- Email: [email protected]
Medical Expulsive Therapy versus Early Endoscopic Stone Removal for Acute Renal Colic: An Instrumental Variable Analysis
Urolithiasis/Endourology
Medical Expulsive Therapy versus Early Endoscopic Stone Removal for Acute Renal Colic: An Instrumental Variable Analysis John M. Hollingsworth,* Edward C. Norton, Samuel R. Kaufman, R. Matt Smith, J. Stuart Wolf, Jr. and Brent K. Hollenbeck From the Department of Urology (JMH, SRK, RMS, JSW, BKH) and Center for Healthcare Outcomes & Policy (JMH, ECN, BKH), University of Michigan Medical School, and Department of Health Management and Policy, University of Michigan School of Public Health (ECN), Ann Arbor, Michigan
Abbreviations and Acronyms ED ⫽ emergency department MET ⫽ medical expulsive therapy POS ⫽ point of service PPO ⫽ preferred provider organization Accepted for publication March 12, 2013. Study received institutional review board approval. Supported by Grant 1K08HS020927-01A1 (JMH) from the Agency for Healthcare Research and Quality. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality. * Correspondence: Department of Urology, North Campus Research Complex, 2800 Plymouth Rd., Bldg 16, 1st Floor, #113W, Ann Arbor, Michigan 48109-2800 (telephone: 734-615-3040; FAX: 734-232-2400; e-mail: [email protected]).
For other articles on related topics see pages 1090 and 1096.
Purpose: The use of medical expulsive therapy to hasten stone passage potentially decreases expenditures around episodes of renal colic. However, these efficiency gains may be mitigated if patients treated with medical expulsive therapy have frequent health care encounters due to pain while waiting for the stones to pass. Materials and Methods: Using claims data (2002 to 2006) we identified adult men with acute renal colic. We compared 6-week payments as well as frequency of hospitalization and emergency department revisits associated with an initial course of medical expulsive therapy with those for early endoscopic stone removal. To account for unmeasured confounding we performed an instrumental variable analysis, exploiting variation in recommended treatments based on the day of the week that a patient’s first emergency department visit occurred. Results: Overall 1,835 and 4,397 men underwent medical expulsive therapy or early endoscopic stone removal, respectively. Although minimal differences existed between men with respect to the day of the week of emergency department presentation, weekend encounters were strongly associated with receiving medical expulsive therapy (p ⬍0.001). Two-stage least squares regression revealed 6-week payments to be tenfold lower for men on medical expulsive therapy who were candidates for either treatment (p ⬍0.001). While there was no difference in frequency of hospitalization, these men were more likely to have a repeat emergency department visit compared to those who underwent endoscopic stone removal (68.8% vs 39.6%, respectively, p ⫽ 0.025). Conclusions: Findings on medical expulsive therapy are mixed, with lower 6-week payments but more frequent repeat emergency department visits. These data inform patients who are candidates for medical expulsive therapy or endoscopic stone removal when making decisions about their care. Key Words: urinary calculi, adrenergic alpha-antagonists, calcium channel blockers, ureteroscopy
882
www.jurology.com
URINARY stone disease is common and costly, accounting for nearly 1 million patient visits to emergency departments each year in the United States alone.1,2 Fewer than 5% of patients with acute renal colic require urgent diversion.3,4 For the remaining pa-
tients conservative nonsurgical management with a trial of a calcium channel blocker or alpha-blocker, commonly referred to as medical expulsive therapy,5,6 may or may not be appropriate depending on case characteristics such as stone size and
0022-5347/13/1903-0882/0 THE JOURNAL OF UROLOGY® © 2013 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION
http://dx.doi.org/10.1016/j.juro.2013.03.040 Vol. 190, 882-887, September 2013 RESEARCH, INC. Printed in U.S.A.
AND
MEDICAL EXPULSIVE THERAPY VERSUS EARLY ENDOSCOPIC STONE REMOVAL
location. Some patients would experience equally positive results from surgery or MET. Numerous clinical trials have shown that MET facilitates urinary stone passage.7 To the extent that MET reliably reduces patient need for endoscopic stone removal, its use is expected to decrease expenditures related to the acute stone episode. If the magnitude of these savings is great, efforts to accelerate the dissemination of MET will have implications for payers. However, any cost efficiency gain could be mitigated if patients treated with MET have frequent repeat ED visits or hospital admissions due to pain while waiting for the stones to pass, in which case prompt endoscopic stone removal may be preferable. To identify MET consequences for health spending and resource use we performed an instrumental variable analysis, exploiting variation in recommended therapies based on the day of the week of the patient’s index ED visit.8 Given that this variation should be unrelated to disease severity but is likely associated with doctors’ preferences for how they spend their time, an instrumental variable approach would achieve pseudo-randomization in a nonexperimental setting.
METHODS Data Source We used the MarketScan® Commercial Claims and Encounters Database which contains information on the health care use of 5.6 million adults and their dependents with employer sponsored benefit plans. We chose these data because urinary stone disease affects mainly working age adults.9 We also chose these data because MarketScan data capture all inpatient and outpatient care, including detailed prescription drug information,10 which was important given our focus on pharmacotherapy.
Study Population We used a previously validated, diagnosis code based algorithm to identify adult men in the United States age 18 to 64 years with at least 1 ED claim for urinary stone disease filed between 2002 and 2006.11 We studied men because urinary stone disease shows a male preponderance.9 In addition, MET agents, which are often used for symptom amelioration in benign prostatic hyperplasia, are more commonly prescribed for men.12 We also observed more day-to-day variation in prescribing patterns for men than for women. Therefore, the instrumental variable study design was stronger for men. We required continuous enrollment in a benefit plan for a minimum of 120 days before and 42 days after the incident ED claim. Since MET is only indicated in the patient who is amenable to conservative management, we excluded those who had pyelonephritis or renal failure within the 7 days before and 3 days after the incident ED claim. For this same reason we excluded 312 men who underwent ureteral stenting or nephrostomy tube placement without stone removal during this time. Because we wanted to
883
understand the MET treatment effect when newly prescribed, we excluded 147 men who had been prescribed calcium channel blockers and alpha-blockers during the 30 days before the incident ED visit.
