Outpatient Antibiotic Use is Not Associated with an Increased Risk of First-Time Symptomatic Kidney Stones

Abstract

Significance Statement

Antibiotics modify human microbiomes and may contribute to kidney stone risk. In a population-based case-control study using 1247 chart-validated first-time symptomatic kidney stone formers and 4024 age- and sex-matched controls, the risk of kidney stones was transiently higher during the first year after antibiotic use. However, this risk was no longer evident after adjustment for comorbidities and excluding participants with prior urinary symptoms. Findings were consistent across antibiotic classes and the number of antibiotic courses received. This suggests that antibiotics are not important risk factors of kidney stones. Rather, kidney stones when they initially cause urinary symptoms are under-recognized, resulting in antibiotic use before a formal diagnosis of kidney stones (i.e., reverse causality).

Background

Antibiotics modify gastrointestinal and urinary microbiomes, which may contribute to kidney stone formation. This study examined whether an increased risk of a first-time symptomatic kidney stone episode follows antibiotic use.

Methods

A population-based case-control study surveyed 1247 chart-validated first-time symptomatic kidney stone formers with a documented obstructing or passed stone (cases) in Olmsted County, Minnesota, from 2008 to 2013 and 4024 age- and sex-matched controls. All prescriptions for outpatient oral antibiotic use within 5 years before the onset of symptomatic stone for the cases and their matched controls were identified. Conditional logistic regression estimated the odds ratio (OR) of a first-time symptomatic kidney stone across time after antibiotic use. Analyses were also performed after excluding cases and controls with prior urinary tract infection or hematuria because urinary symptoms resulting in antibiotic prescription could have been warranted because of undiagnosed kidney stones.

Results

The risk of a symptomatic kidney stone was only increased during the 1-year period after antibiotic use (unadjusted OR, 1.31; P = 0.001), and this risk was attenuated after adjustment for comorbidities (OR, 1.16; P = 0.08). After excluding cases and controls with prior urinary symptoms, there was no increased risk of a symptomatic kidney stone during the 1-year period after antibiotic use (unadjusted OR, 1.04; P = 0.70). Findings were consistent across antibiotic classes and the number of antibiotic courses received.

Conclusions

The increased risk of a first-time symptomatic kidney stone with antibiotic use seems largely due to both comorbidities and prescription of antibiotics for urinary symptoms. Under-recognition of kidney stones that initially cause urinary symptoms resulting in antibiotic use may explain much of the perceived stone risk with antibiotics (i.e., reverse causality).

Introduction

The human microbiome, particularly in the gastrointestinal tract, plays an important role in host metabolism and overall health.¹ Stone formers are known to have a different composition and reduced diversity of their gastrointestinal microbiome compared with healthy participants.^2–5 The disruption of the gastrointestinal microbiome may be involved in the pathogenesis of kidney stone formation through the gut-kidney axis by altering the metabolism, absorption, and secretion within the intestine and ultimately the urinary excretion of metabolites, such as oxalate.^6,7 In addition, the urinary microbiome of stone formers appears distinct from healthy participants^8–11 and may directly influence crystal formation and stone growth within the kidney.⁶ Therefore, it is biologically plausible that antibiotics could increase the risk of kidney stones through the gastrointestinal and urinary microbiome.^12,13

Prior studies have reported an increased risk of kidney stones after antibiotic use. A large, electronic health record–based study demonstrated that the prior use of specific oral antibiotic classes, namely sulfas, cephalosporins, fluoroquinolones, nitrofurantoin/methenamine, and broad-spectrum penicillins, was associated with an increased risk of kidney stones, with a greater risk observed in a more recent period after antibiotic use.¹⁴ A questionnaire-based study found that the use of antibiotics for more than 2 months was associated with a higher risk of kidney stones later in life.¹⁵ However, these observational studies lacked detailed chart validation of kidney stone episodes, which was needed to clarify the timing of the stones and antibiotic use and to determine whether the stone was a first-time symptomatic, recurrent symptomatic, or asymptomatic (incidental on imaging) one. Importantly, urinary tract symptoms (e.g., urgency, dysuria, flank pain) and hematuria are common symptoms among first-time symptomatic stone formers¹⁶ and may result in antibiotic prescription for a perceived or real urinary tract infection (UTI) before a kidney stone is actually seen (by imaging or passage) and diagnosed. Therefore, it is possible that the increased risk of kidney stones associated with antibiotics that was reported in prior studies could be because of reverse causality (i.e., urinary symptoms caused by kidney stones result in antibiotic prescription).

To address these limitations, we performed a population-based case-control study using chart-validated first-time symptomatic kidney stone formers and matched controls. The objective was to determine whether there was an increased risk of a first-time symptomatic kidney stone after oral antibiotic use.

