Daniel Gallacher

Objectives: To undertake a systematic review evaluating the clinical and cost-effectiveness of technology-enhanced positive end-expiratory pressure (PEEP) optimization strategies in adults and children receiving invasive mechanical ventilation on an ICU. Data Sources: We searched key electronic databases (including MEDLINE and Embase) from inception to July 2024. Study Selection: We included randomized studies examining clinical or cost-effectiveness of technology-enhanced PEEP optimization strategies compared with standard care or an alternative PEEP optimization strategy in adults and children. The primary outcome was duration of mechanical ventilation and secondary outcomes were clinical effectiveness (e.g., mortality) and efficacy (e.g., PEEP). Data Extraction: Two reviewers independently assessed eligibility, extracted data, assessed risk of bias (Revised Cochrane tool) and performed Grading of Recommendations Assessment, Development and Evaluation evidence certainty assessments. Data Synthesis: Our database and trial register search retrieved 8845 results, of which 34 studies (2951 patients) were included. Eight studies were at low risk of bias. Across studies, 7 technologies were evaluated, most commonly esophageal balloon measurement of transpulmonary pressure (10 studies), electrical impedance tomography (7 studies), pressure-volume curve analysis (6 studies), and fully automated closed-loop ventilation (5 studies). Meta-analysis used random-effects models. Duration of mechanical ventilation was reported in only three studies (172 patients, two technologies) and there was no effect compared with standard care (mean difference –0.06 d; 95% CI, –0.20 to 0.09; very low-certainty evidence). For 28-day mortality (10 studies; 1,719 patients; six technologies), technology-enhanced PEEP optimization reduced 28-day mortality (risk ratio 0.69; 95% CI, 0.52–0.93; very low-certainty evidence). No significant differences were found for other clinical-effectiveness outcomes. We identified no evidence in children or on cost-effectiveness. Conclusions: Technology-enhanced PEEP optimization strategies did not reduce duration of mechanical ventilation, but these technologies may reduce mortality. Evidence certainty was low or very low, highlighting the urgent need for adequately powered randomized trials. Registration: PROSPERO (CRD42024555390).

Introduction: The use of a 36-mm femoral head in primary total hip replacement (THR) increases the jump distance and offers a wider impingement-free range of motion, theoretically reducing the risk of post-operative dislocation. However, concerns exist regarding its potentially greater impact on polyethylene wear and associated risks of liner fracture, aseptic loosening, and head-neck taper corrosion, compared to 32-mm femoral heads. This meta-analysis aims to compare the risk of revision for dislocation and all-causes when using 36-mm versus 32-mm metal or ceramic femoral heads with polyethylene liners for primary THR. Methods: Medline, Embase, Web of Science, and the Cochrane Library were searched for relevant studies, while annual reports of arthroplasty registries were searched for relevant data. Random effects meta-analysis was performed. Sensitivity analysis was conducted, limiting to THR performed for osteoarthritis and using cross-linked polyethylene liners, or statistically adjusting for this factor. The study was registered in the International Prospective Register of Systematic Reviews (PROSPERO ID CRD42024557895). Results: Four observational studies and two registry reports were identified. Median follow-up ranged from 2.1 to 4.7 years (inter-quartile range values spanned from 0.9 to 7.7 years). The main analyses demonstrated that 36-mm heads were associated with a marginal reduction in the risk of revision for dislocation (HR 0.85, 95%CI 0.72–1.01, p = 0.058; n = 241,136) without an increased risk of all-cause revision (HR 1.06, 95%CI 0.95–1.18, p = 0.287; n = 942,617). Similar results were observed in the sensitivity analyses. Discussion: At early to midterm follow-up, the use of 36-mm heads were not associated with an increased risk of revision for all-causes, compared to 32-mm heads. However, they may offer a protective reduction against revision for dislocation, as a marginal statistically significant reduction was observed, indicating a possible benefit. Further studies investigating these outcomes at longer-term follow up are needed to understand whether these revision risk profiles are maintained.

