Regionalization of Stem Cell Transplant Procedures Into Teaching Hospitals in United States: Are We Ready?

LETTER TO THE EDITOR Regionalization of stem cell transplant procedures into teaching hospitals in United States: are we ready?

Bone Marrow Transplantation (2016) 51, 1004–1006; doi:10.1038/ in year 2007 to a high of 99.7% in year 2010. The characteristics of bmt.2016.33; published online 7 March 2016 patients who underwent SCTs by hospital teaching status are presented in Table 1. Mean age of patients who had SCTs in non-teaching hospitals was 47.7 years (44.1 years for those in Regionalization is a process wherein complex high-risk surgical teaching hospitals), and 78.3% of those who had SCTs in procedures are selectively referred to be performed in selectively non-teaching hospitals were Whites (73% of those in teaching high volume, specialized, teaching hospitals. In the recent past, there hospitals were Whites). Females comprised of 46.1% of patients has been increasing regionalization or concentration of a wide array treated in non-teaching hospitals (41.1% in teaching hospitals). of complex surgical procedures into high volume or teaching Northeastern region accounted for only 0.5% of all SCTs hospitals, the underlying empirical basis being that performing high- performed in non-teaching hospitals (23.2% of all SCTs in teaching risk complex surgical procedures in large hospitals will lead to better hospitals were performed in Northeastern region). outcomes.1–5 There is an increasing number of stem cell transplan- Results from the multivariable logistic regression model are tation (SCT) procedures performed in the United States.6,7 So far, summarized in Table 2. In the regression model, following there are no data on regionalization of SCTs into teaching hospitals. adjustment for multiple confounders, when compared to in year The specific aim of the present study is to examine if performance of 2004, SCT procedures were more likely to be performed in SCT procedures has been regionalized to teaching hospitals over a teaching hospitals in year 2010 (odd ratios (OR) = 8.01, 95% 7-year-period from 2004 to 2010. The present study tests the confidence interval (CI) = 1.22–52.84, P = 0.03) and in year 2008 hypothesis that during the latter years of the study period, more SCT (OR = 7.84, 95% CI = 1.19–51.46, P = 0.03). SCT procedures were procedures were performed in teaching hospitals as opposed to in also more likely to be performed in teaching hospitals in the non-teaching hospitals. A secondary aim is to examine if there is Northeastern region (OR = 54.91, 95% CI = 18.81–160.28, Po0.001) geographic variations in regionalization. compared to in Western regions. The Nationwide Inpatient Sample (NIS) for the years 2004–2010 Findings from our analysis indicate that a vast majority of SCTs was used. The NIS is the largest all-payer hospitalization database are performed in teaching hospitals. During the later years of the in the United States.8 NIS is sponsored by the Healthcare Cost study period, was increasing regionalization of SCTs into teaching and Utilization Project (HCUP) of the Agency for Healthcare hospitals. These findings are consistent with prior studies that Research and Quality (AHRQ). A data user agreement was have looked at the phenomenon of regionalization across multiple 1–5 completed with HCUP–AHRQ before obtaining the data. The complex surgical procedures. A further finding in our study is present study was granted Institutional Review Board Exempt that, there seems to be a geographic variation in regionalization. status as a large publicly available database was used. Regionalization appears to be significantly pronounced in the All patients who underwent a SCT procedure (see Table 1 for the Northeastern regions compared to other geographic areas. This ICD-9-CM procedure codes9 that were used to select cases) during has not been demonstrated in prior studies. the years 2004–2010 were selected for analysis. Performance of Before embarking on regionalization of SCTs into teaching SCT in a teaching hospital was the outcome and used as a binomial hospitals, we should assess the relative merits and trade-offs for variable. As per the NIS, a teaching hospital has an American regionalizing complex procedures (and associated medical care) Medical Association approved residency program, is a member of such as SCT. Although there is unequivocal support that the Council of Teaching Hospitals or has a ratio of full-time performing high-risk complex surgical procedures in high-volume equivalent interns and residents to beds of 0.25 or higher.8 The specialized hospitals or teaching hospitals will lead to better primary independent variables were: year of procedure and outcomes10,11 in terms of lower costs, lower mortality rates and geographic region (Northeast, Midwest, South or Western regions lower complications, these benefits need to be confirmed over of the United States) of hospital location. A multivariable logistic time and for SCT procedures in specific. A major concern regarding regression models was used to examine the association between regionalization is that indiscriminate concentration of procedures the outcome and independent variables. The confounding effects into select centers will overburden these centers. The underlying of age, sex, race, type of admission, type of SCT and comorbid framework that high-volume hospitals spread their fixed costs for burden were adjusted in the multivariable regression model. Year providing surgical care over a larger number of procedures and 2004 was used as the reference as this would enable us to examine thereby achieve economies of scale may not hold true in the long if regionalization occurred in the latter parts of the study period. term.12 Over time, these centers could be overburdened resulting The effect of clustering of outcomes within hospitals was adjusted. in long waiting times for patients to undergo SCTs. This may All statistical tests were two sided and a P-value of o0.05 was compromise the delivery of timely care to patients who will need deemed to be statistically significant. All statistical analyses were SCTs. Burdening the high-volume centers with large numbers of performed using SAS Version 9.3 software (SAS Inc., Cary, NC, USA) complex procedures could also precipitate a scenario wherein they and SAS callable SUDAAN Version 11.0.1 (Research Triangle are unable to provide low-risk procedures and health care services, Institute, Research Triangle Park, NC, USA). which could have lasting financial implications and threaten the During the study period, a total of 101 462 SCTs were very viability of these high-volume centers. A similar phenomenon performed. Of these, 97 301 were performed in teaching hospitals could also occur in low-volume non-teaching hospitals as they and 2954 in non-teaching hospitals. Status of hospital was withdraw from providing complex medical care/procedures. unknown for 1207 SCT procedures. Performance of SCTs in Specialist surgeons/providers may want to leave these hospitals teaching hospitals over the time period varied from a low of 91.5% as opportunities to provide highly specialized care dwindle. Letter to the Editor 1005

