Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Higher caseload improves cervical cancer survival in patients treated with brachytherapy

Higher caseload improves cervical cancer survival in patients treated with brachytherapy Objectives: Increased caseload has been associated with better patient outcomes in many areas of health care, including high-risk surgery and cancer treatment. However, such a positive volume vs. outcome relationship has not yet been validated for cervical cancer brachytherapy. The purpose of this study was to examine the relationship between physician caseload and survival rates in cervical cancer treated with brachytherapy using population-based data. Methods: Between 2005 and 2010, a total of 818 patients were identified using the Taiwan National Health Insurance Research Database. Multivariate analysis using a Cox proportional hazards model and propensity scores was used to assess the relationship between 5-year survival rates and physician caseloads. Results: As the caseload of individual physicians increased, unadjusted 5-year survival rates increased (P = 0.005). Using a Cox proportional hazard model, patients treated by high-volume physicians had better survival rates (P =0.03), after adjusting for comorbidities, hospital type, and treatment modality. When analyzed by propensity score, the adjusted 5-year survival rate differed significantly between patients treated by high/medium-volume physicians vs. patients treated by low/medium-volume physicians (60% vs. 54%, respectively; P =0.04). Conclusions: Provider caseload affected survival rates in cervical cancer patients treated with brachytherapy. Both Cox proportional hazard model analysis and propensity scores showed association between high/medium volume physicians and improved survival. Introduction cervical cancer patients to prevent damage to surround- Cervical cancer remains the most important cause ing normal tissues. of cancer death in women from Taiwan with an age- Brachytherapy is a technically demanding process. The adjusted incidence of 26.2 per one hundred thousand “practice makes perfect” hypothesis may be valid for women [1,2]. The combination of chemotherapy admin- such a procedure, in that increased experience improves istered concurrently with radiotherapy shows survival patient outcomes. The association between increased benefit in patients with bulky and locally advanced cer- caseload and improved patient outcomes has been re- vical cancer [3]. However, dose is related to both local ported for both high-risk surgery and cancer treatment control of tumor growth and overall survival. The risk of [1,2]. Positive correlations between improved treatment tissue toxicity currently limits the external radiation dose outcomes and increased caseload volume have been that can be safely delivered [4]. Thus, brachytherapy is documented for nasopharyngeal cancer, breast cancer, often combined with external beam radiotherapy in oral cancer, and esophageal cancer [2,5-7]. However, such a positive volume-outcome relationship has not been established for cervical cancer brachytherapy. The purpose * Correspondence: DOC31221@ndmctsgh.edu.tw; of this study was to examine the relationship between oncology158@yahoo.com.tw physician caseload and survival rates in cervical cancer pa- Equal contributors Department of Radiation Oncology, Buddhist Dalin Tzu Chi Hospital, 2, Ming tients treated with brachytherapy, using population-based Sheng Road, Dalin, Chiayi, Taiwan data. School of Medicine, Tzu Chi University, Hualien, Taiwan Full list of author information is available at the end of the article © 2014 Lee et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Lee et al. Radiation Oncology 2014, 9:234 Page 2 of 6 http://www.ro-journal.com/content/9/1/234 Materials and methods of Disease, Ninth Revision, Clinical Modification codes Ethics statement 180) were included who received radiotherapy or chemora- The study protocol was approved by the Buddhist Dalin diotherapy between 2005 and 2007. Patients were excluded Tzu Chi General Hospital Institutional Review Boards. who had unclear treatment modalities or incomplete phys- The institutional review board waived the need for writ- ician data. Finally, 818 patients, treated by 93 radiation on- ten informed consent from the participants because the cologists during this 5-year period, were included in our data analyzed consisted of anonymous secondary data analysis. released to the public for research. Physicians were further stratified by their total patient volumes (using the unique physician identifiers in this Patients and study design database) and by their caseload of cervical cancer pa- Between 2005 to 2010, data from the National Health tients. The volume category cutoff points (high, medium, Insurance (NHI) Research Database was used in this and low) were determined by sorting the 818 patients study. This data contained all covered medical benefit into three groups (1–11 cases = low caseload), 12–40 claims for over 23 million people in Taiwan (approximately cases = medium caseload, and ≧41 cases = high caseload), 97 percent of the island’s population). All patients with as previously described [5,8]. The volume category cutoff cervical cancer (as defined by International Classification points were determined by sorting the sample into 3 Table 1 Patients characteristics according to caseload (n = 818) Cervical cancer caseload group Variable Low Medium High P-value (1–11) (12–40) (41–78) (n = 280) (n = 262) (n = 276) Age <0.001 25-44 years 23(8.2) 29(11.1) 40(14.5) 45-54 years 53(18.9) 57(21.8) 74(26.8) 55-64 years 47(16.8) 33(12.6) 58(21.0) 65-74 years 76(27.1) 68(26.0) 60(21.7) ≧75 years 81(28.9) 75(28.6) 44(15.9) Charlson comorbidity index score 0.009 0 103(36.8) 113(43.1) 139(50.4) 1-3 109(38.9) 78(29.8) 80(29.0) ≧4 68(24.3) 71(27.1) 57(20.7) Treatment modality 0.001 Radiotherapy 122(43.6) 82(31.3) 83(30.1) Chemoradiotherapy 158(56.4) 180(68.7) 193(69.9) Geographic location <0.001 North 98(35.0) 108(41.2) 57(20.7) Central 74(26.4) 84(32.1) 129(46.7) Southern and Eastern 108(38.6) 70(26.7) 90(32.6) Enrollee category 0.89 EC 1-2 60(21.4) 57(21.8) 57(20.7) EC 3 109(38.9) 107(40.8) 100(36.2) EC 4 57(20.4) 51(19.5) 58(21.9) Other 54(19.3) 47(17.9) 61(22.1) Urbanization 0.14 Urban 66(23.6) 84(32.1) 73(26.4) Suburban 133(47.5) 102(38.9) 130(47.1) Rural 81(28.9) 76(29.0) 73(26.4) Values are given as number (percentage). Lee et al. Radiation Oncology 2014, 9:234 Page 3 