- Email: [email protected]
Do all-cause revisit rates reflect the quality of pediatric surgical care provided during index encounters?
Journal of Pediatric Surgery 52 (2017) 1050–1055
Contents lists available at ScienceDirect
Journal of Pediatric Surgery journal homepage: www.elsevier.com/locate/jpedsurg
Do all-cause revisit rates reflect the quality of pediatric surgical care provided during index encounters?☆ Danielle B. Cameron a, Dionne A. Graham b, Carly E. Milliren b, Stephanie Serres a, Charity C. Glass a, Adam B. Goldin c, Shawn J. Rangel a,⁎ a b c
Department of Surgery, Boston Children's Hospital - Harvard Medical School, Boston, MA Center for Applied Pediatric Quality Analytics, Boston Children's Hospital, Boston, MA Division of General and Thoracic Surgery, Seattle Children's Hospital - University of Washington School of Medicine, Seattle, WA
a r t i c l e
i n f o
Article history: Received 18 February 2017 Accepted 9 March 2017 Key words: Readmission Pediatric surgery Comparative reporting Revisit diagnoses Quality metric
a b s t r a c t Purpose: The purpose of this study was to compare the relatedness of revisits to the index surgical encounter across different pediatric surgical procedures and to explore whether all-cause revisit rates are an accurate surrogate measure for related revisits in this cohort of children. Methods: We reviewed all-cause revisits occurring within ninety days of the thirty most commonly performed pediatric surgical procedures at 44 children's hospitals between 1/1/2012 and 3/31/2015. For each condition, a team of four surgeons reviewed revisit diagnoses and reached consensus around relatedness to the index surgical encounter. Chi-squared tests were used to test for variation in all-cause and related revisits among procedures. Spearman's correlation coefficient was used to measure the association between rankings of procedures by their all-cause and related revisit rates. Results: 144,535 index encounters were analyzed with an overall revisit rate of 15.0% (21,732). Significant variation was found in both the rates of all-cause revisits among procedures (ranges: 7.6–68.4%, p b 0.0001), and in the relative proportions of revisits related the index surgical encounter (range: 0% to 77%, p b 0.0001). Poor correlation was found between procedure rankings based on all-cause revisit rates and revisit rates related to the index admission (r = 0.33, p = 0.07). Conclusions: The relative proportion of revisits related to the index encounter varies significantly across pediatric surgical conditions, and poor correlation exists at the procedure-level between all-cause and related revisits rates. Level of evidence: IV. © 2017 Elsevier Inc. All rights reserved.
Hospital readmission poses a significant financial burden for hospital systems, patients, and payors alike [1–3]. Reduction of inpatient and emergency department (ED) revisits has therefore become a high-priority focus of many hospital-wide quality improvement (QI) initiatives. While all-cause revisit rates may reflect the quality of broader systems of care within hospitals, they may not accurately reflect the quality of care provided by the surgical team during and following the index surgical admission [4–6]. In this regard, there are many reasons that a child might return to the system following discharge. These include complications of the surgical procedure, complications or scheduled revisits associated with an underlying chronic medical condition (e.g. oncologic diagnoses), or reasons entirely
☆ Source of Funding and Potential Conflict of Interest: No additional funding was obtained for this project. No conflicts of interest were reported by any of the study's authors. ⁎ Corresponding author at: Department of Surgery, Children's Hospital Boston, Harvard Medical School, 300 Longwood Ave. Fegan-3, Boston, MA, 02115. Tel.: +1 617 355 3040; fax: +1 617 730 0298. E-mail address: [email protected] (S.J. Rangel). http://dx.doi.org/10.1016/j.jpedsurg.2017.03.034 0022-3468/© 2017 Elsevier Inc. All rights reserved.
unrelated to the index surgical procedure or underlying chronic condition. The relative distribution of these revisit diagnoses is likely to be quite different across the spectrum of pediatric surgical procedures, bringing into question the validity of using all-cause revisit rates as a surrogate for revisits truly related to the index surgical encounter. With the considerations above, the purpose of the present study was two-fold. First, we sought to characterize the relatedness of revisit encounters to the index admission for commonly performed general pediatric surgical procedures. Second, we aimed to explore whether all-cause revisit rates are an accurate surrogate for revisits considered related to the index surgical encounter. We hypothesized that great variation in the rates of related revisits would be found across different pediatric surgical procedures, and furthermore, that correlation between rates of all-cause and related revisits would be relatively poor. The practical implications of this analysis were to identify procedures where the greatest relative opportunities for revisit prevention by the pediatric surgical team might exist in perioperative and postdischarge care. Furthermore, the results of this study could be used to provide a framework for meaningful comparative performance among
D.B. Cameron et al. / Journal of Pediatric Surgery 52 (2017) 1050–1055
pediatric surgery departments on the basis of revisits relating to surgical care. 1. Methods 1.1. Study design and data source This study was a retrospective cohort analysis using the Pediatric Health Information System (PHIS) database, which contains detailed administrative and billing data from 44 freestanding children's hospitals affiliated with the Children's Hospital Association (Overland Park, KS). Patient-level data available in PHIS includes demographic and payer information, primary and secondary International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic and procedural codes, and date-stamped billing data for a wide range of clinical services including diagnostic tests and therapeutic procedures. All patients in PHIS are assigned a unique identifier that can be used to track patients through subsequent readmission and revisit encounters. Data quality is facilitated by systematic monitoring including consistency reviews, coding consensus meetings, and quarterly data reports. 1.2. Study cohort We identified all patients undergoing surgery for one of the 30 most commonly performed general pediatric procedures from 1/1/2012 to 3/ 31/2015. We chose to focus on the top 30 procedures because of a preliminary analysis that demonstrated that this cohort accounted for greater than 80% of case volume in general pediatric surgery during the study period. The data query was restricted to index encounters with a preoperative length of stay of 2 days or less in order to exclude patients where the indication for hospitalization may have been related to reasons other than those associated with the operative procedure. All patients meeting inclusion criteria were assessed for a revisit to the emergency department or readmission to the inpatient setting within 90 days of discharge from the index encounter.
