Cost and economic determinants of paediatric tonsillectomy

Keywords: healthcare reform, health economics, health funding and financing, health services research, hospital readmissions, paediatric, population health, tonsillectomy.

Tonsillectomy, with or without adenoidectomy, is one of the most common paediatric surgeries¹ and the most frequent reason for paediatric unplanned hospital readmission.² Indications for tonsillectomy can be categorised into infective and obstructive causes, with recurrent tonsillitis and obstructive sleep apnoea being the most common indications.³ Reasons for hospital readmission are predominantly postoperative complications such as haemorrhage, pain, decreased oral intake and respiratory complications.⁴

Due to its prevalence and significant variation in costs among hospitals,⁵ tonsillectomy imposes a large economic burden on healthcare systems.^5,6 Most cost descriptions of tonsillectomy have been based in the United States, where it is ranked the ninth most expensive reason for inpatient care at paediatric hospitals, costing more than US$92 million per year.⁶ There has been no previous description of costs in Australia.

Rising costs and an increased emphasis on patient-centred care have driven an interest in a value-based model of funding.⁷ Value-based healthcare incentivises reimbursement to providers based on patient health outcomes rather than the traditional volume-based, fee-for-service model. However, it requires insight into the costs of health services and associated consequences. Insight into the economic aspects of tonsillectomy is also important for health providers to understand the effect of the surgery on resource utilisation, for patients in health decision making, and for policymakers to allocate resources in the management of obstructive sleep apnoea and recurrent tonsillitis. To date, there have been no large-scale studies investigating the economic effect of paediatric tonsillectomy.

Australia has a universal healthcare system where the government funds public hospitals and 75% of treatment costs in private hospitals through an activity-based (casemix) system.⁸ This study aimed to describe the government-funded costs of paediatric tonsillectomy hospitalisations and subsequent readmissions in Victoria, Australia. Additionally, it sought to identify factors associated with higher cost surgery and model nationwide costs across Australia.

We undertook a population-based longitudinal study using data extracted from all hospitalisations for paediatric tonsillectomy, with or without adenoidectomy, in Victoria. Data were drawn from the Victorian Admitted Episodes Dataset (VAED).⁹ The VAED is managed by the Victorian Department of Health for administrative and billing purposes. It contains information on patient demographics, hospital sector and location, and clinical details that are collected from all Victorian hospital admissions.

We analysed a subset of the VAED containing all tonsillectomy, with or without adenoidectomy, performed between 1 June 2010 and 30 June 2015, at all public and private hospitals in Victoria, among patients aged between 0 and 19 years at the time of surgery. It also contained any resulting hospital readmissions within 30 days of the tonsillectomy. The number of surgeries is slightly higher than the number of patients, due to either more than one procedure being performed on the same child in the study period or in cases where one hospital contracts another to perform the surgery and hospitalisations are recorded by both the contracting hospital and contracted hospital. To best reflect the effect of surgery on governmental costs, we analysed cases at the surgical level rather than the patient level. To protect patient privacy, the supplied dataset listed age in 5-year age bands. To capture all paediatric patients, we included patients aged 0–19 years inclusive.

We defined ‘index tonsillectomy’ as the initial surgical admission, ‘hospital readmission’ as a return and admission to hospital after the index tonsillectomy, and ‘tonsillectomy-related hospitalisations’ to include both ‘index tonsillectomy’ and ‘hospital readmission’ if applicable.

