Risk factors for infection after prostate biopsy in the United States

Abstract

Objectives

To characterize trends in the incidence and severity of post-biopsy infection using nationally representative data, given that population-based data have revealed an increasing incidence of post-prostate biopsy infections through 2007. Little is known about contemporary trends in prostate biopsy related infections and associated risk-factors, particularly considering the increased focus on preventative strategies.

Methods

We used the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked data from 2001 to 2015 and recorded 30-day infection rate and emergency department, hospital and, ICU admissions. We performed an adjusted analysis to account for the effect of age, demographic data, surgeon volume, rectal swab use, and biopsy year.

Results

We found that the overall rate of post-biopsy infections increased from 2001 (5.9%) to 2007 (7.2%) but remained stable through 2015. Despite this, post-biopsy emergency room visits rose from 0.2% to 0.5% (95% CI 0.2%−0.4%, p<0.01), hospitalizations rose from 0.5% to 1.3% (95% CI 0.5%−1.0%, p< 0.01), and intensive care unit admissions increased from 0.1% to 0.3% (95% CI 0.1%−0.3%, p<0.01). Surgeons who performed 25 biopsies per year had lower risk of post-biopsy infection (OR: 0.65; 95% CI: 0.61–0.69) and a lower risk of hospitalization (OR: 0.50; 95% CI: 0.43–0.59) compared to surgeons who performed one biopsy per year. Rectal swab use increased but remained low (1.8% in 2015).

Conclusions

While the overall rate of post-biopsy infections has stabilized since 2007, admissions to the emergency room, hospital, and ICU continue to rise. Increased surgeon volume was associated with a decreased risk of infection.

Introduction

Prostate biopsy is a common procedure that many men in the United States endure during their lifetimes.¹ The United States Preventive Services Task Force (USPSTF) Grade C recommendation for PSA screening considers pain and adverse events associated with biopsy to be among the harms of PSA-based screening.² These adverse events include infectious complications such as cystitis, prostatitis, epididymitis, orchitis, bacteremia, sepsis, and hospitalization.^{3, 4} In rare cases, prostate biopsy can also lead to limb gangrene/amputation, endocarditis, meningitis, disseminated intravascular coagulation (DIC), or death.^5–10

Prior population-based studies from over a decade ago found that the risk of post-biopsy infections has been increasing.^{11, 12} This trend has coincided with a concomitant increase in antimicrobial resistance.^13–17 Established risk factors for post-biopsy infection include non-white race, comorbid conditions including diabetes, immunocompromise, prostate enlargement or inflammation, and recent hospitalization or travel.^{12, 18–21} In order to reduce the rate of infections, the AUA guidelines discussed the use of employing augmented prophylaxis or targeted prophylaxis using rectal swab cultures obtained prior to prostate biopsy in 2012 and then encouraged utilization of a rectal swab in higher risk patients in 2016.³ Here, we aimed to characterize recent trends in post-biopsy infections, to assess the adoption of targeted antimicrobial prophylaxis and to determine risk factors for post-biopsy sepsis in the Surveillance, Epidemiology and End Results (SEER)-Medicare cohort.

Methods

We analyzed (SEER)-Medicare data to identify male Medicare beneficiaries aged over 65 years who underwent prostate biopsy between 2001 and 2015. The primary outcome was the incidence of infection within 30 days of prostate biopsy. Secondary outcomes included 30-day infections requiring admission to the emergency department (ED) and subsequent discharge, 30-day hospitalization for infection, and 30-day intensive care unit (ICU) admission. MRI-guided in-bore biopsies were excluded due to small number (<1%). Average annual surgeon volume was assessed by stratifying based on the Unique Provider Identification Number (UPIN). Baseline characteristics were summarized for the entire cohort. Mean and standard deviations (SD) are presented for age, surgeon volume, and biopsy year. Numbers and percentages are presented for categorical variables. The differences in the risk of any infection and the risk of hospitalization between subgroups stratified by categorical baseline characteristic were assessed using the Chi-square test. The unadjusted nonlinear relationship between infection risk and three continuous covariates (age, biopsy year and surgeon volume) were assessed using a restricted cubic splines model. Models with 3 knots were chosen based on the Akaike information criterion (AIC). The nonlinear relationship was observed in surgeon volume for both any infection and infection requiring admission outcomes, and was also observed in biopsy year for any infection (Figure 2). The restricted cubic spline of surgeon volume and biopsy year were added into multivariable logistic regression for the corresponding outcomes. Other covariates in the model included age, race, registry region for prostate cancer, biopsy type, Charlson comorbidity index and prior rectal swab. Adjusted odds ratios and 95% confidence intervals were calculated for each of the covariates.

