Open versus Minimally-Invasive Surgical Techniques in Pediatric Renal Tumors: A Population-Level Analysis of In-Hospital Outcomes

Abstract

INTRODUCTION:

Minimally-invasive surgery (MIS) has been adopted slowly in pediatric oncology. We attempted to describe contemporary national trends in MIS use; we hypothesized that adolescents (who are more likely to have relatively small renal cell carcinomas) would have a higher proportion of MIS than younger children (who are more likely to have relatively large Wilms tumors) and that this relationship would vary by region.

OBJECTIVE:

To explore whether pediatric urologic oncology outcomes vary by patient age or by surgical technique.

METHODS:

We queried the 1998–2014 National Inpatient Sample (NIS) and included encounters in children aged <= 18y, ICD-9 diagnostic Pcodefor renal tumor, and procedure code for open or MIS partial or radical nephrectomy. All analyses used weighted descriptive statistics and outcomes are compared based on age group (</>10 y) or surgery type; Wald-Chi square test was used for differences in proportions and unadjusted weighted ANOVA was used to test for differences in means.

RESULTS:

9,259 weighted encounters were included; 91% were <10 years old and 50.7% were female. MIS surgery accounted for 1.8% of encounters; there was a difference in proportions by age group (1% <9y vs. 9.9% >9y, p<0.01). The proportion of surgery type was similar across regions within age groups, however. Complications occurred in 13.3% of encounters; mean inpatient length of stay was 8.9 days (SD: 0.3); mean cost was $ 34,457.68 (SD: $1,197.00). There was no evidence of a difference between surgery type and proportion of post-operative complications, mean inpatient length of stay or mean inpatient cost.

DISCUSSION:

The admission-based, retrospective design of NIS left us unable to assess long-term outcomes, repeated admissions, or to track a particular patient across time; this is particularly relevant for oncologic variables on interest such as tumor stage or event-free survival. We were similarly limited in evaluating the effect of pre-surgical referral patterns on patient distributions.

CONCLUSION:

In this preliminary descriptive analysis, MIS techniques were infrequently used in children, but there was a higher proportion of MIS use among adolescents. There were similar proportions of surgery type across geographic regions within the United States. Whether this infrequent usage is appropriate is as yet unclear given the lack of Level I evidence regarding the relative merits of MIS and open surgery for pediatric and adolescent renal tumors.

Introduction

Minimally invasive surgery (MIS) is increasingly used in adult tumor resections. MIS has been shown to be associated with reduced pain and reduced length of stay, with similar oncologic and long-term survival outcomes.^1–4 However, the use of MIS in pediatric cancer cases has been adopted slowly, due in part to concerns about oncologic outcomes including tumor spillage, ability to achieve negative margins, and nodal harvest, in addition to technical concerns due to tumor size in relatively small abdominal compartments.⁵ A recent survey indicated that 88% of pediatric surgeons favor laparoscopic appendectomy, 90% favor laparoscopic fundoplication, but only 13% favor laparoscopic Wilms tumor (WT) resection.⁶ Indeed, a recent Cochrane review found no published clinical trials comparing MIS to open surgery in pediatric abdominal malignancies.⁷

In a previous study using the National Cancer Database (NCDB), our group found that only 5% of WT patients under went MIS tumor resection.⁵ Importantly, there was no evidence of a difference in MIS utilization in terms of patient age, race or ethnicity, tumor size, stage, or metastatic status. However, patient age was dichotomized at age 5, which may have caused this analysis to overlook differences in management of adolescents and younger children. Further, because of the sampling structure of NCDB, we did not examine geographic differences in MIS utilization; NCDB focuses primarily on higher-volume centers and is not a population-based data source.

