Skip to main content
PMC home page
PLOS ONE logoLink to PLOS ONE
. 2020 Aug 10;15(8):e0236783. doi: 10.1371/journal.pone.0236783

Selective dorsal rhizotomy; evidence on cost-effectiveness from England

Mark Pennington 1,2,*, Jennifer Summers 2,3, Bola Coker 2,3, Saskia Eddy 2,3, Muralikrishnan R Kartha 1,2, Karen Edwards 4, Robert Freeman 4, John Goodden 5, Helen Powell 6, Christopher Verity 7, Janet L Peacock 2,3
Editor: Inmaculada Riquelme8
PMCID: PMC7416930  PMID: 32776949

Abstract

Objectives

Selective dorsal rhizotomy (SDR) has gained interest as an intervention to reduce spasticity and pain, and improve quality of life and mobility in children with cerebral palsy mainly affecting the legs (diplegia). We evaluated the cost-effectiveness of SDR in England.

Methods

Cost-effectiveness was quantified with respect to Gross Motor Function Measure (GMFM-66) and the pain dimension of the Cerebral Palsy Quality of Life questionnaire for Children (CPQOL-Child). Data on outcomes following SDR over two years were drawn from a national evaluation in England which included 137 children, mean age 6.6 years at surgery. The incremental impact of SDR on GMFM-66 was determined through comparison with data from a historic Canadian cohort not undergoing SDR. Another single centre provided data on hospital care over ten years for 15 children undergoing SDR at a mean age of 7.0 years, and a comparable cohort managed without SDR. The incremental impact of SDR on pain was determined using a before and after comparison using data from the national evaluation. Missing data were imputed using multiple imputation. Incremental costs of SDR were determined as the difference in costs over 5 years for the patients undergoing SDR and those managed without SDR. Uncertainty was quantified using bootstrapping and reported as the cost-effectiveness acceptability curve.

Results

In the base case, the incremental cost-effectiveness ratios (ICERs) for SDR are £1,382 and £903 with respect to a unit improvement in GMFM-66 and the pain dimension of CPQOL-Child, respectively. Inclusion of data to 10 years indicates SDR is cheaper than management without SDR. Incremental costs and ICERs for SDR rose in sensitivity analysis applying an alternative regression model to cost data.

Conclusions

Data on outcomes from a large observational study of SDR and long-term cost data on children who did and did not receive SDR indicates SDR is cost-effective.

Introduction

Cerebral palsy (CP) is a movement disorder usually arising from disturbances in brain development prior to or around childbirth. [1] It affects 1 in 500 people in Europe. [2] A common manifestation of CP is with spasticity mainly affecting the legs (previously known as spastic diplegia). [2] Children with spasticity from CP experience muscle spasms, shortening and weakening of muscles leading to mobility challenges, pain and joint degeneration. [3] Therapies used to mitigate muscle spasms and contractions include physiotherapy, botulinum toxin (Botox), [4] intrathecal baclofen and Selective Dorsal Rhizotomy (SDR). [5]

Interest in SDR has been increasing since the 1982 publication from Professor Peacock’s group. [6] Optimisation of patient selection criteria and surgical approach have led to renewed interest in SDR for treatment of lower limb spasticity in children with CP, with a particular focus on offering it to those functioning at Gross Motor Function Classification System (GMFCS) levels II and III. Children with spastic diplegic CP are assigned one of five GMFCS levels according to the impact of CP on their ability to move unaided. Children at level II are limited in their ability to run and jump or walk long distances. Children at level III typically require mobility devices to walk.

In SDR, the lumbosacral nerve rootlets are selectively severed to reduce overstimulation of muscles in the lower limbs. [7] Limited evidence from trials and observational studies links SDR with sustained improvements in physiology and anatomy. [8] However, the procedure is expensive and limited access to public funding until recently has led parents of children with CP in the UK to seek private treatment. [9]

In 2014, SDR was included in NHS England’s Commissioning through Evaluation (CtE) programme. [10] The programme provides limited funding for treatments for which the evidence base is insufficient with the aim of collecting evidence on effectiveness and cost-effectiveness to inform a decision on funding. In July 2018, following an interim report on the SDR CtE scheme by King’s Technology Evaluation Centre (KiTEC), NHS England agreed to commission the procedure for children aged 3 to 9 inclusive with CP with spasticity mainly affecting the legs, functioning at GMFCS levels II or III. [11] The clinical findings from the SDR CtE scheme have recently been published. [12]

As part of their evaluation of the SDR CtE scheme, KiTEC undertook a cost-effectiveness analysis of SDR. Data on outcomes following SDR were drawn from the CtE scheme. Long term cost data for children who did or didn’t receive SDR was taken from data supplied by the Robert Jones and Agnes Hunt Orthopaedic Hospital (RJAH), Oswestry, UK. This unit did not participate in the CtE scheme. We estimated counterfactual data for patients undergoing SDR based on published data on children not undergoing SDR. We report the cost-effectiveness analysis here.

Methods

We undertook a cost-effectiveness analysis of SDR from a health sector perspective. The patient population was children aged 3–9 years inclusive, diagnosed with CP with spasticity mainly affecting the legs, and classified as GMFCS level II or III. The comparator consisted of usual care which included physiotherapy, orthopaedic surgery, and drug treatments to alleviate spasticity. The intervention arm includes SDR in addition to treatments provided as part of usual care. We undertook subgroup analysis according to GMFCS level.

Five centres experienced in providing SDR were recruited to the CtE scheme and followed usual clinical practice with respect to SDR surgery and post-operative physiotherapy; 60–70% of the L1 to S1 nerve rootlets were cut in the vast majority of patients. Children received 3-weeks of daily physiotherapy rehabilitation at the treating centre, prior to discharge to their community provider team. This was remunerated as part of the SDR tariff. The majority of children received at least 2–3 hours of physiotherapy a week in the first three months following surgery and at least 1–2 hours in the second three months. Enhanced community physiotherapy continued to 24 months after surgery for the majority of children. [12]

The primary outcome was the Gross Motor Function Measure (GMFM-66). GMFM-66 was designed to assess gross motor function in children with CP. [13] The instrument has 66 items across five dimensions: lying and rolling; sitting; crawling and kneeling; standing; and walking, running and jumping. Each item is scored from 0 to 3. Scores within dimensions are converted to a percentage and a summary score is calculated as the mean of percentage scores on each dimension. We also evaluated a secondary outcome, the pain and impact of disability domain (hereon ‘CPQOL-pain’) of the Cerebral Palsy Quality of Life Questionnaire for Children (CPQOL-Child). [14] The CPQOL-Child has seven domains and is intended to capture the impact of CP on children’s wellbeing. A summary score is not available for the measure. The pain domain has 8 items; the score for each item is converted to a percentage prior to deriving the mean across items. These outcomes were selected prior to data analysis to span the main perceived benefits of SDR.

Data sources

Data on the effectiveness of SDR was taken from the SDR CtE scheme. The scheme enrolled 137 children from 2014 to 2016 at five centres across England: Alder Hey Children’s NHS Foundation Trust; University Hospitals Bristol NHS Foundation Trust; Great Ormond Street Hospital for Children NHS Foundation Trust; Leeds Teaching Hospitals NHS Trust and Nottingham University Hospitals NHS Trust. Children were eligible for SDR if they met the following criteria:

  • Aged between 3 and 9 years

  • GMFCS II or III

  • Dynamic spasticity in lower limbs affecting function and mobility

  • Absence of dystonia

  • MRI showing typical CP changes and no damage to key areas of brain controlling posture and coordination–periventricular leukomalacia or white matter injury of prematurity, without involvement of thalami or basal ganglia

  • No evidence of genetic or neurological progressive illness

  • Mild to moderate lower limb weakness with ability to maintain antigravity postures

  • No significant scoliosis or hip dislocation (Reimer’s index<40%) [15]

The CtE scheme received Health Research Authority approval in September 2014 (Integrated Research Approval System (IRAS) project ID: 162253). Consent to enrol children in the study and collect outcome data was sought from parents after provision of patient information sheets to parents and children. In addition to GMFM-66 and the CPQOL-Child, data collection included gait assessment, spasticity (Modified Ashworth Scale—MAS), motor control (Boyd and Graham test) and adverse events following surgery. [16,17] Evaluation of these outcomes are reported in more detail in the clinical paper. [12] Data on GMFM-66 and CPQOL-Child were collected prior to surgery and at 6, 12 and 24-month follow-up. Data were collected by the five centres and reported to a bespoke database created using Research Electronic Data Capture (REDCap).

