Contemporary Trends in Percutaneous Nephrolithomy Across New York State: A Review of the Statewide Planning and Research Cooperative System

Abstract

Introduction and Objectives: Percutaneous nephrolithotomy (PCNL) is a complex multistep surgery that has shown a steady increase in use for the past decade in the United States. We sought to evaluate the trends and factors associated with PCNL usage across New York State (NYS). Our goal was to characterize patient demographics and socioeconomic factors across high-, medium-, and low-volume institutions.

Materials and Methods: We searched the NYS, Statewide Planning and Research Cooperative System (SPARCS) database from 2006 to 2014 using ICD-9 Procedure Codes 55.04 (percutaneous nephrostomy with fragmentation) for all hospital discharges. Patient demographics including age, gender, race, insurance status, and length of hospital stay were obtained. We characterized each hospital as a low-, medium-, or high-volume center by year. Patient and hospital demographics were compared and reported using chi-square analysis and Student's t-test for categorical and continuous variables, respectively, with statistical significance as a p-value of <0.05.

Results: We identified a total of 4576 procedures performed from 2006 to 2014 at a total of 77 hospitals in NYS (Table 1). Total PCNL volume performed across all NYS hospitals increased in the past decade, with the greatest number of procedures performed in 2012 to 2013. Low-volume institutions were more likely to provide care to minority populations (21.4% vs 17.3%, p < 0.001) and those with Medicaid (25.5% vs 21.5%, p < 0.001). High-volume institutions provided care to patients with private insurance (42.1% vs 34.0%, p < 0.001) and had a shorter length of stay (3.3 days vs 4.1 days, p < 0.001).

Conclusion: Our data provide insight into the patient demographics of those treated at high-, medium-, and low-volume hospitals for PCNL across NYS. Significant differences in race, insurance status, and length of stay were noted between low- and high-volume institutions, indicating that racial and socioeconomic factors play a role in access to care at high-volume centers.

Introduction

Percutaneous nephrolithotomy (PCNL) is the gold standard treatment modality for renal stones >2 cm and lower pole stones >1 cm.^1,2 PCNL is associated with high stone clearance rates up to 95%, but comes at the cost of increased complications.^3,4 Complication rates can be as high as 15%, including severe bleeding, infection, urinary extravasation, and injury to adjacent organs, most notably the colon.^5,6

Leow et al. studied the trends in utilization of PCNL across the United States and noted an increase from 6.7% (2003) to 8.9% (2008) before plateauing at 9.0% (2008–2011), and later declining to 7.2% (2014). Furthermore, the annual incidence of PCNL declined from 5.49/100,000 in 2003 to 5.34/100,000 in 2014. Although utilization has decreased, complication rates of the procedure have increased from 6.6% in 2003 to 33.1% in 2014.⁷ Although these results provide insight regarding the climate of PCNL in the entire nation, it is often difficult to extrapolate the findings to specific states and regions. Furthermore, there is limited data on socioeconomic factors and insurance status playing a role with these findings. Review of a national database revealed that patients undergoing PCNL with public insurance (Medicaid or Medicare) were more likely to experience a complication than privately insured patients.⁸ This is likely attributed to these patients exhibiting more medical comorbidities and can be further examined to prevent readmissions and prevent complications or poor care that may affect Centers of Medicare and Medicaid Services reimbursement rates. Understanding the trends of PCNL on a nationwide as well as regional level is important in evaluating how to deliver more effective high-quality health care. In addition, we can examine strengths and deficits in the delivery of that care.⁹

The New York State (NYS) Statewide Planning and Research Cooperative System (SPARCS) database houses information regarding all hospital discharges across NYS. This database contains information regarding patient demographics/characteristics, insurance status, diagnoses, treatments, procedures, and length of stay. This information is required to be submitted on a monthly basis by an assigned administrator/coordinator.

In our study, we examined the trends of PCNL utilization across NYS, allowing us to examine rural, suburban, and urban regions with high-, medium-, and low-volume institutions to gain a thorough understanding for significant stone burden care in our state.

