- Email: [email protected]
Post-acute care for children with special health care needs
Disability and Health Journal xxx (2017) 1e9
Contents lists available at ScienceDirect
Disability and Health Journal journal homepage: www.disabilityandhealthjnl.com
Post-acute care for children with special health care needs Charles D. Phillips a, *, Chau Truong b, Hye-Chung Kum a, Obioma Nwaiwu c, Robert Ohsfeldt a a
Texas A&M Health Science Center, School of Public Health, Department of Health Policy and Management, USA University of Texas, School of Public Health, Department of Management, Policy, and Community Health, USA c University of Arkansas for Medical Sciences, School of Medicine, Department of Family Medicine, USA b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 12 February 2017 Received in revised form 25 August 2017 Accepted 31 August 2017
Background: Almost all studies of post-acute care (PAC) focus on older persons, frequently those suffering from chronic health problems. Some research is available on PAC for the pediatric population in general. However, very few studies focus on PAC services for children with special health care needs (SHCN). Objective: To investigate factors affecting the provision of PAC to children with SHCN. Methods: Pooled cross-sectional data from Texas Department of State Health Services hospital discharge database from 2011-2014 were analyzed. Publicly available algorithms identified chronic conditions, complex chronic conditions, and the principal problem leading to hospitalization. Analysis involved estimating two logistic regressions, with clustered robust standard errors, concerning the likelihood of receiving PAC and where that PAC was delivered. Models included patient characteristics and conditions, as well as hospital characteristics and location. Results: Only 5.8 percent of discharges for children with SHCN resulted in the provision of PAC. Twothirds of PAC was provided in a health care facility (HCF). Severity of illness and the number of complex chronic conditions, though not the number of chronic problems, made PAC more likely. Patient demographics had no effect on PAC decisions. Hospital type and location also affected PAC decisionmaking. Conclusions: PAC was provided to relatively few children with SHCN, which raises questions concerning the potential underutilization of PAC for children with SHCN. Also, the provision of most PAC in a HCF (66%) seems at odds with professional judgment and family preferences indicating that health care for children with SHCN is best provided in the home. © 2017 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Keywords: Children with special health care needs Post-acute care Disabilities Child and maternal health
Background Post-acute care for older persons has been the focus of considerable research. In part this emphasis stems from the prevalence of PAC for older adults. In 2008, estimates indicate that 38.7 percent of Medicare recipients received post-acute care.1 In addition, PAC expenditures have proved to be the fastest growing major health care spending category.2 Spending on Medicare PAC had risen to $635 billion by 2011.1 In contrast, Jay Berry and his colleagues' recent work focused on
* Corresponding author. Texas A&M Health Science Center, School of Public Health, Adriance Lab Road, Department of Health Policy and Management, College Station, TX, 78543-1266, USA. E-mail address: [email protected] (C.D. Phillips).
an area of post-acute care that they correctly indicate has historically received little attention e hospital discharges for those under 21 years of age.3 Their research involved a sample of over two million 2012 discharges from hospitals in 44-states. Their effort provided an important and much needed foundation for additional analyses of pediatric hospital discharges and of the use of postdischarge services among children and youth. In that sample, just over six percent of discharges resulted in the provision of postacute services. Of that 6.1 percent who received PAC, just 17.6 percent received those services in a health care facility (HCF), while the remainder (82.4%) received home health care (HHC) services. The multivariate analyses in that study indicated how individual characteristics (race, age), the number of chronic diseases, diagnoses of complex chronic conditions, source of payment (Medicaid), the type of hospital (children's), as well as state and
http://dx.doi.org/10.1016/j.dhjo.2017.08.010 1936-6574/© 2017 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/ ).
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
2
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
region of the nation significantly affected the use of different postdischarge services.3 This research complements that earlier effort. However, it approaches the analyses of PAC for children or youth differently. This research focused on post-discharge dispositions for an especially vulnerable pediatric population, children with special health care needs (SHCN). As the 20011/12 National Survey of Children with Special Health Care Needs estimated, 14.6 million children or youth under 18 years of age in the US faced health challenges requiring special health care services. This is almost one of every five (19.8%) persons under 18.4 This population presents special challenges to all elements of healthcare.5e7 They need access to a relatively wide variety of health care services and absorb a disproportionate share of health care expenditures.8e12 In 2009, for example, children in Texas's Early and Periodic Screening Diagnostic and Treatment (EPSDT) program, which is populated by children with chronic conditions, averaged Medicaid expenditures over 18 times ($33,628 vs. $1,834) that of the average Medicaid expenditure for a child or youth in Texas.13 As important as such information is, these prevalence and expenditure figures fail to include an important element in the lives of children with SHCN. Families bear much of the responsibility for care of children with SHCN and provide the equivalent of billions of dollars of family care to these children.14 The burden of that care can at times be overwhelming, and the potential adverse effects of that responsibility on family members and family life can be devastating.15 While family care, frequently called informal care, is an important aspect of post-discharge services, the research presented here will focus more narrowly on the distribution of postdischarge services provided by health care professionals (formal care). Like many issues in health services research, elaborate theoretical frameworks have not driven investigations into the provision of formal PAC services.16 Instead, previous research on PAC provides a basic conceptual model for guidance, identifying factors associated with PAC use in previous analyses of hospital discharge data. A graphical representation of the basic model used in this research appears in Fig. 1. The elements of the model that focus on those variables associated with PAC are relatively standard in such analyses.17e19 This effort, however, focuses on children with SHCN, which differs from other populations in which PAC has been investigated. Therefore, these analyses elaborate on the basic model found in earlier research by using more finely-grained information on the number of complex chronic conditions, condition severity, as well as categories of presenting conditions. The model, unlike others, also allows for the investigation for the effect of time on PAC decisions. In addition, the dependent variables representing PAC decisions in this research differ from that in other analyses. Other efforts consider PAC best defined as a single dependent variable with separate categories for no services and for each setting where PAC is provided. This type of analysis results in separate estimates for receiving no services versus receiving services in a particular setting (e.g., likelihood of no services versus home care or the likelihood of no services versus PAC in an HCF). This strategy was used by earlier analyses of PAC for children and youth.3 However, the analyses here conceptualize PAC provision as a two-stage, rather than one-stage, decision process. These two decisions are: (1) whether any type of PAC will be provided, and (2) for those receiving PAC, where that PAC will be provided. This approach is like that used by Gage and her colleagues in their analyses of PAC for older persons and allows one to investigate directly any important differences between those children receiving PAC at home or those children receiving PAC in a health
care facility.20 This formulation results in two questions concerning formal post-discharge services provided to this vulnerable population. What affects the likelihood that an acute care discharge involving a child or youth with SHCN will result in the provision of formal PAC? For those discharges where children or youth with SHCN receive formal post-discharge services, what differentiates between those discharges receiving HHC and those receiving services in a HCF?
Methods Data The Texas Department of State Health Services maintains public use data files of all hospital discharges. From those data, a pooled cross-sectional database of hospital discharges for those discharges involving patients under the age of 18 for three years (10/2011-3/ 2014) were extracted.21 Of these 1,688,454 discharges for children or youth under the age of eighteen, 476,144 (28.2%) involved a child or youth with at least one chronic condition. After excluding those discharges for children without a chronic condition, those who died, or those who discharged against medical advice, a total of 472,825 discharges involving children or youth with a chronic condition remained. We identified these discharges as discharges involving children with SHCN. All aspects of this project were approved by the Institutional Review Board at Texas A&M University. Measurement Dependent variables The dependent variables in this effort focused on the services identified in the discharge record. They focus on two very clear distinctions in post-discharge services for children with SHCN. As noted earlier, the dependent variable for the first multivariate model defined two categories: those discharges with no formal services and those discharges involving any type of formal services. The second dependent variable focused only on those discharges that involved services. It differentiated between the sites where care was received after discharge, an important distinction for the patient, as well as informal and formal caregivers.22 This dependent variable differentiated between those discharges involving children and youth discharged to home with HHC (i.e., home health or private duty nursing) and those who received care after discharge in a HCF of some type (i.e., rehabilitation facility, nursing facility, residential hospice, mental health facility, specialty hospital, or another acute care facility).23 Independent variables Any discharge involving a child or youth with at least one diagnosis of a chronic condition was included in our study database as a discharge involving a child or youth with special health care needs.24 The presence of a chronic condition was indicated if the principal diagnosis or any of the other diagnoses listed in the discharge record (up to 24) were included in the list of chronic conditions developed in the Healthcare Costs and Utilization Project.25 Complex chronic conditions were identified using Feudtner and his colleagues pediatric complex chronic condition system.26 The standard demographic characteristics of age, sex, and race/ ethnicity were provided in the Texas discharge data. The ages under
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
3
Fig. 1. Model underlying the Investigation of PAC provision for Children with SHCN.