In contrast, the National Inpatient Sample (NIS) is sampled from approximately 97% of all inpatient encounters, including low-volume and non-urban hospitals. Due to the anatomical complexity of performing surgery on young children, we hypothesized that older children and adolescents would have increased proportion of a MIS surgery compared to younger pediatric encounters. As secondary objectives, we hypothesized that the odds of post-operative complications would decrease for MIS vs open surgery, inpatient length of stay would be shorter for MIS, and inpatient costs would be increased for MIS.

Methods

Data Source

The Nationwide Inpatient Sample (NIS) is an all-payer database managed by the Healthcare Cost and Utilization Project (HCUP) and sponsored by the Agency for Healthcare Research and Quality. Data in the NIS are from a 20% stratified probability sample of US hospitals based on five hospital characteristics including ownership status, number of beds, teaching status, urban/rural location, and geographic region. NIS includes post-stratification discharge weights that can be used to make national estimates.

Cohort

Encounters in the NIS that were </= 18 years of age at admission and that had a diagnosis for a renal tumor (ICD-9 189.0) and a surgical intervention (partial or radical nephrectomy, 55.4, 55.5) were included. In 2015, NIS shifted from International Classification of Diseases, 9th edition (ICD-9) to ICD-10-based coding; because of this coding shift, it would be difficult to define what findings were based on actual shifts in practice versus what findings were purely a result of differences in coding. We therefore limited our cohort to include only those encounters resulting in an admission between 1998 and 2014.

Outcomes and Variable definitions

The primary outcome of interest was open surgery vs. MIS, defined by presence or absence of an ICD-9 procedure codes for MIS (54.21, 54.51, and/or 17.4x). Secondary outcomes were post-operative complications, inpatient length of stay, and cost of encounter. For secondary outcomes, MIS was investigated as the exposure. Post-operative complications were determined using ICD-9 code definitions based on complications used in the American College of Surgeons’ National Surgical Quality Improvement Project (NSQIP). These included: surgical site infection, acute renal failure, urinary tract infection, respiratory failure, pneumonia, pulmonary embolism, stroke, myocardial infarction, cardiac arrest, post-operative bleeding and deep vein thrombosis (Appendix 1). Because individual complications occurred relatively infrequently, we defined this outcome as a dichotomous marker of whether any complication occurred during each inpatient encounter. Because this may underestimate the true incidence of complications and could potentially introduce bias, we chose to reduce this potential issue by examining “any complication” rather than specific complications. The presence of at least one NSQIP complication occurring during a patient’s index admission qualified as a “yes” for this outcome, similar to previous HCUP analyses.^8–15

Encounters were divided into two age groups: children (0–9 years old) and adolescents (10–18 years old) based on anatomical and biological differences in renal tumors and renal surgery between adolescents and young children. Hospital region was defined based on US Census Bureau definitions: Northeast, Midwest, South and West. Additionally, these data were compared by gender, race (white, black, Hispanic, Asian or Pacific Islander, Native American or other), primary payer (public, private, or other), hospital type (metropolitan teaching, metropolitan non-teaching and non-metropolitan), and bed size (small, medium or large).

Encounter cost was estimated by using the cost-to-charge ratio (CCR) files provided by HCUP.¹⁶ CCR was only released for 2001 to 2014, thus 1998–2000 were excluded from the cost analysis. We adjusted for inflation based on 2014 USD using the Consumer Price Index (CPI)-standard index, obtained from the Bureau of Labor Statistics.¹⁷

The Van Walraven Index (VWI) was used to account for patient comorbidity.¹⁸ As our cohort was composed entirely of children and adolescents who are typically healthy (cancer diagnosis notwithstanding), VWI is not normally distributed. VWI was therefore dichotomized into two groups: those with low (VWI <=0) and increased comorbidity (VWI>0).

Statistical Methods

As recommended by HCUP, weighted descriptive statistics were used to describe the demographic characteristics by age groups. This Pre-accounted for the complex survey structure of the NIS and to make national level inferences. As per HCUP data use agreement, all variables with cell counts less than 15 were excluded. For simplicity, we refer to weighted encounters as ‘encounters,’ but all analyses presented are weighted. A Wald-chi-squared test was used for categorical variables and an unadjusted weighted ANOVA was used for continuous variables.