Figure 2:

Adjusted odds of infection by surgeon volume (A) and biopsy year (B)

Results

Figure 1 shows that the risk of any post-biopsy infections increased from 5.9% in 2001 to 7.2% in 2007 (95% CI 0.8%– 1.8%, p<0.01) and then stabilized through 2015 at 7.1%. Between 2001 and 2015, post-biopsy emergency room visits rose from 0.2% to 0.5% (95% CI 0.2%−0.4%, p<0.01), hospitalizations rose from 0.5% to 1.3% (95% CI 0.5%−1.0%, p<0.01), and intensive care unit admissions increased from 0.1% to 0.3% (95% CI 0.1%−0.3%, p<0.01). Age at biopsy and surgeon volume predicted infection and hospitalization risk whereas biopsy year predicted hospitalization risk (Supplemental Figure 1). This trend persisted in adjusted analyses; ORs of any post-biopsy infection compared to 2001 were first increased until to 2007 and later barely changed (Figure 3B). The OR of hospitalization per each year increase was 1.07 (95% CI 1.06–1.09, p<0.01, Table 2). The use of rectal swabs for targeted prophylaxis remained low, although it increased significantly over the time period from 0.2% to 1.8% (95% CI 1.4%−1.9%, p<0.01, Figure 1).

Figure 1:

Trend in A.Prostate Biopsies and overall complications within thirty days, and B. Trends in hospital, ICU, and ED visits, as well as the use of pre-biopsy rectal swabs.

Table 2:

Unadjusted and adjusted association between patient characteristics and any infections, or hospitalization for infection, within 30 days of biopsy.

	Any Infection				Hospitalization for Infection
	Infectious Complications	Unadjusted p-value	Adjusted OR (95% CI)	Adjusted p-value	Hospitalizations	Unadjusted p-value	Adjusted OR (95% CI)	p-value
Age at biopsy	-	-	1.01(1.01–1.02)	<0.01	-	-	1.04(1.03–1.05)	<0.01
Biopsy Year	-	-	-	-	-	-	1.07(1.06–1.09)	<0.01
Race		<0.01				<0.01
White	15141(6.6%)		Ref		2162(0.9%)		Ref
Black	2545(8.8%)		1.27(1.20–1.34)	<0.01	410(1.4%)		1.47(1.28–1.69)	<0.01
Hispanic	559(11.7%)		1.50(1.34–1.68)	<0.01	92(1.9%)		1.93(1.46–2.54)	<0.01
Other	1017(7.9%)		1.06(0.97–1.16)	0.19	145(1.1%)		1.17(0.91–1.50)	0.22
SEER Region		<0.01				<0.01
California	6706(8.1%)		Ref		826(1.0%)		Ref
Connecticut	772(5.4%)		0.60(0.54–0.66)	<0.01	121(0.8%)		0.81(0.63–1.05)	0.11
Georgia	2074(6.7%)		0.80(0.75–0.85)	<0.01	289(0.9%)		1.01(0.85–1.21)	0.87
Hawaii	254(7.4%)		0.83(0.70–0.97)	0.02	27(0.8%)		0.76(0.46–1.23)	0.26
Iowa	729(5.1%)		0.61(0.55–0.68)	<0.01	125(0.9%)		1.09(0.85–1.39)	0.51
Kentucky	908(5.7%)		0.69(0.63–0.75)	<0.01	178(1.1%)		1.19(0.95–1.48)	0.13
Louisiana	1182(6.6%)		0.79(0.73–0.86)	<0.01	220(1.2%)		1.38(1.13–1.69)	<0.01
Michigan	2180(10.2%)		1.28(1.20–1.36)	<0.01	316(1.5%)		1.44(1.22–1.71)	<0.01
New Jersey	2844(6.5%)		0.78(0.74–0.83)	<0.01	481(1.1%)		1.08(0.93–1.26)	0.30
New Mexico	449(7.5%)		0.97(0.86–1.09)	0.57	63(1.1%)		1.27(0.91–1.76)	0.15
Utah	469(5.6%)		0.71(0.63–0.80)	<0.01	58(0.7%)		0.86(0.61–1.22)	0.40
Washington	697(4.3%)		0.51(0.46–0.56)	<0.01	105(0.6%)		0.78(0.59–1.01)	0.06
Biopsy Type		0.84				0.76
TRUS	19165(7.0%)		Ref		2796(1.0%)		Ref
Transperineal	99(7.2%)		0.94(0.63–1.41)	0.77	13(0.9%)		0.41(0.10–1.65)	0.21
Charlson’s Score		<0.01				<0.01
0	1803(5.8%)		Ref		179(0.6%)		Ref
1 or 2	6354(5.9%)		0.95(0.89–1.01)	0.10	688(0.6%)		1.00(0.81–1.23)	0.98
3+	11107(8.2%)		1.25(1.18–1.33)	<0.01	1942(1.4%)		2.18(1.80–2.64)	<0.01
Prior Rectal Swab		<0.01				0.21
No	19139(7.0%)		Ref		2791(1.0%)		Ref
Yes	125(9.5%)		1.47(1.12–1.92)	<0.01	18(1.4%)		2.06(1.13–3.76)	0.02

