- Email: [email protected]
Nationwide readmissions after tonsillectomy among pediatric patients - United States
International Journal of Pediatric Otorhinolaryngology 107 (2018) 10–13
Contents lists available at ScienceDirect
International Journal of Pediatric Otorhinolaryngology journal homepage: www.elsevier.com/locate/ijporl
Nationwide readmissions after tonsillectomy among pediatric patients United States
T
Romaine F. Johnsona,b,∗, Andrew Changa, Ron B. Mitchella,b a b
UT Southwestern Medical Center Dallas, Dallas, TX, USA Children's Medical Center Dallas, Children's Health, Dallas, TX, USA
A B S T R A C T Objectives: 1) Investigate incidence and predictors of readmissions after tonsillectomy with or without adenoidectomy (T&A) in children. 2) Identify factors that may predict readmission. Settings: Nationwide cross-sectional survey of US hospital admissions. Subjects: and Methods: The 2013 Nationwide Readmission Database (NRD) was used to examine all-cause readmissions within 30 days of T&A in children (age < 18 years). Logistic regression was used to analyze the associations of demographics, diagnosis, insurance status, length of index stay, and median household income with readmission. Results: 9079 children undergoing T&A resulted in 327 (3.6%) patients requiring readmission. The average age of children readmitted were 5.0 years and they were 51% female. The most common readmission diagnoses were dehydration (47%), hemorrhage (26%), and pain (16%). The average time to readmission was 7.3 days. The average times to readmission for hemorrhage, pain and dehydration were 6.3, 4.5 and 4.1 days, respectively. Children who needed respiratory intubation (OR = 4.0), had a medical or surgical complication (OR = 3.3), or prolonged hospital stay (OR = 1.03) during the index admission were more likely to be readmitted. Age, gender, payer and socioeconomic status and diagnosis of obstructive sleep apnea (OSA) did not increase the odds of readmission. Conclusions: Readmissions in children after T&A were primarily due to dehydration, hemorrhage, and pain. Adequate symptom control in children has the greatest potential to reduce readmission rates following T&A.
1. Introduction There are few cross-sectional studies examining readmission rates following tonsillectomy with or without adenoidectomy (T&A) in children. Mahant and colleagues [1] reported a readmission rate of 7.8% following T&A in a study confined to children's hospitals. Edmonson and colleagues [2] used a state discharge database to report a 2.3% readmission rate in the state of California. Shay and colleagues [3] and Bhattacharyya and colleagues [4] used the Healthcare Cost and Utilization Project (HCUP) to study ambulatory T&A from State Ambulatory Surgery databases in New York, Florida, Iowa, and California for 2010 and linked these cases to the corresponding State Emergency Department and State Inpatient databases within a 14-day postoperative window. They reported readmission and revisit rates of approximately 16% after T&A [3], mostly for fever, nausea, vomiting, or dehydration. Other studies [5–15], at single institutions, have reported readmission rates ranging from 0 to 11% for primarily postoperative hemorrhage.
∗
Readmissions have taken on added significance as a measure of health outcomes and quality. In June 2009, the Centers for Medicare & Medicaid Services (CMS) began publicly reporting hospital readmission rates for certain diagnoses, resulting in readmission rates becoming an important metric to assess quality and efficiency of care [16]. Readmissions impart medicals risks to the patient, create inconvenience for families, and may reflect suboptimal medical care [17]. Using 2005 Medicare data, the Medicare Payment Advisory Commission estimated that 13.3% of readmissions were potentially preventable and cost an additional $12 billion [18]. Another study [19] reported that readmissions at children's hospitals cost $1.7 billion, 27% of which were categorized as “potentially preventable readmissions.” With passage of the Patient Protection and Affordable Care Act, CMS began holding hospitals accountable for their readmission rates and adjusting payments to hospitals accordingly [17]. In 2015, the Nationwide Readmissions Database (NRD) was added to the family of databases for the HCUP to address a deficiency in health
Corresponding author. 2350 N. Stemmons Freeway, F6.207 Dallas, TX, USA. E-mail address: [email protected] (R.F. Johnson).
https://doi.org/10.1016/j.ijporl.2018.01.026 Received 24 October 2017; Received in revised form 17 January 2018; Accepted 18 January 2018 Available online 20 February 2018 0165-5876/ © 2018 Elsevier B.V. All rights reserved.
International Journal of Pediatric Otorhinolaryngology 107 (2018) 10–13
R.F. Johnson et al.