Materials and Methods

The SPARCS database was queried from January 2006 to December 2014 using ICD-9 procedure codes 55.03 (percutaneous nephrostomy without fragmentation) and 55.04 (percutaneous nephrostomy with fragmentation) examining all hospital discharges across NYS. Reports with duplication or incomplete data were excluded. Demographic variables include patient age, gender, race, insurance status, and length of stay. Each hospital was characterized as a low-, medium-, or high-volume center. All hospitals were ranked in order of increasing procedure on a yearly basis and then split into three equivalent groups or tertiles based on whole number cutoff points for the annual volume to characterize them as high- (top tertile), medium- (middle tertile), or low (bottom tertile)-volume institutions. The hospitals were categorized by location within a rural vs suburban vs urban setting. Finally, the teaching status of each designated hospital was examined as follows: nonteaching, minor teaching (academic center that is a teaching hospital without a medical school), and major teaching (academic center that is a teaching hospital with a medical school).

Statistical analysis

Patient and hospital demographics were compared using chi-square analysis and Student's t-test for categorical and continuous variables, respectively. Univariable logistic regression was performed to compare demographics and socioeconomic variables between high-, medium-, and low-volume institutions and also urban vs rural settings. Odds ratios with 95% confidence intervals were reported with p-values <0.05 considered statistically significant. Statistics were calculated using Stata, version 12 (College Station, TX).

Results

Between 2006 and 2014, the number of PCNLs performed across NYS increased by 52% overall, with the highest total in 2012 as given in Table 1. Among low- and medium-volume institutions, there was an average increase of PCNL procedures from 2.30 to 3.16 per year and from 9.36 to 14.29 per year resulting in an increase of 37.3% and 52.7%, respectively. As for high-volume institutions, the average volume increased from 20.29 to 55.0 for an increase of 171%.

Table 1.

Hospital Level Variates

	2006	2007	2008	2009	2010	2011	2012	2013	2014
Definition (minimum–maximum number of procedures per year)
High	15–26	19–35	19–31	24–38	29–65	36–79	38–112	49–105	34–105
Medium	7–14	7–12	7–14	8–23	8–24	11–31	14–32	11–34	11–27
Low	1–6	1–6	1–6	1–7	1–7	1–10	1–13	1–10	1–10
Number of hospitals per category
High	7	5	5	5	4	4	3	3	4
Medium	14	13	14	11	11	12	10	11	14
Low	54	57	54	58	54	56	55	63	59
Number of procedures per category
High	142	129	125	151	154	192	206	205	220
Medium	131	122	132	150	155	211	222	212	200
Low	124	120	140	143	144	181	211	209	183
Average number of procedures per hospital
High	20.29	27.50	25.00	30.20	38.50	48.00	68.67	68.33	55.00
Medium	9.36	9.38	9.43	13.64	14.09	17.58	22.20	19.27	14.29
Low	2.30	2.11	2.59	2.51	2.72	3.29	3.91	3.37	3.16

The patient demographics and characteristics as well as hospital geography and teaching status are shown across institutions with low, medium, or high volume in Table 2. The average patient age (standard deviation) for low-, medium-, and high-volume institutions was 54.0 (16.1), 54.9 (16.0), and 54.4 (15.7) years, respectively, p = 0.31. A higher proportion of female patients (53.4%–56.4%) underwent PCNL regardless of hospital volume status, but this was not significant (p = 0.66). Low-volume center ethnicity rates were 63.9% white, 10.5% black, 4.2% Asian, and 21.4% other. Medium volume distribution was 70.2% white, 8.3% black, 4.1% Asian, and 17.4% other. High-volume centers provided care for 65.1% whites, 12.2% blacks, 5.4% Asian, and 17.3% other. Compared with low-volume (21.4%) and high-volume centers (17.3%), minorities (non-white) were a smaller proportion of the patients treated at the high-volume centers (p < 0.001). As for insurance type, private insurance use was most pronounced at high-volume (42.1%) than at medium- (37.6%) and low-volume centers (34.0%) (p < 0.001). Medicaid patients were more common at low-volume centers (25.5%) than at medium- (21.5%) or high-volume centers (23.5%) (p < 0.001). Length of stay was lowest at high-volume centers with 3.3 days on average compared with 3.6 and 4.1 days at medium- and low-volume centers, respectively (p < 0.001).

Table 2.

