Abstract
Background and Purpose:
To investigate the factors associated with outcomes and medical costs for percutaneous nephrolithotomy (PCNL).
Methods:
The present study uses a subset of the National Health Insurance Research Database (NHIRD), known as the Longitudinal Health Insurance Database 2005 (LHID 2005), which contains the data of all medical benefit claims from 1997 to 2010 for a subset of 1 million enrollees randomly drawn from the population of 22.72 million persons who were enrolled in 2005. The claims data for all subjects with a diagnosis of urolithiasis who underwent PCNL were analyzed. Hospital and surgeon case volume were classified by quartile. The correlations of all patient, surgeon, and hospital variables with the outcomes and medical costs of PCNL were analyzed by generalized estimating equations.
Results:
A total of 995 subjects received PCNL. In univariate analysis, PCNL performed by high-volume surgeons (≥12) cost 26% less ($2684 vs $1986) and resulted in a 34.3% shorter hospital stay (6.5 vs 9.9 days) compared with low-volume surgeons (≤3). In multivariate analysis, surgeon volume was a significant predictor for medical cost, length of stay, and intensive care unit transfer but not complications and mortality.
Conclusions:
Surgeon volume was associated with lower medical costs and shorter length of stay after PCNL. Surgeon volume, however, was not an independent predictor of complications and mortality. Our findings have important implications for urologists and policymakers with regard to the cost and effectiveness of PCNL.
Introduction
U
PCNL has been reported to have a low rate of complications, fast convalescence, and good efficacy. 4 –7 Most complications for PCNL are minor and can be managed conservatively if recognized early. 7 Nevertheless, once complications occur, they lengthen the hospital stay, increase medical costs, and reduce the quality of care of the patients.
Previous studies has indicated that successful PCNL relies heavily on the surgeon's technique and experience, as with other surgeries. 8 Above all, there existed a unique utilization pattern for PCNL which was mostly performed by young urologists in practice 9 and was performed more commonly at teaching and urban hospitals with higher case volumes. 10 Most important of all, few studies have analyzed the impact of the surgeon and hospital volume on the outcomes and medical costs of PCNL together with patient factors.
The high surgeon volume provides the opportunity for surgeon to master technically demanding surgery. Like other urologic cancer and laparoscopic surgeries, PCNL has been reported to have a steep learning curve leading to a higher complication rate in the beginning of the experience. 11
In this study, we used a single-payer, nationwide, population-based health insurance database in Taiwan to investigate the factors associated with the outcomes and medical costs for PCNL.
Methods
National Health Insurance Research Database
The Taiwan National Health Insurance Administration had collected claims records covering all inpatient and outpatient medical benefit claims for nearly the entire population of Taiwan since the inception of its single-payer National Health Insurance (NHI) program in 1995. As of 2005, the Taiwan NHI program covered nearly 99% of the nation's population. This entire data collection program is known as NHIRD. The NHIRD includes a registry for contracted medical facilities, a registry for board-certified specialists, a monthly claims summary for inpatient claims, and details of inpatient orders and principal operational procedures.
The present study used a subset of the NHIRD known as the Longitudinal Health Insurance Database 2005 (LHID2005), which contains all inpatient and outpatient medical benefit claims from 1997 through 2010, for a sample of 1 million beneficiaries randomly drawn from the NHI registry of beneficiaries. The 1 million de-identified persons included in the LHID2005 provide a good statistical representation of the entire population of Taiwan.
Study sample
To assess surgery for PCNL, we first identified all patients enrolled in the NHIRD during the study period (1997–2010) with the diagnosis of upper urinary tract stone disease identified by the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) codes 592.X and 274.11. We then identified those whose claims records included the primary PCNL (76016 B) payment code according to standard payment for surgery from the NHI.
Covariates
For each procedure, we examined all available patient, surgeon, and hospital characteristics that may have been associated with outcomes and medical costs.
Patient characteristics
These included year of surgery, age, sex, and comorbidities based on the Charlson Comorbidity Index (CCI).
Surgeon characteristics
These included sex, age, years of license certification, specialty (urology vs nonurology), and caseload volume.
Hospital characteristics
These included geographic region, urbanization level of the hospital location (metropolitan, nonmetropolitan), ownership of the hospital (public, private, and nonprofit), level of hospital (local, regional, and medical center), and hospital caseload volume.