Table 1. Characteristics of patients having stem cell transplant procedures

Characteristic Response Non-teaching Teaching hospitals: hospitals: N = 2954 N = 97 301

Type of SCT (ICD-9-CM procedure code)—Note: a Bone marrow transplant, not specified 0.6% 0.2% patient may have one more of the SCTs performed. (41.00) Consequently, the individual % values will not add up to 100% Autologous bone marrow transplant 7.8% 4.5% without purging (41.01) Allogeneic bone marrow transplant 1.8% 0.6% with purging (41.02) Allogeneic bone marrow transplant 6.8% 7.5% without purging (41.03) Autologous hematopoietic stem cell 50.6% 52.2% transplant without purging (41.04) Allogeneic hematopoietic stem cell 26.8% 28.8% transplant without purging (41.05) Cord blood stem cell transplant (41.06) 3.2% 3.7% Autologous hematopoietic stem cell 1.9% 1.7% transplant with purging (41.07) Allogeneic hematopoietic stem cell DS 1.3% transplant (41.08) Autologous bone marrow transplant DS 0.1% with purging (41.09) Sex Male 53.9% 58.9% Female 46.1% 41.1% Type of admission Emergency/urgent 19% 20.9% Elective 81% 79.1% Racea White 78.6% 73% Black 7.4% 8.8% Hispanic 9.8% 10.9% Asian/Pacific Islander 1.8% 3% Native American 0.6% 0.5% Other races 1.7% 3.7% Comorbid burdenb 0 22.9% 21.9% 1 33.8% 30.8% 2 21.5% 24.6% 3 11.9% 13.7% 4 6.6% 6% 5 2.2% 2.2% ⩾ 6 1.1% 0.8% Age in years Mean 47.7 years 44.1 years SE of mean 2.56 1.29 Median 51.5 years 49.7 years Hospital region Northeast 0.5% 23.2% Midwest 19.5% 23.8% South 13.4% 31% West 36.6% 21.9% Year of procedure 2004 9.2% 10.3% 2005 19.6% 13.6% 2006 12.2% 8.4% 2007 43.8% 14.3% 2008 2.3% 17.5% 2009 10.8% 17.4% 2010 2% 18.5% Abbreviation: SCT = stem cell transplantation. DS: discharge information is suppressed because individual cell count is ⩽10 (as per data user agreement with HCUP–AHRQ). aInformation on race was available for 2213 patients who had SCTs in non-teaching hospitals and 77 944 patients who had SCTs in teaching hospitals. bComorbid burden was computed by summing the presence of 29 different chronic comorbid conditions. The 29 comorbid conditions were: AIDS, alcohol abuse, deficiency anemias, rheumatoid arthritis/collagen vascular diseases, chronic blood loss anemia, congestive heart failure, chronic pulmonary disease, coagulopathy, depression, diabetes—uncomplicated, diabetes—with chronic complications, drug abuse, hypertension, hypothyroidism, liver disease, lymphoma, fluid and electrolyte disorders, metastatic cancer, neurological disorders, obesity, paralysis, peripheral vascular disorders, psychoses, pulmonary circulatory disorders, renal failure, solid tumor without metastasis, peptic ulcer disease, valvular disease, and weight loss.