Table 1 Demographics tonsillectomy with or without adenoidectomy in the 2013 Nationwide Readmission Database (NRD). Variable
Index Admissiona
95% CI
Readmission
95% CI
P valueb
Overall, N (%)
9079 (100)
8958 to 9199
327 (3.6)
–
Age ≤ 12 years, N (%)
7988 (88)
267 (3.3)
Age > 12 years, N (%)
914 (10)
Female, N (%)
3788 (42)
Male, N (%)
5291 (59)
LOS, days, mean Total charges/case, $, mean Medicaid, N (%)
2.1 22989 5508 (61)
Private Ins., N (%)
3241 (36)
Lowest Inc Qrtl, N (%)
2978 (33)
Highest Inc Qrtl, N (%)
1745 (19)
7828 to 8148 (88–91) 808 to 1021 (9.1–10.5) 3602 to 3974 (39–44) 5085 to 5495 (56–61) 1.9 to 2.2 21405 to 24574 5317 to 5699 (59–63) 2947 to 3313 (34–38) 2806 to 3151 (31–35) 1609 to 1881 (18–21)
258 to 396 (2.9–4.4) 203 to 330 (2.4–3.8) 26 to 76 (3.5–8.9) 114 to 217 (3.2–5.9) 115 to 208 (2.3–4.1) 2.1 to 3.4 23636 to 43997 153 to 261 (2.9–4.9) 68 to 154 (2.3–5.0) 58 to 132 (2.9–4.4) 34 to 99 (2.3–6.1)
51 (5.6) 166 (4.4) 162 (3.1) 2.7 33817 207 (3.8) 111 (3.4) 95 (3.2) 66 (3.8)
.06 .06 .09 .09 .04 .03 .64 .75 .47 .80
Abbreviations: LOS = length of hospital stay in days; SE = standard errors; N = number; Inc Qrtl = income quartile; Ins = insurance. a Index stay defined by all listed tonsillectomy with or without adenoidectomy – Readmission defined as readmission for any cause within 30 days of index discharge. Index percentages are based on the total admissions column; for example, 87% of index admission were ≤12 years old. Readmission percentage are based on the variable rows so 3.3% of patients ≤12 years old are readmitted and so forth. b P value based on linear regression for continuous variables and Pearson X2 or categorical variables.
care analysis – namely the magnitude and burden of hospital readmissions in the United States as a whole. In its initial release, which utilizes 2013 data, the NRD contains information from approximately 14 million discharges from 21 states, with the ability to estimate about 36 million hospital discharges when weighted, accounting for almost 50% of the U.S. population and hospitalizations. These hospitalizations include inpatient stays and exclude patients admitted under observation status although patients changed from observation to inpatient status are included (according to an email from HCUP support, April 2017). Outpatient surgery centers and same day discharges are not included in the dataset. The primary aim of this study is to examine all cause 30-day readmissions after T&A among children in the United States among hospitalized patients captured by the NRD. We hypothesize that readmissions are primarily due to postoperative hemorrhage, pain, and dehydration and are similar among children regardless of age, gender, or socioeconomic status.
calculate readmission if this value is missing), and if the index admission occurred in December since the 30-day window would include January 2014, which falls outside the time period of data collection by the 2013 NRD. In instances of multiple admissions, analysis was limited to the first readmission. It is worth noting that race/ethnicity is not measured in the NRD so no analysis of this could be undertaken. 2.2. Statistical analysis The NRD uses a complex survey design. Observations are collected at the hospital level and then via statistical weighting estimated counts of nationwide observations are calculated using Taylor series linearization. Standard statistics are then performed on those estimated inferences to produce national estimations. Since these are estimated counts, continuous data is presented as means and 95% confidence intervals as well as medians with interquartile ranges (25th percentile to 75th percentile) to help account for outliers. Categorical data is presented as percentages with 95% confidence intervals. Simple logistical regression analysis was performed on the odds of readmission for the following variables: age, gender, primary payer, median income by zip code, indications for surgery (obstructive sleep apnea [OSA], tonsillitis), hemorrhage complicating the procedure, respiratory complications, any complication of medical or surgical care, respiratory intubation and mechanical ventilation. Statistical significance was set as a two-tailed p-value less than or equal to 0.05. All statistics were performed with Stata Statistical Software (Version 14, StataCorp, College Station, Texas).
2. Materials and methods The Institutional Review Board of UT Southwestern approved this study as exempt status. 2.1. Data source We included all patients from the 2013 NRD who underwent T&A (International Classification of Diseases, Ninth Revision, Clinical Modifications [ICD-9 CM] procedural codes 28.2 or 28.3), as the index admission among patients less or equal to 18 years of age. Standard demographic and baseline information was extracted for readmissions including: age, gender, diagnosis and procedures utilizing ICD-9 codes and HCUP's Clinical Classification Software (CCS) [20] (Appendix 1), length of stay, time to readmission, total charges in dollars, primary payer, and estimated median household income quartile of residents in the patient's zip-code. Cases were stratified by age (0–12 years of age; and ≥12 years of age). We excluded cases where patients died during the index admission, the length of stay was missing (the NRD cannot
3. Results An estimated 9079 (95% CI, 8958 to 9199) index admissions for T& A were included in the study. The mean age was 5.0 years (95% confidence interval [CI], 4.8 to 5.2); median age was 3 with interquartile range (IQR) of 2–7. Approximately 90% (95% CI, 88%–91%) were ≤12 years old. The majority were male (58%; 95% CI, 56%–60%). Table 1 presents the demographics of the population. A total of 327, or 3.6% (95% CI, 2.9%–4.4%), required readmission. 11
International Journal of Pediatric Otorhinolaryngology 107 (2018) 10–13
R.F. Johnson et al.
