- Email: [email protected]
Criteria for identification of comprehensive pediatric hospitals and referral regions
CRITERIA FOR IDENTIFICATION OF COMPREHENSIVE PEDIATRIC HOSPITALS AND REFERRAL REGIONS ROBERT K. KANTER, MD, AND FRANKLIN DEXTER, MD, PHD
Objective To identify comprehensive pediatric hospitals on the basis of publicly available data. Study design We developed identification criteria for comprehensive pediatric hospitals, then evaluated the number of hospitals meeting these selection criteria. Criteria for a comprehensive pediatric hospital included pediatric residency accreditation, pediatric inpatient volume, and diversity of pediatric disorders at each hospital. New York State hospital administrative discharge data were analyzed for patients 0 to 14 years of age, excluding neonatal diagnoses.
Results Infants and children (n = 125,588) with 375 different diagnosis-related groups were discharged from 230 hospitals in 2000. Through the use of higher selective criteria (educational accreditation plus both high volume and diversity in the top decile), 11 comprehensive pediatric hospitals were identified. These hospitals serve populations of 1.7 ± 0.3 million (mean ± SD) each, with 8 referral regions throughout the state, collectively providing care for 29% of all pediatric statewide hospitalizations. Conclusions Comprehensive pediatric hospitals serve the population of New York widely and evenly. The ability to identify pediatric hospitals will permit evaluation of the relative quality of care and suggest appropriate regulatory interventions to improve pediatric hospital utilization. (J Pediatr 2005;146:26-9)
ptimal care of children with severe illness and injury requires appropriate organization of hospital resources. Care of the most severe and complex problems is best provided at comprehensive pediatric hospitals.1,2 Referral to a limited number of designated comprehensive hospitals would provide sufficient clinical volume to achieve proficiency in care of high-risk conditions,3 to generate sufficient revenue to support costly services,4 and to avoid redundancy of these resources.5 The lack of an objective method to identify comprehensive pediatric hospitals results in an incomplete understanding of the existing system and may interfere with the use of existing resources.6 Pediatric hospitals identified by national organizations7,8 are based on self-report. Inclusion on such lists may require a membership fee and may reflect hospitals’ marketing interests. Lacking objective definitions, marketing claims about what constitutes a ‘‘real’’ children’s hospital can deteriorate into legal disputes.9 With pediatric hospitals lacking formal status, business interests may threaten patient access to unique regional resources.10 Federal goals for national health care improvement include an increase in the number of states with guidelines that categorize acute care facilities with respect to pediatric personnel and resources.11 One study indicated that accreditation of hospitals for pediatric graduate medical education purposes identified facilities that each tend to care for larger numbers of children than nonaccredited hospitals.12 However, the numerous hospitals identified in this way exceed the local need for comprehensive pediatric hospitals in some regions. From the Department of Pediatrics, We carried out the current study to develop a research method to identify Upstate Medical University, Syracuse, New York; and the Department of comprehensive pediatric hospitals based on publicly available data rather than by self-report.
O
METHODS Characteristics of Hospitals Each hospital’s pediatric inpatient volume was expressed as the annual number of hospital discharges. Diversity of disorders at each hospital was expressed as the probability
DRG PHRR
26
Diagnosis-related group Pediatric hospital referral region
SPARCS
Statewide Planning and Research Cooperative System
Anesthesia, University of Iowa, Iowa City, Iowa. Submitted for publication Dec 16, 2003; last revision received May 12, 2004; accepted Aug 24, 2004. Reprint requests: Robert K. Kanter, MD, Department of Pediatrics, Upstate Medical University, 750 East Adams St, Syracuse, NY 13210. E-mail: [email protected]. 0022-3476/$ - see front matter Copyright ª 2005 Elsevier Inc. All rights reserved. 10.1016/j.jpeds.2004.08.058
that two patients selected at random will have the same federal diagnosis-related group (DRG) and was calculated as the sum of squares of proportions of each DRG.13 Diversity increases as the probability decreases from a maximum of 1 to 0. Educational accreditation of hospitals was defined as sponsorship or major participation in a pediatric residency.14
Patients Data were analyzed for patients 14 years of age and younger. Older patients were excluded because data indicating the survival benefit of pediatric hospital care pertain to younger ages.15,16 Patients with neonatal DRGs were excluded from the study because care of newborn infants is often provided in separate facilities and may be regulated independently from other pediatric care.