The surgeon volume was calculated by the number of procedures performed by each surgeon during the 14-year study period. The number of procedures performed at each hospital during the study period was calculated as the hospital volume. The hospital and surgeon volumes were then stratified into quartiles.
Outcomes
Complications
Complications were identified by ICD-9-CM codes. The main intraoperative complications included those involving the bowel, ureter, and vessel injury, and blood transfusion. The main postoperative complications included cardiac, respiratory, vascular, miscellaneous medical and surgical events, and infection. The main urologic complications were hematuria needing further procedures and septicemia.
Medical costs
The medical costs of PCNL were direct medical costs and calculated from the claim records of study subjects, including all prescriptions, drugs, radiology, laboratory tests, surgery-related costs, and all inpatient medical services during PCNL hospitalization. The medical costs were quoted in US dollars. We apply the Consumer Price Index (CPI) adjustments to account for the rate of inflation in medical costs according to the CPI issued by The Directorate General of Budget, Accounting and Statistics, Executive Yuan, Taiwan.
Other outcomes included in-patient mortality, mortality in 180 days, hospital stay, intensive care unit (ICU) transfer, emergency department visit within 3 days, readmission within 14 days, and subsequent stone surgery (extracorporeal shockwave lithotripsy, ureteroscopic or cystoscopic lithotripsy, and ureterolithotomy) within 30 days.
CCI
The CCI predicts mortality for a patient with a range of up to 17 comorbid conditions. Each condition is assigned a score of 1, and the sum of all of the conditions give a total score that predicts mortality. 12
Level of urbanization
The key independent variable in this study was the urbanization level of residence. The 359 communities in Taiwan were stratified into eight urbanization categories according to standards published by the Institute of Occupational Safety & Health in Taiwan, with code one indicating the most urbanized and code eight the least urbanized area. The standards include population density (persons per km2), percentage of persons with college education or higher, percentage of elderly people (older than 65 years), percentage of agricultural workers in the population, and number of physicians per 100,000 persons in the population. 13 –15 In this study, we used two categories: Metropolitan (level 1–2) and nonmetropolitan area (level 3–8). 15
Statistical analysis
Continuous data are expressed as mean±standard deviation. Chi-square was used for univariate analyses as appropriate on the basis of the data characteristics. Multivariate linear regression models were constructed to further examine associations of the predictors of interest with medical costs and outcomes. Model diagnostics revealed no significant violations of regression assumptions. All statistical tests were two-sided, and P values of ≤0.05 were considered to be significant. All statistical analyses were performed using the SAS statistical software package (version 9.1.3; SAS Institute, Inc., Cary, NC).
Results
In total, 995 patients with renal stones underwent PCNL from 1997 to 2010. The medical costs and length of hospital stay based on patient, surgeon, and hospital characteristics are listed in Table 1. Among the patients, 31 (3.1%) had septicemia, 24 (2.4%) were transferred to the ICU after the procedures, 227 (22.8%) had blood transfusions, 202 (20.3%) had coexisting urinary tract infections, 18 (1.8%) had postoperative infections, and 39 (3.9%) had hematuria needing further management. Thirty-eight (3.8%) subjects visited the emergency department within 3 days after PCNL, and 68 (6.8%) subjects were readmitted within 14 days after PCNL. In-patient mortality was noted in one patient (0.1%), and five subjects (0.5%) died within 180 days after PCNL.
USD=United States dollars; SD=standard deviation; CCI=Charlson Comorbidity Index; ICU=intensive care unit.
As shown in Table 1, the mean medical cost for PCNL was $2256. The medical cost for PCNL was significantly associated with some patient characteristics (age and CCI), surgeon characteristics (surgeon volume, age, specialty, and years of license certification), hospital characteristics (hospital level, urbanization level of the location) (all P<0.05). The mean period of hospital stay was 7.8 days. The length of hospital stay was also significantly associated with some patient characteristics (patient age and CCI), surgeon characteristics (surgeon volume, age, specialty, and years of license certification), hospital characteristics (hospital level, urbanization level of the location) (all P<0.05). PCNL performed by high-volume surgeons (≥12) cost 26% less compared with low-volume surgeons (≤3) ($2684 vs $1986), and resulted in a 34.3% shorter hospital stay (6.5 vs 9.9 days).
