Comparison of Selective Versus Empiric Pharmacologic Preventative Therapy with Kidney Stone Recurrence

Abstract

Objectives:

To assess the effectiveness of an empiric approach to metabolic stone prevention.

Methods:

Using medical claims from a cohort of working age adults with kidney stone diagnoses (2008 to 2017), we identified the subset who were prescribed thiazides, alkali therapy, or allopurinol—collectively known as preventive pharmacological therapy (PPT). We distinguished between those who had 24-hour urine testing prior to initiating PPT (selective therapy) from those without it (empiric therapy). We conducted a survival analysis for time to first recurrence for stone-related events, including ED visits, hospitalizations, and surgery, up to two years after initiating PPT.

Results:

Of 10,125 patients identified, 2,744 (27%) and 7,381 (73%) received selective and empiric therapy, respectively. The overall frequency of any stone-related event was 11%, and this did not differ between the two groups on bivariate analysis (p=0.29). After adjusting for sociodemographic factors, comorbidities, medication class, and adherence; there was no difference in the hazard of a stone-related event between the selective and empiric therapy groups (HR, 0.97; 95% CI, 0.84 to 1.12). When considered individually, the frequency of ED visits, hospitalizations, and surgeries did not differ between groups. Greater adherence to PPT and older age were associated with a lower hazard of a stone-related event (both p<0.05).

Conclusions:

Compared to empiric therapy, PPT guided by 24-hour urine testing, on average, is not associated with a lower hazard of a stone-related event. These results suggest a need to identify kidney stone patients who benefit from 24-hour urine testing.

Introduction

As many as half of all patients with kidney stones will experience a second episode of renal colic within five years of their first, and over 10% can expect to have three or more recurrences.^1,2 Kidney stone recurrence contributes to poorer health outcomes, detriments in health-related quality of life, and costs of care.^3–8 Taken together, kidney stones are best viewed as a chronic disease. Like other chronic diseases, prevention strategies play an important role in kidney stone management. While the mainstay of prevention is increased fluid intake, some patients with kidney stones also benefit from preventive pharmacological therapy (PPT).^9,10 Under contemporary practice guidelines, a selective approach to PPT is recommended, ^11–13 whereby a 24-hour urine is used to identify abnormal urine chemistries.

Yet, while selective therapy can be extremely useful in the academic setting or a research-based kidney stone clinic, it may be impractical in a more clinically focused community practice, where the availability of 24-hour urine testing can be limited, and where many providers may not know how to act on the results from it. This has led some to advocate for an empiric approach, which entails administering PPT based on medical history, serum laboratories, and, if present, stone composition, without identifying any abnormal urinary chemistries.¹⁴ However, for empiric therapy to be a viable alternative, it must be at least equivalent to selective therapy in terms of reducing the risk of recurrent stone events.

To date, the only study comparing selective versus empiric therapy suffered from significant follow-up bias in the selective therapy arm and examined only dietary interventions.¹⁵ To fill this knowledge gap, we conducted an observational study in which we analyzed medical claims data from working age adults with kidney stones. Specifically, we compared the frequency of stone-related events, including ED visits, hospitalizations, and surgery, among those with and without 24-hour urine testing before the initiation of PPT.

Materials and Methods

Data source and study population

For our study, we used Opium’s de-identified Clinformatics® Data Mart Database. This is a commercial and Medicare Advantage US database that captures all inpatient, outpatient, ED, and pharmacy encounters for an estimated 73 million beneficiaries, including children and adults.

The cohort selection is illustrated in Supplementary Figure 1. First, we identified all beneficiaries in the database who were 18 to 64 years of age and had at least two physician-coded diagnoses of kidney stone disease or at least one procedure for kidney stones between January 1, 2008 and December 31, 2017. We considered the date of index stone claim as the occurrence of first of two diagnoses or the procedure. The relevant kidney stone Current Procedural Terminology (CPT) codes and International Classification of Disease (ICD) diagnosis and procedure codes are listed in Supplementary Figure 2. To facilitate comorbidity adjustment, we excluded patients without continuous health insurance coverage in the 12 months prior to their index stone claim. To ensure adequate follow-up data, we excluded patients who did not have continuous insurance coverage at least one year after their index stone claim and excluded those enrolled in a Medicare Advantage plan during either the year prior to or the year after their index stone claim since we cannot track all of their healthcare utilization. To identify patients with kidney stones receiving PPT, we then identified the subset of patients with a prescription filled for a PPT agent (i.e., thiazides, alkali therapy, or allopurinol) within one year after their index stone claim using National Drug Codes.

A complete list of the medications that we included can be found in Supplementary Table 1. We excluded patients who received PPT within the six months preceding their index stone claim and those who did not have continuous enrollment in a drug benefit plan for the year after their index stone claim.