Methods

Study Design and Population

This study was approved by the Mayo Clinic and Olmsted Medical Center institutional review boards with a waiver of informed consent for medical record review. This population-based matched case-control study was conducted using the Rochester Epidemiology Project (REP), which is a medical record linkage system comprising nearly all medical care facilities in Olmsted County, Minnesota.^17,18 Cases were confirmed first-time symptomatic kidney stone formers previously used for similar case-control studies to define clinical associations and risk factors of kidney stones.^19,20 Briefly, manual medical record review by trained nurse abstractors and physicians was performed on all potential kidney stone formers in the county from 1984 to 2013 as initially identified from The International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes (592, 594, and 274.11) to confirm a first-time symptomatic kidney stone episode and to determine the exact date of symptomatic stone onset (index date), which often differed from the date of the ICD-9 diagnosis code. Symptomatic stone formers were defined as having flank/abdominal pain or gross hematuria and either a stone obstructing the ureter on imaging or a voided stone. Study nephrologists and urologists reviewed medical records to adjudicate the diagnosis and onset of symptomatic stones when there was uncertainty. Patients with prior kidney stone episodes, only asymptomatic kidney stones incidentally found on imaging, or only a suspected symptomatic kidney stone not confirmed with an obstructing stone on imaging or by a voided stone were excluded.

Each case was matched on age on the index date (±1 year) and sex to four controls who resided in Olmsted County. Controls were randomly selected from among all the residents of Olmsted County who met the matching criteria using the REP database.^17,18 The index date of stone formers' matched controls was the date of the first symptomatic stone event in the stone formers. Cases could not serve as controls for other cases, even before their index date, and each control was only matched to one case. Controls with a documented history of kidney stone before the index date or who lacked evidence of residency in Olmsted County after the match date were excluded. Because outpatient prescription data in the REP were fully available starting from 2003, only cases and matched controls whose index date was from 2008 to 2013 and who were Olmsted County residents during the 5 years before their index date were included to allow for complete ascertainment of antibiotic use within the 5-year period before the index date. Comorbidities before the index date were identified from ICD-9 codes for obesity, diabetes, hypertension, and CKD (Supplemental Table 1).

Characterization of Outpatient Oral Antibiotic Use

Data on antibiotic prescriptions were obtained electronically from the REP outpatient prescription database. We could not confirm whether prescribed antibiotics were dispensed and used by patients. The exposure of interest was an outpatient oral antibiotic prescribed from 1 day to 5 years before the index date. Topical, otic, and ophthalmic antibiotics were excluded because of their limited systemic bioavailability. Intravenous and intramuscular antibiotics were not available in the outpatient prescription database. If an individual received more than 1 antibiotic prescription in this 5-year period, the one closest to the index date was used for the main analysis. The timing of antibiotic use before the index date was categorized as follows: 1 day–1 year, 1–2 years, 2–5 years, and no antibiotic use within 5 years before the index date. Duration and dosage were not consistently available and thus were not considered in the analyses. Antibiotics were grouped according to class, including sulfas, cephalosporins, penicillins, fluoroquinolones, nitrofurantoin/methenamine/fosfomycin, metronidazole, macrolides, tetracyclines, and clindamycin. The total of number of antibiotic prescriptions during the 5 years before the index date was also assessed.

Sensitivity Analysis to Address Potential Reverse Causality

Potential bias from inaccurate timing of antibiotic prescription and the actual date of onset of a symptomatic kidney stone was largely mitigated by our careful medical record review. Nonetheless, there were many instances in which it was unclear whether patients' urinary symptoms resulting in antibiotic prescription were because of a kidney stone with or without a concurrent UTI on the basis of available information in the medical record. Therefore, we performed sensitivity analyses to exclude patients with urinary symptoms before the index date because these symptoms could potentially be related to an undiagnosed symptomatic kidney stone. First, an analysis was performed after excluding cases and controls who either had a UTI or hematuria diagnosis (on the basis of ICD-9/Hospital International Classification of Disease Adaptation codes, see Supplemental Table 1) or were prescribed antibiotics used exclusively for UTIs (nitrofurantoin, methenamine, or fosfomycin) within 5 years before their index date. The use of these diagnosis codes to identify urinary symptoms was found to have a sensitivity of 60% and specificity of 98% when validated against manual medical record review in a random sample of 207 stone formers. As a second step to capture urinary symptoms missed by diagnosis codes, medical record review was additionally performed to determine indications for all antibiotics prescribed between 1 and 90 days before the index date in both cases and controls. Subsequent analysis was then performed after additionally excluding both cases and controls who received antibiotic prescriptions for urinary symptoms on the basis of medical record review within 90 days before the index date.

Statistical Analysis

Continuous variables were summarized as mean (SD), and categorical variables were summarized as n (%). Conditional logistic regression models were fit to determine the associations between antibiotic use and other patient characteristics with a first-time symptomatic kidney stone while accounting for the matched design for age and sex. The odds ratio (OR) and corresponding 95% confidence interval (CI) for a first-time symptomatic kidney stone during each period of antibiotic use before the index date was calculated using “no antibiotic use within 5 years before the index date” as the reference group. ORs were calculated both unadjusted and adjusted for ethnicity (White versus non-White), obesity, diabetes mellitus, hypertension, and CKD. The association between antibiotic use and a first-time symptomatic kidney stone were additionally analyzed after stratifying by antibiotic class and considering the total number of antibiotic prescriptions received within the 5 years before the index date (0 [reference group], 1, 2, 3, 4, 5+).