18 were grouped into six categories in the discharge records. The principal diagnosis listed in the discharge data was converted, using the HCUP system, into a set of binary indicators indicating the type of presenting problem that resulted in hospitalization.25 The illness severity measure (mild, moderate, severe, extreme), based on AP-DRGs was also available in the discharge database.23 Length of the hospital stay was identified in the discharge record as the number of hospital days. Patients were identified as having had surgery if the principal procedure code for the discharge indicated a surgical procedure and the discharge involved charges for use of operating room services. Hospitals with membership in The Council of Teaching Hospitals were coded as teaching hospitals, and those with membership in The National Association of Children's Hospitals and Related Institutions were coded as children's hospitals. Rural Urban Classification Codes, version 2103, were used to construct a binary variable indicating that the hospital operated in a large metropolitan area.27 Binary variables representing nine of the ten Texas Public Health Regions were also included in the models to adjust for any regional differences in the use of postdischarge services. Analysis strategy All analyses were performed using STATA 14.28 Descriptive statistics for all variables were calculated. No variable used in the analysis had greater than one percent missing data. Population data such as those used in this research are not collected, especially in large scale administrative databases, without error. To account for this reality, the data were treated as a sample from a series of populations, each with differing amounts of processing error. This conceptualization makes the use of standard inferential statistical techniques both necessary and appropriate. In the two estimated logistic regression models, all independent
variables were treated as fixed-effects. The multivariate results reported include odds-ratios and 95% confidence intervals. Confidence intervals were estimated based on clustered robust standard errors, with hospital identified as the cluster. Huber-White robust standard errors were also calculated. However, they were quite similar to the traditionally calculated standard errors. Like the traditional standard errors estimated for these models, they were much smaller than the estimated clustered robust errors. The use of clustered robust errors was chosen as the more conservative strategy. In discussing the results, only those parameters with an alpha of less than 0.01 are considered in order to protect against the elevated risk of Type I error, judging a parameter to be statistically significant when it is not, that occurs when estimating a large number of parameters or making multiple comparisons.29 The most traditional adjustment is a Bonferroni adjustment, but the Bonferroni is an extraordinarily conservative approach to the problem.30 This effort does not use the overly conservative Bonferroni adjustment but does address the issue of multiple comparisons by using an alpha of 0.01 rather than 0.05. In reporting the results of the multivariate analyses, we report two measures of goodness-of-fit. The first is the pseudo-R2 based on a comparison of the log-likelihood of the intercept-only model with the model containing the intercept and the covariates. The second is the C statistic, which indicates the area under the calculated Receiver Operating Characteristics (ROC) curve generated by the model's predictions.28 Results As Table 1 indicates, only 5.8 percent of discharges involving children or youth with special health care needs resulted in the provision of post-discharge services. The overall percentage of PAC
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
4
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
Table 1 Descriptive Statistics for Children with SHCN Discharged from Hospital in Texas With or Without Post-Discharge Services, 2011e2014. Variable
Number of Chronic Disease Diagnoses One Two Three Four or more Complex Chronic Conditions None One Two Three or more Illness Severity Mild Moderate Serious Extreme Demographics Female White Hispanic Black Asian Other Payor Other Medicaid Charity Age Less than 28 days 29-365 days 1-4 years 5-9 years 10-14 years 15-17 years Surgery Performed Yes Type of Acute Care Setting Other Teaching Children's Context Metropolitan Area Length of Acute Care Stay Average Standard Deviation
Population N ¼ 472,825 (Col. %)
Post-Acute Disposition (Row. %) Discharged Home with No Services (N ¼ 445,501) 94.2%
With Post-Discharge Health Services (N ¼ 27,324) 5.8%
50.0 21.7 11.5 16.8
96.1 95.4 93.2 87.6
3.9 4.6 6.8 12.4
64.2 24.2 7.8 3.8
96.5 92.6 87.6 78.6
3.5 7.4 12.4 21.4
36.5 40.3 17.6 5.6
97.3 95.7 90.6 74.6
2.7 4.3 9.4 25.4
47.4 36.5 38.3 15.2 2.1 7.4
94.4 94.0 94.5 94.8 95.0 92.7
5.6 6.0 5.5 5.2 5.0 7.3
43.1 47.6 9.2
94.3 93.8 96.0
5.7 6.2 4.0
22.9 11.6 14.2 13.7 19.6 17.9
94.5 90.5 95.1 95.8 94.6 93.9
5.5 9.5 4.9 4.2 5.4 6.1
23.4
91.7
8.3
53.0 9.5 37.5
94.1 92.3 94.9
5.9 7.7 5.1
98.1
94.3
5.7
8.0 19.0
7.5 18.1
16.8 28.8
use found in Berry and his colleagues 44-state study of general pediatric discharges was 6.1 percent.3 Using a Texas discharge database with similar exclusions to those seen in this earlier research, analyses indicated that 6.4 percent of general pediatric discharges in Texas resulted in post-acute care (authors' analysis). Thus, the prevalence of PAC in Texas for discharges involving children with SHCN was lower than that observed in the multi-state study of general pediatric hospital discharges and below that for general pediatric discharges in Texas as well. Half of all discharges reported in Table 1 involved children with SHCN who had more than one chronic condition. Just over onethird of the discharges involved a patient with more than one complex chronic condition; 64.5 percent of discharges were classified as greater than “mild” in their severity. The modal payor for care was Medicaid (47.6%); the modal race/ethnicity was Hispanics (38.3%), followed closely by discharges for White/Non-Hispanics (36.3%). Almost one-quarter (22.9%) of the discharges involved infants less than 29 days old. Almost one-quarter of those in the sample had a surgical
procedure during their hospital stay, and 98 percent of those discharges lived in urban areas. Over one-third were discharged from a children's hospital. The average length of stay was 8.0 days. The bivariate results presented in Table 1 indicate four individual characteristics that were most likely to result in the provision of post-discharges services in this population. These were the presence of four or more diagnoses of a chronic condition, at least two complex chronic conditions, or extreme illness severity. Table 2 presents information, not on issues of chronic illness, but on the general nature of the patients' principal diagnosis as specified in the discharge record. For the 16 categories included in our analysis, mental health issues and respiratory problem were the most prevalent. Injuries, perinatal issues, neoplasms, and circulatory problems were the most likely to receive post-discharge services. Pregnancy and hematological problem were the least likely to receive PAC. In the earlier multi-state research on PAC for the general
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
5
Table 2 Principal Presenting Problem for Children with SHCN Discharged from Hospital in Texas With or Without Post-Discharge Services, 2011e2014. Principal Presenting Problem
Hematology Circulatory Congenital defect Digestive Endocrine Genital Infectious Injury Musculoskeletal Neoplasm Nervous Skin Perinatal Pregnancy Respiratory Mental health
Population N ¼ 472,825 (Col. %)
2.33 1.32 4.47 4.46 3.60 1.88 1.99 4.09 1.60 1.37 4.38 1.13 2.08 2.12 13.90 19.94
Post-Acute Disposition (Col. %) Discharged Home with No Services (N ¼ 445,501) 94.2%
With Post-Discharge Health Services (N ¼ 27,324) 5.8%
97.4 91.5 93.7 94.6 96.1 96.5 92.3 84.2 92.3 91.4 94.7 95.5 90.0 99.1 95.5 95.4
2.6 8.5 6.3 5.4 3.9 3.5 7.7 17.6 7.7 8.6 5.3 4.5 10.0 0.9 4.5 4.6
