Variation in Quality of Tonsillectomy Perioperative Careand Revisit Rates in Children’s Hospitals

WHAT’S KNOWN ON THIS SUBJECT: Tonsillectomy is one of themost commonly performed surgeries in children and is one of themost cumulatively expensive conditions in pediatric hospital care.Little is known about how the quality of tonsillectomy care variesacross hospitals.

WHAT THIS STUDY ADDS: In a large cohort of low-risk childrenundergoing same-day tonsillectomy, there was substantialvariation in quality measures of process, dexamethasone andantibiotic use, and outcome, revisits to the hospital within thefirst 30 days after surgery.

abstractOBJECTIVE: To describe the quality of care for routine tonsillectomy atUS children’s hospitals.

METHODS: We conducted a retrospective cohort study of low-riskchildren undergoing same-day tonsillectomy between 2004 and 2010at 36 US children’s hospitals that submit data to the PediatricHealth Information System Database. We assessed quality of careby measuring evidence-based processes suggested by nationalguidelines, perioperative dexamethasone and no antibiotic use, andoutcomes, 30-day tonsillectomy-related revisits to hospital.

RESULTS: Of 139 715 children who underwent same-day tonsillectomy,10 868 (7.8%) had a 30-day revisit to hospital. There was significantvariability in the administration of dexamethasone (median 76.2%, range0.3%–98.8%) and antibiotics (median 16.3%, range 2.7%–92.6%) acrosshospitals. The most common reasons for revisits were bleeding (3.0%)and vomiting and dehydration (2.2%). Older age (10–18 vs 1–3 years)was associated with a greater standardized risk of revisits for bleeding anda lower standardized risk of revisits for vomiting and dehydration. Afterstandardizing for differences in patients and year of surgery, there wassignificant variability (P, .001) across hospitals in total revisits (median7.8%, range 3.0%–12.6%), revisits for bleeding (median 3.0%, range1.0%–8.8%), and revisits for vomiting and dehydration (median 1.9%,range 0.3%–4.4%).

CONCLUSIONS: Substantial variation exists in the quality of care forroutine tonsillectomy across US children’s hospitals as measuredby perioperative dexamethasone and antibiotic use and revisits tohospital. These data on evidence-based processes and relevantpatient outcomes should be useful for hospitals’ tonsillectomyquality improvement efforts. Pediatrics 2014;133:280–288

AUTHORS: Sanjay Mahant, MD, FRCPC, MSc,a Ron Keren,MD, MPH,b,c Russell Localio, PhD,b,c Xianqun Luan, MS,b

Lihai Song, MS,b Samir S. Shah, MD, MSCE,d Joel S.Tieder, MD, MPH,e Karen M. Wilson, MD, MPH,f LisaElden, MD, MSc,g,h and Rajendu Srivastava, MD, FRCPC,MPH,i,j for the Pediatric Research in Inpatient Settings(PRIS) NetworkaDivision of Pediatric Medicine, Pediatric Outcomes ResearchTeam (PORT), Department of Pediatrics, Institute of Health Policy,Evaluation and Management, University of Toronto, Child HealthEvaluative Sciences, Research Institute, Hospital for SickChildren, Toronto, Canada; bDivision of General Pediatrics, Centerfor Pediatric Clinical Effectiveness, Children’s Hospital ofPhiladelphia, Philadelphia, Pennsylvania; cDepartment ofBiostatistics and Epidemiology, Perelman School of Medicine,University of Pennsylvania, Philadelphia, Pennsylvania; dDivisionsof Hospital Medicine and Infectious Diseases, CincinnatiChildren’s Hospital Medical Cente r, Cincinnati, Ohio; eDivision ofInpatient Medicine, Department of Pediatrics, University ofWashington and Seattle Children’s Hospital, Seattle, Washington;fSection of Pediatric Hospital Medicine, Children’s HospitalColorado, Aurora, Colorado; gDivision of Pediatric Otolaryngology,Children’s Hospital of Philadelphia, and hPerelman School ofMedicine, University of Pennsylvania, Philadelphia, Pennsylvania;iDivision of Inpatient Medicine, Department of Pediatrics,University of Utah Health Sciences Center, Salt Lake City, Utah;and jInstitute for Healthcare Delivery Research, IntermountainHealthcare Inc, Salt Lake City, Utah

KEY WORDStonsillectomy, perioperative, quality of care, dexamethasone,children’s hospitals, readmissions, antibiotics, regional variation

ABBREVIATIONSAAO-HNS—American Academy of Otolaryngology–Head and NeckSurgeryAPR-DRG—All Patient-Refined Diagnosis-Related GroupsCI—confidence intervalED—emergency departmentICD-9-CM—International Classification of Diseases, NinthRevision, Clinical ModificationOR—odds ratioPHIS—Pediatric Health Information System

(Continued on last page)

280 MAHANT et al by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from

More than 500 000 tonsillectomies areperformed each year in children in theUnited States,most commonly for sleep-disordered breathing and recurrentthroat infections.1,2 It is the secondmostcommon and the ninth most cumula-tively expensive reason for care in USchildren’s hospitals.3 The 2011 AmericanAcademy of Otolaryngology–Head andNeck Surgery (AAO-HNS) practice guide-lines delineate high-quality tonsillec-tomy perioperative care.2 The guidelinesrecommend routine perioperative ad-ministration of dexamethasone and noantibiotics. These recommendationsare based on evidence gathered fromtrials over the past 2 decades,2,4,5 whichshowed that dexamethasone, adminis-tered on the day of surgery, reducespostoperative nausea, vomiting, and pain,4

whereas perioperative antibiotics donot reduce postoperative bleeding,pain, or infections.5 The guideline rec-ommends monitoring bleeding rates,as bleeding is the most common andserious complication.6,7

To understand potential opportunitiesfor quality improvement around tonsil-lectomy care, information is needed onperioperative evidence-based processessuggested by guidelines, such as dexa-methasone and antibiotic use. Further-more, information isneededonoutcomes,such as the prevalence, patterns, andrisk factors for revisits to the hospitalafter tonsillectomy and variations inrevisit rates across hospitals. The ob-jective of this study was to describe thequality of perioperative tonsillectomycare by assessing perioperative dexa-methasone and antibiotic use and re-visit rates in a large multicenter cohortof low-risk children undergoing same-day tonsillectomy at US children’shospitals.

