Nationwide Readmission Rates Following Percutaneous Nephrolithotomy: Does Age Matter?

Abstract

Introduction/Objectives:

Despite minimal evidence that evaluates the effect of age on percutaneous nephrolithotomy (PCNL) morbidity, pediatric and elderly patients are considered high-risk groups. Our objective was to assess the effect of the extremes of ages on PCNL readmission and postoperative complication rates.

Methods:

We identified all PCNL encounters in the 2013 and 2014 Nationwide Readmission Database. Encounters were divided into five age groups: pediatric (<18 years old), young adult (18–25 years old), adult (26–64 years old), geriatric (65–74 years old), and elderly (≥75 years old). Weighted descriptive statistics were used to describe population demographics. We fit an adjusted weighted logistic regression model for 30-day readmission and complication rates.

Results:

We identified 23,357 encounters. Testing average effect of pediatric and elderly encounters to all other age groups did not reveal a difference in odds for 30-day readmissions, but did result in increased odds for 30-day GU readmissions (odds ratio: 17.7 [95% confidence interval (CI): 2.65–118.9]; p = 0.003). Compared to all other age groups, elderly encounters had 7.5 (95% CI: 2.5–22.7; p = 0.0004) times the odds of a 30-day readmission and 68.3 (95% CI: 29.1–160.4; p < 0.0001) times the odds of a postoperative complication.

Conclusions:

When comparing the average effect of the extremes of ages to all other age groups, we did not find evidence to suggest a difference in odds for 30-day GU readmissions, but did find increased odds for complications. Further examination revealed that PCNL encounters of elderly patients had significantly increased odds for both readmission and complications, whereas PCNL encounters of pediatric patients did not.

Introduction

According to the American Urological Association/Endourology Society Guidelines, the recommended treatment for stone burden greater than 2 cm or staghorn calculi, regardless of age, is percutaneous nephrolithotomy (PCNL).¹ PCNL has been proved to be a superior surgical technique for larger stone burden²; however, it comes with an increased risk of complications. Compared to ureteroscopy, complications have been found to be 40% more likely to occur with PCNL.³ PCNL-related complications result in unplanned care in 15% of cases costing on average $19,370.⁴

PCNL is still recommended for the treatment of both children and adults. In children, PCNL has significantly higher stone-free rates compared to ureteroscopy in the treatment of stone burden >2 cm.⁵ Despite this evidence in support of PCNL in pediatric patients, concerns persist regarding the surgical complexity, small kidney size, and risk of bleeding in children.⁶

At the other end of the age spectrum, PCNL has been increasingly utilized in older adults.⁷ As the use of PCNL continues to rise in older adults, questions arise regarding complication risks for a population with more medical comorbidities. In the adult population, age has been associated with increased risks of 90-day complication rates and prolonged length of stay (LoS) following PCNL.⁷

Despite a gap in the literature that directly evaluates whether the extremes of ages are associated with PCNL-related morbidity, pediatric and elderly patients are considered high-risk groups for PCNL. Therefore, our objective was to assess the effect of the extremes of ages on PCNL readmission rates and postoperative complication rates. We hypothesized that the average effect of pediatric and elderly encounters would be higher than the combined average effect of the other age groups for both readmission and postoperative complication rates.

Methods

Data source

We analyzed the 2013 and 2014 Nationwide Readmission Database (NRD), an all-payer database managed by the Healthcare Cost and Utilization Project (HCUP) and sponsored by the Agency for Healthcare Research and Quality. NRD is derived from 21 HCUP State Inpatient Databases accounting for 49% of all U.S. hospitalizations.⁸