1051
appendectomy). A revisit diagnosis was considered very unlikely to be related to index surgical care if the surgical procedure would likely not have contributed to the diagnosis (e.g. otitis media following appendectomy). 1.5. Analysis of all-cause revisit rates and the relative proportion of related revisits All-cause revisit rates for each procedure were calculated by dividing the number of patients that had at least one revisit (ED or inpatient) within ninety days of discharge from the index admission by the total number of patients undergoing the procedure. Only the first revisit occurring within the 90-day follow-up period was included in the analysis. We then calculated the proportion of related revisits for each condition based on revisit encounters that were categorized as very likely related. Chi-Squared tests were used to test for variation in the proportion of related revisits across procedures. Spearman's correlation coefficient was used to measure the association between rankings of procedures by their all-cause and related revisit rates. In order to provide a more conservative assessment as to whether all-cause revisit rates were an accurate surrogate for related revisit rates, we repeated the analysis after redefining relatedness to include revisit encounters that were categorized as possibly related as well as very likely related. All statistical analysis was performed using SAS v. 9.4 (SAS Institute; Cary, NC). The Institutional Review Board of Boston Children's Hospital deemed this study exempt from review under 45 CFR 46.102(f). 2. Results We analyzed 144,535 index surgical encounters meeting inclusion criteria, of which 15.0% (21,732) had at least one revisit within 90 days of discharge. Of these, 13,860 encounters were associated with a revisit diagnosis that contributed more than 1% of all revisit encounters for a specific procedure. (Fig. 1). 2.1. All-cause revisit rates and the relative proportion of related revisits
1.3. Diagnosis-procedure pairs as the unit basis for readmission analysis Prior to analyzing revisit rates, we first reviewed all ICD-9 procedure and diagnosis codes associated with index encounters to create clinically relevant diagnosis-procedural pairings for each condition. This was necessary to distinguish different disease processes requiring the same procedure that may have different implications for readmission risk (e.g. colectomy for Hirschsprung's disease versus colectomy for ulcerative colitis). Encounters where more than one of the 30 targeted procedures was performed at the same time were excluded in order to provide a more accurate assessment of revisit risk attributable to specific procedures. The single exception to this exclusion was an encounter when a gastrostomy was combined with a fundoplication. We included these cases as a unique cohort as we felt that the risk of revisits was particularly high owing to the procedures and unique cohort of children undergoing these procedures. 1.4. Defining relatedness to index surgical encounter For each condition, a team of four surgeons reviewed principal ICD-9 diagnosis codes that were associated with at least 1% of all revisit encounters for that condition, and then established consensus on the degree of relatedness to the index surgical encounter. The surgical team met on a weekly basis to collectively review revisit diagnoses and assign one of three possible categories of relatedness: very likely, possibly, or very unlikely related. A revisit diagnosis was considered very likely related if it could not have occurred without the index surgical procedure (e.g. wound infection following appendectomy) and possibly related if the diagnosis could be attributed to either a post-operative complication or unrelated condition (e.g. fever following
All-cause revisit rates ranged from 7.6% to 68.4% across procedures, with the highest rates associated with ulcerative colitis (68.4%), hepatoblastoma (65.3%), fundoplication combined with gastrostomy (63.6%), neuroblastoma (61.4%), and Wilms tumors (54.6%). Conditions with the lowest all-cause revisit rates included undescended testes (7.6%), umbilical hernia (7.9%), inguinal hernia (8.9%), pectus excavatum (10.3%), and ovarian torsion (10.6%). Of the 13,860 revisit encounters reviewed, 37.8% (5236) were categorized as very likely related, 43.0% (5967) possibly related, and 19.2% (2657) very unlikely related. The proportion of revisits likely related to the index hospitalization was significantly different across procedures (range: 0% to 77%, p b 0.0001), with the highest proportions associated with complicated appendicitis (77%), imperforate anus (66%), and uncomplicated appendicitis (61%) (Fig. 2). When using the more conservative definition of relatedness that included revisit diagnoses that were categorized as possibly related as well as very likely related, the proportion of related revisits remained significantly different across procedures (range: 28% to 100%, p b 0.0001). All revisits following management of complicated appendicitis, pectus excavatum, and pilonidal disease were considered related using this definition, and very high rates of related revisit rates were also found with Hirschsprung's disease (97.8%), ovarian cyst (97.4%), imperforate anus (96.1%), and uncomplicated appendicitis (93.1%). 2.2. Correlation between all-cause revisit rates and related revisits Revisit rates very likely related to the index surgical admission were on average 83% lower than all-cause revisit rates (range of decrease by procedure: 39–100%), and relatively poor correlation was found
1052
D.B. Cameron et al. / Journal of Pediatric Surgery 52 (2017) 1050–1055
Fig. 1. Assembly of study cohort. *Readmission encounters that were reviewed to determine relatedness included those where the revisit diagnosis accounted for at least 1% of all revisits for a specific procedure.
between ranking of procedures based on all-cause revisit rates and by revisit rates likely related to index admission (r = 0.33, p = 0.07). When using the more conservative definition of relatedness, the correlation was relatively stronger but remained modest across all procedures (r = 0.79, p b 0.001). The relationship between all-cause readmission rates and the proportion of revisits categorized as being related is presented in Fig. 3. 3. Discussion In this retrospective analysis of 144,535 encounters from 44 children's hospitals, the proportion of revisits related to the index surgical encounter varied significantly across procedures and relatively poor correlation was found between all-cause and related revisit rates. Furthermore, the correlation between rates of all cause and related revisits remained relatively weak even after redefining related
diagnoses to include those categorized as only possibly related. In aggregate, these results provide a compelling argument that all-cause revisit rates are a relatively poor surrogate for revisit rates plausibly related to the index surgical encounter. Previous studies have characterized variation in pediatric surgical readmission rates and factors associated with readmission; however, these studies have either been limited to single center analyses or relatively few pediatric surgical conditions [7–9]. Data surrounding the relationship of revisit encounters to care provided during index pediatric surgical care remain limited. Burjonrappa and colleagues conducted a review of 2217 pediatric surgical admissions at two tertiary-referral children's centers and characterized revisit rates, causes of readmission, and revisit length of stay. Of the 145 readmissions analyzed, less than half of revisits were found to be related to the procedure performed during the index admission. Appendectomy and procedures associated with the management of intestinal obstruction were among the
Fig. 2. Relative proportion of revisits categorized as either being very likely related, possibly related, or likely unrelated to the index surgical encounter for the thirty most common pediatric surgical conditions performed during the study period. Abbreviations: TEF/EA – tracheoesophageal fistula/esophageal atresia, CCAM – congenital cystic adenomatoid malformation, GT – gastrostomy tube, CDH – congenital diaphragmatic hernia.