VAED data were received with each public hospital having a unique code (not identifiable to the authors), while all private hospitals were coded as ‘zero’ and therefore could not be distinguished from one another. Hospitals were also classified as metropolitan or regional, predefined by the Department of Health.¹⁰ Surgical information included month and year of hospital admission and discharge, day case or overnight status, surgery type (tonsillectomy or tonsillectomy with adenoidectomy) and indication for surgery. Surgical indication and reason for readmission were recorded using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM; https://www.ihpa.gov.au/publications/icd-10-amachiacs-tenth-edition). We categorised ICD-10-AM codes into groups of ‘cardiovascular complication’ (e.g. arrhythmias), ‘dehydration’ (e.g. decreased oral intake), ‘pain’, ‘nausea’, ‘haemorrhage’, ‘airway compromise’, ‘upper respiratory complication’ (includes upper respiratory-related infections and inflammation, e.g. infection of the tonsil bed), ‘lower respiratory complication’ (includes lower respiratory-related infections and inflammation), ‘infection’ (infection other than respiratory related), ‘anaesthesia-related complication’, ‘surgical burn’, ‘unspecified surgical complication’ (e.g. malaise and fatigue, attention to surgical dressings and sutures) and ‘other’ (e.g. sleep disorder, unspecified).

The Australian Bureau of Statistics releases several metrics of socioeconomic status at various levels of Australian geographic areas. We used the 2011 Index of Relative Socio-economic Disadvantage at the level of ‘Statistical Local Area’¹¹ because this was the smallest residential area available in the surgical dataset for linkage. Areas were ranked and divided into quintiles, where quintile 1 represented the lowest socioeconomic group and quintile 5 the highest.

All estimates of hospital-related costs adopted the perspective of the Australian government. We used a ‘bottom-up’ costing approach which considers activity-based costing at the patient level, whereas the ‘top-down’ method considers an overall expenditure for a large healthcare system, then divides this expenditure using formulae to estimate costs of individual services. Every hospital admission in Victoria is assigned a weighted inlier equivalent separation (WIES) value to represent the admission’s relative cost.¹² The main drivers of WIES value are: (1) the primary reason for admission (diagnosis related group); (2) length of hospital stay; (3) any time in the intensive care unit; and (4) any time requiring mechanical ventilation. A price for each unit of WIES, defined by the government, is multiplied by the WIES value ascribed to an admission to determine costs to be reimbursed to the hospital for providing care during that admission. WIES unit prices vary by year, as well as by a hospital’s public/private and metropolitan/regional status. Supplementary Table S1 lists the value of the fixed prices. For reference, the WIES for a typical non-complicated overnight tonsillectomy for infection is 0.53 and for obstruction is 0.69. Other procedures could have been performed in the same admission with tonsillectomy, most commonly myringotomy. When adjusted for other cost-increasing factors, there was no difference in the WIES value of an admission for tonsillectomy and an admission for tonsillectomy with myringotomy. We excluded cases if the WIES value was zero, indicating that the surgery was incomplete or a coding problem had occurred.

Costs were extrapolated to the whole of Australia using weighted adjustment, based on published data on the total number of paediatric tonsillectomy surgeries across all states and territories in 2012/2013.¹³ To estimate costs for non-Victorian states and territories, the ratios of the number of admissions in each state or territory to the number of admissions in Victoria were multiplied by the annual cost of tonsillectomy in Victoria in 2014/2015.

The aim of our study was to evaluate direct costs to the government, so costs were not adjusted for inflation, unless stated otherwise. Any inflation-adjusted costs were calculated according to the total health price index¹⁴ with 2014/2015 as the chosen reference year. Costs were expressed in Australian Dollars (A$).

We conducted all statistical analyses using Stata (release 15.1, StataCorp LLC, College Station, TX, USA). To analyse cost growth during the 5-year period of the study and trend of cost over months, we used autoregressive models with a lag of one, with dependent variable WIES and independent variables ‘time unit (years or months)’ and ‘number of tonsillectomies’. Univariate and multivariable logistic regression models were used to identify factors associated with higher (upper quartile) WIES for tonsillectomy. WIES values were used as the outcome variable instead of actual cost because it is a fixed outcome that remains relatively similar across time periods and between public and private sectors. Cases were excluded from regression analysis if the patient had an overseas residential address because no Australian socioeconomic status value could be attached.

This study was approved by the Monash Health Human Research Ethics Committee (16066Q) as a Quality Assurance activity.