Outcome and variable definitions

The primary outcome was 30-day readmission rates defined as encounters within 30 days of a patient's discharge from the inpatient care setting with a genitourinary (GU)-related diagnoses (Appendix Table A1). Our secondary outcome was postoperative complications defined as at least one complication during the inpatient encounter. Complication rates were determined by National Surgical Quality Improvement Program's (NSQIP) definition using International Classification of Diseases, ninth revision, Clinical Modification (ICD-9) codes (Appendix Table A2). Patients were divided into five age groups: pediatric (<18 years old), young adult (18–25 years old), adult (26–64 years old), geriatric (65–74 years old), and elderly (≥75 years old). This categorical variable was used as our primary predictor for modeling. In addition, the NRD collects demographic features such as gender, expected primary payer (public, private, or other), median household income based on zip codes (Q1–Q4), hospital type (metropolitan teaching, metropolitan nonteaching, and nonmetropolitan), LoS, and cost of encounter. Cost of encounter was estimated by using the cost-to-charge ratio files provided by HCUP. This provides a hospital level adjustment factor that is multiplied to the total charge of an encounter. The Van Walraven Index (VWI) was used to account for comorbidity.⁹

Cohort selection

We included all patients with upper tract stones (ICD-9 code 592.0) who underwent PCNL (ICD-9 procedure codes 55.01, 55.02, 55.03, 55.04). We excluded patients with potentially complicating diagnoses: hypertonic bladder (696.51), low bladder compliance (596.52), bladder paralysis (596.53), detrusor-sphincter dyssynergia (596.55), other bladder function disorder (596.59), spina bifida (74.10, 74.19, 756.17), congenital urethral strictures (75.36), epispadias (752.62), renal & ureteral dis necrotizing enteroColitis (NEC) (593.89), neurogenic bladder (596.54, 344.61), ureterocele (753.23), megaureter (593.89, 753.22), posterior urethral valves (753.6), bladder exstrophy (753.5), kidney transplant (55.61, 55.69), or prune belly syndrome (756.71). For simplicity, we refer to weighted encounters as “encounters,” but all analyses use weighted statistics as per HCUP recommendations.

Statistical methods

As recommended by HCUP, we used weighted descriptive statistics to describe the demographic characteristics.⁸ This allowed us to account for the complex survey structure of the NRD and to make national level inferences. As per HCUP data use agreement, all variables with cell counts less than 15 were excluded.

For our primary and secondary outcomes, we fit a weighted logistic regression model. The weighting scheme allowed us to account for the correlation structure of the dataset by allowing the variance within each hospital to change by year. Our primary predictor was age group. The main model under study adjusts for gender, insurance type (private vs other/missing), median household income, and hospital type (metropolitan teaching vs other/missing) were included as covariates. Inclusion of median household income in the 30-day readmissions model resulted in the model failing to converge; because this was not our primary predictor of interest, we chose to exclude this covariate in the interest of model parsimony and simplicity. VWI was not used as a covariate in the model as it is a possible mediator between age and 30-day readmission rates (encounters in the older age groups may have a higher VWI value and encounters with higher VWI may have higher 30-day readmission rates). Thus, we believed that including it in the model may attenuate the effect of age on the outcome. To examine this possible attenuation, we analyzed this model with and without VWI as a covariate.

To test our primary and secondary hypotheses, we used contrasts to test if the average effect of pediatric and elderly encounters were higher than the average effect of all other age groups on the 30-day readmissions and postoperative complications. A postdesign exploratory evaluation investigated if the elderly or pediatric group was driving the average effect. This was done by testing both the average effect of the elderly group compared to all other age groups and the pediatric group compared to all other age groups for both outcomes.

Our primary analyses used an alpha of 0.05 and 95% confidence intervals (CI) as criteria for statistical significance. Bonferroni correction was used to account for multiple testing during our postdesign exploratory analysis. This resulted in an adjusted alpha level of 0.025. All analyses were performed using SAS, version 9.4 (SAS Institute, Cary, NC).

Results

There was a total of 23,357 PCNL encounters between 2013 and 2014 (Table 1). The mean age was 56.1 years (SE: 0.2) ranging between 0 and 90 years old. Adult encounters represented the largest cohort (62.3%), followed by geriatric (21%), elderly (12.1%), young adults (3.4%), and pediatric encounters (1.2%). Of our cohort, 44.3% was male and 51% had public insurance. Thirty-seven percent had a VWI that was one or greater, with a mean score of 1.9 (SE: 0.1); 0.7% of encounters resulted in in-patient death.