Data Data were obtained from the New York Statewide Planning and Research Cooperative System (SPARCS17) for 2000. The data are collected by hospitals for administrative purposes and reported to the state. The data are publicly available, with no individual patient identifiers. Hospitals are required to report the information, and the data are audited according to New York State law. This analysis of publicly available data was considered exempt from institutional review board review.
Analysis Hospitals reporting any pediatric discharges were studied. Diversity of disorders was evaluated for its consistency with other markers of comprehensive pediatric hospitals. Correlation of high diversity with high clinical volume was expressed as the Spearman correlation coefficient. The probability that the slope of the line for this relation differed from 0 was determined. The diversity of disorders was compared for hospitals with and without educational accreditation by the Mann-Whitney test (P < .05 was considered to represent a statistically significant difference). Criteria were considered for identification of comprehensive pediatric hospitals. The criteria included educational accreditation, clinical volume, and diversity of disorders. Residency accreditation is the only existing tool to evaluate hospitals according to uniform national standards for a range of medical and surgical services.6 Volume is an indirect marker consistent with quality.3 Diversity distinguishes comprehensive from specialty hospitals.13 Numbers of hospitals meeting percentile levels of selectivity for combinations of criteria were determined. Pediatric hospital referral regions (PHRRs) were determined on the basis of patient referral patterns identified in this study. The PHRRs were formulated after identifying comprehensive pediatric hospitals by a single set of selective criteria. The county of residency was determined for each child hospitalized at a comprehensive pediatric hospital. For each county, we identified the comprehensive pediatric hospital serving the largest number of children residing in that county.
Counties referring patients to a single comprehensive hospital were considered to form a PHRR. If more than one comprehensive hospital was located in a single county, all the counties referring patients to those comprehensive hospitals were considered to form a PHRR. If a comprehensive pediatric hospital did not attract the largest number of patients from within its own county, then that county was included in a PHRR with the other county serving the largest number of its patients.
RESULTS A total of 125,588 infants and children with 375 different DRGs were hospitalized at 230 hospitals, statewide in 2000. Greater diversity of disorders was correlated with greater numbers of hospital discharges (Figure 1; Spearman r = 20.61, P < .001). The diversity of disorders was greater at hospitals with accredited pediatric residencies than at nonaccredited institutions (Figure 1; median = 0.11 vs 0.58, respectively, P < .05). Among all hospitals, the annual number of pediatric discharges was 1715, 469, and 172 for institutions at the 90th, 75th, and 50th percentiles, respectively. Diversity of diagnoses (the probability that two patients had the same DRG) was 0.053, 0.073, and 0.105, corresponding to 158, 75, and 38 different DRGs per hospital, for institutions at the 90th, 75th, and 50th percentiles, respectively. The numbers of hospitals satisfying both volume and diversity criteria at 50th, 75th, and 90th percentile levels are shown in Figure 2. Forty-two hospitals had educational accreditation. Few hospitals that lack residency accreditation satisfied high volume or diversity criteria, and none met both volume and diversity criteria in the top decile. Eleven hospitals with residency educational accreditation have both volume and diversity in the top decile. These are considered to represent the comprehensive pediatric hospitals for purposes of describing existing regional services. The 11 comprehensive pediatric hospitals identified in this study provided care for 36,119 infants and children (29% of pediatric hospitalizations) in 2000. These hospitals were widely distributed throughout the state. The existing patterns of referrals of children to comprehensive hospitals formed 8 PHRRs (Figure 3). Each hospital served populations of 1.7 ± 0.3 million (mean ± SD). Each region was served by at least one pediatric hospital, with a maximum of three in the New York City region. One percent of hospitalized children resided outside the state.