of 6 http://www.ro-journal.com/content/9/1/234 Table 2 Physician characteristics (n = 93) approximately equal groups, so that each group would have approximately equal numbers of patients. These Physician caseload group cervical cancer patients were then linked to death data Variable Low Medium High P -value extracted from the records covering the years between (1–11) (12–40) (41–78) 1996 and 2010. Total no. of physicians 74 14 5 Age (years) 0.90 Measurements Mean ± SD 42 ± 8 41 ± 6 41 ± 4 The key dependent variable of interest was the 5-year Gender 0.71 survival rate. The key independent variables were the Male 64(86) 13(92) 4(80) cervical cancer caseloads (low, medium, or high). Other Female 10(13) 1(7) 1(20) physician characteristics included age (≦40, 41–50, ≧51 Caseload <0.001 years) and gender. Patient characteristics included age, Mean ± SD 3 ± 2 18 ± 7 55 ± 13 geographic location, treatment modality, severity of disease, enrollee category (EC), and urbanization. The Values are given as number (percentage). Abbreviation: SD = standard deviation. disease severity in each patient was assessed using the modified Charlson comorbidity index score, which has adjusting for hospital type, surgeon characteristics, and been widely used, in recent years, for risk adjustment in patient demographics. administrative claims data sets [9]. This study used EC as a proxy measure of socioeco- Propensity score nomic status, which is an important prognostic factor in Propensity analysis was used to reduce the effect of selec- cancer patients [10,11]. Patients with cervical cancer tion bias on our hypothesis, as described by Rosenbaum were classified into four subgroups: EC 1 (civil servants, and Rubin [13-15]. Propensity score stratification replaced full-time, or regular paid personnel with a government the many confounding factors that might be present in an affiliation), EC 2 (employees of privately owned institu- observational study with such a variety of factors. To cal- tions), EC 3 (self-employed individuals, other employees, culate the propensity score in this study, patient character- and members of farmers’ or fishermen’s associations), istics were entered into a logistic regression model that EC 4 (veterans, low-income families, and substitute ser- predicted selection for high/medium-volume surgeons. vice draftees), and other [12]. In Taiwan, government These patient characteristics included the year in which affiliated workers have stable job occupation and fixed the patient was diagnosed, their age, gender, Charlson salary in every month than self-employed. On average, comorbidity index score, geographic area of residence, government affiliated workers’ payroll-related amount enrollee category, and treatment modality. The study for the health insurance was highest. population was then divided into five discrete strata based The hospitals were categorized by ownership (public, on propensity score. The effect of caseload assignment on not-for-profit, or for-profit) and hospital type (medical 5-year survival rates was analyzed within each quintile. center, regional hospital, or district hospital). Statistical analysis The SAS statistical package (version 9.2; SAS Institute, Inc., Cary, N.C.) and SPSS (version 21, SPSS Inc., Chicago, IL, USA) was used for data analysis. A two-sided P value < 0.05 was used to determine statistical significance. The cumulative 5-year survival rates and the survival curves for each group were compared by the log-rank test. Survival was measured from the time of cervical cancer diagnosis to the time of death. Cox proportional regression model and survival analysis using propensity score stratification were used to compare outcomes between different groups based on caseload. Cox proportional hazards model The Cox proportional regression model was used to Figure 1 Cervical cancer survival rates by physician caseload. evaluate the effect of caseload on survival rate after Lee et al. Radiation Oncology 2014, 9:234 Page 4 of 6 http://www.ro-journal.com/content/9/1/234 Table 3 Cervical cancer survival rate and adjusted hazard Table 3 Cervical cancer survival rate and adjusted hazard ratios by physician caseload groups and the ratios by physician caseload groups and the characteristics of the patients and providers (n = 818) characteristics of the patients and providers (n = 818) (Continued) Variable Adjusted 95% CI P-value hazard ratio Urbanization Physician characteristics Urban 1 Physician volume Suburban 0.73 (0.56-0.95) 0.02 Low (1–11) 1 Rural 0.66 (0.48-0.91) 0.01 Medium (12–40) 0.90 (0.69-1.19) 0.49 95% CI, 95% confidence interval. High (41–78) 0.69 (0.50-0.97) 0.03 Physician age The Mantel-Haenszel odds ratio was calculated in ≦40 years 1 addition to the Cochran-Mantel-Haenszel χ statistic. 41-50 years 0.94 (0.72-1.24) 0.70 Results ≧51 years 0.88 (0.56-1.36) 0.56 A total of 346 out of 818 patients (42%) died after under- Hospital characteristics going treatment between 2005 and 2007. A total of 93 Hospital ownership radiation oncologists were included in the analysis. The Public 1 characteristics of the physicians and patients are summa- Non-for-profit 0.95 (0.72-1.25) 0.74 rized in Tables 1 and 2. Patients in the low-volume phys- For-profit 0.98 (0.69-1.40) 0.95 ician group were more likely to undergo radiotherapy, reside in Southern and Eastern Taiwan, and have higher Hospital level comorbidity score, than their counterparts in other Medical center 1 groups. There were 74 radiation oncologists (80%) in the Regional hospital 0.70 (0.71-1.26) 0.70 low-volume group, 14 physicians (15%) in the medium- District hospital 1.59 (1.01-2.49) 0.04 volume group, and five (5%) physicians in the high- Patient characteristics volume group. The mean age of all physicians was 41 ± 6 Patient age years. There was no significant difference among physi- cians who comprised these three caseload groups with 25-44 years 1 regards to age (P =0.90). 