Table 1.

Nationwide Readmission Database 2013 and 2014 Percutaneous Nephrolithotomy Encounter Demographics

	Total (N = 23,357)
Age (years)
Mean (SD)	56.1 (0.2)
Median (Q1, Q3)	57.1 (45.2, 66.8)
Range	(0.0–90.0)
Age groups
Pediatric (<18 years)	285 (1.2%)
Young adult (18–25 years)	804 (3.4%)
Adult (26–64 years)	14,547 (62.3%)
Geriatric (65–74 years)	4910 (21%)
Elderly (75+ years)	2811 (12.1%)
Gender
Male	10,355 (44.3%)
Primary payer
Public	11,994 (51.4%)
Private	10,154 (43.5%)
Other	1184 (5%)
Missing	26 (0.1%)
Median household income
0–25th percentile	5789 (24.8%)
26th–50th percentile	5883 (25.2%)
51st–75th percentile	5946 (25.5%)
76th–100th percentile	5358 (23%)
Missing	359 (1.5%)
Van Walraven Index
≤−1	4379 (18.7%)
0	10,310 (44.1%)
1	454 (1.9%)
2	469 (2%)
3	1560 (6.7%)
4	452 (1.9%)
5	1595 (6.8%)
≥6	4138 (17.7%)
Mean (SD)	1.9 (0.1)
Median (Q1, Q3)	−0.3 (−0.9, 3.8)
Range	(−13.0–34.0)
Hospital type
Metropolitan nonteaching	6029 (25.8%)
Metropolitan teaching	16,791 (71.9%)
Nonmetropolitan hospital	537 (2.3%)

PCNL-related outcomes by age group

The elderly and geriatric encounters had the highest rates of mortality compared to the other groups (3.2% and 0.7%, respectively, Table 2). Pediatric encounters had the highest mean LoS (11.3 days, SE: 4.8). The median LoS for elderly encounters was higher than the pediatric group (4.6 days; interquartile range [IQR]: 1.8–8.9 vs 3.6 days; IQR: 1.8–6). The other age groups had a mean LoS between 4 and 5 days and a median LoS between 2 and 2.5 days. Pediatric encounters had both the highest mean cost $35,361 (SE: $11,362) and median cost $17,142 (IQR: $11,551–$27,356). The elderly group had the second highest with a mean of $20,188 (SE: $502.94) and median of $14,671 (IQR: $10,304–$24,107).

Table 2.

Percutaneous Nephrolithotomy Encounter Outcomes by Age Group

	Pediatric (<18 years)	Young adult (18–25 years)	Adult (26–64 years)	Geriatric (65–74 years)	Elderly (75+ years)
30-day GU-readmissions	20 (7.1%)	109 (13.5%)	1744 (12%)	618 (12.6%)	525 (18.7%)
Postoperative complications	87 (30.6%)	240 (29.9%)	4831 (33.2%)	2032 (41.4%)	1777 (63.2%)
Inpatient death	0	0	35 (0.2%)	35 (0.7%)	89 (3.2%)
Length of stay
Mean (SD)	11.3 (4.8)	4.7 (0.6)	4.1 (0.1)	5.3 (0.2)	7.2 (0.2)
Median (Q1, Q3)	3.6 (1.8, 6)	2.5 (1.3, 4.5)	2.1 (1.1, 3.9)	2.4 (1.2, 5.1)	4.6 (1.8, 8.9)
Range	0–153	0–91	0–361	0–291	0–90
Cost of admission
Mean (SD)	$35,361 (11,362)	$15,937 (2103)	$14,841 (252)	$17,555 (429)	$20,188 (503)
Median (Q1, Q3)	$17,142 (11,551, 27,356)	$11,434 (7556.2, 16,781)	$11,837 (8434, 16,827)	$13,260 (9456, 19,309)	$14,671 (10,304, 24,107)
Range	$4989.5–$357,455	$1959.62–$296,591	$272.19–$358,899	$592.38–$282,731	$1174.48–$195,506

GU = genitourinary.