D.B. Cameron et al. / Journal of Pediatric Surgery 52 (2017) 1050–1055
1053
Fig. 3. Association between all-cause revisit rates and revisit rates related to care provided during the index surgical encounter. For the analysis represented in the figure, related revisit diagnosis were defined as those categorized as being very likely related or possibly related. Abbreviations: TEF/EA – tracheoesophageal fistula/esophageal atresia, CCAM – congenital cystic adenomatoid malformation, GT – gastrostomy tube, CDH – congenital diaphragmatic hernia.
operations most likely to be associated with procedure-related revisits in their study [10]. Our current analysis suggests similar findings; however our primary aim was to characterize relatedness across a much broader spectrum of pediatric surgical procedures using a more granular assessment of relatedness for this purpose. Furthermore, our study cohort was derived from 44 children's hospitals, providing the most comprehensive and generalizable assessment of revisit relatedness associated with pediatric surgical practice to date. Previously published studies have also challenged the validity of readmission rates as a quality measure for care provided during the index hospitalization, suggesting that revisits may be more dependent on non-modifiable factors rather than on the care provided during the index hospitalization [5,11–13]. In the context of pediatric surgical conditions, these considerations may have particular relevance when assessing quality across different conditions where the rates of related revisits and all-cause revisits vary considerably. For example, all-cause readmission rates may more directly reflect the quality of care provided during the index encounter for conditions such as appendicitis and pilonidal disease where nearly all revisits were found to be related. This is in contrast to oncologic surgical conditions where previous studies have shown that all-cause revisit rates are often unrelated to care provided during the index hospitalization [14,15]. Our results were consistent with these earlier findings, with hepatoblastoma, Wilms tumor, and neuroblastoma all having among the highest all-cause revisit rates in our study but relatively modest related revisit rates. These findings may be somewhat intuitive when considering the relatively high proportion of patients in this cohort that are readmitted for scheduled adjuvant therapy (as well as the complications that arise from such therapy). Discordance between all-cause and related revisit rates were also found for procedures associated with underlying chronic medical conditions, many of which are associated with revisits and readmissions [16–18]. For example, a child with neurological impairment who undergoes gastrostomy tube placement might return to the ED with gastrostomy related complications (e.g. gastrostomy site infection or dislodgement). Alternatively, the same patient might be readmitted following a seizure that is likely unrelated to surgery but related to the underlying condition. The former exemplifies a scenario that might be
addressed through QI initiatives targeted at surgical care pathways and post-discharge protocols for gastrostomy care, while the latter relates to quality assurance and management considerations outside of the scope of pediatric surgical practice. The results of this study may have important implications for the prioritization of efforts to reduce revisits within the influence of pediatric surgical practice, and for meaningful comparative performance reporting among pediatric surgery departments on the basis of revisit rates. With respect to the former, identification of conditions where both the rate and burden of related revisits are relatively high (e.g. appendicitis) may guide prioritization of readmission reduction efforts within and across surgery departments. With respect to comparative performance reporting, data from this study may be used to identify a portfolio of procedures where allcause revisit rates may be a reasonable surrogate for related revisit rates, while at the same time also identifying procedures where caution should be exercised when extrapolating all-cause revisit rates as a potential reflection of surgical quality. The results of this study must be considered in the context of its limitations. These data were derived from a network of free-standing children's hospitals and may not be generalizable to other hospitals. Although the PHIS database is subject to continuous quality assurance audits, administrative coding errors and misclassification of revisit diagnoses and procedures are possible. With respect to our study cohort, we limited the analysis to the 30 most commonly performed general pediatric surgical procedures, and further limited our analysis to revisit diagnoses that contributed at least 1% of all revisit encounters for specific procedures. Although this approach did not capture all pediatric surgical cases, we do believe it is comprehensive in that it captured more than 80% of all pediatric surgical cases during the study period and strongly supports our hypothesis that all-cause revisit rates are not an accurate surrogate for those related to care during the index encounter. By limiting the preoperative length of stay to 2 days or less, we may be undercounting semi-urgent procedures (and therefore revisits attributable to these procedures) where preoperative treatment delay may occur because of prolonged diagnostic evaluation and scheduling logistics (e.g. pull-through procedure for newly diagnosed Hirschsprung's
1054
D.B. Cameron et al. / Journal of Pediatric Surgery 52 (2017) 1050–1055
disease). With respect to our cohort definitions, we found a relatively large proportion of revisits categorized as “possibly” related for many surgical procedures. This observation reflects the non-specific nature of many symptom-based readmission diagnoses (e.g. fever and nausea), and highlights the challenges with using administrative data even when the diagnostic coding is accurate. We attempted to adjust for this uncertainty by using a more “conservative” definition of relatedness as a sensitivity analysis. Even when using this more conservative definition, correlation between all-cause and related revisit rates remained only modest and significant variation persisted among procedures with respect to their related revisit rates. In conclusion, significant variation exists in the rates of related revisits for pediatric surgical procedures, and all-cause revisit rates appear to be poor surrogates for revisit rates that are plausibly related to care provided during the index hospitalization. These findings may have important implications for the prioritization of conditions for preventing readmission encounters sensitive to surgical care, and for fair comparative reporting of revisit rates among surgical departments. Finally, it is important to note that relatedness of revisits is not necessarily synonymous with preventability. Future efforts will need to characterize preventability within the cohort of procedures where the rates and burden of related revisits and readmissions are particularly high. Appendix A. Discussions Danielle B. Cameron, MD, MPH DANIELLE B. CAMERON, MD, MPH: Members and guests, thank you for the opportunity to present our work. I have no disclosures. Hospital revisit rates are considered important measures of quality. While all-cause revisit rates may reflect the quality of broader systems of care, they may not be an accurate reflection of the quality of care provided by the surgical team during and following the index admission. In this regard, little data exists surrounding the relationship between revisits and care provided during the index pediatric surgical encounter. With these considerations, the purpose of our study was two-fold. First, we sought to characterize the relatedness of revisit encounters to the index admission for pediatric surgical procedures. Second, we aimed to explore whether all cause revisit rates accurately reflect revisit rates plausibly related to the index surgical encounter. Using the Pediatric Health Information System (PHIS) database we retrospectively reviewed 90-day emergency department and inpatient revisits following index encounters for the 30 most commonly performed pediatric surgical procedures between January 1, 2012 and March 31, 2015. For each condition a team of four surgeons reviewed revisit diagnoses to determine relatedness to care provided during the index surgical encounter. A revisit diagnosis was considered very likely related if it could not have occurred without the index surgical procedure, (e.g. wound infection following appendectomy). Diagnoses considered possibly related were those that could be attributed to either a post-operative complication or an unrelated condition (e.g. fever following appendectomy). Diagnoses considered very unlikely to be related were those that were implausibly related to the surgical procedure (e.g. otitis media following appendectomy). We used Chi-squared tests to characterize variation in rates of all cause and related revisits among procedures. Spearman's correlation coefficient was used to explore whether all-cause revisit rates were an accurate surrogate for revisit rates related to care provided during the index surgical encounter. We analyzed over 140,000 index surgical encounters of which
