Pacemakers (PMs), automatic implantable cardioverter-defibrillators (AICDs), and cardiac resynchronization therapy systems (CRTs) are medical devices that regulate cardiac rate and rhythm and coordinate myocardial contraction.1 Overall, these cardiac devices have been shown to improve symptoms, quality of life, and survival,2,3 fueling increasing enthusiasm for their use.4‐8 The increased use, along with concerns about safety and effectiveness of the devices and the financial incentives associated with their use, has increased the need for data to track utilization and outcomes.
Multiple sources of data on cardiac devices exist. First, clinical trials are continually being conducted to study new indications and technologies.3,9‐13 Second, reports of case series and retrospective reviews of medical records at local institutions offer lessons learned from practice about operative procedures and prevention of complications.14‐18 Third, device registries have been established to collect information on devices, operators, and implantation techniques, as well as on some aspects of outcomes.19 For example, the Center for Medicare & Medicaid Services (CMS) established an AICD registry in January 2005 and required hospitals to submit data on every implantation for Medicare payment.20,21 Fourth, the Food and Drug Administration (FDA), through MedSun and other reporting mechanisms, collects data from operators and manufacturers on device flaws, malfunctions, and adverse events, and issues advisories and recalls.22‐26 Fifth, surveys of operators and manufacturers have been conducted to monitor use in the Unites States4,5,27 and worldwide.28 Lastly, administrative data or hospital claims data have been used to track utilization in broad patient populations.6,7
Anzeige
Each of these data sources has advantages and weakness. Clinical trials provide robust data on efficacy, but are limited by their choice of patients and clinical settings. Case reports and retrospective reviews are rooted in real experience, but are limited in generalizability and by the size of observation sets. Registries usually have narrow focuses, such as on device flaws and malfunction. FDA voluntary reports are limited to what is voluntarily reported. Although lacking in clinical details and susceptible to coding errors, administrative data have several advantages over other data sources: primarily large size and nationwide coverage.29 Administrative data-based analysis is a convenient and efficient method when properly approached and can provide valuable information to supplement surveys, registries, case series, and clinical trials to study utilization, patient and hospital characteristics, patient outcomes, and associated factors.
This study explores the use of nationwide administrative data to assess the incidence of PM, AICD and CRT implantations, patient and hospital characteristics, selected perioperative outcomes, and relationships between them for the period 1997–2004. This study does not address clinical indications or the clinical benefits of the devices because of the nature of the data but rather focuses on nationwide utilization, characteristics and outcomes ascertainable from administrative data to provide a comprehensive scan of cardiac device utilization in the United States and pave the way for refined analyses that focus on specific clinical issues.
METHODS
Data and Variables
The primary source of data for this study was the 1997–2004 Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Samples (NIS) developed by the Agency for Healthcare Research and Quality (AHRQ).30 Each annual sample contained about 8 million uniform hospital discharge abstracts from more than 900 short-term general hospitals across more than 30 participating states, approximating a 20% stratified sample of nonfederal acute care hospitals in the United States. NIS includes variables on source and type of admission, 15 diagnosis codes as classified in the International Classification of Diseases, 9th Edition, Clinical Modification (ICD-9-CM), 15 ICD-9-CM procedure codes, procedure days from admission, discharge status, length of stay (LOS), total charges, patient demographic characteristics, insurance coverage, and a few hospital characteristic variables (e.g., hospital procedure volume, size, ownership, and location). The database also included sampling weights and design variables for generating national estimates.
Three additional variables—costs, comorbidities, and hospital volume—were created based on variables from HCUP data. Hospital-specific cost-to-charge ratios, developed by HCUP,30 were used to convert charges to costs. The Elixhauser method31 was used to define 30 comorbid diseases based on diagnosis codes. Defining hospital volume was not as straightforward. Despite recognition of the importance of surgeon procedure volume,32‐34 no agreement exists as to what constitutes high, medium, or low volume. Furthermore, because NIS does not identify surgeons, only hospital volume could be constructed. Based on the literature and initial exploration of the data, we summed the number of elective total cardiac device implantations performed, including primary systems of pacemaker, CRT, and AICD, by each hospital to categorize hospital volume. A “low-volume hospital” performed fewer than 100 new cardiac device implantations; a “high-volume hospital” performed 300 or more.