following T&A. These findings emphasize that patients who have complications after T&A are at highest risk for readmissions and that early interventions in these patients could reduce readmission rates in children overall. Early interventions should be tailored toward the provider's population but could include some approaches aimed specifically at higher risk patients. These interventions could include recommending scheduled pain medication dosing instead of the more common as needed ("prn") dosing. Another intervention could be nursing phone call follow-up 24–48 h after discharge to ensure that pain control and hydration are adequate. A final intervention could be making discharge criteria more stringent among high-risk patients. Interestingly many variables, like OSA as an indication for T&A, showed no increased association with readmission. Other studies have reported an association between OSA severity and respiratory complications [20] in the perioperative period [21]. Perhaps in this series, patients may have only had mild OSA, that is not associated with perioperative respiratory complications, or had their care optimized during the index admission resulting in no increased in risk of readmission. Equally, even severe OSA may only by associated with complications in the perioperative period when the patient is already hospitalized and not reflected in a risk of readmission. There are some limitations to this study that should be considered. First, this database excludes surgeries that were done as outpatients. Since, the majorities of T&As are done as outpatients this data is likely to be biased towards sicker children. This data also uses discharge abstracts, which are billing documents, produced by hospitals for reimbursement. These documents can contain ICD-9 code errors that could under or over-estimate conditions and procedures. Therefore, the reader should use the 95% confidence intervals in interpreting the results whereby small confidence intervals suggest more precise results. Another limitation of this study is that the readmission benchmark presented is only for inpatient T&A and is not applicable to outpatient or observation cases. Lastly, as this is the first attempt at a nationwide readmission database, there are likely to be sampling issues that will be recognized and improved with future releases. This data should therefore be viewed as just one source of readmission data and not the only source. Nonetheless, this represents the largest cross-sectional study to date of readmissions after T&A that includes an entire representative population of patients. This study also adds to the growing literature concerning the quality and efficiency of healthcare, especially with the increasing spotlight on readmissions as a measure of quality of care. A high readmission rate can adversely affect hospital revenue and prestige. Though post-T&A hemorrhage is most widely reported, the current study reports dehydration and pain as equally common admitting diagnosis and the areas where interventions could potentially reduce readmissions.
Table 2 Simple logistic regression for readmission after tonsillectomy with or without adenoidectomy–2013 Nationwide Readmission Database (NRD). Variable
Odds Ratioa
95% CI
P value
Age, years Age ≤12 years Age > 12 years Male Female Medicaid Private Ins. Lowest Inc. Qrtl. Highest Inc. Qrtl. Respiratory Complicationb Complication of careb Intubation & mechanical ventilationb OSAb Tonsillitisb Index LOS
1.00 0.58 1.7 0.68 1.5 1.1 0.92 0.84 1.1 1.5 3.3 4.0 1.1 0.71 1.03
0.98 to 1.1 0.33 to 1.0 0.98 to 3.0 0.44 to 1.1 0.93 to 2.3 0.70 to 1.8 0.58 to 1.5 0.52 to 1.4 0.61 to 1.9 0.45 to 5.6 1.5 to 7.3 1.3 to 12.4 0.70 to 1.8 0.32 to 1.6 1.00 to 1.06
.25 .06 .06 .10 .10 .64 .75 .47 .80 .48 .002 .02 .60 .41 .02
Abbreviations: LOS = length of hospital stay in days; Inc Qrtl = income quartile; Ins = insurance; OSA = obstructive sleep apnea. a Odds ratio based on simple logistic regression. b See ICD9 or CCS codes in Appendix for definitions.
The mean age among readmissions was 5.6 years (95% CI, 4.4 to 6.7); median age 4 (IQR, 2 to 8). Of those readmitted, 84% (95% CI, 75%–90%) were younger than 12 years old. The majority of readmissions were female (51%; 95% CI 40%–61%). Among readmitted patients, 26% (95% CI, 18%–36%) had a postoperative hemorrhage. Other diagnoses included dehydration (47%; 95% CI, 36%–57%) and pain (16%; 95% CI, 9.8%–26%). Less frequent diagnoses include postoperative infection (1.6%; 95% CI, 0.4%–5.8%) and pneumonia (1.3%; 95% CI 0.3%–5.2%). The average time to readmission was 7.3 days (95% CI, 5.6 to 8.9). The median was 5 days with an interquartile range of 3–8 days. The average time to readmission for children with postoperative pain was 4.5 days (95% CI, 3 to 6), median 4 days (IQR 2 to 5); postoperative dehydration was 4 days (95% CI, 3 to 5), median 3 days (IQR 2 to 6); postoperative hemorrhage was 6 days (95% CI, 5 to 8), median 6 days (IQR, 5 to 7); and postoperative infection was 8 days (95% CI, 6 to 10), median 9 days (IQR, 6 to 9). Simple logistic regression analysis for readmission (Table 2) found length of stay (OR, 1.03; 95% CI, 1.0 to 1.06), need for respiratory ventilation (e.g., mechanical ventilation) during the index admission (OR, 4.0; 95% CI [1.3 to 12.4]), medical or surgical complication (OR, 3.3; 95% CI, 1.5 to 7.3) increased the likelihood of readmission (see Appendix 1 for associated ICD-9 codes). The following variables were not associated with increased risk of readmission: age, gender, payer status, median income of zip code and indication for T&A (OSA versus tonsillitis).
5. Conclusion The current study uses a newly available database to quantify, at a national level, readmissions following T&A in children. The most common admitting diagnoses were pain, dehydration, and hemorrhage. Further studies are needed to look at ways of reducing pain, dehydration, and hemorrhage after T&A in children.
4. Discussion Our analysis of the Nationwide Readmission Database for readmissions after T&A demonstrated a readmission rate of 3.6%, consistent with the 2.3–11% reported in other studies [1–3]. Our study also found that readmissions were most commonly due to postoperative pain, dehydration, and bleeding. Children with post-operative pain and dehydration presented several days earlier than patients with postoperative bleeding. The current study presents an updated and contemporary analysis on readmissions following T&A that can be used to benchmark national readmission rates since previous studies have been limited to children's hospitals [1] or to a single [2] or a few [3] states. Besides affirming the causes of readmission, our analysis showed that children with longer lengths of hospital stays, patients requiring airway support, and those with a medical or surgical complication during the index admissions were at increased risk of readmission
Disclosures None to report. Conflicts of interest None to report. External funding None. 12
International Journal of Pediatric Otorhinolaryngology 107 (2018) 10–13
R.F. Johnson et al.