15% had at least one revisit within 90 days of discharge. All cause revisit rates ranged from 7.6 to 68.4% across procedures, and this figure illustrates variation in the proportion of revisits across procedures likely related to care provided during or after the index surgical hospitalization. Conditions with the highest relative proportions of related revisits shown in pink and towards the left side of the figure were appendicitis and imperforate anus. Conditions with revisits very unlikely to be related located on the right side of the figure were oncologic conditions such as Wilm's tumor and neuroblastoma. This bubble plot illustrates the relationship between all-cause revisit rates on the Y axis and the proportion of revisits likely or possibly related to care provided during the index surgical encounter on the X axis. For this graph we combined very likely and possibly related diagnoses as a more conservative estimate of relatedness. Overall there was variable correlation between all cause and related revisits, with nearly perfect correlation for conditions such as appendicitis and pilonidal disease on the right side of the figure, while relatively poor correlation was seen for oncologic conditions such as hepatoblastoma and those associated with chronic underlying medical conditions (e.g. CDH and children requiring antireflux procedures) seen on the left side of the figure. The correlation between all-cause and related revisit rates across all conditions was 0.79, suggesting that all-cause revisits are a relatively week surrogate for related revisits in pediatric general surgery. The results of this study must be considered in the context of its limitations. The analysis was derived from a cohort of free-standing children's hospitals and thus the results may not be generalizable. In addition, administrative data is vulnerable to coding errors and misclassification of revisit diagnoses and procedures. Despite these limitations, we can conclude that significant variation exists in the relatedness of revisits for pediatric surgical procedures and all-cause revisit rates appear to be poor surrogates for revisits for related to care provided during the index hospitalization. Thank you for your attention. [applause] MODERATOR: This paper is open for discussion. I'll start. I wonder if you have any insight whether this could point towards potentially preventable causes of readmission? DR. CAMERON: Thank you for your question. The results of this study provide insight regarding which pediatric surgical procedures are more likely to be followed by revisits related to care provided during index surgical care and relatedness is not necessarily synonymous with preventability. However, for those conditions where there was a relatively high proportion of related revisits, these may serve as a good starting point for future efforts to characterize the nature of the revisit and whether it is potentially preventable. FEMALE VOICE: Thank you so much for this very elegant and very comprehensive view for our revisit. The question I have is why did you choose 90 days as opposed to the typical -30-day correlation, and if you looked at that, was there a difference in whether or not these are related or not related? The other thing is, are the revisits planned or unplanned because Wilm's tumors and – blastomas, they come back for their chemotherapy? Then the last thing is did you look to see whether
D.B. Cameron et al. / Journal of Pediatric Surgery 52 (2017) 1050–1055
or not there is clinical correlation? So for instance since this is an administrative database did you look at a biopsy of patients and look at certain institutions just to make sure that what was being coded in the administrative database is actually clinically relevant. Thank you again. DR. CAMERON: Thank you for your questions. The first question is why did we choose 90-day revisit intervals? We chose this because there are some conditions, particularly those related to chronic medical conditions like gastrostomy tubes, where the revisit period can extend far beyond the 30-day period, and may still have relevance to the pediatric surgical encounter. The second question was, are some of these planned revisits, and the answer is likely yes. We looked at all-cause revisits which contributes to why we see poor correlation for conditions such as oncologic conditions, where a large proportion of revisits are planned admissions for chemotherapy (and were therefore considered unrelated to index surgical care). Finally with regard to clinical correlation, we did not do any specific chart biopsies, but that would be a potential next step as part of a more focused analysis on preventability.
References [1] Berry JG, Toomey SL, Zaslavsky AM, et al. Pediatric readmission prevalence and variability across hospitals. JAMA 2013;309:372–80. http://dx.doi.org/10.1001/jama.2012.188351. [2] Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med 2009;360:1418–28. http://dx.doi. org/10.1056/NEJMsa0803563. [3] Nakamura MM, Toomey SL, Zaslavsky AM, et al. Measuring pediatric hospital readmission rates to drive quality improvement. Acad Pediatr 2014;14:S39–46. http://dx.doi.org/10.1016/j.acap.2014.06.012.
1055
[4] Bardach NS, Vittinghoff E, Asteria-Peñaloza R, et al. Measuring hospital quality using pediatric readmission and revisit rates. Pediatrics 2013;132:429–36. http://dx.doi. org/10.1542/peds.2012-3527. [5] Dimick JB, Ghaferi AA. Hospital readmission as a quality measure in surgery. JAMA 2015;313:512–3. http://dx.doi.org/10.1001/jama.2014.14179. [6] Feudtner C, Pati S, Goodman DM, et al. State-level child health system performance and the likelihood of readmission to children's hospitals. J Pediatr 2010;157:98–102. [7] Thibaudeau S, Anari JB, Carducci N, et al. 30-day readmission after pediatric upper extremity surgery: analysis of the NSQIP database. J Pediatr Surg 2016 Aug;51(8): 1370–4. http://dx.doi.org/10.1016/j.jpedsurg.2016.04.012. [8] Sinha CK, Decker E, Rex D, et al. Thirty-days readmissions in pediatric surgery: the first U.K. experience. J Pediatr Surg November 2016;51(11):1877–80. http://dx.doi. org/10.1016/j.jpedsurg.2016.06.015. [9] Rice-Townsend S, Hall M, Barnes JN, et al. Variation in risk-adjusted hospital readmission after treatment of appendicitis at 38 children's hospitals: an opportunity for collaborative quality improvement. Ann Surg 2013;257:758–65. http://dx.doi. org/10.1097/SLA.0b013e318268a663. [10] Burjonrappa S, Theodorus A, Shah A, et al. Pediatric surgery readmissions: a root cause analysis. Pediatr Surg Int 2015;31:551–5. http://dx.doi.org/10.1007/s00383015-3701-4. [11] Ludke RL, Booth BM, Lewis-Beck JA. Relationship between early readmission and hospital quality of care indicators. Inquiry 1993;30:95–103. [12] Benbassat J, Taragin M. Hospital readmissions as a measure of quality of health care: advantages and limitations. Arch Intern Med 2000;160:1074–81. [13] Berry JG, Ziniel SI, Freeman L, et al. Hospital readmission and parent perceptions of their child's hospital discharge. Int J Qual Health Care 2013;25: 573–81. http://dx.doi.org/10.1093/intqhc/mzt051. [14] Shulman DS, London WB, Guo D, et al. Incidence and causes of hospital readmission in pediatric patients after hematopoietic cell transplantation. - PubMed - NCBI. Biol Blood Marrow Transplant 2015;21:913–9. [15] Akenroye AT, Thurm CW, Neuman MI, et al. Prevalence and predictors of return visits to pediatric emergency departments. J Hosp Med 2014;9:779–87. http://dx.doi.org/10.1002/jhm.2273. [16] Berry JG, Hall DE, Kuo DZ, et al. Hospital utilization and characteristics of patients experiencing recurrent readmissions within children's hospitals. JAMA 2011;305: 682–90. http://dx.doi.org/10.1001/jama.2011.122. [17] Jurgens V, Spaeder MC, Pavuluri P, et al. Hospital readmission in children with complex chronic conditions discharged from subacute care. Hosp Pediatr 2014;4:153–8. http://dx.doi.org/10.1542/hpeds.2013-0094. [18] Cohen E, Berry JG, Camacho X, et al. Patterns and costs of health care use of children with medical complexity. Pediatrics 2012;130:e1463–70. http://dx.doi.org/10.1542/ peds.2012-0175.