Anzeige
Identification of Patients with Cardiac Device Implantation
Cardiac device implantation was identified from ICD-9-CM procedure codes. Initial exploration showed that of 71,201 discharges with cardiac device procedure codes in the 2003 NIS, 37% had 1 code, 60% had 2, and 3% had 3 to 5 cardiac device procedure codes. The Appendix classifies these patients into 5 groups based on the code combination in each discharge record, and the footnote lists all specific procedure codes. If any uncertainty existed, a patient was placed into the “Other” group. The validity of these codes or this grouping has not been examined, but, given the clinical value of these procedures, it is unlikely that the procedure would not be coded in the discharge summary if it were performed, or, conversely, that the procedure would be coded if it were not performed. However, errors in choosing specific codes might occur, therefore some misclassification of patients was expected.
Patient Outcome Measures
Three groups of outcome measures were constructed. The first group included length of stay, hospital charges, and in-hospital deaths, which were available from the source data. The second group was measures of complications or adverse events based on the AHRQ Patient Safety Indicators (PSIs).35 The AHRQ PSIs include 20 indicators with reasonable validity, specificity, and potential for fostering quality improvement. Details on the development and validation of these indicators, the variables and ICD-9-CM codes used to define these PSIs, and the computer programs applying the PSIs to hospital discharge data were downloaded from the AHRQ website.35 Our analysis included the following 6 PSIs of primary concern to patients who undergo cardiac device implantation: iatrogenic pneumothorax (PSI 6), postoperative hemorrhage or hematoma (PSI 9), postoperative pulmonary embolism or deep vein thrombosis (PSI 12), infection caused by medical care (PSI 7), postoperative sepsis (PSI 13), and accidental puncture/laceration (PSI 15). The third group of outcome measures was based on ICD-9-CM codes for mechanical complications caused by PM (99601) or by AICD (99604), recorded in the current hospitalization as a secondary diagnosis. Finally, a composite outcome measure was created to indicate whether a patient had any of the 7 types of complications identified through the PSIs and the ICD-9-CM codes for mechanical complications.
Statistical Method
National estimates on the numbers of cardiac device implantations by patient and hospital characteristics were estimated and tabulated. The diagnoses codes were examined and tabulated. Weighted estimates and standard errors were calculated for patient outcome measures. Student’s t tests were used to determine statistically significant difference between 2 estimates when needed.
We conducted a series of multivariable regressions to explore the relationship between patient outcomes and patient and hospital characteristics such as patient comorbid conditions and hospital volume. General linear regressions were used to estimate the effects of patient and hospital characteristics on continuous outcomes variables (i.e., LOS and costs), and logistic regressions were used to estimate the effects of the characteristics on dichotomous outcomes variables (i.e., mortality and complication measures). P < 0.05 and p < 0.01 were considered statistically significant and highly significant, respectively.
RESULTS
National Estimates of Cardiac Device Implantation and Patient Characteristics
Table 1 shows the national estimates of primary PM, AICD, and CRT systems implanted in 1997 to 2004 in the United States. The last column presents the number of patients who had 1 or more cardiac device procedures but could not be placed with certainty in the first 4 groups. From 1997 through 2004 there was a 145% increase in the yearly number of new AICD implantations, but an increase of only 24% in pacemaker implantations. PM implantations leveled off after 2001 and demonstrated a small decline in 2004. New AICD implants also leveled off a year later, after 2002. After approval by the FDA in 2001, CRT-D and CRT-P implantations increased quickly.
Table 1
National Estimates of Cardiac Device Implantation, 1997–2004
Year | CRT-D | CRT-P | AICD | Pacemaker | Other |
|---|---|---|---|---|---|
1997 | 0 (0) | 0 (0) | 26,922 (2,333) | 144,765 (5,416) | 50,918 (2,735) |
1998 | 0 (0) | 0 (0) | 28,260 (2,255) | 147,695 (5,949) | 41,387 (1,955) |
1999 | 0 (0) | 0 (0) | 32,944 (3,168) | 155,182 (6,438) | 46,595 (2,711) |
2000 | 0 (0) | 0 (0) | 39,334 (2,698) | 164,845 (6,054) | 49,488 (2,436) |
2001 | 0 (0) | 0 (0) | 47,962 (3,988) | 188,358 (7,096) | 55,435 (3,092) |
2002 | 1,623 (220) | 734 (104) | 66,528 (5,670) | 188,224 (7,534) | 67,593 (4,752) |
2003 | 18,761 (1,714) | 6,697 (536) | 62,200 (4,190) | 182,597 (6,590) | 70,089 (2,643) |
2004 | 32,737 (2,760) | 7,325 (644) | 66,545 (4,416) | 178,816 (6,528) | 60,118 (3,200) |
Table 2 displays the associated patient and hospital characteristics for cardiac device implants performed in 2004. Patients aged 65 or over accounted for 70% of CRT patients, 60% of AICD patients, and over 85% of PM patients. Three-quarters of CRT and AICD patients were male, whereas about half of the PM patients were male. Whites accounted for more than half of the patients. At least 75% of the patients had 1 or more chronic conditions. Approximately half of CRT implants were planned admissions, whereas most AICD and PM implants were during emergent admissions. Medicare was billed for three-quarters of all cardiac device implants. Most CRT and AICD implants were performed in large teaching hospitals and hospitals that implanted more than 300 primary PM, AICD, or CRT systems a year. PM implants were most often done in urban nonteaching hospitals, and equally distributed among low, medium, and high-volume hospitals.