Background: Computer navigation and patient-specific instrumentation in unicompartmental knee replacement (UKR) improve the precision of implant positioning, but there is limited information regarding their impact on implant survival and patient-reported outcomes. We aimed to compare postoperative implant survival, Oxford Knee Score (OKS) values, health-related quality of life (measured using the EuroQol-5 Dimension 3-level version [EQ-5D-3L]), and intraoperative complications between UKRs performed using computer navigation or patient-specific instrumentation versus conventional instrumentation. Methods: Using National Joint Registry data, an observational study of patients who underwent primary UKR for osteoarthritis between 2003 and 2020 was performed. The primary analyses focused on all-cause revision, and the secondary analyses focused on differences in the OKS and EQ-5D-3L at 6 to 12 months postoperatively. To account for several covariates, weights based on propensity scores were generated. Cox proportional hazards models and generalized linear models were used to assess for differences in revision risk, and OKS and EQ-5D-3L change scores, respectively, between patient groups. Sensitivity analyses accounting for body mass index were performed. Effective sample sizes (ESSs) were computed, representing the statistical power comparable with that of an unweighted sample. Results: Compared with conventional instrumentation, the hazard ratio (HR) for all-cause revision was 1.126 (95% confidence interval [CI], 0.909 to 1.395; p = 0.277; ESS, 4,273) with computer navigation and 0.805 (95% CI, 0.442 to 1.467; p = 0.478; ESS, 1,199) with patient-specific instrumentation. No difference was found in the change in OKS between the groups (−1.287; 95% CI, −2.851 to 0.278; p = 0.107; ESS, 470), although improvement in the EQ-5D-3L scores was relatively lower for computer-navigated UKR compared with conventional instrumentation (−0.049, 95% CI, −0.093 to −0.005; p = 0.028; ESS, 455). However, sensitivity analyses demonstrated that computer navigation was associated with an increased risk of all-cause revision (HR, 1.446; 95% CI, 1.102 to 1.898; p = 0.008; ESS, 3,011) and relatively smaller improvements in the OKS (−2.845; 95% CI, −5.006 to −0.684; p = 0.010; ESS, 272) and EQ-5D-3L scores (−0.087; 95% CI, −0.145 to −0.030; p = 0.003; ESS, 286). There were no differences in intraoperative complications (p = 0.073). Conclusions: This study found no clinically meaningful differences in patient-reported outcomes following computer-navigated UKR. Although likely underpowered, the primary analyses showed no difference in implant survival. While a sensitivity analysis suggested that computer navigation could worsen implant survival, this analysis had a smaller sample size. These findings highlight potential signals that warrant further investigation. Level of Evidence: Therapeutic Level III . See Instructions for Authors for a complete description of levels of evidence.

Background: Lung cancer is one of the most common types of cancer and the leading cause of cancer death in the United Kingdom. Artificial intelligence-based software has been developed to reduce the number of missed or misdiagnosed lung nodules on computed tomography images. Objective: To assess the accuracy, clinical effectiveness and cost-effectiveness of using software with artificial intelligence-derived algorithms to assist in the detection and analysis of lung nodules in computed tomography scans of the chest compared with unassisted reading. Design: Systematic review and de novo cost-effectiveness analysis. Methods: Searches were undertaken from 2012 to January 2022. Company submissions were accepted until 31 August 2022. Study quality was assessed using the revised tool for the quality assessment of diagnostic accuracy studies (QUADAS-2), the extension to QUADAS-2 for assessing risk of bias in comparative accuracy studies (QUADAS-C) and the COnsensus-based Standards for the selection of health status Measurement INstruments (COSMIN) checklist. Outcomes were synthesised narratively. Two decision trees were used for cost-effectiveness: (1) a simple decision tree for the detection of actionable nodules and (2) a decision tree reflecting the full clinical pathways for people undergoing chest computed tomography scans. Models estimated incremental cost-effectiveness ratios, cost per correct detection of an actionable nodule, and cost per cancer detected and treated. We undertook scenario and sensitivity analyses. Results: Twenty-seven studies were included. All were rated as being at high risk of bias. Twenty-four of the included studies used retrospective data sets. Seventeen compared readers with and without artificial intelligence software. One reported prospective screening experiences before and after artificial intelligence software implementation. The remaining studies either evaluated stand-alone artificial intelligence or provided only non-comparative evidence. (1) Artificial intelligence assistance generally improved the detection of any nodules compared with unaided reading (three studies; average per-person sensitivity 0.43–0.68 for unaided and 0.79–0.99 for artificial intelligence-assisted reading), with similar or lower specificity (three studies; 0.77–1.00 for unaided and 0.81–0.97 for artificial intelligence-assisted reading). Nodule diameters were similar or significantly larger with semiautomatic measurements than with manual measurements. Intra-reader and inter-reader agreement in nodule size measurement and in risk classification generally improved with artificial intelligence assistance or were comparable to those with unaided reading. However, the effect on measurement accuracy is unclear. (2) Radiologist reading time generally decreased with artificial intelligence assistance in research settings. (3) Artificial intelligence assistance tended to increase allocated risk categories as defined by clinical guidelines. (4) No relevant clinical effectiveness and cost-effectiveness studies were identified. (5) The de novo cost-effectiveness analysis suggested that for symptomatic and incidental populations, artificial intelligence-assisted computed tomography image analysis dominated the unaided radiologist in cost per correct detection of an actionable nodule. However, when relevant costs and quality-adjusted life-years from the full clinical pathway were included, artificial intelligence-assisted computed tomography reading was dominated by the unaided reader. For screening, artificial intelligence-assisted computed tomography image analysis was cost-effective in the base case and all sensitivity and scenario analyses. Limitations: Due to the heterogeneity, sparseness, low quality and low applicability of the clinical effectiveness evidence and the major challenges in linking test accuracy evidence to clinical and economic outcomes, the findings presented here are highly uncertain and provide indicators/frameworks for future assessment. Conclusions: Artificial intelligence-assisted analysis of computed tomography scan images may reduce variability of and improve consistency in the measurement and clinical management of lung nodules. Artificial intelligence may increase nodule and cancer detection but may also increase the number of patients undergoing computed tomography surveillance unnecessarily. No direct comparative evidence was found, and nor was any direct evidence found on clinical outcomes and cost-effectiveness. Artificial intelligence-assisted image analysis may be cost-effective in screening for lung cancer but not for symptomatic populations. However, reliable estimates of cost-effectiveness cannot be obtained with current evidence. Study registration: This study is registered as PROSPERO CRD42021298449. Funding: This award was funded by the National Institute for Health and Care Research (NIHR) Evidence Synthesis programme (NIHR award ref: NIHR135325) and is published in full in Health Technology Assessment; Vol. 29, No. 14. See the NIHR Funding and Awards website for further award information.