Table 2 shows the variables associated with medical costs and hospital stay for PCNL by multivariate analysis. A higher surgeon volume (4–6, 7–11 and ≥12) vs low volume (≤3) was significantly associated with lower medical costs (all P<0.05). When PCNL was performed by a urologist compared with other specialties, the medical costs were significantly lower (estimate, -$3461; P<0.0001). When PCNL was performed at a hospital located in a metropolitan area (estimate, 321; P<0.001) and medical center (estimate, 252; P=0.007), the medical costs were significantly higher. In addition, ICU transfer (estimate, $4103; P<0.0001), septicemia (estimate, $1284; P=0.003) and blood transfusions (estimate, $710; P<0.0001) were significantly associated with higher medical costs for PCNL. We then further subcategorize the direct medical costs to total, surgery-related, drug and prescription, and other costs to further analyze the correlations between surgeon volume and individual costs. Consistently, surgeon volume was inversely correlated with total and individual costs as well as length of stay (data not shown).
Statistical method: Stepwise multiple regression model.
The medical costs were US dollars and inflation adjusted at 2006 Consumer Price Index level, USD:NTD=1:32.53
USD=United States dollars; SE=standard error; CI=confidence interval; ICU=intensive care unit.
In terms of the hospital stay for PCNL, a higher surgeon volume (4–6, 7–11, and ≥12) was significantly associated with a shorter hospital stay (estimate, -1.1, -1.4, -2.1; all P<0.05). Similarly, when PCNL was performed by a urologist, the hospital stay was significantly shorter (estimate, -15.7; P<0.0001). When PCNL was performed at a hospital located in a metropolitan area, the hospital stay was significantly longer (estimate, 2.1; P<0.0001). ICU transfer (estimate, 13.6; P<0.0001), septicemia (estimate, 6.6; P=0.002), and blood transfusions (estimate, 2.4; P<0.0001) were significantly associated with a longer hospital stay for PCNL. The inverse correlations existed in surgeon volume vs medical costs and surgeon volume vs hospital stay are shown in supplementary figures (Fig. S1 and S2; Supplementary Data are available online at
Table 3 shows the variables associated with ICU transfer after PCNL. Older patients were more likely to be transferred to the ICU after PCNL (OR, 1.07; P<0.001). Higher surgeon volume and insurance fee ($600∼$699 and ≥$700) were less likely to be transferred to the ICU after PCNL (OR<1; P<0.05). The patients who received blood transfusions and had septicemia were more likely to be transferred to the ICU after PCNL (OR, 10.1 and 39.5; P<0.001, respectively).
Statistical method: Stepwise logistic regression model.
OR=odds ratio; CI=confidence interval; ICU: intensive care unit.
Discussion
The relationship between surgeon volume and outcomes has been widely discussed in various surgical procedures. 16 Most previous studies have reported that the outcomes and complications are inversely related to surgeon volume, and that this is especially true for procedures that depend greatly on technique, such as cardiovascular and cancer surgery. 17 –22 In the field of urology, the association between volume and outcomes has been widely studied in some high-risk oncologic procedures, notably radical prostatectomy and cystectomy, 23,24 yet the relationship between volume and outcome for PCNL has not been well established.
To the best of our knowledge, this is one of the few studies to examine caseload volume on the outcomes of PCNL. This study is also the first to demonstrate the association between surgeon volume and medical costs by using a nationwide, population-based database. A recent study using the Nationwide Inpatient Sample database showed that hospital volume was significantly associated with a shorter hospital stay but not complications or transfusions. 25 The surgeon volume, however, was not included in the analysis of this study. Another previous study from the Clinical Research Office of the Endourological Society (CROES) PCNL group included 3933 patients treated at 96 global centers and found a shorter operative time, lower perforation rate, higher stone-free rate, fewer complications, and lower mortality rate in high-volume centers. 26
In the present study, surgeon volume was a significant predictor for medical costs, length of stay, and ICU transfer but not complications or mortality. It is worth noting that the CROES study excluded all centers with fewer than 10 PCNL procedures, whereas we included all levels of hospitals where PCNL was performed.
Surgeon volume in the present study was inversely correlated with medical costs, length of hospital stay, and transfer to ICU after PCNL but not complications or mortality. PCNL performed by high-volume surgeons (≥12) cost 26% less and had a 34.3% shorter hospital stay compared with low-volume surgeons (≤3). Therefore, if patients undergoing PCNL are treated by high-volume surgeons, cost savings and a shorter hospital stay can be expected.