45-54 years 1.26 (0.82-1.92) 0.28 55-64 years 1.11 (0.70-1.74) 0.64 Analysis using a Cox proportional hazards model 65-74 years 0.96 (0.62-1.48) 0.85 The 5-year survival rates, by physician caseload group, ≧75 years 1.23 (078–1.94) 0.36 are shown in Figure 1. The 5-year survival rates were Comorbidity index score 48%, 54%, and 64% for low-, medium-, and high-volume surgeons, respectively (P = 0.005). Table 3 shows the adjusted hazard ratios (calculated using the Cox pro- 1-3 1.40 (1.07-1.83) 0.01 portional hazards regression model) after adjusting for ≧4 2.52 (1.93-3.29) <0.001 patient comorbidities, hospital type, and treatment mo- Treatment modality dality. Physicians’ age and 5-year survival have no associ- Chemoradiotherapy 1 ation (P > 0.05). The hazard ratio for age 41–50, and ≧51 Radiotherapy 1.23 (1.08-1.40) 0.002 during the 5-year follow-up was 0.94 (P = 0.70) and 0.88- times (P = 0.56) lower than in ≦40 years respectively. Geographic location The positive association between survival and physician North 1 caseload remained statistically significant after multivari- Central 1.17 (0.85-1.62) 0.32 ate analysis. Patients treated by high-volume physicians Southern and Eastern 1.12 (0.81-1.55) 0.47 had better survival rates (hazard ratio [HR] = 0.69; 95% Enrollee category confidence interval [CI], 0.50-0.97; P = 0.03), after Other 1 adjusting for other factors. EC 1-2 0.92 (0.65-1.30) 0.65 Analysis using propensity scores EC 3 1.02 (0.74-1.39) 0.88 Patients were stratified by propensity score and the ef- EC 4 1.09 (0.77-1.54) 0.61 fect of physician caseload on survival was assessed. The population was stratified into propensity quintiles, as Lee et al. Radiation Oncology 2014, 9:234 Page 5 of 6 http://www.ro-journal.com/content/9/1/234 previously described. Table 4 shows the survival rates for The quality of the risk-adjustment techniques used in caseload groups after stratification. The percentage of analyzing administrative information is an important issue. patients treated by low-volume physicians decreased In the first part of this study, a Cox proportional hazard from the first propensity quintile to the fifth, as pre- model was used to compare the effects of high/medium dicted by the propensity model. In each of the five strata, volume versus low volume on survival rates. We found patients treated by high-volume physicians had a higher that treatment by high/medium-volume physicians was sig- 5-year survival rate. While controlling for propensity nificantly associated with a lower adjusted hazard ratio for score (with fewer patients dying who were treated by death. Patients treated by high-volume physicians were high/medium-volume physicians), the P value equaled found to have a 31% lower risk of death after adjusting for 0.04 using Cochran-Mantel-Haenszel statistics. This ana- comorbidities and other confounding factors. However, lysis demonstrated a significant difference in survival be- there were differences in clinical conditions between case- tween patients treated by low vs. high/medium-volume load groups. In the second part of our series, propensity physicians, (adjusted odds ratio = 0.71, 95% CI, 0.51-0.99). score was used to stratify patients into five strata with simi- The adjusted 5-year survival rates for low vs. high/medium- lar propensity score in order to reduce the effect of selec- volume physicians were 54% vs. 60%, respectively (P = tion bias on caseload groups [14,15,17]. Patients treated 0.04). by high/medium-volume physicians were found to have In summary, cervical cancer patients treated by higher a 6% relative improvement in adjusted 5-year survival volume physicians showed improved survival. The ro- rate (P = 0.04) compared to those treated by low-volume bustness of this result was demonstrated by two different physicians. multivariate analyses, the Cox proportional regression Several hypotheses have been proposed regarding the model and stratification by propensity score. relationship between caseload volume and outcome. They suggest that increased caseload may help physicians or Discussion hospital staff improve their ability to perform treatment Improved patient outcomes have been correlated with procedures, such as planning and manipulation of the higher caseload volumes. However, there is limited data radioactive source to target treatment sites, gauze packing, about the use of brachytherapy in cervical cancer patients. dose calculation or computerized planning. Careful ma- Although the Royal College of Radiologists has made the nipulation of the target volume is important for treatment pragmatic decision to maintain sufficient experience and of cervical cancer with brachytherapy. A team that is com- expertise, they are not backed by any published evidence fortable with a higher caseload volume may be more adept [16]. We used a Cox proportional hazards model and pro- at administering radiation dosage which improves loco- pensity score to evaluate the relative patient benefit of treat- regional control of cancer and reduces the risk of toxicity ment by high/medium-volume physicians vs. low -volume to nearby normal organs and tissues. physicians using cervical cancer brachytherapy. From these Although our study revealed some issues that may be results of both forms of multivariate analyses, we found useful for policy makers, further research is necessary that the 5-year survival rates for brachytherapy patients to identify differences in care and treatment strategies treated by high/medium -volume physicians were signifi- among low-, medium-, and high-volume physicians. In cantly better compared to patients treated by low-volume our study, nearly 33% of patients were treated by only physicians. five high-volume radiation oncologists. The viewpoints Table 4 5-year survival rates of cervical patients according to propensity score strata; low-volume vs. high/medium- volume physicians Propensity score stratum Low-volume physician group High/medium-volume physician group No. % of stratum Survival rate (%) No. % of stratum Survival rate (%) P-value 1 112 68 50 51 31 52 0.07 2 84 51 50 80 48 62 0.64 3 50 30 42 114 69 60 0.41 4 19 11 68 145 88 59 0.44 5 15 9 60 148 90 66 0.02 Total 280 54 538 60 0.09 0.04 Stratum 1 had the strongest propensity for low-volume physician; Stratum 5, for high/medium-volume physicians. Cochran-Mantel-Haenszel statistics; adjusted odds ratio = 0.71,95% confidence interval = 0.51-0.99. Lee et al. Radiation Oncology 2014, 9:234 Page 6 of 6 http://www.ro-journal.com/content/9/1/234 of high-volume physicians may influence the development Received: 13 October 2013 Accepted: 10 October 2014 of effective protocols and clinical practice guidelines. Furthermore, the treatment strategies of high-volume References physicians should be analyzed and adopted, throughout 1. Chang CM, Huang KY, Hsu TW, Su YC, Yang WZ, Chen TC, Chou P, Lee CC: Multivariate analyses to assess the effects of surgeon and hospital the country, to improve survival rates. volume on cancer survival rates: a nationwide