Postoperative Emergency Department Visits After Urinary Stone Surgery: Variation Based on Surgical Modality

Introduction

Urinary stone disease has an estimated prevalence of nearly 9% in the United States. ¹ A significant portion of these patients undergo some form of surgical intervention in their lifetime. ² Indeed, total annual direct and indirect costs related to urolithiasis diagnosis and treatment are estimated to exceed $10 billion, ranking it among the most costly urologic diseases. ³

There is increasing regulatory and cost pressure to reduce unplanned episodes of care, particularly after elective surgery. Although the health policy discourse surrounding unplanned perioperative resource utilization has focused largely on hospital readmission, ^4,5 an increasing number of regulators and insurers are also examining unplanned post-procedure emergency department (ED) visits as an important and relevant quality metric. ^{6 –9} However, the frequency of postoperative ED visits following different modalities of urinary stone surgery is not well characterized.

We aimed at quantifying the frequency and reasons for postoperative ED visits after percutaneous nephrolithotomy (PCNL), ureteroscopy (URS), and extracorporeal shockwave lithotripsy (SWL). In addition, we sought to characterize the clinical and demographic factors that may be associated with unplanned ED visits after stone surgery.

Methods

The Agency for Healthcare Research and Quality's (AHRQ) Healthcare Cost and Utilization Project (HCUP) state databases were used to identify patients undergoing urinary stone surgery. This included patients undergoing PCNL (Current Procedural Terminology [CPT]: 50080, 50081, 50561; International Classification of Disease [ICD]-9: 55.03, 55.04), URS (CPT: 52352, 52353, 52356; ICD-9: 56.0, 56.31), and SWL (CPT: 50590; ICD-9: 98.51) in Florida (2010–2012), Iowa (2010–2012), California (2010–2011), and New York (2006–2012).

The primary study outcome was postoperative ED visits within 30 days of surgery for ambulatory procedures or within 30 days of hospital discharge for inpatient procedures. Reasons for ED visits were identified by using the AHRQ Clinical Classification Software (CCS), which categorizes more than 14,000 ICD-9 diagnosis codes into ∼260 clinically meaningful groupings to facilitate analyses of large datasets. Comorbidity burden was assessed by using the number of clinically significant chronic diagnoses at the time of surgery. A clinically significant comorbidity is defined as one that lasts longer than 12 months and either places limitations on patients' ability to perform self-care or results in the need for ongoing intervention with medical products, services, or special equipment. Urban–rural designation was based on the United States Department of Agriculture Urban Influence Codes for the patient's county of residence and was categorized into large metropolitan (>1 million residents), small metropolitan (<1 million residents), micropolitan (with an urban core of >10,000 residents), or rural (all counties not meeting other criteria). Income was estimated from the median household income of patients' residential zip code and categorized into quartiles.

The HCUP is a collection of national and state health care databases maintained by the AHRQ. Patients undergoing stone surgery were identified by using the State Ambulatory Surgery Database (SASD) and the State Inpatient Database (SID). Postoperative ED visits were identified in the State Emergency Department Database (SEDD). All three datasets were linked by using a de-identified unique patient identifier that applies across all three state datasets across all study years. All HCUP databases include all payers.

Baseline demographic and clinical characteristics of patients undergoing PCNL, URS, and SWL were compared by using chi-square for categorical variables and analysis of variance or Kruskal-Wallis for normally distributed and non-normally distributed continuous variables, respectively. Postoperative ED visit rates were compared across surgery types by using chi-square. Multivariate logistic regression was used to assess the association between surgery type on postoperative ED visits while adjusting for clinical and sociodemographic covariates. A secondary analysis within the URS cohort used chi-square to compare ED visits for patients undergoing URS with concomitant ureteral stent placement (CPT: 52332; ICD-9: 59.8) vs URS alone. The URS+stent subgroup was further stratified into those who had cystoscopic stent removal (CPT: 52310, 52315) within 30 days of original surgery and those who had stent removed via tethering string (defined as no stent removal codes within 30 days of surgery). Statistical significance was defined as p < 0.05. All analyses were performed by using SAS Studio 3.5 (SAS Institute, Inc., Cary, NC). This study was granted exempt status by the Cleveland Clinic Institutional Review Board.