Contents lists available at ScienceDirect
Journal of Pediatric Surgery journal homepage: www.elsevier.com/locate/jpedsurg
Do all-cause revisit rates reflect the quality of pediatric surgical care provided during index encounters?☆ Danielle B. Cameron a, Dionne A. Graham b, Carly E. Milliren b, Stephanie Serres a, Charity C. Glass a, Adam B. Goldin c, Shawn J. Rangel a,⁎ a b c
Department of Surgery, Boston Children's Hospital - Harvard Medical School, Boston, MA Center for Applied Pediatric Quality Analytics, Boston Children's Hospital, Boston, MA Division of General and Thoracic Surgery, Seattle Children's Hospital - University of Washington School of Medicine, Seattle, WA
a r t i c l e
i n f o
Article history: Received 18 February 2017 Accepted 9 March 2017 Key words: Readmission Pediatric surgery Comparative reporting Revisit diagnoses Quality metric
a b s t r a c t Purpose: The purpose of this study was to compare the relatedness of revisits to the index surgical encounter across different pediatric surgical procedures and to explore whether all-cause revisit rates are an accurate surrogate measure for related revisits in this cohort of children. Methods: We reviewed all-cause revisits occurring within ninety days of the thirty most commonly performed pediatric surgical procedures at 44 children's hospitals between 1/1/2012 and 3/31/2015. For each condition, a team of four surgeons reviewed revisit diagnoses and reached consensus around relatedness to the index surgical encounter. Chi-squared tests were used to test for variation in all-cause and related revisits among procedures. Spearman's correlation coefficient was used to measure the association between rankings of procedures by their all-cause and related revisit rates. Results: 144,535 index encounters were analyzed with an overall revisit rate of 15.0% (21,732). Significant variation was found in both the rates of all-cause revisits among procedures (ranges: 7.6–68.4%, p b 0.0001), and in the relative proportions of revisits related the index surgical encounter (range: 0% to 77%, p b 0.0001). Poor correlation was found between procedure rankings based on all-cause revisit rates and revisit rates related to the index admission (r = 0.33, p = 0.07). Conclusions: The relative proportion of revisits related to the index encounter varies significantly across pediatric surgical conditions, and poor correlation exists at the procedure-level between all-cause and related revisits rates. Level of evidence: IV. © 2017 Elsevier Inc. All rights reserved.
Hospital readmission poses a significant financial burden for hospital systems, patients, and payors alike [1–3]. Reduction of inpatient and emergency department (ED) revisits has therefore become a high-priority focus of many hospital-wide quality improvement (QI) initiatives. While all-cause revisit rates may reflect the quality of broader systems of care within hospitals, they may not accurately reflect the quality of care provided by the surgical team during and following the index surgical admission [4–6]. In this regard, there are many reasons that a child might return to the system following discharge. These include complications of the surgical procedure, complications or scheduled revisits associated with an underlying chronic medical condition (e.g. oncologic diagnoses), or reasons entirely
☆ Source of Funding and Potential Conflict of Interest: No additional funding was obtained for this project. No conflicts of interest were reported by any of the study's authors. ⁎ Corresponding author at: Department of Surgery, Children's Hospital Boston, Harvard Medical School, 300 Longwood Ave. Fegan-3, Boston, MA, 02115. Tel.: +1 617 355 3040; fax: +1 617 730 0298. E-mail address: [email protected] (S.J. Rangel). http://dx.doi.org/10.1016/j.jpedsurg.2017.03.034 0022-3468/© 2017 Elsevier Inc. All rights reserved.
unrelated to the index surgical procedure or underlying chronic condition. The relative distribution of these revisit diagnoses is likely to be quite different across the spectrum of pediatric surgical procedures, bringing into question the validity of using all-cause revisit rates as a surrogate for revisits truly related to the index surgical encounter. With the considerations above, the purpose of the present study was two-fold. First, we sought to characterize the relatedness of revisit encounters to the index admission for commonly performed general pediatric surgical procedures. Second, we aimed to explore whether all-cause revisit rates are an accurate surrogate for revisits considered related to the index surgical encounter. We hypothesized that great variation in the rates of related revisits would be found across different pediatric surgical procedures, and furthermore, that correlation between rates of all-cause and related revisits would be relatively poor. The practical implications of this analysis were to identify procedures where the greatest relative opportunities for revisit prevention by the pediatric surgical team might exist in perioperative and postdischarge care. Furthermore, the results of this study could be used to provide a framework for meaningful comparative performance among
D.B. Cameron et al. / Journal of Pediatric Surgery 52 (2017) 1050–1055
pediatric surgery departments on the basis of revisits relating to surgical care. 1. Methods 1.1. Study design and data source This study was a retrospective cohort analysis using the Pediatric Health Information System (PHIS) database, which contains detailed administrative and billing data from 44 freestanding children's hospitals affiliated with the Children's Hospital Association (Overland Park, KS). Patient-level data available in PHIS includes demographic and payer information, primary and secondary International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic and procedural codes, and date-stamped billing data for a wide range of clinical services including diagnostic tests and therapeutic procedures. All patients in PHIS are assigned a unique identifier that can be used to track patients through subsequent readmission and revisit encounters. Data quality is facilitated by systematic monitoring including consistency reviews, coding consensus meetings, and quarterly data reports. 1.2. Study cohort We identified all patients undergoing surgery for one of the 30 most commonly performed general pediatric procedures from 1/1/2012 to 3/ 31/2015. We chose to focus on the top 30 procedures because of a preliminary analysis that demonstrated that this cohort accounted for greater than 80% of case volume in general pediatric surgery during the study period. The data query was restricted to index encounters with a preoperative length of stay of 2 days or less in order to exclude patients where the indication for hospitalization may have been related to reasons other than those associated with the operative procedure. All patients meeting inclusion criteria were assessed for a revisit to the emergency department or readmission to the inpatient setting within 90 days of discharge from the index encounter.