Demographics of Hospitals and Patients

Hospital volume	Low	Medium	High	p
Total procedures (N)	1507	1535	1534
Hospital classification				<0.001
Urban	1156 (76.7)	1207 (78.6)	1442 (94.0)
Suburban	343 (22.8)	328 (21.4)	92 (6.0)
Rural	8 (0.5)	0 (0)	0 (0)
Teaching status				<0.001
Major teaching	520 (34.5)	940 (61.2)	1202 (78.4)
Minor teaching	646 (42.9)	457 (29.8)	332 (21.6)
Nonteaching	339 (22.6)	138 (9.0)	0 (0)
Race (%)
White	963 (63.9)	1078 (70.2)	998 (65.1)	<0.001
Black	158 (10.5)	127 (8.3)	187 (12.2)
Asian	64 (4.2)	63 (4.1)	83 (5.4)
Other	322 (21.4)	267 (17.4)	266 (17.3)
Gender (%)				0.657
Male	684 (45.4)	721 (46)	715 (46.6)
Female	823 (54.6)	814 (54)	819 (53.4)
Age (SD)	54.0 (16.1)	54.9 (16.0)	54.4 (15.7)	0.308
Insurance type (%)				<0.001
Private	513 (34.0)	577 (37.6)	646 (42.1)
Medicare	496 (33.0)	529 (34.4)	489 (31.9)
Medicaid	384 (25.5)	330 (21.5)	360 (23.5)
Military	11 (0.7)	6 (0.4)	1 (0.1)
Self-pay	73 (4.8)	48 (3.1)	33 (2.2)
Other	31 (2.0)	45 (3.0)	5 (0.3)
Initial LOS (SD)	4.1 (5.5)	3.6 (5.0)	3.3 (4.1)	<0.001

LOS = length of stay; SD = standard deviation.

When hospital geography was examined, urban hospital settings comprised the majority of PCNL procedures in all locations: low- (76.7%), medium- (78.6%), and high-volume centers (94.0%). No rural hospital was able to qualify as a high- or medium-volume institution. Almost all high-volume institutions were urban (94.0%) compared with suburban (6.0%) (p < 0.001). Similar trends were seen with teaching status with major teaching institutions comprising 34.5%, 61.2%, and 78.4% of low-, medium-, and high-volume centers, respectively. No nonteaching hospital was designated as a high-volume center, whereas major teaching institutions were more pronounced among medium- and high-volume centers (p < 0.001). Interestingly, the minor teaching institutions comprised the greatest proportion of low-volume hospitals (p < 0.001).

Discussion

PCNL is considered a complex endourologic procedure, which only a minority of urologists perform routinely, and of these, only 11% obtain percutaneous access by themselves.¹⁰ Furthermore, PCNL is considered a challenging procedure in inexperienced hands and/or on complex patients. As such, it would not be unusual to find a high level of regionalization in this field as well. Indeed, our study found an average PCNL volume increase of 171% for NYS high-volume institutions, compared with 37.3% and 52.7% for low- and medium-volume institutions, respectively, over 8 years. Stern et al. examined the national inpatient sample and found close to a twofold increase in PCNL utilization from 1998 to 2011.¹¹ The lifetime prevalence of nephrolithiasis is increasing from 7.14% to 11.62%.¹² It is possible that this increase in stone disease is attributed to increased imaging or increasing rates of diabetes mellitus and obesity. By 2030, it is estimated that stone prevalence will increase by 0.36% caused by obesity and by 0.72% for diabetes mellitus.¹³

Furthermore, our study also highlighted the impact of insurance status on the level of health care. Historically, the most notable disabling condition precluding access to quality health care in the United States has been inadequate health insurance.¹⁴ In our study, private insurance use was most significantly pronounced at high-volume (42.1%) compared with medium- (37.6%) and low-volume centers (34.0%) (p < 0.001). Medicaid patients were more commonly treated at low-volume (25.5%) than at medium- (21.5%) or high-volume centers (23.5%) (p < 0.001). In addition, almost all the high-volume-PCNL institutions were urban (94.0%) compared with suburban (6.0%) (p < 0.001). These data confirm the substantial role that socioeconomic and geographic factors play in accessing PCNL.

The consequences of this unequal access to care are obvious, being reflected in our study by the differences found in length of stay after PCNL between high-volume (3.3 days) and medium- and low-volume centers (3.6 and 4.1 days, p < 0.001, respectively), further emphasizing the importance of patient's demographic and socioeconomic factors that can explain differences in access of care to this challenging procedure.

Access to care is a complex topic with multiple individual and contextual variables that unequally influence general population trends in health and, ultimately, affect patient outcomes. These factors, which can be resource dependent (socioeconomic class and access) or nonresource dependent (e.g., psychologic, genetic, and cultural), interact dynamically in a complicated social network, rendering topics such as inequity and disparity difficult to analyze and address.¹⁵ The resource-dependent factors have become vital to guarantee access to high-quality medical care. This shift is significant and profoundly affects patient outcome.

