northeast | 19% |
| Mid West/North Central | 24.09% |
| South | 46.08% |
| West | 10.80% |
| Teaching hospital | 60.3% |
| Admission types | |
| Emergent/urgent | 29% |
| Elective | 69.8% |
| Weekend admission (%) | 3.7% |
| Length of stay (means ± SE) (days) | 2.52 ± 0.02 |
| Total charges § ($) (means ± SE) | 14,366 ± 46 |
| Peri-procedural complications ¶ | 8.0% |
| Disposition | |
| Home | 92.4% |
| Facility | 6.9% |
| Death | 0.5% |
∗ Race was missing in 20.6% of the study population.
† The Charlson/Deyo co-morbidity index was calculated using ICD-9 codes—Deyo RA, Cherkin DC, Coil MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–619.
‡ This represents a quartile classification of the estimated median household income of residents in the patient’s zip code. These values are derived from zip code demographic data obtained from Claritas. The quartiles are identified by values of 1 to 4, indicating the poorest to wealthiest populations. Because these estimates are updated annually, the value ranges vary by year. Available at: http://www.hcup-us.ahrq.gov/db/vars/zipinc_qrtl/nisnote.jsp .
§ For each year, cost was adjusted for inflation according to reference 2010 cost. Cost was missing in 16.5% of cases.
¶ Periprocedural complications were coded using PSI codes http://www.qualityindicators.ahrq.gov/Modules/psi_overview.aspx and ICD-9 diagnosis codes, for complications not coded by PSI.
The overall postprocedural mortality was low at 0.5% and remained relatively steady over the time frame of the study. Similarly, the complication rate was also steady around an overall rate of 8% from 2006 to 2010 ( Supplementary Table 3 ). However, both the rate of postprocedural mortality and complications were noticed to decrease among symptomatic patients ( Figure 1 ) from 2006 to 2010. The observed rate of iatrogenic stroke after the procedure was 1.6%. Only 2% of the patients had iatrogenic cardiac complications after the procedure. The overall rate of vascular (3.7%) and renal (0.1%) complications was low, and only 0.8% of the patients had any significant postoperative hemorrhage requiring transfusion.
Higher CCI was associated with a higher rate of death or complications (for CCI ≥2: odds ratio [OR] 1.34, 95% confidence interval [CI] 1.09 to 1.66, p value <0.01; Table 2 ). Female gender was predictive of a higher rate of postoperative mortality and complications (OR 1.47, 95% CI 1.23 to 1.76, p value <0.001). A higher annual operator volume was associated with a lower rate of mortality and complications (tertile 2 [5 to 13 annual procedures]: OR 0.74, 95% CI 0.59 to 0.92, p value = 0.01; tertile 3 [14 to 68 annual procedures]: OR 0.70, 95% CI 0.54 to 0.91, p value = 0.01). Besides, symptomatic patients had a higher rate of postprocedural complications and mortality (OR 2.17, 95% CI 1.78 to 2.65, p value <0.001). In subgroup analysis, higher annual operator volume continues to be a significant independent predictor of lesser postprocedural mortality and complication in the following subgroups: highest (third) annual hospital volume tertile and asymptomatic patients ( Table 3 ).
| Variable | Any Complication/Death | Length of Stay | Cost of Hospitalization | |||
|---|---|---|---|---|---|---|
| OR (95% CI) | p Value | Days (95% CI) | p Value | Dollar ($) (95% CI) | p Value | |
| Any complication | +2.48 (2.20–2.76) | <0.001 | +7466 (5818–9115) | <0.001 | ||
| Age | 1.01 (0.997–1.01) | 0.16 | +0.01 (0.00–0.01) | 0.22 | −126 (−185 to −67) | <0.001 |
| Female | 1.47 (1.23–1.76) | <0.001 | +0.08 (−0.07 to 0.23) | 0.30 | +36 (−1091 to 1163) | 0.95 |
| CCI (referent = 0) | ||||||
| 1 | 0.94 (0.74–1.18) | 0.57 | +0.09 (−0.09 to 0.27) | 0.32 | +232 (−1191 to 1656) | 0.749 |
| ≥2 | 1.34 (1.09–1.66) | 0.01 | +0.76 (0.58–0.94) | <0.001 | +1603 (236–2970) | 0.022 |
| Primary payer (referent = Medicare/Medicaid) | ||||||