Treatment Variable and Patient Covariates Using National Drug Codes we defined patients treated with an initial course of MET as those prescribed nifedipine, tamsulosin, alfuzosin, terazosin, doxazosin or prazosin during the 3 days after the incident ED claim.13 If patients had a claim for ureteroscopy with laser lithotripsy or removal/manipulation of calculus during this same period, then we said that they underwent early endoscopic stone removal. We characterized patients by age, employee classification (salaried vs nonsalaried), employment status (fulltime vs part-time), benefit plan (comprehensive, PPO, health maintenance organization, POS, other), urban/rural status and geographic region of residence. We assessed comorbidity using a modification of the Charlson index,14 examining claims submitted during the 120 days before the incident ED visit.
Outcome Measures For each patient we calculated total payments made around the acute stone episode using a claims window that began on the incident ED visit date and extended out 6 weeks. This time frame allowed us to capture costs related to stone treatment before average payments returned to baseline.15 We categorized the major components of payments (ie those for ED visit, and inpatient, outpatient, pharmacy and surgical care) and inflationadjusted all payments to 2006 U.S. dollars using the Consumer Price Index.16 Given the skewed distribution, cost data were log transformed for subsequent analyses. In addition, we measured the number, if any, of repeat ED visits and hospitalizations that occurred after the incident ED visit.
Statistical Analysis We evaluated associations between the type of stone treatment (an initial course of MET vs early endoscopic stone removal) and the patient level covariates previously described. We then used t tests to test for differences in average total and component payments between the 2 groups. We also used chi-square tests to compare the proportions of patients in each group who had a repeat ED visit and a hospitalization. One of our concerns was that the decision to treat with MET as opposed to early endoscopic stone removal could be affected by unmeasured (and, therefore, omitted) patient characteristics.17 To address unmeasured confounding we used the econometric technique of instrumental variable analysis.18 For example, if patients prescribed MET tended to have less severe disease (eg smaller or fewer stones) than those treated with early endoscopic stone removal, our estimates of the treatment effect would be biased. In this example one might conclude that MET reduced costs (the average cost being lower), when instead the reduction derived from the relatively healthier state of patients who received MET in the first place. The focus of the instrumental variable analysis is on marginal patients, those for whom either type of treat-
884
MEDICAL EXPULSIVE THERAPY VERSUS EARLY ENDOSCOPIC STONE REMOVAL
ment would have been appropriate. In other words, there are some patients who will almost always undergo surgery (eg an obstructing 1 cm proximal ureteral stone) and some for whom conservative management is most appropriate (eg a nonobstructing 3 mm distal stone). On the margins are patients for whom treatment could go either way (eg a 5 mm mid ureteral stone). The 2 conditions that must be satisfied for an instrument to be considered valid are 1) it must be highly associated with the exposure (an initial course of MET in our case) and 2) it cannot be related to the outcomes of interest except through its effect on the treatment received. The instrument that we selected was based on the day of the week that a patient’s incident ED visit occurred.8 For a variety of medical conditions weekend hospitalizations are associated with less use of health services than those on weekdays.19 –22 In particular, for urinary stone disease, patients admitted during the weekend with renal colic have significantly lower odds of undergoing surgical intervention.19 As we demonstrate, this variation seems unrelated to disease severity and gives us some pseudo-randomization in the data. Factors underlying this weekend effect include resource availability, levels of staffing and even convenience. Based on these practice norms for elective surgery, we constructed indicator variables capturing when the incident ED visit occurred, whether on a weekday (Monday, Tuesday or Wednesday), during the early weekend (Thursday, Friday or Saturday) or on Sunday. We created a separate indicator for Sunday because we anticipated that the decision for surgery would depend on the time of day that the ED visit occurred (which is not captured in the MarketScan data). In addition, patients whose initial ED visit occurred on a Sunday had characteristics different from those whose visit was during other parts of the week, appearing to be a mix of the 2 other groups. To assess instrument validity we evaluated whether incident ED visits in the early weekend and on Sunday were associated with an initial course of MET. Using 2-stage least squares regression we examined the effect of MET on episode payments. We regressed the treatment variable on our instruments and all patient covariates. We obtained predicted values of the treatment variable which were then inserted into our second stage linear regression model along with patient covariates to estimate the logarithm of expenditures. To draw inferences about the mean we transformed model derived values back to their dollar scale by exponentiating the linear index and then multiplying by Duan’s smearing factor to correct for nonnormality in the error term.23 The estimated errors were nonnormal but did not exhibit strong heteroskedasticity in the observed covariates. To determine the effect of MET on the frequency of repeat ED visits and hospitalizations, we also used instrumental variables. However, because having a repeat ED visit or a hospitalization is a dichotomous outcome, we used a bivariate probit model.24 This model uses the instrumental variables to predict whether someone receives MET. However, unlike 2-stage least squares, it is appropriate for binary main outcomes. To test the robustness of our findings we conducted 3 sensitivity analyses. 1) A small number of patients in our
Table 1. Comparison of patients who underwent early endoscopic stone removal vs an initial course of medical expulsive therapy during an acute stone episode
Mean pt age (SD) % Benefit plan type: Comprehensive Exclusive provider organization/PPO POS Other noncapitated % Salaried % Full-time employment % Urban % Region of residence: Northeast Midwest South West % Charlson score: 0 1 2 3 or Greater
Early Endoscopic Stone Removal
Initial Course of MET
47.5 (10.6)
47.1 (10.9)
10.0 0.7
11.2 0.8
12.6 76.7 15.9 50.4 69.4
12.5 75.5 25.7 60.4 82.6
6.9 30.5 54.8 7.9
9.1 32.6 46.5 11.8
88.0 9.4 1.7 1.0
88.4 8.7 2.1 0.8
p Value 0.174 0.590
⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001
0.425
cohort had a prescription fill and endoscopic stone removal within 3 days of the incident ED visit (11, 2.5%). In our primary analysis we placed these patients in the surgery group. We repeated our analysis reclassifying these patients as having been treated with an initial course of MET. 2) We then identified a subgroup of patients with diagnosis codes that have the highest positive predictive value for ureteral stone disease,11 and we fitted separate models for them. 3) Finally, because our coding of a patient as being in one of 2 treatment groups could depend on the length of time we examined the record after the incident ED date, we varied that date, moving it from 3 days to 2 days and then to 4 days. Since results from these analyses were consistent, only varying in the extent of statistical significance, we report only the findings from the main analysis. We performed all analyses using SAS® (version 9.3) and Stata® (version 12.1) statistical software. All tests were 2-tailed with the probability of Type 1 error at 0.05. The institutional review board of the University of Michigan approved this study.