The SDR CtE scheme did not include the collection of data on resource use and in particular did not provide data on a comparison group of children with CP who did not undergo SDR.

We were able to access inpatient data relating to treatment for CP for 26 children at RJAH. The data consisted of all assessments and treatments for CP at RJAH including physiotherapy, surgery and Botox injections. The data did not include primary care and oral drug treatments. Hospital remuneration for each procedure was available as the 2016/17 locally agreed tariff. The data were collected over the period from 1994 to 2017 and children were included if they had a continuous record of care at RJAH for at least four years. Children referred to RJAH for consideration for SDR underwent assessments including history, examination, 3D instrumented gait analysis and discussion at an MDT meeting. Eleven patients who met the inclusion criteria set by the SDR CtE protocol did not go on to have SDR, either due to a lack of funding (4 children), or due to stricter criteria set by the RJAH team including raised BMI, or a subjective impression that the degree of weakness or selective control would result in a poorer outcome. Consent for use of their data in research had previously been obtained for each child by RJAH.

Assessment of incremental effectiveness

The incremental effectiveness of SDR on GMFM-66 was determined as the difference in GMFM-66 observed at two year follow-up in the CtE cohort and that predicted from published growth curves. Gross motor function increases rapidly in the early years after birth as children develop. However, for children with CP the rate of increase is slower and for those with more severe disease, gross motor function may decline in adolescence. The change in GMFM-66 between baseline and two-year follow-up for children receiving SDR in the CtE scheme was assessed against expected changes in the absence of surgery for children with CP of the same age and GMFCS level using longitudinal data on GMFM-66 in a cohort of Canadian children with CP recruited over the period 1996–2001 (the CanChild cohort). These data had been analysed and developmental curves for children in each of the five GMFCS categories reported. [18] We used the published growth curve models to predict GMFM-66 at two-year follow-up in the absence of SDR, for children in the CtE scheme on the basis of age, GMFCS level and baseline GMFM-66. This approach differed slightly to the analysis of GMFM-66 in the clinical evaluation in which observed GMFM-66 scores for children at baseline and two years were compared with the distribution of scores observed in the CanChild cohort. [12] The prediction models used here allowed us to estimate GMFM-66 for each child in the absence of SDR and hence to construct a counterfactual outcome for each child undergoing SDR.

The incremental effectiveness of SDR on CPQOL-pain was estimated as the change in the domain score between baseline assessment prior to SDR and two-year follow-up.

Missing data were imputed using Multiple Imputation (MI). [19] We exploited data from intervening follow-ups at 6 and 12 months in addition to data on mobility and gait to impute missing outcome data.

Uncertainty in the estimates of effectiveness in the primary (GMFM-66) and secondary outcome (CPQOL-Child) was quantified by bootstrapping. We applied a two-stage bootstrap which allows for the clustering of data within the five centres partaking in the SDR CtE scheme. [20] Following MI, two-stage bootstrapping was applied to each imputed dataset and results were combined across the datasets using Rubin’s rules. [21] Further details on the methods, including prediction of GMFM-66 and imputation of missing data, are provided in the S1 File.

Assessment of incremental cost

The incremental cost of SDR was determined as the difference in cost for children who did and did not receive SDR in the RJAH cohort. We had pre-specified the primary analysis as the difference in costs for children who did and did not receive SDR (RJAH cohort) over ten years from assessment for SDR, with adjustment for age at assessment and GMFCS level. However, we had insufficient data beyond five years in the control group to support a robust analysis. Consequently, we present our main findings over five years and include data to ten years in a sensitivity analysis. In further sensitivity analysis, we adjusted for children declined SDR on clinical grounds. Data were considered complete for the years in which there was evidence that children were under the care of RJAH for at least 6 months. Missing costs by year were imputed using MI with Predictive Mean Matching. [22] Costs were discounted at 3.5% per annum as recommended by the National Institute of Health and Care Excellence (NICE). [23]

We investigated the distribution of the cost data with the intention of applying linear regression to adjust for case-mix provided the data were not highly skewed. The mean incremental cost of SDR and the standard error were determined by combining regression results across imputed datasets using Rubin’s rules. [21] More details of the analysis of cost data are given in the S1 File.

Assessment of cost-effectiveness

Cost-effectiveness is reported as the Incremental Cost-Effectiveness Ratio (ICER) and the Cost-Effectiveness Acceptability Curve (CEAC). The ICER is the ratio of incremental cost and incremental effectiveness and reports the cost per additional unit of health gained from the intervention. An ICER was calculated where one comparator was both more effective and more costly than the other. The CEAC plots the likelihood that an intervention is cost-effective across a range of maximum values the decision maker is prepared to pay for a unit improvement in health. It integrates the impact of uncertainty in cost and outcome data.

We simulated 1,000 estimates of the incremental cost of SDR by sampling values from a Normal distribution with mean and variance determined from regression analysis of the cost data from the RJAH cohort. We simulated 1,000 estimates of the incremental effectiveness of SDR on GMFM-66 from a Normal distribution with mean and variance determined from the bootstrap replicates of the change in GMFM-66 scores at two year follow-up in the CtE cohort; the change in GMFM-66 was calculated as the observed datum minus the predicted value from the CanChild growth curve. Each of the 1,000 incremental cost estimates was paired at random with one of the 1,000 incremental effectiveness estimates to generate 1,000 pairs. The paired data were used to generate a CEAC (details in S1 File). The same procedure was used to generate a CEAC for the secondary outcome measure, and for the subgroup analysis.

Results

Table 1 reports baseline characteristics of children in the CtE cohort, and children in the RJAH cohort according to whether they received SDR. Children receiving SDR in the RJAH cohort were similar to children in the CtE cohort. Children who did not receive SDR in the RJAH cohort were slightly older at assessment and more likely to be GMFCS III. Table 2 reports GMFM-66, CPQOL-pain, gait analysis and measures of mobility at baseline and follow-up in the CtE cohort. Function improved over time. Pain scores improved at six months after SDR and then plateaued at subsequent assessments. Less than 10% of data were missing with the exception of gait scores. Compared to predicted GMFM-66 at 24-month follow-up the mean (SD) incremental gain is 5.2 (0.53). The mean (SD) improvement in CPQOL-pain at 24 months is 7.9 (1.82). The subgroup analysis indicates larger improvements in both GMFM-66 and CPQOL-pain in patients in GMFCS level II compared with those in GMFCS level III (details in S1 File).

Table 1. Baseline characteristics of patients in the CtE and RJAH cohorts.

CtE cohort RJAH cohort
n = 137 SDR (n = 15) No SDR (n = 11)
Mean age at assessment in years (SD) 6.54 (1.92) 6.58 (1.11) 7.41 (1.13)
Mean age at SDR surgery in years (SD) 6.59 (1.92) 7.04 (1.16) -
GMFCS level II* (number) 38.0% (52) 33.3% (5) 18.2% (2)
Wheelchair/buggy use 67.2% 73.3% 90.9%
Open in a new tab

*The remaining proportion of the cohort was GMFCS level III.

Table 2. Gross motor function and quality of life at baseline and follow-up in the CtE cohort (n = 137).

