A health economic evaluation of screening and treatment in patients with adolescent idiopathic scoliosis

Summary of background data Adolescent idiopathic scoliosis can progress and affect the health related quality of life of the patients. Research shows that screening is effective in early detection, which allows for bracing and reduced surgical rates, and may save costs, but is still controversial from a health economic perspective. Study design Model based cost minimisation analysis using hospital’s costs, administrative data, and market prices to estimate costs in screening, bracing and surgical treatment. Uncertainty was characterised by deterministic and probabilistic sensitivity analyses. Time horizon was 6 years from first screening at 11 years of age. Objective To compare estimated costs in screening and non-screening scenarios (reduced treatment rates of 90%, 80%, 70% of screening, and non-screening Norway 2012). Methods Data was based on screening and treatment costs in primary health care and in hospital care settings. Participants were 4000, 12-year old children screened in Norway, 115190 children screened in Hong Kong and 112 children treated for scoliosis in Norway in 2012. We assumed equivalent outcome of health related quality of life, and compared only relative costs in screening and non-screening settings. Incremental cost was defined as positive when a non-screening scenario was more expensive relative to screening. Results Screening per child was € 8.4 (95% CrI 6.6 to10.6), € 10350 (8690 to 12180) per patient braced, and € 45880 (39040 to 55400) per child operated. Incremental cost per child in non-screening scenario of 90% treatment rate was € 13.3 (1 to 27), increasing from € 1.3 (−8 to 11) to € 27.6 (14 to 44) as surgical rates relative to bracing increased from 40% to 80%. For the 80% treatment rate non-screening scenario, incremental cost was € 5.5 (−6 to 18) when screening all, and € 11.3 (2 to 22) when screening girls only. For the non-screening Norwegian scenario, incremental cost per child was € -0.1(−14 to 16). Bracing and surgery were the main cost drivers and contributed most to uncertainty. Conclusions With the assumptions applied in the present study, screening is cost saving when performed in girls only, and when it leads to reduced treatment rates. Cost of surgery was dominating in non-screening whilst cost of bracing was dominating in screening. The economic gain of screening increases when it leads to higher rates of bracing and reduced surgical rates. Electronic supplementary material The online version of this article (doi:10.1186/s13013-014-0021-8) contains supplementary material, which is available to authorized users.


A.1: Introduction
In this supplementary data, we show the core equation on which the simulation model was based. We begun by presenting the equations for estimating the cost of the different interventions: screening, diagnosis of scoliosis, confirming scoliosis > 20º, brace treatment and surgery. Then we estimated the fraction of children receiving each category of interventions in the various scenarios. In the end we merged the estimated costs and the estimated fractions to estimate the cost pr child for each category of intervention and for the different scenarios.
The methodology used in the cost-minimizing analysis and discounting are presented in the main text of the manuscript and based on general literature on health economic evaluation like Drummond et al [1] or Hunink et al [2]. Methods for performing decision models probabilistic are based on Briggs et al. [3] The simulation model was built in Microsoft Excel. For the probabilistic sensitivity analysis we used the software @risk which is a part of the Decision Tools Suite software. The software @risk works is an extension to Excel.

A2: Estimation of cost of screening, brace treatment and surgery -all scenarios
Estimating the cost of the school screening per examination: C s = (u 1 • uc 1 ) + m + s u 1 = Number of minutes (units) used pr child pr examination (se row 1 in Table 1). uc 1 = Cost pr minute (unit cost) used pr examination (se row 1 in Table 1). m = Cost of materials and supplies per examination. s = Cost of scoliometer pr examination.
Estimating the cost of diagnosing one child for scoliosis: C con = t con + rad con t con = Cost of transportation to/from X-ray exam (se row 4 in Table 1). rad con = Cost of radiographs (se row 5 in Table 1).
Estimating the cost of brace per treatment: Where j = 9 to 16 in table 1. For example for j = 11 are u 11 • uc 11 equal to 3 hospital hotel days multiplied with €212 per day in hospital hotel, and likewise for the other cost components of brace treatment.
Estimating the cost per operation: Where i = 18 to 21 in Table 1, and j = 22 to 30 in Table 1.
For each child receiving an operation, 15% were assumed to be re-operated. So, for per child operated, the cost will be 100% + 15% of the costs estimated by the equation above. Here, j = 31 to 34, were 31 means diagnosing scoliosis, 32 means confirming scoliosis > 20º, 33 means brace treatment and 34 means surgery.

A3.2.1Non-screening scenario Norway
The fraction of children receiving surgery or brace treatment: To estimate the number of children 2012 in Norway receiving surgery (TrN 34 ) we took the number of children receiving surgery as the first treatment option and added the 10% of the children receiving bracing as first treatment option because these children receive in addition surgery later on.
The fraction of the non-screened children confirmed for scoliosis or scoliosis > 20º:

FnscN j = (((TrHK j /ChHK) • ChN) • Fr-conf) / ChN) = ((TrHK j /ChHK) • Fr-conf)
Fr-conf = The fraction of the screened children confirmed for scoliosis or scoliosis > 20º, who also would be confirmed for scoliosis or scoliosis > 20º if the same group when not screened. Fr-conf was assumed to be 0.15. The treatment rate for the Norwegian scenario was 73%. To change this according to the scenarios with different treatment rates, we adjusted the FnscN j to fit for the 80% scenario by multiplying FnscN j with 0.8, and for 70% scenario by multiplying with 0.7.
Here, j = 31 and 32, were 31 means confirmed for scoliosis and 32 means confirmed for scoliosis > 20º.
The fraction of the children in a year cohort (or the chance pr child) receiving surgery or brace treatment for the 80% non-screening scenarios: Fnsc80 j = The fraction of children receiving category j treatment for the 80% non-screening scenario.
Tr80 j = The number of children receiving category j of treatment in the 80% non-screening scenario.
Here, j = 33 and 34, were 33 means brace treatment and 34 means surgery.
In the 80% non-screening scenario, number receiving brace treatment and surgery, respectively: Here, j = 33 and 34, were 33 means brace treatment and 34 means surgery.
Note that, when we use the notion "treatment rate" in the main text, we do not "double-count" the cases of surgery. Instead we refer to the rate of children treated by brace or surgery, where those who are both receiving braced and surgery are included among the braced.