RESULTS We identified 1,835 (29.4%) and 4,397 (70.6%) men who underwent either an initial course of MET or early endoscopic stone removal, respectively. While there were no differences between the 2 groups with respect to patient age and comorbid status, men treated with MET were more often salaried, had full-time employment and resided in an urban area (p ⬍0.001 for each comparison, table 1). Surgery was more common in the South (p ⬍0.001). We observed greater balance in these covariates when we looked across the day-of-the-week instru-
MEDICAL EXPULSIVE THERAPY VERSUS EARLY ENDOSCOPIC STONE REMOVAL
Weekday No. pts Mean pt age (SD) % Benefit plan type: Comprehensive Exclusive provider organization/PPO POS Other noncapitated % Salaried % Full-time employment % Urban % Region of residence: Northeast Midwest South West % Charlson score: 0 1 2 3 or Greater % ED visits at which MET was prescribed
Early Weekend
Sunday
2,868 2,382 982 47.3 (10.7) 47.7 (10.6) 47.3 (10.7) 10.5 0.6
10.2 0.9
10.3 0.8
12.8 76.2 18.7 52.7 73.3
12.2 76.7 20.1 54.4 73.4
12.8 76.1 16.1 52.9 73.2
7.6 32.0 51.4 9.0
7.5 32.1 51.4 9.1
7.4 26.1 57.5 9.0
88.6 8.9 1.7 0.8 26.1
87.6 9.7 1.8 0.9 34.1
87.9 8.7 2.1 1.3 27.9
p Value 0.469 0.856
0.024 0.420 0.993 0.020
0.550
⬍0.001
ments (table 2). Most important, our instruments were strongly associated with an initial course of MET (p ⬍0.001) such that patients seen in the early weekend or on Sunday were more likely to receive this treatment type by about 6 percentage points (fig. 1). This relationship persisted in a multivariable model adjusting for all measured patient characteristics (chi-square statistic 38.6, p ⬍0.001). Only 24.4% of men treated with MET required surgery after the incident ED visit. In our unadjusted analyses patients on MET were less likely to have a repeat ED visit or to be hospitalized (p ⬍0.001 for each
Percentage of ED Visits With Expulsive Therapy
Early Weekend 35%
30% Weekday
25%
20%
Monday
Tuesday
Wednesday Thursday
Friday
Saturday
Sunday
Day of the Week
Figure 1. Percentage of ED visits for acute renal colic at which MET is prescribed stratified by day of week.
80%
Percentage of Patients with Outcome
Table 2. Comparison of patients based on the day of week of their incident ED visit
885
70% 60% 50% 40% 30% 20% 10% 0% ED Revisit
Hospitalization
Unadjusted Results
ED Revisit
Hospitalization
Results Adjusted for Measured and Unmeasured Confounders
Figure 2. Likelihood of repeat ED visit and hospitalization for patients treated with MET (gray) vs early stone directed surgery (black). Adjusted results are from our instrumental variable analysis.
comparison, fig. 2). However, after accounting for unmeasured confounding, no difference was noted in the frequency of hospitalization. Moreover, repeat ED visits were more common among those prescribed MET (p ⫽ 0.025). Regarding episode payments, total expenditures were significantly lower after MET vs endoscopic stone removal ($4,012 vs $9,923, respectively). Similar differences were noted across all component payments except pharmacy claims (table 3). Our 2-stage least squares regression revealed episode payments to be more than tenfold lower for men on MET (p ⬍0.001), which was driven largely by less spending on outpatient care.
DISCUSSION Our analysis suggests that patients who receive an initial course of MET for acute renal colic incur significantly lower expenditures than those treated with early endoscopic stone removal. In light of this finding, efforts to increase MET uptake in the broader medical community are likely to reduce costs associated with the management of urinary stone disease. However, there is a tradeoff for patients as MET use may lead to more repeat ED visits. Studies of MET effects on health spending and resource use have been limited. Consistent with our work, simulation models reveal a fourfold cost advantage of MET compared to endoscopic stone removal.25 However, the rates of repeat ED visits and hospitalizations that we found among MET users were substantially higher than what has been reported in the clinical trial setting.26 Furthermore, pooled data from 7 randomized controlled trials suggest that the frequency of unplanned medical en-
886
MEDICAL EXPULSIVE THERAPY VERSUS EARLY ENDOSCOPIC STONE REMOVAL
Table 3. Major components of payments during an acute stone episode Results Adjusted for Measured ⫹ Unmeasured Confounders (95% CI)
Unadjusted Results (95% CI) Early Endoscopic Stone Removal
Initial Course of MET
Early Endoscopic Stone Removal
Initial Course of MET
$336 ($298–$374)* $941 ($840–$1,042) $3,876 ($3,723–$4,028)* $4,590 ($4,422–$4,759)* $180 ($169–$191)*
$269 ($229–$309)* $731 ($526–$935) $2,025 ($1,863–$2,187)* $763 ($647–$879)* $224 ($204–$244)*
$78 ($58–$165) $101 ($72–$268) $10,271 ($5,014–$34,986)* $2,998 ($2,801–$4,091) $181 ($140–$356)
$162 ($70–$446) $251 ($69–$1,234) $699 ($406–$1,071)* $360 ($37–$4,123) $217 ($104–$580)
ED Inpatient Outpatient Surgical Pharmacy
Adjusted results are from our instrumental variable analysis. * p ⬍ 0.05 for difference between endoscopic stone removal and expulsive therapy.
counters following ureteroscopy is much lower (7%) than what we observed.27 A possible explanation for these discrepancies relates to the difference between efficacy and effectiveness.28 By design, clinical trials are highly constrained so as to establish efficacy (or the therapeutic effect) of a given intervention. The medical claims data that we analyzed reflect the performance of MET and early endoscopic stone removal under real-world conditions, in which patients have multiple, sometimes competing, diagnoses and needs (ie the intervention’s effectiveness). Several limitations merit discussion. Our analysis could be confounded if patients presenting during the weekend were different from those seen on weekdays. However, we observed good covariate balance when stratifying by our instruments. Moreover, the number of ED visits did not vary substantially by day of the week. In addition, the external validity of our findings is limited to working age men. Nevertheless, this is the same group in whom urinary stone disease is most common. In addition, we performed parallel analyses on a combined cohort that included men and women who otherwise met the study requirements. While our instrument did not perform well for women, preventing its use here, multivariable regression results did not differ materially for any of the outcomes (see supplementary Appendix, http://jurology.com/). We also measured MET using pharmacy claims, but a prescription fill does not necessarily indicate that the
medication was taken. As such, our measures of the cost efficiency gains associated with MET use are likely conservative. Finally, while our payment data included all expenditures made on behalf of the beneficiaries, important indirect costs (eg workplace absenteeism and lost productivity) were not captured. As these costs are arguably more important to patients, additional studies examining them are needed.