The extremes of ages represented the highest and lowest rates of 30-day readmission. Elderly encounters had the highest readmissions rate (18.7%) vs pediatric encounters, which had the lowest rate (7.1%). Nearly two-thirds (63.2%) of elderly encounters experienced a postoperative complication, the highest rate of all the groups. Pediatrics, young adult, and adult encounters had similar complication rates (∼30%).

Thirty-day GU readmission

In our multivariate analysis, the compared average effect of combining the extremes of ages, the elderly and pediatric encounters, against all other combined age groups did not have significantly higher odds of 30-day readmission (odds ratio [OR]: 0.42 [95% CI: 0.03–5.8]; p = 0.51; Table 3). When including VWI in the model, the average effect of the elderly and pediatric group was attenuated (OR: 0.28 [95% CI: 0.02–3.91]; p = 0.34).

Table 3.

Average Effects of Age on 30-Day Genitourinary-Readmission Rates for Percutaneous Nephrolithotomy Encounters

	Adjusted OR ^a (95% CI)	p-Value	Adjusted OR ^b (95% CI)	p-Value
Pediatric and elderly vs all other ages	0.42 (0.03–5.8)	0.51	0.28 (0.02–3.91)	0.34
Elderly vs all other ages	7.5 (2.5–22.7)	0.0004	4.2 (1.4–12.8)	0.0115
Pediatric vs all other ages	0.06 (0.002–1.5)	0.08	0.07 (0.003–1.8)	0.12

Model adjusted for gender, insurance type (private vs other/missing), and hospital type (metropolitan teaching vs other/missing).

Model adjusted for gender, insurance type (private vs other/missing), hospital type (metropolitan teaching vs other/missing), and VWI.

CI = confidence interval; OR = odds ratio; VWI = Van Walraven Index.

Postoperative complications

Modeling for postoperative complications, the average effect of the combined elderly and pediatric encounters had significantly higher odds, 17.7 times, of having a postoperative complication compared to all other combined age groups (95% CI: 2.65–118.9; p = 0.003; Table 4). When including VWI in the model, the average effect of the elderly and pediatric group was attenuated (OR: 6.86 [95% CI: 1.10–42.72]; p = 0.04).

Table 4.

Average Effects of Age on Postoperative Complication Rates for Percutaneous Nephrolithotomy Encounters

	Adjusted OR ^a (95% CI)	p-Value	Adjusted OR ^b (95% CI)	p-Value
Pediatric and elderly vs all other ages	17.7 (2.65–118.9)	0.003	6.86 (1.10–42.72)	0.04
Elderly vs all other ages	68.3 (29.1–160.4)	<0.0001	17.29 (7.32–40.8)	<0.01
Pediatric vs all other ages	0.26 (0.02–2.8)	0.27	0.4 (0.04–3.9)	0.43

Model adjusted for gender, insurance type (private vs other/missing), median quarterly income, and hospital type (metropolitan teaching vs other/missing).

Model adjusted for gender, insurance type (private vs other/missing), median quarterly income, hospital type (metropolitan teaching vs other/missing), and VWI.

Postdesign exploratory analysis

Interestingly, while investigating the average effect of only elderly encounters, compared with all other age groups, the elderly group did have significantly higher odds, 7.5 times (95% CI: 2.5–22.7; p = 0.0004), of a 30-day readmission. In addition, elderly encounters had 68.3 (95% CI: 29.1–160.4; p < 0.0001) times the odds of a postoperative complications compared with the other age groups. Including VWI in both models attenuated the effect of elderly encounters as expected.

While investigating the average effect of pediatric encounters compared with all other age groups, we did not find evidence of a difference in odds for either 30-day readmissions (OR: 0.06 [95% CI: 0.002–1.5]; p = 0.08) or postoperative complications (OR: 0.26 [95% CI: 0.02–2.8]; p = 0.27). Including VWI in the models did not produce statistically significant results.