METHODS

Study Design and Data Sources

We conducted a retrospective multi-center observational study of children

undergoing same-day tonsillectomy atUS children’s hospitals from 2004 to2010. Data were obtained from the Pe-diatric Health Information System (PHIS),a database of administrative andbilling data from 43 freestanding chil-dren’s hospitals in 18 states and theDistrict of Columbia, affiliated with theChildren’s Hospital Association. The da-tabase contains detailed information ondemographics, diagnosis and pro-cedure codes, service locations, andtheir charges. Charges are mapped toa common set of clinical transactioncodes, which include imaging studies,clinical services, laboratory tests, phar-macy, supplies, and room charges.Patients have unique identifiers, whichpermit longitudinal analysis of hospitalencounters. Data quality and reliabilityare ensured through a joint effort be-tween the Children’s Hospital Associa-tion and participating hospitals.8

The institutional review board of theChildren’s Hospital of Philadelphiadeemed this study exempt from reviewunder 45 CFR 46.102(bf), as the par-ticipants were not readily identifiable.

Study Population

Eligibility criteria were chosen to as-semble a relatively homogeneous, low-risk cohort of children for which theAAO-HNS guidelines recommend routineuse of dexamethasone and no anti-biotics.2 Children 1 to 18 years who un-derwent tonsillectomy with or withoutadenoidectomyandwhoweredischargedon the same day as the surgery ata hospital that submitted data to PHISfrom January 1, 2004, to December 31,2010, were eligible. Children wereidentified using International Classifi-cation of Diseases, Ninth Revision,Clinical Modification (ICD-9-CM) pro-cedure codes of tonsillectomy (28.2) ortonsillectomy and adenoidectomy (28.3).We excluded children if they had tonsil-lectomy and/or adenoidectomy in theprevious 2 years; a peritonsillar abscess

or malignancy; an emergency de-partment (ED) charge or admissionthrough the ED; additional proceduresincluding tympanostomy tubes; or a com-plex chronic condition,9 craniofacial ab-normalities,10 diabetes, or a disorder inhemostasis (Supplemental Table 4). Wealso excluded children if the hospitalsubmitted incomplete or no data for theindex same-day surgery or revisits. Toexclude records inwhich therewasahighlikelihood of coding errors, we excludedcost outliers at the bottomand top 0.1%oftotal standardized cost3 and also thosesurgeries without a perioperative roomor clinical service charge.

Covariates

Demographic characteristics analyzedwere age (1–3, 4–9, and 10–18 years),gender, race (white, black, other), andinsurance type (government or non-government, including commercial, orself-pay). Age categories were chosenbased on known associations with in-creased risk of complications (eg, age,3 or 4)11 and to ensure homogeneousrisk within categories. A diagnosisof asthma, assessed using ICD-9-CMcodes,12 was included as a covariate,as it is associated with increased risk ofcomplications,13,14 is prevalent, and isnot an excluded chronic condition in theAAO-HNS guidelines. Patient illness se-verity was classified using the APR-DRG(All Patient-Refined Diagnosis-RelatedGroups) severity classification system15

into 1 of 2 severity levels: minor (level 1)or nonminor (level 2 [moderate], 3[major], 4 [extreme]). We classified theindication for surgery into 2 categories:infection (children who had an infection-related code) or not infection related(children who had an airway obstruc-tion or other diagnosis code), based ona previous ICD-9-CM classification.1

Outcome Measures

Evidence-based Process Measures

We determined whether children re-ceived dexamethasone and antibiotics

ARTICLE

PEDIATRICS Volume 133, Number 2, February 2014 281 by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from

on the day of surgery based on thepresence of pharmacy clinical trans-action codes. For each hospital, wecalculated a summary performancescore,16 defined as the number of timesa hospital administered dexametha-sone but not antibiotics to an individualchild, divided by the number of childrenwho had a tonsillectomy at the hospi-tal, multiplied by 100.

Revisits to Hospital

We evaluated revisits to the hospital byday after discharge to 30 days after theindexsurgery.Revisits included those to

the ED and admission to the hospital.Two of the investigators (S.M., R.S.)reviewed all ICD-9-CM principal di-agnoses related to revisits and classi-fied them as tonsillectomy-related orunrelated. Only tonsillectomy-relatedrevisits were included. For example,arevisit for fractureof thehumeruswascategorized as an unrelated revisit andexcluded. Tonsillectomy-related revisitswere then classified into 5 reasonsbased on the ICD-9-CM principal di-agnosiscodeon therevisit: (1)bleeding,(2) vomiting and dehydration (codesrelated to dehydration, vomiting, or

poor feeding), (3) pain, (4) infection(infections related to tonsillectomy orthose that could be acquired duringhospital care), (5) respiratory, and (6)other (including unspecified surgicalcomplications and drug reactions).