pediatric population, the most common form of PAC was discharge home with home health services (84.2%). Only 17.6 percent of discharges with services resulted in a discharge to a health facility.3 The results presented here are quite different. Of those 5.8 percent of discharges with services, 66 percent resulted in post-discharge services in some type of HCF, while only 34 percent of discharges resulted in the provision of care in the home. Table 3 displays the multivariate results indicating what factors had a significant impact on the distribution of PAC to children with SHCN. The first model, estimating the odds of receiving any formal post-discharge services, had a pseudo-R2 of 0.14 and a C statistic of 0.79. The second model included only those discharges in which the patient received formal post-discharge services and examined the factors that affected whether those services were received in the home or in a HCF. That model had a pseudo-R2 of 0.36 and a C statistics of 0.88. Unlike other pediatric hospital discharges,3 among discharges involving children or youth with special health care needs, the number of chronic conditions had no significant impact on whether formal post-discharge services were provided. The same holds true for the model estimating the likelihood of receiving post-discharge services in a HCF versus in the home. Instead, the individual factors that had the greatest effect on the receipt of formal post-discharge services were the number of complex chronic conditions and the severity of the illness during the hospital stay. The presence of complex chronic conditions made it much more likely a patient would receive formal services after discharge. In the model for those receiving services, however, those with one of more complex chronic conditions were significantly less likely to receive services in a HCF. It was those with no complex chronic conditions or who had less than extremely serious illnesses that brought them to the hospital who were more likely to have received care in a HCF. Only those patients who presented with extremely severe conditions were much more likely (OR ¼ 2.28, CI 1.73e2.99) to receive post-discharge services in a health care setting. Sex and race/ethnicity had no significant effect on whether formal services were received after discharge from hospital or where it was received. Payment source had no effect on whether the patient received post-discharge care. However, those with no source of payment who received post-discharge services were much more likely (OR ¼ 1.89, CI ¼ 1.25e2.86) than those with
identifiable payors to receive post discharge care in a health facility than at home. The effect of age on the likelihood of receiving formal postdischarge care was somewhat complex. Infants (0e28 days) and older children and youth (10e17) had basically the same odds of receiving services. Those infants and children with ages in what might be considered the middle range (29 dayse9 years) had significantly lower odds of receiving services. However, the effect of age on where post discharge services were received was much simpler. Compared to those less than 29 days old, all other children and youth were less likely to receive PAC services in a health care setting. Teaching hospitals did not differ from other hospitals in their likelihood of discharging a child or youth with special needs with services, though they were significantly less likely to send a child to a HCF for services. Children's’ hospitals were significantly less likely than other acute care hospitals to provide a child with PAC services and less likely than other acute care settings to discharge a patient to a HCF. The effect of having a surgical procedure had no effect on whether a patient received services, but its presence significantly reduced the likelihood of any post discharge services being provided in a HCF. The effects of the patient's length of stay in hospital also followed that pattern. It had no effect of the likelihood of services, but, for those who did receive services, a longer stay reduced the odds of receiving services in a health care setting. Table 4 displays other results from the same equations as those estimated for Table 3, but the emphasis here is on the effect of different types of presenting, or principal, problems on decisions concerning care after discharge. Only three of these diagnosis categories had consistent results across both models. Circulatory problems, an injury, and the presence of neoplasms all increased the likelihood of services and of those services being provided in a heath care setting. The other ten types of primary problems exhibited different effects between the likelihood of receiving services and the likelihood of services being provided in a health setting. Notably, children or youth that presented with mental health problems represent a relatively large proportion of discharges. These patients are no more or less likely to receive PAC. However, when PAC is provided, the odds that it would be in a HCF seem almost ridiculously high (OR ¼ 35.7, CI 11.5e110.82). However, when one looks at those who received PAC and had a mental health
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
6
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
Table 3 Logistic Regression Results Reflecting Post-Discharge Service Results for Children with Special Health Care Needs. Variable
Odds of Being Discharged with Services pseudo-R2 ¼ 0.24/C ¼ 0.79 Odds-Ratio (95% CI)
Number of Chronic Disease Diagnoses One e Two 0.82 (0.57e1.20) Three 0.93 (0.67e1.30) Four or more 1.00 (0.64e1.57) Complex Chronic Conditions None e One 2.23*** (1.77e2.83) Two 3.16*** (2.26e4.40) Three or more 4.89*** (3.40e7.02) Illness Severity Mild e Moderate 1.47 (1.01e2.14) Serious 2.62*** (1.80e3.83) Extreme 6.51*** (3.80e11.16) Demographics Female 1.01 (0.88e1.16) White e Hispanic 0.70 (0.52e0.95) Black 0.94 (0.88e1.01) Asian 0.88 (0.75e1.05) Payor Other e Medicaid 1.11 (1.00e1.23) Charity 0.87 (0.48e1.56) Age Less than 28 days e 29-364 days 0.67** (0.50e0.89) 1-4 years 0.51*** (0.39e0.66) 5-9 years 0.51*** (0.38e0.67) 10-14 years 0.71 (0.54e0.93) 15-17 years 0.84 (0.64e1.11) Surgery 0.91 (0.75e1.11) Type of Acute Care Setting Other e Teaching 0.89 (0.60e1.33) Children's 0.41*** (0.29e0.60) Metropolitan Area 0.45*** (0.36e0.54) Length Stay (log) 1.19 (0.86e1.49) Time (quarter) 0.98 (0.96e1.01)
If Receiving Services, Odds of Receiving Services in a Heath Care Facility pseudo-R2 ¼ 0.36/C ¼ 0.88 Odds-Ratio (95% CI) e 0.92 (0.81e1.04) 0.90 (0.78e1.04) 0.79 (0.63e0.99) e 0.70*** (0.57e0.85) 0.50*** (0.41e0.60) 0.31*** (0.23e0.41) e 0.97 (0.83e1.13) 1.06 (0.90e1.25) 2.28*** (1.74e2.99) 1.03 (0.95e1.13) e 0.95 (0.80e1.11) 1.12 (0.98e1.27) 0.82 (0.66e1.02) e 0.92 (0.79e1.06) 1.89** (1.25e2.86) – 0.28*** (0.19e0.42) 0.15*** (0.11e0.22) 0.17*** (0.120.26) 0.27*** (0.18e0.41) 0.36*** (0.24e0.53) 0.73** (0.59e0.89) e 0.45** (0.25e0.79) 0.36*** (0.25e0.52) 0.30*** (0.18e0.50) 0.71** (0.61e0.84) 1.00 (0.97e1.03)
** p <. 01 *** p <. 001.
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
7
Table 4 Logistic Regression Results Related to The Child With Special Health Care Need's Principal Presenting Problem. Principal Presenting Problem
Odds Ratios for Receiving Any Services Odds-Ratio (95% C.I.)
Odds Ratios for Receiving Services in a Health Care Facility Odds-Ratio (95% C.I.)
Hematologic
0.70** (0.55e0.89) 1.82*** (1.37e2.41) 1.21 (0.96e1.52) 1.49** (1.14e1.94) 1.16 (0.82e1.66) 0.89 (0.62e1.27) 1.48** (1.12e1.94) 5.30*** (3.99e7.05) 2.92*** (2.15e3.95) 1.57** (1.19e2.06) 1.49 (0.97e2.28) 1.87*** (1.40e2.48) 1.04 (0.82e1.34) 0.23*** (0.14e0.39) 1.34 (0.96e1.88) 1.73 (0.75e4.02)
1.00 (0.63e1.58) 2.49*** (1.80e3.44) 0.96 (0.74e1.23) 0.57*** (0.44e0.75) 0.69 (0.48e0.99) 0.60*** (0.45e0.80) 0.64 (0.45e0.91) 2.01*** (1.50e2.68) 0.43 (0.22e0.82) 2.28*** (1.48e3.51) 1.61** 2.01e4.81 0.30** (0.15e0.60) 1.04 (0.78e1.38) 0.62 (0.36e1.08) 0.54** (0.38e0.77) 35.74*** (11.53e110.82)
Circulatory Congenital defect Digestive Endocrine Genital Infectious Injury Musculoskeletal Neoplasm Nervous Skin Perinatal Pregnancy Respiratory Mental health ** p <. 01 *** p <. 001.
issue as their principal presenting problem, almost 99 percent received their PAC in a HCF. Discussion While considerable health services research has focused on PAC for older persons, very little has addressed the post-acute care services received by other segments of the population. Recent research has contributed to our knowledge on PAC for younger persons in general.3 The research presented here constitutes an attempt to add to the more general issue of pediatric PAC a similar concern with post discharge services for some of the most vulnerable of younger persons, children with SHCN. The results presented here indicate that the decision-making concerning post discharge services for children with SHCN may differ in many ways from those dynamics in the more general pediatric population. Berry and his colleagues found that the number of chronic condition was a major predictor of the type of PAC received,3 while among discharges for children with SHCN, the number of conditions was not a significant predictor of whether or where PAC was received. Earlier research found differences based on the patient race or ethnicity that were not observed in this research. The effects of age and payer also were more complex in these data than in those concerning general pediatric post-acute care. The seriousness of the illness that precipitated the acute care stay, a previously unexplored factor in pediatric discharges, and the number of complex chronic conditions were, for this population, the major individual characteristics affecting decisions about services following an acute care stay.