For the primary objective, a cross tabulation between hospital region, age group and MIS were calculated using weighted descriptive statistics. No formal statistical test was used to compare MIS utilization within age groups by hospital region due to the low event rates for each group limiting the degrees of freedom in a potential model.

Secondary objectives were modeled using weighted generalized estimating equations (GEE), which accounted for both survey weighting and correlation present in the dataset by allowing the variance of hospitals within strata to change by year. The primary predictor for all models was surgical technique (MIS vs open). For postoperative complications, the GEE model was fit with a logit link function and binomial distribution and was adjusted for age group, race, gender and VWI. For length of stay, the GEE model was fit with a negative binomial distribution and log link function and was adjusted for age group, race, gender, VWI, and insurance payor. For inflation-adjusted cost, the GEE model was fit with a Gamma distribution and log link function and was adjusted for race, gender and VWI. Age was excluded as we did not find evidence of a difference in cost by age, consistent with the intuition that age should have a limited impact on cost in a pediatric cohort.

Results

Demographic

There were a total of 9,259 pediatric encounters between 1998 and 2014 with a renal tumor diagnosis and a renal tumor surgical intervention (Table 1). Of these, 91% of pediatric encounters were <9 years old, 50.7% were female, 13.5% were black, 14.5% were Hispanic, and 60.4% had private insurance. VWI was >/= 1 in 32.1% of encounters, with a mean score of 2.6.

Table 1:

Patient demographics, insurance status, comorbidity score, and hospital size and type.

Variables	Children (n = 8436)	Adolescent (n=823)	Total (n = 9259)	P-value
Gender				0.6011
Female	4259 (50.49%)	434 (52.70%)	4693 (50.69%)
Male	4167 (49.39%)	389 (47.30%)	4556 (49.21%)
Race				0.9601
Missing	1586 (18.80%)	156 (19.01%)	1742 (18.82%)
White	3957 (46.91%)	399 (48.46%)	4356 (47.04%)
Black	1106 (13.11%)	143 (17.40%)	1249 (13.49%)
Hispanic	1258 (14.91%)	81 (9.79%)	1338 (14.46%)
Other	530 (6.28%)	44 (5.35%)	574 (6.20%)
Insurance				0.4101
Public	3018 (35.78%)	266 (32.37%)	3285 (35.47%)
Private	5064 (60.02%)	524 (63.67%)	5588 (60.35%)
Other	336 (3.98%)	33 (3.96%)	369 (3.98%)
Comorbidity Score
Mean (SD)	2.49 (0.13)	3.29 (0.42)	2.57 (0.13)	0.0662
Median (IQR)	−0.30 (−0.71,2.85)	−0.47 (−1.33,5.57)	−0.29 (−0.71,2.96)
<=0	5764 (68.33%)	523 (63.56%)	6287 (67.91%)
>=1	2672 (31.67%)	300 (36.44%)	2972 (32.09%)
Hospital bed size				0.1081
Small	1485 (17.60%)	126 (15.28%)	1611 (17.40%)
Medium	1903 (22.56%)	146 (17.74%)	2049 (22.13%)
Large	4942 (58.58%)	543 (65.96%)	5484 (59.23%)
Hospital Type				0.0021
Rural	74 (0.88%)	22 (2.70%)	97 (1.04%)
Urban nonteaching	436 (5.17%)	84 (10.22%)	521 (5.62%)
Urban teaching	7819 (92.69%)	708 (86.06%)	8527 (92.10%)

^{1 –}Wald-Chi-squared

^{2 –}Weighted unadjusted ANOVA

Outcomes

Overall, 1.7% of encounters in children involved use of MIS for renal tumors. Adolescent encounters had a higher proportion of MIS compared to younger children (9.9% vs 1%, p<0.001). The proportion of open surgery to MIS was similar across census region within age groups. For children, the proportion of MIS was ~1% for all four census regions. In contrast, the proportion of MIS for adolescent was ~10% for all regions (data not shown due to sparse outcomes in some regions, in compliance with HCUP reporting restrictions). There were no significant differences between the open and MIS cohorts in terms of race, ethnicity, surgical technique (partial or radical nephrectomy), gender, or insurance status.