DISCUSSION Proposed selective criteria for identification of comprehensive pediatric hospitals include educational accreditation and both pediatric volume and diagnostic diversity in the top decile. These criteria identified existing facilities that are widely and evenly distributed in relation to the statewide population while eliminating redundancy in any region. Validity of the proposed criteria can be evaluated in the following ways. Content validity includes face validity (the subjective impression that criteria identify comprehensive
Criteria For Identification Of Comprehensive Pediatric Hospitals And Referral Regions
27
Figure 1. Relation between diversity of disorders and number of discharges at each hospital. Diversity is the probability that two patients chosen at random have the same DRG (= sum of squares of proportions of each DRG). Diversity increases as probability decreases from 1 to 0. Hospitals with (:) and without (s) pediatric residency educational accreditation are indicated. For this relation, Spearman r = 20.61, P < .001. Diversity of disorders was greater at accredited hospitals (P < .05). (Figure 1 is available in color online at www.us.elsevierhealth.com/jpeds)
pediatric hospitals in a reasonable way) and sampling validity (the inclusion of pertinent aspects of the phenomenon being measured). These are both achieved by the study criteria; all 11 hospitals identified are large teaching hospitals with diverse patient populations. It might be argued that some large teaching hospitals were excluded unreasonably. Sampling validity would be stronger if hospital characteristics and personnel necessary for comprehensive services could be better identified from publicly available data. Construct validity involves observations that are consistent with theoretical concepts about the phenomenon being studied. Construct validity is supported by the homogeneity of regional population size served by each of the 11 hospitals (coefficient of variation for regional population served per comprehensive hospital = 17%). Construct validity is also supported by the similarity of regional populations per hospital determined in this study and the population previously found to be necessary to financially support pediatric hospitals.4 Concurrent validity is the extent to which a measure is consistent with an independent measure of the same phenomenon. The lack of a gold standard to identify pediatric hospitals is the reason for undertaking this study. Among the 11 comprehensive pediatric hospitals identified in this study, 9 (Reference 7) or 11 (Reference 6) were independently identified as having a pediatric intensive care unit. One (Reference 7) or 4 (Reference 8) were independently identified as children’s hospitals. Thus, other published identification lists of pediatric hospitals may be incomplete or may involve even more selective criteria than ours. New York State only had one free-standing children’s hospital exclusively caring for 28 Kanter and Dexter
Figure 2. Number of hospitals in New York State meeting both volume and diversity criteria at indicated percentile levels. Results are shown separately for hospitals with (:) and without (s) residency accreditation. (Figure 2 is available in color online at www.us. elsevierhealth.com/jpeds)
Figure 3. Map indicates counties (light lines), pediatric hospital referral regions (bold lines), and comprehensive pediatric hospitals (d) by selective criteria. (Figure 3 is available in color online at www.us.elsevierhealth.com/jpeds)
children. Although the 11 identified hospitals are the busiest pediatric facilities in the state, children only accounted for an average of 16% of their discharges, the remainder being adult medical, surgical, obstetrics, psychiatry, and neonatal patients. Thus, exclusive care of children cannot be used as a criterion for pediatric hospital identification in New York. The single free-standing children’s hospital in the state was identified by our criteria as one of the 11 comprehensive pediatric hospitals. Empirical investigation will be needed to determine how well the proposed method to identify comprehensive pediatric hospitals can be generalized to other states as a health services The Journal of Pediatrics January 2005
research tool. New York has a uniquely large population, with a disproportionately large number of medical schools, only one free-standing children’s hospital, and no for-profit hospitals. Some states cannot satisfy the requirement for an accredited pediatric residency. Rural areas with extremely large distances are lacking in this state sample. Regulatory environments may vary among states. In other respects, New York shares features representative of other states, with a range of large metropolitan, small urban, suburban, and rural areas. The following limitations of the study are recognized. The study analyzed data collected for administrative rather than clinical purposes. The quality of hospital care was not assessed. Specific hospital resources, services, and personnel cannot be identified in hospital discharge databases. Nevertheless, high volume is associated with high quality, and high volume may serve as an indirect marker of hospitals with high capabilities until better measures of quality are available.18 No SPARCS data are available regarding children residing in New York who are admitted to hospitals outside the state. PHRRs, determined from patterns of referral to the subgroup of facilities identified as comprehensive pediatric hospitals, are defined functionally as those geographic subunits surrounding a hospital used most by the population in those areas.19 Previous descriptions of hospital referral regions have been related to complete samples of hospitals and the small areas (towns or zip codes) from which they draw patients.19,20 For some marketing research purposes, competition in overlapping small areas has been a focus of study.21 Our method differs from previous reports in that only a subset of hospitals is considered in formulating regions and counties rather than smaller areas are considered the smallest unit of analysis. Geographic units of analysis smaller than counties would result in many unclassifiable areas with no patients referred to a comprehensive hospital. Referral regions for the subset of comprehensive hospitals, as determined in this study, serve larger populations and areas than the referral regions determined for each facility in a complete sample of all hospitals. The PHRRs determined in this study are generally similar to health service areas defined by the State of New York without regard to age (compared with SPARCS health service areas,17 only 8 of 62 counties are rearranged to describe PHRRs, with the same total number of statewide regions). This study proposes the identification of a subset of existing hospitals as comprehensive regional pediatric resources, based on simple and selective administrative criteria. Further investigation to evaluate the proposed criteria in a broader national sample is warranted. The ability to identify comprehensive pediatric hospitals will permit evaluation of the relative quality of care at such facilities and might
suggest regulatory interventions to improve their appropriate utilization.