The nature of the acute care setting was also important. Urban hospitals were less likely to use PAC. The average acute care hospital was significantly more likely to discharge children with SHCN with services than were teaching or children's hospitals. Also, general hospitals and hospitals in rural areas were significantly more likely to use post-discharge services in health care facilities. The results noted above are interesting, but limited to the effects of specific factors on PAC. One more general conclusion, however, seems discernable from these results. The use of post-acute care services for children with SHCN appears to be surprisingly low in data focusing specifically on a relative fragile population with chronic conditions, many with multiple complex conditions or severe illnesses. The 5.8 percent of discharges for children with SHCN receiving PAC in these data is approximately five percent lower than the overall prevalence (6.1%) of PAC for general pediatric discharges observed in the data from 44 states.3 It is also almost ten percent lower than the prevalence of PAC (6.4%) observed in a database of pediatric hospital discharges in Texas constructed to mirror the 44state data (authors' analysis). This result raises questions of possible under-utilization of PAC for children with SHCN that this research cannot, unfortunately, address. What the result suggests is a pressing need for outcome studies. The question of under-utilization of PAC for children with SHCN can only be investigated by research focusing on the effects in this special population of going home after an acute care stay without services versus receiving various types and levels of services after an acute care discharge. In the 44-state data only 17.6 percent of discharges from the general children population resulted in the provision of services in
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
8
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
a health care setting.3 In the Texas data, post-acute treatment of 66 percent of discharges for children with SHCN was in health care facilities. One source of this difference may be that the earlier study excluded newborns and patients with a mental health diagnosis. However, even when those two types of patients are excluded from the Texas database used in this research, 31 percent of discharges for children with SHCN who received PAC received those services in a health care setting (authors' analysis). It may be that in all settings, children with SHCN, unlike the more general pediatric population, are significantly more likely to receive services after an acute care stay in a HCF than they are to receive services at home. However, whether the prevalence is 66 percent or 31 percent, such results fly in the face of the near universal professional judgement that any care other than acute care for this population is best provided in the home.22 Studies of children with SHCN in other settings or states might reveal whether the Texas results are anomalous or are in fact generally reflective of PAC for patients in this population. If the results of similar studies in other settings indicate that these results are not anomalous, then smaller scale studies would seem to be required to better understand the use of post-discharge services for children with SHCN. These more-limited studies might involve quantitative work with more finely-grained data about patient conditions and service providers. Qualitative research that would provide more information about the dynamics of decision-making involving this special population could also prove quite useful. In this vein, both qualitative and quantitative (mixed-method) studies investigating the role of the availability of services, patients' and families' preferences, as well as the family's ability and willingness to assist in needed care, might be important considerations.14,15 However, the results presented here may differ significantly from similar studies in other sites e other studies might find very different prevalence for the use of PAC and the sites in which it is delivered. Should that be the case, then one suspects a search would begin for what aspects of the broader context affect different patterns of PAC for children with special health care needs. One obvious axis of such analyses might be variation in the availability of different types and intensities of PAC services across settings. Another issue might be the degree to which PAC has been substituted for inpatient care in these different settings.32 Other investigators would do well to be mindful of the limitations of this study. The study database was limited to a three-year period in one state, albeit one with a large population. The use of state administrative data gathered largely for the purposes of payment limited the amount of clinical and provider detail available on each discharge. As a purely quantitative secondary analysis, this research lacks the richness that might be found in a mixedmethods approach that integrated qualitative insights into the quantitative element of such research. Such research might help with understanding the results observed for age, type of hospital, and presenting problems. In addition, our multivariate models addressing these issues need improvement in later research efforts. Both logistic regression models displayed only a moderate ability to make correct predictions. The positive predictive value for the model distinguishing between no formal care and the provision of formal care had a positive predictive value of only 48.1 percent and correctly predicted the outcome 94.2 percent of cases. The model differentiating between sites of formal care had a positive predictive value of 69.5 percent and correctly predicted 78 percent of case outcomes. Further testing of the models indicated that the models may not be properly specified.31 These tests other research suggest that the models used in this research might benefit from additional data drawn from other sources.32
With these limitations in mind, this research offered a useful strategy for classifying and analyzing the distribution of pediatric PAC services, which may be of assistance to other investigators. It also made at least an initial contribution to an important issue that has, as of yet, received little attention in health services research e decisions concerning the provision of different types of health services to children with SHCN, a vulnerable population composed of relatively heavy users both of formal health services and family resources. Conflicts of interests The authors have no conflicts-of-interests, and no external or special funding was used to support this research. References 1. Morley M, Bogasky S, Gage B, Flood S, Ingber MJ. Medicare post-acute care episodes and payment bundling. Medicare Medicaid Res Rev. 2014;4(1). mmrr.004.01.b02. 2. Chandra A, Dalton MA, Holmes J. Large increases in spending on postacute care in Medicare point to the potential for cost savings in these settings. Health Aff. 2013;32(5):864e872. 3. Berry JG, Hall M, Dumas H, et al. Pediatric hospital discharges to home health and postacute facility care: a national study. JAMA Pediatr. 2016;170(4): 326e333. 4. Children and Adolescent Health Measurement Institute. Who Are Children with Special Health Care Needs (CSHCN). Data Resource Center supported by Cooperative Agreement 1-U59-MC06980-01 from the U.S. Department of Health and Human Services. Health Resources and Services Adminstration (HRSA), Maternal and Child Health Bureau (MCHB); 2012. Available at: www. childheath.org. Accessed December 30, 2016. Revised 4/3/2012. 5. Willits KA, Platonova EA, Nies MA, Racine EF, Troutman ML, Harris HL. Medical home and pediatric primary care utilization among Children with Special Health Care Needs. J Pediatr Health Care. 2013;27(3):202e208. 6. Bethell CD, Newacheck PW, Fine A, et al. Optimizing health and health care systems for Children with Special Health Care Needs using the life course perspective. Mater Child Health J. 2014;18(2):467e477. 7. Van Cleave J, Boudreau AA, McAllister J, Cooley WC, Maxwell A, Kuhlthau K. Care coordination over time in medical homes for children with special health care needs. Pediatrics. 2015;135(6):1018e1026. 8. Schieve LA, Gonzalez V, Boulet SL, et al. Concurrent medical conditions and health care use and needs among children with learning and behavioral developmental disabilities, National Health Interview Survey, 2006e2010. Res Dev Disabil. 2012;33(2):467e476. 9. Calcedo C. Children with Special Health Care Needs: child health and functioning outcomes and health care service use. J Pediatr Health Care. 2016;30(6): 590e598. 10. Newacheck PW, Kim SE. A national profile of health care utilization and expenditures for Children with Special Health Care Needs. Arch Pediatr Adolesc Med. 2005;159(1):10e17. 11. Cohen E, Berry JG, Camacho X, Anderson G, Wodchis W, Guttmann A. Patterns and costs of health care use of children with medical complexity. Pediatrics. 2012;130(6):e1463ee1470. 12. Agrawal R, Hall M, Cohen E, et al. Trends in health care spending for children in Medicaid with high resource use. Pediatrics. 2016;138(4):e20160682. 13. Miller TR, Elliott TR, McMaughan DM, et al. Personal care services provided to children with special health care needs (CSHCN) and their subsequent use of physician services. Disabil Health J. 2013;6(4):317e324. 14. Romley JA, Shah AK, Chung PJ, Elliott MN, Vestal KD, Schuster MA. Familyprovided health care for children with special health care needs. Pediatrics. 2017;139(1):e20161287. 15. Cousino MK, Hazen RA. Parenting Stress among caregivers of children with chronic illness: a systematic review. J Pediatr Psychol. 2013;38(8):809e828. 16. Phillips CD. What do you do for a living: toward a more succinct definition of health services research. BMC Health Serv Res. 2006;6:117. 17. Bowles KH, Holmes JH, Ratcliffe SJ, Liberatore M, Nydick R, Naylor MD. Factors identified by experts to support decision making in post acute referral. Nurs Res. 2009;58(2), 115e112. 18. Burke RE, Juarez-Colunga E, Levy C, Prochazka AV, Coleman EA, Ginde AA. Patient and hospitalization characteristics associated with increased post-acute care facility discharges from US hospitals. Med Care. 2015;53(6):492e500. 19. Phillips CD, Truong C, Kum H, Nwaiwu O, Ohsfeldt O. Post-Acute care for children and youth in Texas, 2011-2014. Clin Med Insights Pediatr. 2017;11. http://dx.doi.org/10.1177/1179556517711445. 20. Gage B, Morley M, Spain P, Ingber M. Final Report: Post Acute Care in an Integrated Hospital System. 2009. Waltham, MA: RTI International; 2009. 21. Texas Department of State Health Services. Hospital Inpatient Public Use Data File. [accessed on May 5, 2016]. Available at: https://www.dshs.texas.gov/thcic/ hospitals/Inpatientpudf.shtm.
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9 22. Committee on Children with Disabilities. American Academy of Pediatrics: guidelines for home care of infants, children, and adolescents with chronic disease. Pediatrics. 1995;96(1):161e164. 23. Texas Department of State Health Services. User Manual: Texas Hospital Inpatient Discharge File (PUDF); 2015. Available at: https://www.dshs.texas.gov/ thcic/hospitals/Inpatientpudf.shtm. Accessed April 14, 2016. 24. McPherson M, Arango P, Fox H, et al. A new definition of children with special health care needs. Pediatrics. 1998;102:137e140. 25. Healthcare Cost and Utilization Project (HCUP). HCUP Chronic Condition Indicator; 2016. Available at: https://www.hcupus.ahrq.gov/toolssoftware/chronic/ chronic.jsp. Accessed October 1, 2016. 26. Feudtner C, Feinstein JA, Zhong W, Hall M, Dai D. Pediatric complex chronic conditions classification system version 2: updated for ICD-10 and complex medical technology dependence and transplantation. Biomed Cent Pediatr.
9
2014;14(199):1. 27. United States Department of Agriculture Economic Research Service. Ruralurban Continuum Codes; 2013. Available at: https://www.ers.usda.gov/dataproducts/rural-urban-continuum-codes/. Accessed October 22, 2016. 28. StataCorp. STATA (Release 14) Statistical Software. College Station, TX: StataCorp LP; 2015. 29. Mittelhammer, Ron C, Judge, George G, Miller, Douglas J. Econometric Foundations. Cambridge: Cambridge University Press; 2000. 30. Perneger T. What's wrong with Bonferroni adjustments? BMJ. 1998;316, 1236e8. 31. Hosmer DW, Lemeshow S. Applied Regression Analysis. second ed. New York: John Wiley & Sons; 2000. 32. Buntin MB, Garten Ad, Paddock S, Saliba D, Totten M, Escare JJ. How much of post acute use is affected by its availability. HSR. 2005;40(2):413e434.