Appendix 1. ICD-9 conditions and procedural codes
Clinical Classification Software Codea
Variable
ICD-9 Code
Tonsillectomy with or without adenoidectomy Hemorrhage complicating procedure Respiratory intubation and mechanical ventilation Complications of surgical procedures or medical care Postoperative Infection Respiratory Complication Dehydration Acute Postoperative Pain Obstructive Sleep Apnea Tonsillitis
282, 283 998.11 9390, 9604, 9605, 9670, 9671, 9672
a
238 9985, 99851, 99859 9971, 99731, 99732, 99739 27651 33818 32723 47400, 47402, 463, 0340
Clinical classification software: https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccsfactsheet.jsp.
Trends 27 (3) (1995) 98–100. [12] W.C. Lee, J.F. Sharp, Complications of paediatric tonsillectomy post-discharge, J. Laryngol. Otol. 110 (2) (1996) 136–140. [13] A.C. Norrington, L.M. Flood, T. Meek, M.R. Tremlett, Does day case pediatric tonsillectomy increase postoperative pain compared to overnight stay pediatric tonsillectomy? a prospective comparative audit, Paediatr. Anaesth. 23 (8) (2013) 697–701. [14] V. Raut, N. Bhat, J. Kinsella, J.G. Toner, A.R. Sinnathuray, M. Stevenson, Bipolar scissors versus cold dissection tonsillectomy: a prospective, randomized, multi-unit study, Laryngoscope 111 (12) (2001) 2178–2182. [15] Y.M. Takwoingi, M. Shykhon, M. Wake, Effect of post-operative analgesia duration on post-tonsillectomy readmission rate: comparison of five-day and 14-day regime, J. Laryngol. Otol. 121 (10) (2007) 968–972. [16] R.N. Axon, M.V. Williams, Hospital readmission as an accountability measure, J. Am. Med. Assoc. 305 (5) (2011) 504–505. [17] N.R. Payne, A. Flood, Preventing pediatric readmissions: which ones and how? J. Pediatr. 166 (3) (2015) 519–520. [18] J.C. Gay, R. Agrawal, K.A. Auger, et al., Rates and Impact of potentially preventable readmissions at children's hospitals, J. Pediatr. 166 (3) (2015) 613–619 e5. [19] HCUP, Clinical Classifications Software (CCS) for ICD-9-CM. Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality, Rockville, MD, 2006-2009www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp Accessed March, 2016. [20] M.M. Statham, R.G. Elluru, R. Buncher, M. Kalra, Adentonsillectomy for obstructive sleep apnea syndrome in young children: prevalence of pulmonary complications, Arch. Otolaryngol. Head Neck Surg. 132 (5) (2006) 476–480. [21] A. Belloso, A. Chidambaram, P. Morar, M.S. Timms, Coblation tonsillectomy versus dissection tonsillectomy: postoperative hemorrhage, Laryngoscope 113 (11) (2010) 2010–2013.
References [1] S. Mahant, R. Keren, R. Localio, et al., Variation in quality of tonsillectomy perioperative care and revisit rates in Children's hospitals, Pediatrics 133 (2) (2014) 280–288. [2] M.B. Edmonson, J.C. Eickhoff, C. Zhang, A population-based study of acute care revisits following tonsillectomy, J. Pediatr. 166 (3) (2015) 607–612 e5. [3] S. Shay, N.L. Shapiro, N. Bhattacharyya, Revisit rates and diagnoses following pediatric tonsillectomy in a large multistate population, Laryngoscope 125 (2) (2015) 457–461. [4] N. Bhattacharyya, L.J. Kepnes, Revisits and postoperative hemorrhage after adult tonsillectomy, Laryngoscope 124 (7) (2014) 1554–1556. [5] R. Benson-Mitchell, A.R. Maw, Assessment of sequelae at home following adenotonsillectomy. a basis for day-case management? Clin. Otolaryngol. Allied Sci. 18 (4) (1993) 282–284. [6] N. Bhattacharyya, Evaluation of post-tonsillectomy bleeding in the adult population, Ear Nose Throat J. 80 (8) (2001) 544–549. [7] A.S. Evans, A.M. Khan, D. Young, R. Adamson, Assessment of secondary haemorrhage rates following adult tonsillectomy – a telephone survey and literature review, Clin. Otolaryngol. Allied Sci. 28 (6) (2003) 489–491. [8] K. Ghufoor, A. Frosh, G. Sandhu, J. Hanif, Post-tonsillectomy patient care in the community, Int. J. Clin. Pract. 54 (7) (2000) 420–423. [9] R.L. Harris, J.E. Mitchell, D.A. Jonathan, A telephone audit in parallel with the UK national tonsillectomy audit to investigate re-admission as a measure of secondary haemorrhage rate, Auris Nasus Larynx 35 (2) (2008) 220–223. [10] E.K. Hoddeson, C.G. Gourin, Adult tonsillectomy: current indications and outcomes, Otolaryngol. Head Neck Surg. 140 (1) (2009) 19–22. [11] M. Kuo, D. Hegarty, A. Johnson, S. Stevenson, Early post-tonsillectomy morbidity following hospital discharge: do patients and GPs know what to expect? Health
13
Contents lists available at ScienceDirect
International Journal of Pediatric Otorhinolaryngology journal homepage: www.elsevier.com/locate/ijporl