REFERENCES 1. American Academy of Pediatrics, Committee on Pediatric Emergency Medicine, American College of Critical Care Medicine, Society of Critical Care Medicine, Pediatric Section. Consensus report for regionalization of services for critically ill or injured children. Pediatrics 2000;105:152-5. 2. American Academy of Pediatrics, Committee on Pediatric Emergency Medicine. Guidelines for pediatric emergency care facilities. Pediatrics 1995; 96:526-37. 3. Birkmeyer JD, Siewers AE, Finlayson EVA, Stukel TA, Lucas FL, Batista I, et al. Hospital volume and surgical mortality in the United States. N Engl J Med 2002;346:1128-37. 4. Billi JE, Wise CG, Bills EA, Mitchell RL. Potential effects of managed care on specialty practice at a university medical center. N Engl J Med 1995; 333:979-83. 5. Richardson DK, Reed K, Cutler C, Boardman RC, Goodman K, Moynihan T, et al. Perinatal regionalization versus hospital competition: the Hartford example. Pediatrics 1995;96:417-23. 6. Kanter RK. Regional variation in child mortality at hospitals lacking a pediatric intensive care unit. Crit Care Med 2002;30:94-9. 7. American Hospital Association. AHA Guide 2000-2001, Health Forum, One North Franklin, Chicago, Ill, 2000. 8. National Association for Children’s Hospitals and Related Interests. Children’s Hospital Profiles. Alexandria, Va, Internet access: 10/21/03. www.children’shospital.net/index.htm. 9. Bellandi D. Stand-alone vs unit: a battle is on over what constitutes a children’s hospital. Mod Healthcare 1997;27:68. 10. Becker C. Antitrust action averted: agreement with Pennsylvania attorney general allows UPMC merger to proceed. Mod Healthcare 2001; 31:14-5. 11. US Department of Health and Human Services. Healthy People 2010: understanding and improving health. 2nd edition. Washington, DC; US Government Printing Office, November 2000. Goal 1-14b. p. 1-33 to 1-34. 12. Kanter RK, Egan M. Utilization of pediatric hospitals in New York State. Pediatrics 2003;111:1068-71. 13. Dayhoff DA, Cromwell J. Measuring differences and similarities in hospital caseloads: a conceptual and empirical analysis. Health Serv Res 1993; 28:293. 14. American Medical Association. Graduate Medical Education Directory, 1999-2000. Chicago, Ill: American Medical Association. 15. Hannan EL, Racz M, Kavey RE, Quaegebeur JM, Williams R. Pediatric cardiac surgery: the effect of hospital and surgeon volume on inhospital mortality. Pediatrics 1998;101:963-9. 16. Tilford JM, Simpson PM, Green JW, Lensing S, Fiser DH. Volume outcome relationship in pediatric intensive care units. Pediatrics 2000;106: 289-92. 17. New York State Department of Health. New York Statewide Planning and Research Cooperative System, Albany, 2001. 18. Epstein AM. Volume and outcome: it is time to move ahead. N Engl J Med 2002;346:1161-3. 19. Wennberg J, Gittelsohn A. Small area variations in health care delivery. Science 1973;182:1102-8. 20. Guagliardo MF, Jablonski KA, Joseph JG, Goodman DC. Do pediatric hospitalizations have a unique geography? BMC Health Serv Res 2004;4:2. 21. Sohn MW. A relational approach to measuring competition among hospitals. Health Serv Res 2002;37:457-82.