Background: Computer navigation and patient-specific instrumentation have been in use over the past 2 decades for total knee replacement (TKR). However, their effects on implant survival and patient-reported outcomes remain under debate. We aimed to investigate their influence on implant survival, outcomes of the Oxford Knee Score (OKS) and health-related quality of life (EQ-5D-3L), intraoperative complications, and postoperative mortality compared with conventional instrumentation, across a real-world population. Methods: This observational study used National Joint Registry (NJR) data and included adult patients who underwent primary TKR for osteoarthritis between April 1, 2003, and December 31, 2020. The primary analysis evaluated revision for all causes, and secondary analyses evaluated differences in the OKS and EQ-5D-3L at 6 months postoperatively, and mortality within 1 year postoperatively. Weights based on propensity scores were generated, accounting for several covariates. A Cox proportional hazards model was used to assess revision and mortality outcomes. Generalized linear models were used to evaluate differences in the OKS and EQ-5D-3L. Effective sample sizes were computed and represent the statistical power comparable with an unweighted sample. Results: Compared to conventional instrumentation, the hazard ratios (HRs) for all-cause revision following TKR performed using computer navigation and patient-specific instrumentation were 0.937 (95% confidence interval [CI], 0.860 to 1.021; p = 0.136; effective sample size [ESS] = 91,607) and 0.960 (95% CI, 0.735 to 1.252; p = 0.761; ESS = 13,297), respectively. No differences were observed in the OKS and EQ-5D-3L between conventional and computer-navigated TKR (OKS, −0.134 [95% CI, −0.331 to 0.063]; p = 0.183; ESS = 29,135; and EQ-5D-3L, 0.000 [95% CI, −0.005 to 0.005]; p = 0.929; ESS = 28,396) and between conventional TKR and TKR with patient-specific instrumentation (OKS, 0.363 [95% CI, −0.104 to 0.830]; p = 0.127; ESS = 4,412; and EQ-5D-3L, 0.004 [95% CI, −0.009 to 0.018]; p = 0.511; ESS = 4,285). Mortality within 1 year postoperatively was similar between conventional instrumentation and either computer navigation or patient-specific instrumentation (HR, 1.020 [95% CI, 0.989 to 1.052]; p = 0.212; ESS = 110,125). Conclusions: On the basis of this large registry study, we conclude that computer navigation and patient-specific instrumentation have no statistically or clinically meaningful effect on the risk of revision, patient-reported outcomes, or mortality following primary TKR. Level of Evidence: Therapeutic Level II. See Instructions for Authors for a complete description of levels of evidence.

