Handling Missing Data in Within-Trial Cost-Effectiveness Analysis: A Review with Future Recommendations

Gabrio, Andrea; Mason, Alexina J.; Baio, Gianluca

doi:10.1007/s41669-017-0015-6

“…l Missing data in baseline resource use, quality of life and other covariates were handled with multiple imputation using chained equations, using guidelines from Gabrio et al 68 In line with primary outcomes analysis, the base case was a complete-case analysis.…”

Section: Health Economic Analysesmentioning

confidence: 99%

A contingency management intervention to reduce cannabis use and time to relapse in early psychosis: the CIRCLE RCT

Rains

¹

,

Marwaha

²

,

Strang

³

et al. 2019

Health Technol Assess

View full text Add to dashboard Cite

Background Cannabis is the most prevalent illicit substance among people with psychosis, and its use is associated with poorer clinical and social outcomes. However, so far, there has been limited evidence that any treatment is effective for reducing use. Contingency management (CM) is an incentive-based intervention for substance misuse that has a substantial evidence base across a range of substances and cohorts. However, to date there have been no randomised controlled trials (RCTs) of CM as a treatment for cannabis use specifically in psychosis. Objective To conduct a RCT investigating the clinical effectiveness and cost-effectiveness of CM in reducing cannabis use among Early Intervention in Psychosis (EIP) service users. Design The CIRCLE (Contingency Intervention for Reduction of Cannabis in Early Psychosis) trial was a rater-blinded, multicentre RCT with two arms. Participants were randomised 1 : 1 to either an CM arm, in which participants received CM for cannabis use alongside an optimised treatment-as-usual programme including structured psychoeducation, or a control arm in which participants received the treatment as usual only. Setting EIP services across the Midlands and the south-east of England. Participants The main eligibility criteria were EIP service users with a history of psychosis, aged 18–36 years, and having used cannabis at least once per week during 12 of the previous 24 weeks. Intervention The CM intervention offered financial incentives (i.e. shopping vouchers) for cannabis abstinence over 12 once-weekly sessions, confirmed using urinalysis. The maximum value in vouchers that participants could receive was £240. Main outcome measures The main outcome was time to relapse, operationalised as admission to an acute mental health service or hospital. The primary outcome was assessed at 18 months post inclusion using electronic patient records. Secondary outcomes assessed the clinical effectiveness and cost-effectiveness of the intervention, for which data were collected at 3 and 18 months. Results A total of 278 participants were randomised to the CM arm and 273 were randomised to the control arm. In total, 530 (96%) participants were followed up for the primary outcome. There was no significant difference in time to admission between trial arms by 18 months following consent (hazard ratio 1.03, 95% confidence interval 0.76 to 1.40). There were no statistically significant differences in most secondary outcomes, including cannabis use, at either follow-up assessment. There were 58 serious adverse events, comprising 52 inpatient episodes, five deaths and one arrest. Limitations Participant retention was low at 18 months, limiting the assessment of secondary outcomes. A different CM intervention design or reward level may have been effective. Conclusions The CM intervention did not appear to be effective in reducing cannabis use and acute relapse among people with early psychosis and problematic cannabis use. Future work Cannabis use is still a significant clinical concern in this population. A pressing need remains to identify suitable treatments. A wider perspective on the social circumstances of young people with psychosis may be needed for a successful intervention to be found. Trial registration Current Controlled Trials ISRCTN33576045. Funding This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 23, No. 45. See the NIHR Journals Library website for further project information.

show abstract

“…Both cost‐utility and cost‐effectiveness analyses were repeated using a societal perspective. To test the robustness of the assumption that data were ‘missing at random’ as required for multiple imputation and recommended for the reference assumption, cost‐utility was also estimated for complete cases only (no missing cost (MBS and PBS) or outcome data (SF6D)). One‐way sensitivity analyses were performed using costs associated with emergency department presentations and hospital admissions, which comprise the largest proportion of healthcare costs, with modification of this component (+10% and −10%) to assess the impact on cost‐utility estimates.…”

Section: Methodsmentioning

confidence: 99%

Home‐based pulmonary rehabilitation for COPD using minimal resources: An economic analysis

Burge

¹

,

Holland

²

,

McDonald

³

et al. 2019

View full text Add to dashboard Cite

Background and objective This study aimed to compare the cost‐effectiveness and cost‐utility of home and centre‐based pulmonary rehabilitation for adults with stable chronic obstructive pulmonary disease (COPD). Methods Prospective economic analyses were undertaken from a health system perspective alongside a randomized controlled equivalence trial in which participants referred to pulmonary rehabilitation undertook a standard 8‐week outpatient centre‐based or a new home‐based programme. Participants underwent clinical assessment prior to programme commencement, immediately following completion and 12 months following programme completion. They provided data for utility (quality‐adjusted life years (QALY) determined using SF6D (utility scores for health states) calculated from 36‐Item Short Form Health Survey version 2) and effectiveness (change in distance walked on 6‐min walk test (Δ6MWD) following pulmonary rehabilitation ). Individual‐level cost data for the 12 months following programme completion was sourced from healthcare administration and government databases. Results Between‐group mean difference point estimates for cost (−$4497 (95% CI: −$12 250 to $3257), utility (0.025 (−0.038 to 0.086) QALY) and effectiveness (14 m (−11 to 39) Δ6MWD) favoured the home‐based group. Cost‐utility analyses demonstrated 63% of estimates falling in the dominant southeast quadrant and the probability that the new home‐based model was cost‐effective at a $0 threshold for willingness to pay was 78%. Results were robust to a range of sensitivity analyses. Programme completion was associated with significantly lower healthcare costs in the following 12 months. Conclusion Home‐based pulmonary rehabilitation provides a cost‐effective alternative model for people with COPD who cannot access traditional centre‐based programmes.