| Private | 1.03 (0.81–1.32) | 0.80 | −0.24 (−0.44 to −0.03) | 0.02 | −1644 (−3168 to −120) | 0.034 |
| Other | 1.03 (0.63–1.67) | 0.92 | +0.30 (−0.13 to 0.72) | 0.17 | −3337 (−6225 to −448) | 0.024 |
| Weekend admission | 1.27 (0.86–1.88) | 0.24 | +2.29 (1.88–2.70) | <0.001 | +2828 (748–4908) | 0.008 |
| Hospital region (referent = West) | ||||||
| North East | 1.06 (0.73–1.54) | 0.74 | +0.14 (−0.36 to 0.63) | 0.59 | −2029 (−4735 to 678) | 0.142 |
| Mid West/North Central | 1.08 (0.72–1.62) | 0.72 | −0.22 (−0.75 to 0.32) | 0.43 | −2683 (−4869 to −497) | 0.016 |
| South | 1.05 (0.76–1.46) | 0.77 | −0.10 (−0.54 to 0.34) | 0.65 | −742 (−3505 to 2020) | 0.598 |
| Hospital volume tertile {referent = 1 (1–18)} | ||||||
| 2 (19–38) | 0.98 (0.77–1.26) | 0.88 | +0.29 (0.02–0.57) | 0.04 | −54 (−1828 to 1720) | 0.952 |
| 3 (39–142) | 1.01 (0.77–1.33) | 0.90 | +0.26 (−0.06 to 0.58) | 0.12 | −566 (−2680 to 1549) | 0.6 |
| Operator volume tertile {referent = 1 (1–4)} | ||||||
| 2 (5–13) | 0.74 (0.59–0.92) | 0.01 | −1.11 (−1.31 to −0.91) | <0.001 | −2956 (−4479 to −1434) | <0.001 |
| 3 (14–68) | 0.70 (0.54–0.91) | 0.01 | −1.17 (−1.40 to −0.94) | <0.001 | −2778 (−4508 to −1047) | 0.002 |
| Elective admission | 0.92 (0.74–1.13) | 0.42 | −2.34 (−2.54 to −2.14) | <0.001 | −4234 (−5609 to −2858) | <0.001 |
| Symptomatic | 2.17 (1.78–2.65) | <0.001 | +1.30 (1.11–1.50) | <0.001 | +3223 (1918–4529) | <0.001 |
| Highest (Third) Annual Hospital Volume Tertile | Symptomatic | Asymptomatic | ||||
|---|---|---|---|---|---|---|
| OR (95% CI) | p Value | OR (95% CI) | p Value | OR (95% CI) | p Value | |
| Age | 1.02 (1.00–1.03) | 0.07 | 1.02 (1.00–1.04) | 0.02 | 1.00 (0.99–1.01) | 0.90 |
| Female | 1.65 (1.21–2.25) | <0.01 | 1.34 (0.99–1.83) | 0.06 | 1.53 (1.23–1.91) | <0.001 |
| CCI (referent = 0) | ||||||
| 1 | 0.87 (0.58–1.30) | 0.49 | 0.86 (0.57–1.31) | 0.49 | 1.00 (0.76–1.31) | 0.98 |
| ≥2 | 1.24 (0.86–1.78) | 0.25 | 1.25 (0.88–1.79) | 0.22 | 1.42 (1.09–1.84) | 0.01 |
| Operator volume tertile {referent = 1 (1–4)} | ||||||
| 2 (5–13) | 0.59 (0.37–0.94) | 0.02 | 0.62 (0.41–0.95) | 0.03 | 0.79 (0.60–1.04) | 0.10 |
| 3 (14–68) | 0.63 (0.42–0.92) | 0.02 | 0.74 (0.48–1.14) | 0.17 | 0.71 (0.52–0.97) | 0.03 |
| Symptomatic | 2.41 (1.70–3.42) | <0.001 | ||||
The average LOS was 2.52 ± 0.02 days, and the mean cost of hospitalization was $14,366 ± 46. As seen in Figure 2 , although the average LOS has decreased since 2006, the cost of hospitalization has increased from 2006 to 2010, mostly due to increase in the cost of hospitalization for symptomatic patients. A high burden of co-morbidities represented by a high CCI score was predictive of lengthier hospital stay (for CCI ≥2: OR 0.76, estimate +0.76 days, 95% CI 0.58 to 0.94, p value <0.001) and increased the cost of hospitalization by $1,603. Any postprocedural complications increased LOS by an average of 2.48 days (95% CI 2.2 to 2.76, p value <0.001) and increased the cost of hospitalization by $7,466 ( Table 2 ). Private insurance holders had shorter LOS by 0.24 days (95% CI −0.44 to −0.03, p value = 0.02) with hospitalization costs reduced by $1,644. Weekend admissions were associated with an increased LOS by an average of 2.29 days (95% CI 1.88 to 2.70, p value <0.001) and increased cost ($2,828), whereas elective admission reduced the LOS by 2.34 days (95% CI −2.54 to −2.14, p value <0.001) and hospitalization costs by $4,234. Similarly, symptomatic patients had a longer LOS by an average of 1.30 days (95% CI 1.11 to 1.50, p value <0.001) and greater hospitalization costs ($3,223). Importantly, a higher annual operator volume predicted shorter LOS (tertile 2 [5 to 13 annual procedures]: estimate −1.11 days, 95% CI −1.31 to −0.91, p value <0.01 and tertile 3 [14 to 68 annual procedures]: estimate −1.17 days, 95% CI −1.40 to −0.94, p value <0.001) and reduced cost of hospitalization (tertile 2: −$2,956 and tertile 3: −$2,778; Tables 4 and 5 ).