CONCLUSIONS These limitations notwithstanding, our findings have implications for the treatment of acute renal colic for patients and payers. By demonstrating that treatment with MET, in an effort to avoid surgical intervention and its attendant risks, comes at the expense of more repeat ED visits, physicians will be better able to inform patients making decisions about their own care. Knowing this, in certain scenarios some patients may opt for early endoscopic stone removal. For payers our findings highlight the potential savings associated with greater MET use. Survey data from the United States suggest that MET is prescribed at just a fraction of ED visits for acute renal colic.29 Targeted educational interventions aimed at providers,30 who account for a substantial proportion of the variation in MET use,12 could help rein in health expenditures for urinary stone disease.
REFERENCES 1. Litwin MS and Saigal CS: Urologic Diseases in America. US Department of Health and Human Services, Public Health Service, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. Washington, DC: US Government Printing Office 2012. 2. Foster G, Stocks C and Borofsky MS: Emergency department visits and hospital admissions for kidney stone disease, 2009. HCUP Statistical Brief #139. Agency for Healthcare Research and Quality 2012. Available at www.hcup-
us.ahrq.gov/reports/statbriefs/sb139.pdf. Accessed May 30, 2013.
5. Preminger GM, Tiselius HG, Assimos DG et al: 2007 Guideline for the management of ureteral calculi. Eur Urol 2007; 52: 1610.
3. Gottlieb RH, La TC, Erturk EN et al: CT in detecting urinary tract calculi: influence on patient imaging and clinical outcomes. Radiology 2002; 225: 441.
6. Preminger GM, Tiselius HG, Assimos DG et al: 2007 Guideline for the management of ureteral calculi. J Urol 2007; 178: 2418.
4. Yoshimura K, Utsunomiya N, Ichioka K et al: Emergency drainage for urosepsis associated with upper urinary tract calculi. J Urol 2005; 173: 458.
7. Hollingsworth JM, Rogers MA, Kaufman SR et al: Medical therapy to facilitate urinary stone passage: a meta-analysis. Lancet 2006; 368: 1171. 8. Ho V, Hamilton BH and Roos LL: Multiple approaches to assessing the effects of delays for
MEDICAL EXPULSIVE THERAPY VERSUS EARLY ENDOSCOPIC STONE REMOVAL
hip fracture patients in the United States and Canada. Health Serv Res 2000; 34: 1499. 9. Scales CD Jr, Smith AC, Hanley JM et al: Prevalence of kidney stones in the United States. Eur Urol 2012; 62: 160. 10. Adamson DM, Chang S and Hansen LG: Health research data for the real world: the MarketScan® databases. Ann Arbor: Thomson Medstat 2006. 11. Semins MJ, Trock BJ and Matlaga BR: Validity of administrative coding in identifying patients with upper urinary tract calculi. J Urol 2010; 184: 190. 12. Hollingsworth JM, Wolf JS Jr, Faerber GJ et al: Understanding the barriers to the dissemination of medical expulsive therapy. J Urol 2010; 184: 2368. 13. U.S. Food and Drug Administration: National Drug Code Directory 2012. Available at www.accessdata. fda.gov/scripts/cder/ndc/default.cfm. Accessed May 30, 2013.
16. US Department of Labor, Bureau of Labor Statistics: Consumer Price Index 2012. Available at www.bls.gov/cpi/. Accessed May 30, 2013. 17. Stukel TA, Fisher ES, Wennberg DE et al: Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods. JAMA 2007; 297: 278.
887
23. Duan N: Smearing estimate: a nonparametric retransformation method. J Am Stat Assoc 1983; 78: 605. 24. Bhattacharya J, Goldman D and McCaffrey D: Estimating probit models with self-selected treatments. Stat Med 2006; 25: 389. 25. Bensalah K, Pearle M and Lotan Y: Cost-effectiveness of medical expulsive therapy using alpha-blockers for the treatment of distal ureteral stones. Eur Urol 2008; 53: 411.
18. Angrist JD, Imbens GW and Rubin DB: Identification of causal effects using instrumental variables. J Am Stat Assoc 1996; 91: 444.
26. Cooper JT, Stack GM and Cooper TP: Intensive medical management of ureteral calculi. Urology 2000; 56: 575.
19. Strope SA, Ye Z, Hollingsworth JM et al: Weekend admission and treatment of patients with renal colic: a case of avoidable variation? Urology 2009; 73: 720.
27. Nabi G, Cook J, N=Dow J et al: Outcomes of stenting after uncomplicated ureteroscopy: systematic review and meta-analysis. BMJ 2007; 334: 572.
20. Kostis WJ, Demissie K, Marcella SW et al: Weekend versus weekday admission and mortality from myocardial infarction. N Engl J Med 2007; 356: 1099.
14. Klabunde CN, Potosky AL, Legler JM et al: Development of a comorbidity index using physician claims data. J Clin Epidemiol 2000; 53: 1258.
21. Barnett MJ, Kaboli PJ, Sirio CA et al: Day of the week of intensive care admission and patient outcomes: a multisite regional evaluation. Med Care 2002; 40: 530.
15. Hollingsworth JM, Saigal CS, Lai JC et al: Medicare payments for outpatient urological surgery by location of care. J Urol 2012; 188: 2323.
22. Bell CM and Redelmeier DA: Waiting for urgent procedures on the weekend among emergently hospitalized patients. Am J Med 2004; 117: 175.
28. Harris KM and Remler DK: Who is the marginal patient? Understanding instrumental variables estimates of treatment effects. Health Serv Res 1998; 33: 1337. 29. Hollingsworth JM, Davis MM, West BT et al: Trends in medical expulsive therapy use for urinary stone disease in U.S. emergency departments. Urology 2009; 74: 1206. 30. Brede C, Hollingsworth JM, Faerber GJ et al: Medical expulsive therapy for ureteral calculi in the real world: targeted education increases use and improves patient outcome. J Urol 2010; 183: 585.