Table 2
Patient and Hospital Characteristics for Cardiac Device Implantations, 2004
CRT-D | CRT-P | AICD | Pacemaker | Other | |
|---|---|---|---|---|---|
Total Sample Size | 6,752 | 1,509 | 13,577 | 36,818 | 12,370 |
Age (%) | |||||
0–24 | – | – | 1 | – | 2 |
25–64 | 28 | 20 | 41 | 13 | 23 |
65–74 | 33 | 26 | 30 | 22 | 24 |
75–84 | 34 | 38 | 25 | 42 | 35 |
85+ | 4 | 15 | 3 | 22 | 16 |
Sex (%) | |||||
Male | 76 | 61 | 76 | 50 | 59 |
Race (%) | |||||
White | 59 | 58 | 57 | 63 | 62 |
Black | 6 | 6 | 8 | 6 | 7 |
Other or missing | 34 | 36 | 35 | 31 | 31 |
Comorbid diseases (%) | |||||
0 | 24 | 25 | 22 | 15 | 24 |
1 | 32 | 29 | 32 | 31 | 31 |
2 | 26 | 25 | 25 | 28 | 25 |
3 or more | 19 | 20 | 21 | 27 | 20 |
Admission type (%) | |||||
Emergency | 44 | 47 | 54 | 65 | 49 |
Planned | 50 | 48 | 38 | 25 | 42 |
Primary payer (%) | |||||
Medicare | 73 | 79 | 60 | 82 | 74 |
Medicaid | 3 | 4 | 2 | 2 | 5 |
Private | 21 | 15 | 30 | 13 | 18 |
Other | 3 | 2 | 5 | 2 | 3 |
Hospital control (%) | |||||
Government (n = 77) | 6 | 9 | 7 | 9 | 9 |
Private, not for profit (n = 417) | 83 | 78 | 85 | 76 | 78 |
Private, investor-owned (n = 113) | 10 | 13 | 8 | 15 | 12 |
Hospital size (%) | |||||
Small (N = 131) | 13 | 14 | 7 | 10 | 10 |
Medium (N = 200) | 14 | 16 | 13 | 21 | 16 |
Large (N = 284) | 73 | 70 | 80 | 69 | 74 |
Location/teaching status (%) | |||||
Rural (N = 146) | 3 | 8 | 2 | 10 | 6 |
Urban nonteaching (N = 312) | 33 | 34 | 31 | 48 | 39 |
Urban teaching (N = 157) | 64 | 58 | 66 | 43 | 55 |
New Cardiac Device Implantations per Year | |||||
<100 (N = 444) | 7 | 12 | 7 | 32 | 23 |
100–299 (N = 123) | 35 | 42 | 37 | 37 | 37 |
300 or more (N = 48) | 60 | 46 | 56 | 30 | 40 |
More than 60% of patients with a new CRT had congestive heart failure as the principal diagnosis (table available upon request). The proportion of patients who had congestive heart failure as either the principal or a secondary diagnosis was 93%, 88%, 51%, 28%, and 41% in the 5 groups, respectively. Cardiac dysrhythmias were the most frequent primary diagnosis for new AICD and PM implants.