Analysis

To estimate the hazards of revisits overtime from surgery, we implementeda series of reason-specific discrete time-failure models using logistic re-gression.17,18 Data were processed intoperson-period format, with an obser-vation representing each day from dis-charge to day 30 for each child. Wefurther expanded the data to allow fora separate observation for each poten-tial reason for revisit (eg, bleeding,vomiting, and dehydration). Using thesedata, we then constructed discrete time-failure models to estimate the hazard ofrevisit as a function of in-hospital pro-cesses (dexamethasone, antibiotics),the day postdischarge, the patient-levelcovariates, year of surgery, and hospi-tal.17,18 We divided the 30 days into 5-daycategories (1–2, 3–5, 6–7, 8–14, 15 ormore). We modeled the hazard of revisitseparately by reason for revisit to per-mit maximum flexibility in estimatingreason-specific associations of patient-level covariates and outcome.

For estimating the risk for revisit at 30days, we used logistic regression withacovariate foreachhospital toadjust forconfounding by hospital arising out ofthe differences across hospitals in therisk of revisit and the concomitant var-iation in the use of treatments acrosshospitals.19,20 For that reason, all esti-mates represent the within-hospitalassociation of the risk of revisit andin-hospital processes or patient-levelcovariates. Results from logistic re-gression were standardized, using pre-dictive margins,21 by other patient-levelcovariates, as well as by hospital andyear of admission to produce adjustedprobabilities and their 95% confidence

FIGURE 1Cohort flow diagram. a Includes inpatient, ambulatory, observation, and ED patient types in the PHISdatabase.

282 MAHANT et al by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from

bounds of revisit according to givenpatient and treatment characteristics.Standardization was to the character-istics of the population of childrenacross all hospitals.

Weestimatedvariationacrosshospitals inthe risk for revisit at 30 days, and therelativerankingsofhospitalsbyreasonforrevisit, by using mixed effects logistic re-gression with fixed effects for patientcharacteristics and random interceptsfor hospitals. These models generatedhospital-level predictions of revisit ratesthat not only adjust for interhospital dif-ferencesinpatientcharacteristicsbutalsocompensate for the inherent instability ofraw hospital rates arising out of finitenumbers of revisit events per hospital.22,23

To assess the trends in antibiotic anddexamethasone use and total revisitrates over time, we divided the studyperiodinto6-monthintervals.Weestimatedthe trend over time within hospitalsusing both marginal and fixed-effectsmodels adjusting for the patient-levelcovariates. Then we used likelihoodratio tests to evaluate the possibility ofdifferent trendsacrosshospitals. Finally,we estimated probabilities of antibioticand dexamethasone use and revisits atthe start and end of the study period,standardized by patient characteristics.This allowedus toevaluatehospital-levelchanges over time adjusted for possibledifferences in patients across hospitalsand over time.

Loss of data to missing covariates wasminimal (n = 7). Observations withmissing data were excluded in multi-variable models. All models were gen-erated by using Stata version 12.1(Stata Corp, College Station, TX).

RESULTS

Study Population

Of 297 155 children who underwenttonsillectomy during the study period,139 715 children from 36 hospitals metthe eligibility criteria (Fig 1 and Table 1).

The mean age of the cohort was 7.0years (SD 3.8), and the illness severitywas low in 94%. Almost all patients hadtonsillectomywith adenoidectomy (91.0%)and the indication for surgery was mostcommonly airway obstruction (58.4%)and infection (33.4%).

Evidence-based Process Measures

Of the 139 715 admissions for tonsil-lectomy, 69.6% received dexametha-sone and 31.1% received antibiotics onthe day of surgery. Hospitals varied intheir use of both dexamethasone andantibiotics (Fig 2). At the hospital level,the median percentage of patients whoreceived dexamethasone was 76.2%(range 0.3%–98.8%) and antibioticswas 16.3% (range 2.7%–92.6%). Themedian summary performance scoreacross hospitals was 43.4% (range0.2%–91.3%).

Revisits to Hospital

Patient Level

A total of 10 868 (7.8%) children hada revisit within 30 days of the surgery,6897 (63.5%) to the ED and 3971 (36.5%)admitted to the hospital; 4182 (3.0%)had a revisit for bleeding, 3011 (2.2%)for vomiting and dehydration, 1073(0.8%) for pain, 1060 (0.8%) for infec-tions, 542 (0.5%) for respiratory prob-lems, and 1000 (0.7%) for other reasons.

Most revisits (93.8%) occurredwithin 15days after surgery and our analysisidentified distinct patterns of revisits forthe 2 most common reasons: bleedingand vomiting and dehydration (Fig 3).The highest rate of revisits for vomitingand dehydration was on days 1 and 2(4.52 per 1000 days of follow-up [95%confidence interval (CI) 4.27–4.77]) and98.6% of revisits occurred by day 15. Incontrast, the highest rate of revisits forbleeding was delayed occurring at theday 6 and 7 time interval (4.32 per 1000days of follow-up [95% CI 4.08–4.56])and 97.8%of revisits occurred by day 15.

Revisits and Patient-level Covariates

Age was the covariate most associatedwith standardized riskof revisits for themost common reasons for revisits.Older age, relative to youngest age (1–3years), was associated with an in-creased risk of revisits for bleeding: age

TABLE 1 Study Population

Patientcharacteristics,n = 139 715

Surgery type, n (%)Tonsillectomy with

adenoidectomy127 147 (91.0)

Tonsillectomy 12 568 (9.0)Gender, n (%)Female 72 280 (51.7)

Age, y, n (%)mean (SD) 7.0 (3.8)1–3 22 146 (15.9)4–9 86 764 (62.1)10–18 30 805 (22.0)

Payer, n (%)Government 50 339 (36.0)

Race, n (%)White 98 739 (70.7)Black 18 549 (13.3)Other 22 427 (16.0)