Discussion

Given the lack of data evaluating the relationship of age to PCNL outcomes, our study sought to determine differences in readmission and complication rates between age groups. While we did not find evidence of a difference in the combined effects of the extremes of ages compared to all other ages for 30-day readmissions, we did find evidence of a difference in postoperative complications. In addition, we found that elderly encounters had significantly increased odds of 30-day GU-readmission and postoperative complications compared to all other age groups (including pediatric), both greater than the odds of the combined extremes of ages. In comparing the effect of pediatric patients against all other ages, we did not find evidence of increased odds for both readmissions and complications. This indicates that elderly patients may actually have higher odds for readmission and complications while pediatric patients do not.

Although the American Urological Association guidelines recommend PCNL for large stone burden in pediatric patients, there are still concerns of increased complication rates in this unique patient population. However, our study may serve to mitigate these concerns as pediatric encounters had the lowest readmission rates compared with every other age group. Furthermore, we did not find evidence of a difference in odds for 30-day readmissions in pediatric encounters compared with all other age groups. With regard to postoperative complications, while pediatric encounters did not have the lowest complication rates at 30.6%, their complication rates were comparable to those of young adults and adults. In addition, pediatric encounters were not at increased odds of postoperative complications compared to all other age groups. Our findings are supported by the literature with previously reported post-PCNL complication rates of 34% to 41% for pediatric patients.^5,10 A single-institution, retrospective study using Clavien–Dindo complication grading, also reported similar complication rates between children (34.4%) and adults (28.0%).¹⁰ The similar rate of complications for pediatric encounters to the adult encounters in our study could be due to the division of age cohorts. The anatomical complexity of infants or toddlers may not be detected since it was necessary to group all patients <18 years old due to data sparseness and HCUP's data use agreement prohibiting data reports with occurrences lower than 15.

We found that elderly encounters had the highest rates of readmissions (18.7%) and complications (63.2%) following PCNL. In addition, our postdesign exploratory analysis indicated increased odds for both 30-day readmissions and postoperative complications for elderly encounters. To our knowledge, this is first reported readmission rate following PCNL for encounters 75 years and older. Similar to our study, Leow et al. sampled the Premier database for post-PCNL complications in adults and found that increasing age is a risk factor for complications.⁷ It is worth noting that NRD, in contrast to Premier, is an all-payer database of over 36 million weighted discharges per year, or almost half of all hospitalizations in the United States in a given year. Thus, these data represent (to the best of our knowledge) the first analysis of PCNL in a nationally representative database capturing use of these surgeries across all age groups. Our findings do not contradict the recommendation for PCNL in older adults. Instead, these data should aid in the design of future prospective studies that allow the investigation of causal pathways related to these increased risks for this vulnerable patient population.

Interestingly, pediatric PCNL encounters had longer median LoS and the highest median costs compared to the other adult age groups, excluding elderly encounters with the highest LoS. The longer LoS could be protective against readmission. Longer hospital stays would allow for optimization of pain control, and earlier detection and management of postoperative complications. Of note, LoS may have been affected by planned “second-look” PCNL procedures during the primary admission vs a subsequent admission. Controlling for this limitation is not possible with NRD.

Our study should be evaluated in the context of its design limitations. As a retrospective study, our inference is limited to associations. Since NRD is an administrative coding database, our results are subject to misclassification bias. The NRD includes data from 21 states, and thus may not be generalizable to all regions of the country or other parts of the world. The NRD does not allow for longitudinal tracking of patients between calendar years. To account for this limitation, we excluded patients from the month of December to allow for 30-day follow-up. Although we used the best available comorbidity adjustment index,¹¹ it is possible that our models did not complete account for confounding effects from comorbid diagnoses. Although unlikely, preselected NSQIP complication ICD-9 codes could potentially be incorrectly coded and interpreted to represent a patient's medical history rather than a postoperative complication, therefore, complication rates could potentially be overestimated. To reduce the chances of this error occurring, we strictly followed HCUP guidelines and all published HCUP Methods reports. As such, we are confident in the veracity of our analysis and of our ability to distinguish general trends in complication rates.