Medical Expulsive Therapy versus Early Endoscopic Stone Removal for Acute Renal Colic: An Instrumental Variable Analysis John M. Hollingsworth,* Edward C. Norton, Samuel R. Kaufman, R. Matt Smith, J. Stuart Wolf, Jr. and Brent K. Hollenbeck From the Department of Urology (JMH, SRK, RMS, JSW, BKH) and Center for Healthcare Outcomes & Policy (JMH, ECN, BKH), University of Michigan Medical School, and Department of Health Management and Policy, University of Michigan School of Public Health (ECN), Ann Arbor, Michigan
Abbreviations and Acronyms ED ⫽ emergency department MET ⫽ medical expulsive therapy POS ⫽ point of service PPO ⫽ preferred provider organization Accepted for publication March 12, 2013. Study received institutional review board approval. Supported by Grant 1K08HS020927-01A1 (JMH) from the Agency for Healthcare Research and Quality. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality. * Correspondence: Department of Urology, North Campus Research Complex, 2800 Plymouth Rd., Bldg 16, 1st Floor, #113W, Ann Arbor, Michigan 48109-2800 (telephone: 734-615-3040; FAX: 734-232-2400; e-mail: [email protected]).
For other articles on related topics see pages 1090 and 1096.
Purpose: The use of medical expulsive therapy to hasten stone passage potentially decreases expenditures around episodes of renal colic. However, these efficiency gains may be mitigated if patients treated with medical expulsive therapy have frequent health care encounters due to pain while waiting for the stones to pass. Materials and Methods: Using claims data (2002 to 2006) we identified adult men with acute renal colic. We compared 6-week payments as well as frequency of hospitalization and emergency department revisits associated with an initial course of medical expulsive therapy with those for early endoscopic stone removal. To account for unmeasured confounding we performed an instrumental variable analysis, exploiting variation in recommended treatments based on the day of the week that a patient’s first emergency department visit occurred. Results: Overall 1,835 and 4,397 men underwent medical expulsive therapy or early endoscopic stone removal, respectively. Although minimal differences existed between men with respect to the day of the week of emergency department presentation, weekend encounters were strongly associated with receiving medical expulsive therapy (p ⬍0.001). Two-stage least squares regression revealed 6-week payments to be tenfold lower for men on medical expulsive therapy who were candidates for either treatment (p ⬍0.001). While there was no difference in frequency of hospitalization, these men were more likely to have a repeat emergency department visit compared to those who underwent endoscopic stone removal (68.8% vs 39.6%, respectively, p ⫽ 0.025). Conclusions: Findings on medical expulsive therapy are mixed, with lower 6-week payments but more frequent repeat emergency department visits. These data inform patients who are candidates for medical expulsive therapy or endoscopic stone removal when making decisions about their care. Key Words: urinary calculi, adrenergic alpha-antagonists, calcium channel blockers, ureteroscopy
882
www.jurology.com
URINARY stone disease is common and costly, accounting for nearly 1 million patient visits to emergency departments each year in the United States alone.1,2 Fewer than 5% of patients with acute renal colic require urgent diversion.3,4 For the remaining pa-
tients conservative nonsurgical management with a trial of a calcium channel blocker or alpha-blocker, commonly referred to as medical expulsive therapy,5,6 may or may not be appropriate depending on case characteristics such as stone size and
0022-5347/13/1903-0882/0 THE JOURNAL OF UROLOGY® © 2013 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION
http://dx.doi.org/10.1016/j.juro.2013.03.040 Vol. 190, 882-887, September 2013 RESEARCH, INC. Printed in U.S.A.
AND
MEDICAL EXPULSIVE THERAPY VERSUS EARLY ENDOSCOPIC STONE REMOVAL
location. Some patients would experience equally positive results from surgery or MET. Numerous clinical trials have shown that MET facilitates urinary stone passage.7 To the extent that MET reliably reduces patient need for endoscopic stone removal, its use is expected to decrease expenditures related to the acute stone episode. If the magnitude of these savings is great, efforts to accelerate the dissemination of MET will have implications for payers. However, any cost efficiency gain could be mitigated if patients treated with MET have frequent repeat ED visits or hospital admissions due to pain while waiting for the stones to pass, in which case prompt endoscopic stone removal may be preferable. To identify MET consequences for health spending and resource use we performed an instrumental variable analysis, exploiting variation in recommended therapies based on the day of the week of the patient’s index ED visit.8 Given that this variation should be unrelated to disease severity but is likely associated with doctors’ preferences for how they spend their time, an instrumental variable approach would achieve pseudo-randomization in a nonexperimental setting.
METHODS Data Source We used the MarketScan® Commercial Claims and Encounters Database which contains information on the health care use of 5.6 million adults and their dependents with employer sponsored benefit plans. We chose these data because urinary stone disease affects mainly working age adults.9 We also chose these data because MarketScan data capture all inpatient and outpatient care, including detailed prescription drug information,10 which was important given our focus on pharmacotherapy.
Study Population We used a previously validated, diagnosis code based algorithm to identify adult men in the United States age 18 to 64 years with at least 1 ED claim for urinary stone disease filed between 2002 and 2006.11 We studied men because urinary stone disease shows a male preponderance.9 In addition, MET agents, which are often used for symptom amelioration in benign prostatic hyperplasia, are more commonly prescribed for men.12 We also observed more day-to-day variation in prescribing patterns for men than for women. Therefore, the instrumental variable study design was stronger for men. We required continuous enrollment in a benefit plan for a minimum of 120 days before and 42 days after the incident ED claim. Since MET is only indicated in the patient who is amenable to conservative management, we excluded those who had pyelonephritis or renal failure within the 7 days before and 3 days after the incident ED claim. For this same reason we excluded 312 men who underwent ureteral stenting or nephrostomy tube placement without stone removal during this time. Because we wanted to
883
understand the MET treatment effect when newly prescribed, we excluded 147 men who had been prescribed calcium channel blockers and alpha-blockers during the 30 days before the incident ED visit.