Exposure

We defined patients on selective therapy as those who performed a 24-hour urine collection before their first PPT prescription fill. We determined that a patient performed a collection if there was a claim filed for urine oxalate (CPT 83945), which is specifically used for the biochemical evaluation of kidney stone disease.¹⁶ Those without a 24-hour urine collection prior to PPT initiation were defined as receiving empiric therapy.

Assessing Clinical Health Outcomes

Next, we assessed for recurrence free probability, which we defined as whether each patient had an ED visit, hospitalization, or surgery for stone disease between six months and two years after their initial PPT prescription fill. We used relevant place of service codes from Optum to identify ED visits and hospitalizations associated with a diagnosis of kidney stone disease. We defined the occurrence of stone-directed surgery using both CPT and ICD procedure codes (see Supplementary Figure 2).^17,18

Statistical Analysis

For our initial analytic step, we compared patients receiving selective and empiric PPT across a range of sociodemographic factors, including age, gender, race/ethnicity, education level, region of residence, and level of comorbid illness (assessed with the modified Charlson comorbidity index¹⁹). We also compared patients based on medication factors, including adherence (defined as >80% days covered from start of PPT to 6 months),²⁰ medication classes received, receipt of combination therapy, and whether there were concurrent diagnoses putting the patient at higher risk for kidney stone recurrence (see Supplementary Table 2).²¹ We made bivariate comparisons using t-tests for continuous variables and chi-square tests for categorical variables.

We performed Kaplan Meier survival analysis to evaluate recurrence-free probability from the time of the index stone event, comparing both groups using the log-rank test. We calculated unadjusted frequencies of stone-related ED visit, hospitalization, both overall and individually, among patients receiving selective and empiric PPT with chi-square tests. We then calculated adjusted frequencies of the outcomes by fitting multivariable Cox proportional hazard regression models for our composite outcome and for ED visit, hospitalization, and surgery individually. We confirmed that the proportional hazards assumption was met. Our independent variable was an indicator variable for whether the patient received selective PPT. We controlled for the patient factors described above.

We performed three distinct sensitivity analyses to test the robustness of the results. First, we added hypertension and gout diagnoses to the regression model, since patients receiving empiric PPT were expected to have higher rates of these conditions. In the second analysis, we excluded thiazide class combinations that would typically be expected to be primarily prescribed for hypertension. Finally, we required having both 24-hour urine oxalate and 24-hour urine citrate (CPT 82507) obtained within the same week to be categorized as having received selective therapy.

We conducted all analyses using SAS software, Version 9.4 (SAS Institute Inc., Cary, NC). We performed two-sided significance testing with alpha set at 0.05. The Institutional Review Board at the University of Michigan Health System deemed that this study was exempt from its oversight.

Results

There were 10,125 patients identified meeting inclusion and exclusion criteria (Supplementary Figure 1). Overall, there were 2,744 (27%) who received selective therapy and 7,381 (73%) who received empiric therapy with a median follow up of 730 days (interquartile range, 227 days for empiric group and 236 days for selective group). Table 1 shows patient characteristics by group. Patients receiving empiric PPT tended to be older, more often male, non-white, and had more comorbid conditions when compared to those on selective PPT. Additionally, there were significant differences in the level of education and region of residence between the two groups. Patients on selective therapy were most commonly prescribed citrate, while patients on empiric therapy were most commonly prescribed thiazides. Concurrent diagnoses of hypertension were higher among those receiving empiric PPT than selective PPT (40% vs 13%, respectively; p<0.01).

Table 1.

Comparing patients on selective therapy vs empiric therapy.

Characteristic	Empiric Therapy (n=7,381)	Selective Therapy (n=2,744)	P-Value
Follow up time in days, median (IQR)	730 (227)	730 (236)	0.60
Age in years (%)			<0.001
18 to 34	629 (9)	428 (16)
35 to 44	1,422 (19)	606 (22)
45 to 54	2,570 (35)	867 (32)
55 to 64	2,760 (37)	843 (31)
Female gender (%)	2,804 (38)	1,142 (42)	<0.001
Race/ethnicity (%)			<0.001
White	5,630 (76)	2,257 (82)
Black	715 (10)	179 (7)
Other	1,036 (14)	308 (11)
Education (%)			<0.001
High school or less	2,083 (28)	573 (21)
Some college	4,091 (55)	1,530 (56)
College or more	1,207 (16)	641 (23)
Region of residence (%)			<0.001
East North Central	1,120 (15)	417 (15)
East South Central	341 (5)	66 (2)
Middle Atlantic	438 (6)	174 (6)
Mountain	544 (7)	264 (10)
New England	193 (3)	70 (3)
Pacific	665 (9)	192 (7)
South Atlantic	2,270 (31)	736 (27)
West North Central	677 (9)	335 (12)
West South Central	1,133 (15)	490 (18)
Charlson comorbidity index, mean (SE)	0.5 (0.01)	0.3 (0.02)	<0.001
Hypertension (%)	3,001 (41)	361 (13)	<0.01
High risk stone former (%)	2,160 (29)	749 (27)	0.052
Adherent (%)	2,182 (30)	820 (30)	0.753
Thiazides (%)	4,354 (59)	1,144 (42)	<0.001
Allopurinol (%)	1,540 (21)	445 (16)	<0.001
Alkali Citrate (%)	1,963 (27)	1,547 (56)	<0.001
Combination therapy (%)	454 (6)	368 (13)	<0.001