population-based study in Our study had several limitations. First, we could not as- Taiwan. PLoS One 2012, 7:e40590. sess the relationship of caseload to stage, tumor size or 2. Lee CC, Huang TT, Lee MS, Su YC, Chou P, Hsiao SH, Chiou WY, Lin HY, Chien SH, Hung SK: Survival rate in nasopharyngeal carcinoma improved local control rate because this information was not avail- by high caseload volume: a nationwide population-based study in able from the database. Although this limitation may Taiwan. Radiat Oncol 2011, 6:92. influence our results, Begg et al., using a SEER-Medicare 3. Peters WA 3rd, Liu PY, Barrett RJ 2nd, Stock RJ, Monk BJ, Berek JS, Souhami L, Grigsby P, Gordon W Jr, Alberts DS: Concurrent chemotherapy and linked database, reported that cancer stage and patient age pelvic radiation therapy compared with pelvic radiation therapy alone as were independent of caseload volume [18]. Though the adjuvant therapy after radical surgery in high-risk early-stage cancer of health system in Taiwan is not complete the same as the the cervix. J Clin Oncol 2000, 18:1606–1613. 4. Ferrigno R, dos Santos Novaes PE, Pellizzon AC, Maia MA, Fogarolli RC, one in USA, patient in Taiwan are also free to choose any Gentil AC, Salvajoli JV: High-dose-rate brachytherapy in the treatment of physician no matter the disease severity, stage or comor- uterine cervix cancer. Analysis of dose effectiveness and late bidity. In addition, selective referral bias must also be con- complications. Int J Radiat Oncol Biol Phys 2001, 50:1123–1135. 5. Lin CC, Lin HC: Effects of surgeon and hospital volume on 5-year survival sidered, i.e., healthier patients or patients with early-stage rates following oral cancer resections: the experience of an Asian country. disease may tend to be referred to the high-volume physi- Surgery 2008, 143:343–351. cians. Although healthier patients tend to be wealthier 6. Peltoniemi P, Huhtala H, Holli K, Pylkkanen L: Effect of surgeon's caseload on the quality of surgery and breast cancer recurrence. Breast 2012, and they advocate for themselves, the National Health In- 21:539–543. surance covered approximately 97 percent of the island’s 7. Markar SR, Karthikesalingam A, Thrumurthy S, Low DE: Volume-outcome population and the hospitals are dispersion in Taiwan. relationship in surgery for esophageal malignancy: systematic review and meta-analysis 2000–2011. J Gastrointest Surg 2012, 16:1055–1063. The probability for patients’ choice is average. Second, the 8. Goodney PP, Stukel TA, Lucas FL, Finlayson EV, Birkmeyer JD: Hospital database does not contain information regarding tobacco volume, length of stay, and readmission rates in high-risk surgery. use, dietary habits, and body mass index, which might be Ann Surg 2003, 238:161–167. 9. Deyo RA, Cherkin DC, Ciol MA: Adapting a clinical comorbidity index for risk factors for cervical cancer. Taken together, given the use with ICD-9-CM administrative databases. J Clin Epidemiol 1992, robustness of both the evidence and statistical analyses 45:613–619. used in this study, these limitations are unlikely to have 10. Braaten T, Weiderpass E, Lund E: Socioeconomic differences in cancer survival: the Norwegian women and cancer study. BMC Public Health compromised our results. 2009, 9:178. In summary, using analyses based on a Cox proportional 11. Kwok J, Langevin SM, Argiris A, Grandis JR, Gooding WE, Taioli E: The hazard model and propensity score, we found an associ- impact of health insurance status on the survival of patients with head and neck cancer. Cancer 2010, 116:476–485. ation between higher caseload volume and improved sur- 12. Chen CY, Liu CY, Su WC, Huang SL, Lin KM: Factors associated with vival in cervical cancer patients treated with brachytherapy the diagnosis of neurodevelopmental disorders: a population-based using population-based data. In conclusion, higher provider longitudinal study. Pediatrics 2007, 119:e435–e443. 13. Joffe MM, Rosenbaum PR: Invited commentary: propensity scores. Am J caseload affects survival in cervical cancer patients treated Epidemiol 1999, 150:327–333. with brachytherapy. 14. Rubin DB: Tasks in statistical inference for studying variation in medicine. Med Care 1993, 31:YS103–YS110. Competing interests 15. Rubin DB: Estimating causal effects from large data sets using propensity The authors declare that they have no competing interests. scores. Ann Intern Med 1997, 127:757–763. 16. Symonds P, Davidson S, Vale C, Drinkwater K: Caseload and outcome after Authors’ contributions brachytherapy. Clin Oncol 2013, 25:519–521. LMS, TSJ, LHY and HSK developed the ideas for these studies, performed 17. D'Agostino RB Jr: Propensity score methods for bias reduction in the much of the work, and drafted the manuscript. LCC, SYC, CWY and LHY comparison of a treatment to a non-randomized control group. designed the study, managed and interpreted the data. LCC performed the Stat Med 1998, 17:2265–2281. statistical analysis. All authors read and approved the final manuscript. 18. Begg CB, Cramer LD, Hoskins WJ, Brennan MF: Impact of hospital volume on operative mortality for major cancer surgery. JAMA 1998, Acknowledgments 280:1747–1751. This study is based in part on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, doi:10.1186/s13014-014-0234-2 Department of Health and managed by the National Health Research Cite this article as: Lee et al.: Higher caseload improves cervical cancer Institutes (Registry number 99029). The interpretation and conclusions survival in patients treated with brachytherapy. Radiation Oncology contained herein do not represent those of the Bureau of National Health 2014 9:234. Insurance, Department of Health, or National Health Research Institutes. Author details Department of Radiation Oncology, Buddhist Dalin Tzu Chi Hospital, 2, Ming Sheng Road, Dalin, Chiayi, Taiwan. Department of Otolaryngology, Buddhist Dalin Tzu Chi Hospital, Chiayi, Taiwan. Department of Hematology Oncology, Buddhist Dalin Tzu Chi Hospital, Chiayi, Taiwan. School of Medicine, Tzu Chi University, Hualien, Taiwan. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Radiation Oncology Springer Journals

Higher caseload improves cervical cancer survival in patients treated with brachytherapy

Loading next page...
 
/lp/springer-journals/higher-caseload-improves-cervical-cancer-survival-in-patients-treated-ZS0j4C230B

References (18)

Publisher
Springer Journals
Copyright
Copyright © 2014 by Lee et al.; licensee BioMed Central Ltd.