Additional analyses were performed to assess the association between a first-time symptomatic kidney stone and antibiotic exposure, defined as having antibiotic use from 1 day to 1 year before the index date, both overall and within each of the following subgroups: age (<45 versus 45+ years), sex (male versus female), obesity, diabetes mellitus, hypertension, and CKD (yes versus no). Subgroup analyses were performed using unconditional logistic regression because matching was broken.

Among antibiotic users within 5 years before their index date, smoothing splines using 10 degrees of freedom were constructed to estimate the odds of being a symptomatic kidney stone former after antibiotic use, both in the full cohort and in the cohort excluding patients with prior urinary symptoms. P-values <0.05 were considered statistically significant. All statistical analyses were performed using the SAS Version 9.4 software package (SAS Institute) and with R version 4.1.2 using the rms package.

Results

Study Sample

There were 1247 confirmed first-time symptomatic kidney stone formers (cases) matched 4:1 to 4988 random controls from among all residents in Olmsted County. After excluding 118 controls for a kidney stone before the index date and 846 controls for lack of evident residency in Olmsted County at least 5 years before and any time after the index date, there were 4024 age- and sex-matched controls. The final samples of cases and controls used in analyses are shown in Figure 1. The mean (SD) age was 47 (16) years, and 2937 (56%) were male. There were 829 (66%) stone formers and 2610 (65%) controls who received antibiotic prescriptions within 5 years before their index date. Table 1 compares clinical characteristics and comorbidities of stone formers and controls at the time of their index date. Stone formers were more likely to have obesity, diabetes, hypertension, and CKD. Stone formers were also more likely to have prior urinary symptoms and to have received five or more courses of antibiotics in the past 5 years. Within 90 days before the index date, stone formers were more likely to receive antibiotic prescriptions for urinary tract symptoms than controls (22% versus 12%) while there was comparable proportion of stone formers and controls who received antibiotic prescription for respiratory tract symptoms (41% versus 42%) or skin/soft-tissue infection (21% versus 20%).

Figure 1.

Flowchart of inclusion and exclusion criteria. *Urinary symptom is defined as having a diagnostic code of UTI or hematuria or receiving a prescription of antibiotics with an indication exclusively for UTIs (including nitrofurantoin, methenamine, or fosfomycin).

Table 1.

Clinical characteristics and comorbidities of all first-time symptomatic stone formers (cases) and matched controls (N=5271)

Patient Characteristic	Total (N=5271)	Stone Formers (N=1247)	Controls (N=4024)	P Value
On the index date
Age, yr, mean (SD)	47.4 (16.0)	46.1 (16.0)	47.8 (15.9)	a
Male sex, n (%)	2937 (55.7)	698 (56.0)	2239 (55.6)	a
White, n (%)	4685 (89.5)	1095 (89.1)	3590 (89.6)	0.54
Obesity, n (%)	979 (18.6)	317 (25.4)	662 (16.5)	<0.001b
Diabetes, n (%)	848 (16.1)	250 (20.0)	598 (14.9)	<0.001b
Hypertension, n (%)	1359 (25.8)	336 (26.9)	1023 (25.4)	0.004b
CKD, n (%)	292 (5.5)	90 (7.2)	202 (5.0)	<0.001b
In past 5 yr, n (%)
Prior urinary symptomsc	977 (18.5)	355 (28.5)	622 (15.5)	<0.001b
Any antibiotics	3439 (65.2)	829 (66.5)	2610 (64.9)	0.10
Number of antibiotics, n (%)
0	1832 (34.8)	418 (33.5)	1414 (35.1)	REF
1	1027 (19.5)	223 (17.9)	804 (20.0)	0.75
2	702 (13.3)	163 (13.1)	539 (13.4)	0.54
3	482 (9.1)	103 (8.3)	379 (9.4)	0.85
4	314 (6.0)	78 (6.3)	236 (5.9)	0.22
5+	914 (17.3)	262 (21.1)	652 (16.2)	<0.001b

P-values derived from conditional logistic regression models to account for age and sex matching.

^aNo P-values reported as variables used in matching and were effectively adjusted for in the conditional logistic regression.

^bP-values denote statistical significance at the 0.05 α level.

^cDefined as having a diagnosis code of a urinary tract infection or hematuria or receiving prescription of antibiotics with an indication exclusively for urinary tract infections (including nitrofurantoin, methenamine, or fosfomycin).

Among the stone formers, 580 (46.5%) had a stone composition analysis. Among those with a known stone composition, the defining compositions based on mutually exclusive categories²¹ were majority calcium oxalate monohydrate (n=385, 66.4%), majority calcium oxalate dihydrate (n=50, 8.6%), majority hydroxyapatite (n=108, 18.6%), any uric acid (n=27, 4.7%), any struvite (n=7, 1.21%), any brushite (n=2, 0.34%), and any cystine (n=1, 0.17%) stones.

