Missing data in trial‐based cost‐effectiveness analysis: An incomplete journey

Cost‐effectiveness analyses (CEA) conducted alongside randomised trials provide key evidence for informing healthcare decision making, but missing data pose substantive challenges. Recently, there have been a number of developments in methods and guidelines addressing missing data in trials. However, it is unclear whether these developments have permeated CEA practice. This paper critically reviews the extent of and methods used to address missing data in recently published trial‐based CEA. Issues of the Health Technology Assessment journal from 2013 to 2015 were searched. Fifty‐two eligible studies were identified. Missing data were very common; the median proportion of trial participants with complete cost‐effectiveness data was 63% (interquartile range: 47%–81%). The most common approach for the primary analysis was to restrict analysis to those with complete data (43%), followed by multiple imputation (30%). Half of the studies conducted some sort of sensitivity analyses, but only 2 (4%) considered possible departures from the missing‐at‐random assumption. Further improvements are needed to address missing data in cost‐effectiveness analyses conducted alongside randomised trials. These should focus on limiting the extent of missing data, choosing an appropriate method for the primary analysis that is valid under contextually plausible assumptions, and conducting sensitivity analyses to departures from the missing‐at‐random assumption.     

Moderating factors for the effectiveness of group art therapy for schizophrenia: secondary analysis of data from the MATISSE randomised controlled trial

Purpose

Although some studies suggest that art therapy may be useful in the treatment of negative symptoms of schizophrenia, a recent large trial of group art therapy found no clinical advantage over standard care, but the study population was heterogeneous and uptake of the intervention was poor. This study aimed to investigate whether art therapy was more effective for specific subgroups of patients.

Methods

Secondary analysis of data from a randomised controlled trial of group art therapy as an adjunctive treatment for schizophrenia (n = 140) versus standard care alone (n = 137). Positive and Negative Syndrome Scale scores at 12 months were compared between trial arms. Interaction between intervention effect and different subgroups, including those with more severe negative symptoms of schizophrenia, and those who expressed a preference for art therapy prior to randomisation, was tested using a linear mixed model.

Results

The clinical effectiveness of group art therapy did not significantly differ between participants with more or less severe negative symptoms [interaction for difference in PANSS = 1.7, 95 % CI (?8.6 to 12.1), P = 0.741], or between those who did and did not express a preference for art therapy [interaction = 3.9, 95 % CI (?6.7 to 14.5), P = 0.473]. None of the other exploratory subgroups suggested differences in intervention effect.

Conclusions

There was no evidence of greater improvement in clinical symptoms of schizophrenia for those with more severe negative symptoms or those with a preference for art therapy. Identification of patients with schizophrenia who may benefit most from group art therapy remains elusive.     

Aortic valvuloplasty in the elderly. Immediate and medium-term results

Between January 86 and February 88 we attempted percutaneous aortic valvuloplasty in 100 patients. 68 patients could be dilated with hemodynamic success: the aortic valve area increased from 0.52 +/- 0.17 cm2 to 0.78 +/- 0.27 cm2 (p less than 0.001). Cardiac output did not change significantly (4 +/- 1.1 l/mn). In the remaining 32 patients: 8 patients (8%) died in hospital and the morbidity rate was 11%: arterial injury (9), stroke (1) and major cardiac event (1). Sixty patients regarded as initial success were followed for 6 months: 3 patients (5%) died without cardiac events, functional improvement persisted in 41 patients (68%) and clinical failure was observed in 16 patients (27%): 4 patients died and in the remaining 12 patients, the recurrence of symptoms required medical treatment (5), surgery (5) or repeat percutaneous valvuloplasty (2). Echo-Doppler maximum aortic valve gradient was significantly less in improved patients (55 +/- 21 mmHg) than in deteriorated patients (78 +/- 36 mmHg) (p less than 0.02). One year after valvuloplasty: actuarial survival was 74 +/- 6%, clinical improvement was 53 +/- 7% and clinical recurrence was 41 +/- 7%. Thus percutaneous aortic valvuloplasty may be proposed only to non surgical patients.     