Baseline 6 months 12 months 24 months
Mean GMFM-66 score (% missing) 59.0 (0%) 61.7(0%) 63.6 (1.4%) 66.0 (3.6%)
Mean CPQOL-pain (% missing) 36.4 (2.9%) 25.3 (5.1%) 28.8 (5.1%) 27.6 (8.0%)
Mean CPQOL function (% missing) 70.5 (2.9%) 77.6 (5.1%) 78.0 (5.1%) 78.5 (7.3%)
Open in a new tab

nr—not recorded.

Cost data were available for all children in the RJAH cohort receiving SDR up to 8 years after assessment and for 11 children (73%) at ten years. Nine children (82%) not receiving SDR had complete data at five years after assessment and two (18%) had complete data at ten years. For children with complete data, mean costs at five years were £33,282 in those receiving SDR and £31,030 in those not receiving SDR. Fig 1 displays the mean cost by year for children with non-missing data according to SDR status. Interventions received by children in the SDR group and the non-SDR group over the first five years are tabulated in the appendix. Costs in the SDR group are considerably higher in the first year, driven by the cost of the surgery and post-operative rehabilitation (£22,650). (Note: some children received SDR in the second year after the initial assessment.) At year 3 and beyond, mean costs are consistently lower each year in the SDR group.

Fig 1. Mean costs per child over time (years) by treatment status.

Fig 1

Open in a new tab

The distribution of cost data was not strongly skewed. Table 3 presents the results of linear regression analyses to estimate the incremental cost of SDR. The base case analysis suggests the initial cost of SDR and 3-weeks’ initial rehabilitation (£22,650) is partially offset at five years. Costs associated with SDR rise after adjustment for clinical criteria contraindicating SDR. Analysis of costs at ten years suggests the cost of SDR is entirely offset by reduction in the subsequent cost of supporting children. Confidence intervals are wide, and all include zero.

Table 3. Estimates of the impact of treatment (SDR surgery) on mean cost per child at 5 and 10 years.

Incremental cost of SDR 95% CI
Complete case, 5 years, raw £2,252 -£7,641 to £12,145
Complete case, 5 years, adjusted* £5,041 -£6,057 to £16,139
Imputed, 5 years, raw £4,849 -£5,250 to £14,949
Imputed, 5 years, adjusted* £7,160 -£3,998 to £18,318
Imputed, 5 years, sensitivity analysis# £12,035 -£1,982 to £26,052
Imputed, 10 years, raw -£9,132 -£26,648 to £8,385
Imputed, 10 years, adjusted* -£5,426 -£23,788 to £12,936
Imputed, 10 years, sensitivity analysis# £2,271 -£24,407 to £28,950
Open in a new tab

*Adjusted for age and GMFCS

#Adjusted for age, GMFCS and clinical criteria contraindicating SDR.

In the base case analysis ICERs for SDR with respect to GMFM-66 and CPQOL-pain are £1,382 and £903, respectively. This means that it costs around £1400 to generate each unit improvement in GMFM-66 and £900 to generate each unit improvement in CPQOL-pain. Figs 2 and 3 report the CEACs for the primary and secondary outcomes under the base case, and in sensitivity analysis controlling for clinical criteria contraindicating SDR. In the base case, the likelihood that SDR is cost-effective rises above 95% at a value of £3,150 for a unit gain in GMFM-66 and £2,350 for a unit reduction in CPQOL-pain. Subgroup analysis indicates a higher likelihood of cost-effectiveness for patients in GMFCS II and a lower likelihood for those in GMFCS III (CEACs are provided in the S1 File). In sensitivity analysis applying a time horizon of ten years, SDR dominates (it is cheaper and more effective than management without SDR). In sensitivity analysis controlling for clinical criteria contraindicating SDR, ICERs for SDR with respect to GMFM-66 and CPQOL-pain are £2,323 and £1,517, respectively; the likelihood that SDR is cost-effective reaches 95% at a value of £4,750 and £3,550 for unit gains in GMFM-66 and CPQOL-pain, respectively.

Fig 2. Cost-Effectiveness Acceptability Curve (CEAC) for primary outcome (GMFM-66).

Fig 2

Open in a new tab

Fig 3. Cost-Effectiveness Acceptability Curve (CEAC) for secondary outcome (CPQOL-pain).

Fig 3

Open in a new tab

Discussion

Our analysis indicates SDR in eligible children is likely to be cost-effective if decision makers value a unit gain in GMFM-66 at more than £1400 and a concomitant improvement in CPQoL-Pain at more than £900. Children in the CtE cohort showed a greater improvement in mean GMFM-66 at two-year follow-up than values predicted using data from the CanChild cohort and bootstrap resampling indicated the finding was unlikely to be due to chance. [18] Reported pain was also significantly reduced at two-year follow-up. Data from RJAH suggests that the cost of SDR is partially offset by a reduction in the costs of caring for children over a period of five years following surgery, and may be completely offset at ten years. Whilst the sample size from RJAH was small, uncertainty in cost-effectiveness due to sampling is low if decision makers value a unit gain in GMFM-66 at £3,000 or more. However, the use of a non-randomized comparison in the evaluation of both costs and outcomes increases the potential for bias, and introduces uncertainty into the findings that is not captured in the CEACs we report.

The judgment of whether SDR is cost-effective and the impact of sampling uncertainty in these findings depends on the value placed on the primary and secondary outcomes. In judging whether a health system should pay £1,000 to improve GMFM-66 by one point or reduce CPQOL-pain by one point, two considerations are pertinent. Firstly, both scales run from 0 to 100; one point is a very modest change. Second, our data and the available literature suggests that improvements in pain and function are maintained over time. [24] Hence the benefits of any change potentially accrue over each child’s lifetime.

This is the first study to examine the cost-effectiveness of SDR. Multiple studies have reported the cost-effectiveness of intrathecal baclofen and Botox injections. The evidence indicates intrathecal baclofen is cost-effective but finds little evidence of improved outcomes with Botox. [25–30]

One previous study compared the costs of patients undergoing SDR with a matched group receiving an intrathecal baclofen pump and concluded that SDR was cheaper. [31] However, it should be noted that intrathecal baclofen is typically offered to children with greater disability levels, functioning at GMFCS levels IV and V. Our evidence on outcomes from the CtE scheme which were reported in the clinical paper, [12] and formed the basis of the cost-effectiveness analysis is consistent with the available evidence. [32–34] A review of long term outcomes following SDR found improvements in physiology and anatomy, but noted the weakness of the evidence base. [8] A matched analysis of 13 patients undergoing SDR at a median age of 5 years with 8 patients who were eligible but did not undergo SDR did not find superior outcomes in patients receiving SDR after 10 years, but did note more frequent surgical treatment during follow-up in the non-SDR group. [35]

Evidence of the impact of SDR on costs is weaker. Cross-study comparisons of incidence of orthopaedic surgery following SDR are limited by differences in practice styles. Comparison of children undergoing SDR above and below the age of 6 years suggested a greater potential for SDR to reduce the incidence of orthopaedic surgery when was performed prior to adolescence. [36]

Our analysis has a number of strengths. Data on outcomes are taken from a large prospectively recruited sample of children which applied standardised inclusion criteria reflecting clinical evidence regarding which children could derive the greatest benefit from SDR. Missing data were limited due to CtE funding requirements to submit data and extensive efforts by the evaluators to support collection. Data on costs were taken from a centre with longstanding expertise in SDR which applied the same or tighter inclusion criteria to those of the SDR CtE scheme. Cost data included children undergoing SDR and similar children receiving supportive care from the same centre eliminating confounding from differences in practice styles or unit costs. Cost data were mostly complete to five years after assessment and data were available up to ten years after assessment.