After adjusting for baseline characteristics (summarized in Table 1), older age was associated with higher risk of any infection as well as hospitalization (Table 2). However, the average age at the time of biopsy decreased over time (74.3 in 2001 and 73.4 in 2015), suggesting that the average age of the cohort does not explain the rise in severe infections. Black and Hispanic men had an increased risk of infection compared to White men. Additionally, men with a Charlson score of 3 or greater had a higher risk of any infection or hospitalization compared to those with Charlson score of 1.

Table 1:

Characteristics of study population.

	# of Biopsies (% of total in column)
Total	274858
Mean Age at Biopsy (std)	74.0 (5.6)
Mean Surgeon Volume (std)	22.3 (31.7)
Mean Biopsy Year (std)	2007 (4)
Race
White	228237(83.1%)
Black	28830(10.5%)
Hispanic	4801(1.7%)
Other	12946(4.7%)
SEER Region
California	82413(30.0%)
Connecticut	14254(5.2%)
Georgia	31048(11.3%)
Hawaii	3429(1.2%)
Iowa	14180(5.2%)
Kentucky	15895(5.8%)
Louisiana	17946(6.5%)
Michigan	21413(7.8%)
New Jersey	43587(15.9%)
New Mexico	5962(2.2%)
Utah	8322(3.0%)
Washington	16409(6.0%)
Biopsy Type
TRUS	273472(99.5%)
Transperineal	1386(0.5%)
Charlson’s Score
0	31054(11.3%)
1 or 2	107414(39.1%)
3+	136390(49.6%)
Prior Rectal Swab
No	273546(99.5%)
Yes	1312(0.5%)
Surgery Center*
Yes	16290(8.6%)
No	173375(91.4%)

^*31.0% missing value for surgeon volume and surgery center

Higher surgeon volume was associated with lower odds of infection and hospitalization in a cubic spline model and on multivariable analysis, while increasing age and year of biopsy were associated with an increase in infection risk (Table 2 and Figure 2). Figure 2b demonstrates an inflection point in the risk of infection and hospitalization due to infection relative to surgeon volume around 25 biopsies per year. Compared to patients treated by surgeons who performed 1 biopsy annually, patients treated by surgeons who annually performed 25 biopsies or more had a lower odds of post-biopsy infection (OR: 0.65; 95% CI: 0.61–0.69) and hospitalization (OR: 0.50; 95% CI: 0.43–0.59). In subanalysis examining only higher volume (>25 biopsies per year) surgeons, the relationship between volume and infections persisted, with the odds of infection or hospitalization continuing to decrease with increasing surgeon volume (Supplemental Figure 2).Compared to patients treated by surgeons who annually performed 1 biopsy, patients treated by surgeons who annually performed 25 biopsies or more had a lower risk of post-biopsy infection (OR: 0.65; 95% CI: 0.61–0.69) and a lower risk of hospitalization (OR: 0.50; 95% CI: 0.43–0.59).

Discussion

Loeb et al. previously demonstrated an increase of post-biopsy infections using SEER-Medicare data between 1991 and 2007.¹² We used current, nationally representative, population-based data to demonstrate that the rate of post-prostate biopsy infections subsequently stabilized, however the rate of hospitalizations, ED, and ICU admissions continues to increase annually even when controlling for surgeon volume and patient race, age, region, and CCI. This may be related to increasing antimicrobial resistance, resulting in increased severity of infections when they do occur. This represents a significant public health concern, particularly in light of the regulatory emphasis placed on reducing health-care associated infections. Although it is beyond the scope of our study design to determine causation for this sustained increase in the risk of post-biopsy infections, our findings demonstrate a need to identify strategies to effectively combat severe infectious complications.