A total of 47 054 index tonsillectomies were performed in 45 571 patients aged 0–19 years in Victoria between 2010 and 2015. Our analysis included 46 969 index tonsillectomy admissions because 85 cases were excluded due to having a WIES value of zero. Of the sample, there were 2750 hospital readmissions within 30 days of the index tonsillectomy surgery. However, three readmissions were excluded as they had WIES values of zero. Therefore, there were 2747 hospital readmissions in our analysis, giving an incidence of 5.83%. Table 1 summarises the patient demographics, hospital characteristics and surgery types of both the index tonsillectomy and resulting hospital readmissions. The most common diagnosis related group (reason for admission) coded for infective indication were ‘D11Z–Tonsillectomy and Adenoidectomy’ (94.08%) and ‘D10Z–Nasal Procedures’ (5.15%), while for obstructive indications they were ‘E02C–Other Respiratory System or Procedures Without Complication and/or Comorbidity’ (64.14%) and ‘D11Z–Tonsillectomy and Adenoidectomy’ (30.35%).

Table 1.  Patient demographics, hospital characteristics and surgical characteristics of tonsillectomy and resulting hospital readmissions in Victoria, 2010–2015.

The median WIES value of index tonsillectomy was 0.55 (interquartile range (IQR) 0.53–0.69; range 0.18–25.09). The total governmental expenditure on index hospitalisation for tonsillectomy surgery over the 5-year period in Victoria was A$109 685 776, averaging to an annual cost of A$21 937 155 for 9394 admissions per year. The median cost per admission over the 5-year period was A$2224 (IQR 1826–2560).

There were fewer admissions involving day case surgery (N = 4753; 10.12%) compared to overnight stay or longer (N = 42 216; 89.88%). The median cost of a day case admission was A$2110 (IQR 1802–2423), while for an overnight stay or longer it was A$2224 (IQR 1842–2564).

Temporal analysis revealed a significantly increasing total annual cost of tonsillectomy over the 5-year period from A$17 812 647 in 2010/2011 to A$27 046 098 in 2014/2015 (P < 0.001). This rise was independent of the increasing number of surgeries (P < 0.001). Table 2 presents annual tonsillectomy costs during 2010–2015. Fig. 1 illustrates the trend of costs by month over the 5-year period.

Table 2.  Annual cost (A$) of tonsillectomy hospitalisations in Victoria, 2010–2015.

Fig. 1.  Cost of hospitalisation for tonsillectomy and tonsillectomy with adenoidectomy in Victoria, 2010–2015. Black line = total cost of tonsillectomy; dashed line = trend line of cost; dotted line = total number of tonsillectomy. Surgical admissions performed in June 2010 were omitted because data were not collected for the entire month.

Even when adjusted for inflation using the total health price index, costs from 2010/2011 to 2014/2015 significantly increased over the 5-year period (P < 0.001): A$19 236 120 (2010/2011), A$19 346 533 (2011/2012), A$21 541 201 (2012/2013), A$26 235 100 (2013/2014), A$34 844 648.34 (2014/2015).

The total cost of hospital readmissions following index tonsillectomy over the 5-year period in Victoria was A$7 138 579, with an average annual cost of A$1 427 716 for 549 readmissions per year. The median cost of a hospital readmission was A$2411 (IQR 1936–2732).

Table 3 presents costs according to the reason for hospital readmission. Haemorrhage was the reason for the greatest number of readmissions and was associated with the highest cost of all complications: N = 1868; overall cost A$5 113 493; median A$2413 (IQR 2209–2732). Although airway compromise was uncommon, it had the highest median cost: N = 5; overall cost A$78 903; median A$19 318 (IQR 2455–28 075).

Table 3.  Costs (A$) by reason for hospital readmission post-tonsillectomy in Victoria, 2010–2015.

The overall cost of tonsillectomy-related hospitalisations (index tonsillectomy surgeries and resulting hospital readmissions) over the 5-year period amounted to A$116 824 355, giving an annual average of A$23 364 871.