Our analysis is limited by the lack of concurrent comparator data on outcomes and the small number of children with data on costs. Clinicians and parents of children with CP believed strongly in the benefits of SDR, rendering a randomised controlled trial unfeasible. Consequently, we undertook comparisons with historical data for GMFM-66 and a before and after comparison for CPQOL-pain collected as part of the CtE scheme. It is possible that expectations of the success of SDR biased both comparisons. Pain may have improved with conventional treatments such as Botulinum toxin. The CanChild cohort, which provided comparison data on GMFM-66 was recruited two decades ago. [18] Since that time, the management of CP has progressed and GMFM-66 trajectory in the absence of SDR may have improved. The availability of comparator data was a strength of the analysis of costs, but the RJAH cohort was small and follow-up limited beyond five years. The available data beyond five years suggests the trend of higher costs in the absence of SDR continues, indicating that our analysis may have overestimated the incremental cost of SDR. The comparator data included children declined SDR for clinical criteria. We had no a priori reason to believe these criteria would influence the cost of caring for these children but it is a potential confounder. Our cost data included only secondary health care. Any impact of SDR on primary health care, social care and productivity costs has not been captured. Whilst we were able to examine costs up to ten years after assessment for SDR our analysis of outcomes is restricted to a two-year follow-up period since this was the duration of CtE data collection. Finally, our analysis of cost-effectiveness ignores any potential correlation between cost and outcome data.

Conclusion

Evidence from England suggests SDR is cost-effective. The impact of SDR on costs may be offset by reductions in the cost of supportive care over the decade following surgery. Uncertainty in this finding is lower if commissioners value a percent improvement in GMFM-66 or a percent reduction in pain at £3,000 or more, but persists in the absence of a randomised comparison. Evidence from the SDR CtE scheme and independent cost data supports the recent decision by NHS England to commission SDR. Further research on the long-term costs and outcomes of SDR is needed to address some of the limitations in the current analysis.

Supporting information

S1 File

(DOCX)

Click here for additional data file. (220.1KB, docx)

Data Availability

The individual patient data which forms the basis for our analysis was collected from children partaking in the NHS England Commissioning through Evaluation programme (https://www.england.nhs.uk/commissioning/spec-services/npc-crg/comm-eval/). Ethical approval to analyse the data was provided by the National Research Ethics Service (NRES) East of England committee (REC reference 14/EE/1155). Consent for analysis of the data was sought on the basis that data would be stored securely, and access limited to the research team. As such, we are unable to upload the data to a public repository. The data is held at King’s Technology Evaluation Centre (KiTEC) whose director is Steve Keevil. The data is owned by the five hospitals contributing to the database: Leeds General Infirmary, Leeds (John Goodden); Great Ormond Street Hospital for Children, London (Kristian Aquilira); Bristol Royal Hospital for Children, Bristol (Richard Edwards); Alder Hey Children’s Hospital, Liverpool (Benedetta Pettorini); Nottingham University Hospitals, Nottingham (Michael Vloeberghs). The data is called the SDR database. The data on resource use for children with Cerebral Palsy is owned and resides at the Robert Jones and Agnes Hunt Orthopaedic Hospital, Oswestry. Access requests should be directed to Caroline Stewart, Manager Orthotic research & Locomotor Assessment Unit at the Robert Jones and Agnes Hunt Orthopaedic Hospital, Oswestry. The contact details for Steve Keevil are: Stephen Keevil Professor of medical physics School of Biomedical Engineering & Imaging Sciences 5th Floor, Becket House 1 Lambeth Palace Road London SE1 7EU 020 7188 3812 stephen.keevil@kcl.ac.uk The contact details for Caroline Stewart are: Caroline Stewart Senior Bioengineer/ORLAU Manager ORLAU RJAH Orthopaedic Hospital Oswestry Shropshire SY10 7AG 01691 404666 Caroline.Stewart9@nhs.net.

Funding Statement

HP is employed by the National Institute for Health and Care Excellence (NICE) and was contracted by NHS England to oversee the study. MP, JS, BC, SE, MK, and JP were employed by King's College London, London, UK, in partnership with NICE and NHS England, funded by NICE, and supported by the National Institute for Health Research Biomedical Research Centre based at Guy's and St Thomas' NHS Foundation Trust and King's College London. NHS England provided funding for the intervention and the evaluation and influenced the design of the evaluation of outcomes of surgery. NHS England also paid centres to provides data to the national evaluation. The funders had no additional role in data collection or any role in the analysis, decision to publish or preparation of the manuscript. https://www.england.nhs.uk/commissioning/spec-services/npc-crg/comm-eval/.