Criteria For Identification Of Comprehensive Pediatric Hospitals And Referral Regions
29
Objective To identify comprehensive pediatric hospitals on the basis of publicly available data. Study design We developed identification criteria for comprehensive pediatric hospitals, then evaluated the number of hospitals meeting these selection criteria. Criteria for a comprehensive pediatric hospital included pediatric residency accreditation, pediatric inpatient volume, and diversity of pediatric disorders at each hospital. New York State hospital administrative discharge data were analyzed for patients 0 to 14 years of age, excluding neonatal diagnoses.
Results Infants and children (n = 125,588) with 375 different diagnosis-related groups were discharged from 230 hospitals in 2000. Through the use of higher selective criteria (educational accreditation plus both high volume and diversity in the top decile), 11 comprehensive pediatric hospitals were identified. These hospitals serve populations of 1.7 ± 0.3 million (mean ± SD) each, with 8 referral regions throughout the state, collectively providing care for 29% of all pediatric statewide hospitalizations. Conclusions Comprehensive pediatric hospitals serve the population of New York widely and evenly. The ability to identify pediatric hospitals will permit evaluation of the relative quality of care and suggest appropriate regulatory interventions to improve pediatric hospital utilization. (J Pediatr 2005;146:26-9)
ptimal care of children with severe illness and injury requires appropriate organization of hospital resources. Care of the most severe and complex problems is best provided at comprehensive pediatric hospitals.1,2 Referral to a limited number of designated comprehensive hospitals would provide sufficient clinical volume to achieve proficiency in care of high-risk conditions,3 to generate sufficient revenue to support costly services,4 and to avoid redundancy of these resources.5 The lack of an objective method to identify comprehensive pediatric hospitals results in an incomplete understanding of the existing system and may interfere with the use of existing resources.6 Pediatric hospitals identified by national organizations7,8 are based on self-report. Inclusion on such lists may require a membership fee and may reflect hospitals’ marketing interests. Lacking objective definitions, marketing claims about what constitutes a ‘‘real’’ children’s hospital can deteriorate into legal disputes.9 With pediatric hospitals lacking formal status, business interests may threaten patient access to unique regional resources.10 Federal goals for national health care improvement include an increase in the number of states with guidelines that categorize acute care facilities with respect to pediatric personnel and resources.11 One study indicated that accreditation of hospitals for pediatric graduate medical education purposes identified facilities that each tend to care for larger numbers of children than nonaccredited hospitals.12 However, the numerous hospitals identified in this way exceed the local need for comprehensive pediatric hospitals in some regions. From the Department of Pediatrics, We carried out the current study to develop a research method to identify Upstate Medical University, Syracuse, New York; and the Department of comprehensive pediatric hospitals based on publicly available data rather than by self-report.
O
METHODS Characteristics of Hospitals Each hospital’s pediatric inpatient volume was expressed as the annual number of hospital discharges. Diversity of disorders at each hospital was expressed as the probability
DRG PHRR
26
Diagnosis-related group Pediatric hospital referral region
SPARCS
Statewide Planning and Research Cooperative System
Anesthesia, University of Iowa, Iowa City, Iowa. Submitted for publication Dec 16, 2003; last revision received May 12, 2004; accepted Aug 24, 2004. Reprint requests: Robert K. Kanter, MD, Department of Pediatrics, Upstate Medical University, 750 East Adams St, Syracuse, NY 13210. E-mail: [email protected]. 0022-3476/$ - see front matter Copyright ª 2005 Elsevier Inc. All rights reserved. 10.1016/j.jpeds.2004.08.058
that two patients selected at random will have the same federal diagnosis-related group (DRG) and was calculated as the sum of squares of proportions of each DRG.13 Diversity increases as the probability decreases from a maximum of 1 to 0. Educational accreditation of hospitals was defined as sponsorship or major participation in a pediatric residency.14
Patients Data were analyzed for patients 14 years of age and younger. Older patients were excluded because data indicating the survival benefit of pediatric hospital care pertain to younger ages.15,16 Patients with neonatal DRGs were excluded from the study because care of newborn infants is often provided in separate facilities and may be regulated independently from other pediatric care.