Nationwide readmissions after tonsillectomy among pediatric patients United States
T
Romaine F. Johnsona,b,∗, Andrew Changa, Ron B. Mitchella,b a b
UT Southwestern Medical Center Dallas, Dallas, TX, USA Children's Medical Center Dallas, Children's Health, Dallas, TX, USA
A B S T R A C T Objectives: 1) Investigate incidence and predictors of readmissions after tonsillectomy with or without adenoidectomy (T&A) in children. 2) Identify factors that may predict readmission. Settings: Nationwide cross-sectional survey of US hospital admissions. Subjects: and Methods: The 2013 Nationwide Readmission Database (NRD) was used to examine all-cause readmissions within 30 days of T&A in children (age < 18 years). Logistic regression was used to analyze the associations of demographics, diagnosis, insurance status, length of index stay, and median household income with readmission. Results: 9079 children undergoing T&A resulted in 327 (3.6%) patients requiring readmission. The average age of children readmitted were 5.0 years and they were 51% female. The most common readmission diagnoses were dehydration (47%), hemorrhage (26%), and pain (16%). The average time to readmission was 7.3 days. The average times to readmission for hemorrhage, pain and dehydration were 6.3, 4.5 and 4.1 days, respectively. Children who needed respiratory intubation (OR = 4.0), had a medical or surgical complication (OR = 3.3), or prolonged hospital stay (OR = 1.03) during the index admission were more likely to be readmitted. Age, gender, payer and socioeconomic status and diagnosis of obstructive sleep apnea (OSA) did not increase the odds of readmission. Conclusions: Readmissions in children after T&A were primarily due to dehydration, hemorrhage, and pain. Adequate symptom control in children has the greatest potential to reduce readmission rates following T&A.
1. Introduction There are few cross-sectional studies examining readmission rates following tonsillectomy with or without adenoidectomy (T&A) in children. Mahant and colleagues [1] reported a readmission rate of 7.8% following T&A in a study confined to children's hospitals. Edmonson and colleagues [2] used a state discharge database to report a 2.3% readmission rate in the state of California. Shay and colleagues [3] and Bhattacharyya and colleagues [4] used the Healthcare Cost and Utilization Project (HCUP) to study ambulatory T&A from State Ambulatory Surgery databases in New York, Florida, Iowa, and California for 2010 and linked these cases to the corresponding State Emergency Department and State Inpatient databases within a 14-day postoperative window. They reported readmission and revisit rates of approximately 16% after T&A [3], mostly for fever, nausea, vomiting, or dehydration. Other studies [5–15], at single institutions, have reported readmission rates ranging from 0 to 11% for primarily postoperative hemorrhage.
∗
Readmissions have taken on added significance as a measure of health outcomes and quality. In June 2009, the Centers for Medicare & Medicaid Services (CMS) began publicly reporting hospital readmission rates for certain diagnoses, resulting in readmission rates becoming an important metric to assess quality and efficiency of care [16]. Readmissions impart medicals risks to the patient, create inconvenience for families, and may reflect suboptimal medical care [17]. Using 2005 Medicare data, the Medicare Payment Advisory Commission estimated that 13.3% of readmissions were potentially preventable and cost an additional $12 billion [18]. Another study [19] reported that readmissions at children's hospitals cost $1.7 billion, 27% of which were categorized as “potentially preventable readmissions.” With passage of the Patient Protection and Affordable Care Act, CMS began holding hospitals accountable for their readmission rates and adjusting payments to hospitals accordingly [17]. In 2015, the Nationwide Readmissions Database (NRD) was added to the family of databases for the HCUP to address a deficiency in health
Corresponding author. 2350 N. Stemmons Freeway, F6.207 Dallas, TX, USA. E-mail address: [email protected] (R.F. Johnson).
https://doi.org/10.1016/j.ijporl.2018.01.026 Received 24 October 2017; Received in revised form 17 January 2018; Accepted 18 January 2018 Available online 20 February 2018 0165-5876/ © 2018 Elsevier B.V. All rights reserved.
International Journal of Pediatric Otorhinolaryngology 107 (2018) 10–13
R.F. Johnson et al.