Short-term Patient Outcomes following Cardiac Device Implantations
Table 3 presents the national estimates for short-term patient outcomes for 2004. The length of stay was about 6 days for implanting primary PM, AICD, or CRT systems, but the charges and costs for CRT and AICD implantation were double that for PMs. In-hospital mortality risks were about 1% for CRTs and AICDs and slightly higher for PMs, but the difference was not statistically significant (p > 0.05). The 7 measures of complications each occurred in 1% or fewer of admissions. One or more complications occurred in 2–4% of new implantations. Mechanical complications were substantially more frequently reported in hospitalizations with cardiac device procedures that could not be determined to represent new implantations.
Table 3
In-hospital Patient Outcomes After Cardiac Device Implantation, Unadjusted, 2004
Patient outcomes | CRT-D | CRT-P | New AICD | Pacemaker | Other |
|---|---|---|---|---|---|
Mean LOS (days) | 5.2 (0.2) | 5.5 (0.2) | 6.2 (0.2) | 5.9 (0.1) | 6.5 (0.2) |
Mean charges ($1,000) | 117 (6) | 76 (4) | 110 (5) | 52 (1) | 66 (2) |
Mean costs ($1000) | 42 (1) | 28 (1) | 40 (1) | 19 (1) | 24 (1) |
In-hospital mortality (%) | 0.93 (0.12) | 1.40 (0.32) | 0.75 (0.08) | 1.12 (0.07) | 1.86 (0.16) |
Iatrogenic pneumothorax (%) | 0.94 (0.14) | 0.66 (0.20) | 0.77 (0.08) | 1.04 (0.06) | 0.87 (0.08) |
Postoperative hemorrhage or hematoma (%) | 0.28 (0.06) | 0.06 (0.06) | 0.19 (0.04) | 0.04 (0.01) | 0.52 (0.08) |
Postoperative pulmonary embolism or deep venous thrombosis (%) | 1.17 (0.14) | 1.48 (0.29) | 1.13 (0.10) | 0.24 (0.03) | 1.80 (0.15) |
Infection due to medical care (%) | 0.27 (0.06) | 0.40 (0.18) | 0.45 (0.05) | 0.28 (0.03) | 0.38 (0.06) |
Postoperative sepsis (%) | 0.16 (0.05) | 0.20 (0.11) | 0.07 (0.03) | 0.05 (0.01) | 0.26 (0.06) |
Accidental puncture/laceration (%) | 0.44 (0.08) | 0.26 (0.13) | 0.36 (0.06) | 0.31 (0.04) | 0.57 (0.08) |
Mechanical complication (%) | 0.90 (0.14) | 1.16 (0.28) | 0.39 (0.06) | 0.23 (0.03) | 11.35 (0.49) |
Any of the above 7 complications (%) | 3.96 (0.32) | 3.76 (0.45) | 3.20 (0.19) | 2.14 (0.09) | 14.94 (0.54) |
Anzeige
Table 4 highlights selected outcome estimates for AICD patients for the years 1997 to 2003 (tables for other comes and other patient groups available upon request). Mean LOS decreased continuously, whereas charges nearly doubled. In-hospital mortality rates decreased and complication rates fluctuated slightly during the period.
Table 4
In-hospital Patient Outcomes After AICD Implantation, Unadjusted, 1997–2003
Patient outcomes | 1997 | 1998 | 1999 | 2000 | 2001 | 2002 | 2003 |
|---|---|---|---|---|---|---|---|
Mean LOS (days) | 9.9(0.4) | 8.1(0.2) | 8.3(0.3) | 7.7(0.2) | 7.4(0.2) | 6.8(0.2) | 6.4(0.2) |
Mean Charges ($1000) | 66 (2) | 69 (2) | 72 (2) | 78 (2) | 86 (3) | 101 (4) | 115 (4) |
In-hospital mortality (%) | 0.7(0.1) | 0.8(0.1) | 1.0(0.1) | 0.9(0.1) | 0.8(0.1) | 1.0(0.1) | 0.9(0.1) |
Any of above 7 complications (%) | 2.8(0.3) | 2.9(0.3) | 3.1(0.3) | 2.3(0.2) | 3.0(0.2) | 3.4(0.2) | 3.3(0.2) |
Relationship between Patient Outcomes and Patient and Hospital Characteristics
Table 5 presents the effects of patient and hospital characteristics on in-hospital mortality for the 5 patient groups. Across the 5 patient groups, in-hospital mortality was higher among patients who were older, under Medicare coverage, with more comorbid diseases, or admitted through the emergency department. Smaller hospitals appeared to have lower in-hospital mortality rates, whereas urban teaching hospitals had higher in-hospital mortality rates. The data did not show a consistent volume–mortality relationship across the 5 patient groups.