Data Data were obtained from the New York Statewide Planning and Research Cooperative System (SPARCS17) for 2000. The data are collected by hospitals for administrative purposes and reported to the state. The data are publicly available, with no individual patient identifiers. Hospitals are required to report the information, and the data are audited according to New York State law. This analysis of publicly available data was considered exempt from institutional review board review.
Analysis Hospitals reporting any pediatric discharges were studied. Diversity of disorders was evaluated for its consistency with other markers of comprehensive pediatric hospitals. Correlation of high diversity with high clinical volume was expressed as the Spearman correlation coefficient. The probability that the slope of the line for this relation differed from 0 was determined. The diversity of disorders was compared for hospitals with and without educational accreditation by the Mann-Whitney test (P < .05 was considered to represent a statistically significant difference). Criteria were considered for identification of comprehensive pediatric hospitals. The criteria included educational accreditation, clinical volume, and diversity of disorders. Residency accreditation is the only existing tool to evaluate hospitals according to uniform national standards for a range of medical and surgical services.6 Volume is an indirect marker consistent with quality.3 Diversity distinguishes comprehensive from specialty hospitals.13 Numbers of hospitals meeting percentile levels of selectivity for combinations of criteria were determined. Pediatric hospital referral regions (PHRRs) were determined on the basis of patient referral patterns identified in this study. The PHRRs were formulated after identifying comprehensive pediatric hospitals by a single set of selective criteria. The county of residency was determined for each child hospitalized at a comprehensive pediatric hospital. For each county, we identified the comprehensive pediatric hospital serving the largest number of children residing in that county.
Counties referring patients to a single comprehensive hospital were considered to form a PHRR. If more than one comprehensive hospital was located in a single county, all the counties referring patients to those comprehensive hospitals were considered to form a PHRR. If a comprehensive pediatric hospital did not attract the largest number of patients from within its own county, then that county was included in a PHRR with the other county serving the largest number of its patients.
RESULTS A total of 125,588 infants and children with 375 different DRGs were hospitalized at 230 hospitals, statewide in 2000. Greater diversity of disorders was correlated with greater numbers of hospital discharges (Figure 1; Spearman r = 20.61, P < .001). The diversity of disorders was greater at hospitals with accredited pediatric residencies than at nonaccredited institutions (Figure 1; median = 0.11 vs 0.58, respectively, P < .05). Among all hospitals, the annual number of pediatric discharges was 1715, 469, and 172 for institutions at the 90th, 75th, and 50th percentiles, respectively. Diversity of diagnoses (the probability that two patients had the same DRG) was 0.053, 0.073, and 0.105, corresponding to 158, 75, and 38 different DRGs per hospital, for institutions at the 90th, 75th, and 50th percentiles, respectively. The numbers of hospitals satisfying both volume and diversity criteria at 50th, 75th, and 90th percentile levels are shown in Figure 2. Forty-two hospitals had educational accreditation. Few hospitals that lack residency accreditation satisfied high volume or diversity criteria, and none met both volume and diversity criteria in the top decile. Eleven hospitals with residency educational accreditation have both volume and diversity in the top decile. These are considered to represent the comprehensive pediatric hospitals for purposes of describing existing regional services. The 11 comprehensive pediatric hospitals identified in this study provided care for 36,119 infants and children (29% of pediatric hospitalizations) in 2000. These hospitals were widely distributed throughout the state. The existing patterns of referrals of children to comprehensive hospitals formed 8 PHRRs (Figure 3). Each hospital served populations of 1.7 ± 0.3 million (mean ± SD). Each region was served by at least one pediatric hospital, with a maximum of three in the New York City region. One percent of hospitalized children resided outside the state.