Results

The HCUP state databases were used to identify 321,899 patients undergoing URS (n = 151,006, 46.9%), SWL (n = 128,040, 39.8%), and PCNL (n = 42,853, 13.3%). Table 1 illustrates the baseline clinical and sociodemographic characteristics of the three study groups. The cohort undergoing PCNL was older, had more comorbidities, had a smaller proportion of privately insured patients, and had a greater proportion of racial minorities and low-income patients (p < 0.0001).

The overall postoperative 30-day ED visit rate across all surgery types was 9.7%. Postoperative ED visit rates were highest for PCNL (5636/42,853, 13.2%), followed by URS (16,008/151,006, 10.6%), and were lowest for SWL (9637/128,040, 7.5%; p < 0.0001). Postoperative ED visits occurred a median of 6 (inter-quartile range [IQR] 2–16) days after SWL, 9 (IQR 4–18) days after PCNL, and 4 (IQR 1–11) days after URS (p < 0.0001).

Table 2 outlines the results of multivariate logistic regression analyses to identify factors associated with postoperative ED visits. Even after adjusting for relevant clinical and sociodemographic characteristics, both PCNL (odds ratio [OR] 1.62, 95% confidence interval [CI] 1.56–1.69) and URS (OR 1.33, 95% CI 1.30–1.37) were associated with a higher likelihood of 30-day postoperative ED visits than SWL.

Additional factors associated with postoperative ED visits on multivariate analyses included greater number of comorbidities (OR 1.06, 95% CI 1.06–1.07 per additional significant diagnosis), Hispanic ethnicity (OR 1.06, 95% CI 1.02–1.10), and rural residence (OR 1.28, 95% CI 1.19–1.38). Privately insured patients (OR 0.82, 95% CI 0.79–0.85) and those with the highest income (OR 0.95, 95% CI 0.92–0.99) were less likely to experience postoperative ED visits. Gender was not associated with ED visit rates (p = 0.833).

Table 3 outlines the most common reasons for postoperative ED visits in the first 30 days after PCNL, URS, and SWL. Residual stone was the most common reason for ED visits in the SWL (42.7%) and URS (33.9%) groups, but it was relatively infrequent in the PCNL group (8.1%). Revisits for surgical complications were largely unique to PCNL (9.2%). Abdominal pain, urinary tract infection, and voiding symptoms or hematuria were common across all surgery types.

A subgroup analysis of concomitant stent placement was performed within the URS cohort. There was no difference in 30-day postoperative ED visits in patients who underwent URS alone vs URS with concurrent stent (10.8% vs 10.6%, p = 0.2478). Within the URS with stent subgroup, patients who had stent removed via tethering string had a slightly lower crude ED visit rate than those who underwent cystoscopic removal (10.4% vs 11.6%, p = 0.01). However, this difference was eliminated after adjusting for between-group differences in clinical and sociodemographic characteristics on multivariate regression (OR 0.96, 95% CI 0.88–1.06, p = 0.44).

Discussion

Postoperative ED visits are gaining scrutiny as global health care costs and quality become more prominent issues in the U. S. health care system. In this analysis of HCUP data, we examined 321,899 patients who underwent urinary stone surgery and report an overall 30-day postoperative ED visit rate of 9.7%. PCNL patients visited the ED most frequently at 13.2%, followed by URS and SWL at 10.6% and 7.5%, respectively. This variation across surgical modality persisted on multivariate analyses even after adjusting for relevant clinical and sociodemographic variables.