Treatment Variable and Patient Covariates Using National Drug Codes we defined patients treated with an initial course of MET as those prescribed nifedipine, tamsulosin, alfuzosin, terazosin, doxazosin or prazosin during the 3 days after the incident ED claim.13 If patients had a claim for ureteroscopy with laser lithotripsy or removal/manipulation of calculus during this same period, then we said that they underwent early endoscopic stone removal. We characterized patients by age, employee classification (salaried vs nonsalaried), employment status (fulltime vs part-time), benefit plan (comprehensive, PPO, health maintenance organization, POS, other), urban/rural status and geographic region of residence. We assessed comorbidity using a modification of the Charlson index,14 examining claims submitted during the 120 days before the incident ED visit.
Outcome Measures For each patient we calculated total payments made around the acute stone episode using a claims window that began on the incident ED visit date and extended out 6 weeks. This time frame allowed us to capture costs related to stone treatment before average payments returned to baseline.15 We categorized the major components of payments (ie those for ED visit, and inpatient, outpatient, pharmacy and surgical care) and inflationadjusted all payments to 2006 U.S. dollars using the Consumer Price Index.16 Given the skewed distribution, cost data were log transformed for subsequent analyses. In addition, we measured the number, if any, of repeat ED visits and hospitalizations that occurred after the incident ED visit.
Statistical Analysis We evaluated associations between the type of stone treatment (an initial course of MET vs early endoscopic stone removal) and the patient level covariates previously described. We then used t tests to test for differences in average total and component payments between the 2 groups. We also used chi-square tests to compare the proportions of patients in each group who had a repeat ED visit and a hospitalization. One of our concerns was that the decision to treat with MET as opposed to early endoscopic stone removal could be affected by unmeasured (and, therefore, omitted) patient characteristics.17 To address unmeasured confounding we used the econometric technique of instrumental variable analysis.18 For example, if patients prescribed MET tended to have less severe disease (eg smaller or fewer stones) than those treated with early endoscopic stone removal, our estimates of the treatment effect would be biased. In this example one might conclude that MET reduced costs (the average cost being lower), when instead the reduction derived from the relatively healthier state of patients who received MET in the first place. The focus of the instrumental variable analysis is on marginal patients, those for whom either type of treat-
884
MEDICAL EXPULSIVE THERAPY VERSUS EARLY ENDOSCOPIC STONE REMOVAL
ment would have been appropriate. In other words, there are some patients who will almost always undergo surgery (eg an obstructing 1 cm proximal ureteral stone) and some for whom conservative management is most appropriate (eg a nonobstructing 3 mm distal stone). On the margins are patients for whom treatment could go either way (eg a 5 mm mid ureteral stone). The 2 conditions that must be satisfied for an instrument to be considered valid are 1) it must be highly associated with the exposure (an initial course of MET in our case) and 2) it cannot be related to the outcomes of interest except through its effect on the treatment received. The instrument that we selected was based on the day of the week that a patient’s incident ED visit occurred.8 For a variety of medical conditions weekend hospitalizations are associated with less use of health services than those on weekdays.19 –22 In particular, for urinary stone disease, patients admitted during the weekend with renal colic have significantly lower odds of undergoing surgical intervention.19 As we demonstrate, this variation seems unrelated to disease severity and gives us some pseudo-randomization in the data. Factors underlying this weekend effect include resource availability, levels of staffing and even convenience. Based on these practice norms for elective surgery, we constructed indicator variables capturing when the incident ED visit occurred, whether on a weekday (Monday, Tuesday or Wednesday), during the early weekend (Thursday, Friday or Saturday) or on Sunday. We created a separate indicator for Sunday because we anticipated that the decision for surgery would depend on the time of day that the ED visit occurred (which is not captured in the MarketScan data). In addition, patients whose initial ED visit occurred on a Sunday had characteristics different from those whose visit was during other parts of the week, appearing to be a mix of the 2 other groups. To assess instrument validity we evaluated whether incident ED visits in the early weekend and on Sunday were associated with an initial course of MET. Using 2-stage least squares regression we examined the effect of MET on episode payments. We regressed the treatment variable on our instruments and all patient covariates. We obtained predicted values of the treatment variable which were then inserted into our second stage linear regression model along with patient covariates to estimate the logarithm of expenditures. To draw inferences about the mean we transformed model derived values back to their dollar scale by exponentiating the linear index and then multiplying by Duan’s smearing factor to correct for nonnormality in the error term.23 The estimated errors were nonnormal but did not exhibit strong heteroskedasticity in the observed covariates. To determine the effect of MET on the frequency of repeat ED visits and hospitalizations, we also used instrumental variables. However, because having a repeat ED visit or a hospitalization is a dichotomous outcome, we used a bivariate probit model.24 This model uses the instrumental variables to predict whether someone receives MET. However, unlike 2-stage least squares, it is appropriate for binary main outcomes. To test the robustness of our findings we conducted 3 sensitivity analyses. 1) A small number of patients in our
Table 1. Comparison of patients who underwent early endoscopic stone removal vs an initial course of medical expulsive therapy during an acute stone episode
Mean pt age (SD) % Benefit plan type: Comprehensive Exclusive provider organization/PPO POS Other noncapitated % Salaried % Full-time employment % Urban % Region of residence: Northeast Midwest South West % Charlson score: 0 1 2 3 or Greater
Early Endoscopic Stone Removal
Initial Course of MET
47.5 (10.6)
47.1 (10.9)
10.0 0.7
11.2 0.8
12.6 76.7 15.9 50.4 69.4
12.5 75.5 25.7 60.4 82.6
6.9 30.5 54.8 7.9
9.1 32.6 46.5 11.8
88.0 9.4 1.7 1.0
88.4 8.7 2.1 0.8
p Value 0.174 0.590
⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001
0.425
cohort had a prescription fill and endoscopic stone removal within 3 days of the incident ED visit (11, 2.5%). In our primary analysis we placed these patients in the surgery group. We repeated our analysis reclassifying these patients as having been treated with an initial course of MET. 2) We then identified a subgroup of patients with diagnosis codes that have the highest positive predictive value for ureteral stone disease,11 and we fitted separate models for them. 3) Finally, because our coding of a patient as being in one of 2 treatment groups could depend on the length of time we examined the record after the incident ED date, we varied that date, moving it from 3 days to 2 days and then to 4 days. Since results from these analyses were consistent, only varying in the extent of statistical significance, we report only the findings from the main analysis. We performed all analyses using SAS® (version 9.3) and Stata® (version 12.1) statistical software. All tests were 2-tailed with the probability of Type 1 error at 0.05. The institutional review board of the University of Michigan approved this study.
