The first 2 authors contributed equally to this article.
Improvement of racial disparities with respect to the utilization of minimally invasive radical prostatectomy in the United States†
Version of Record online: 6 SEP 2011
Copyright © 2011 American Cancer Society
Volume 118, Issue 7, pages 1894–1900, 1 April 2012
How to Cite
Trinh, Q.-D., Schmitges, J., Sun, M., Sukumar, S., Sammon, J., Shariat, S. F., Jeldres, C., Bianchi, M., Tian, Z., Perrotte, P., Rogers, C. G., Graefen, M., Peabody, J. O., Menon, M. and Karakiewicz, P. I. (2012), Improvement of racial disparities with respect to the utilization of minimally invasive radical prostatectomy in the United States. Cancer, 118: 1894–1900. doi: 10.1002/cncr.26527
- Issue online: 19 MAR 2012
- Version of Record online: 6 SEP 2011
- Manuscript Accepted: 21 JUL 2011
- Manuscript Revised: 20 JUL 2011
- Manuscript Received: 2 MAR 2011
- prostatic neoplasms;
- minimally invasive;
- racial disparities
Race represents an established barrier to health care access in the United States and elsewhere. We examined whether race affects the utilization rate of minimally invasive radical prostatectomy (MIRP) in a population-based sample of individuals from the United States.
Within the Healthcare Cost and Utilization Project Nationwide Inpatient Sample (NIS), we focused on patients in whom MIRP and open radical prostatectomy (ORP) were performed between 2001 and 2007. We assessed the proportions and temporal trends in race distributions between MIRP and ORP. Multivariable logistic regression analyses further adjusted for age, year of surgery, baseline Charlson Comorbidity Index, annual hospital caseload tertiles, hospital region, insurance status, and median zip code income.
Of 65,148 radical prostatectomies, 3581 (5.5%) were MIRPs. African Americans accounted for 11.4% of patients versus 78.8% for Caucasians versus 9.9% for others. Between 2001 and 2007, the annual proportions of Caucasian patients treated with MIRP were 2.2%, 0.9%, 2.6%, 7.2%, 4.7%, 9.3%, and 11.6%, respectively (chi-square trend p<0.001). For the same years in African American patients, the proportions were 0.8, 0.3, 1.4, 4.4, 3.5, 9.0 and 8.4% (chi-square trend P < .001). In multivariable analyses relative to Caucasian patients, African American patients were 14% less likely to undergo MIRP (P = .01). After period stratification between years 2001-2005 versus 2006-2007, African Americans were 22% less likely to undergo a MIRP in the early period (P = .007) versus 11% less likely to have a MIRP in the contemporary period (P = .1).
The racial discrepancies in MIRP utilization rates are gradually improving. Cancer 2012;. © 2011 American Cancer Society.
Race represents an established barrier to health care access in the United States.1 Many authors have reported differences in cancer-related deaths among certain racial backgrounds, namely African American, when compared with their Caucasian counterparts.2, 3 Although higher mortality may be secondary to biological predisposition, as demonstrated by higher tumor grade and stage at presentation, disparities in the use of definitive therapy may actually explain the difference in mortality.4-6 These observed racial disparities have been attributed to lack of public awareness and medical access in the African American community. Interestingly, similar trends were recorded in the Hispanic population,7 which further supports the hypothesis that race occupies a central role in prostate cancer management.
Minimally invasive radical prostatectomy (MIRP), which includes laparoscopic and robot-assisted radical prostatectomy, has become an increasingly popular definitive treatment modality for prostate cancer management and clinical decision making due to less morbidity and faster recovery.8 Despite these important advantages, MIRP may not be offered to some racial groups at the same rate as to others. For example, African Americans may be less likely to undergo this potentially less morbid surgical procedure.
In this study, we examined the effect of race on utilization rates of MIRP. Specifically, we compared utilization rates among African American and Caucasian patients. Our hypothesis was that African American patients were less likely to be treated with MIRP than their Caucasian counterparts.