1051
appendectomy). A revisit diagnosis was considered very unlikely to be related to index surgical care if the surgical procedure would likely not have contributed to the diagnosis (e.g. otitis media following appendectomy). 1.5. Analysis of all-cause revisit rates and the relative proportion of related revisits All-cause revisit rates for each procedure were calculated by dividing the number of patients that had at least one revisit (ED or inpatient) within ninety days of discharge from the index admission by the total number of patients undergoing the procedure. Only the first revisit occurring within the 90-day follow-up period was included in the analysis. We then calculated the proportion of related revisits for each condition based on revisit encounters that were categorized as very likely related. Chi-Squared tests were used to test for variation in the proportion of related revisits across procedures. Spearman's correlation coefficient was used to measure the association between rankings of procedures by their all-cause and related revisit rates. In order to provide a more conservative assessment as to whether all-cause revisit rates were an accurate surrogate for related revisit rates, we repeated the analysis after redefining relatedness to include revisit encounters that were categorized as possibly related as well as very likely related. All statistical analysis was performed using SAS v. 9.4 (SAS Institute; Cary, NC). The Institutional Review Board of Boston Children's Hospital deemed this study exempt from review under 45 CFR 46.102(f). 2. Results We analyzed 144,535 index surgical encounters meeting inclusion criteria, of which 15.0% (21,732) had at least one revisit within 90 days of discharge. Of these, 13,860 encounters were associated with a revisit diagnosis that contributed more than 1% of all revisit encounters for a specific procedure. (Fig. 1). 2.1. All-cause revisit rates and the relative proportion of related revisits
1.3. Diagnosis-procedure pairs as the unit basis for readmission analysis Prior to analyzing revisit rates, we first reviewed all ICD-9 procedure and diagnosis codes associated with index encounters to create clinically relevant diagnosis-procedural pairings for each condition. This was necessary to distinguish different disease processes requiring the same procedure that may have different implications for readmission risk (e.g. colectomy for Hirschsprung's disease versus colectomy for ulcerative colitis). Encounters where more than one of the 30 targeted procedures was performed at the same time were excluded in order to provide a more accurate assessment of revisit risk attributable to specific procedures. The single exception to this exclusion was an encounter when a gastrostomy was combined with a fundoplication. We included these cases as a unique cohort as we felt that the risk of revisits was particularly high owing to the procedures and unique cohort of children undergoing these procedures. 1.4. Defining relatedness to index surgical encounter For each condition, a team of four surgeons reviewed principal ICD-9 diagnosis codes that were associated with at least 1% of all revisit encounters for that condition, and then established consensus on the degree of relatedness to the index surgical encounter. The surgical team met on a weekly basis to collectively review revisit diagnoses and assign one of three possible categories of relatedness: very likely, possibly, or very unlikely related. A revisit diagnosis was considered very likely related if it could not have occurred without the index surgical procedure (e.g. wound infection following appendectomy) and possibly related if the diagnosis could be attributed to either a post-operative complication or unrelated condition (e.g. fever following
All-cause revisit rates ranged from 7.6% to 68.4% across procedures, with the highest rates associated with ulcerative colitis (68.4%), hepatoblastoma (65.3%), fundoplication combined with gastrostomy (63.6%), neuroblastoma (61.4%), and Wilms tumors (54.6%). Conditions with the lowest all-cause revisit rates included undescended testes (7.6%), umbilical hernia (7.9%), inguinal hernia (8.9%), pectus excavatum (10.3%), and ovarian torsion (10.6%). Of the 13,860 revisit encounters reviewed, 37.8% (5236) were categorized as very likely related, 43.0% (5967) possibly related, and 19.2% (2657) very unlikely related. The proportion of revisits likely related to the index hospitalization was significantly different across procedures (range: 0% to 77%, p b 0.0001), with the highest proportions associated with complicated appendicitis (77%), imperforate anus (66%), and uncomplicated appendicitis (61%) (Fig. 2). When using the more conservative definition of relatedness that included revisit diagnoses that were categorized as possibly related as well as very likely related, the proportion of related revisits remained significantly different across procedures (range: 28% to 100%, p b 0.0001). All revisits following management of complicated appendicitis, pectus excavatum, and pilonidal disease were considered related using this definition, and very high rates of related revisit rates were also found with Hirschsprung's disease (97.8%), ovarian cyst (97.4%), imperforate anus (96.1%), and uncomplicated appendicitis (93.1%). 2.2. Correlation between all-cause revisit rates and related revisits Revisit rates very likely related to the index surgical admission were on average 83% lower than all-cause revisit rates (range of decrease by procedure: 39–100%), and relatively poor correlation was found
1052
D.B. Cameron et al. / Journal of Pediatric Surgery 52 (2017) 1050–1055
Fig. 1. Assembly of study cohort. *Readmission encounters that were reviewed to determine relatedness included those where the revisit diagnosis accounted for at least 1% of all revisits for a specific procedure.
between ranking of procedures based on all-cause revisit rates and by revisit rates likely related to index admission (r = 0.33, p = 0.07). When using the more conservative definition of relatedness, the correlation was relatively stronger but remained modest across all procedures (r = 0.79, p b 0.001). The relationship between all-cause readmission rates and the proportion of revisits categorized as being related is presented in Fig. 3. 3. Discussion In this retrospective analysis of 144,535 encounters from 44 children's hospitals, the proportion of revisits related to the index surgical encounter varied significantly across procedures and relatively poor correlation was found between all-cause and related revisit rates. Furthermore, the correlation between rates of all cause and related revisits remained relatively weak even after redefining related
diagnoses to include those categorized as only possibly related. In aggregate, these results provide a compelling argument that all-cause revisit rates are a relatively poor surrogate for revisit rates plausibly related to the index surgical encounter. Previous studies have characterized variation in pediatric surgical readmission rates and factors associated with readmission; however, these studies have either been limited to single center analyses or relatively few pediatric surgical conditions [7–9]. Data surrounding the relationship of revisit encounters to care provided during index pediatric surgical care remain limited. Burjonrappa and colleagues conducted a review of 2217 pediatric surgical admissions at two tertiary-referral children's centers and characterized revisit rates, causes of readmission, and revisit length of stay. Of the 145 readmissions analyzed, less than half of revisits were found to be related to the procedure performed during the index admission. Appendectomy and procedures associated with the management of intestinal obstruction were among the
Fig. 2. Relative proportion of revisits categorized as either being very likely related, possibly related, or likely unrelated to the index surgical encounter for the thirty most common pediatric surgical conditions performed during the study period. Abbreviations: TEF/EA – tracheoesophageal fistula/esophageal atresia, CCAM – congenital cystic adenomatoid malformation, GT – gastrostomy tube, CDH – congenital diaphragmatic hernia.