Abbreviation: SE, standard error of the mean. IQR, interquartile range.

Results from the Kaplan Meier survival analysis are depicted in Figure 1, demonstrating no difference in recurrence-free probability at two years between both groups (0.87 vs 0.88, p=0.25). Figure 2 displays the unadjusted and adjusted frequencies of overall kidney-stone related events. These did not differ significantly between the selective and empiric therapy groups (unadjusted 11% versus 10%, adjusted 12% versus 12%, respectively). When examining ED visits, hospitalizations, and surgery individually, the unadjusted and adjusted frequencies, comparing selective and empiric therapy, also did not differ (ED visits: unadjusted 6% versus 6%, adjusted 7% versus 7%; hospitalizations: unadjusted 2% versus 2%, adjusted 2% versus 2%; surgery: unadjusted 7% vs 6%, adjusted 8% vs 7%).

Figure 1.

Kaplan Meier symptomatic stone recurrence-free probability comparing selective versus empiric pharmacologic preventative therapy

Figure 2.

Unadjusted and adjusted rates of symptomatic stone recurrence over 2 years comparing selective versus empiric pharmacologic preventative therapy

Full multivariate model results are displayed in Table 2. While selective therapy (compared to empiric therapy) was not associated with a lower hazard of a stone-related event (HR, 0.97; 95% CI, 0.84 to 1.12), adherence to pharmacologic therapy (HR, 0.82; 95% CI, 0.71 to 0.95) and older age were (categories of age ≥35 years versus 18 to 34 years, HRs ranged from 0.76 to 0.78, p<0.05 for each comparison). There was no association between the hazard of stone-related event and medication class or with use of combination therapy. All three sensitivity analyses performed similarly showed that selective therapy was not associated with a lower hazard of a stone-related event.

Table 2.

Multivariate Cox proportional hazards model of predictors of composite stone events.

Characteristic	Hazard Ratio	95% Confidence Interval
Selective Therapy (reference: empiric therapy)	0.97	0.84 to 1.12
Combination medical therapy (reference: single agent)	1.04	0.43 to 2.55
Adherent	0.82	0.71 to 0.95
Alkali	1.33	0.58 to 3.05
Thiazide	0.93	0.40 to 2.12
Allopurinol	1.12	0.49 to 2.57
High risk condition	1.04	0.91 to 1.19
Age (reference: 18 to 34)
35 to 44	0.76	0.62 to 0.94
45 to 54	0.78	0.64 to 0.95
55 to 64	0.76	0.62 to 0.93
Female gender	1.01	0.89 to 1.15
Race/ethnicity (reference: white)
Black	0.89	0.71 to 1.11
Other	1.00	0.83 to 1.20
Education (reference: High school or less)
Some college	0.90	0.78 to 1.04
College or more	0.92	0.77 to 1.11
Region of residence (reference: East North Central)
East South Central	0.89	0.63 to 1.26
Middle Atlantic	0.68	0.50 to 0.94
Mountain	0.82	0.62 to 1.07
New England	1.33	0.95 to 1.88
Pacific	0.82	0.62 to 1.07
South Atlantic	1.02	0.85 to 1.23
West North Central	0.95	0.75 to 1.22
West South Central	0.92	0.74 to 1.14
Charlson comorbidity index	1.03	0.98 to 1.08

Comment

There are three principal findings of this study. First, among a cohort of commercially insured adults with kidney stones on PPT, an empiric approach was observed in three-quarters of cases. Since a concomitant diagnosis of hypertension and thiazide use were more common in the empiric PPT cohort, it is likely that many patients in this group were prescribed medications for reasons other than stone prevention. Second, compared to those receiving selective PPT, patients receiving empiric PPT tended to be older, more often male, non-white, and have more comorbid conditions. Citrate was more commonly utilized with selective PPT, while thiazides were more commonly utilized with empiric PPT. Third, there was no observed difference in the frequency of symptomatic stone events between those managed with empiric versus selective PPT by two years.