Risk of a Kidney Stone after Antibiotic Use

Compared with no antibiotic use, the unadjusted risk of a first-time symptomatic kidney stone was significantly increased within 1 year after antibiotic use (OR, 1.31; P = 0.001); however, the risk was not significantly higher within 1–5 years after antibiotic use (Table 2). After multivariable adjustment, the risk of a first-time symptomatic kidney stone within 1 year after antibiotic use was attenuated and no longer statistically significant (OR, 1.16; P = 0.08). When the analysis was stratified by antibiotic class, the risk of a first-time symptomatic kidney stone was significantly increased for certain antibiotic classes: within 1 year after sulfas and within 2 years for penicillins, fluoroquinolones, nitrofurantoin/methenamine/fosfomycin, metronidazole, and macrolides (Supplemental Table 2). These associations, however, were attenuated and no longer statistically significant after multivariable adjustment, except for sulfas, fluoroquinolones, nitrofurantoin/methenamine/fosfomycin, and metronidazole.

Table 2.

Association of a first-time symptomatic kidney stone with different periods of antibiotic use among all included stone formers and matched controls (N=5271)

Time from Antibiotic to the Index Date	Stone Formers (N=1247)	Controls (N=4024)	Unadjusted		Adjusteda
			OR (95% CI)	P Value	OR (95% CI)	P Value
1 d–1 yr, n (%)	416 (33.4)	1139 (28.3)	1.31 (1.12 to 1.54)	0.001b	1.16 (0.98 to 1.37)	0.08
1–2 yr, n (%)	177 (14.2)	570 (14.2)	1.09 (0.89 to 1.34)	0.39	1.02 (0.83 to 1.26)	0.85
2–5 yr, n (%)	236 (18.9)	901 (22.4)	0.92 (0.77 to 1.11)	0.39	0.87 (0.72 to 1.04)	0.12
No antibiotics,c n (%)	418 (33.5)	1414 (35.1)	REF	REF	REF	REF

P-values derived from conditional logistic regression models to account for age and sex matching. OR, odds ratio; CI, confidence interval.

^aAdjusted for ethnicity, obesity, diabetes, hypertension, CKD.

^bP-values denote statistical significance at the 0.05 α level.

^cNo antibiotics prescribed within 5 years before the index date.

Risk of a Kidney Stone after Antibiotic Use in the Absence of Urinary Symptoms

When the study sample was limited to 892 stone formers and 3402 controls with no prior urinary symptoms before the index date (Figure 1), 515 stone formers (58%) and 2028 controls (60%) received antibiotics within the prior 5 years. The exclusion of prior urinary symptoms considerably attenuated the risk of a first-time symptomatic kidney stone in the early period after antibiotic use (Figure 2). Antibiotic use in any of the periods within 5 years before the index date was not significantly associated with the risk of a first-time symptomatic kidney stone, in unadjusted or adjusted analyses (Table 3). Similarly, there were no significant associations of the risk of a first-time symptomatic kidney stone with the use of any specific antibiotic classes in any periods (Table 4) or with the number of antibiotic courses received during the previous 5 years (Supplemental Table 3). There was no significant association between antibiotic use and the risk of calcium-based stones (Supplemental Table 4) or calcium oxalate stones (Supplemental Table 5). Results were also consistent in subgroup analyses stratified by age, sex, diabetes, hypertension, and CKD (P for interaction >0.05 for all) (Table 5). The risk of a symptomatic kidney stone with antibiotic exposure was higher in obese participants, compared with nonobese participants (OR, 1.35 versus 0.85, respectively; P for interaction=0.02), although the risk remained not statistically significant in obese participants (OR, 1.35; P = 0.08).

Figure 2.

Odds of a being stone former relative to the time between antibiotic use and index date among antibiotic users. Lines are estimated using smoothing splines with 10 degrees of freedom and shown from 5 years to 1 day before the index date. Gray solid line—full cohort (N=3439). Black dotted line—cohort after exclusion of patients with prior urinary symptoms (N=2543).

Table 3.

Association of a first-time symptomatic kidney stone with different periods of antibiotic use among stone formers and matched controls with no prior urinary symptoms within 5 years (N=4294)

Time from Last Antibiotic to Index Date	Stone Formers (N=892)	Controls (N=3402)	Unadjusted		Adjusteda
			OR (95% CI)	P Value	OR (95% CI)	P Value
1 d–1 yr, n (%)	226 (25.3)	845 (24.8)	1.04 (0.85 to 1.27)	0.70	0.95 (0.78 to 1.17)	0.64
1–2 yr, n (%)	115 (12.9)	438 (12.9)	1.01 (0.79 to 1.29)	0.94	0.96 (0.75 to 1.24)	0.78
2–5 yr, n (%)	174 (19.5)	745 (21.9)	0.91 (0.74 to 1.13)	0.39	0.87 (0.70 to 1.08)	0.20
No antibiotics,b n (%)	377 (42.3)	1374 (40.4)	REF	REF	REF	REF

P-values derived from conditional logistic regression models to account for age and sex matching. P-values denote statistical significance at the 0.05 α level. OR, odds ratio; CI, confidence interval.