A previous analysis by Scales and colleagues of 93,523 postoperative stone patients by using the MarketScan database characterized rates of 30-day ED visits and found that SWL had the lowest unplanned visit rate (12%) compared with URS (15%) and PCNL (15%). ⁶ These trends mirrored the findings of this study. With the increasing utilization of national clinical and administrative databases for the purposes of tracking perioperative outcomes in surgery, the differences between these datasets are important to acknowledge. ¹⁰ The MarketScan database utilized by Scales and colleagues is a private analytic database primarily comprising data from employer-sponsored health plans. ¹⁰ This is a different patient population from the HCUP database, which included uninsured, Medicaid, and Medicare patients who might be expected to have different ED utilization tendencies. This is supported by our observation that privately insured patients were significantly less likely to experience postoperative ED visits. The current findings appear more broadly applicable across a variety of insurance types and patient populations. In addition, the current cohort is more than three times larger than the MarketScan cohort used by Scales and colleagues, further adding to its generalizability.

An analysis of 194,781 stone procedures in Ontario examined ED visits within 7 days of surgery as a secondary study outcome. ¹¹ The overall ED visit rate increased significantly from 7% in 2000 to 11% in 2010, which was suggested as being related to the rise in URS and fall in SWL over the same period. ¹¹ The applicability of these trends to the United States is limited by the single-payer system in Canada as well as to differences in access to care. In addition, limiting analyses to the first 7 postoperative days may not adequately capture postoperative ED visits occurring beyond the first week after surgery.

We identified multiple clinical and sociodemographic factors that were associated with postoperative ED visits. Insurance status was associated with returning to the ED, with higher risk for uninsured and Medicaid patients and lower risk for privately insured patients. This trend persisted even after adjusting for comorbidity burden and could reflect differences in access to urologic follow-up care between these groups. Previous literature has suggested higher rates of postoperative ED visits among Medicare patients after a variety of surgery types, ¹² including after URS among patients with public insurance. ⁸ In the current analysis, patients with a higher number of comorbidities were also more likely to visit the ED, a trend previously observed by Scales and colleagues ⁶ Both Scales and colleagues and the current analysis found that higher income was associated with fewer unplanned ED visits, further suggesting a socioeconomic disparity possibly related to health care access.

Using CCS of ICD diagnosis codes enabled us to characterize the reasons for postoperative ED visits, a previously unreported finding in a large database study of PCNL, URS, and SWL patients. Retained stone was the most common reason for postoperative ED visits after SWL and URS, comprising 42% and 33% of visits, respectively. Conversely, only 8% of postoperative ED visits after PCNL were for retained stone. This mirrors existing literature suggesting higher stone-free rates for PCNL than for URS or SWL. ¹³ Interestingly, abdominal pain presentations were slightly more common among URS and SWL, which may reflect the passage of residual stone fragments after surgery. Infection and voiding symptoms were common across all three surgery types. It should be noted that some small portion of the observed ED visits may have been secondary to patient difficulty with accessing appropriate outpatient office-based care, suggesting that access issues may play a role in driving ED visits after stone surgery.

PCNL is a more definitive surgical treatment with correspondingly higher stone-free rates along with greater morbidity. ¹⁴ In our sample, 9% of PCNL patients presented to the ED with surgical complications and 7% presented with implant complications, the latter being a risk unique to PCNL and possibly representative of issues related to nephrostomy tubes. A previous analysis of trends in PCNL observed an overall complication rate of 13.6%, including urinary tract and wound infections, cardiac and vascular events, and miscellaneous medical events. ¹⁵ Overall, this appeared roughly similar to the current ED visit rate of 13.2% for PCNL.

The placement of stents after routine URS is a widely debated topic, primarily due to concerns regarding patient quality of life and discomfort after stent placement. ¹⁶ The patients in our dataset who underwent URS with stent placement did not demonstrate any significant difference in ED visitation rates compared with those without stents. Multiple meta-analyses have found no significant difference in unplanned medical visits within 30 days between these groups, although their criteria for unplanned visits varied and were primarily composed of small, single-center trials. ^{17 –20} A large, retrospective population-based analysis did not find any association between stent placement and ED visit rates or inpatient admissions in the first 7 or 30 days after URS for 16,060 patients. ⁸ A nonrandomized trial found that unplanned ED return rates within 2 weeks of surgery were similar between stented and unstented patients. ²¹ In conjunction with these previously reported findings, it would appear that any patient discomfort attributable to stents does not appear to increase postoperative ED visits. However, due to the nature of our study, we were unable to account for stone burden nor complexity as well as other outcomes such as phone calls, outpatient visits, or early stent removals.