Table 1 Demographics tonsillectomy with or without adenoidectomy in the 2013 Nationwide Readmission Database (NRD). Variable
Index Admissiona
95% CI
Readmission
95% CI
P valueb
Overall, N (%)
9079 (100)
8958 to 9199
327 (3.6)
–
Age ≤ 12 years, N (%)
7988 (88)
267 (3.3)
Age > 12 years, N (%)
914 (10)
Female, N (%)
3788 (42)
Male, N (%)
5291 (59)
LOS, days, mean Total charges/case, $, mean Medicaid, N (%)
2.1 22989 5508 (61)
Private Ins., N (%)
3241 (36)
Lowest Inc Qrtl, N (%)
2978 (33)
Highest Inc Qrtl, N (%)
1745 (19)
7828 to 8148 (88–91) 808 to 1021 (9.1–10.5) 3602 to 3974 (39–44) 5085 to 5495 (56–61) 1.9 to 2.2 21405 to 24574 5317 to 5699 (59–63) 2947 to 3313 (34–38) 2806 to 3151 (31–35) 1609 to 1881 (18–21)
258 to 396 (2.9–4.4) 203 to 330 (2.4–3.8) 26 to 76 (3.5–8.9) 114 to 217 (3.2–5.9) 115 to 208 (2.3–4.1) 2.1 to 3.4 23636 to 43997 153 to 261 (2.9–4.9) 68 to 154 (2.3–5.0) 58 to 132 (2.9–4.4) 34 to 99 (2.3–6.1)
51 (5.6) 166 (4.4) 162 (3.1) 2.7 33817 207 (3.8) 111 (3.4) 95 (3.2) 66 (3.8)
.06 .06 .09 .09 .04 .03 .64 .75 .47 .80
Abbreviations: LOS = length of hospital stay in days; SE = standard errors; N = number; Inc Qrtl = income quartile; Ins = insurance. a Index stay defined by all listed tonsillectomy with or without adenoidectomy – Readmission defined as readmission for any cause within 30 days of index discharge. Index percentages are based on the total admissions column; for example, 87% of index admission were ≤12 years old. Readmission percentage are based on the variable rows so 3.3% of patients ≤12 years old are readmitted and so forth. b P value based on linear regression for continuous variables and Pearson X2 or categorical variables.
care analysis – namely the magnitude and burden of hospital readmissions in the United States as a whole. In its initial release, which utilizes 2013 data, the NRD contains information from approximately 14 million discharges from 21 states, with the ability to estimate about 36 million hospital discharges when weighted, accounting for almost 50% of the U.S. population and hospitalizations. These hospitalizations include inpatient stays and exclude patients admitted under observation status although patients changed from observation to inpatient status are included (according to an email from HCUP support, April 2017). Outpatient surgery centers and same day discharges are not included in the dataset. The primary aim of this study is to examine all cause 30-day readmissions after T&A among children in the United States among hospitalized patients captured by the NRD. We hypothesize that readmissions are primarily due to postoperative hemorrhage, pain, and dehydration and are similar among children regardless of age, gender, or socioeconomic status.
calculate readmission if this value is missing), and if the index admission occurred in December since the 30-day window would include January 2014, which falls outside the time period of data collection by the 2013 NRD. In instances of multiple admissions, analysis was limited to the first readmission. It is worth noting that race/ethnicity is not measured in the NRD so no analysis of this could be undertaken. 2.2. Statistical analysis The NRD uses a complex survey design. Observations are collected at the hospital level and then via statistical weighting estimated counts of nationwide observations are calculated using Taylor series linearization. Standard statistics are then performed on those estimated inferences to produce national estimations. Since these are estimated counts, continuous data is presented as means and 95% confidence intervals as well as medians with interquartile ranges (25th percentile to 75th percentile) to help account for outliers. Categorical data is presented as percentages with 95% confidence intervals. Simple logistical regression analysis was performed on the odds of readmission for the following variables: age, gender, primary payer, median income by zip code, indications for surgery (obstructive sleep apnea [OSA], tonsillitis), hemorrhage complicating the procedure, respiratory complications, any complication of medical or surgical care, respiratory intubation and mechanical ventilation. Statistical significance was set as a two-tailed p-value less than or equal to 0.05. All statistics were performed with Stata Statistical Software (Version 14, StataCorp, College Station, Texas).
2. Materials and methods The Institutional Review Board of UT Southwestern approved this study as exempt status. 2.1. Data source We included all patients from the 2013 NRD who underwent T&A (International Classification of Diseases, Ninth Revision, Clinical Modifications [ICD-9 CM] procedural codes 28.2 or 28.3), as the index admission among patients less or equal to 18 years of age. Standard demographic and baseline information was extracted for readmissions including: age, gender, diagnosis and procedures utilizing ICD-9 codes and HCUP's Clinical Classification Software (CCS) [20] (Appendix 1), length of stay, time to readmission, total charges in dollars, primary payer, and estimated median household income quartile of residents in the patient's zip-code. Cases were stratified by age (0–12 years of age; and ≥12 years of age). We excluded cases where patients died during the index admission, the length of stay was missing (the NRD cannot
3. Results An estimated 9079 (95% CI, 8958 to 9199) index admissions for T& A were included in the study. The mean age was 5.0 years (95% confidence interval [CI], 4.8 to 5.2); median age was 3 with interquartile range (IQR) of 2–7. Approximately 90% (95% CI, 88%–91%) were ≤12 years old. The majority were male (58%; 95% CI, 56%–60%). Table 1 presents the demographics of the population. A total of 327, or 3.6% (95% CI, 2.9%–4.4%), required readmission. 11
International Journal of Pediatric Otorhinolaryngology 107 (2018) 10–13
R.F. Johnson et al.