Asthma, n (%) 13 975 (10.0)APR-DRG severity,

n (%)Minor 131 366 (94.0)Nonminor 8349 (6.0)

Indication diagnosiscodes, n (%)Airway obstruction 81 656 (58.4)Infection 46 695 (33.4)Both infection

and airwayobstruction

10 145 (7.3)

Other 1219 (0.9)Hospitalcharacteristics,n = 36Region, n (%)

of hospitalsMidwest 10 (27.8)Northeast 5 (13.9)South 13 (36.1)West 8 (22.2)

Tonsillectomyvolume peryear, median(range)2004 483 (154–1980)2005 535 (108–2013)2006 528 (85–2149)2007 555 (136–2235)2008 560 (130–2020)2009 682 (125–1761)2010 707(113–1679)

ARTICLE

PEDIATRICS Volume 133, Number 2, February 2014 283 by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from

4–9 (odds ratio [OR] 1.36, 95% CI 1.05–1.78; P = .02), age 10–18 (OR 2.74, 95% CI2.08–3.60; P , .001). In contrast, olderage was associated with a reduced riskof revisits for vomiting and dehydration:age 4–9 (OR 0.67, 95% CI 0.55–0.84; P =.005), age 10–18 (OR 0.59, 95% CI 0.45–0.76; P = .001). Government insurance,black race, nonminor illness severity,infection indication, and asthma wereassociated with increased cumulative30-day risk of total revisits (Table 2).

Hospital Level

The hospital median 30-day total revisitrate was 7.8%, but this varied acrosshospitals from 3.0% to 12.6%. Thehospital median revisit rate for bleed-ing was 3.0% (range 1.0%–8.8%), andfor vomiting and dehydration was 1.9%

(range 0.3%–4.4%). Variability in allthese outcomes remained after stan-dardizing for patient-level covariatesand year (P , .001) (Fig 4).

Trends Over Time

Dexamethasone usage was flat overtime, with an OR of 1.00 for each 1-yearchange in time. By contrast, antibioticusage dropped significantly over time(OR of 0.97 per year). For both processmeasures, trends in usage as well aslevel of usage varied widely acrosshospitals. Total revisit rates did notchangeover time (OR1.00). The levels ofrates did vary significantly acrosshospitals, however. Variation in trendsacross hospital remained significant,but far less so than with the antibioticsand dexamethasone trends.

Evidence-based ProcessesMeasures and Revisits

There was no significant associationbetween dexamethasone or antibioticuse and 30-day cumulative risk of totalrevisits (see Table 3).

DISCUSSION

In this largemulticenter low-risk cohortof children undergoing same-day ton-sillectomy, we observed substantialvariation across US children’s hospitalsin the quality of tonsillectomy care,measured in terms of perioperativeprocess measures (use of dexametha-sone and antibiotics) and outcomes(standardized revisits to hospital).Some hospitals provided almost nopatientswith the current recommended

FIGURE 2Variation in perioperative dexamethasone and antibiotic use across 36 children’s hospitals. Each hospital’s dexamethasone use, antibiotic use, and summaryscore is indicated by a shaded circle, open square, and horizontal line, respectively. The dexamethasone use indicates the percentage of patients in eachhospital who received perioperative dexamethasone. The antibiotic use indicates the percentage of patients in each hospital who received perioperativeantibiotics. The summary score is the number of times a hospital performed the appropriate action; that is, the number of times a hospital administereddexamethasone but not antibiotics to an individual child, divided by the number of children who had a tonsillectomy at the hospital, multiplied by 100.

284 MAHANT et al by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from

care of dexamethasone and no anti-biotics compared with 91% at otherhospitals. Revisits occurred in 7.8% ofchildren within 30 days after this elec-tive surgery, with bleeding and vomitingand dehydration accounting for three-quarters of total revisits. There wasa fourfold difference across hospitals inrevisit rates even after accounting fortime and types of patients.

The 3.0% rate of revisits for bleedingfound in our study is similar to the ratesof secondary bleeding requiring hos-pital care reported in prospectivestudies.6,24–26 Also consistent with pre-vious studies,6,26 we observed a de-layed peak rate in secondary bleedingthat corresponds with sloughing of thefibrin clot that coats the tonsillar fos-sae formed as part of the healingprocess. A UK prospective multicenterstudy that included 21 063 childrenreported a secondary bleeding rate of1.9% to 3.0% depending on age category.6

It did not include ED revisits and found

FIGURE 3Variation in 30-dayprobability of revisits across 36 children’s hospitals. Variation in 30-dayprobability ofrevisits by reason for revisit (all reasons, bleeding, or vomiting and dehydration) across 36 children’shospitals after adjusting for patient-level factors. A dark circle indicates each hospital’s 30-dayprobability of revisit by reason. The dashed lines connecting the 3 probabilities (all reasons, bleeding,and vomiting and dehydration) indicate values from a single hospital. Probabilities were estimatedusing mixed effects logistic regression, models with a random intercept for hospital, and with fixedeffects for patient age, race, insurance (government or private), diagnosis of asthma, indication fortonsillectomy, and year of surgery to account for possible differences across hospitals and over time intypes of patients treated. Probabilities are further adjusted to improve prediction (of these proba-bilities) and to avoid overstating the degree of interhospital variation, via a weighted average ofhospital-specific estimates and the overall estimates of all hospitals. For all revisits, and those frombleeding and dehydration/vomiting, the dispersion across hospitals of these adjusted probabilitieswas significantly greater (P , .001) than expected at random.