^aAdjusted for ethnicity, obesity, diabetes, hypertension, and CKD.

^bNo antibiotics prescribed within 5 years before the index date.

Table 4.

Association of a first-time symptomatic kidney stone with different periods of antibiotic use by antibiotic classes among stone formers and matched controls with no prior urinary symptoms within 5 years (N=4294)

Antibiotic Class	Stone Formers (N=892)	Controls (N=3402)	Unadjusted		Adjusteda
			OR (95% CI)	P Value	OR (95% CI)	P Value
Sulfa, n (%)	56 (6.3)	226 (6.6)
1 d–1 yr	20 (2.2)	59 (1.7)	1.01 (0.59 to 1.75)	0.96	0.91 (0.52 to 1.59)	0.73
1–2 yr	8 (0.9)	55 (1.6)	0.59 (0.27 to 1.29)	0.19	0.61 (0.28 to 1.33)	0.21
2–5 yr	28 (3.1)	112 (3.3)	1.04 (0.67 to 1.63)	0.86	1.02 (0.65 to 1.61)	0.92
No antibioticsb	836 (93.7)	3176 (93.4)	REF	REF	REF	REF
Cephalosporin, n (%)	154 (17.3)	592 (17.4)
1 d–1 yr	37 (4.2)	156 (4.6)	0.83 (0.56 to 1.22)	0.34	0.75 (0.51 to 1.11)	0.15
1–2 yr	24 (2.7)	118 (3.5)	0.79 (0.49 to 1.26)	0.31	0.73 (0.46 to 1.18)	0.20
2–5 yr	93 (10.4)	318 (9.4)	1.15 (0.89 to 1.50)	0.29	1.05 (0.81 to 1.38)	0.70
No antibioticsb	738 (82.7)	2810 (82.6)	REF	REF	REF	REF
Penicillin, n (%)	256 (28.7)	968 (28.4)
1 d–1 yr	81 (9.1)	308 (9.1)	1.06 (0.81 to 1.40)	0.66	0.96 (0.73 to 1.28)	0.79
1–2 yr	60 (6.7)	203 (6.0)	1.19 (0.87 to 1.64)	0.28	1.08 (0.78 to 1.50)	0.64
2–5 yr	115 (12.9)	457 (13.4)	1.05 (0.83 to 1.33)	0.68	1.02 (0.80 to 1.29)	0.90
No antibioticsb	636 (71.3)	2434 (71.6)	REF	REF	REF	REF
Fluoroquinolone, n (%)	149 (16.7)	563 (16.5)
1 d–1 yr	34 (3.8)	162 (4.8)	0.92 (0.62 to 1.37)	0.67	0.82 (0.56 to 1.24)	0.35
1–2 yr	37 (4.2)	115 (3.4)	1.26 (0.84 to 1.89)	0.26	1.13 (0.75 to 1.72)	0.55
2–5 yr	78 (8.7)	286 (8.4)	1.15 (0.87 to 1.53)	0.34	1.09 (0.82 to 1.45)	0.57
No antibioticsb	743 (83.3)	2839 (83.5)	REF	REF	REF	REF
Metronidazole, n (%)	40 (4.5)	133 (3.9)
1 d–1 yr	12 (1.4)	33 (1.0)	1.46 (0.71 to 3.01)	0.31	1.43 (0.69 to 2.97)	0.34
1–2 yr	10 (1.1)	29 (0.9)	1.12 (0.53 to 2.39)	0.77	0.98 (0.45 to 2.12)	0.96
2–5 yr	18 (2.0)	71 (2.1)	1.08 (0.62 to 1.90)	0.78	1.00 (0.57 to 1.77)	0.99
No antibioticsb	852 (95.5)	3269 (96.1)	REF	REF	REF	REF
Macrolide, n (%)	257 (28.8)	952 (28.0)
1 d–1 yr	84 (9.4)	286 (8.4)	1.20 (0.91 to 1.59)	0.20	1.13 (0.85 to 1.51)	0.39
1–2 yr	57 (6.4)	179 (5.3)	1.40 (0.99 to 1.97)	0.054	1.33 (0.94 to 1.88)	0.11
2–5 yr	116 (13.0)	487 (14.3)	0.99 (0.78 to 1.25)	0.93	0.96 (0.76 to 1.22)	0.73
No antibioticsb	635 (71.2)	2450 (72.0)	REF	REF	REF	REF
Tetracycline, n (%)	74 (8.3)	302 (8.9)
1 d–1 yr	23 (2.6)	82 (2.4)	1.02 (0.62 to 1.67)	0.95	1.01 (0.61 to 1.66)	0.98
1–2 yr	17 (1.9)	65 (1.9)	1.10 (0.62 to 1.95)	0.75	1.01 (0.57 to 1.81)	0.97
2–5 yr	34 (3.8)	155 (4.6)	0.90 (0.60 to 1.34)	0.61	0.85 (0.56 to 1.27)	0.42
No antibioticsb	818 (91.7)	3100 (91.1)	REF	REF	REF	REF
Clindamycin, n (%)	17 (1.9)	66 (1.9)
1 d–1 yr	5 (0.6)	15 (0.4)	1.25 (0.43 to 3.65)	0.68	1.09 (0.37 to 3.24)	0.88
1–2 yr	3 (0.3)	14 (0.4)	1.07 (0.28 to 4.08)	0.92	0.77 (0.20 to 2.99)	0.71
2–5 yr	9 (1.0)	37 (1.1)	1.08 (0.50 to 2.37)	0.84	1.15 (0.52 to 2.53)	0.74
No antibioticsb	875 (98.1)	3336 (98.1)	REF	REF	REF	REF