following T&A. These findings emphasize that patients who have complications after T&A are at highest risk for readmissions and that early interventions in these patients could reduce readmission rates in children overall. Early interventions should be tailored toward the provider's population but could include some approaches aimed specifically at higher risk patients. These interventions could include recommending scheduled pain medication dosing instead of the more common as needed ("prn") dosing. Another intervention could be nursing phone call follow-up 24–48 h after discharge to ensure that pain control and hydration are adequate. A final intervention could be making discharge criteria more stringent among high-risk patients. Interestingly many variables, like OSA as an indication for T&A, showed no increased association with readmission. Other studies have reported an association between OSA severity and respiratory complications [20] in the perioperative period [21]. Perhaps in this series, patients may have only had mild OSA, that is not associated with perioperative respiratory complications, or had their care optimized during the index admission resulting in no increased in risk of readmission. Equally, even severe OSA may only by associated with complications in the perioperative period when the patient is already hospitalized and not reflected in a risk of readmission. There are some limitations to this study that should be considered. First, this database excludes surgeries that were done as outpatients. Since, the majorities of T&As are done as outpatients this data is likely to be biased towards sicker children. This data also uses discharge abstracts, which are billing documents, produced by hospitals for reimbursement. These documents can contain ICD-9 code errors that could under or over-estimate conditions and procedures. Therefore, the reader should use the 95% confidence intervals in interpreting the results whereby small confidence intervals suggest more precise results. Another limitation of this study is that the readmission benchmark presented is only for inpatient T&A and is not applicable to outpatient or observation cases. Lastly, as this is the first attempt at a nationwide readmission database, there are likely to be sampling issues that will be recognized and improved with future releases. This data should therefore be viewed as just one source of readmission data and not the only source. Nonetheless, this represents the largest cross-sectional study to date of readmissions after T&A that includes an entire representative population of patients. This study also adds to the growing literature concerning the quality and efficiency of healthcare, especially with the increasing spotlight on readmissions as a measure of quality of care. A high readmission rate can adversely affect hospital revenue and prestige. Though post-T&A hemorrhage is most widely reported, the current study reports dehydration and pain as equally common admitting diagnosis and the areas where interventions could potentially reduce readmissions.
Table 2 Simple logistic regression for readmission after tonsillectomy with or without adenoidectomy–2013 Nationwide Readmission Database (NRD). Variable
Odds Ratioa
95% CI
P value
Age, years Age ≤12 years Age > 12 years Male Female Medicaid Private Ins. Lowest Inc. Qrtl. Highest Inc. Qrtl. Respiratory Complicationb Complication of careb Intubation & mechanical ventilationb OSAb Tonsillitisb Index LOS
1.00 0.58 1.7 0.68 1.5 1.1 0.92 0.84 1.1 1.5 3.3 4.0 1.1 0.71 1.03
0.98 to 1.1 0.33 to 1.0 0.98 to 3.0 0.44 to 1.1 0.93 to 2.3 0.70 to 1.8 0.58 to 1.5 0.52 to 1.4 0.61 to 1.9 0.45 to 5.6 1.5 to 7.3 1.3 to 12.4 0.70 to 1.8 0.32 to 1.6 1.00 to 1.06
.25 .06 .06 .10 .10 .64 .75 .47 .80 .48 .002 .02 .60 .41 .02
Abbreviations: LOS = length of hospital stay in days; Inc Qrtl = income quartile; Ins = insurance; OSA = obstructive sleep apnea. a Odds ratio based on simple logistic regression. b See ICD9 or CCS codes in Appendix for definitions.
The mean age among readmissions was 5.6 years (95% CI, 4.4 to 6.7); median age 4 (IQR, 2 to 8). Of those readmitted, 84% (95% CI, 75%–90%) were younger than 12 years old. The majority of readmissions were female (51%; 95% CI 40%–61%). Among readmitted patients, 26% (95% CI, 18%–36%) had a postoperative hemorrhage. Other diagnoses included dehydration (47%; 95% CI, 36%–57%) and pain (16%; 95% CI, 9.8%–26%). Less frequent diagnoses include postoperative infection (1.6%; 95% CI, 0.4%–5.8%) and pneumonia (1.3%; 95% CI 0.3%–5.2%). The average time to readmission was 7.3 days (95% CI, 5.6 to 8.9). The median was 5 days with an interquartile range of 3–8 days. The average time to readmission for children with postoperative pain was 4.5 days (95% CI, 3 to 6), median 4 days (IQR 2 to 5); postoperative dehydration was 4 days (95% CI, 3 to 5), median 3 days (IQR 2 to 6); postoperative hemorrhage was 6 days (95% CI, 5 to 8), median 6 days (IQR, 5 to 7); and postoperative infection was 8 days (95% CI, 6 to 10), median 9 days (IQR, 6 to 9). Simple logistic regression analysis for readmission (Table 2) found length of stay (OR, 1.03; 95% CI, 1.0 to 1.06), need for respiratory ventilation (e.g., mechanical ventilation) during the index admission (OR, 4.0; 95% CI [1.3 to 12.4]), medical or surgical complication (OR, 3.3; 95% CI, 1.5 to 7.3) increased the likelihood of readmission (see Appendix 1 for associated ICD-9 codes). The following variables were not associated with increased risk of readmission: age, gender, payer status, median income of zip code and indication for T&A (OSA versus tonsillitis).
5. Conclusion The current study uses a newly available database to quantify, at a national level, readmissions following T&A in children. The most common admitting diagnoses were pain, dehydration, and hemorrhage. Further studies are needed to look at ways of reducing pain, dehydration, and hemorrhage after T&A in children.
4. Discussion Our analysis of the Nationwide Readmission Database for readmissions after T&A demonstrated a readmission rate of 3.6%, consistent with the 2.3–11% reported in other studies [1–3]. Our study also found that readmissions were most commonly due to postoperative pain, dehydration, and bleeding. Children with post-operative pain and dehydration presented several days earlier than patients with postoperative bleeding. The current study presents an updated and contemporary analysis on readmissions following T&A that can be used to benchmark national readmission rates since previous studies have been limited to children's hospitals [1] or to a single [2] or a few [3] states. Besides affirming the causes of readmission, our analysis showed that children with longer lengths of hospital stays, patients requiring airway support, and those with a medical or surgical complication during the index admissions were at increased risk of readmission
Disclosures None to report. Conflicts of interest None to report. External funding None. 12
International Journal of Pediatric Otorhinolaryngology 107 (2018) 10–13
R.F. Johnson et al.