Subject
Medicine & Public Health; Oncology; Radiotherapy
eISSN
1748-717X
DOI
10.1186/s13014-014-0234-2
pmid
25344121
Publisher site
See Article on Publisher Site

Abstract

Objectives: Increased caseload has been associated with better patient outcomes in many areas of health care, including high-risk surgery and cancer treatment. However, such a positive volume vs. outcome relationship has not yet been validated for cervical cancer brachytherapy. The purpose of this study was to examine the relationship between physician caseload and survival rates in cervical cancer treated with brachytherapy using population-based data. Methods: Between 2005 and 2010, a total of 818 patients were identified using the Taiwan National Health Insurance Research Database. Multivariate analysis using a Cox proportional hazards model and propensity scores was used to assess the relationship between 5-year survival rates and physician caseloads. Results: As the caseload of individual physicians increased, unadjusted 5-year survival rates increased (P = 0.005). Using a Cox proportional hazard model, patients treated by high-volume physicians had better survival rates (P =0.03), after adjusting for comorbidities, hospital type, and treatment modality. When analyzed by propensity score, the adjusted 5-year survival rate differed significantly between patients treated by high/medium-volume physicians vs. patients treated by low/medium-volume physicians (60% vs. 54%, respectively; P =0.04). Conclusions: Provider caseload affected survival rates in cervical cancer patients treated with brachytherapy. Both Cox proportional hazard model analysis and propensity scores showed association between high/medium volume physicians and improved survival. Introduction cervical cancer patients to prevent damage to surround- Cervical cancer remains the most important cause ing normal tissues. of cancer death in women from Taiwan with an age- Brachytherapy is a technically demanding process. The adjusted incidence of 26.2 per one hundred thousand “practice makes perfect” hypothesis may be valid for women [1,2]. The combination of chemotherapy admin- such a procedure, in that increased experience improves istered concurrently with radiotherapy shows survival patient outcomes. The association between increased benefit in patients with bulky and locally advanced cer- caseload and improved patient outcomes has been re- vical cancer [3]. However, dose is related to both local ported for both high-risk surgery and cancer treatment control of tumor growth and overall survival. The risk of [1,2]. Positive correlations between improved treatment tissue toxicity currently limits the external radiation dose outcomes and increased caseload volume have been that can be safely delivered [4]. Thus, brachytherapy is documented for nasopharyngeal cancer, breast cancer, often combined with external beam radiotherapy in oral cancer, and esophageal cancer [2,5-7]. However, such a positive volume-outcome relationship has not been established for cervical cancer brachytherapy. The purpose * Correspondence: DOC31221@ndmctsgh.edu.tw; of this study was to examine the relationship between oncology158@yahoo.com.tw physician caseload and survival rates in cervical cancer pa- Equal contributors Department of Radiation Oncology, Buddhist Dalin Tzu Chi Hospital, 2, Ming tients treated with brachytherapy, using population-based Sheng Road, Dalin, Chiayi, Taiwan data. School of Medicine, Tzu Chi University, Hualien, Taiwan Full list of author information is available at the end of the article © 2014 Lee et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Lee et al. Radiation Oncology 2014, 9:234 Page 2 of 6 http://www.ro-journal.com/content/9/1/234 Materials and methods of Disease, Ninth Revision, Clinical Modification codes Ethics statement 180) were included who received radiotherapy or chemora- The study protocol was approved by the Buddhist Dalin diotherapy between 2005 and 2007. Patients were excluded Tzu Chi General Hospital Institutional Review Boards. who had unclear treatment modalities or incomplete phys- The institutional review board waived the need for writ- ician data. Finally, 818 patients, treated by 93 radiation on- ten informed consent from the participants because the cologists during this 5-year period, were included in our data analyzed consisted of anonymous secondary data analysis. released to the public for research. Physicians were further stratified by their total patient volumes (using the unique physician identifiers in this Patients and study design database) and by their caseload of cervical cancer pa- Between 2005 to 2010, data from the National Health tients. The volume category cutoff points (high, medium, Insurance (NHI) Research Database was used in this and low) were determined by sorting the 818 patients study. This data contained all covered medical benefit into three groups (1–11 cases = low caseload), 12–40 claims for over 23 million people in Taiwan (approximately cases = medium caseload, and ≧41 cases = high caseload), 97 percent of the island’s population). All patients with as previously described [5,8]. The volume category cutoff cervical cancer (as defined by International Classification points were determined by sorting the sample into 3 Table 1 Patients characteristics according to caseload (n = 818) Cervical cancer caseload group Variable Low Medium High P-value (1–11) (12–40) (41–78) (n = 280) (n = 262) (n = 276) Age <0.001 25-44 years 23(8.2) 29(11.1) 40(14.5) 45-54 years 53(18.9) 57(21.8) 74(26.8) 55-64 years 47(16.8) 33(12.6) 58(21.0) 65-74 years 76(27.1) 68(26.0) 60(21.7) ≧75 years 81(28.9) 75(28.6) 44(15.9) Charlson comorbidity index score 0.009 0 103(36.8) 113(43.1) 139(50.4) 1-3 109(38.9) 78(29.8) 80(29.0) ≧4 68(24.3) 71(27.1) 57(20.7) Treatment modality 0.001 Radiotherapy 122(43.6) 82(31.3) 83(30.1) Chemoradiotherapy 158(56.4) 180(68.7) 193(69.9) Geographic location <0.001 North 98(35.0) 108(41.2) 57(20.7) Central 74(26.4) 84(32.1) 129(46.7) Southern and Eastern 108(38.6) 70(26.7) 90(32.6) Enrollee category 0.89 EC 1-2 60(21.4) 57(21.8) 57(20.7) EC 3 109(38.9) 107(40.8) 100(36.2) EC 4 57(20.4) 51(19.5) 58(21.9) Other 54(19.3) 47(17.9) 61(22.1) Urbanization 0.14 Urban 66(23.6) 84(32.1) 73(26.4) Suburban 133(47.5) 102(38.9) 130(47.1) Rural 81(28.9) 76(29.0) 73(26.4) Values are given as number (percentage). Lee et al. Radiation