Table 5
The Effects of Patient and Hospital Characteristics on In-hospital Mortality (Odds Ratios)
CRT-D | CRT-P | AICD | Pacemaker | Other | |
|---|---|---|---|---|---|
Age (%) | |||||
25–64* | Ref | Ref | Ref | Ref | Ref |
65–74 | 0.76 | 0.61 | 0.66 | 0.88 | 1.14 |
75–84 | 1.43† | 0.73 | 1.18‡ | 1.02 | 1.07 |
85+ | 2.32‡ | 0.90 | 0.85 | 1.56‡ | 0.96 |
Sex (%) | |||||
Male | Ref | Ref | Ref | Ref | Ref |
Female | 0.68‡ | 0.67 | 1.43‡ | 0.96 | 0.71‡ |
Race (%) | |||||
White | Ref | Ref | Ref | Ref | Ref |
Black | 0.50† | 1.98† | 1.04 | 1.24‡ | 1.08 |
Other or missing | 1.39‡ | 1.23 | 0.94 | 0.93 | 1.05 |
Comorbid diseases (%) | |||||
0 | Ref | Ref | Ref | Ref | Ref |
1 | 0.47‡ | 1.56 | 1.03 | 1.46‡ | 1.57‡ |
2 | 1.83‡ | 2.15† | 1.31 | 1.90‡ | 2.31‡ |
3 or more | 3.96‡ | 4.87‡ | 2.97‡ | 3.30‡ | 4.27‡ |
Admission type (%) | |||||
Planned | Ref | Ref | Ref | Ref | Ref |
Emergency | 1.54‡ | 2.50‡ | 1.72‡ | 1.70‡ | 1.91‡ |
Primary payer (%) | |||||
Private | Ref | Ref | Ref | Ref | Ref |
Medicare | 1.01 | 1.17 | 2.64‡ | 1.46‡ | 1.03 |
Medicaid | 0.93 | 0.23 | 1.43 | 1.08 | 1.11 |
Other | 0.22 | 2.48 | 2.07‡ | 1.29 | 1.47‡ |
Hospital control (%) | |||||
Government (n = 77) | Ref | Ref | Ref | Ref | Ref |
Private, not for profit (n = 417) | 0.71 | 2.40 | 0.78 | 0.94 | 1.09 |
Private, Investor-owned (n = 113) | 1.00 | 3.04 | 0.79 | 0.72‡ | 1.42‡ |
Hospital size (%) | |||||
Small (N = 131) | 0.48‡ | 0.90 | 0.97 | 0.94‡ | 0.72† |
Medium (N = 200) | Ref | Ref | Ref | Ref | Ref |
Large (N = 284) | 0.58‡ | 2.32† | 0.98 | 1.09 | 1.11 |
Location/teaching status (%) | |||||
Rural (N = 146) | 2.31‡ | 1.79 | 0.46 | 0.96 | 1.05 |
Urban nonteaching (N = 312) | Ref | Ref | Ref | Ref | Ref |
Urban teaching (N = 157) | 1.18 | 1.69 | 1.48‡ | 1.22‡ | 1.62‡ |
New Cardiac Device Implantations per Year | |||||
<100 (N = 444) | Ref | Ref | ref | ref | ref |
100–299 (N = 123) | 2.22† | 0.75 | 1.18 | 0.78‡ | 1.05 |
300 or more (N = 48) | 1.52 | 0.82 | 1.08 | 0.69‡ | 0.94 |
Table 6 presents the effects of patient and hospital characteristics on occurrence of complications. Again, patient frailty indicated by advanced age and comorbidity was significant. The relationships between other outcome measures and patient and hospital characteristics are not presented, but are available from the authors upon request. Overall, the estimates were similar to that for in-hospital mortality with some notable differences; for example, patients with more comorbid conditions were less likely to suffer from iatrogenic pneumothorax.