There was no evidence of a difference between proportion of post-operative complications, mean length of stay or mean estimated cost by age group (Table 2). The proportion of post-operative complications for the cohort was 13.3% (Table 3). The mean length of stay was 8.9 (SD:0.3) and the median length of stay was 6.4 (IQR: 4.6–9). The mean estimated cost was $34,458 (SE: $1,197) and median estimated cost was $25,491 (IQR: $17,986–38,830).

Table 2.

Outcomes by Patient Encounter Age Group

Pariables	Children (n = 8436)	Adolescent (n=823)	Total (n = 9259)	P-value
MIS	83 (0.99%)	82 (9.91%)	165 (1.78%)	<0.0011
Post-op Complications	1116 (13.23%)	112 (13.59%)	1228 (13.26%)	0.8951
LoS (days in inpatient)
Mean (SD)	8.86 (0.25)	9.13 (1.23)	8.88 (0.27)	0.8252
Median (IQR)	6.53 (4.73,9.00)	5.11 (3.40,8.12)	6.43 (4.60,8.95)
Cost 3
Mean (SD)	$34,389.14 ($1,164.77)	$35,090.36 ($3,422.75)	$34,457.68 ($1,197.00)	0.8302
Median (IQR)	$25,752.25 ($18,276.03, $38,850.49)	$22,486.04 ($16,091.30, $37,766.62)	$25,490.54 ($17,986.30, $38,829.77)

^{1 –}Wald-Chi-squared

^{2 –}Weighted Unadjusted ANOVA

^{3 –}Cost was adjusted to 2014 USD

Table 3.

Secondary Outcomes by Surgical Technique

Variables	MIS (n = 165)	Open (n = 9094)	P-value
Post-operative Complications	22 (13.28%)	1206 (13.26%)	0.9981
LOS (days in inpatient)	8.59 (1.71)	8.89 (0.27)
Mean (SD)	3.87 (2.14,10.33)	6.46 (4.64,8.94)	0.8582
Median (IQR)	8.59 (1.71)	8.89 (0.27)
Cost 3
Mean (SD)	$33,720.41 ($5,964.18)	$34,474.09 ($1,201.40)	0.8992
Median (IQR)	$19,161.54 ($12,947.64, $35,241.45)	$25,676.64 ($18,152.27, $38,836.14)

^{1 –}Wald-Chi-squared

^{2 –}Weighted Unadjusted ANOVA

^{3 –}Cost was adjusted to 2014 USD

Multivariable analysis

When adjusting for covariates, we did not find evidence of a difference in odds between surgery type and post-operative complications (OR: 0.48; 95% CI: 0.16–1.47), a percent change between surgery type and hospital length of stay (RR: 0.80; 95% CI: 0.54–1.18) or a percent change between surgery type and cost (RR: 0.78; 95% CI: 0.57–1.06).

Discussion

In this analysis of the NIS database, we observed that children underwent MIS for resection of renal tumors at a low rate overall, and that MIS utilization varied by age group. While adolescents underwent MIS at higher rates than younger children, >90% of surgical encounters were still performed as an open technique. While there are many reasons to select MIS or open surgical techniques, these data suggest underutilization of MIS among children and adolescents. Importantly, we did not find any significant differences between MIS and open surgeries in terms of geographic disparities in utilization, nor in terms of postoperative complications, LOS or cost.