Prior studies have provided indirect evidence to support empiric approaches for PPT. The identification of baseline urinary calcium, oxalate, and citrate did not predict symptomatic kidney stone recurrence in several randomized controlled diet and pharmacologic trials.¹⁰ Some positive trials of citrate therapy did not require the intervention group to have hypocitraturia.^22,23 The same was true for several thiazide trials that did not require hypercalciuria as an inclusion criteria.^24,25 Our finding of no observed difference in outcomes related to empiric and selective PPT approaches were also similar when examining stone event outcomes separately as stone-related ED visit, hospitalization, or surgery. In this study population, it is likely that the empiric PPT cohort may not have received the medications as the primary intended strategy, since hypertension and use of thiazides were much more prevalent among the empiric PPT group. Yet even so, this should have biased the results to favor the selective PPT cohort, who would have been expected to have benefited from the exposure to intentional clinical encounters for stone prevention, which has been well-described as the “stone clinic effect.”^26,27

Our findings must be interpreted within the context of its limitations. One potential reason that we found no difference in outcomes between empiric and selective approaches to PPT may be inadequate length of follow-up. While we observed patients for up to two years after PPT initiation, prior PPT trials followed patients for three to five years. Due to the relatively low frequency of stone recurrence over time, a longer follow-up period may be needed to detect a difference in stone events between the empiric and selective therapy groups. However, it is worth noting that our study population was considerably larger than any of the prior studies on PPT. As such, we have greater ability to detect even a small difference if one had emerged by two years. While we accounted for observed differences in sociodemographic and medication factors and patient recurrence risk, we cannot exclude the possibility of residual and unmeasured confounding. It is possible that patients on selective therapy were at higher risk of recurrence, though not well captured in administrative data. Administrative data is susceptible to omitted variable bias. The findings of this study of commercially insured adults may not be generalizable to other populations, most notably those who lack health insurance coverage. In addition, we lacked granular clinical information to assess the occurrence of adverse events related to PPT use. There may be misclassification bias of our primary outcome and of the exposure using prescription fills. However, we use an accepted measure of adherence and show that PPT adherence is associated with reduction in recurrence rates, as in a prior administrative claims study.²⁰ We lacked information on dietary interventions for stone prevention, which are often first-line therapy for most stone formers. It is possible that their use may differ between the two groups. The findings of this study should not be used to compare the effectiveness of selective dietary therapy to empiric dietary therapy. Additionally, we did not have stone composition data. Often this information is used to initiate empiric therapy or combined with 24-hour urine chemistry data to guide treatment. Patients may also receive more intensive treatment and follow-up for uncommon stone composition types. For instance, use of alkali therapy, a mainstay of treatment for patients with uric acid and cystine stone types, requires periodic monitoring of urine pH.

Notwithstanding these limitations, our findings have important clinical implications. Our finding of no difference in stone recurrence events comparing selective to empiric PPT is consistent with the American College of Physicians clinical practice guideline supporting preventative pharmacologic monotherapy with a thiazide, citrate, or allopurinol without knowing urine chemistry.²⁸ Importantly, among those receiving PPT, we identified younger age and lower adherence to PPT associating with higher stone recurrence rates. This would indicate that sub-populations of kidney stone formers–particularly those diagnosed at younger age–may benefit from earlier workup and intervention that may include 24-hour urine testing. Along these lines, clinical guidelines from urology and nephrology disciplines recommend the determination of high-risk status and performing testing among high-risk populations.^11–13 Here, we did not identify high-risk diagnoses as a predictor of recurrence in the multivariate model. This may be due to the heterogeneity of diagnoses we examined and inherent limitations of administrative coding. Still, more evidence is needed to elucidate which kidney stone patient populations would benefit from physiologic knowledge of urine abnormalities in order to guide treatment. For example, urine pH and calcium phosphate supersaturation may be important to monitor among patients receiving citrate therapy.²⁹

Future studies could examine longer follow-up periods beyond two years. Additionally, data to support selective therapy, especially among the highest risk populations of kidney stone patients, would contribute to a greater use of 24-hour urine testing and lead to improved health outcomes. Factors that predict reduced recurrence risk among those that receive selective PPT could be examined. Finally, these data can be used to compare the effectiveness of thiazide, allopurinol, and alkali therapy on kidney stone recurrence. Overall, our results support the need for randomized, controlled trials that evaluate the selective strategy of biochemical 24-hour urine testing on stone recurrence outcomes.

Conclusions

Among privately insured patients with kidney stones, biochemical 24-hour urine testing as part of a strategy for selective PPT is not associated with any difference in stone recurrence when compared to empiric pharmacologic therapy. Non-adherence to PPT and younger age are associated with higher recurrence risk. These data support future studies to examine sub-populations in whom physiologic knowledge of urine abnormalities improves clinical outcomes.

Abbreviations

CCI

Charlson Comorbidity Index

CPT

Current Procedural Terminology

ED

Emergency Department

PPT

Preventive Pharmacological Therapy