P-values derived from conditional logistic regression models to account for age and sex matching. P-values denote statistical significance at the 0.05 α level. OR, odds ratio; CI, confidence interval.

^aAdjusted for ethnicity, obesity, diabetes, hypertension, and CKD.

^b.No antibiotics prescribed within 5 years before the index date.

Table 5.

Association of a first-time symptomatic kidney stone with antibiotic exposure across different subgroups among stone formers and matched controls with no prior urinary symptoms within 5 years (N=4294)

Subgroup	Stone Formers (N=892)		Controls (N=3402)
	Antibiotic exposurea	No Antibiotic exposurea	Antibiotic exposurea	No Antibiotic exposurea	OR (95% CI)	P Value	P for Interaction
Overall, n (%)	226 (25.3)	666 (74.7)	845 (24.8)	2557 (75.2)	1.03 (0.87 to 1.22)	0.76
Age, n (%)							0.82
<45	109 (24.7)	332 (75.3)	356 (24.6)	1093 (75.4)	1.01 (0.79 to 1.29)	0.95
>45	117 (25.9)	334 (74.1)	489 (25.0)	1464 (75.0)	1.06 (0.83 to 1.33)	0.69
Sex, n (%)							0.52
M	123 (21.8)	441 (78.2)	448 (22.0)	1593 (78.1)	0.99 (0.79 to 1.24)	0.94
F	103 (31.4)	225 (68.6)	397 (29.2)	964 (70.8)	1.11 (0.86 to 1.44)	0.43
Obesity, n (%)							0.02b
No	139 (20.2)	548 (79.8)	662 (23.0)	2223 (77.1)	0.85 (0.69 to 1.05)	0.13
Yes	87 (42.4)	118 (57.6)	183 (35.4)	334 (64.6)	1.35 (0.97 to 1.87)	0.08
Diabetes, n (%)							0.30
No	170 (23.0)	569 (77.0)	693 (23.6)	2238 (76.4)	0.97 (0.80 to 1.17)	0.72
Yes	56 (36.6)	97 (63.4)	152 (32.3)	319 (67.7)	1.21 (0.83 to 1.77)	0.32
Hypertension, n (%)							0.59
No	157 (22.9)	529 (77.1)	599 (22.9)	2015 (77.1)	1.00 (0.82 to 1.22)	0.99
Yes	69 (33.5)	137 (66.5)	246 (31.2)	542 (68.8)	1.11 (0.80 to 1.54)	0.53
CKD, n (%)							0.79
No	211 (24.8)	639 (75.2)	796 (24.3)	2475 (75.7)	1.03 (0.86 to 1.22)	0.77
Yes	15 (35.7)	27 (64.3)	49 (37.4)	82 (62.6)	0.93 (0.45 to 1.92)	0.84

P-values derived from unconditional logistic regression because matching is broken with subset cohorts. OR, odds ratio; CI, confidence interval.

^aAntibiotic exposure defined as having been prescribed an antibiotic from 1 day to 1 year before the index date.

^bP-values denote statistical significance at the 0.05 α level.

Finally, in analysis that further excluded cases and controls with urinary symptoms in the previous 90 days identified by manual medical record review, there were 847 stone formers and 3289 controls, of which 60 (7%) and 271 (8%) had antibiotic prescriptions in the previous 90 days, respectively. No significant associations between the risk of kidney stones and antibiotic use were observed, with or without multivariable adjustment (Supplemental Table 6).

Discussion

The findings of this study suggest that oral antibiotics do not contribute to a first-time symptomatic kidney stone. When using medical record validation to identify and accurately determine the timing of first-time symptomatic stone episodes, the unadjusted risk of stones with antibiotics was modest and only higher within 1 year after antibiotic use. These first-time symptomatic stone formers were more likely to have obesity, diabetes, hypertension, and CKD compared with controls,^22–27 and these comorbidities seemed to confound the association between kidney stones and antibiotic use. There was also compelling evidence of reverse causality. When the study sample was limited to antibiotic use for nonurinary symptoms, there was no evidence of an increased risk of symptomatic kidney stone with antibiotic use. Overall, these findings suggest that much of the observed risk of kidney stones attributed to antibiotics is better explained by kidney stones as an under-recognized cause or cocondition for urinary symptoms attributed to UTIs that lead to oral antibiotic prescriptions.