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
Contents lists available at ScienceDirect
Disability and Health Journal journal homepage: www.disabilityandhealthjnl.com
Post-acute care for children with special health care needs Charles D. Phillips a, *, Chau Truong b, Hye-Chung Kum a, Obioma Nwaiwu c, Robert Ohsfeldt a a
Texas A&M Health Science Center, School of Public Health, Department of Health Policy and Management, USA University of Texas, School of Public Health, Department of Management, Policy, and Community Health, USA c University of Arkansas for Medical Sciences, School of Medicine, Department of Family Medicine, USA b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 12 February 2017 Received in revised form 25 August 2017 Accepted 31 August 2017
Background: Almost all studies of post-acute care (PAC) focus on older persons, frequently those suffering from chronic health problems. Some research is available on PAC for the pediatric population in general. However, very few studies focus on PAC services for children with special health care needs (SHCN). Objective: To investigate factors affecting the provision of PAC to children with SHCN. Methods: Pooled cross-sectional data from Texas Department of State Health Services hospital discharge database from 2011-2014 were analyzed. Publicly available algorithms identified chronic conditions, complex chronic conditions, and the principal problem leading to hospitalization. Analysis involved estimating two logistic regressions, with clustered robust standard errors, concerning the likelihood of receiving PAC and where that PAC was delivered. Models included patient characteristics and conditions, as well as hospital characteristics and location. Results: Only 5.8 percent of discharges for children with SHCN resulted in the provision of PAC. Twothirds of PAC was provided in a health care facility (HCF). Severity of illness and the number of complex chronic conditions, though not the number of chronic problems, made PAC more likely. Patient demographics had no effect on PAC decisions. Hospital type and location also affected PAC decisionmaking. Conclusions: PAC was provided to relatively few children with SHCN, which raises questions concerning the potential underutilization of PAC for children with SHCN. Also, the provision of most PAC in a HCF (66%) seems at odds with professional judgment and family preferences indicating that health care for children with SHCN is best provided in the home. © 2017 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Keywords: Children with special health care needs Post-acute care Disabilities Child and maternal health
Background Post-acute care for older persons has been the focus of considerable research. In part this emphasis stems from the prevalence of PAC for older adults. In 2008, estimates indicate that 38.7 percent of Medicare recipients received post-acute care.1 In addition, PAC expenditures have proved to be the fastest growing major health care spending category.2 Spending on Medicare PAC had risen to $635 billion by 2011.1 In contrast, Jay Berry and his colleagues' recent work focused on
* Corresponding author. Texas A&M Health Science Center, School of Public Health, Adriance Lab Road, Department of Health Policy and Management, College Station, TX, 78543-1266, USA. E-mail address: [email protected] (C.D. Phillips).
an area of post-acute care that they correctly indicate has historically received little attention e hospital discharges for those under 21 years of age.3 Their research involved a sample of over two million 2012 discharges from hospitals in 44-states. Their effort provided an important and much needed foundation for additional analyses of pediatric hospital discharges and of the use of postdischarge services among children and youth. In that sample, just over six percent of discharges resulted in the provision of postacute services. Of that 6.1 percent who received PAC, just 17.6 percent received those services in a health care facility (HCF), while the remainder (82.4%) received home health care (HHC) services. The multivariate analyses in that study indicated how individual characteristics (race, age), the number of chronic diseases, diagnoses of complex chronic conditions, source of payment (Medicaid), the type of hospital (children's), as well as state and
http://dx.doi.org/10.1016/j.dhjo.2017.08.010 1936-6574/© 2017 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/ ).
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
2
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
region of the nation significantly affected the use of different postdischarge services.3 This research complements that earlier effort. However, it approaches the analyses of PAC for children or youth differently. This research focused on post-discharge dispositions for an especially vulnerable pediatric population, children with special health care needs (SHCN). As the 20011/12 National Survey of Children with Special Health Care Needs estimated, 14.6 million children or youth under 18 years of age in the US faced health challenges requiring special health care services. This is almost one of every five (19.8%) persons under 18.4 This population presents special challenges to all elements of healthcare.5e7 They need access to a relatively wide variety of health care services and absorb a disproportionate share of health care expenditures.8e12 In 2009, for example, children in Texas's Early and Periodic Screening Diagnostic and Treatment (EPSDT) program, which is populated by children with chronic conditions, averaged Medicaid expenditures over 18 times ($33,628 vs. $1,834) that of the average Medicaid expenditure for a child or youth in Texas.13 As important as such information is, these prevalence and expenditure figures fail to include an important element in the lives of children with SHCN. Families bear much of the responsibility for care of children with SHCN and provide the equivalent of billions of dollars of family care to these children.14 The burden of that care can at times be overwhelming, and the potential adverse effects of that responsibility on family members and family life can be devastating.15 While family care, frequently called informal care, is an important aspect of post-discharge services, the research presented here will focus more narrowly on the distribution of postdischarge services provided by health care professionals (formal care). Like many issues in health services research, elaborate theoretical frameworks have not driven investigations into the provision of formal PAC services.16 Instead, previous research on PAC provides a basic conceptual model for guidance, identifying factors associated with PAC use in previous analyses of hospital discharge data. A graphical representation of the basic model used in this research appears in Fig. 1. The elements of the model that focus on those variables associated with PAC are relatively standard in such analyses.17e19 This effort, however, focuses on children with SHCN, which differs from other populations in which PAC has been investigated. Therefore, these analyses elaborate on the basic model found in earlier research by using more finely-grained information on the number of complex chronic conditions, condition severity, as well as categories of presenting conditions. The model, unlike others, also allows for the investigation for the effect of time on PAC decisions. In addition, the dependent variables representing PAC decisions in this research differ from that in other analyses. Other efforts consider PAC best defined as a single dependent variable with separate categories for no services and for each setting where PAC is provided. This type of analysis results in separate estimates for receiving no services versus receiving services in a particular setting (e.g., likelihood of no services versus home care or the likelihood of no services versus PAC in an HCF). This strategy was used by earlier analyses of PAC for children and youth.3 However, the analyses here conceptualize PAC provision as a two-stage, rather than one-stage, decision process. These two decisions are: (1) whether any type of PAC will be provided, and (2) for those receiving PAC, where that PAC will be provided. This approach is like that used by Gage and her colleagues in their analyses of PAC for older persons and allows one to investigate directly any important differences between those children receiving PAC at home or those children receiving PAC in a health
care facility.20 This formulation results in two questions concerning formal post-discharge services provided to this vulnerable population. What affects the likelihood that an acute care discharge involving a child or youth with SHCN will result in the provision of formal PAC? For those discharges where children or youth with SHCN receive formal post-discharge services, what differentiates between those discharges receiving HHC and those receiving services in a HCF?
Methods Data The Texas Department of State Health Services maintains public use data files of all hospital discharges. From those data, a pooled cross-sectional database of hospital discharges for those discharges involving patients under the age of 18 for three years (10/2011-3/ 2014) were extracted.21 Of these 1,688,454 discharges for children or youth under the age of eighteen, 476,144 (28.2%) involved a child or youth with at least one chronic condition. After excluding those discharges for children without a chronic condition, those who died, or those who discharged against medical advice, a total of 472,825 discharges involving children or youth with a chronic condition remained. We identified these discharges as discharges involving children with SHCN. All aspects of this project were approved by the Institutional Review Board at Texas A&M University. Measurement Dependent variables The dependent variables in this effort focused on the services identified in the discharge record. They focus on two very clear distinctions in post-discharge services for children with SHCN. As noted earlier, the dependent variable for the first multivariate model defined two categories: those discharges with no formal services and those discharges involving any type of formal services. The second dependent variable focused only on those discharges that involved services. It differentiated between the sites where care was received after discharge, an important distinction for the patient, as well as informal and formal caregivers.22 This dependent variable differentiated between those discharges involving children and youth discharged to home with HHC (i.e., home health or private duty nursing) and those who received care after discharge in a HCF of some type (i.e., rehabilitation facility, nursing facility, residential hospice, mental health facility, specialty hospital, or another acute care facility).23 Independent variables Any discharge involving a child or youth with at least one diagnosis of a chronic condition was included in our study database as a discharge involving a child or youth with special health care needs.24 The presence of a chronic condition was indicated if the principal diagnosis or any of the other diagnoses listed in the discharge record (up to 24) were included in the list of chronic conditions developed in the Healthcare Costs and Utilization Project.25 Complex chronic conditions were identified using Feudtner and his colleagues pediatric complex chronic condition system.26 The standard demographic characteristics of age, sex, and race/ ethnicity were provided in the Texas discharge data. The ages under
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
3
Fig. 1. Model underlying the Investigation of PAC provision for Children with SHCN.