TABLE 2 Patient-Level Covariates and 30-Day Cumulative Risk of Revisits

Characteristic Total Revisits Bleeding Revisits Vomiting-Dehydration Revisits

30-d CumulativeRisk (95% CI)

Adjusted 30-d Cumulative Risk Adjusted 30-d Cumulative Risk Adjusted

OR (95% CI) (95% CI) OR (95% CI) (95% CI) OR (95% CI)

Age, y1–3 8.2 (7.8–8.5) 1 [Reference] 1.8 (1.6–2.0) 1 [Reference] 2.7 (2.5–3.0) 1 [Reference]4–9 7.2 (7.0–7.4) 0.84 (0.74–0.95)** 2.6 (2.5–2.7) 1.36 (1.05–1.78)* 2.2 (2.1–2.3) 0.67 (0.55–0.84)***10–18 9.1 (8.8–9.4) 1.05 (0.91–1.21) 4.9 (4.7–5.2) 2.74 (2.08–3.60)*** 1.8 (1.6–1.9) 0.59 (0.45–0.76)***

GenderMale 7.8 (7.6–8.0) 1 [Reference] 3.1 (3.0–3.2) 1 [Reference] 2.0 (1.9–2.1) 1 [Reference]Female 7.8 (7.6–8.0) 1.01 (0.97–1.04) 2.9 (2.8–3.0) 0.93 (0.87–0.99)* 2.3 (2.2–2.4) 1.13 (1.05–1.21)**

SeverityLevel 1 7.7 (7.6–7.8) 1 [Reference] 3.0 (2.9–3.0) 1 [Reference] 2.1 (2.1–2.2) 1 [Reference]Level 2, 3 8.8 (8.3–9.4) 1.16 (1.08–1.26)*** 3.0 (2.6–3.3) 1.00 (0.88–1.13) 2.6 (2.2–2.9) 1.20 (1.04–1.39)*

AsthmaAbsent 7.5 (7.4–7.7) 1 [Reference] 3.0 (3.9–3.1) 1 [Reference] 2.0 (2.0–2.2) 1 [Reference]Present 9.7 (9.3–10.0) 1.33 (1.25–1.41)*** 3.3 (3.0–3.6) 1.12 (1.01–1.23)* 2.7 (2.4–3.0) 1.31 (1.17–1.46)

RaceWhite 7.7 (7.5–7.9) 1 [Reference] 3.0 (2.9–3.2) 1 [Reference] 2.2 (2.1–2.3) 1 [Reference]Black 9.1 (8.7–9.5) 1.20 (1.13–1.27)*** 3.0 (2.7–3.2) 0.97 (0.88–1.07) 2.5 (2.3–2.8) 1.17 (1.05–1.31)**Other 7.0 (6.6–7.4) 0.90 (0.84–0.95)** 2.8 (2.5–3.0) 0.90 (0.81–0.99)* 1.7 (1.6–1.9) 0.80 (0.70–0.90)***

PayerNongovernment 7.0 (6.8–7.2) 1 [Reference] 3.0 (2.9–3.1) 1 [Reference] 1.9 (1.8–2.0) 1 [Reference]Government 9.2 (8.9–9.5) 1.35 (1.30–1.42)*** 3.0 (2.9–3.2) 1.02 (1.00–1.15) 2.6 (2.4–2.7) 1.36 (1.26–1.48)***

IndicationNon–infection related 7.5 (7.3–7.7) 1 [Reference] 3.1 (3.0–3.2) 1 [Reference] 2.0 (1.9–2.2) 1 [Reference]Infection related 8.1 (7.9–8.3) 1.08 (1.04–1.13)*** 2.9 (2.7–3.0) 1.07(1.00–1.15)* 2.3 (2.2–2.5) 1.15 (1.06–1.25)**

Cumulative risk per 100 patients and ORs standardized (adjusted) for all other patient levels, covariates, year of admission, and hospital by using logistic regression and predictive margins.* P , .05; ** P , .01; *** P , .001.

ARTICLE

PEDIATRICS Volume 133, Number 2, February 2014 285 by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from

that the most common reasons for re-admission in children and adults werefor bleeding followed by pain, vomiting,and then fever. This study also observedwide variation in bleeding complications,from 0% to 12%, after adjusting for pa-tient factors, surgical technique, and sur-geon experience across a large numberof hospitals.

We did not detect an important asso-ciation between process measures andthe outcome of total revisits. Althoughdexamethasone is effective in reducing

postoperative nausea and vomiting,4

this may not translate into reductionsin revisits. Outcomes, such as length ofpostoperative stay and unplanned over-night admission, may be more closelylinked to a reduction in postoperativenausea and vomiting. As antibioticshave not been shown to reduce post-tonsillectomy morbidity,5 one wouldnot expect to see a reduction in revisits,unless there were a significant numberof antibiotic-associated adverse eventsthat resulted in revisits.

Further investigation is necessary tounderstand the reasons for the sub-stantial variation in dexamethasone andantibiotic use and revisits across hos-pitals. Given the relatively few hospitalsin our cohort, we had limited statisticalpower to examine hospital factors as-sociated with better performance. Thewidevariation indexamethasoneusemayrelate to a concern of increased bleedingrisk as reported in 1 trial.27 However, theAAO-HNS guidelines,2 and a subsequenttrial,28 have concluded that there is noclinically important increased risk. Thevariation in revisits may indicate differ-ences in processes of care during theindex hospitalization (eg, surgical tech-nique,25 anesthesia, pain management,discharge education) or postdischarge(eg, follow-up care, pain management).Although the variation might also reflectdifferences in patient factors such ashealth status, this explanation is lesslikely as we restricted our cohort toa low-risk population and accounted forhospital differences in patients in ouranalyses. Whatever the reasons, it is safeto say that hospitals with 12.6% revisitrates would be interested in un-derstanding how to reduce their revisitrates down to the 3.0% rate observed atsome of their peer institutions.