MATERIALS AND METHODS
We relied on the Nationwide Inpatient Sample (NIS) discharge records. The NIS is a part of the Healthcare Cost and Utilization Project, sponsored by the Agency for Healthcare Research and Quality, and is considered the largest all-payer inpatient care database in the United States. It contains data from approximately 8 million hospital stays each year, approximating a 20% stratified sample of community hospitals from participating states, including academic and specialty hospitals within the United States.9 Relying on discharge records, all patients with a primary diagnosis of prostate cancer (ICD-9-CM code 185) were considered for the study. The prostatectomy procedure code (ICD-9-CM 60.5) resulted in the identification of 89,970 patients who underwent radical prostatectomy between 2001 and 2007. A total of 24,822 men were excluded due to missing racial classification, leaving 65,148 patients for analyses. Of these patients, 3581 were further classified as having undergone MIRP using the minimally invasive modifier code (ICD-9-CM 54.21). The remaining 61,567 patients were considered to have undergone ORP.
For all patients, the following variables were available: race, age, year of surgery, Charlson Comorbidity Index (CCI), annual hospital caseload (AHC) tertiles, hospital region, insurance status, and median zip code income. Information about hospital region was obtained from the American Hospital Association Annual Survey of Hospitals, and defined by the United States Census Bureau.10 This is comprised of the following: South (Delaware, Maryland, District of Columbia, Virginia, Carolina, Georgia, Florida, Kentucky, Tennessee, Alabama, Mississippi, Arkansas, Louisiana, Oklahoma, Texas), West (Montana, Idaho, Wyoming, Colorado, New Mexico, Arizona, Utah, Nevada, Washington, Oregon, California, Alaska, Hawaii), Northeast (Maine, New Hampshire, Massachusetts, Rhode Island, Connecticut, New York, New Jersey, Pennsylvania), and Midwest (Ohio, Indiana, Illinois, Michigan, Wisconsin, Minnesota, Iowa, Missouri, Dakota, Nebraska, Kansas). CCI was derived from ICD-9 codes according to previously established criteria11 and was stratified according to 2 levels: 0 versus ≥1. AHC was defined according to the number of procedures performed at each participating institution during each study calendar year. Hospitals were divided into caseload tertiles, including low (≤34), intermediate (35-90), and high (≥91) open radical prostatectomies (ORPs). Race is self-reported by the patient at diagnosis, and defined as Caucasian, African American, Hispanic, Asian or Pacific Islander, Native American, and other. Race ascertainment in the NIS is considered highly accurate and representative of the United States. For purpose of analyses, race was categorized as Caucasian, African American, and other in the current study. Median annual household income in the patient's ZIP code of residence adjusted for inflation based on projections of census data was reported in the NIS in 4 quartiles. To ensure uniformity of coding across data sources, detailed insurance categories are combined into general groups, namely private insurance, Medicare, Medicaid, and self-pay. In particular, private insurance includes Blue Cross, commercial carriers, private health maintenance organizations (HMO), and preferred provider organizations (PPO). Medicare includes both fee-for-service and managed care Medicare patients; Medicaid includes both fee-for-service and managed care Medicaid patients.