18 were grouped into six categories in the discharge records. The principal diagnosis listed in the discharge data was converted, using the HCUP system, into a set of binary indicators indicating the type of presenting problem that resulted in hospitalization.25 The illness severity measure (mild, moderate, severe, extreme), based on AP-DRGs was also available in the discharge database.23 Length of the hospital stay was identified in the discharge record as the number of hospital days. Patients were identified as having had surgery if the principal procedure code for the discharge indicated a surgical procedure and the discharge involved charges for use of operating room services. Hospitals with membership in The Council of Teaching Hospitals were coded as teaching hospitals, and those with membership in The National Association of Children's Hospitals and Related Institutions were coded as children's hospitals. Rural Urban Classification Codes, version 2103, were used to construct a binary variable indicating that the hospital operated in a large metropolitan area.27 Binary variables representing nine of the ten Texas Public Health Regions were also included in the models to adjust for any regional differences in the use of postdischarge services. Analysis strategy All analyses were performed using STATA 14.28 Descriptive statistics for all variables were calculated. No variable used in the analysis had greater than one percent missing data. Population data such as those used in this research are not collected, especially in large scale administrative databases, without error. To account for this reality, the data were treated as a sample from a series of populations, each with differing amounts of processing error. This conceptualization makes the use of standard inferential statistical techniques both necessary and appropriate. In the two estimated logistic regression models, all independent
variables were treated as fixed-effects. The multivariate results reported include odds-ratios and 95% confidence intervals. Confidence intervals were estimated based on clustered robust standard errors, with hospital identified as the cluster. Huber-White robust standard errors were also calculated. However, they were quite similar to the traditionally calculated standard errors. Like the traditional standard errors estimated for these models, they were much smaller than the estimated clustered robust errors. The use of clustered robust errors was chosen as the more conservative strategy. In discussing the results, only those parameters with an alpha of less than 0.01 are considered in order to protect against the elevated risk of Type I error, judging a parameter to be statistically significant when it is not, that occurs when estimating a large number of parameters or making multiple comparisons.29 The most traditional adjustment is a Bonferroni adjustment, but the Bonferroni is an extraordinarily conservative approach to the problem.30 This effort does not use the overly conservative Bonferroni adjustment but does address the issue of multiple comparisons by using an alpha of 0.01 rather than 0.05. In reporting the results of the multivariate analyses, we report two measures of goodness-of-fit. The first is the pseudo-R2 based on a comparison of the log-likelihood of the intercept-only model with the model containing the intercept and the covariates. The second is the C statistic, which indicates the area under the calculated Receiver Operating Characteristics (ROC) curve generated by the model's predictions.28 Results As Table 1 indicates, only 5.8 percent of discharges involving children or youth with special health care needs resulted in the provision of post-discharge services. The overall percentage of PAC
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
4
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
Table 1 Descriptive Statistics for Children with SHCN Discharged from Hospital in Texas With or Without Post-Discharge Services, 2011e2014. Variable
Number of Chronic Disease Diagnoses One Two Three Four or more Complex Chronic Conditions None One Two Three or more Illness Severity Mild Moderate Serious Extreme Demographics Female White Hispanic Black Asian Other Payor Other Medicaid Charity Age Less than 28 days 29-365 days 1-4 years 5-9 years 10-14 years 15-17 years Surgery Performed Yes Type of Acute Care Setting Other Teaching Children's Context Metropolitan Area Length of Acute Care Stay Average Standard Deviation
Population N ¼ 472,825 (Col. %)
Post-Acute Disposition (Row. %) Discharged Home with No Services (N ¼ 445,501) 94.2%
With Post-Discharge Health Services (N ¼ 27,324) 5.8%
50.0 21.7 11.5 16.8
96.1 95.4 93.2 87.6
3.9 4.6 6.8 12.4
64.2 24.2 7.8 3.8
96.5 92.6 87.6 78.6
3.5 7.4 12.4 21.4
36.5 40.3 17.6 5.6
97.3 95.7 90.6 74.6
2.7 4.3 9.4 25.4
47.4 36.5 38.3 15.2 2.1 7.4
94.4 94.0 94.5 94.8 95.0 92.7
5.6 6.0 5.5 5.2 5.0 7.3
43.1 47.6 9.2
94.3 93.8 96.0
5.7 6.2 4.0
22.9 11.6 14.2 13.7 19.6 17.9
94.5 90.5 95.1 95.8 94.6 93.9
5.5 9.5 4.9 4.2 5.4 6.1
23.4
91.7
8.3
53.0 9.5 37.5
94.1 92.3 94.9
5.9 7.7 5.1
98.1
94.3
5.7
8.0 19.0
7.5 18.1
16.8 28.8
use found in Berry and his colleagues 44-state study of general pediatric discharges was 6.1 percent.3 Using a Texas discharge database with similar exclusions to those seen in this earlier research, analyses indicated that 6.4 percent of general pediatric discharges in Texas resulted in post-acute care (authors' analysis). Thus, the prevalence of PAC in Texas for discharges involving children with SHCN was lower than that observed in the multi-state study of general pediatric hospital discharges and below that for general pediatric discharges in Texas as well. Half of all discharges reported in Table 1 involved children with SHCN who had more than one chronic condition. Just over onethird of the discharges involved a patient with more than one complex chronic condition; 64.5 percent of discharges were classified as greater than “mild” in their severity. The modal payor for care was Medicaid (47.6%); the modal race/ethnicity was Hispanics (38.3%), followed closely by discharges for White/Non-Hispanics (36.3%). Almost one-quarter (22.9%) of the discharges involved infants less than 29 days old. Almost one-quarter of those in the sample had a surgical
procedure during their hospital stay, and 98 percent of those discharges lived in urban areas. Over one-third were discharged from a children's hospital. The average length of stay was 8.0 days. The bivariate results presented in Table 1 indicate four individual characteristics that were most likely to result in the provision of post-discharges services in this population. These were the presence of four or more diagnoses of a chronic condition, at least two complex chronic conditions, or extreme illness severity. Table 2 presents information, not on issues of chronic illness, but on the general nature of the patients' principal diagnosis as specified in the discharge record. For the 16 categories included in our analysis, mental health issues and respiratory problem were the most prevalent. Injuries, perinatal issues, neoplasms, and circulatory problems were the most likely to receive post-discharge services. Pregnancy and hematological problem were the least likely to receive PAC. In the earlier multi-state research on PAC for the general
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
5
Table 2 Principal Presenting Problem for Children with SHCN Discharged from Hospital in Texas With or Without Post-Discharge Services, 2011e2014. Principal Presenting Problem
Hematology Circulatory Congenital defect Digestive Endocrine Genital Infectious Injury Musculoskeletal Neoplasm Nervous Skin Perinatal Pregnancy Respiratory Mental health
Population N ¼ 472,825 (Col. %)
2.33 1.32 4.47 4.46 3.60 1.88 1.99 4.09 1.60 1.37 4.38 1.13 2.08 2.12 13.90 19.94
Post-Acute Disposition (Col. %) Discharged Home with No Services (N ¼ 445,501) 94.2%
With Post-Discharge Health Services (N ¼ 27,324) 5.8%
97.4 91.5 93.7 94.6 96.1 96.5 92.3 84.2 92.3 91.4 94.7 95.5 90.0 99.1 95.5 95.4
2.6 8.5 6.3 5.4 3.9 3.5 7.7 17.6 7.7 8.6 5.3 4.5 10.0 0.9 4.5 4.6