Our detailed data on tonsillectomy-related revisits in children can beused in several ways to inform qualitymeasurement around tonsillectomycare. For example, given thatmore than90% of revisits occurred in the first 15days after surgery, this time period(versus 30 days) may be more appro-priate for assessment of revisits. Also,our data on patient-level covariates re-veal that age is an important variable forrisk adjustment when reporting tonsil-lectomy revisit data. Furthermore, as themost common reasons for revisits(bleeding and vomiting and dehydration)are driven in different directions by thispatient-level factor (ie, higher versuslowerage), havedifferent ratesof revisits

FIGURE 4Standardized rates of revisit by reason for revisit between days 1 through 30 after discharge. Stan-dardized rates of revisit between days 1 through 30 after discharge per 1000 patient days of observationand 95% confidence bounds by reason for revisit. Rates were estimated separately for each reason forrevisitbyusingdiscrete time failuremodel implementedusing logistic regressionandwith timeofrevisitcategorized by day (1–2, 3–5, 6–7, 8–14, and 15–30) to allow for the effect of time to vary without therestriction of a particular parametric form (such as a linear trend). Adjacent days with similar rates ofrevisits were groups to avoid instability of estimates from small sample sizes. All rates are marginallystandardized by patient characteristics (age, gender, race), severity (APR-DRG), indication for tonsil-lectomy, insurance (government or private), year of surgery, and hospital to represent within-hospitaladjusted rates and avoid confounding by these factors or by hospital characteristics.

TABLE 3 Antibiotic and Dexamethasone Use and 30-Day Cumulative Risk of Total Revisits

Treatment No. (n = 139 708) 30-d CumulativeRisk of Total

Revisits (95% CI)

Risk Difference(95%CI)

P Value AdjustedOR (95% CI)

Antibiotic 43 425 7.6 (7.3 to 7.8) 20.3 (20.7 to 0.1) .12 0.96 (0.91 to 1.01)No antibiotic 96 283 7.9 (7.2 to 7.9)

Dexamethasone 97 242 7.9 (7.7 to 8.1) 0.3 (20.1 to 0.8) .12 1.02 (0.90 to 1.15)No dexamethasone 42 466 7.5 (7.2 to 7.9)

Cumulative risk per 100 patients and ORs standardized (adjusted) for patient-level covariates (age, gender, severity level,asthma, race, payer, indication), year of admission, and hospital by using logistic regression and predictive margins.

286 MAHANT et al by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from

over time, and are likely driven by dif-ferentprocessesofcare, itwouldbemostuseful if these data on revisits arereported by reasons, in addition to totalrevisits, for hospitals’ improvementefforts. Finally, revisits for vomiting anddehydration also account for a significantproportion of total revisits, and in thecontext of this common surgery shouldalso be given priority for quality im-provement measurement and reporting.

Ourstudyfindingsshouldbeinterpretedinlight of several limitations. First, this studydoes not address the entire continuum oftonsillectomycare.Thedatasourcedidnotpermit us to assess surgical appropri-ateness or important outcomes, such asresolution of sleep-disordered breathing,recurrent infections, or quality of life.Second, our data are representative of USchildren’s hospitals, which account forone-third of US pediatric hospital-izations,29 andmay not be generalizable toother hospital settings. Third, we did notinclude children with chronic complexdiseases or children admitted throughthe ED for emergent tonsillectomy. Wemight have observed greater variation incare had we included these groups.Fourth, we did not examine the issue ofvariation in post-tonsillectomy admissionpractices, which is an important quality of

care issue.30 Fifth, although parents andchildren would likely return to the samehospital for complications, they may havesought follow-up at different hospitals,leading to underestimation of revisitcounts. Sixth, given the administrativedata source, there may be misclassifi-cation in measures due to coding differ-ences across hospitals.

CONCLUSIONS

Nearly 40 years ago, Wennberg andGittelsohn31 demonstrated small areasof variation in rates of tonsillectomydespite no apparent differencesamong populations of patients in theUnited States. Our study, which focuseson 1 aspect of tonsillectomy care,shows that variation in care processesand outcomes continues today. Recentdevelopments in US health care reformand financing may serve to nudgehospitals in the right direction. Asevolving payment models, such asbundled payments that reward high-value care, are implemented, hospi-tals will need to deliver quality carewhile avoiding revisits for complica-tions to maintain their profit mar-gins.32 Quality improvement initiativesareneeded to implement current evidence

into practice, understand and dissem-inate the practices of high-performinghospitals, and improve the value ofhealth care delivered for childrenundergoing tonsillectomy.33

ACKNOWLEDGMENTSWe thank Jaime Blank, Manager for thePediatric Research in Inpatient SettingsNetwork, and Debra Hillman, managerof the Children’s Hospital of PhiladelphiaCenter for Pediatric Clinical Effective-ness, both of whom provided overallproject management for this researcheffort; members of the Pediatric Re-search in Inpatient Settings ExecutiveCouncil, Christopher Landrigan, MD,MPH, associate professor of medicineand pediatrics, Division of General Pe-diatrics and Sleep Medicine, Depart-ments of Medicine, Boston Children’sHospital and Brigham and Women’sHospital, Harvard Medical School, andTamara D. Simon, MD, MSPH, assistantprofessor, University of Washington,Seattle Children’s Hospital, who pro-vided advice on presentation of thedata; and Jacqueline Kueser, MHA,and Matt Hall, PhD, Children’s HospitalAssociation, Overland Park, KS, whoprovided advice about the data andits presentation.