Early non-obstructive thrombosis of mechanical mitral valve prostheses

Transoesophageal echocardiography has shown a high incidence on non-obstructive thrombosis after mitral valve replacement with a mechanical prosthesis. The unpredictable outcome and the period during which the complication arises make treatment difficult. The aim of this study was to assess the tolerance and efficacy of the association of long-term heparin and oral anticoagulation, as recommended in this indication. All patients undergoing mitral valve replacement with a mechanical prosthesis between June 1999 and July 2001 were systematically included and studied by transoesophageal echocardiography in the immediate postoperative period. Those with non-obstructive thrombosis at least 5 mm in size were treated by heparin and oral coagulation until the thrombus disappeared on transoesophageal echocardiography. One hundred and fourteen patients undergoing 120 mitral valve replacements (6 reoperations) underwent transoesophageal echocardiography and non-obstructive thrombi measuring at least 5 mm were found on 26 occasions (21.7%). The association of heparin and oral coagulation was maintained for 7 to 115 days (average 20 days). No thromboembolic or haemorrhagic complications and no deaths were observed during this period. Two patients were treated with danaparoid and oral anticoagulation because of heparin-induced thrombocytopenia before the diagnosis. None of the patients died during follow-up (average 49 months); there were 4 recurrent non-obstructive thromboses, three of which were complicated by thromboembolic events with no sequellae in the first 8 months, again treated effectively with the association of heparin and oral anticoagulants; two cerebral embolic events without sequellae were observed without a demonstrable non-obstructive thrombus on transoesophageal echocardiography. The authors conclude that the association of heparin and oral anticoagulants seems well tolerated and effective in this small population and this would justify a large scale clinical trial.     

Direct coronary stenting without balloon predilation of lesions requiring long stents: immediate and 6-month results of a multicenter prospective registry.

To assess the outcomes of direct coronary stenting (DS) using long stents and examine predictive factors of DS failure, this prospective multicenter registry included 128 consecutive patients who underwent the implantation of stents >or= 18 mm in length without balloon predilation for de novo coronary artery stenoses. Mean lesion and stent lengths were 20.7 +/- 5.4 and 21.4 +/- 3.8 mm, respectively. Rates of DS success, lesion success, and primary success were 82%, 99%, and 97.7%, respectively. At 6 months, rates of MACE and TVR were 12.5% and 6.3%, respectively. In multivariate analysis, factors predictive of DS failure vs. DS success were presence of calcifications (78% vs. 45%; P = 0.004) and reference vessel diameter (2.77 +/- 0.4 vs. 3.13 +/- 0.42 mm; P = 0.0002). DS of complex lesions with stents >or= 18 mm in length was performed safely and with a high success rate. This strategy was less successful in the treatment of small vessels and in presence of calcifications.     

Comparative results of coronary angioplasty performed at the time of coronarography and at a later time

Between January 1986 and December 1988, 558 patients underwent percutaneous transluminal coronary angioplasty (PTCA) of whom 40 per cent were dilated at the time of diagnostic coronary arteriography. In order to assess the value of this therapeutic strategy we compared the results of 221 patients dilated at the time of diagnostic coronary arteriography (Group 1) with those of 337 patients who underwent deferred PTCA. In Group 1, the incidence of stable angina was lower (26.7% vs 46.3%, p less than 10-5), that of thrombolysed myocardial infarction was higher (24% vs 2.7%, p less than 10-9) and a higher proportion of patients had previously undergone PTCA (29.4% vs 3.2%, p less than 10-9). The proportion of patients with single vessel disease was higher in Group 1 (84.6% vs 74.7%, p less than 0.01) as was that of angioplasty of a single lesion (97.7% vs 88.1%, p less than 10-4). There were fewer dilations of the left circumflex artery in Group 1 (17.2% vs 27.3%, p less than 0.05) which was compensated by a higher number involving the right coronary artery (26.1% vs 15.5%, p less than 0.01). The immediate results were comparable in the two groups with 87.8 per cent primary successes, 3.6 per cent of myocardial infarcts and 1.3 per cent of coronary bypass operations with no fatalities in Group 1. These favorable results encourage the development of PTCA at the time of diagnostic coronary arteriography in the following indications: unstable angina, thrombolysed myocardial infarction and restenosis irrespective of the patient's symptomatology.(ABSTRACT TRUNCATED AT 250 WORDS)