DISCUSSION Proposed selective criteria for identification of comprehensive pediatric hospitals include educational accreditation and both pediatric volume and diagnostic diversity in the top decile. These criteria identified existing facilities that are widely and evenly distributed in relation to the statewide population while eliminating redundancy in any region. Validity of the proposed criteria can be evaluated in the following ways. Content validity includes face validity (the subjective impression that criteria identify comprehensive
Criteria For Identification Of Comprehensive Pediatric Hospitals And Referral Regions
27
Figure 1. Relation between diversity of disorders and number of discharges at each hospital. Diversity is the probability that two patients chosen at random have the same DRG (= sum of squares of proportions of each DRG). Diversity increases as probability decreases from 1 to 0. Hospitals with (:) and without (s) pediatric residency educational accreditation are indicated. For this relation, Spearman r = 20.61, P < .001. Diversity of disorders was greater at accredited hospitals (P < .05). (Figure 1 is available in color online at www.us.elsevierhealth.com/jpeds)
pediatric hospitals in a reasonable way) and sampling validity (the inclusion of pertinent aspects of the phenomenon being measured). These are both achieved by the study criteria; all 11 hospitals identified are large teaching hospitals with diverse patient populations. It might be argued that some large teaching hospitals were excluded unreasonably. Sampling validity would be stronger if hospital characteristics and personnel necessary for comprehensive services could be better identified from publicly available data. Construct validity involves observations that are consistent with theoretical concepts about the phenomenon being studied. Construct validity is supported by the homogeneity of regional population size served by each of the 11 hospitals (coefficient of variation for regional population served per comprehensive hospital = 17%). Construct validity is also supported by the similarity of regional populations per hospital determined in this study and the population previously found to be necessary to financially support pediatric hospitals.4 Concurrent validity is the extent to which a measure is consistent with an independent measure of the same phenomenon. The lack of a gold standard to identify pediatric hospitals is the reason for undertaking this study. Among the 11 comprehensive pediatric hospitals identified in this study, 9 (Reference 7) or 11 (Reference 6) were independently identified as having a pediatric intensive care unit. One (Reference 7) or 4 (Reference 8) were independently identified as children’s hospitals. Thus, other published identification lists of pediatric hospitals may be incomplete or may involve even more selective criteria than ours. New York State only had one free-standing children’s hospital exclusively caring for 28 Kanter and Dexter
Figure 2. Number of hospitals in New York State meeting both volume and diversity criteria at indicated percentile levels. Results are shown separately for hospitals with (:) and without (s) residency accreditation. (Figure 2 is available in color online at www.us. elsevierhealth.com/jpeds)
Figure 3. Map indicates counties (light lines), pediatric hospital referral regions (bold lines), and comprehensive pediatric hospitals (d) by selective criteria. (Figure 3 is available in color online at www.us.elsevierhealth.com/jpeds)
children. Although the 11 identified hospitals are the busiest pediatric facilities in the state, children only accounted for an average of 16% of their discharges, the remainder being adult medical, surgical, obstetrics, psychiatry, and neonatal patients. Thus, exclusive care of children cannot be used as a criterion for pediatric hospital identification in New York. The single free-standing children’s hospital in the state was identified by our criteria as one of the 11 comprehensive pediatric hospitals. Empirical investigation will be needed to determine how well the proposed method to identify comprehensive pediatric hospitals can be generalized to other states as a health services The Journal of Pediatrics January 2005
research tool. New York has a uniquely large population, with a disproportionately large number of medical schools, only one free-standing children’s hospital, and no for-profit hospitals. Some states cannot satisfy the requirement for an accredited pediatric residency. Rural areas with extremely large distances are lacking in this state sample. Regulatory environments may vary among states. In other respects, New York shares features representative of other states, with a range of large metropolitan, small urban, suburban, and rural areas. The following limitations of the study are recognized. The study analyzed data collected for administrative rather than clinical purposes. The quality of hospital care was not assessed. Specific hospital resources, services, and personnel cannot be identified in hospital discharge databases. Nevertheless, high volume is associated with high quality, and high volume may serve as an indirect marker of hospitals with high capabilities until better measures of quality are available.18 No SPARCS data are available regarding children residing in New York who are admitted to hospitals outside the state. PHRRs, determined from patterns of referral to the subgroup of facilities identified as comprehensive pediatric hospitals, are defined functionally as those geographic subunits surrounding a hospital used most by the population in those areas.19 Previous descriptions of hospital referral regions have been related to complete samples of hospitals and the small areas (towns or zip codes) from which they draw patients.19,20 For some marketing research purposes, competition in overlapping small areas has been a focus of study.21 Our method differs from previous reports in that only a subset of hospitals is considered in formulating regions and counties rather than smaller areas are considered the smallest unit of analysis. Geographic units of analysis smaller than counties would result in many unclassifiable areas with no patients referred to a comprehensive hospital. Referral regions for the subset of comprehensive hospitals, as determined in this study, serve larger populations and areas than the referral regions determined for each facility in a complete sample of all hospitals. The PHRRs determined in this study are generally similar to health service areas defined by the State of New York without regard to age (compared with SPARCS health service areas,17 only 8 of 62 counties are rearranged to describe PHRRs, with the same total number of statewide regions). This study proposes the identification of a subset of existing hospitals as comprehensive regional pediatric resources, based on simple and selective administrative criteria. Further investigation to evaluate the proposed criteria in a broader national sample is warranted. The ability to identify comprehensive pediatric hospitals will permit evaluation of the relative quality of care at such facilities and might
suggest regulatory interventions to improve their appropriate utilization.