|Overall No. (%)||Caucasian||African American||Other|
|MIRP No. (%)||ORP No. (%)||P||MIRP No. (%)||ORP No. (%)||P||MIRP No. (%)||ORP No. (%)||P|
|65148 (100)||2921 (5.7)||48400 (94.3)||308 (4.2)||7100 (95.8)||352 (5.5)||6067 (94.5)|
|Year of surgery||<.001||<.001|
|2001||9135 (14.5)||157 (2.2)||7138 (97.8)||8 (0.8)||1011 (99.2)||29 (3.5)||792 (96.5)|
|2002||9665 (14.8)||67 (0.9)||7518 (99.1)||3 (0.3)||1104 (99.7)||5 (0.5)||968 (99.5)|
|2003||8780 (13.5)||180 (2.6)||6696 (97.4)||14 (1.4)||1010 (98.6)||25 (2.8)||855 (97.2)|
|2004||8326 (12.8)||478 (7.2)||6145 (92.8)||44 (4.4)||959 (95.6)||31 (4.4)||669 (95.6)|
|2005||8231 (12.6)||308 (4.7)||6210 (95.3)||30 (3.5)||823 (96.5)||62 (7.2)||798 (92.8)|
|2006||9767 (15.0)||709 (9.3)||6896 (90.7)||101 (9.0)||1016 (91.0)||75 (7.2)||970 (92.8)|
|2007||11,244 (17.3)||1022 (11.6)||7797 (88.4)||108 (8.4)||1177 (91.6)||125 (11.0)||1015 (89.0)|
|Mean (median)||61.1 (62.0)||61.0 (61.0)||61.4 (62)||58.3 (58.5)||59.0 (59.0)||60.6 (61.0)||61.0 (62.0)|
|0||51,973 (79.8)||2464 (5.9)||39,052 (94.1)||241 (4.4)||5232 (95.6)||283 (5.7)||4701 (94.3)|
|≥1||13,175 (20.2)||457 (4.7)||9348 (95.3)||67 (3.5)||1868 (96.5)||69 (4.8)||1366 (95.2)|
|Annual hospital caseload||<.001||<.001||<.001|
|Low (1-34)||21,961 (33.7)||94 (0.6)||16,693 (99.4)||22 (0.8)||2756 (99.2)||19 (0.8)||2377 (99.2)|
|Intermediate (35-90)||21,575 (32.2)||500 (2.9)||16,593 (97.1)||70 (3.0)||2299 (97.0)||81 (3.8)||2032 (96.2)|
|High (>90)||21,612 (34.1)||2327 (13.3)||15,114 (86.7)||216 (9.6)||2045 (90.4)||252 (13.2)||1658 (86.8)|
|Northeast||14,997 (23.0)||800 (6.7)||11,087 (93.3)||121 (7.4)||1506 (92.6)||142 (9.6)||1341 (90.4)|
|Midwest||10,637 (16.3)||289 (3.1)||9100 (96.9)||24 (2.6)||882 (97.4)||12 (3.5)||330 (96.5)|
|South||25,647 (39.4)||930 (4.8)||18,390 (95.2)||130 (3.2)||3923 (96.8)||89 (3.9)||2185 (96.1)|
|West||13,867 (21.3)||902 (8.4)||9823 (91.6)||33 (4.0)||789 (96.0)||109 (4.7)||2211 (95.3)|
|Private||41,762 (64.1)||1945 (5.9)||31,201 (94.1)||202 (4.2)||4655 (95.8)||225 (6.0)||3534 (94.0)|
|Medicaid||1078 (1.7)||59 (13.0)||395 (87.0)||17 (5.7)||282 (94.3)||20 (6.2)||305 (93.8)|
|Medicare||19,749 (30.3)||811 (5.0)||15,343 (95.0)||68 (3.8)||1704 (96.2)||85 (4.7)||1738 (95.8)|
|Other||2559 (3.9)||106 (6.8)||1461 (93.2)||21 (4.4)||459 (95.6)||22 (4.3)||490 (95.7)|
|Median zip code income||<.001||.001||.2|
|1st quartile||8678 (13.3)||250 (4.5)||5246 (95.5)||104 (4.9)||2016 (95.1)||60 (5.6)||1002 (94.4)|
|2nd quartile||13,112 (20.1)||458 (4.5)||9727 (95.5)||53 (3.0)||1690 (97.0)||52 (4.4)||1132 (95.6)|
|3rd quartile||16,196 (24.9)||628 (4.8)||12,408 (95.2)||58 (3.5)||1580 (96.5)||82 (5.4)||1440 (94.6)|
|4th quartile||25,531 (39.2)||1501 (7.0)||19,911 (93.0)||77 (4.5)||1633 (95.5)||138 (5.7)||2271 (94.3)|
|Missing||1631 (2.5)||84 (7.0)||1108 (93.0)||16 (8.1)||181 (91.9)||20 (9.3)||222 (91.7)|
Descriptive statistics focused on frequencies and proportions for categorical variables. Means, medians, and ranges were reported for continuously coded variables. Chi-square and independent-sample t tests were used to compare the statistical significance of differences in proportions and means, respectively. The chi-square trend test was used to assess temporal trends.