Extrapolating from Victorian data, the estimated total annual expenditure of all tonsillectomy-related hospitalisations in Australia in 2015 amounted to A$126 705 989. Index tonsillectomy accounted for 94.95% of these costs (A$119 039 881), with the remaining 6.05% being for hospital readmissions (A$7 666 108). Table 4 presents the estimation of total tonsillectomy hospitalisation cost in Australia and by state and territory.

Table 4.  Estimated annual cost (A$) of tonsillectomy in Australia.

The 75th percentile of WIES was 0.69. Table 5 presents results of the univariate and multivariable logistic regression analyses. Tonsillectomy with a WIES in the upper quartile was associated with younger age, male sex, lower socioeconomic status, surgery for reasons other than infection alone (i.e. for obstruction, both obstruction and infection, and other reasons), day case vs overnight stay, public hospitals and metropolitan hospitals.

Table 5.  Factors associated with upper quartile WIES for index tonsillectomy hospitalisations.

To the best of our knowledge, our study is the first to describe nationwide healthcare utilisation and cost of paediatric tonsillectomy. This high-volume procedure is associated with a 5.83% incidence of readmission within 30 days and carries considerable costs to the government.

Our estimated median cost for tonsillectomy surgery of A$2224 in an Australian setting is very similar to estimates for other countries. In New Zealand, which has a similar universal healthcare system to Australia, the estimated cost of each surgery is NZ$2750 (A$2548‬).¹⁵ In the United States, where health care is largely privately funded and serviced, Mahant et al.⁵ found significant variation in same-day paediatric tonsillectomy costs among different surgeons and hospitals, but reported a higher mean cost of US$2392 (IQR: US$1827–2793) (~A$3103 [IQR: A$2370–3624]). There are no data available describing the cost of other paediatric surgeries in Australia. For reference, the average cost of a medical admission for paediatric constipation as the principal diagnosis was A$1736 during 2002–2009 in Victoria, and the WIES value was 0.39.^16,17 While this presents an illustrative reference point, this type of medical admission cannot be meaningfully compared to elective tonsillectomy.

Between 2010 and 2015, the cost of index tonsillectomy hospitalisations significantly increased, even after adjusting for inflation. This increase was independent of the more gradual increase in the actual number of surgeries over the period. This rise is likely due to our finding that tonsillectomy for obstructive reasons had a significant association with higher (upper quartile) cost surgery, having a likelihood of 17.55 times compared to surgery for infective causes alone. As previously shown, there is a significant temporal rise in tonsillectomy performed for obstruction and a significant fall for infective causes in Victoria.¹⁸ This shift is likely due to increased recognition of symptoms of airway obstruction during sleep in the community, and both local¹⁹ and international²⁰ guidelines advocating for increased access to surgery for children with moderate and severe obstructive sleep apnoea. In addition, guidelines supporting watchful waiting for recurrent tonsillitis unless the Paradise criteria were met (≥7 episodes in the past year, or ≥5 episodes per year in the past 2 years or ≥3 episodes per year in the past 3 years) likely lead to a fall in surgery for that indication.^19,20 The higher cost of surgery for obstruction could be due to the more commonly coded ‘E02C’ rather than ‘D11Z’ for infective indication, which has a higher WIES value when calculated as an overnight stay rather than a day case. Increased attention to detail of coding in individual hospitals in recent years may have contributed to the observed increase in WIES.²¹ Research also shows that tonsillectomy for obstructive sleep apnoea is associated with greater cost due to increased risk of respiratory compromise, longer stay in hospital, and occasional need for mechanical ventilation and transfer to the intensive care unit – all factors which directly increase the WIES value and therefore costs.²² Therefore, while clinical need for surgery is the key factor in decision making, healthcare providers should be aware of the costs associated with tonsillectomy for obstructive sleep apnoea when making clinical decisions.