In a subgroup analysis, we found an increased risk of a first-time symptomatic kidney stone with the use of several antibiotic classes, but associations were diminished and not significant after excluding patients with urinary symptoms. Indeed, three of the four antibiotic classes that were associated with kidney stone are either commonly or exclusively used to treat UTIs: sulfas, fluoroquinolones, and nitrofurantoin/methenamine/fosfomycin. We did not observe that the risk of kidney stone with antibiotics depended on their sensitivity to Oxalobacter formigenes, an organism that metabolizes oxalate and is sensitive to macrolides, tetracycline, metronidazole, and fluoroquinolone.²⁸ In contrast to our findings, a prior population-based matched case-control study demonstrated a higher risk of kidney stones up to at least 5 years after the use of sulfas, cephalosporins, fluoroquinolones, nitrofurantoin/methenamine, and broad-spectrum penicillins.¹⁴ However, this study used diagnosis codes to identify kidney stones and lacked medical record review to confirm and then validate the timing of first-time symptomatic stone episodes. Kidney stones frequently present with UTI-mimicking symptoms and are thus often initially misdiagnosed as a UTI before stones are later discovered or co-occur with an actual or presumed UTI, leading to antibiotic prescriptions. Importantly, kidney stones themselves increase the risk of UTI while UTI can also cause stones.²⁹ Therefore, the observed higher risk of kidney stones associated with antibiotic use, particularly for urinary symptoms, might instead reflect reverse causality (i.e., stone symptoms as the cause of antibiotic prescriptions). Even with medical record review, available information can often be inadequate to adjudicate whether urinary symptoms were due to an unrecognized stone, a UTI, or both, because of insufficient clinical evaluation (i.e., no imaging or urine cultures at the time of antibiotic prescription for urinary symptoms).

The risk of a first-time symptomatic kidney stone was highest right after antibiotic use (Figure 2). This observation potentially runs counter to the hypothesis that antibiotics increase stone formation by altering urine metabolite excretion through perturbation of the gastrointestinal microbiome^6,7 because this process would take weeks to months for stones to form and grow before their symptomatic passage.^20,30 Because the cause of urinary symptoms can be uncertain, the only way to ensure reverse causality does not affect the risk of kidney stones with antibiotics is to exclude patients with prior urinary symptoms, including UTIs, from the study. In this study, after doing so, the increased risk of a first-time symptomatic kidney stone in the early period after antibiotic use was no longer evident, and no antibiotic class was associated with an increased risk of kidney stones. Although a prior study demonstrated that certain antibiotic classes remained risk factors of kidney stones after excluding participants with any prior UTI by diagnosis codes,¹⁴ one of these high-risk antibiotic classes for kidney stones was nitrofurantoin, which is exclusively prescribed for a UTI. Thus, it is likely that UTIs were not adequately excluded from the analyses using diagnosis codes. Misclassification of urinary symptoms from a stone as a UTI may also have biased a survey-based study, which found an increased risk of kidney stones with antibiotic use.¹⁵ Specifically, participants were excluded only if UTI was the main reason (rather than ever a reason) for prior antibiotic use. Our study thoroughly excluded all persons with UTIs or urinary symptoms by using a comprehensive set of diagnosis codes for UTI and hematuria, by identifying antibiotics in which indication was exclusively for UTI, and by manual review of medical records for urinary symptoms. Despite a smaller sample size, the upper limit of 95% CI in our study remained lower than the risk estimates reported in the prior study.¹⁴ Thus, differences in statistical power alone would not explain differences in these data compared with prior work. Certainly, a small risk of stones with antibiotics may be present with a much larger sample size, but it likely lacks clinical significance given the upper limit of the 95% CI was 1.17 for the adjusted OR among those without prior UTI symptoms during the 1-year period after antibiotic use. While antibiotic stewardship is encouraged because of several potential harms of antibiotics, our findings can provide additional insight into antibiotic use, particularly in circumstances when the risk and benefit are not clear. For example, some providers might be reluctant to give antibiotic prophylaxis in at-risk patients who undergo urologic procedures,³¹ including when kidney stone is the indication, because of concern for kidney stone risk. Therefore, it is reassuring that antibiotics given before the procedure or periprocedure are unlikely to lead to more future stone formation.

Our study found that the risk of a first-time symptomatic kidney stone in participants with antibiotic exposure may be higher in obese participants, compared with nonobese participants. The synergistic effect between antibiotic use and obesity on kidney stone risk might be explained by reports that antibiotic exposure significantly increases the risk of weight gain and obesity^32,33 and weight gain and obesity significantly increase the risk of kidney stone formation.^23,34 However, this would suggest that efforts to treat obesity rather than limit oral antibiotic prescriptions may be more impactful in preventing kidney stone disease in obese patients.