Another factor to be taken into account is the travel time for Patients’ local area providers may be reluctant to treat them patients and their families. This could be particularly pronounced because of their lack of familiarity with managing complications for those residing in rural areas, racial minorities and the low- following SCTs. It would be unreasonable to expect these patients income groups.13,14 There is evidence to conclude that utilization to travel to the facilities that provided their medical care to have of health care services reduces as the travel times to health care every post discharge complication treated. Finally, regionalization facilities increase.15 Procedures such as SCT are highly complex and could adversely affect the traditional referral behaviors and patients may suffer from prolonged lengths of stay in hospitals and patterns of health care providers, and could have lasting impact experience perioperative and post discharge complications. on the United States health care system.13

© 2016 Macmillan Publishers Limited Bone Marrow Transplantation (2016) 1004 – 1006 Letter to the Editor 1006 their availability in the NIS database. These are major limitations of Table 2. Odds of stem cell transplant procedures being performed in the present study. a teaching hospital vs non-teaching hospital (estimates from In conclusion, over the study period, there is evidence to multivariable logistic regression analyses) suggest that there is increasing regionalization of SCT procedures Independent variables Model 1 into teaching hospitals. Furthermore, it appears that concentration of procedures into teaching hospitals could also be inﬂuenced by Odds ratio (95% CI) P-value geographic location. This may have implications for access to care.

Year of procedure Year 2010 8.01 (1.22–52.84) 0.03 CONFLICT OF INTEREST Year 2009 1.20 (0.19–7.49) 0.85 The authors declare no conﬂict of interest. Year 2008 7.84 (1.19–51.46) 0.03 – Year 2007 0.51 (0.08 3.15) 0.47 V Allareddy1, S Rampa2, N Martinez-Schlurmann3, Year 2006 0.54 (0.13–2.34) 0.41 4 3,5 – RP Nalliah and V Allareddy Year 2005 0.57 (0.10 3.41) 0.54 1 Year 2004 Reference year Department of Orthodontics, College of Dentistry, The University of Iowa, Iowa City, IA, USA; 2 Geographic region Department of Health Management and Policy, College of Public Northeast region 54.91 (18.81–160.28) o0.0001 Health, University of Nebraska Medical Center, Omaha, NE, USA; Midwest region 0.90 (0.21–3.99) 0.89 3Division of Pediatric Critical Care, Rainbow Babies & South region 0.56 (0.15–2.13) 0.39 Children’s Hospital, Case Western Reserve University West region Reference region School of Medicine, Cleveland, OH, USA; 4University of Michigan, Ann Arbor, MI, USA and Age 5 ’ Each 1-year increase 0.99 (0.97–1.01) 0.29 Pediatric Cardiology, Boston Children s Hospital, Harvard University, Boston, MA, USA Sex E-mail: [email protected] or veerasathpurush- Female 0.80 (0.59–1.09) 0.16 [email protected] Male Reference