REFERENCES

1. Boss EF, Marsteller JA, Simon AE. Out-patient tonsillectomy in children: de-mographic and geographic variation in theUnited States, 2006. J Pediatr. 2012;160(5):814–819

2. Baugh RF, Archer SM, Mitchell RB, et al;American Academy of Otolaryngology-Headand Neck Surgery Foundation. Clinicalpractice guideline: tonsillectomy in chil-dren. Otolaryngol Head Neck Surg. 2011;144(suppl 1):S1–S30

3. Keren R, Luan X, Localio R, et al; PediatricResearch in Inpatient Settings (PRIS) Net-work. Prioritization of comparative effec-tiveness research topics in hospitalpediatrics. Arch Pediatr Adolesc Med. 2012;166(12):1155–1164

4. Steward DL, Grisel J, Meinzen-Derr J. Steroidsfor improving recovery following tonsillec-tomy in children. Cochrane Database SystRev. 2011;(8):CD003997

5. Dhiwakar M, Clement WA, Supriya M,McKerrow W. Antibiotics to reduce post-tonsillectomy morbidity. Cochrane Data-base Syst Rev. 2010;(7):CD005607

6. The Royal College of Surgeons of EnglandNational Prospective Tonsillectomy Audit 2005.Available at: www.rcseng.ac.uk/publications/docs/national_prospective.html. AccessedOctober 20, 2012

7. Windfuhr JP, Schloendorff G, Baburi D,Kremer B. Lethal outcome of post-tonsillectomy hemorrhage. Eur Arch Oto-rhinolaryngol. 2008;265(12):1527–1534

8. Mongelluzzo J, Mohamad Z, Ten Have TR,Shah SS. Corticosteroids and mortality inchildren with bacterial meningitis. JAMA.

2008;299(17):2048–2055

9. Feudtner C, Hays RM, Haynes G, Geyer JR,Neff JM, Koepsell TD. Deaths attributed to

pediatric complex chronic conditions: na-tional trends and implications for sup-portive care services. Pediatrics. 2001;107(6). Available at: www.pediatrics.org/cgi/content/full/107/6/e99

10. Lam DJ, Jensen CC, Mueller BA, Starr JR,Cunningham ML, Weaver EM. Pediatricsleep apnea and craniofacial anomalies:a population-based case-control study. La-ryngoscope. 2010;120(10):2098–2105

ARTICLE

PEDIATRICS Volume 133, Number 2, February 2014 287 by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from

11. Brigger MT, Brietzke SE. Outpatient tonsillec-tomy in children: a systematic review. Oto-rhinolaryngol Head Neck Surg.2006;135(1):1–7

12. Morse RB, Hall M, Fieldston ES, et al.Hospital-level compliance with asthma carequality measures at children’s hospitalsand subsequent asthma-related outcomes.JAMA. 2011;306(13):1454–1460

13. Kalra M, Buncher R, Amin RS. Asthma asa risk factor for respiratory complicationsafter adenotonsillectomy in children withobstructive breathing during sleep. Ann Al-lergy Asthma Immunol. 2005;94(5):549–552

14. Kieran S, Gorman C, Kirby A, et al. Risk fac-tors for desaturation after tonsillectomy:analysis of 4092 consecutive pediatric cases.Laryngoscope. 2013;123(10):2554–2559

15. Averill RF, Goldfield N, Hughes JS, et al. AllPatient Refined Diagnosis Related Groups(APR-DRGs). Version 25.0. Wallingford, CT: 3MHealth Information Systems; 2007

16. Jha AK, Orav EJ, Dobson A, Book RA, EpsteinAM. Measuring efficiency: the association ofhospital costs and quality of care. HealthAff (Millwood). 2009;28(3):897–906

17. D’Agostino RB, Lee ML, Belanger AJ, CupplesLA, Anderson K, Kannel WB. Relation ofpooled logistic regression to time dependentCox regression analysis: the FraminghamHeart Study. Stat Med. 1990;9(12):1501–1515

18. Singer JD, Willett JB. Applied LongitudinalData Analysis. New York, NY: Oxford Uni-versity Press; 2003

19. Neuhaus JM, Kalbfleisch JD. Between- andwithin-cluster covariate effects in the anal-ysis of clustered data. Biometrics. 1998;54(2):638–645

20. Berlin JA, Kimmel SE, Ten Have TR, SammelMD. An empirical comparison of severalclustered data approaches under con-founding due to cluster effects in theanalysis of complications of coronary an-gioplasty. Biometrics. 1999;55(2):470–476

21. Korn EL, Graubard BI. Analysis of Health Sur-veys. New York, NY: John Wiley & Sons; 1999

22. Normand SLT, Glickman ME, Gatsonis CA.Statistical methods for profiling providersof medical care: issues and applications.J Am Stat Assoc. 1997;92(439):803–814

23. Normand SLT, Shahian DM. Statistical andclinical aspects of hospital outcomes pro-filing. Stat Sci. 2007;22(2):206–226

24. Shargorodsky J, Hartnick CJ, Lee GS. Dexa-methasone and postoperative bleeding aftertonsillectomy and adenotonsillectomy in chil-dren: a meta-analysis of prospective studies.Laryngoscope. 2012;122(5):1158–1164

25. Lowe D, van der Meulen J; National ProspectiveTonsillectomy Audit. Tonsillectomy technique asa risk factor for postoperative haemorrhage.Lancet. 2004;364(9435):697–702

26. Sarny S, Ossimitz G, Habermann W, Stamm-berger H. Hemorrhage following tonsil sur-gery: a multicenter prospective study.Laryngoscope. 2011;121(12):2553–2560

27. Czarnetzki C, Elia N, Lysakowski C, et al.Dexamethasone and risk of nausea andvomiting and postoperative bleeding aftertonsillectomy in children: a randomizedtrial. JAMA. 2008;300(22):2621–2630