Our study also had several potential limitations. Our study sample consisted of mostly White adults in one county of Southeast Minnesota, and findings from our study might not be generalizable to children, other ethnic groups, or other geographic locations. The exact date of stone formation cannot be ascertained in symptomatic stone formers with the current existing technology, and this inherently limits the scientific study of the timing between risk factors and kidney stone disease. Regardless, if antibiotics do not increase the risk of symptomatic stones, then the role of antibiotics in subclinical stone formation and growth is of less clinical importance. Our study did not capture either inpatient antibiotics or outpatient intravenous antibiotics. It is possible that more potent intravenous antibiotics would have a more pronounced effect on the microbiome, and thus possibly stone risk. However, most antibiotic prescriptions are in outpatient settings and not intravenous.³⁵ Data on cumulative duration and dose of antibiotics were not available in our study. Some antibiotic prescriptions might not actually have been taken or completed, and this could bias the results toward the null. Despite matching and statistical adjustment, residual confounders might remain in our analyses. Our study did not account for differences in all comorbid conditions, medication usage, health care encounters, or dietary intake between stone formers and controls. However, antibiotic users tended to have more risk factors of kidney stones and more health care encounters. Therefore, an increased risk of kidney stones with antibiotic use would likely remain not evident with adjustment for additional factors.

In conclusion, the association between antibiotic use and the risk of a first-time symptomatic kidney stone is complicated because nonspecific urinary symptoms and urinalysis finding can be due to unrecognized kidney stones and lead to antibiotic use for a perceived UTI. In this study, after adjusting for comorbidities and in the absence of urinary symptoms, there was no increased risk of symptomatic kidney stones with antibiotic use. These data suggest that antibiotics are not an important risk factor of kidney stones.

Supplementary Material

jasn-34-1399-s001.pdf ^{(246.7KB, pdf)}

Disclosures

E.F. Barreto reports Consultancy: Wolters Kluwer; Research Funding: AHRQ and NIAID; and Honoraria: Vifor Pharma. K. Koo reports Other Interests or Relationships: UpToDate. J.C. Lieske reports Consultancy: Allena, Alnylam, American Board of Internal Medicine, BioMarin, Chinook, Dicerna, Federation Bio, Intellia, Mirium, Novobiome, NovoNordisc, Orfan, Oxidien, OxThera, Siemens, and Synlogic; Research Funding: Allena, Alnylam, Dicerna, Novobiome, OxThera, Retrophin, Siemens, and Synlogic; Honoraria: American Board of Internal Medicine and Up to Date; and Advisory or Leadership Role: ABIM and Kidney International. A.D. Rule reports Patents or Royalties: UpToDate; and Advisory or Leadership Role: JASN—Associate Editor, Mayo Clinic Proceedings—Section Editor, and NIDDK—Urological Diseases of America Contract Management Board. All remaining authors have nothing to disclose.

Funding

This work was partially supported by grants from the National Institute of Diabetes and Digestive and Kidney Diseases (Mayo Clinic O'Brien Urology Research Center, DK100227 and DK83007) and made possible by the Rochester Epidemiology Project (AG034676) from the National Institutes of Health, US Public Health Service, and the CTSA Grant UL1 TR002377 from the National Center for Advancing Translational Sciences, and National Institute of Allergy and Infectious Diseases of the National Institutes of Health under Award Number K23AI143882 (PI; EFB). The funding sources had no role in study design; data collection, analysis, or interpretation; writing the report; or the decision to submit the report for publication. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.

Author Contributions

Conceptualization: Andrew D. Rule, Charat Thongprayoon.

Data curation: Ramila A. Mehta.

Formal analysis: Lisa E. Vaughan.

Funding acquisition: Andrew D. Rule.

Investigation: Erin F. Barreto, Kevin Koo, John C. Lieske, Andrew D. Rule, Phillip J. Schulte, Charat Thongprayoon.

Methodology: Andrew D. Rule, Charat Thongprayoon.

Supervision: Andrew D. Rule.

Visualization: Lisa E. Vaughan.

Writing – original draft: Charat Thongprayoon.

Writing – review & editing: Erin F. Barreto, Kevin Koo, John C. Lieske, Ramila A. Mehta, Andrew D. Rule, Phillip J. Schulte, Lisa E. Vaughan.

Data Sharing Statement

All data are included in the manuscript and/or supporting information.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/JSN/E432.

Supplemental Table 1. Diagnosis codes for comorbidities.

Supplemental Table 2. Association of a first-time symptomatic kidney stone with different periods of antibiotic use by antibiotic classes (N=5,271).

Supplemental Table 3. Association of a first-time symptomatic kidney stone with the number of antibiotic courses within 5 years before the index date among stone formers and matched controls with no prior urinary symptoms within 5 years (N=4,294).

Supplemental Table 4. Association of a first-time symptomatic kidney stone with different periods of antibiotic use among calcium-based stone formers and matched controls with no prior urinary symptoms within 5 years (N=1,423).

Supplemental Table 5. Association of a first-time symptomatic kidney stone with different time periods of antibiotic use among calcium oxalate stone formers and matched controls with no prior urinary symptoms within 5 years (N=1,207).

Supplemental Table 6. Association of a first-time symptomatic kidney stone with different periods within 90 days of antibiotic use among stone formers and matched controls with no prior urinary symptoms within 5 years, after additionally excluding participants who were found to have received antibiotics for urinary symptoms within 90 days before the index date on the basis of manual review (N=4,236).