Oncology 2014, 9:234 Page 3 of 6 http://www.ro-journal.com/content/9/1/234 Table 2 Physician characteristics (n = 93) approximately equal groups, so that each group would have approximately equal numbers of patients. These Physician caseload group cervical cancer patients were then linked to death data Variable Low Medium High P -value extracted from the records covering the years between (1–11) (12–40) (41–78) 1996 and 2010. Total no. of physicians 74 14 5 Age (years) 0.90 Measurements Mean ± SD 42 ± 8 41 ± 6 41 ± 4 The key dependent variable of interest was the 5-year Gender 0.71 survival rate. The key independent variables were the Male 64(86) 13(92) 4(80) cervical cancer caseloads (low, medium, or high). Other Female 10(13) 1(7) 1(20) physician characteristics included age (≦40, 41–50, ≧51 Caseload <0.001 years) and gender. Patient characteristics included age, Mean ± SD 3 ± 2 18 ± 7 55 ± 13 geographic location, treatment modality, severity of disease, enrollee category (EC), and urbanization. The Values are given as number (percentage). Abbreviation: SD = standard deviation. disease severity in each patient was assessed using the modified Charlson comorbidity index score, which has adjusting for hospital type, surgeon characteristics, and been widely used, in recent years, for risk adjustment in patient demographics. administrative claims data sets [9]. This study used EC as a proxy measure of socioeco- Propensity score nomic status, which is an important prognostic factor in Propensity analysis was used to reduce the effect of selec- cancer patients [10,11]. Patients with cervical cancer tion bias on our hypothesis, as described by Rosenbaum were classified into four subgroups: EC 1 (civil servants, and Rubin [13-15]. Propensity score stratification replaced full-time, or regular paid personnel with a government the many confounding factors that might be present in an affiliation), EC 2 (employees of privately owned institu- observational study with such a variety of factors. To cal- tions), EC 3 (self-employed individuals, other employees, culate the propensity score in this study, patient character- and members of farmers’ or fishermen’s associations), istics were entered into a logistic regression model that EC 4 (veterans, low-income families, and substitute ser- predicted selection for high/medium-volume surgeons. vice draftees), and other [12]. In Taiwan, government These patient characteristics included the year in which affiliated workers have stable job occupation and fixed the patient was diagnosed, their age, gender, Charlson salary in every month than self-employed. On average, comorbidity index score, geographic area of residence, government affiliated workers’ payroll-related amount enrollee category, and treatment modality. The study for the health insurance was highest. population was then divided into five discrete strata based The hospitals were categorized by ownership (public, on propensity score. The effect of caseload assignment on not-for-profit, or for-profit) and hospital type (medical 5-year survival rates was analyzed within each quintile. center, regional hospital, or district hospital). Statistical analysis The SAS statistical package (version 9.2; SAS Institute, Inc., Cary, N.C.) and SPSS (version 21, SPSS Inc., Chicago, IL, USA) was used for data analysis. A two-sided P value < 0.05 was used to determine statistical significance. The cumulative 5-year survival rates and the survival curves for each group were compared by the log-rank test. Survival was measured from the time of cervical cancer diagnosis to the time of death. Cox proportional regression model and survival analysis using propensity score stratification were used to compare outcomes between different groups based on caseload. Cox proportional hazards model The Cox proportional regression model was used to Figure 1 Cervical cancer survival rates by physician caseload. evaluate the effect of caseload on survival rate after Lee et al. Radiation Oncology 2014, 9:234 Page 4 of 6 http://www.ro-journal.com/content/9/1/234 Table 3 Cervical cancer survival rate and adjusted hazard Table 3 Cervical cancer survival rate and adjusted hazard ratios by physician caseload groups and the ratios by physician caseload groups and the characteristics of the patients and providers (n = 818) characteristics of the patients and providers (n = 818) (Continued) Variable Adjusted 95% CI P-value hazard ratio Urbanization Physician characteristics Urban 1 Physician volume Suburban 0.73 (0.56-0.95) 0.02 Low (1–11) 1 Rural 0.66 (0.48-0.91) 0.01 Medium (12–40) 0.90 (0.69-1.19) 0.49 95% CI, 95% confidence interval. High (41–78) 0.69 (0.50-0.97) 0.03 Physician age The Mantel-Haenszel odds ratio was calculated in ≦40 years 1 addition to the Cochran-Mantel-Haenszel χ statistic. 41-50 years 0.94 (0.72-1.24) 0.70 Results ≧51 years 0.88 (0.56-1.36) 0.56 A total of 346 out of 818 patients (42%) died after under- Hospital characteristics going treatment between 2005 and 2007. A total of 93 Hospital ownership radiation oncologists were included in the analysis. The Public 1 characteristics of the physicians and patients are summa- Non-for-profit 0.95 (0.72-1.25) 0.74 rized in Tables 1 and 2. Patients in the low-volume phys- For-profit 0.98 (0.69-1.40) 0.95 ician group were more likely to undergo radiotherapy, reside in Southern and Eastern Taiwan, and have higher Hospital level comorbidity score, than their counterparts in other Medical center 1 groups. There were 74 radiation oncologists (80%) in the Regional hospital 0.70 (0.71-1.26) 0.70 low-volume group, 14 physicians (15%) in the medium- District hospital 1.59 (1.01-2.49) 0.04 volume group, and five (5%) physicians in the high- Patient characteristics volume group. The mean age of all physicians was 41 ± 6 Patient age years. There was no significant difference among physi- cians who comprised these three caseload groups with 25-44 years 1 regards to age (P =0.90). 