28. Gallagher TQ, Hill C, Ojha S, et al. Peri-operative dexamethasone administrationand risk of bleeding following tonsillectomyin children: a randomized controlled trial.JAMA. 2012;308(12):1221–1226

29. Berry JG, Toomey SL, Zaslavsky AM, et al.Pediatric readmission prevalence and var-iability across hospitals [published cor-rection appears in JAMA. 2013;309(10):986].JAMA. 2013;309(4):372–380

30. Goyal SS, Shah R, Roberson DW, SchwartzML. Variation in post-adenotonsillectomyadmission practices in 24 pediatric hospi-tals. Laryngoscope. 2013;123(10):2560–2566

31. Wennberg J, Gittelsohn A. Small area var-iations in health care delivery. Science.1973;182(4117):1102–1108

32. Epstein AM. Geographic variation in Medicarespending. N Engl J Med. 2010;363(1):85–86

33. Conway PH, Clancy C. Transformation ofhealth care at the front line. JAMA. 2009;301(7):763–765

(Continued from first page)

Dr Mahant conceptualized and designed the study, contributed to data analysis and interpretation of data, and drafted and revised the article; Dr Keren contributedto the concept and design of the study, interpretation of the data, and drafting and critical revision of the manuscript; Dr Localio contributed to the design of thestudy, conducted the data analysis, and contributed to drafting and critical revision of the manuscript; Mr Luan and Mr Song contributed to the design of the study,acquisition of data, data analysis, interpretation of the data, and critical revision of the manuscript; Drs Shah, Tieder, Wilson, and Elden contributed to the design ofthe study, interpretation of the data, and critical revision of the manuscript; Dr Srivastava conceptualized and designed the study, contributed to data analysis andinterpretation of data, and critically revised the article; and all authors gave final approval of the manuscript as submitted.

www.pediatrics.org/cgi/doi/10.1542/peds.2013-1884

doi:10.1542/peds.2013-1884

Accepted for publication Nov 22, 2013

Address correspondence to Sanjay Mahant, MD, FRCPC, MSc, Department of Pediatrics, Hospital for Sick Children, 555 University Ave, Toronto, ON M5G 1X8, Canada.E-mail: sanjay.mahant@sickkids.ca

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright © 2014 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: Pediatric Research in Inpatient Settings has received support from the Children’s Hospital Association, Academic Pediatric Association,American Academy of Pediatrics, and the Society of Hospital Medicine. Drs Mahant, Keren, Tieder, Wilson, Shah, and Srivastava have support from the Children’sHospital Association for the submitted work, and Dr Keren also has support from the Pennsylvania Department of Public Health for the submitted work; Dr Kerenhas served as a member of the CVS Medical Health Advisory Board and received fees; Dr Shah has received grants from the Agency for Healthcare Research andQuality, the Patient-Centered Outcomes Research Institute, the Robert Wood Johnson Foundation, and the National Institute for Allergy and Infectious Diseases,outside the submitted work; and Dr Srivastava has received grants from the Agency for Healthcare Research and Quality, the Patient-Centered Outcomes ResearchInstitute, and the National Institutes of Health, outside the submitted work. The other authors have indicated they have no financial relationships relevant to thisarticle to disclose.

FUNDING: Supported by Health Research Formula grant 41000050891 from the Pennsylvania Department of Public Health Commonwealth Universal ResearchEnhancement Program and a grant from the Children’s Hospital Association.

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

288 MAHANT et al by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from

DOI: 10.1542/peds.2013-1884; originally published online January 20, 2014;Pediatrics

Shah, Joel S. Tieder, Karen M. Wilson, Lisa Elden and Rajendu SrivastavaSanjay Mahant, Ron Keren, Russell Localio, Xianqun Luan, Lihai Song, Samir S.

Children's HospitalsVariation in Quality of Tonsillectomy Perioperative Care and Revisit Rates in

  

ServicesUpdated Information &

/peds.2013-1884http://pediatrics.aappublications.org/content/early/2014/01/15including high resolution figures, can be found at:

Supplementary Material

5/peds.2013-1884.DCSupplemental.htmlhttp://pediatrics.aappublications.org/content/suppl/2014/01/1Supplementary material can be found at:

Citations

/peds.2013-1884#related-urlshttp://pediatrics.aappublications.org/content/early/2014/01/15This article has been cited by 8 HighWire-hosted articles:

Permissions & Licensing

tmlhttp://pediatrics.aappublications.org/site/misc/Permissions.xhtables) or in its entirety can be found online at: Information about reproducing this article in parts (figures,

Reprints http://pediatrics.aappublications.org/site/misc/reprints.xhtml

Information about ordering reprints can be found online:

rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.Grove Village, Illinois, 60007. Copyright © 2014 by the American Academy of Pediatrics. All and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elkpublication, it has been published continuously since 1948. PEDIATRICS is owned, published, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from

DOI: 10.1542/peds.2013-1884; originally published online January 20, 2014;Pediatrics

Shah, Joel S. Tieder, Karen M. Wilson, Lisa Elden and Rajendu SrivastavaSanjay Mahant, Ron Keren, Russell Localio, Xianqun Luan, Lihai Song, Samir S.

Children's HospitalsVariation in Quality of Tonsillectomy Perioperative Care and Revisit Rates in

  

  http://pediatrics.aappublications.org/content/early/2014/01/15/peds.2013-1884

located on the World Wide Web at: The online version of this article, along with updated information and services, is

 

of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2014 by the American Academy published, and trademarked by the American Academy of Pediatrics, 141 Northwest Pointpublication, it has been published continuously since 1948. PEDIATRICS is owned, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

by guest on September 4, 2015pediatrics.aappublications.orgDownloaded from