First, we assessed the rate of MIRP use over time. Univariable and multivariable logistic regression models were used to predict MIRP status. Covariates included race, age, year of surgery, baseline CCI, AHC tertiles, hospital region, insurance status, and median zip code income. Multivariable models predicting MIRP status were fitted for each individual AHC tertiles. Then, logistic regression models relying on generalized estimating equations (GEE models) further adjusted for clustering among hospitals.12 All analyses were repeated after stratification according to the years 2001-2005 verus 2006-2007.
|Multivariable Logistic Regression 2001-2007 (1359 Institutions)||Multivariable Logistic Regression 2001-2005 (1174 Institutions)||Multivariable Logistic Regression 2006-2007 (572 institutions)|
|Odds Ratio (95% CI)||P||Odds Ratio (95% CI)||P||Odds Ratio (95% CI)||P|
|African American||0.86 (0.77-0.97)||.01||0.78 (0.65-0.93)||.007||0.89 (0.77-1.06)||.1|
|Other||1.03 (0.93-1.13)||.6||1.03 (0.90-1.17)||.7||0.97 (0.83-1.11)||.7|
|Year of surgery|
|Age||1.00 (1.00-1.01)||.3||1.00 (0.99-1.02)||.4||1.00 (0.99-1.01)||.7|
|Charlson score (≥1 vs 0)||0.97 (0.89-1.05)||.4||0.94 (0.83-1.06)||.3||0.97 (0.86-1.10)||.6|
|Annual hospital caseload|
|Intermediate||4.91 (3.72-6.50)||<.001||8.29 (4.81-14.26)||<.001||3.48 (2.22-5.99)||<.001|
|High||20.40 (15.50-26.84)||<.001||52.8 (30.8-90.7)||<.001||10.3 (6.9-17.6)||<.001|
|Midwest||0.37 (0.28-0.51)||<.001||0.63 (0.32-1.23)||.2||0.26 (0.16-0.51)||<.001|
|South||0.59 (0.48-0.73)||<.001||0.82 (0.48-1.40)||.4||0.55 (0.38-0.94)||<.001|
|West||0.82 (0.64-1.04)||.1||1.85 (1.05-3.24)||.03||0.50 (0.33-0.87)||<.001|
|Medicaid||1.34 (0.94-1.92)||.1||0.51 (0.24-1.09)||.08||1.46 (1.15-3.09)||.002|
|Medicare||0.97 (0.80-1.19)||.8||0.94 (0.81-1.08)||.4||1.03 (0.87-1.19)||.7|
|Other||1.21 (0.92-1.60)||.2||0.99 (0.76-1.29)||1.0||1.12 (0.86-1.45)||.4|
|Median zip code income|
|2nd quartile||0.88 (0.67-1.16)||.4||0.75 (0.41-1.38)||.4||1.01 (0.63-1.87)||1.0|
|3rd quartile||1.05 (0.81-1.35)||.7||1.03 (0.58-1.82)||.9||1.20 (0.77-2.13)||.4|
|4th quartile||1.12 (0.87-1.45)||.4||0.96 (0.54-1.69)||.9||1.39 (0.87-2.46)||.2|
|Missing||1.32 (0.94-1.85)||.1||1.24 (0.64-2.41)||.5||1.50 (0.86-2.92)||0.2|
All statistical analyses were performed using the R statistical package system (R Foundation for Statistical Computing, Vienna, Austria), with a 2-sided significance level set at P < .05.
Between 2001 and 2007, 89,970 radical prostatectomies were recorded within the National Inpatient Sample. After exclusion of 24,822 patients for whom race information was unavailable, we identified 3581 MIRPs (5.5%). This cohort represents the focus of the analyses (Table 1). For the overall cohort, the average patient age was 61.1 years (median, 62.0 years, range 28 years-89 years). Caucasians accounted for the majority of patients (78.8%), followed by African Americans (11.4%) and other racial groups (9.9%). Most patients had a CCI of 0 (79.8%) and had private insurance (64.1%). MIRP rates were 0.7%, 3.5%, and 12.2% for patients treated at, respectively, low, intermediate, and high AHC institutions (P < .001).
After stratification according to race, African American patients were younger (mean age 59.0 years vs 61.4 years) and more likely to have a CCI ≥1 (26.1% vs 19.1%) than their Caucasian counterparts. After stratification according to MIRP vs ORP, MIRP patients were younger (mean age 60.7 vs 61.1 years) and more likely to have a CCI of 0 (83.4% vs 79.6%). The comparison of African American MIRP patients with their Caucasian counterparts showed that they were on average younger (58.3 years vs 61.0 years, P < .001), less likely to have a CCI of 0 (78.2% vs 84.4%, P = .006), less likely to be treated at a high AHC institution (70.1% vs 79.7%, P < .001), less likely to have private insurance (65.6% vs 66.6%, P < .001), less likely to be in the 4th quartile of median zip code income (25.0% vs 51.4%, P < .001), but more likely to be treated in the Northeast region (39.2% vs 27.4%, P < .001).