References

  • 1.Rosenbaum P, Paneth N, Leviton A, Goldstein M, Bax M, Damiano D, et al. A report: the definition and classification of cerebral palsy April 2006. Developmental Medicine & Child Neurology 2007;109(suppl):8–14. [PubMed] [Google Scholar]
  • 2.Johnson A. Prevalence and characteristics of children with cerebral palsy in Europe. Developmental medicine and child neurology. 2002;44(9):633–40. [PubMed] [Google Scholar]
  • 3.Koman LA, Smith BP, Shilt JS (2004) Cerebral palsy. The Lancet.;363:1619–3. [DOI] [PubMed] [Google Scholar]
  • 4.Jefferson RJ. botulinum toxin in the management of cerebral palsy. Developmental medicine and child neurology. 2004;46(7):491–9. 10.1017/s0012162204000817 [DOI] [PubMed] [Google Scholar]
  • 5.Albright AL. Intrathecal baclofen in cerebral palsy movement disorders. Journal of child neurology. 1996;11(1):S29–35. [DOI] [PubMed] [Google Scholar]
  • 6.Peacock WJ, Arens LJ. Selective posterior rhizotomy for the relief of spasticity in cerebral palsy. South African Medical Journal. 1982;62:119–124. [PubMed] [Google Scholar]
  • 7.Park TS, Johnston JM. Surgical techniques of selective dorsal rhizotomy for spastic cerebral palsy. Neurosurgical focus. 2006;21(2):1–6. [PubMed] [Google Scholar]
  • 8.Grunt S, Becher JG, Vermeulen RJ. Long‐term outcome and adverse effects of selective dorsal rhizotomy in children with cerebral palsy: a systematic review. Developmental Medicine & Child Neurology. 2011;53(6):490–8. [DOI] [PubMed] [Google Scholar]
  • 9.https://www.telegraph.co.uk/news/health/children/10178774/How-selective-dorsal-rhizotomy-SDR-allowed-a-cerebral-palsy-sufferer-the-chance-to-walk.html (Accessed 20/05/19)
  • 10.https://www.england.nhs.uk/commissioning/spec-services/npc-crg/comm-eval/ (Accessed 20/05/19)
  • 11.https://www.england.nhs.uk/wp-content/uploads/2019/03/Selective-dorsal-rhizotomy-for-the-treatment-of-spasticity-in-cerebral-palsy-children-aged-3-9-years.pdf (Accessed 01/08/19)
  • 12.Summers J, Coker B, Eddy S, Elstad M, Bunce C, Bourmpaki E, et al. Selective dorsal rhizotomy in ambulant children with cerebral palsy: an observational cohort study. The Lancet Child & Adolescent Health. 2019;3(7):455–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Harvey AR. The Gross Motor Function Measure (GMFM). Journal of physiotherapy. 2017;63(3):187 10.1016/j.jphys.2017.05.007 [DOI] [PubMed] [Google Scholar]
  • 14.Waters E, Davis E, Mackinnon A, Boyd R, Graham HK, Kai Lo S, et al. Psychometric properties of the quality of life questionnaire for children with CP. Developmental Medicine & Child Neurology. 2007;49(1):49–55. [DOI] [PubMed] [Google Scholar]
  • 15.Reimers J and Poulsen S, Adductor transfer versus tenotomy for stability of the hip in spastic cerebral palsy. Journal of Pediatric Orthopaedics, 1984;4(1):52–54. 10.1097/01241398-198401000-00012 [DOI] [PubMed] [Google Scholar]
  • 16.Scholtes VAB, Becher JG, Beelen A, Lankhorst GJ. Clinical assessment of spasticity in children with cerebral palsy a critical review of available instruments. Developmental Medicine & Child Neurology, 2006;48(1):64–73. [DOI] [PubMed] [Google Scholar]
  • 17.Boyd RN, Graham HK. Objective measurement of clinical findings in the use of botulinum toxin type A for the management of children with cerebral palsy. European Journal of Neurology, 1999;6:s23–s35 [Google Scholar]
  • 18.Hanna SE, Rosenbaum PL, Bartlett DJ, Palisano RJ, Walter SD, et al. Stability and decline in gross motor function among children and youth with cerebral palsy aged 2 to 21 years. Developmental Medicine & Child Neurology. 2009;51(4):295–302. [DOI] [PubMed] [Google Scholar]
  • 19.White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Statistics in medicine. 2011;30(4):377–99. 10.1002/sim.4067 [DOI] [PubMed] [Google Scholar]
  • 20.Ng ES, Grieve R, Carpenter JR. Two-stage nonparametric bootstrap sampling with shrinkage correction for clustered data. The Stata Journal. 2013;13(1):141–64. [Google Scholar]
  • 21.Rubin DB. Multiple imputation for nonresponse in surveys. John Wiley & Sons; 2004. [Google Scholar]
  • 22.Vink G, Frank LE, Pannekoek J, Van Buuren S. Predictive mean matching imputation of semicontinuous variables. Statistica Neerlandica. 2014;68(1):61–90. [Google Scholar]
  • 23.NICE 2013. Guide to the methods of technology appraisal 2013. Available at: https://www.nice.org.uk/process/pmg9/resources/guide-to-the-methods-of-technology-appraisal-2013-pdf-2007975843781 (Accessed 20/05/19). [Google Scholar]
  • 24.Aquilina K, Graham D, Wimalasundera N. Selective dorsal rhizotomy: an old treatment re-emerging. Archives of disease in childhood. 2015;100(8):798–802. 10.1136/archdischild-2014-306874 [DOI] [PubMed] [Google Scholar]
  • 25.Walczak J, Jóźwiak M, Jasińska A, Garbacka M, Obrzut G, Dardziński W. A cost-effectiveness analysis of programmable baclofen pump therapy in children with spastic cerebral palsy. Journal of Health Policy and Outcomes Research. 2012(1):81–8. [Google Scholar]
  • 26.Bensmail D, Ward AB, Wissel J, Motta F, Saltuari L, Lissens J, et al. Cost-effectiveness modeling of intrathecal baclofen therapy versus other interventions for disabling spasticity. Neurorehabilitation and neural repair. 2009;23(6):546–52. 10.1177/1545968308328724 [DOI] [PubMed] [Google Scholar]
  • 27.de Lissovoy G, Matza LS, Green H, Werner M, Edgar T. Cost-effectiveness of intrathecal baclofen therapy for the treatment of severe spasticity associated with cerebral palsy. Journal of child neurology. 2007;22(1):49–59. 10.1177/0883073807299976 [DOI] [PubMed] [Google Scholar]
  • 28.Sampson FC, Hayward A, Evans G, Morton R, Collett B. Functional benefits and cost/benefit analysis of continuous intrathecal baclofen infusion for the management of severe spasticity. Journal of neurosurgery. 2002;96(6):1052–7. 10.3171/jns.2002.96.6.1052 [DOI] [PubMed] [Google Scholar]
  • 29.Houltram J, Noble I, Boyd RN, Corry I, Flett P, Graham HK. Botulinum toxin type A in the management of equinus in children with cerebral palsy: an evidence‐based economic evaluation. European journal of neurology. 2001;8:194–202. 10.1046/j.1468-1331.2001.00052.x [DOI] [PubMed] [Google Scholar]
  • 30.Schasfoort F, Dallmeijer A, Pangalila R, Catsman C, Stam H, Becher J, et al. Value of botulinum toxin injections preceding a comprehensive rehabilitation period for children with spastic cerebral palsy: a cost-effectiveness study. Journal of rehabilitation medicine. 2018;50(1):22–9. 10.2340/16501977-2267 [DOI] [PubMed] [Google Scholar]
  • 31.Steinbok P, Daneshvar H, Evans D, Kestle JR. Cost analysis of continuous intrathecal baclofen versus selective functional posterior rhizotomy in the treatment of spastic quadriplegia associated with cerebral palsy. Pediatric Neurosurgery. 1995;22(5):255–65. 10.1159/000120911 [DOI] [PubMed] [Google Scholar]
  • 32.Kan P, Gooch J, Amini A, Ploeger D, Grams B, Oberg W, et al. Surgical treatment of spasticity in children: comparison of selective dorsal rhizotomy and intrathecal baclofen pump implantation. Child's Nervous System. 2008;24(2):239–43. 10.1007/s00381-007-0457-8 [DOI] [PubMed] [Google Scholar]
  • 33.Dudley RW, Parolin M, Gagnon B, Saluja R, Yap R, Montpetit K, et al. Long-term functional benefits of selective dorsal rhizotomy for spastic cerebral palsy. Journal of Neurosurgery: Pediatrics. 2013;12(2):142–50. 10.3171/2013.4.PEDS12539 [DOI] [PubMed] [Google Scholar]
  • 34.Bolster EA, Van Schie PE, Becher JG, Van Ouwerkerk WJ, Strijers RL, Vermeulen RJ. Long‐term effect of selective dorsal rhizotomy on gross motor function in ambulant children with spastic bilateral cerebral palsy, compared with reference centiles. Developmental Medicine & Child Neurology. 2013;55(7):610–6. [DOI] [PubMed] [Google Scholar]
  • 35.Munger ME, Aldahondo N, Krach LE, Novacheck TF, Schwartz MH. Long‐term outcomes after selective dorsal rhizotomy: a retrospective matched cohort study. Developmental Medicine & Child Neurology. 2017;59(11):1196–203. [DOI] [PubMed] [Google Scholar]
  • 36.O'Brien DF, Park TS, Puglisi JA, Collins DR, Leuthardt EC. Effect of selective dorsal rhizotomy on need for orthopedic surgery for spastic quadriplegic cerebral palsy: long-term outcome analysis in relation to age. Journal of Neurosurgery: Pediatrics. 2004;101(2):59–63. [DOI] [PubMed] [Google Scholar]

Decision Letter 0

Inmaculada Riquelme

10 Mar 2020

PONE-D-19-22434

Selective dorsal rhizotomy; evidence on cost-effectiveness from England

PLOS ONE

Dear author,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The topic of the manuscript is relevant; however, reviewers have expressed important methodological bias which make the manuscript unacceptable in its current form. I sugest the authors a deep reformulation of the manuscript according to the reviewers' suggestions in methods and consequently, in the discussion and conclusions. A new control cohort for comparison would be advisable and efforts to address analyses bias are encouraged.

We would appreciate receiving your revised manuscript by May 9th. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Inmaculada Riquelme

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1) Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2) Thank you for stating the following in the Competing Interests section:

[CV has received payments or expenses for acting as the clinical chair for NHS England

Commissioning through Evaluation for SDR. MP has received personal fees from

Merck. The other authors declare no competing interests.].

Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to  PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests).  If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

Please know it is PLOS ONE policy for corresponding authors to declare, on behalf of all authors, all potential competing interests for the purposes of transparency. PLOS defines a competing interest as anything that interferes with, or could reasonably be perceived as interfering with, the full and objective presentation, peer review, editorial decision-making, or publication of research or non-research articles submitted to one of the journals. Competing interests can be financial or non-financial, professional, or personal. Competing interests can arise in relationship to an organization or another person. Please follow this link to our website for more details on competing interests: http://journals.plos.org/plosone/s/competing-interests

3) Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

4)  We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: I Don't Know

Reviewer #3: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This report by Pennington et al. is a cost-effectiveness analysis of children undergoing selective dorsal rhizotomy. This current report is a subanalysis of economic and effectiveness outcomes as part of a larger clinical trial through the Commissioning through Evaluation program. The comparator group was a separate cohort of children from a previous (decades earlier) trial of Canadian children.