Appendix 1. ICD-9 conditions and procedural codes
Clinical Classification Software Codea
Variable
ICD-9 Code
Tonsillectomy with or without adenoidectomy Hemorrhage complicating procedure Respiratory intubation and mechanical ventilation Complications of surgical procedures or medical care Postoperative Infection Respiratory Complication Dehydration Acute Postoperative Pain Obstructive Sleep Apnea Tonsillitis
282, 283 998.11 9390, 9604, 9605, 9670, 9671, 9672
a
238 9985, 99851, 99859 9971, 99731, 99732, 99739 27651 33818 32723 47400, 47402, 463, 0340
Clinical classification software: https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccsfactsheet.jsp.
Trends 27 (3) (1995) 98–100. [12] W.C. Lee, J.F. Sharp, Complications of paediatric tonsillectomy post-discharge, J. Laryngol. Otol. 110 (2) (1996) 136–140. [13] A.C. Norrington, L.M. Flood, T. Meek, M.R. Tremlett, Does day case pediatric tonsillectomy increase postoperative pain compared to overnight stay pediatric tonsillectomy? a prospective comparative audit, Paediatr. Anaesth. 23 (8) (2013) 697–701. [14] V. Raut, N. Bhat, J. Kinsella, J.G. Toner, A.R. Sinnathuray, M. Stevenson, Bipolar scissors versus cold dissection tonsillectomy: a prospective, randomized, multi-unit study, Laryngoscope 111 (12) (2001) 2178–2182. [15] Y.M. Takwoingi, M. Shykhon, M. Wake, Effect of post-operative analgesia duration on post-tonsillectomy readmission rate: comparison of five-day and 14-day regime, J. Laryngol. Otol. 121 (10) (2007) 968–972. [16] R.N. Axon, M.V. Williams, Hospital readmission as an accountability measure, J. Am. Med. Assoc. 305 (5) (2011) 504–505. [17] N.R. Payne, A. Flood, Preventing pediatric readmissions: which ones and how? J. Pediatr. 166 (3) (2015) 519–520. [18] J.C. Gay, R. Agrawal, K.A. Auger, et al., Rates and Impact of potentially preventable readmissions at children's hospitals, J. Pediatr. 166 (3) (2015) 613–619 e5. [19] HCUP, Clinical Classifications Software (CCS) for ICD-9-CM. Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality, Rockville, MD, 2006-2009www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp Accessed March, 2016. [20] M.M. Statham, R.G. Elluru, R. Buncher, M. Kalra, Adentonsillectomy for obstructive sleep apnea syndrome in young children: prevalence of pulmonary complications, Arch. Otolaryngol. Head Neck Surg. 132 (5) (2006) 476–480. [21] A. Belloso, A. Chidambaram, P. Morar, M.S. Timms, Coblation tonsillectomy versus dissection tonsillectomy: postoperative hemorrhage, Laryngoscope 113 (11) (2010) 2010–2013.
References [1] S. Mahant, R. Keren, R. Localio, et al., Variation in quality of tonsillectomy perioperative care and revisit rates in Children's hospitals, Pediatrics 133 (2) (2014) 280–288. [2] M.B. Edmonson, J.C. Eickhoff, C. Zhang, A population-based study of acute care revisits following tonsillectomy, J. Pediatr. 166 (3) (2015) 607–612 e5. [3] S. Shay, N.L. Shapiro, N. Bhattacharyya, Revisit rates and diagnoses following pediatric tonsillectomy in a large multistate population, Laryngoscope 125 (2) (2015) 457–461. [4] N. Bhattacharyya, L.J. Kepnes, Revisits and postoperative hemorrhage after adult tonsillectomy, Laryngoscope 124 (7) (2014) 1554–1556. [5] R. Benson-Mitchell, A.R. Maw, Assessment of sequelae at home following adenotonsillectomy. a basis for day-case management? Clin. Otolaryngol. Allied Sci. 18 (4) (1993) 282–284. [6] N. Bhattacharyya, Evaluation of post-tonsillectomy bleeding in the adult population, Ear Nose Throat J. 80 (8) (2001) 544–549. [7] A.S. Evans, A.M. Khan, D. Young, R. Adamson, Assessment of secondary haemorrhage rates following adult tonsillectomy – a telephone survey and literature review, Clin. Otolaryngol. Allied Sci. 28 (6) (2003) 489–491. [8] K. Ghufoor, A. Frosh, G. Sandhu, J. Hanif, Post-tonsillectomy patient care in the community, Int. J. Clin. Pract. 54 (7) (2000) 420–423. [9] R.L. Harris, J.E. Mitchell, D.A. Jonathan, A telephone audit in parallel with the UK national tonsillectomy audit to investigate re-admission as a measure of secondary haemorrhage rate, Auris Nasus Larynx 35 (2) (2008) 220–223. [10] E.K. Hoddeson, C.G. Gourin, Adult tonsillectomy: current indications and outcomes, Otolaryngol. Head Neck Surg. 140 (1) (2009) 19–22. [11] M. Kuo, D. Hegarty, A. Johnson, S. Stevenson, Early post-tonsillectomy morbidity following hospital discharge: do patients and GPs know what to expect? Health
13