In African American individuals, the MIRP utilization rates ranged from 0.3% to 9.0% between 2001 and 2007 (Fig. 1). For the same years, MIRP utilization rates in Caucasian patients ranged from 0.9% to 11.6%. The utilization rate increased 10.5-fold in African American versus 5.3-fold in Caucasian patients (chi-square trend, P <. 001 both).
In multivariable analyses adjusted for age, year of surgery (2001-2005 vs 2006-2007), baseline CCI, AHC, hospital region, insurance status, and median zip code income, African Americans were 14% less likely to undergo MIRP than their Caucasian counterparts if the entire study period was considered (P = .01) (Table 2). Since the utilization rate became virtually equal in 2006, we stratified the temporal trends according to two separate periods: 2001-2005 versus 2006-2007. We repeated the multivariable analyses in both subgroups. In the early period, African Americans were 22% less likely to undergo MIRP than their Caucasian counterparts (P = .007). Conversely, in the late period, the rate difference dropped to 11% and was not statistically significant (P = .1).
In the current study, we tested the hypothesis that racial disparities exist in the utilization rates of MIRP. Our results showed that racial disparities existed in the utilization rates of MIRP in the early study years (2001-2005). Specifically, African Americans were less likely to be treated with MIRP than Caucasians. For example, between 2001 and 2005, the MIRP rates in African Americans increased significantly from 0.8% to 3.5% versus 2.2% to 4.7% in Caucasians. In 2006, the rates in African Americans virtually equaled those recorded in Caucasians. The multivariable analyses corroborated the improvement of racial disparities in MIRP utilization rates. Specifically, the overall multivariable analyses showed a 14% lower rate of MIRP in African Americans relative to Caucasians. This difference became even more pronounced after period stratification. In the early period (2001-2005), African Americans were 22% less likely to have a MIRP. In the late study period (2006-2007), African Americans were 11% less likely to undergo a MIRP, relative to their Caucasian counterparts. Taken together, our findings demonstrate that the racial disparities decreased substantially over time. In the 2 most contemporary study years, the difference was not statistically significant after adjustment for multiple potential confounders. This implies that minimally invasive surgery increased as the urologic community became more sensitized to the need of equal access to care.
The presence and the magnitude of the observed racial disparities should be interpreted with respect to the influence of patient preferences toward treatment outcomes. Patient preferences represent an important source of bias in the utilization rates of various diagnostic and therapeutic options for localized prostate cancer. Utilization of MIRP is invariably affected by patient preferences as well. Unfortunately, the effect of patient preferences tends not to be systematic or structured. For example, Steginga et al showed that men do not use information about medical treatments comprehensively or systematically when making treatment decisions, and their processing of medical information was biased by their previous beliefs about cancer and health.30 These findings have implications for the provision of informational and decisional support to men considering prostate cancer treatment. Similarly, others have reported that the single strongest predictor of treatment choice was the type of physician seen.17 This variable is not accounted for in the NIS.
Several explanations may also be advanced to ascertain the racial disparities in the overall cohort. These consist of other patient variables (disease characteristics, education), insurance status, and socioeconomic determinants. For example, patients with more aggressive disease may be directed toward ORP. Some data indicate that African Americans may harbor more aggressive prostate cancer variants.17 In consequence, more ORPs relative to MIRP may be performed in African American patients. Surgeon preferences and institutional administrative considerations also affect the choice of prostatectomy type.18 Experience and familiarity with MIRP will affect whether MIRPs are performed and to what extent they are performed. For example, surgeons with moderate experience in MIRP may elect to perform some but not all RP in the minimally invasive fashion. Institutional considerations may further confound the MIRP delivery rates. Moreover, based on cost-control considerations, some institutions may encourage MIRPs while others may prioritize ORPs. Although many institutional series have reported superior outcomes of MIRP,19 concerns regarding late complications,20 steeper learning curve,21 and cost-effectiveness22 have been raised. Nonetheless, Abdollah et al showed that total hospital charges are substantially lower when MIRPs are performed by high-volume surgeons versus low-volume surgeons.23 In consequence, institutions with high AHC may encourage more MIRP use than institutions where MIRP expertise is limited or not available, for it is important to note that in the current study, African American patients were more often treated at low-AHC institutions.