Overall the manuscript is quite well prepared, with clear descriptors of the methods, statistics, and analysis. Although not a strict criteria for publication, the topic of this report is quite important and impactful, as cost-effectiveness studies continue to gain importance in health system analyses. However, I do have some concerns that may benefit from consideration by the authors for clarification. Specifically:

1) I would defer to review by a health economics expert, but I question some of the terminology and methods used for this analysis. As I understand this report, the test group of children is compared to a cohort of children using previously published data (data not collected in the context of the CtE trial). In this scenario, calculation of ICERs is not appropriate, rather one should calculate the marginal cost-utility ratio. It is unclear to me whether the authors so such, including measuring the effect of treatment in each cohort in terms of a standardized metric (such as QALYs, DALYs, etc. rather than just the GMFM-66 scale or GMFCS levels) to allow for appropriate comparison across treatments.

2) The large difference in sample sizes between the cohorts really is a challenge, particularly for comparison of costs. I realize the authors addressed this in part by bootstrapping and assessment of confidence intervals, and not surprisingly the confidence intervals were quite wide. I am not sure how to best address this, and the authors do acknowledge this challenge in the description of limitations in the paper. However, I would certainly soften the conclusions of the report to recognize this limitation, as the implication is that the cost-effectiveness of SDR is clear and convincing. I am not sure this is fair to policymakers, who will interpret these statements without clear understanding of these limitations.

3) And finally, the use of such a remote historical control population (from decades earlier) is also a major limitations, both in terms of clinical effectiveness and cost analyses. The authors again acknowledge this limitation, although they should soften the conclusions accordingly. I would encourage that this data be considered pilot in nature, raising the question of whether a clinical trial (even non-randomized) would more conclusively address the question of cost-effectiveness of SDR.

Reviewer #2: This paper attempts to estimate the cost benefits of SDR. This was done by examining post-SDR outcomes (GMFM-66 and CPQOL pain score) compared to published data in the absence of surgery. The cost differential was then examined by looking at costs in an SDR group from RJAH and a group who did not have SDR in RJAH over a 10 year period from baseline assessment.

It's an interesting idea but I think that while some limitations are acknowledged, there are too many limitations and missing details to support the conclusions made.

The major flaw in this study appears to be in estimating the cost differential in those treated with SDR compared to those who were not. There area number of issues with this mainly the very small number in the non-SDR group (n=2) for whom data was available at 10 years. Looking at Figure 1 a spike in costs in the SDR group in clear at year one corresponding to the surgery. A similar spike is in the non-SDR group but no attempt is made to explain this? Was this orthopaedic surgery? Overall, there is no analysis or reporting of the standard treatment received by the control group (surgical or otherwise). Figure 1 also shows that costs are lower in the SDR group from year 3 onwards but as seen in Table 1 this 'No SDR' group were older and more involved (higher % GMFCS III and more likely to use wheelchair/buggy) so is the slightly higher costs in this very small number of children just related to that?

The benefits of SDR are established compared to published GMFM-66 prediction curves. However, this possibly over-estimates the benefits of SDR as really the benefits (or otherwise) of SDR should be when compared to standard/orthopaedic surgery which over a ten year period from baseline (at age 6-7 years) most CP children would probably have had if they did not have SDR?

There is no control data for the CPQOL pain score so it might be that the pain score improved less that it would have in the absence of surgery?

The Discussion suggests that the available literature suggests that improvements in pain and function are maintained over time. However, Munger's 2017 paper suggest no real long-term benefit in any outcome measure when comparing SDR to standard intervention and this is not mention at all.

In terms out outcome measures used here, only the pain score from the CPQOL assessment was used. Why is this and why not use other domains? This is not explained or justified.

Table 2 lists changes in GMFM 66 and CPQOL-pain over 24 months in the CtE SDR group. However, CPQOL function is also listed but not referenced at all in the paper. If listing here why not use in results? Likewise, a gait score is listed at Baseline and at 24 months but never mentioned in the paper.

Reviewer #3: The authors report on the cost-effectiveness of SDR in England.

Abstracts:

The abstract lacks the description of the underlying data set:

Age of children, number of children

Introduction: clear

Methods: the “”control”group is somewhat small, and the description of the group in table 1 is in my opinion inappropriate. Therefore, comparison of costs seems troublesome

Results

Table 2, can the authors add the number of evaluated subjects, or were data complete??

Dicussion:

The authors use a model to predict the costs of SDR as shown in figure 1. However, the follow-up was short and this assumption should get more attention in the discussion than is currently provided by the authors

“Confidence intervals are wide, and all include zero.”

(page 13)

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Henry Rice

Reviewer #2: No

Reviewer #3: Yes: R. Jeroen Vermeulen, MD, PhD

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2020 Aug 10;15(8):e0236783. doi: 10.1371/journal.pone.0236783.r002

Author response to Decision Letter 0

26 Jun 2020

Reviewer #1: This report by Pennington et al. is a cost-effectiveness analysis of children undergoing selective dorsal rhizotomy. This current report is a sub-analysis of economic and effectiveness outcomes as part of a larger clinical trial through the Commissioning through Evaluation program. The comparator group was a separate cohort of children from a previous (decades earlier) trial of Canadian children.

Overall the manuscript is quite well prepared, with clear descriptors of the methods, statistics, and analysis. Although not a strict criteria for publication, the topic of this report is quite important and impactful, as cost-effectiveness studies continue to gain importance in health system analyses. However, I do have some concerns that may benefit from consideration by the authors for clarification. Specifically:

1) I would defer to review by a health economics expert, but I question some of the terminology and methods used for this analysis. As I understand this report, the test group of children is compared to a cohort of children using previously published data (data not collected in the context of the CtE trial). In this scenario, calculation of ICERs is not appropriate, rather one should calculate the marginal cost-utility ratio. It is unclear to me whether the authors so such, including measuring the effect of treatment in each cohort in terms of a standardized metric (such as QALYs, DALYs, etc. rather than just the GMFM-66 scale or GMFCS levels) to allow for appropriate comparison across treatments.

Our analysis compared gross motor function in children receiving SDR with published data from a cohort of children who did not receive SDR. The measure of gross motor function, the GMFM-66, is the most commonly used measure. Data on children receiving SDR were taken from a National study undertaken to inform clinical decision making on the provision of SDR by the NHS in England. As was fully acknowledged in our published clinical paper, this study was not designed to include the collection of data on a comparison group of children not in receipt of SDR. Indeed, such a study is very likely to have met with ethical and practical barriers to recruitment and follow-up since the clinical and public view was firmly set at a belief that SDR is effective and that SDR may have age-sensitive outcomes. For similar reasons, it is highly unlikely that a clinical trial in which children are randomised to SDR or “standard” care without SDR will be possible due to the invasiveness of the procedure, the potential for age-sensitive outcomes and the evidence that SDR reduces spasticity. As a consequence, we compared data on changes in gross motor function over time for children in the NHS England Commissioning through Evaluation (CtE) study with published data for children with cerebral palsy who had not undergone SDR. The published data dates from before 2000 but provides information on the change in GMFM-66 as children mature according to their level of disability as assessed by Gross Motor Function Classification System (GMFCS). Growth models of the change in GMFM-66 with age according to GMFCS have been published. It is the most definitive data published on the motor function in children with CP. These data informed the assessment of the impact of SDR on gross motor function in the CtE evaluation and the subsequent report published in 2019 in the Lancet Child and Adolescent Health.