Race REFERENCES – White 0.94 (0.68 1.30) 0.70 1 Allareddy V. Is there an increasing regionalization of surgical repair of Other races Reference craniosynostosis procedures into teaching hospitals? implications of regionalization. Cleft Palate Craniofac J 2016; 53:197–202. Type of admission 2 Allareddy V, Ackerman MB, Venugopalan SR, Yadav S, Nanda VS, Nanda R. – Elective 0.95 (0.30 3.01) 0.93 Longitudinal trends in discharge patterns of orthognathic surgeries: is there a Emergency/urgent Reference regionalization of procedures in teaching hospitals? Oral Surg Oral Med Oral Pathol Oral Radiol 2013; 115: 583–588. Comorbid burden 3 Bhattacharyya N, Abemayor E. Combined changing patterns of hospital utilization – Each 1 unit increase 1.01 (0.78 1.32) 0.92 for head and neck cancer care: implications for future care. JAMA Otolaryngol Head Neck Surg 2013; 139: 1043–1047. Type of stem cell transplant procedure 4 Cooperberg MR, Modak S, Konety BR. Trends in regionalization of inpatient – Allogeneic 1.15 (0.77 1.70) 0.49 care for urological malignancies, 1988 to 2002. J Urol 2007; 178: Others Reference 2103–2108 (discussion 2108). Abbreviation: CI = confidence interval. 5 Losina E, Kessler CL, Wright EA, Creel AH, Barrett JA, Fossel AH et al. Geographic diversity of low-volume hospitals in total knee replacement: implications for regionalization policies. Med Care 2006; 44:637–645. 6 Majhail N. CIBMTR slides and reports. http://www.cibmtr.org/ReferenceCenter/ The present study has several limitations attributed to the use SlidesReports/Documents/CIBMTR_Center_Survey_Report.pdf. of secondary hospital discharge database and the cross-sectional 7 Allareddy V, Roy A, Rampa S, Lee MK, Nalliah RP, Allareddy V et al. Outcomes of nature of the study design. The utility and limitations of the NIS stem cell transplant patients with acute respiratory failure requiring mechanical dataset have been well described in literature.7 What we ventilation in the United States. Bone Marrow Transplant 2014; 49: 1278–1286. demonstrated is a temporal association between concentration 8 HCUP Nationwide Inpatient Sample (NIS). Healthcare Cost and Utilization Project of SCTs in teaching hospitals and time. We are unable to (HCUP). 2004 - 2010. Agency for Healthcare Research and Quality: Rockville, MD, USA. determine the actual etiological factors precipitating regionaliza- Available at http://www.hcup-us.ahrq.gov/nisoverview.jsp. Accessed 4 January 2014. fi tion of SCTs into teaching hospitals. At best, we could speculate 9 Centers for Disease Control and Prevention. Classi cation of diseases, functioning, and disability. International Classification of Diseases, Ninth Revision, Clinical that selective referrals into teaching hospitals could be attributed Modification (ICD-9-CM). Available at www.cdc.gov/nchs/icd/icd9cm.htm#ftp. to multiple factors, including but not limited to: increasing Accessed 03 June 2014. awareness of health care providers that large-volume hospitals 10 Birkmeyer JD, Stukel TA, Siewers AE, Goodney PP, Wennberg DE, Lucas FL. are better positioned to deliver high-risk care, pressure from Surgeon volume and operative mortality in the United States. N Engl J Med 2003; insurance carriers, preferences of patients, availability of resources, 349:2117–2127. familiarity with care of patients with complex comorbidities and 11 Luft HS. The relation between surgical volume and mortality: an exploration of long standing existing referral patterns, so on. Future studies must causal factors and alternative models. Med Care 1980; 18: 940–959. focus on identifying factors associated with regionalization, long- 12 Menke TJ, Wray NP. When does regionalization of expensive medical care term effects of regionalization on the health care system, and if save money? Health Serv Manage Res 2001; 14:116–124. regionalization indeed improves outcomes specifically for SCT 13 Birkmeyer JD, Siewers AE, Marth NJ, Goodman DC. Regionalization of high-risk surgery and implications for patient travel times. JAMA 2003; 290: 2703–2708. procedures. For SCT patients, the hematopoietic cell transplant co- 14 Kansagra SM, Curtis LH, Schulman KA. Regionalization of percutaneous morbidity index (HCT-CI) is the gold standard for baseline health. transluminal coronary angioplasty and implications for patient travel distance. This information was not available in the NIS and hence could not JAMA 2004; 292: 1717–1723. be examined. Finally, outcomes such as quality of life and 1-year 15 McGuirk MA, Porell FW. Spatial patterns of hospital utilization: the impact of post discharge survival rate were not examined because of lack of distance and time. Inquiry 1984; 21:84–95.