D.B. Cameron et al. / Journal of Pediatric Surgery 52 (2017) 1050–1055
1053
Fig. 3. Association between all-cause revisit rates and revisit rates related to care provided during the index surgical encounter. For the analysis represented in the figure, related revisit diagnosis were defined as those categorized as being very likely related or possibly related. Abbreviations: TEF/EA – tracheoesophageal fistula/esophageal atresia, CCAM – congenital cystic adenomatoid malformation, GT – gastrostomy tube, CDH – congenital diaphragmatic hernia.
operations most likely to be associated with procedure-related revisits in their study [10]. Our current analysis suggests similar findings; however our primary aim was to characterize relatedness across a much broader spectrum of pediatric surgical procedures using a more granular assessment of relatedness for this purpose. Furthermore, our study cohort was derived from 44 children's hospitals, providing the most comprehensive and generalizable assessment of revisit relatedness associated with pediatric surgical practice to date. Previously published studies have also challenged the validity of readmission rates as a quality measure for care provided during the index hospitalization, suggesting that revisits may be more dependent on non-modifiable factors rather than on the care provided during the index hospitalization [5,11–13]. In the context of pediatric surgical conditions, these considerations may have particular relevance when assessing quality across different conditions where the rates of related revisits and all-cause revisits vary considerably. For example, all-cause readmission rates may more directly reflect the quality of care provided during the index encounter for conditions such as appendicitis and pilonidal disease where nearly all revisits were found to be related. This is in contrast to oncologic surgical conditions where previous studies have shown that all-cause revisit rates are often unrelated to care provided during the index hospitalization [14,15]. Our results were consistent with these earlier findings, with hepatoblastoma, Wilms tumor, and neuroblastoma all having among the highest all-cause revisit rates in our study but relatively modest related revisit rates. These findings may be somewhat intuitive when considering the relatively high proportion of patients in this cohort that are readmitted for scheduled adjuvant therapy (as well as the complications that arise from such therapy). Discordance between all-cause and related revisit rates were also found for procedures associated with underlying chronic medical conditions, many of which are associated with revisits and readmissions [16–18]. For example, a child with neurological impairment who undergoes gastrostomy tube placement might return to the ED with gastrostomy related complications (e.g. gastrostomy site infection or dislodgement). Alternatively, the same patient might be readmitted following a seizure that is likely unrelated to surgery but related to the underlying condition. The former exemplifies a scenario that might be
addressed through QI initiatives targeted at surgical care pathways and post-discharge protocols for gastrostomy care, while the latter relates to quality assurance and management considerations outside of the scope of pediatric surgical practice. The results of this study may have important implications for the prioritization of efforts to reduce revisits within the influence of pediatric surgical practice, and for meaningful comparative performance reporting among pediatric surgery departments on the basis of revisit rates. With respect to the former, identification of conditions where both the rate and burden of related revisits are relatively high (e.g. appendicitis) may guide prioritization of readmission reduction efforts within and across surgery departments. With respect to comparative performance reporting, data from this study may be used to identify a portfolio of procedures where allcause revisit rates may be a reasonable surrogate for related revisit rates, while at the same time also identifying procedures where caution should be exercised when extrapolating all-cause revisit rates as a potential reflection of surgical quality. The results of this study must be considered in the context of its limitations. These data were derived from a network of free-standing children's hospitals and may not be generalizable to other hospitals. Although the PHIS database is subject to continuous quality assurance audits, administrative coding errors and misclassification of revisit diagnoses and procedures are possible. With respect to our study cohort, we limited the analysis to the 30 most commonly performed general pediatric surgical procedures, and further limited our analysis to revisit diagnoses that contributed at least 1% of all revisit encounters for specific procedures. Although this approach did not capture all pediatric surgical cases, we do believe it is comprehensive in that it captured more than 80% of all pediatric surgical cases during the study period and strongly supports our hypothesis that all-cause revisit rates are not an accurate surrogate for those related to care during the index encounter. By limiting the preoperative length of stay to 2 days or less, we may be undercounting semi-urgent procedures (and therefore revisits attributable to these procedures) where preoperative treatment delay may occur because of prolonged diagnostic evaluation and scheduling logistics (e.g. pull-through procedure for newly diagnosed Hirschsprung's
1054
D.B. Cameron et al. / Journal of Pediatric Surgery 52 (2017) 1050–1055
disease). With respect to our cohort definitions, we found a relatively large proportion of revisits categorized as “possibly” related for many surgical procedures. This observation reflects the non-specific nature of many symptom-based readmission diagnoses (e.g. fever and nausea), and highlights the challenges with using administrative data even when the diagnostic coding is accurate. We attempted to adjust for this uncertainty by using a more “conservative” definition of relatedness as a sensitivity analysis. Even when using this more conservative definition, correlation between all-cause and related revisit rates remained only modest and significant variation persisted among procedures with respect to their related revisit rates. In conclusion, significant variation exists in the rates of related revisits for pediatric surgical procedures, and all-cause revisit rates appear to be poor surrogates for revisit rates that are plausibly related to care provided during the index hospitalization. These findings may have important implications for the prioritization of conditions for preventing readmission encounters sensitive to surgical care, and for fair comparative reporting of revisit rates among surgical departments. Finally, it is important to note that relatedness of revisits is not necessarily synonymous with preventability. Future efforts will need to characterize preventability within the cohort of procedures where the rates and burden of related revisits and readmissions are particularly high. Appendix A. Discussions Danielle B. Cameron, MD, MPH DANIELLE B. CAMERON, MD, MPH: Members and guests, thank you for the opportunity to present our work. I have no disclosures. Hospital revisit rates are considered important measures of quality. While all-cause revisit rates may reflect the quality of broader systems of care, they may not be an accurate reflection of the quality of care provided by the surgical team during and following the index admission. In this regard, little data exists surrounding the relationship between revisits and care provided during the index pediatric surgical encounter. With these considerations, the purpose of our study was two-fold. First, we sought to characterize the relatedness of revisit encounters to the index admission for pediatric surgical procedures. Second, we aimed to explore whether all cause revisit rates accurately reflect revisit rates plausibly related to the index surgical encounter. Using the Pediatric Health Information System (PHIS) database we retrospectively reviewed 90-day emergency department and inpatient revisits following index encounters for the 30 most commonly performed pediatric surgical procedures between January 1, 2012 and March 31, 2015. For each condition a team of four surgeons reviewed revisit diagnoses to determine relatedness to care provided during the index surgical encounter. A revisit diagnosis was considered very likely related if it could not have occurred without the index surgical procedure, (e.g. wound infection following appendectomy). Diagnoses considered possibly related were those that could be attributed to either a post-operative complication or an unrelated condition (e.g. fever following appendectomy). Diagnoses considered very unlikely to be related were those that were implausibly related to the surgical procedure (e.g. otitis media following appendectomy). We used Chi-squared tests to characterize variation in rates of all cause and related revisits among procedures. Spearman's correlation coefficient was used to explore whether all-cause revisit rates were an accurate surrogate for revisit rates related to care provided during the index surgical encounter. We analyzed over 140,000 index surgical encounters of which