We used the published growth models to estimate the change in GMFM-66 that we would expect to see in children who had not undergone SDR in order to derive an estimate the incremental gain associated with SDR over two years. We accept that the use of these models is inferior to a control group in a randomised trial and increases the risk of bias but we took every measure possible to minimise the bias by covariate adjustment in statistical modelling with sensitivity analyses, plus we have reported confidence intervals as for all key estimates in order to be transparent about precision. We highlight these issues in the discussion. Whilst the use of a historical control is inferior to comparison with a randomised control arm, it does not change the appropriate methodology for undertaking the economic evaluation. The incremental cost-effectiveness ratio (ICER) is in essence a measure of the marginal cost-effectiveness of an intervention. Where outcomes are quantified in QALYs or DALYs, that methodology would allow a quantification of the marginal cost-utility. We chose to undertake a cost-effectiveness analysis based on GMFM-66 primarily because we lacked data on quality of life for children not undergoing SDR, although we note that measures of quality of life and associated health state utility values required to calculate QALYs are poorly developed for children. We accept that a cost-utility analysis would have been preferable. However, cost-effectiveness analyses are frequently reported and used to aid decision making in the absence of a cost-utility analysis, and we believe that our analysis makes best use of the available data to inform decision making.

2) The large difference in sample sizes between the cohorts really is a challenge, particularly for comparison of costs. I realize the authors addressed this in part by bootstrapping and assessment of confidence intervals, and not surprisingly the confidence intervals were quite wide. I am not sure how to best address this, and the authors do acknowledge this challenge in the description of limitations in the paper. However, I would certainly soften the conclusions of the report to recognize this limitation, as the implication is that the cost-effectiveness of SDR is clear and convincing. I am not sure this is fair to policymakers, who will interpret these statements without clear understanding of these limitations.

We accept that the sample of patients available to estimate the long term impact of SDR on costs is small. We used the most appropriate techniques to accommodate the missing data and quantify the impact of the size of the sample on the uncertainty in the estimate of the impact of SDR on costs. Whilst the sample used to estimate costs had some strengths, notably the long follow-up, and comparison of data from the same centre for intervention and control, it is fair to say that the non-randomized sample introduces risk of bias. We have amended our discussion to emphasise this risk and to temper the strength of our inference. We now state:

Our analysis indicates SDR in eligible children is likely to be cost-effective if decision makers value a unit gain in GMFM-66 at more than £1400 and a concomitant improvement in CPQoL-Pain at more than £900. Children in the CtE cohort showed a greater improvement in mean GMFM-66 at two-year follow-up than values predicted using data from the CanChild cohort and bootstrap resampling indicated the finding was unlikely to be due to chance.(18) Reported pain was also significantly reduced at two-year follow-up. Data from RJAH suggests that the cost of SDR is partially offset by a reduction in the costs of caring for children over a period of five years following surgery, and may be completely offset at ten years. Whilst the sample size from RJAH was small, uncertainty in cost-effectiveness due to sampling is low if decision makers value a unit gain in GMFM-66 at £3,000 or more. However, the use of a non-randomized comparison in the evaluation of both costs and outcomes increases the potential for bias, and introduces uncertainty into the findings that is not captured in the cost-effectiveness acceptability curves we report.

We have also amended the conclusion to acknowledge the remaining uncertainty:

Evidence from England suggests SDR is cost-effective. The impact of SDR on costs may be offset by reductions in the cost of supportive care over the decade following surgery. Uncertainty in this finding is lower if commissioners value a percent improvement in GMFM-66 or a percent reduction in pain at £3,000 or more, and in the absence of a randomised comparison is the best evidence that we have. Evidence from the SDR CtE scheme and independent cost data supports the recent decision by NHS England to commission SDR. Further research on the long-term costs and outcomes of SDR is needed to address some of the limitations in the current analysis.

We have also chosen to emphasise our comparison of costs at five years after intervention. Data on resource use is rarely reported beyond five years in economic evaluations alongside clinical trials. We remain of the view that data beyond five years is a strength of our data source. However, we accept the considerable extent of missing data beyond five years for children not in receipt of SDR. For this reason we now emphasise the comparison of costs at five years and report the comparison of costs at ten years in a sensitivity analysis.

3) And finally, the use of such a remote historical control population (from decades earlier) is also a major limitations, both in terms of clinical effectiveness and cost analyses. The authors again acknowledge this limitation, although they should soften the conclusions accordingly. I would encourage that this data be considered pilot in nature, raising the question of whether a clinical trial (even non-randomized) would more conclusively address the question of cost-effectiveness of SDR.

We agree that a trial would be preferable to the CtE study to support decision making. However, as we have discussed above, there is a lack of equipoise on the clinical benefits of SDR in clinicians and parents of children with CP in the UK and worldwide and so neither of these parties would agree to participate in a randomised trial. As such, it is hard to see how such a trial will ever be undertaken. Indeed, the decision of NHS England to recommend the use of SDR following the CtE study would challenge any future attempt to collect data on children eligible for SDR but not receiving it. We accept that we cannot conclusively answer the question on the cost-effectiveness of SDR and we have highlighted this in the conclusion as described above.

However, our analysis provides the most thorough examination of the cost-effectiveness of SDR to date. It is of value to inform decision making. It is also of value in informing any future trial which might incorporate an economic evaluation.

Reviewer #2: This paper attempts to estimate the cost benefits of SDR. This was done by examining post-SDR outcomes (GMFM-66 and CPQOL pain score) compared to published data in the absence of surgery. The cost differential was then examined by looking at costs in an SDR group from RJAH and a group who did not have SDR in RJAH over a 10 year period from baseline assessment.

It's an interesting idea but I think that while some limitations are acknowledged, there are too many limitations and missing details to support the conclusions made.

The major flaw in this study appears to be in estimating the cost differential in those treated with SDR compared to those who were not. There are a number of issues with this mainly the very small number in the non-SDR group (n=2) for whom data was available at 10 years. Looking at Figure 1 a spike in costs in the SDR group in clear at year one corresponding to the surgery. A similar spike is in the non-SDR group but no attempt is made to explain this? Was this orthopaedic surgery? Overall, there is no analysis or reporting of the standard treatment received by the control group (surgical or otherwise). Figure 1 also shows that costs are lower in the SDR group from year 3 onwards but as seen in Table 1 this 'No SDR' group were older and more involved (higher % GMFCS III and more likely to use wheelchair/buggy) so is the slightly higher costs in this very small number of children just related to that?

We acknowledge the limitation in the availability of data on resource use in the control group. This was particularly acute at ten years in the control group. For this reason, we have now? chosen to present costs comparisons at 5 years as the base case in our revised manuscript. We believe the data beyond five years provides some modest further evidence of lower resource use in children who have undergone SDR compared to children who have not. However, we accept the extent of missing data limits the value of this additional information. We note that a five year follow-up is longer than that observed in many randomised trials.

We now provide clarification of treatments received by both intervention and control group children in the supplementary material in Table 1S.

We acknowledge that there were some differences in the characteristics of patients in the treatment and control groups for the cohort of children from RJAH. We controlled for differences in age and GMFCS category in regression analysis of the data. We accept that this cannot exclude the possibility that such differences may have biased the analysis but we have attempted to mitigate the impact.

The benefits of SDR are established compared to published GMFM-66 prediction curves. However, this possibly over-estimates the benefits of SDR as really the benefits (or otherwise) of SDR should be when compared to standard/orthopaedic surgery which over a ten year period from baseline (at age 6-7 years) most CP children would probably have had if they did not have SDR?

The prediction curves are based on recorded GMFM-66 scores for a cohort of children who were followed for up to ten years. These children were eligible for standard/orthopaedic surgery and would have received routine care as appropriate. They did not receive SDR. We accept that the use of this historic cohort brings some limitations and have discussed this in our revised paper. It is possible that routine care has improved since these data were collected and we acknowledge this point in the limitations section of the discussion,

‘The CanChild cohort, which provided comparison data on GMFM-66 was recruited two decades ago.(18) Since that time, the management of CP has progressed and GMFM-66 trajectory in the absence of SDR may have improved.’