Socioeconomic status may also affect the choice of prostatectomy type.18 Although the limitations of our data set does not allow for adjustment of socioeconomic status, we performed adjustment for median zip code income, which has been previously described as an appropriate surrogate for patient socioeconomic status.24 Moreover, this topic has been widely discussed. In a study from the California Cancer Registry, Morris et al concluded that black men were the least likely to be treated with prostatectomy regardless of the income level.25 Moreover, African Americans may be less attracted to MIRP than their Caucasian counterparts because it is known that racial disparities also exist in access to other laparoscopic procedures.26, 27 Educational differences are known to exist between these 2 groups and may add to the differences.28 Finally, insurance status may affect access.7 Some payers may not reimburse institutions that offer MIRP. Previous data show that African Americans were frequently uninsured and had less universal insurance plans than their Caucasian counterparts.29 Nonetheless, insurance status was not an independent predictor of MIRP utilization in our analyses. Taken together, it is clear that residual biases or confounders that cannot be fully controlled or accounted for will affect any type of objective analysis. It is likely that any or many of these variables may have changed during the course of the current study.
Hu et al reported that black and Hispanic versus white and Asian men were less likely to undergo MIRP versus ORP.13 The authors concluded that the observed racial and sociodemographic variation might result from the highly successful robotic-assisted MIRP marketing campaign disseminated via media channels likely to be frequented by men of higher socioeconomic status. However, the population of their data set consists of individuals aged 65 years or older with Medicare coverage. These 2 criteria hamper the generalizability of this study population to all men treated with MIRP in the United States. Moreover, the reported racial disparities were only assessed in unadjusted statistical analyses. Differences in patient characteristics may spuriously inflate the magnitude of the observed racial disparities. For example, underlying comorbidities may disqualify patients from consideration for MIRP. In consequence, adjusted multivariable models should ideally be used when racial disparities are examined. In the current study, we relied on multivariable models. In addition, we relied on GEE models to account for the potential effect of clustering within hospitals. Such measure is equally important to multivariable adjustment, as patients with similar characteristics may be operated at the same institution.
Other investigators have also tested for differences in the utilization of several treatment modalities for localized prostate cancer. For example, Schapira et al reported a significant racial difference in the utilization of select treatment modalities for localized prostate carcinoma.5 Using data from the Surveillance, Epidemiology, and End Results (SEER) registry from 1988 to 1989, they reported that African Americans were less likely than Caucasians to receive aggressive therapy. Furthermore, among those receiving aggressive therapy, African Americans were less likely than Caucasians to have a radical prostatectomy. Adjustment for patient, disease, and community health care availability characteristics did not change these results. Others have since reported similar findings.3, 6, 14-16 Taken together, race affects MIRP utilization, as well as other aspects of prostate cancer care.
The primary implication of the current report consists of highly encouraging findings that indicate improvement in MIRP utilization rates in African Americans versus Caucasians. Ideally, more patients in both groups would be treated with MIRP if the advantages of MIRP continue being corroborated relative to ORP. Although we demonstrated an improvement of MIRP utilization rates, we cannot generalize our conclusions about equal access to other diagnostic or therapeutic prostate cancer processes. Evaluation of those processes will require a complete independent analysis. Nonetheless, our findings are highly encouraging.
In summary, our findings indicate that MIRP utilization rates are increasing at a faster rate in African Americans than in Caucasians. These differences stemmed from lower baseline MIRP rates in African American than in Caucasian patients. These disparities improved significantly in the most contemporary years. Nonetheless, monitoring of epidemiologic trends in access to health care by racial minorities and programs aimed at providing equal access to care are still needed.
P.I.K. is partially supported by the University of Montreal Health Center Urology Specialists, Fonds de la Recherche en Sante du Quebec, the University of Montreal Department of Surgery, and the University of Montreal Health Center (CHUM) Foundation.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 10United States Census Bureau. 2000 [cited 2010]. Available at: http://www.census.gov