REFERENCES 1. American Academy of Pediatrics, Committee on Pediatric Emergency Medicine, American College of Critical Care Medicine, Society of Critical Care Medicine, Pediatric Section. Consensus report for regionalization of services for critically ill or injured children. Pediatrics 2000;105:152-5. 2. American Academy of Pediatrics, Committee on Pediatric Emergency Medicine. Guidelines for pediatric emergency care facilities. Pediatrics 1995; 96:526-37. 3. Birkmeyer JD, Siewers AE, Finlayson EVA, Stukel TA, Lucas FL, Batista I, et al. Hospital volume and surgical mortality in the United States. N Engl J Med 2002;346:1128-37. 4. Billi JE, Wise CG, Bills EA, Mitchell RL. Potential effects of managed care on specialty practice at a university medical center. N Engl J Med 1995; 333:979-83. 5. Richardson DK, Reed K, Cutler C, Boardman RC, Goodman K, Moynihan T, et al. Perinatal regionalization versus hospital competition: the Hartford example. Pediatrics 1995;96:417-23. 6. Kanter RK. Regional variation in child mortality at hospitals lacking a pediatric intensive care unit. Crit Care Med 2002;30:94-9. 7. American Hospital Association. AHA Guide 2000-2001, Health Forum, One North Franklin, Chicago, Ill, 2000. 8. National Association for Children’s Hospitals and Related Interests. Children’s Hospital Profiles. Alexandria, Va, Internet access: 10/21/03. www.children’shospital.net/index.htm. 9. Bellandi D. Stand-alone vs unit: a battle is on over what constitutes a children’s hospital. Mod Healthcare 1997;27:68. 10. Becker C. Antitrust action averted: agreement with Pennsylvania attorney general allows UPMC merger to proceed. Mod Healthcare 2001; 31:14-5. 11. US Department of Health and Human Services. Healthy People 2010: understanding and improving health. 2nd edition. Washington, DC; US Government Printing Office, November 2000. Goal 1-14b. p. 1-33 to 1-34. 12. Kanter RK, Egan M. Utilization of pediatric hospitals in New York State. Pediatrics 2003;111:1068-71. 13. Dayhoff DA, Cromwell J. Measuring differences and similarities in hospital caseloads: a conceptual and empirical analysis. Health Serv Res 1993; 28:293. 14. American Medical Association. Graduate Medical Education Directory, 1999-2000. Chicago, Ill: American Medical Association. 15. Hannan EL, Racz M, Kavey RE, Quaegebeur JM, Williams R. Pediatric cardiac surgery: the effect of hospital and surgeon volume on inhospital mortality. Pediatrics 1998;101:963-9. 16. Tilford JM, Simpson PM, Green JW, Lensing S, Fiser DH. Volume outcome relationship in pediatric intensive care units. Pediatrics 2000;106: 289-92. 17. New York State Department of Health. New York Statewide Planning and Research Cooperative System, Albany, 2001. 18. Epstein AM. Volume and outcome: it is time to move ahead. N Engl J Med 2002;346:1161-3. 19. Wennberg J, Gittelsohn A. Small area variations in health care delivery. Science 1973;182:1102-8. 20. Guagliardo MF, Jablonski KA, Joseph JG, Goodman DC. Do pediatric hospitalizations have a unique geography? BMC Health Serv Res 2004;4:2. 21. Sohn MW. A relational approach to measuring competition among hospitals. Health Serv Res 2002;37:457-82.
Criteria For Identification Of Comprehensive Pediatric Hospitals And Referral Regions
29