45-54 years 1.26 (0.82-1.92) 0.28 55-64 years 1.11 (0.70-1.74) 0.64 Analysis using a Cox proportional hazards model 65-74 years 0.96 (0.62-1.48) 0.85 The 5-year survival rates, by physician caseload group, ≧75 years 1.23 (078–1.94) 0.36 are shown in Figure 1. The 5-year survival rates were Comorbidity index score 48%, 54%, and 64% for low-, medium-, and high-volume surgeons, respectively (P = 0.005). Table 3 shows the adjusted hazard ratios (calculated using the Cox pro- 1-3 1.40 (1.07-1.83) 0.01 portional hazards regression model) after adjusting for ≧4 2.52 (1.93-3.29) <0.001 patient comorbidities, hospital type, and treatment mo- Treatment modality dality. Physicians’ age and 5-year survival have no associ- Chemoradiotherapy 1 ation (P > 0.05). The hazard ratio for age 41–50, and ≧51 Radiotherapy 1.23 (1.08-1.40) 0.002 during the 5-year follow-up was 0.94 (P = 0.70) and 0.88- times (P = 0.56) lower than in ≦40 years respectively. Geographic location The positive association between survival and physician North 1 caseload remained statistically significant after multivari- Central 1.17 (0.85-1.62) 0.32 ate analysis. Patients treated by high-volume physicians Southern and Eastern 1.12 (0.81-1.55) 0.47 had better survival rates (hazard ratio [HR] = 0.69; 95% Enrollee category confidence interval [CI], 0.50-0.97; P = 0.03), after Other 1 adjusting for other factors. EC 1-2 0.92 (0.65-1.30) 0.65 Analysis using propensity scores EC 3 1.02 (0.74-1.39) 0.88 Patients were stratified by propensity score and the ef- EC 4 1.09 (0.77-1.54) 0.61 fect of physician caseload on survival was assessed. The population was stratified into propensity quintiles, as Lee et al. Radiation Oncology 2014, 9:234 Page 5 of 6 http://www.ro-journal.com/content/9/1/234 previously described. Table 4 shows the survival rates for The quality of the risk-adjustment techniques used in caseload groups after stratification. The percentage of analyzing administrative information is an important issue. patients treated by low-volume physicians decreased In the first part of this study, a Cox proportional hazard from the first propensity quintile to the fifth, as pre- model was used to compare the effects of high/medium dicted by the propensity model. In each of the five strata, volume versus low volume on survival rates. We found patients treated by high-volume physicians had a higher that treatment by high/medium-volume physicians was sig- 5-year survival rate. While controlling for propensity nificantly associated with a lower adjusted hazard ratio for score (with fewer patients dying who were treated by death. Patients treated by high-volume physicians were high/medium-volume physicians), the P value equaled found to have a 31% lower risk of death after adjusting for 0.04 using Cochran-Mantel-Haenszel statistics. This ana- comorbidities and other confounding factors. However, lysis demonstrated a significant difference in survival be- there were differences in clinical conditions between case- tween patients treated by low vs. high/medium-volume load groups. In the second part of our series, propensity physicians, (adjusted odds ratio = 0.71, 95% CI, 0.51-0.99). score was used to stratify patients into five strata with simi- The adjusted 5-year survival rates for low vs. high/medium- lar propensity score in order to reduce the effect of selec- volume physicians were 54% vs. 60%, respectively (P = tion bias on caseload groups [14,15,17]. Patients treated 0.04). by high/medium-volume physicians were found to have In summary, cervical cancer patients treated by higher a 6% relative improvement in adjusted 5-year survival volume physicians showed improved survival. The ro- rate (P = 0.04) compared to those treated by low-volume bustness of this result was demonstrated by two different physicians. multivariate analyses, the Cox proportional regression Several hypotheses have been proposed regarding the model and stratification by propensity score. relationship between caseload volume and outcome. They suggest that increased caseload may help physicians or Discussion hospital staff improve their ability to perform treatment Improved patient outcomes have been correlated with procedures, such as planning and manipulation of the higher caseload volumes. However, there is limited data radioactive source to target treatment sites, gauze packing, about the use of brachytherapy in cervical cancer patients. dose calculation or computerized planning. Careful ma- Although the Royal College of Radiologists has made the nipulation of the target volume is important for treatment pragmatic decision to maintain sufficient experience and of cervical cancer with brachytherapy. A team that is com- expertise, they are not backed by any published evidence fortable with a higher caseload volume may be more adept [16]. We used a Cox proportional hazards model and pro- at administering radiation dosage which improves loco- pensity score to evaluate the relative patient benefit of treat- regional control of cancer and reduces the risk of toxicity ment by high/medium-volume physicians vs. low -volume to nearby normal organs and tissues. physicians using cervical cancer brachytherapy. From these Although our study revealed some issues that may be results of both forms of multivariate analyses, we found useful for policy makers, further research is necessary that the 5-year survival rates for brachytherapy patients to identify differences in care and treatment strategies treated by high/medium -volume physicians were signifi- among low-, medium-, and high-volume physicians. In cantly better compared to patients treated by low-volume our study, nearly 33% of patients were treated by only physicians. five high-volume radiation oncologists. The viewpoints Table 4 5-year survival rates of cervical patients according to propensity score strata; low-volume vs. high/medium- volume physicians Propensity score stratum Low-volume physician group High/medium-volume physician group No. % of stratum Survival rate (%) No. % of stratum Survival rate (%) P-value 1 112 68 50 51 31 52 0.07 2 84 51 50 80 48 62 0.64 3 50 30 42 114 69 60 0.41 4 19 11 68 145 88 59 0.44 5 15 9 60 148 90 66 0.02 Total 280 54 538 60 0.09 0.04 Stratum 1 had the strongest propensity for low-volume physician; Stratum 5, for high/medium-volume physicians. Cochran-Mantel-Haenszel statistics; adjusted odds ratio = 0.71,95% confidence interval = 0.51-0.99. Lee et al. Radiation Oncology 2014, 9:234 Page 6 of 6 http://www.ro-journal.com/content/9/1/234 of high-volume physicians may influence the development Received: 13 October 2013 Accepted: 10 October 2014 of effective protocols and clinical practice guidelines. Furthermore, the treatment strategies of high-volume References physicians should be analyzed and adopted, throughout 1. Chang