RESULTS We identified 1,835 (29.4%) and 4,397 (70.6%) men who underwent either an initial course of MET or early endoscopic stone removal, respectively. While there were no differences between the 2 groups with respect to patient age and comorbid status, men treated with MET were more often salaried, had full-time employment and resided in an urban area (p ⬍0.001 for each comparison, table 1). Surgery was more common in the South (p ⬍0.001). We observed greater balance in these covariates when we looked across the day-of-the-week instru-
MEDICAL EXPULSIVE THERAPY VERSUS EARLY ENDOSCOPIC STONE REMOVAL
Weekday No. pts Mean pt age (SD) % Benefit plan type: Comprehensive Exclusive provider organization/PPO POS Other noncapitated % Salaried % Full-time employment % Urban % Region of residence: Northeast Midwest South West % Charlson score: 0 1 2 3 or Greater % ED visits at which MET was prescribed
Early Weekend
Sunday
2,868 2,382 982 47.3 (10.7) 47.7 (10.6) 47.3 (10.7) 10.5 0.6
10.2 0.9
10.3 0.8
12.8 76.2 18.7 52.7 73.3
12.2 76.7 20.1 54.4 73.4
12.8 76.1 16.1 52.9 73.2
7.6 32.0 51.4 9.0
7.5 32.1 51.4 9.1
7.4 26.1 57.5 9.0
88.6 8.9 1.7 0.8 26.1
87.6 9.7 1.8 0.9 34.1
87.9 8.7 2.1 1.3 27.9
p Value 0.469 0.856
0.024 0.420 0.993 0.020
0.550
⬍0.001
ments (table 2). Most important, our instruments were strongly associated with an initial course of MET (p ⬍0.001) such that patients seen in the early weekend or on Sunday were more likely to receive this treatment type by about 6 percentage points (fig. 1). This relationship persisted in a multivariable model adjusting for all measured patient characteristics (chi-square statistic 38.6, p ⬍0.001). Only 24.4% of men treated with MET required surgery after the incident ED visit. In our unadjusted analyses patients on MET were less likely to have a repeat ED visit or to be hospitalized (p ⬍0.001 for each
Percentage of ED Visits With Expulsive Therapy
Early Weekend 35%
30% Weekday
25%
20%
Monday
Tuesday
Wednesday Thursday
Friday
Saturday
Sunday
Day of the Week
Figure 1. Percentage of ED visits for acute renal colic at which MET is prescribed stratified by day of week.
80%
Percentage of Patients with Outcome
Table 2. Comparison of patients based on the day of week of their incident ED visit
885
70% 60% 50% 40% 30% 20% 10% 0% ED Revisit
Hospitalization
Unadjusted Results
ED Revisit
Hospitalization
Results Adjusted for Measured and Unmeasured Confounders
Figure 2. Likelihood of repeat ED visit and hospitalization for patients treated with MET (gray) vs early stone directed surgery (black). Adjusted results are from our instrumental variable analysis.
comparison, fig. 2). However, after accounting for unmeasured confounding, no difference was noted in the frequency of hospitalization. Moreover, repeat ED visits were more common among those prescribed MET (p ⫽ 0.025). Regarding episode payments, total expenditures were significantly lower after MET vs endoscopic stone removal ($4,012 vs $9,923, respectively). Similar differences were noted across all component payments except pharmacy claims (table 3). Our 2-stage least squares regression revealed episode payments to be more than tenfold lower for men on MET (p ⬍0.001), which was driven largely by less spending on outpatient care.
DISCUSSION Our analysis suggests that patients who receive an initial course of MET for acute renal colic incur significantly lower expenditures than those treated with early endoscopic stone removal. In light of this finding, efforts to increase MET uptake in the broader medical community are likely to reduce costs associated with the management of urinary stone disease. However, there is a tradeoff for patients as MET use may lead to more repeat ED visits. Studies of MET effects on health spending and resource use have been limited. Consistent with our work, simulation models reveal a fourfold cost advantage of MET compared to endoscopic stone removal.25 However, the rates of repeat ED visits and hospitalizations that we found among MET users were substantially higher than what has been reported in the clinical trial setting.26 Furthermore, pooled data from 7 randomized controlled trials suggest that the frequency of unplanned medical en-
886
MEDICAL EXPULSIVE THERAPY VERSUS EARLY ENDOSCOPIC STONE REMOVAL
Table 3. Major components of payments during an acute stone episode Results Adjusted for Measured ⫹ Unmeasured Confounders (95% CI)
Unadjusted Results (95% CI) Early Endoscopic Stone Removal
Initial Course of MET
Early Endoscopic Stone Removal
Initial Course of MET
$336 ($298–$374)* $941 ($840–$1,042) $3,876 ($3,723–$4,028)* $4,590 ($4,422–$4,759)* $180 ($169–$191)*
$269 ($229–$309)* $731 ($526–$935) $2,025 ($1,863–$2,187)* $763 ($647–$879)* $224 ($204–$244)*
$78 ($58–$165) $101 ($72–$268) $10,271 ($5,014–$34,986)* $2,998 ($2,801–$4,091) $181 ($140–$356)
$162 ($70–$446) $251 ($69–$1,234) $699 ($406–$1,071)* $360 ($37–$4,123) $217 ($104–$580)
ED Inpatient Outpatient Surgical Pharmacy
Adjusted results are from our instrumental variable analysis. * p ⬍ 0.05 for difference between endoscopic stone removal and expulsive therapy.
counters following ureteroscopy is much lower (7%) than what we observed.27 A possible explanation for these discrepancies relates to the difference between efficacy and effectiveness.28 By design, clinical trials are highly constrained so as to establish efficacy (or the therapeutic effect) of a given intervention. The medical claims data that we analyzed reflect the performance of MET and early endoscopic stone removal under real-world conditions, in which patients have multiple, sometimes competing, diagnoses and needs (ie the intervention’s effectiveness). Several limitations merit discussion. Our analysis could be confounded if patients presenting during the weekend were different from those seen on weekdays. However, we observed good covariate balance when stratifying by our instruments. Moreover, the number of ED visits did not vary substantially by day of the week. In addition, the external validity of our findings is limited to working age men. Nevertheless, this is the same group in whom urinary stone disease is most common. In addition, we performed parallel analyses on a combined cohort that included men and women who otherwise met the study requirements. While our instrument did not perform well for women, preventing its use here, multivariable regression results did not differ materially for any of the outcomes (see supplementary Appendix, http://jurology.com/). We also measured MET using pharmacy claims, but a prescription fill does not necessarily indicate that the
medication was taken. As such, our measures of the cost efficiency gains associated with MET use are likely conservative. Finally, while our payment data included all expenditures made on behalf of the beneficiaries, important indirect costs (eg workplace absenteeism and lost productivity) were not captured. As these costs are arguably more important to patients, additional studies examining them are needed.