Table 6
The Effects of Patient and Hospital Characteristics on Having Any Complication (Odds Ratios)
CRT-D | CRT-P | AICD | Pacemaker | Other | |
|---|---|---|---|---|---|
Age (%)0 | |||||
25–64* | Ref | Ref | Ref | Ref | Ref |
65–74 | 1.08 | 0.79 | 0.79† | 0.88 | 1.11‡ |
75–84 | 0.97 | 0.86 | 1.04 | 1.03 | 1.16† |
85+ | 1.09 | 0.45† | 1.34‡ | 0.86‡ | 0.96 |
Sex (%) | |||||
Male | Ref | ref | ref | ref | ref |
Female | 1.28† | 0.64† | 1.58† | 1.62† | 1.02 |
Race (%) | |||||
White | Ref | ref | ref | ref | ref |
Black | 0.79 | 0.94 | 1.01 | 1.00 | 0.99 |
Other or missing | 0.76† | 0.83 | 0.99 | 1.06 | 1.08† |
Comorbid diseases (%) | |||||
0 | Ref | ref | ref | ref | ref |
1 | 1.18† | 0.90 | 0.87‡ | 0.84† | 1.15† |
2 | 0.99 | 1.75† | 0.99 | 0.84† | 1.30† |
3or more | 1.31† | 1.08 | 1.03 | 0.96 | 1.63† |
Admission type (%) | |||||
Planned | Ref | ref | ref | Ref | Ref |
Emergency | 1.09 | 2.14† | 1.40† | 0.95 | 1.39† |
Primary payer (%) | |||||
Private | Ref | ref | ref | ref | ref |
Medicare | 1.05 | 1.36 | 0.99† | 1.23† | 0.91† |
Medicaid | 0.88 | 1.68 | 1.12 | 1.05 | 1.02 |
Other | 1.39 | 0.82 | 1.39† | 0.97 | 0.96 |
Hospital control (%) | |||||
Government (n = 77) | Ref | ref | ref | ref | Ref |
Private, not for profit (n = 417) | 0.52† | 0.75 | 0.85‡ | 0.90 | 1.02 |
Private, Investor-owned (n = 113) | 0.89 | 0.71 | 1.11 | 0.94 | 1.14† |
Hospital size (%) | |||||
Small (N = 131) | 0.75‡ | 1.64‡ | 0.98 | 0.84‡ | 0.91‡ |
Medium (N = 200) | Ref | ref | ref | ref | ref |
Large (N = 284) | 1.01 | 0.97 | 1.01 | 1.19† | 0.82† |
Location/teaching status (%) | |||||
Rural (N = 146) | 0.92 | 0.88 | 0.16† | 0.86† | 1.28† |
Urban nonteaching (N = 312) | Ref | ref | ref | ref | ref |
Urban teaching (N = 157) | 1.06 | 0.82 | 1.15‡ | 1.14† | 0.85† |
New Cardiac Device Implantations per Year | |||||
<100 (N = 444) | Ref | Ref | Ref | Ref | Ref |
100–299 (N = 123) | 1.23 | 0.80 | 1.10 | 1.39† | 1.16† |
300 or more (N = 48) | 1.59† | 0.78 | 0.83 | 1.42† | 0.94 |
DISCUSSION
This study showed steady increases in the number of AICDs and PMs implanted between 1997 and 2001. After the FDA approval of CRT in 2001, the growth in AICDs and PMs leveled off, and CRT implantations rapidly increased. In 2004, about 33,000 new CRT-D, 7,000 CRT-P, 67,000 new AICD, and 179,000 new PM systems were implanted the United States.
Anzeige
A survey of physicians and device companies estimated that 153,000 PMs and 29,000 AICDs were implanted in 1997.5 Our administrative data-based estimates for the same year were 145,000 PMs and 27,000 AICDs, suggesting that administrative data are an alternative to assessment of cardiac device utilization in the United States.
Most patients who underwent cardiac device implantations were elderly whites with multiple chronic conditions. Most patients had principal diagnoses of congestive heart failure, cardiac dysrhythmia, or conduction disorder, and almost 100% of CRT-D patients had a primary or secondary diagnosis of congestive heart failure. These data suggest that administrative data could be used, to some extent, to assess whether cardiac devices are implanted in patients with proper indications. Further studies could be conducted in narrowly defined patient groups to examine patient indications.
Whereas most CRT and AICD implantations were done in large teaching hospitals that implanted 300 or more primary systems a year, PM implants were done more among nonteaching and low-volume hospitals. More research could be conducted to examine how patient characteristics and outcomes differ by hospital type, for example, whether patient outcomes are better in high-volume hospitals than in low-volume hospitals.