Socioeconomic factors have been demonstrated as a disparity to care across multiple disciplines in medicine. Similar to our study, there have been differences in access to care and poorer outcomes in patients with regard to race, insurance status, treatment at high- vs low-volume settings, and lower median incomes. Feng et al. queried the SPARCS database identifying greater proportion of African American patients, Medicaid, and self-pay patients being treated at low-volume institutions where these groups were found to have higher rates of postoperative bleeding and in-hospital mortality rates for cervical spinal fusion.¹⁶ Similar findings were noted for hip-replacement procedures wherein patients of black race, lower median income, public insurance (Medicaid or Medicare), and treatment at a low-volume hospital had between 23% and 58% increased chance of 30-day readmission.¹⁷

When comparing these findings with patients who have undergone procedures for urologic conditions, similar conclusions can be drawn. Halpern et al. queried the SPARCS database and identified that non-white patients undergoing prostate biopsy had higher rates of infection and urinary retention within 30 days of biopsy.¹⁸ Brubaker et al. recently examined payer type and race with regard to the timing for stone surgery after emergency room (ER) discharge.¹⁹ Patients with Medicaid, Medicare, and self-pay had an increase of 46%, 42%, and 60% for the time to surgery, respectively, compared with their private insurance counterparts. Furthermore, when examining patients of black and Hispanic ethnicity, there was an increase of 36% and 20% compared with those with Caucasian descent. Kirshenbaum and coworkers examined the factors associated with delay of upfront ureteroscopy during ER admission.²⁰ On multivariate regression models, they found that minorities, younger patients, publically/uninsured patients, and more comorbid patients had lower odds of undergoing upfront ureteroscopy. Similarly, higher complication rates with Medicare (17.1%) and Medicaid (16.9%) patients have been seen than private patients (12.3%, p < 0.001) for PCNL.⁸

In 2003, Birkmeyer et al. found that surgeon volume was inversely related to operative mortality rates among 474,108 patients who underwent one of eight procedures or cancer resections, including radical cystectomy, lung resection, and pancreatic resection.²¹ Other studies have found similar results regarding morbidity and mortality rates related to hospital and surgeon volume, in diverse surgical specialties.^22–24 In urology, this phenomenon has been studied as well. Nuttall and coworkers reviewed the literature relating hospital or surgeon volume to health outcomes for three urologic cancer procedures.²⁵ In their study, it was found that outcomes after radical prostatectomy and cystectomy are on average likely to be better if these procedures are performed by more experienced surgeons at high-volume centers. Another review showed evidence suggesting that increasing volume improves outcomes for radical prostatectomy.²⁶ As one of the most complex stone surgery techniques, good proficiency in PCNL is established after 60 cases and excellence after >100 cases.²⁷ Based on these findings, it may be appropriate to refer complex surgical patients to high-volume centers whenever possible.

Several limitations of our study must be acknowledged. The SPARCS database is a statewide database maintained by the New York Bureau of Health Informatics. It is unclear whether there is standardization with regard to data or whether reporting practices varied between all the institutions involved. It is possible these differences may have affected the data we collected. Although we had initially sought to use current procedural terminology (CPT) codes to accurately capture the patients who underwent PCNL, this was not possible given the database parameters. We instead used ICD-9 codes to capture patients who underwent PCNL and these codes were cross-referenced with our billing department to ensure we accurately captured the correct patient population. Furthermore, this is a retrospective study of NYS and our findings cannot be extrapolated on a national level. The hospitals involved were diverse in geographic location, size, patient population, and density, as well as academic status. Information specific to each individual surgeon such as case volume, overall experience, fellowship training, and particular case mix index/case complexity was not available. It is unclear whether a single surgeon could have also operated at another institution and this could have affected the data. We agree that these variables play a role in the quality of care and complication rates; however, we were unable to capture these with the database. Further studies with larger cohorts and on a national level can help support our findings and reveal any other disparities with access to care.

Conclusion

Our study helps demonstrate the trends in utilization of PCNL across NYS and factors related to access to care. During our study period, the rate of PCNL increased significantly and it was most pronounced in high-volume institutions. Furthermore, various differences in race, insurance status, and length of stay were noted between high- and low-volume institutions. Our study shows a regionalization of care, but also racial and socioeconomic factors that pose as barriers for access to care. Understanding of these disparities can help decrease patient-related morbidity and improve surgical outcomes.

Abbreviations Used

ER

emergency room

NYS

New York State

PCNL

percutaneous nephrolithotomy

SPARCS

Statewide Planning and Research Cooperative System

Author Disclosure Statement

No competing financial interests exist.