CONCLUSIONS These limitations notwithstanding, our findings have implications for the treatment of acute renal colic for patients and payers. By demonstrating that treatment with MET, in an effort to avoid surgical intervention and its attendant risks, comes at the expense of more repeat ED visits, physicians will be better able to inform patients making decisions about their own care. Knowing this, in certain scenarios some patients may opt for early endoscopic stone removal. For payers our findings highlight the potential savings associated with greater MET use. Survey data from the United States suggest that MET is prescribed at just a fraction of ED visits for acute renal colic.29 Targeted educational interventions aimed at providers,30 who account for a substantial proportion of the variation in MET use,12 could help rein in health expenditures for urinary stone disease.
REFERENCES 1. Litwin MS and Saigal CS: Urologic Diseases in America. US Department of Health and Human Services, Public Health Service, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. Washington, DC: US Government Printing Office 2012. 2. Foster G, Stocks C and Borofsky MS: Emergency department visits and hospital admissions for kidney stone disease, 2009. HCUP Statistical Brief #139. Agency for Healthcare Research and Quality 2012. Available at www.hcup-
us.ahrq.gov/reports/statbriefs/sb139.pdf. Accessed May 30, 2013.
5. Preminger GM, Tiselius HG, Assimos DG et al: 2007 Guideline for the management of ureteral calculi. Eur Urol 2007; 52: 1610.
3. Gottlieb RH, La TC, Erturk EN et al: CT in detecting urinary tract calculi: influence on patient imaging and clinical outcomes. Radiology 2002; 225: 441.
6. Preminger GM, Tiselius HG, Assimos DG et al: 2007 Guideline for the management of ureteral calculi. J Urol 2007; 178: 2418.
4. Yoshimura K, Utsunomiya N, Ichioka K et al: Emergency drainage for urosepsis associated with upper urinary tract calculi. J Urol 2005; 173: 458.
7. Hollingsworth JM, Rogers MA, Kaufman SR et al: Medical therapy to facilitate urinary stone passage: a meta-analysis. Lancet 2006; 368: 1171. 8. Ho V, Hamilton BH and Roos LL: Multiple approaches to assessing the effects of delays for
MEDICAL EXPULSIVE THERAPY VERSUS EARLY ENDOSCOPIC STONE REMOVAL
hip fracture patients in the United States and Canada. Health Serv Res 2000; 34: 1499. 9. Scales CD Jr, Smith AC, Hanley JM et al: Prevalence of kidney stones in the United States. Eur Urol 2012; 62: 160. 10. Adamson DM, Chang S and Hansen LG: Health research data for the real world: the MarketScan® databases. Ann Arbor: Thomson Medstat 2006. 11. Semins MJ, Trock BJ and Matlaga BR: Validity of administrative coding in identifying patients with upper urinary tract calculi. J Urol 2010; 184: 190. 12. Hollingsworth JM, Wolf JS Jr, Faerber GJ et al: Understanding the barriers to the dissemination of medical expulsive therapy. J Urol 2010; 184: 2368. 13. U.S. Food and Drug Administration: National Drug Code Directory 2012. Available at www.accessdata. fda.gov/scripts/cder/ndc/default.cfm. Accessed May 30, 2013.
16. US Department of Labor, Bureau of Labor Statistics: Consumer Price Index 2012. Available at www.bls.gov/cpi/. Accessed May 30, 2013. 17. Stukel TA, Fisher ES, Wennberg DE et al: Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods. JAMA 2007; 297: 278.
887
23. Duan N: Smearing estimate: a nonparametric retransformation method. J Am Stat Assoc 1983; 78: 605. 24. Bhattacharya J, Goldman D and McCaffrey D: Estimating probit models with self-selected treatments. Stat Med 2006; 25: 389. 25. Bensalah K, Pearle M and Lotan Y: Cost-effectiveness of medical expulsive therapy using alpha-blockers for the treatment of distal ureteral stones. Eur Urol 2008; 53: 411.
18. Angrist JD, Imbens GW and Rubin DB: Identification of causal effects using instrumental variables. J Am Stat Assoc 1996; 91: 444.
26. Cooper JT, Stack GM and Cooper TP: Intensive medical management of ureteral calculi. Urology 2000; 56: 575.
19. Strope SA, Ye Z, Hollingsworth JM et al: Weekend admission and treatment of patients with renal colic: a case of avoidable variation? Urology 2009; 73: 720.
27. Nabi G, Cook J, N=Dow J et al: Outcomes of stenting after uncomplicated ureteroscopy: systematic review and meta-analysis. BMJ 2007; 334: 572.
20. Kostis WJ, Demissie K, Marcella SW et al: Weekend versus weekday admission and mortality from myocardial infarction. N Engl J Med 2007; 356: 1099.
14. Klabunde CN, Potosky AL, Legler JM et al: Development of a comorbidity index using physician claims data. J Clin Epidemiol 2000; 53: 1258.
21. Barnett MJ, Kaboli PJ, Sirio CA et al: Day of the week of intensive care admission and patient outcomes: a multisite regional evaluation. Med Care 2002; 40: 530.
15. Hollingsworth JM, Saigal CS, Lai JC et al: Medicare payments for outpatient urological surgery by location of care. J Urol 2012; 188: 2323.
22. Bell CM and Redelmeier DA: Waiting for urgent procedures on the weekend among emergently hospitalized patients. Am J Med 2004; 117: 175.
28. Harris KM and Remler DK: Who is the marginal patient? Understanding instrumental variables estimates of treatment effects. Health Serv Res 1998; 33: 1337. 29. Hollingsworth JM, Davis MM, West BT et al: Trends in medical expulsive therapy use for urinary stone disease in U.S. emergency departments. Urology 2009; 74: 1206. 30. Brede C, Hollingsworth JM, Faerber GJ et al: Medical expulsive therapy for ureteral calculi in the real world: targeted education increases use and improves patient outcome. J Urol 2010; 183: 585.