Patients stayed in hospitals for about 5 days for AICD implantation in 2004, a substantial decrease from 9.19 days in 1997. In the meantime, the charges and costs increased steadily, from an average charge of $66,530 for an AICD in 1997 to $114,782 in 2004. CRT-D and AICD implantations were substantially more costly than CRT-P and PM. These cardiac device procedures had a substantially lower than average cost-to-charge ratio of 0.50, suggesting that patient admissions for cardiac device procedures might be more profitable than other hospital admissions. The rapid diffusion, especially the increase in CRT implantations, coupled with the profit potential, raises a question about the proper use of these devices.
Anzeige
Fewer than 2% of the patients died, and fewer than 4% of the patients had complications during hospitalization. As expected, iatrogenic pneumothorax was substantially more frequent in cardiac device implantations compared with that in surgical patients, which was reported at 0.09% in 2003,36 whereas the rates of other complications in cardiac device patients were comparable to those in general surgical patients as documented by the 2006 National Healthcare Quality Report, where postoperative hemorrhage or hematoma, DVT/PE, and accidental puncture were reported at 0.2%, 0.9%, and 0.4%, respectively.36 Our analysis further showed that adverse outcomes were mostly associated with patient frailty indicated by advanced age, comorbidities, and emergency admission, and associated with the complexity of device implantations (i.e., AICDs and CRTs compared to pacemaker implantations). Given that the patients were mostly elderly with multiple chronic conditions, these low risks of death and complications suggest that the implantation procedures were fairly safe.
Administrative data have many limitations when used to study a clinical intervention. Previous studies identified many cardiac device-related complications, such as pocket hematoma,15,37,38 pocket infections,7,38‐43 iatrogenic pneumothorax,40,41 arterial puncture,40 venous thrombosis and stenosis,41 electrode displacement,40 lead dislodgements, undersensing,40,42 cardiac device endocarditis,44,45 interference by an electronic antitheft-surveillance device46 or a Personal Digital Assistant,17 and twiddle-induced torsion of leads.41 Administrative data may not able to identify many of these specific types of complications. The data do not capture deaths or complications that are detected after discharge. The data do not capture the risk associated with the experience of the surgeon, type of device used (such as dual-chamber pacemaker or biventricular pacemaker, both of which have different risks of complications),47 the experience of operating room staff, type of anesthesia used, and other clinical variables. Nevertheless, our study demonstrates that administrative data provide an efficient and reliable source to track utilization of cardiac devices, to evaluate associated patient and hospital characteristics, and to offer valuable insights into patient risks and outcomes after implantations.
Acknowledgments
The authors thank Zhengyi Fang, Social and Scientific Systems, Inc., for excellent computing support. We thank Parivash Nourjah, PhD, and Joyce Basu, PhD, the Agency for Healthcare Research and Quality, for useful comments. The authors would also like to acknowledge the statewide data organizations that participate in the HCUP Nationwide Inpatient Sample: California Office of Statewide Health Planning & Development; Colorado Health & Hospital Association; Connecticut - Chime, Inc.; Florida Agency for Health Care Administration; Georgia—GHA: An Association of Hospitals & Health Systems; Hawaii Health Information Corporation; Illinois Department of Public Health; Iowa Hospital Association; Kansas Hospital Association; Kentucky Department for Public Health; Maine Health Data Organization; Maryland Health Services Cost Review Commission; Massachusetts Division of Health Care Finance and Policy; Michigan Health and Hospital Association; Minnesota Hospital Association; Missouri Hospital Industry Data Institute; Nebraska Hospital Association; Nevada Department of Human Resources; New Jersey Department of Health & Senior Services; New York State Department of Health; North Carolina Department of Health and Human Services; Ohio Hospital Association; Oregon Association of Hospitals & Health Systems; Pennsylvania Health Care Cost Containment Council; Rhode Island Department of Health; South Carolina State Budget & Control Board; South Dakota Association of Healthcare Organizations; Tennessee Hospital Association; Texas Health Care Information Council; Utah Department of Health; Vermont Association of Hospitals and Health Systems; Virginia Health Information; Washington State Department of Health; West Virginia Health Care Authority; Wisconsin Department of Health & Family Services.
Disclaimer
The authors of this article are responsible for its contents. No statement in this article should be construed as an official position of the Agency for Healthcare Research and Quality, the U.S. Department of Health and Human Services.
Conflict of Interest
None disclosed.
Open Access This is an open access article distributed under the terms of the Creative Commons Attribution Noncommercial License (
https://creativecommons.org/licenses/by-nc/2.0
), which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.