show abstract

“…The appropriate method for dealing with missing cost data was dependent on the number of missing data and the likely mechanism of missingness. 50 Costs relating to hospitalisations were primarily sourced from PLICS data. If PLICS data were not available or missing, the use of hospital services was based on entries in CRFs, or otherwise from participants' resource use questionnaires.…”

Section: Analytic Approachmentioning

confidence: 99%

Silver-impregnated, antibiotic-impregnated or non-impregnated ventriculoperitoneal shunts to prevent shunt infection: the BASICS three-arm RCT

Mallucci

¹

,

Jenkinson

²

,

Conroy

³

et al. 2020

Health Technol Assess

View full text Add to dashboard Cite

Background Insertion of a ventriculoperitoneal shunt to treat hydrocephalus is one of the most common neurosurgical procedures worldwide. Shunt infection affects up to 15% of patients, resulting in long hospital stays, multiple surgeries and reduced cognition and quality of life. Objectives The aim of this trial was to determine whether or not antibiotic-impregnated ventriculoperitoneal shunts (hereafter referred to as antibiotic shunts) (e.g. impregnated with rifampicin and clindamycin) or silver-impregnated ventriculoperitoneal shunts (hereafter referred to as silver shunts) reduce infection compared with standard ventriculoperitoneal shunts (hereafter referred to as standard shunts). Design This was a three-arm, superiority, multicentre, parallel-group randomised controlled trial. Patients and a central primary outcome review panel, but not surgeons or operating staff, were blinded to the type of ventriculoperitoneal shunt inserted. Setting The trial was set in 21 neurosurgical wards across the UK and the Republic of Ireland. Participants Participants were patients with hydrocephalus of any aetiology who were undergoing insertion of their first ventriculoperitoneal shunt. Interventions Participants were allocated 1 : 1 : 1 by pressure-sealed envelope to receive a standard non-impregnated, silver-impregnated or antibiotic-impregnated ventriculoperitoneal shunt at the time of insertion. Ventriculoperitoneal shunts are medical devices, and were used in accordance with the manufacturer’s instructions for their intended purpose. Main outcome measures The primary outcome was time to ventriculoperitoneal shunt failure due to infection. Secondary outcomes were time to failure for any cause, reason for failure (infection, mechanical), types of ventriculoperitoneal shunt infection, rate of infection after first clean (non-infected) revision and health economics. Outcomes were analysed by intention to treat. Results Between 26 June 2013 and 9 October 2017, 1605 patients from neonate to 91 years of age were randomised to the trial: n = 36 to the standard shunt, n = 538 to the antibiotic shunt and n = 531 to the silver shunt. Patients who did not receive a ventriculoperitoneal shunt (n = 4) or who had an infection at the time of insertion (n = 7) were not assessed for the primary outcome. Infection occurred in 6.0% (n = 32/533) of those who received the standard shunt, in 2.2% (n = 12/535) of those who received the antibiotic shunt and in 5.9% (n = 31/526) of those who received the silver shunt. Compared with the standard shunt, antibiotic shunts were associated with a lower rate of infection (cause-specific hazard ratio 0.38, 97.5% confidence interval 0.18 to 0.80) and a decreased probability of infection (subdistribution hazard ratio 0.38, 97.5% confidence interval 0.18 to 0.80). Silver shunts were not associated with a lower rate of infection than standard shunts (cause-specific hazard ratio 0.99, 97.5% confidence interval 0.56 to 1.74). The ventriculoperitoneal shunt failure rate attributable to any cause was 25.0% overall and did not differ between arms. Antibiotic shunts save £135,753 per infection avoided. There were no serious adverse events. Limitations It was not possible to blind treating neurosurgeons to the ventriculoperitoneal shunt type. The return rate for patient-reported outcomes was low. Limitations to the economic evaluation included failure to obtain Hospital Episode Statistics data from NHS Digital, as per protocol. Reliance on patient-level information and costing systems data mitigated these limitations. Conclusions Antibiotic shunts have a reduced infection rate compared with standard shunts, whereas silver shunts do not. Antibiotic shunts are cost-saving. Future work A sample collection has been established that will enable the study of surrogate markers of ventriculoperitoneal shunt infection in cerebrospinal fluid or blood using molecular techniques. A post hoc analysis to study factors related to shunt failure will be performed as part of a future study. An impact analysis to assess change in practice is planned. Trial registration Current Controlled Trials ISRCTN49474281. Funding This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 24, No. 17. See the NIHR Journals Library website for further project information.

show abstract

Handling Missing Data in Within-Trial Cost-Effectiveness Analysis: A Review with Future Recommendations

Cited by 46 publications

References 198 publications

A contingency management intervention to reduce cannabis use and time to relapse in early psychosis: the CIRCLE RCT

A contingency management intervention to reduce cannabis use and time to relapse in early psychosis: the CIRCLE RCT

Home‐based pulmonary rehabilitation for COPD using minimal resources: An economic analysis

Silver-impregnated, antibiotic-impregnated or non-impregnated ventriculoperitoneal shunts to prevent shunt infection: the BASICS three-arm RCT

Contact Info

Product

Resources

About