To date, many studies examining MIS use in pediatric oncology have been single-institution case reports or series.^19–21 Our group previously used the National Cancer Database to compare MIS versus open surgery in children with neuroblastoma or WT. In that analysis, we found MIS rates for WT to be slightly higher at 5%; MIS was more frequently used in older children (>5 years old) with smaller tumors (<10cm).⁵ However, unlike the current analysis, NCDB is not a nationally-representative, population-based sample; it therefore does not necessarily capture low-volume institutions, which one would expect to predominantly perform open rather than MIS procedures. This likely explains the reduced rate of MIS we herein report. Similarly, in a previous analysis we queried NIS to investigate pediatric urologic MIS utilization and found an MIS rate of 5.7% across both benign and oncologic urology cases; again, these rates are likely conservative because hospitals performing <5 MIS procedures per year were excluded.¹³ Taken together, these findings suggest an overall under-utilization of MIS in pediatric patients with renal tumors.

One reported advantage of MIS for treatment of renal tumors is postoperative complication rate.^22–24 In adults, Semerjian et al examined 30-day outcomes between MIS and open kidney procedures in the NSQIP database, noting that. MIS procedures had lower complication rates and return to operating room rates events.²³ This trend of improved postoperative outcomes is also seen in pediatric patients.^7,13 In a previous analysis, we found that MIS had significantly lower odds (OR 0.70) of post-operative complications compared to open surgery for benign urologic surgeries.¹³ In the present analysis focusing solely on renal surgeries, there was no evidence of an association between surgical technique and postoperative complications, even when adjusting for age group.

Other studies have found that MIS for renal tumors was associated with a shorter LOS. Phelps et al found that MIS resection was associated with decreased blood loss, decreased operating time, and shorter hospital stays. The authors suggested that MIS may promote earlier initiation of adjuvant chemotherapy.²⁵ Romao et al compared laparoscopic to open radical nephrectomy in children with renal tumors and found that mean LOS was significantly shorter for the laparoscopic patients (3 vs 6 days).²¹ Of note, renal tumors undergoing laparoscopic nephrectomy in this study were significantly smaller than open (6.6 vs.11.0 cm), potentially predisposing the latter group to a prolonged postoperative stay. By contrast, we did not find an association between surgery type and LOS, potentially because we were not able to account for tumor size as this variable is not available in NIS.

Another common issue with MIS use is its frequently higher operative cost. Interestingly, we did not find evidence of a significant difference in mean cost between MIS and open surgeries ($33,720 vs $34,474, respectively). This is in contrast to historical trends, as MIS is frequently reported to be more expensive than open.^26,27 However, with more surgeon experience, cheaper surgical instruments, and lower complication rates, costs associated with MIS procedures are declining and in this analysis at least appeared comparable to open procedures.^28–30 It is important to consider that hospital costs can originate from multiple silos beyond the operating room; NIS only reports aggregate costs and is incapable of teasing apart these various sources.

The findings of our study must be interpreted in the context of study limitations. Perhaps most notably, we chose to include data only from NIS years 1998–2014. This decision was based on our concerns with administrative data quality related to changes in coding methods that began in 2015 concurrent with a shift to ICD-10. However, as a result of this decision, these data do not reflect one of the more important recent changes in the field of child and adolescent renal tumors: the 2016 UMBRELLA protocol from the Renal Tumor Study Group of the International Society of Pediatric Oncology (SIOP-RTSG).³¹ The UMBRELLA protocol has explicit criteria to guide the use of MIS and open surgical techniques. While we would anticipate an increase in MIS use after 2016, we unfortunately cannot comment on trends after 2014 based on this analysis.