We also found that hospitals in the public sector and those located in metropolitan areas were associated with upper quartile cost for tonsillectomy. This is likely because the two tertiary hospitals in Victoria, which accept the most complex patients, are both public hospitals situated in metropolitan Melbourne. Treating complex patients results in a higher risk of peri-operative complications. Day case surgery was associated with lower cost, with 10% of tonsillectomy in Victoria being performed as a day case procedure over the study period, which is lower than that reported by some other centres.²³ Lower socioeconomic status was found to increase cost of tonsillectomy. This may be due to an increased surgical complication risk in this population.²⁴ We also found younger age to be associated with higher costing index tonsillectomy. Younger age is associated with poor postoperative oral intake and respiratory complications,²⁵ potentially contributing to increased length of stay and cost. Moreover, pain may be more difficult to control in younger children because codeine is avoided post-tonsillectomy for children with obstructive sleep apnoea due to the risk of respiratory depression.²⁶ Male sex was also a factor associated with higher cost surgery, but no explanation is available from the literature as to why this might be.

Interestingly, readmission rate was proportionally higher in the oldest age group (15–19 years). Several studies have shown an association between older aged children and increasing complication rates, especially for haemorrhage.^4,27 No potential explanations have been presented, but a contributor could be a lower understanding and compliance with postoperative diet, activity and pain management in older children since they have less direct adult supervision. Healthcare workers should therefore tailor postoperative education to the needs of this age group.

A major strength of our study is that it comprised a census for Victoria, capturing data from every paediatric patient in the state who underwent tonsillectomy and every resulting hospital readmission between 2010 and 2015. Furthermore, our ‘bottom-up’ costing approach provided a more accurate method of cost calculation compared to the commonly used ‘top-down’ approach.²⁸

Our study had several limitations. First, our analyses were limited to hospital administrative data and more granular clinical information was not available, such as the provided peri-operative care. Administrative data are also reliant on the accuracy of data entry. Second, our cost analysis for tonsillectomy was limited to hospitalisations relating to the index tonsillectomy and hospital readmissions. We did not analyse ambulance costs, emergency department expenditure, cost of medications, related costs prior to the surgery (e.g. specialist outpatient consultations) and patient out-of-pocket costs (particularly for private patients). Third, our estimation of the national cost and costs of other states and territories involved an extrapolation method of the Victorian cost based on weighted samples.

It is not our intention to advocate for reduced rates of tonsillectomy in order to reduce costs, given established evidence for benefit of the procedure. Tonsillectomy for recurrent tonsillitis decreases the number and severity of subsequent episodes,²⁹ while obstructive sleep apnoea surgery has been shown to improve sleep quality,^30,31 behaviour,³⁰ general quality of life,³² and subsequent healthcare costs.³³ Overall, there are few studies investigating the cost-effectiveness of tonsillectomy,^34,35 and none for obstructive sleep apnoea in the general paediatric population. Future studies should evaluate benefits and cost savings alongside costs of the procedure, to provide a truer picture of this balance. The purpose of this paper is to present costs for context in the discussion of future cost-saving methods in areas of associated high cost. For example, high cost was associated with surgery for obstructive sleep apnoea. Discussions and further research into reducing costs in Australia could follow international trends, such as moving towards day case surgery,²³ and measures to reduce pain levels and risk of haemorrhage, such as potential variations to surgical technique.³⁶ Further research could assess the use of tranexamic acid in reducing return to theatre after secondary haemorrhage³⁷ and low-cost strategies aimed at reducing readmission, such as telephone support after hospital discharge.³⁸

The costs of tonsillectomy are rising, even after adjusting for inflation. Higher cost surgery was associated with younger age, male sex, lower socioeconomic status, surgery for reasons other than infection alone, overnight vs day case surgery, public hospitals and metropolitan hospitals. These findings will inform healthcare reform that aims to contain costs while optimising patient outcomes.

The data that support the findings of this study are available from the Victorian Agency for Health Information but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. However, data are available upon request from the Victorian Agency for Health Information.

The authors declare that they have no conflicts of interest.

This study was supported by Equity Trustees and the Victorian Government’s Research Infrastructure Support Program. AHLT was supported by the Australian Government’s Research Training Program Stipend and Monash University’s Graduate Excellence Scholarship.

Supplementary material is available online.