Conclusions

When comparing the average effect of the extremes of ages, pediatric and elderly, for PCNL encounters to all other age groups, we did not find evidence to suggest a difference in odds for 30-day GU readmissions, but did find increased odds for postoperative complications. Further examination revealed that PCNL encounters of elderly patients had significantly increased odds for both 30-day GU readmission and postoperative complications, whereas PCNL encounters of pediatric patients did not. Our findings support existing evidence that while PCNL can be associated with notable morbidity, the risks are comparable for pediatric and young- and middle-aged adult. However, elderly patients should be appropriately counseled on the increased risk of readmissions and postoperative complications. Additional studies are needed to identify causal relationships between PCNL-complications and specific risk factors in this vulnerable patient population.

Footnotes

Author Disclosure Statement

Dr. Routh is supported by grant number K08-DK100534 from the National Institute of Diabetes and Digestive and Kidney Diseases. The funding sources had no role in the collection, analysis, or interpretation of data; in the writing of the article; or in the decision to submit the article for publication.

Abbreviations Used

Appendix Table A2.

National Surgical Quality Improvement Program Postoperative Complications

Diagnosis	ICD-9 code
Surgical site infection	99832, 99831
Peritoneal abscess	56722
Acute renal failure	584, 586
Urinary tract infection	5990
Urinary complication	9975
Respiratory complication	9973, 51881, 51882, 514, 5184
Pneumonia	481, 482, 483, 484, 485, 486, 487, 507
Postoperative respiratory complications	5185
Acute respiratory insufficiency	51882
Acute respiratory failure	51881
Systemic sepsis	7907, 038
Pulmonary emboli infract	4151, 41511, 41519
Postoperative CVA	99702
Cardiac complications	9971
Myocardial infarction	410
Cardiac arrest	4275
Postoperative bleeding	2851, 99811
Deep vein thrombosis	4534, 45340, 4539

CVA = cerebrovascular accident.

References

Assimos

, Krambeck

, Miller

, et al: Surgical Management of Stones: American Urological Association/Endourological Society Guideline, PART I. J Urol, 2016; 196:1153–1160.

Karakoyunlu

, Goktug

, Şener

, et al: A comparison of standard PCNL and staged retrograde FURS in pelvis stones over 2 cm in diameter: A prospective randomized study. Urolithiasis, 2015; 43:283–287.

, Autorino

, Kim

, et al: Percutaneous nephrolithotomy versus retrograde intrarenal surgery: A systematic review and meta-analysis. Eur Urol, 2015; 67:125–137.

Scales

, Saigal

, Hanley

, et al: The impact of unplanned postprocedure visits in the management of patients with urinary stones. Surgery, 2014; 155:769–775.

Saad

KSM

, Youssif

, Hamdy

SAIN

, et al: Percutaneous nephrolithotomy vs retrograde intrarenal surgery for large renal stones in pediatric patients: A randomized controlled trial. J Urol, 2015; 194:1716–1720.

Guven

, Frattini

, Onal

, et al: Percutaneous nephrolithotomy in children in different age groups: Data from the Clinical Research Office of the Endourological Society (CROES) Percutaneous Nephrolithotomy Global Study. BJU Int, 2013; 111:148–156.

Leow

, Meyer

, Wang

, et al: Contemporary Trends in utilization and perioperative outcomes of percutaneous nephrolithotomy in the United States from 2003 to 2014. J Endourol, 2017; 31:742–750.

Anon: The HCUP Nationwide Readmissions Database (NRD). 2018. Available at: https://www.hcup-us.ahrq.gov/db/nation/nrd/Introduction_NRD_2010–2016.jsp, accessed March 30, 2019 .

van Walraven

, Austin

, Jennings

, et al: A modification of the Elixhauser comorbidity measures into a point system for hospital death using administrative data. Med Care, 2009; 47:626–633.

10.

Kumar

, Keshavamurthy

, Karthikeyan

, et al: Complications after prone pCNL in pediatric, adult and geriatric patients—A single center experience over 7 years. Int Braz J Urol, 2017; 43:704–712.

11.

Jiang

, Wolf

, Alkazemi

, et al: The evaluation of three comorbidity indices in predicting postoperative complications and readmissions in pediatric urology. J Pediatr Urol, 2018; 14:244.e1–244.e7.