15% had at least one revisit within 90 days of discharge. All cause revisit rates ranged from 7.6 to 68.4% across procedures, and this figure illustrates variation in the proportion of revisits across procedures likely related to care provided during or after the index surgical hospitalization. Conditions with the highest relative proportions of related revisits shown in pink and towards the left side of the figure were appendicitis and imperforate anus. Conditions with revisits very unlikely to be related located on the right side of the figure were oncologic conditions such as Wilm's tumor and neuroblastoma. This bubble plot illustrates the relationship between all-cause revisit rates on the Y axis and the proportion of revisits likely or possibly related to care provided during the index surgical encounter on the X axis. For this graph we combined very likely and possibly related diagnoses as a more conservative estimate of relatedness. Overall there was variable correlation between all cause and related revisits, with nearly perfect correlation for conditions such as appendicitis and pilonidal disease on the right side of the figure, while relatively poor correlation was seen for oncologic conditions such as hepatoblastoma and those associated with chronic underlying medical conditions (e.g. CDH and children requiring antireflux procedures) seen on the left side of the figure. The correlation between all-cause and related revisit rates across all conditions was 0.79, suggesting that all-cause revisits are a relatively week surrogate for related revisits in pediatric general surgery. The results of this study must be considered in the context of its limitations. The analysis was derived from a cohort of free-standing children's hospitals and thus the results may not be generalizable. In addition, administrative data is vulnerable to coding errors and misclassification of revisit diagnoses and procedures. Despite these limitations, we can conclude that significant variation exists in the relatedness of revisits for pediatric surgical procedures and all-cause revisit rates appear to be poor surrogates for revisits for related to care provided during the index hospitalization. Thank you for your attention. [applause] MODERATOR: This paper is open for discussion. I'll start. I wonder if you have any insight whether this could point towards potentially preventable causes of readmission? DR. CAMERON: Thank you for your question. The results of this study provide insight regarding which pediatric surgical procedures are more likely to be followed by revisits related to care provided during index surgical care and relatedness is not necessarily synonymous with preventability. However, for those conditions where there was a relatively high proportion of related revisits, these may serve as a good starting point for future efforts to characterize the nature of the revisit and whether it is potentially preventable. FEMALE VOICE: Thank you so much for this very elegant and very comprehensive view for our revisit. The question I have is why did you choose 90 days as opposed to the typical -30-day correlation, and if you looked at that, was there a difference in whether or not these are related or not related? The other thing is, are the revisits planned or unplanned because Wilm's tumors and – blastomas, they come back for their chemotherapy? Then the last thing is did you look to see whether
D.B. Cameron et al. / Journal of Pediatric Surgery 52 (2017) 1050–1055
or not there is clinical correlation? So for instance since this is an administrative database did you look at a biopsy of patients and look at certain institutions just to make sure that what was being coded in the administrative database is actually clinically relevant. Thank you again. DR. CAMERON: Thank you for your questions. The first question is why did we choose 90-day revisit intervals? We chose this because there are some conditions, particularly those related to chronic medical conditions like gastrostomy tubes, where the revisit period can extend far beyond the 30-day period, and may still have relevance to the pediatric surgical encounter. The second question was, are some of these planned revisits, and the answer is likely yes. We looked at all-cause revisits which contributes to why we see poor correlation for conditions such as oncologic conditions, where a large proportion of revisits are planned admissions for chemotherapy (and were therefore considered unrelated to index surgical care). Finally with regard to clinical correlation, we did not do any specific chart biopsies, but that would be a potential next step as part of a more focused analysis on preventability.
References [1] Berry JG, Toomey SL, Zaslavsky AM, et al. Pediatric readmission prevalence and variability across hospitals. JAMA 2013;309:372–80. http://dx.doi.org/10.1001/jama.2012.188351. [2] Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med 2009;360:1418–28. http://dx.doi. org/10.1056/NEJMsa0803563. [3] Nakamura MM, Toomey SL, Zaslavsky AM, et al. Measuring pediatric hospital readmission rates to drive quality improvement. Acad Pediatr 2014;14:S39–46. http://dx.doi.org/10.1016/j.acap.2014.06.012.
1055
[4] Bardach NS, Vittinghoff E, Asteria-Peñaloza R, et al. Measuring hospital quality using pediatric readmission and revisit rates. Pediatrics 2013;132:429–36. http://dx.doi. org/10.1542/peds.2012-3527. [5] Dimick JB, Ghaferi AA. Hospital readmission as a quality measure in surgery. JAMA 2015;313:512–3. http://dx.doi.org/10.1001/jama.2014.14179. [6] Feudtner C, Pati S, Goodman DM, et al. State-level child health system performance and the likelihood of readmission to children's hospitals. J Pediatr 2010;157:98–102. [7] Thibaudeau S, Anari JB, Carducci N, et al. 30-day readmission after pediatric upper extremity surgery: analysis of the NSQIP database. J Pediatr Surg 2016 Aug;51(8): 1370–4. http://dx.doi.org/10.1016/j.jpedsurg.2016.04.012. [8] Sinha CK, Decker E, Rex D, et al. Thirty-days readmissions in pediatric surgery: the first U.K. experience. J Pediatr Surg November 2016;51(11):1877–80. http://dx.doi. org/10.1016/j.jpedsurg.2016.06.015. [9] Rice-Townsend S, Hall M, Barnes JN, et al. Variation in risk-adjusted hospital readmission after treatment of appendicitis at 38 children's hospitals: an opportunity for collaborative quality improvement. Ann Surg 2013;257:758–65. http://dx.doi. org/10.1097/SLA.0b013e318268a663. [10] Burjonrappa S, Theodorus A, Shah A, et al. Pediatric surgery readmissions: a root cause analysis. Pediatr Surg Int 2015;31:551–5. http://dx.doi.org/10.1007/s00383015-3701-4. [11] Ludke RL, Booth BM, Lewis-Beck JA. Relationship between early readmission and hospital quality of care indicators. Inquiry 1993;30:95–103. [12] Benbassat J, Taragin M. Hospital readmissions as a measure of quality of health care: advantages and limitations. Arch Intern Med 2000;160:1074–81. [13] Berry JG, Ziniel SI, Freeman L, et al. Hospital readmission and parent perceptions of their child's hospital discharge. Int J Qual Health Care 2013;25: 573–81. http://dx.doi.org/10.1093/intqhc/mzt051. [14] Shulman DS, London WB, Guo D, et al. Incidence and causes of hospital readmission in pediatric patients after hematopoietic cell transplantation. - PubMed - NCBI. Biol Blood Marrow Transplant 2015;21:913–9. [15] Akenroye AT, Thurm CW, Neuman MI, et al. Prevalence and predictors of return visits to pediatric emergency departments. J Hosp Med 2014;9:779–87. http://dx.doi.org/10.1002/jhm.2273. [16] Berry JG, Hall DE, Kuo DZ, et al. Hospital utilization and characteristics of patients experiencing recurrent readmissions within children's hospitals. JAMA 2011;305: 682–90. http://dx.doi.org/10.1001/jama.2011.122. [17] Jurgens V, Spaeder MC, Pavuluri P, et al. Hospital readmission in children with complex chronic conditions discharged from subacute care. Hosp Pediatr 2014;4:153–8. http://dx.doi.org/10.1542/hpeds.2013-0094. [18] Cohen E, Berry JG, Camacho X, et al. Patterns and costs of health care use of children with medical complexity. Pediatrics 2012;130:e1463–70. http://dx.doi.org/10.1542/ peds.2012-0175.