pediatric population, the most common form of PAC was discharge home with home health services (84.2%). Only 17.6 percent of discharges with services resulted in a discharge to a health facility.3 The results presented here are quite different. Of those 5.8 percent of discharges with services, 66 percent resulted in post-discharge services in some type of HCF, while only 34 percent of discharges resulted in the provision of care in the home. Table 3 displays the multivariate results indicating what factors had a significant impact on the distribution of PAC to children with SHCN. The first model, estimating the odds of receiving any formal post-discharge services, had a pseudo-R2 of 0.14 and a C statistic of 0.79. The second model included only those discharges in which the patient received formal post-discharge services and examined the factors that affected whether those services were received in the home or in a HCF. That model had a pseudo-R2 of 0.36 and a C statistics of 0.88. Unlike other pediatric hospital discharges,3 among discharges involving children or youth with special health care needs, the number of chronic conditions had no significant impact on whether formal post-discharge services were provided. The same holds true for the model estimating the likelihood of receiving post-discharge services in a HCF versus in the home. Instead, the individual factors that had the greatest effect on the receipt of formal post-discharge services were the number of complex chronic conditions and the severity of the illness during the hospital stay. The presence of complex chronic conditions made it much more likely a patient would receive formal services after discharge. In the model for those receiving services, however, those with one of more complex chronic conditions were significantly less likely to receive services in a HCF. It was those with no complex chronic conditions or who had less than extremely serious illnesses that brought them to the hospital who were more likely to have received care in a HCF. Only those patients who presented with extremely severe conditions were much more likely (OR ¼ 2.28, CI 1.73e2.99) to receive post-discharge services in a health care setting. Sex and race/ethnicity had no significant effect on whether formal services were received after discharge from hospital or where it was received. Payment source had no effect on whether the patient received post-discharge care. However, those with no source of payment who received post-discharge services were much more likely (OR ¼ 1.89, CI ¼ 1.25e2.86) than those with
identifiable payors to receive post discharge care in a health facility than at home. The effect of age on the likelihood of receiving formal postdischarge care was somewhat complex. Infants (0e28 days) and older children and youth (10e17) had basically the same odds of receiving services. Those infants and children with ages in what might be considered the middle range (29 dayse9 years) had significantly lower odds of receiving services. However, the effect of age on where post discharge services were received was much simpler. Compared to those less than 29 days old, all other children and youth were less likely to receive PAC services in a health care setting. Teaching hospitals did not differ from other hospitals in their likelihood of discharging a child or youth with special needs with services, though they were significantly less likely to send a child to a HCF for services. Children's’ hospitals were significantly less likely than other acute care hospitals to provide a child with PAC services and less likely than other acute care settings to discharge a patient to a HCF. The effect of having a surgical procedure had no effect on whether a patient received services, but its presence significantly reduced the likelihood of any post discharge services being provided in a HCF. The effects of the patient's length of stay in hospital also followed that pattern. It had no effect of the likelihood of services, but, for those who did receive services, a longer stay reduced the odds of receiving services in a health care setting. Table 4 displays other results from the same equations as those estimated for Table 3, but the emphasis here is on the effect of different types of presenting, or principal, problems on decisions concerning care after discharge. Only three of these diagnosis categories had consistent results across both models. Circulatory problems, an injury, and the presence of neoplasms all increased the likelihood of services and of those services being provided in a heath care setting. The other ten types of primary problems exhibited different effects between the likelihood of receiving services and the likelihood of services being provided in a health setting. Notably, children or youth that presented with mental health problems represent a relatively large proportion of discharges. These patients are no more or less likely to receive PAC. However, when PAC is provided, the odds that it would be in a HCF seem almost ridiculously high (OR ¼ 35.7, CI 11.5e110.82). However, when one looks at those who received PAC and had a mental health
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
6
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
Table 3 Logistic Regression Results Reflecting Post-Discharge Service Results for Children with Special Health Care Needs. Variable
Odds of Being Discharged with Services pseudo-R2 ¼ 0.24/C ¼ 0.79 Odds-Ratio (95% CI)
Number of Chronic Disease Diagnoses One e Two 0.82 (0.57e1.20) Three 0.93 (0.67e1.30) Four or more 1.00 (0.64e1.57) Complex Chronic Conditions None e One 2.23*** (1.77e2.83) Two 3.16*** (2.26e4.40) Three or more 4.89*** (3.40e7.02) Illness Severity Mild e Moderate 1.47 (1.01e2.14) Serious 2.62*** (1.80e3.83) Extreme 6.51*** (3.80e11.16) Demographics Female 1.01 (0.88e1.16) White e Hispanic 0.70 (0.52e0.95) Black 0.94 (0.88e1.01) Asian 0.88 (0.75e1.05) Payor Other e Medicaid 1.11 (1.00e1.23) Charity 0.87 (0.48e1.56) Age Less than 28 days e 29-364 days 0.67** (0.50e0.89) 1-4 years 0.51*** (0.39e0.66) 5-9 years 0.51*** (0.38e0.67) 10-14 years 0.71 (0.54e0.93) 15-17 years 0.84 (0.64e1.11) Surgery 0.91 (0.75e1.11) Type of Acute Care Setting Other e Teaching 0.89 (0.60e1.33) Children's 0.41*** (0.29e0.60) Metropolitan Area 0.45*** (0.36e0.54) Length Stay (log) 1.19 (0.86e1.49) Time (quarter) 0.98 (0.96e1.01)
If Receiving Services, Odds of Receiving Services in a Heath Care Facility pseudo-R2 ¼ 0.36/C ¼ 0.88 Odds-Ratio (95% CI) e 0.92 (0.81e1.04) 0.90 (0.78e1.04) 0.79 (0.63e0.99) e 0.70*** (0.57e0.85) 0.50*** (0.41e0.60) 0.31*** (0.23e0.41) e 0.97 (0.83e1.13) 1.06 (0.90e1.25) 2.28*** (1.74e2.99) 1.03 (0.95e1.13) e 0.95 (0.80e1.11) 1.12 (0.98e1.27) 0.82 (0.66e1.02) e 0.92 (0.79e1.06) 1.89** (1.25e2.86) – 0.28*** (0.19e0.42) 0.15*** (0.11e0.22) 0.17*** (0.120.26) 0.27*** (0.18e0.41) 0.36*** (0.24e0.53) 0.73** (0.59e0.89) e 0.45** (0.25e0.79) 0.36*** (0.25e0.52) 0.30*** (0.18e0.50) 0.71** (0.61e0.84) 1.00 (0.97e1.03)
** p <. 01 *** p <. 001.
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
7
Table 4 Logistic Regression Results Related to The Child With Special Health Care Need's Principal Presenting Problem. Principal Presenting Problem
Odds Ratios for Receiving Any Services Odds-Ratio (95% C.I.)
Odds Ratios for Receiving Services in a Health Care Facility Odds-Ratio (95% C.I.)
Hematologic
0.70** (0.55e0.89) 1.82*** (1.37e2.41) 1.21 (0.96e1.52) 1.49** (1.14e1.94) 1.16 (0.82e1.66) 0.89 (0.62e1.27) 1.48** (1.12e1.94) 5.30*** (3.99e7.05) 2.92*** (2.15e3.95) 1.57** (1.19e2.06) 1.49 (0.97e2.28) 1.87*** (1.40e2.48) 1.04 (0.82e1.34) 0.23*** (0.14e0.39) 1.34 (0.96e1.88) 1.73 (0.75e4.02)
1.00 (0.63e1.58) 2.49*** (1.80e3.44) 0.96 (0.74e1.23) 0.57*** (0.44e0.75) 0.69 (0.48e0.99) 0.60*** (0.45e0.80) 0.64 (0.45e0.91) 2.01*** (1.50e2.68) 0.43 (0.22e0.82) 2.28*** (1.48e3.51) 1.61** 2.01e4.81 0.30** (0.15e0.60) 1.04 (0.78e1.38) 0.62 (0.36e1.08) 0.54** (0.38e0.77) 35.74*** (11.53e110.82)
Circulatory Congenital defect Digestive Endocrine Genital Infectious Injury Musculoskeletal Neoplasm Nervous Skin Perinatal Pregnancy Respiratory Mental health ** p <. 01 *** p <. 001.
issue as their principal presenting problem, almost 99 percent received their PAC in a HCF. Discussion While considerable health services research has focused on PAC for older persons, very little has addressed the post-acute care services received by other segments of the population. Recent research has contributed to our knowledge on PAC for younger persons in general.3 The research presented here constitutes an attempt to add to the more general issue of pediatric PAC a similar concern with post discharge services for some of the most vulnerable of younger persons, children with SHCN. The results presented here indicate that the decision-making concerning post discharge services for children with SHCN may differ in many ways from those dynamics in the more general pediatric population. Berry and his colleagues found that the number of chronic condition was a major predictor of the type of PAC received,3 while among discharges for children with SHCN, the number of conditions was not a significant predictor of whether or where PAC was received. Earlier research found differences based on the patient race or ethnicity that were not observed in this research. The effects of age and payer also were more complex in these data than in those concerning general pediatric post-acute care. The seriousness of the illness that precipitated the acute care stay, a previously unexplored factor in pediatric discharges, and the number of complex chronic conditions were, for this population, the major individual characteristics affecting decisions about services following an acute care stay.