There is no control data for the CPQOL pain score so it might be that the pain score improved less that it would have in the absence of surgery?

This was a major limitation of our secondary outcome measure. It is possible that the use of surgery or the use of Baclofen or Botox might have led to a reduction in pain in the control group. We have added to the limitations section of the discussion to highlight this issue. We now say,

‘Pain may have improved with conventional treatments such as Botulinum toxin.’

We had no a priori reason to believe that pain scores would improve in this population over time in the absence of surgery.

The Discussion suggests that the available literature suggests that improvements in pain and function are maintained over time. However, Munger's 2017 paper suggest no real long-term benefit in any outcome measure when comparing SDR to standard intervention and this is not mention at all.

We focussed our discussion of the literature to those papers which reported on economic evaluations of interventions for CP. The broader literature on outcomes is only briefly mentioned. The study by Munger does suggest that the benefits of SDR are limited. However, this is a small study and there are many others, some of which have more positive findings. We feel that the literature is appropriately summarised by the review we cite. Nevertheless, we now highlight this study with the following addition.

‘A matched analysis of 13 patients undergoing SDR at a median age of 5 years with 8 patients who were eligible but did not undergo SDR did not find superior outcomes in patients receiving SDR after 10 years, but did note more frequent surgical treatment during follow-up in the non-SDR group.(36)’

In terms out outcome measures used here, only the pain score from the CPQOL assessment was used. Why is this and why not use other domains? This is not explained or justified.

The pain score was selected as our secondary outcome measure prior to undertaking analysis of the data. We chose this particular domain because clinicians considered pain relief to be one of the primary goals of surgery, and such an impact would not be captured in our primary outcome measuring gross motor function. We now clarify this in the methods with the following addition:

‘These outcomes were selected prior to data analysis to span the main perceived benefits of SDR.’

Table 2 lists changes in GMFM 66 and CPQOL-pain over 24 months in the CtE SDR group. However, CPQOL function is also listed but not referenced at all in the paper. If listing here why not use in results? Likewise, a gait score is listed at Baseline and at 24 months but never mentioned in the paper.

We included this data to contextualise the changes we observed in our primary and secondary outcome measures and because we used data from those measures in the imputation models that were applied to accommodate missing data. We do not discuss these measures in the main paper because they are ancillary to the pre-planned analysis To aid clarity and focus on the cost-effectiveness analysis we have now removed this data from the table.

Reviewer #3: The authors report on the cost-effectiveness of SDR in England.

Abstracts:

The abstract lacks the description of the underlying data set:

Age of children, number of children

We now include this information for children in the national evaluation and children in the RJAH cohort providing data on costs.

Introduction: clear

Methods: the ”control” group is somewhat small, and the description of the group in table 1 is in my opinion inappropriate. Therefore, comparison of costs seems troublesome

We accept the control group was small and we highlight this point in the limitations. However, the data had strengths in that patients were observed for up to ten years and the data comes from a single centre reducing the risk of confounding from differences in clinical practice or unit costs. We used bootstrapping to quantify the impact of sampling uncertainty. The small sample size increases uncertainty in our estimates of incremental cost. We capture and report the uncertainty in both costs and outcomes in the cost-effectiveness acceptability curves. These do indeed indicate uncertainty in the likelihood that SDR is cost-effective. However, we believe they provide the best possible evidence for commissioners of services at this time..

Results

Table 2, can the authors add the number of evaluated subjects, or were data complete??

The table reports data for all 137 patients in the national evaluation. We report the percentage of missing observations in brackets beside each data point. Missing data was below 10% at all observation points for both the primary and secondary outcomes.

Discussion:

The authors use a model to predict the costs of SDR as shown in figure 1. However, the follow-up was short and this assumption should get more attention in the discussion than is currently provided by the authors

We were fortunate to have cost data up to 10 years on patients in the RJAH cohort. However, data were missing for a number of patients beyond 5 years and we have now amended the manuscript to report the main cost comparison on data to 5 years. This length of follow-up is longer than many clinical trials. Nevertheless, our data suggests that lower costs observed in the SDR group in the years after the initial SDR surgery compared to the comparator group persists for ten years. Hence we may have overestimated the incremental cost of SDR. We have amended the discussion to make this point explicit. We now write,

‘The availability of comparator data was a strength for the analysis of costs, but the RJAH cohort was small and follow-up limited beyond five years. The available data beyond five years suggests the trend of higher costs in the absence of SDR continues, indicating that our analysis may have overestimated the incremental cost of SDR.’

“Confidence intervals are wide, and all include zero.”

(page 13)

The purpose of our analysis was to estimate the cost-effectiveness of SDR which required estimation of the additional cost of treating children with SDR and the additional benefits of treatment. We did not set out to test whether SDR is more expensive than management of children without SDR. The confidence intervals indicate a possibility that SDR is no more expensive than management without SDR. We convey this likelihood in the cost-effectiveness acceptability curve along with the probability that SDR is cost-effective according to the value the decision maker places on improving clinical outcomes.

Attachment

Submitted filename: Response to reviewers v2.docx

Click here for additional data file. (32.2KB, docx)

Decision Letter 1

Inmaculada Riquelme

15 Jul 2020

Selective dorsal rhizotomy; evidence on cost-effectiveness from England

PONE-D-19-22434R1

Dear Dr. Pennington,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Inmaculada Riquelme

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #3: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

Reviewer #3: the authors did a nice job to investigate the cost-effectiveness of selective dorsal rhizotomy for England! I think adding England does add value because health systems differ much. I agre with the previous reviewers that the control group is extremely small which is apotential bias in their calculations.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Henry Rice

Reviewer #3: Yes: R. Jeroen Vermeulen

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 File

    (DOCX)

    Click here for additional data file. (220.1KB, docx)
    Attachment

    Submitted filename: Response to reviewers v2.docx

    Click here for additional data file. (32.2KB, docx)

    Data Availability Statement

    The individual patient data which forms the basis for our analysis was collected from children partaking in the NHS England Commissioning through Evaluation programme (https://www.england.nhs.uk/commissioning/spec-services/npc-crg/comm-eval/). Ethical approval to analyse the data was provided by the National Research Ethics Service (NRES) East of England committee (REC reference 14/EE/1155). Consent for analysis of the data was sought on the basis that data would be stored securely, and access limited to the research team. As such, we are unable to upload the data to a public repository. The data is held at King’s Technology Evaluation Centre (KiTEC) whose director is Steve Keevil. The data is owned by the five hospitals contributing to the database: Leeds General Infirmary, Leeds (John Goodden); Great Ormond Street Hospital for Children, London (Kristian Aquilira); Bristol Royal Hospital for Children, Bristol (Richard Edwards); Alder Hey Children’s Hospital, Liverpool (Benedetta Pettorini); Nottingham University Hospitals, Nottingham (Michael Vloeberghs). The data is called the SDR database. The data on resource use for children with Cerebral Palsy is owned and resides at the Robert Jones and Agnes Hunt Orthopaedic Hospital, Oswestry. Access requests should be directed to Caroline Stewart, Manager Orthotic research & Locomotor Assessment Unit at the Robert Jones and Agnes Hunt Orthopaedic Hospital, Oswestry. The contact details for Steve Keevil are: Stephen Keevil Professor of medical physics School of Biomedical Engineering & Imaging Sciences 5th Floor, Becket House 1 Lambeth Palace Road London SE1 7EU 020 7188 3812 stephen.keevil@kcl.ac.uk The contact details for Caroline Stewart are: Caroline Stewart Senior Bioengineer/ORLAU Manager ORLAU RJAH Orthopaedic Hospital Oswestry Shropshire SY10 7AG 01691 404666 Caroline.Stewart9@nhs.net.

    Articles from PLoS ONE are provided here courtesy of PLOS

    RESOURCES