In 2024, the National Institute for Health and Care Excellence (NICE) recommended two new health technologies for paroxysmal nocturnal haemoglobinuria. This review systematically compares the clinical and cost-effectiveness evidence considered within the NICE single technology appraisals of iptacopan, danicopan and pegcetacoplan, examines the consistency of the clinical evidence and economic modelling, and considers whether single technology appraisals are a suitable apparatus for consistent decision making. The studies used different follow-up lengths and used different definitions for reporting breakthrough haemolysis (BTH), but otherwise reported similar outcomes and found a significant benefit for their interventions. A lack of direct evidence and unreliable indirect comparisons meant that naïve comparisons across trials were carried into the economic modelling despite differences in their control arms. Approaches to modelling BTH and associated dose escalation differed across appraisals, despite information for pegcetacoplan coming from the same source in each appraisal, which had a large impact on the economic results. This review raises the question of whether NICE should implement multiple technology appraisals more frequently to reduce these inconsistences. Additionally, we recommend the development of a framework for revisiting positive recommendations when the implementation of health technologies deviates from assumptions made in the economic modelling to ensure cost-effective healthcare is preserved.

Objectives To examine the accuracy and impact of artificial intelligence (AI) software assistance in lung cancer screening using CT. Methods A systematic review of CE-marked, AI-based software for automated detection and analysis of nodules in CT lung cancer screening was conducted. Multiple databases including Medline, Embase and Cochrane CENTRAL were searched from 2012 to March 2023. Primary research reporting test accuracy or impact on reading time or clinical management was included. QUADAS-2 and QUADAS-C were used to assess risk of bias. We undertook narrative synthesis. Results Eleven studies evaluating six different AI-based software and reporting on 19 770 patients were eligible. All were at high risk of bias with multiple applicability concerns. Compared with unaided reading, AI-assisted reading was faster and generally improved sensitivity (+5% to +20% for detecting/categorising actionable nodules; +3% to +15% for detecting/categorising malignant nodules), with lower specificity (−7% to −3% for correctly detecting/categorising people without actionable nodules; −8% to −6% for correctly detecting/categorising people without malignant nodules). AI assistance tended to increase the proportion of nodules allocated to higher risk categories. Assuming 0.5% cancer prevalence, these results would translate into additional 150–750 cancers detected per million people attending screening but lead to an additional 59 700 to 79 600 people attending screening without cancer receiving unnecessary CT surveillance. Conclusions AI assistance in lung cancer screening may improve sensitivity but increases the number of false-positive results and unnecessary surveillance. Future research needs to increase the specificity of AI-assisted reading and minimise risk of bias and applicability concerns through improved study design. PROSPERO registration number CRD42021298449.

Background: Economic evaluation of emerging health technologies is mandated by agencies such as the National Institute for Health and Care Excellence (NICE) to ensure their cost is proportional to their benefit. To avoid bias, NICE stipulate that the benefit of a treatment is assessed across the lifetime of the patient population, which can be many decades. Unfortunately, follow-up from a clinical trial will not usually cover the required period and the observed follow-up will require extrapolation. For survival data this is often done by selecting a preferred model from a set of candidate parametric models. This approach is limited in that the choice of model is restricted to those originally fitted. What if none of the models are consistent with clinical prediction or external data? Method/Results: This paper introduces SurvInt, a tool that estimates the parameters of common parametric survival models which interpolate key survival time co-ordinates specified by the user, which could come from external trials, real world data or expert clinical opinion. This is achieved by solving simultaneous equations based on the survival functions of the parametric models. The application of SurvInt is shown through two examples where traditional parametric modelling did not produce models that were consistent with external data or clinical opinion. Additional features include model averaging, mixture cure models, background mortality, piecewise modelling, restricted mean survival time estimation and probabilistic sensitivity analysis. Conclusions: SurvInt allows precise parametric survival models to be estimated and carried forward into economic models. It provides access to extrapolations that are consistent with multiple data sources such as observed data and clinical predictions, opening the door to precise exploration of regions of uncertainty/disagreement. SurvInt could avoid the need for post-hoc adjustments for complications such as treatment switching, which are often applied to obtain a plausible survival model but at the cost of introducing additional uncertainty. Phase III clinical trials are not designed with extrapolation in mind, and so it is sensible to consider alternative approaches to predict future survival that incorporate external information.