The nature of the acute care setting was also important. Urban hospitals were less likely to use PAC. The average acute care hospital was significantly more likely to discharge children with SHCN with services than were teaching or children's hospitals. Also, general hospitals and hospitals in rural areas were significantly more likely to use post-discharge services in health care facilities. The results noted above are interesting, but limited to the effects of specific factors on PAC. One more general conclusion, however, seems discernable from these results. The use of post-acute care services for children with SHCN appears to be surprisingly low in data focusing specifically on a relative fragile population with chronic conditions, many with multiple complex conditions or severe illnesses. The 5.8 percent of discharges for children with SHCN receiving PAC in these data is approximately five percent lower than the overall prevalence (6.1%) of PAC for general pediatric discharges observed in the data from 44 states.3 It is also almost ten percent lower than the prevalence of PAC (6.4%) observed in a database of pediatric hospital discharges in Texas constructed to mirror the 44state data (authors' analysis). This result raises questions of possible under-utilization of PAC for children with SHCN that this research cannot, unfortunately, address. What the result suggests is a pressing need for outcome studies. The question of under-utilization of PAC for children with SHCN can only be investigated by research focusing on the effects in this special population of going home after an acute care stay without services versus receiving various types and levels of services after an acute care discharge. In the 44-state data only 17.6 percent of discharges from the general children population resulted in the provision of services in
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
8
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9
a health care setting.3 In the Texas data, post-acute treatment of 66 percent of discharges for children with SHCN was in health care facilities. One source of this difference may be that the earlier study excluded newborns and patients with a mental health diagnosis. However, even when those two types of patients are excluded from the Texas database used in this research, 31 percent of discharges for children with SHCN who received PAC received those services in a health care setting (authors' analysis). It may be that in all settings, children with SHCN, unlike the more general pediatric population, are significantly more likely to receive services after an acute care stay in a HCF than they are to receive services at home. However, whether the prevalence is 66 percent or 31 percent, such results fly in the face of the near universal professional judgement that any care other than acute care for this population is best provided in the home.22 Studies of children with SHCN in other settings or states might reveal whether the Texas results are anomalous or are in fact generally reflective of PAC for patients in this population. If the results of similar studies in other settings indicate that these results are not anomalous, then smaller scale studies would seem to be required to better understand the use of post-discharge services for children with SHCN. These more-limited studies might involve quantitative work with more finely-grained data about patient conditions and service providers. Qualitative research that would provide more information about the dynamics of decision-making involving this special population could also prove quite useful. In this vein, both qualitative and quantitative (mixed-method) studies investigating the role of the availability of services, patients' and families' preferences, as well as the family's ability and willingness to assist in needed care, might be important considerations.14,15 However, the results presented here may differ significantly from similar studies in other sites e other studies might find very different prevalence for the use of PAC and the sites in which it is delivered. Should that be the case, then one suspects a search would begin for what aspects of the broader context affect different patterns of PAC for children with special health care needs. One obvious axis of such analyses might be variation in the availability of different types and intensities of PAC services across settings. Another issue might be the degree to which PAC has been substituted for inpatient care in these different settings.32 Other investigators would do well to be mindful of the limitations of this study. The study database was limited to a three-year period in one state, albeit one with a large population. The use of state administrative data gathered largely for the purposes of payment limited the amount of clinical and provider detail available on each discharge. As a purely quantitative secondary analysis, this research lacks the richness that might be found in a mixedmethods approach that integrated qualitative insights into the quantitative element of such research. Such research might help with understanding the results observed for age, type of hospital, and presenting problems. In addition, our multivariate models addressing these issues need improvement in later research efforts. Both logistic regression models displayed only a moderate ability to make correct predictions. The positive predictive value for the model distinguishing between no formal care and the provision of formal care had a positive predictive value of only 48.1 percent and correctly predicted the outcome 94.2 percent of cases. The model differentiating between sites of formal care had a positive predictive value of 69.5 percent and correctly predicted 78 percent of case outcomes. Further testing of the models indicated that the models may not be properly specified.31 These tests other research suggest that the models used in this research might benefit from additional data drawn from other sources.32
With these limitations in mind, this research offered a useful strategy for classifying and analyzing the distribution of pediatric PAC services, which may be of assistance to other investigators. It also made at least an initial contribution to an important issue that has, as of yet, received little attention in health services research e decisions concerning the provision of different types of health services to children with SHCN, a vulnerable population composed of relatively heavy users both of formal health services and family resources. Conflicts of interests The authors have no conflicts-of-interests, and no external or special funding was used to support this research. References 1. Morley M, Bogasky S, Gage B, Flood S, Ingber MJ. Medicare post-acute care episodes and payment bundling. Medicare Medicaid Res Rev. 2014;4(1). mmrr.004.01.b02. 2. Chandra A, Dalton MA, Holmes J. Large increases in spending on postacute care in Medicare point to the potential for cost savings in these settings. Health Aff. 2013;32(5):864e872. 3. Berry JG, Hall M, Dumas H, et al. Pediatric hospital discharges to home health and postacute facility care: a national study. JAMA Pediatr. 2016;170(4): 326e333. 4. Children and Adolescent Health Measurement Institute. Who Are Children with Special Health Care Needs (CSHCN). Data Resource Center supported by Cooperative Agreement 1-U59-MC06980-01 from the U.S. Department of Health and Human Services. Health Resources and Services Adminstration (HRSA), Maternal and Child Health Bureau (MCHB); 2012. Available at: www. childheath.org. Accessed December 30, 2016. Revised 4/3/2012. 5. Willits KA, Platonova EA, Nies MA, Racine EF, Troutman ML, Harris HL. Medical home and pediatric primary care utilization among Children with Special Health Care Needs. J Pediatr Health Care. 2013;27(3):202e208. 6. Bethell CD, Newacheck PW, Fine A, et al. Optimizing health and health care systems for Children with Special Health Care Needs using the life course perspective. Mater Child Health J. 2014;18(2):467e477. 7. Van Cleave J, Boudreau AA, McAllister J, Cooley WC, Maxwell A, Kuhlthau K. Care coordination over time in medical homes for children with special health care needs. Pediatrics. 2015;135(6):1018e1026. 8. Schieve LA, Gonzalez V, Boulet SL, et al. Concurrent medical conditions and health care use and needs among children with learning and behavioral developmental disabilities, National Health Interview Survey, 2006e2010. Res Dev Disabil. 2012;33(2):467e476. 9. Calcedo C. Children with Special Health Care Needs: child health and functioning outcomes and health care service use. J Pediatr Health Care. 2016;30(6): 590e598. 10. Newacheck PW, Kim SE. A national profile of health care utilization and expenditures for Children with Special Health Care Needs. Arch Pediatr Adolesc Med. 2005;159(1):10e17. 11. Cohen E, Berry JG, Camacho X, Anderson G, Wodchis W, Guttmann A. Patterns and costs of health care use of children with medical complexity. Pediatrics. 2012;130(6):e1463ee1470. 12. Agrawal R, Hall M, Cohen E, et al. Trends in health care spending for children in Medicaid with high resource use. Pediatrics. 2016;138(4):e20160682. 13. Miller TR, Elliott TR, McMaughan DM, et al. Personal care services provided to children with special health care needs (CSHCN) and their subsequent use of physician services. Disabil Health J. 2013;6(4):317e324. 14. Romley JA, Shah AK, Chung PJ, Elliott MN, Vestal KD, Schuster MA. Familyprovided health care for children with special health care needs. Pediatrics. 2017;139(1):e20161287. 15. Cousino MK, Hazen RA. Parenting Stress among caregivers of children with chronic illness: a systematic review. J Pediatr Psychol. 2013;38(8):809e828. 16. Phillips CD. What do you do for a living: toward a more succinct definition of health services research. BMC Health Serv Res. 2006;6:117. 17. Bowles KH, Holmes JH, Ratcliffe SJ, Liberatore M, Nydick R, Naylor MD. Factors identified by experts to support decision making in post acute referral. Nurs Res. 2009;58(2), 115e112. 18. Burke RE, Juarez-Colunga E, Levy C, Prochazka AV, Coleman EA, Ginde AA. Patient and hospitalization characteristics associated with increased post-acute care facility discharges from US hospitals. Med Care. 2015;53(6):492e500. 19. Phillips CD, Truong C, Kum H, Nwaiwu O, Ohsfeldt O. Post-Acute care for children and youth in Texas, 2011-2014. Clin Med Insights Pediatr. 2017;11. http://dx.doi.org/10.1177/1179556517711445. 20. Gage B, Morley M, Spain P, Ingber M. Final Report: Post Acute Care in an Integrated Hospital System. 2009. Waltham, MA: RTI International; 2009. 21. Texas Department of State Health Services. Hospital Inpatient Public Use Data File. [accessed on May 5, 2016]. Available at: https://www.dshs.texas.gov/thcic/ hospitals/Inpatientpudf.shtm.
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010
C.D. Phillips et al. / Disability and Health Journal xxx (2017) 1e9 22. Committee on Children with Disabilities. American Academy of Pediatrics: guidelines for home care of infants, children, and adolescents with chronic disease. Pediatrics. 1995;96(1):161e164. 23. Texas Department of State Health Services. User Manual: Texas Hospital Inpatient Discharge File (PUDF); 2015. Available at: https://www.dshs.texas.gov/ thcic/hospitals/Inpatientpudf.shtm. Accessed April 14, 2016. 24. McPherson M, Arango P, Fox H, et al. A new definition of children with special health care needs. Pediatrics. 1998;102:137e140. 25. Healthcare Cost and Utilization Project (HCUP). HCUP Chronic Condition Indicator; 2016. Available at: https://www.hcupus.ahrq.gov/toolssoftware/chronic/ chronic.jsp. Accessed October 1, 2016. 26. Feudtner C, Feinstein JA, Zhong W, Hall M, Dai D. Pediatric complex chronic conditions classification system version 2: updated for ICD-10 and complex medical technology dependence and transplantation. Biomed Cent Pediatr.
9
2014;14(199):1. 27. United States Department of Agriculture Economic Research Service. Ruralurban Continuum Codes; 2013. Available at: https://www.ers.usda.gov/dataproducts/rural-urban-continuum-codes/. Accessed October 22, 2016. 28. StataCorp. STATA (Release 14) Statistical Software. College Station, TX: StataCorp LP; 2015. 29. Mittelhammer, Ron C, Judge, George G, Miller, Douglas J. Econometric Foundations. Cambridge: Cambridge University Press; 2000. 30. Perneger T. What's wrong with Bonferroni adjustments? BMJ. 1998;316, 1236e8. 31. Hosmer DW, Lemeshow S. Applied Regression Analysis. second ed. New York: John Wiley & Sons; 2000. 32. Buntin MB, Garten Ad, Paddock S, Saliba D, Totten M, Escare JJ. How much of post acute use is affected by its availability. HSR. 2005;40(2):413e434.
Please cite this article in press as: Phillips CD, et al., Post-acute care for children with special health care needs, Disability and Health Journal (2017), http://dx.doi.org/10.1016/j.dhjo.2017.08.010