With regards to stent removal via cystoscopy vs tethering string, this study suggests that extraction strings are not associated with increased ER visits on multivariate analyses. This is consistent with prior literature suggesting that strings do not increase unplanned office nor ER visits. ²² Further, Oliver and coworkers previously demonstrated that strings do not impact pain scores nor urinary symptoms, but do entail a 10% risk of stent dislodgement. ²³ Thus, the decision to utilize strings must be tailored to individual surgeon and patient preferences.

Hospital readmission rates have been given substantial attention from both policymakers and insurers as a core surgical quality metric, as illustrated by programs such as the Hospital Readmissions Reduction Program (HRRP). However, it is likely that legislative efforts to reduce unplanned episodes of care will also extend to postoperative ED visits. Indeed, postoperative ED visits are a relevant and important outcome with impacts on cost, quality, and patient morbidity. ED visits add to postoperative costs, which is especially relevant as payment models look to expand beyond the individual episode of surgery to include all global perioperative costs. Postoperative ED visits also have relevance as a surgical quality metric, as these episodes may reflect a breakdown in upstream care transitions and outpatient care coordination. ¹² Finally, and perhaps most importantly, ED visits are a marker for patient morbidity, and should be considered a relevant patient-centered outcome to be minimized after elective surgery.

These study results must be interpreted in the context of limitations in study design. First, we cannot assess stone size nor location, thus limiting correlations between preoperative stone burden and postoperative ED visits. It is possible that some degree of the observed variation in ED visits may be a function of stone complexity rather than a surgical approach. Second, the primary study outcome includes all postoperative ED visits within 30 days of surgery, which may capture visits unrelated to stone disease. However, a review of reasons for postoperative ED visits suggests that a substantial proportion of these visits are, indeed, stone related. Third, although this dataset captures all postoperative ED visits statewide, it cannot identify patients who undergo surgery and have ED visits in different states. However, this is still more robust than institutional analyses, which often cannot capture ED visits outside the study hospital system. Fourth, the reasons for ED visits is inferred based on ICD-9 diagnosis codes, which is an imperfect method. Fifth, our subgroup analysis on stent removal via cystoscopy vs string may not capture stents that are removed during a second staged stone procedure, nor does it take into account patients who may have been pre-stented before their index stone surgery. Sixth, due to lack of granularity in the dataset, we were unable to explore the impacts of various PCNL exit strategies (i.e., nephrostomy tube vs stent) on postoperative ER visits. Seventh, the dataset does not allow for an analysis of practice setting (community vs academic) on postoperative ER visitation rates. In addition, the need for secondary procedures was not included as an outcome, since the dataset does not allow for distinction between planned/staged reoperations and unplanned reoperations that result from incomplete stone clearance. Finally, the final disposition of patients from the ED (admission vs discharge) cannot be determined by using the current dataset, precluding a more in-depth analysis of readmissions.

Despite these methodologic considerations, this study offers novel insight into variations in postoperative ED visits after urinary stone surgery. The rates and reasons for postoperative ED visits reported here could help inform how urologists counsel patients before surgery as well as assist with expectation-setting during the shared decision-making process. This is especially true in the context of shifting trends in the surgical management of stones away from SWL toward increasing PCNL and URS utilization, ^15,24,25 both of which were associated with higher rates of postoperative ED visits than SWL in this study.

Conclusion

Among kidney stone surgeries, PCNL has the highest rate of 30-day postoperative ED visits, whereas SWL has the lowest. These findings highlight an important outcome for both patients and surgeons in the surgical management of stones. The observed differences across surgical modality could help guide the preoperative shared decision-making process regarding the surgical management of stones.