Similarly, NIS represents a 20% stratified sample of US hospital admissions. As such, our reported results may not be generalizable to encounters not in the sample pool. However, NIS does provide meticulous tracking of discharge and hospital weights in order to minimize the risk of sampling bias.³² Additionally, NIS is a retrospective, administrative database that might be affected by miscoding bias. Our analysis relies on the accuracy of the diagnostic and procedure codes included in NIS; in particular, we relied on accurate utilization of MIS codes in order to define this cohort. While the accuracy of NIS is high for an administrative database, it is possible at least some portion of our cohort may be incorrectly coded. Based on our previous analyses with NIS,^8–13,15 it is unlikely that this potential error is a significant source of bias; however, a higher-than-expected degree of coding error (specifically, misidentification of MIS as open surgery) could have biased our results towards the null. Similarly, the significant difference between cohort sizes of the open and MIS groups is a potential source of bias and sampling error; particularly because of the small size of the MIS cohort, even a small number of misclassified surgeries could have had an impact on our findings.

The complications that we identified may represent associated co-morbidities and not true postoperative complications, as NIS does not provide temporal relationships between different diagnosis codes. Additionally, our analysis may underestimate the true incidence of postoperative events. Because NIS represents admission-based rather than patient-based data, it is impossible to track a given patient across time. We were able to assess neither post-hospital outcomes nor whether individual patients were readmitted or underwent repeat procedures. Our picture of postoperative complications is limited by the inpatient-only nature of NIS, precluding analysis of outpatient encounters not captured in the database.

The retrospective nature of NIS also limits available data and possible analyses. Unlike analyses of NCDB or clinical reviews, NIS does not define oncologic outcomes; because these data are not captured in NIS, we are unable to comment on this important aspect of pediatric care. Additionally, we are unable to identify cancer staging and tumor size, which may be important confounders to account for while modeling surgical outcomes. We would note that both MIS and open techniques are allowed under both Children’s Oncology Group (COG) and International Society of Pediatric Oncology (SIOP) renal tumor protocols. As noted previously, choice of surgical modality is likely multifactorial and provider/setting-specific. Either open or MIS surgery may be a more appropriate approach for a particular child based on institutional resources, provider proficiency and surgeon/parental preferences. While our results do not indicate which modality Journalis ‘better,’ they do reflect contemporary clinical practice patterns and outcomes.

Conclusion

MIS techniques are infrequently used in children with renal masses; only 10% of adolescents and 1% of young children undergo MIS for their renal tumor. Whether this infrequent usage is appropriate is as yet unclear given the lack of Level I evidence regarding the relative merits of MIS and open surgery for pediatric and adolescent renal tumors. We did not find evidence of a difference in post-operative complications, inpatient LOS or cost between open and MIS techniques, nor did we find any evidence of regional variation in MIS use.

Abbreviations:

MIS

minimally-invasive surgery

NIS

National Inpatient Sample

SD

Standard Deviation

SE

StandardError

WT

Wilms Tumor

NCDB

National Cancer Database

HCUP

Healthcare Cost and Utilization Project

ICD-9

International Classification of Diseases, 9^th edition

CCR

cost-to-charge ratio

CPI

Consumer Price Index

VWI

Van Walraven Index

IQR

InterQuartile Range

OR

Odds Ratio

RR

Risk Ratio

CI

Confidence Interval

LOS

Length of Stay

Appendix 1 –

NSQIP-Defined Postoperative Complication ICD-9-CM Codes

Complications	ICD-9 codes
Surgical Site Infection (superficial)	998.32
Surgical Site Infection (deep)	998.31
Peritoneal abscess	567.22
Urinary Tract Infection	599
Urinary complications	997.5
Acute Renal Failure	584.x, 586.x
Respiratory complications	997.3
Pneumonia	481–487, 507
Post-operative respiratory insufficiency	518.5
Acute Respiratory Distress Syndrome	518.82
Systemic Sepsis	790.7, 038.x
Pulmonary Emboli	415.1, 415.11, 415.19
Mechanical Ventilation > 96 hours	96.72
Cerebrovascular Accident	997.02
Cardiac Complications	997.1
Myocardial infarction	410.x
Cardiac arrest	427.5
Bleeding	285.1, 998.11
Deep Vein Thrombosis	453.4, 453.40, 453.9