PCNL has been reported to have a high success rate and low morbidity rate. 7 As shown in the current study, morbidity and mortality related to PCNL are relatively uncommon compared with other major surgeries. Nevertheless, Desai and associates 8 described the learning curve of PCNL and indicated associations between outcomes and the experience of the surgeons. In their study of 773 patients who underwent PCNL for staghorn stones, there was a marked learning curve for PCNL, and the operative time and bleeding continued to decrease with improvements in technique. With increasing experience, there was an exponential decrease in the number of stages needed for stone clearance and the number of tracts needed. The authors concluded that PCNL for staghorn calculi necessitates considerable expertise. 8
These findings are also supported in our present study. A high surgeon volume provides an opportunity for surgeons to master technically demanding surgery. Moreover, these surgeons can coordinate the therapeutic elements and discharge planning, leading to further reductions in hospital stay and costs. In addition, selective referral by physicians or self-referral by patients may be helpful to lower medical costs and hospital stay. This is unlikely within the current health benefit coverage in Taiwan, however, because of fee-for-service reimbursements for surgery and very low out-of-pocket copayments for inpatient care from the patients. New strategies for healthcare reforms to reduce PCNL-related medical expenditure are warranted.
The present study is based on records generated by the very accessible and well-used NHI system in Taiwan. The NHIRD offers some advantages for this study. First, it is a nationwide, population-based database, which includes all medical claims for more than 99% of the population of Taiwan. The results can therefore represent the general population as a whole. Second, the database contains original claims records for the studied population, and it was not designed for academic study; this attenuates bias in patient selection.
There are some limitations to this study. First, detailed demographic and clinical variables were not available in the NHIRD database. For example, we did not have information regarding stone size, position, composition, or the intraoperative findings in the claims records. To ensure the homogeneity of our patient cohort, we only included subjects undergoing primary PCNL for upper tract stones. The larger and complex stone definitely increase the costs and length of stay. Moreover, high volume surgeons have the tendency to operate on larger and complicated stones as is common at academic centers. These PCNLs are more likely to need second-look PCNL. The lack of second-look PCNL data is another limitation to our study.
Nevertheless, the treatment guidelines for renal stones and reimbursement are strictly regulated by Taiwan's NHI Administration, and this could minimize the bias from the surgeons' preferences to ensure uniform indications for PCNL. Moreover, the surgeon volume in this study was calculated by the number of procedures performed by each surgeon during the 14-year study period and then categorized into quartiles. The definition on surgeon volume may not be an exact reflection of surgeon experience; however, it may be only ever at best an imperfect proxy.
In addition, higher transfusion rate and longer hospital stay in this study were exceptive phenomena in Taiwan's NHI system. These findings themselves were unable to extrapolate results to other patient populations in other countries. Most importantly, the NHI Administration has established a uniform system to control the quality of medical services and codes. The medical and claims records of beneficiaries from the contracted medical care institutions are spot checked by the Professional Peer Review Committee. When disputes arise, a Disputes Settlement Board examines whether the claims comply with the provisions of the NHI Act.
Third, the claims data do not necessarily reflect “charge” costs instead of “actual” costs. The NHI Bureau has detailed lists of prices for various hospital service items and consumables as well as audit mechanisms in place that are activated when a hospital's charges are out of the general norm. Therefore, there is no reason to believe that this factor could have materially influenced our findings.
Finally, the differences in post-PCNL morbidity and complication rates between the high-volume and low-volume hospitals may be limited because of the low complication rate attributable to PCNL. It is reasonable to assume that technical differences in PCNL may yield different efficacies of stone clearance. We could not evaluate the stone-free rate, however, because we did not have access to detailed medical records. We analyzed the incidence of subsequent stone-related surgeries within 30 days after PCNL, which did not show a significant correlation with surgeon volume.
Conclusions
According to the results from this population-based cohort study, a high surgeon volume is significantly associated with medical costs, hospital stay, and ICU transfer for primary PCNL in treating patients with renal stone disease. Our findings have important implications for urologists and policymakers with regard to the cost effectiveness of PCNL.
Footnotes
Acknowledgments
This study is based in part on data from the NHIRD provided by the NHI Administration, Ministry of Health and Welfare, Taiwan, and managed by the National Health Research Institutes. The interpretations and conclusions contained herein do not represent those of the NHI Administration, Ministry of Health and Welfare, or the National Health Research Institutes.
Disclosure Statement
All authors have no conflicts of interest to declare.
Abbreviations Used
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