CM, Huang KY, Hsu TW, Su YC, Yang WZ, Chen TC, Chou P, Lee CC: Multivariate analyses to assess the effects of surgeon and hospital the country, to improve survival rates. volume on cancer survival rates: a nationwide population-based study in Our study had several limitations. First, we could not as- Taiwan. PLoS One 2012, 7:e40590. sess the relationship of caseload to stage, tumor size or 2. Lee CC, Huang TT, Lee MS, Su YC, Chou P, Hsiao SH, Chiou WY, Lin HY, Chien SH, Hung SK: Survival rate in nasopharyngeal carcinoma improved local control rate because this information was not avail- by high caseload volume: a nationwide population-based study in able from the database. Although this limitation may Taiwan. Radiat Oncol 2011, 6:92. influence our results, Begg et al., using a SEER-Medicare 3. Peters WA 3rd, Liu PY, Barrett RJ 2nd, Stock RJ, Monk BJ, Berek JS, Souhami L, Grigsby P, Gordon W Jr, Alberts DS: Concurrent chemotherapy and linked database, reported that cancer stage and patient age pelvic radiation therapy compared with pelvic radiation therapy alone as were independent of caseload volume [18]. Though the adjuvant therapy after radical surgery in high-risk early-stage cancer of health system in Taiwan is not complete the same as the the cervix. J Clin Oncol 2000, 18:1606–1613. 4. Ferrigno R, dos Santos Novaes PE, Pellizzon AC, Maia MA, Fogarolli RC, one in USA, patient in Taiwan are also free to choose any Gentil AC, Salvajoli JV: High-dose-rate brachytherapy in the treatment of physician no matter the disease severity, stage or comor- uterine cervix cancer. Analysis of dose effectiveness and late bidity. In addition, selective referral bias must also be con- complications. Int J Radiat Oncol Biol Phys 2001, 50:1123–1135. 5. Lin CC, Lin HC: Effects of surgeon and hospital volume on 5-year survival sidered, i.e., healthier patients or patients with early-stage rates following oral cancer resections: the experience of an Asian country. disease may tend to be referred to the high-volume physi- Surgery 2008, 143:343–351. cians. Although healthier patients tend to be wealthier 6. Peltoniemi P, Huhtala H, Holli K, Pylkkanen L: Effect of surgeon's caseload on the quality of surgery and breast cancer recurrence. Breast 2012, and they advocate for themselves, the National Health In- 21:539–543. surance covered approximately 97 percent of the island’s 7. Markar SR, Karthikesalingam A, Thrumurthy S, Low DE: Volume-outcome population and the hospitals are dispersion in Taiwan. relationship in surgery for esophageal malignancy: systematic review and meta-analysis 2000–2011. J Gastrointest Surg 2012, 16:1055–1063. The probability for patients’ choice is average. Second, the 8. Goodney PP, Stukel TA, Lucas FL, Finlayson EV, Birkmeyer JD: Hospital database does not contain information regarding tobacco volume, length of stay, and readmission rates in high-risk surgery. use, dietary habits, and body mass index, which might be Ann Surg 2003, 238:161–167. 9. Deyo RA, Cherkin DC, Ciol MA: Adapting a clinical comorbidity index for risk factors for cervical cancer. Taken together, given the use with ICD-9-CM administrative databases. J Clin Epidemiol 1992, robustness of both the evidence and statistical analyses 45:613–619. used in this study, these limitations are unlikely to have 10. Braaten T, Weiderpass E, Lund E: Socioeconomic differences in cancer survival: the Norwegian women and cancer study. BMC Public Health compromised our results. 2009, 9:178. In summary, using analyses based on a Cox proportional 11. Kwok J, Langevin SM, Argiris A, Grandis JR, Gooding WE, Taioli E: The hazard model and propensity score, we found an associ- impact of health insurance status on the survival of patients with head and neck cancer. Cancer 2010, 116:476–485. ation between higher caseload volume and improved sur- 12. Chen CY, Liu CY, Su WC, Huang SL, Lin KM: Factors associated with vival in cervical cancer patients treated with brachytherapy the diagnosis of neurodevelopmental disorders: a population-based using population-based data. In conclusion, higher provider longitudinal study. Pediatrics 2007, 119:e435–e443. 13. Joffe MM, Rosenbaum PR: Invited commentary: propensity scores. Am J caseload affects survival in cervical cancer patients treated Epidemiol 1999, 150:327–333. with brachytherapy. 14. Rubin DB: Tasks in statistical inference for studying variation in medicine. Med Care 1993, 31:YS103–YS110. Competing interests 15. Rubin DB: Estimating causal effects from large data sets using propensity The authors declare that they have no competing interests. scores. Ann Intern Med 1997, 127:757–763. 16. Symonds P, Davidson S, Vale C, Drinkwater K: Caseload and outcome after Authors’ contributions brachytherapy. Clin Oncol 2013, 25:519–521. LMS, TSJ, LHY and HSK developed the ideas for these studies, performed 17. D'Agostino RB Jr: Propensity score methods for bias reduction in the much of the work, and drafted the manuscript. LCC, SYC, CWY and LHY comparison of a treatment to a non-randomized control group. designed the study, managed and interpreted the data. LCC performed the Stat Med 1998, 17:2265–2281. statistical analysis. All authors read and approved the final manuscript. 18. Begg CB, Cramer LD, Hoskins WJ, Brennan MF: Impact of hospital volume on operative mortality for major cancer surgery. JAMA 1998, Acknowledgments 280:1747–1751. This study is based in part on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, doi:10.1186/s13014-014-0234-2 Department of Health and managed by the National Health Research Cite this article as: Lee et al.: Higher caseload improves cervical cancer Institutes (Registry number 99029). The interpretation and conclusions survival in patients treated with brachytherapy. Radiation Oncology contained herein do not represent those of the Bureau of National Health 2014 9:234. Insurance, Department of Health, or National Health Research Institutes. Author details Department of Radiation Oncology, Buddhist Dalin Tzu Chi Hospital, 2, Ming Sheng Road, Dalin, Chiayi, Taiwan. Department of Otolaryngology, Buddhist Dalin Tzu Chi Hospital, Chiayi, Taiwan. Department of Hematology Oncology, Buddhist Dalin Tzu Chi Hospital, Chiayi, Taiwan. School of Medicine, Tzu Chi University, Hualien, Taiwan.

Journal

Radiation OncologySpringer Journals

Published: Oct 25, 2014

There are no references for this article.