Overall, the rate of infection (7.1% in 2015) and hospitalization due to infections (1.3% in 2015) we report is similar to the expected range published in the AUA guidelines (5–7% and 1.3% respectively). Our more inclusive definition of infectious complications and reliance on use of billing codes perhaps contributed to our slightly higher estimates. While some series have reported techniques to decrease infection rates, our results reflect nationwide practices.

Evidence based strategies for reducing infection rate after prostate biopsy include augmented prophylaxis (using multiple antibiotics in opposition to principles of antimicrobial stewardship) and targeted prophylaxis (using rectal swabs to identify men at risk of fluoroquinolone resistance). The American Urological Association (AUA) currently recommends culturing the rectal flora from high risk patients for antibiotic resistant bacteria prior to biopsy to direct antibiotic choices.³ Several studies have found this method to be effective at reducing infections and this has led to increased utilization in the clinic.²² However, other studies demonstrate mixed results.²³ Regardless, our data show that use of pre-procedural rectal swabs remains low overall with a utilization rate of 1.8% in 2015. Given the recommendation to consider a rectal swab in men at higher risk of infection in the 2016 AUA guidelines, this rate may increase. Regardless, the low utilization in this cohort is unlikely to have affected the overall trend of increased infection rates and severity of infections that is evident in this cohort.

Consistent with other reports, our study finds comorbid conditions and non-White race to be independent risk factors for post-biopsy infection. Risk factors for antimicrobial resistance in general include both the use and misuse of antimicrobial drugs, travel, migration, and population density as well as provider prescribing practices.^{24, 25} A large study of men in the Kaiser system undergoing prostate biopsy found that Latino men had higher rates of antibiotic resistance compared to White men.²⁶ While any of these confounding factors may explain this result, without data regarding antibiotic prophylaxis prescriptions and culture data, we cannot offer firm conclusions about why this is occurring independent of age, region, and CCI.

Notably, our study also finds higher surgeon volume to be strongly associated with a lower risk of infection. Providers who had performed 1 biopsy per year had a significantly higher infection and hospitalization rate than providers who performed 25 or more biopsies per year. Our cubic spline model adjusted for age and year, demonstrated decreased odds of infection and hospitalization as surgeon volume increased with an inflection point around 25 biopsies per year. However even among only higher volume surgeons (>25 biopsies per year), this relationship persisted. In the absence of an unforeseen confounding effect, our data suggest there is a component of surgeon volume contributing to infectious risks. While the underlying reason for this association is unknown, technical skill, familiarity with prophylaxis protocols, office workflow, and post biopsy management are hypothetical contributing factors that deserve further exploration.

Our findings must be interpreted within the context of the study design. First, we are not able to identify antibiotic prophylaxis patterns of care. Second, our use of Medicare data only allows us to assess a proportion of prostate biopsy volume by surgeons. However, it is very likely that biopsy volume in the elderly is strongly correlated to prostate biopsy volume irrespective of age. Third, our use of Medicare claims data may underestimate the true incidence of infectious complications that do not receive a diagnosis code, such as minor or self-limited prostatitis, cystitis and epididymitis. Furthermore, we also attempted to be inclusive of our definition of infectious complications, which could capture events unrelated to the biopsy itself. SEER-Medicare does not allow us to confirm infectious diagnoses with culture data which may overestimate the true incidence of infection after prostate biopsy.

In summary, higher surgeon volume is associated with a lower risk of post-biopsy infections, and while the overall rate of infections has stabilized, hospitalizations, ICU, and emergency room visits continue to rise independently. Moreover, there has been limited adoption of targeted prophylaxis. Taken together, our findings on prostate biopsy patterns of care inform physicians, policy makers, and payers in terms of formulating incentives or policies to encourage the uptake of strategies to combat the increasing incidence of post-biopsy infectious complications.

Supplementary Material

1

Supplemental Figure 1: Trends in the risk of any infection (top row), or hospitalization for infection (bottom row) according to surgeon biopsy volume (A), patient age (B) and biopsy year (C).

2

Supplemental Figure 2: Adjusted Odds Ratio of Infection and Hospitalization Compared to Surgeon volume Excluding Low Volume Surgeons