What impact do assumptions about missing data have on conclusions? A practical sensitivity analysis for a cancer survival registry.

BACKGROUND: Within epidemiological and clinical research, missing data are a common issue and often over looked in publications. When the issue of missing observations is addressed it is usually assumed that the missing data are 'missing at random' (MAR). This assumption should be checked for plausibility, however it is untestable, thus inferences should be assessed for robustness to departures from missing at random. METHODS: We highlight the method of pattern mixture sensitivity analysis after multiple imputation using colorectal cancer data as an example. We focus on the Dukes' stage variable which has the highest proportion of missing observations. First, we find the probability of being in each Dukes' stage given the MAR imputed dataset. We use these probabilities in a questionnaire to elicit prior beliefs from experts on what they believe the probability would be in the missing data. The questionnaire responses are then used in a Dirichlet draw to create a Bayesian 'missing not at random' (MNAR) prior to impute the missing observations. The model of interest is applied and inferences are compared to those from the MAR imputed data. RESULTS: The inferences were largely insensitive to departure from MAR. Inferences under MNAR suggested a smaller association between Dukes' stage and death, though the association remained positive and with similarly low p values. CONCLUSIONS: We conclude by discussing the positives and negatives of our method and highlight the importance of making people aware of the need to test the MAR assumption.

Patients' views on an education booklet following spinal surgery.

PURPOSE: This study evaluated an evidence-based education booklet developed for patients undergoing spinal surgery which was used as a treatment intervention in a multi-centre, factorial, randomised controlled trial (FASTER: Function after spinal treatment, exercise and rehabilitation) investigating the post-operative management of spinal surgery patients. This study sought to determine the acceptability and content of the booklet to patients. METHODS: Patients receiving the educational booklet before discharge from hospital as part of the FASTER study were asked to complete an evaluation, which rated the booklet "Your Back Operation" with regard to content, information, usability, etc. using forced and open questions. This assessment was conducted at the same time as the initial 6-week post-operative review performed as part of the larger study. RESULTS: Therefore, 97% of the 117 trial participants who returned their 6-week evaluation and randomised to receive a booklet returned their questionnaire. The booklet was highly rated receiving an overall rating of 7 or more out of 10 from 101/111 (91%), and high ratings for content, readability and information. The booklet's key messages were clear to the majority of patients; however, many patients highlighted deficiencies with respect to content particularly in relation to wound care and exercise. CONCLUSIONS: Patients valued the booklet and rated its content highly. Many suggested that the booklet be developed further and there was a clear desire for specific exercises to be included even though there is no evidence to support specific exercise prescription.

The reliability and variation of measurements of the os calcis angles in children.

The databases of the Picture Archiving and Communication Systems of two hospitals were searched and all children who had a lateral radiograph of the ankle during their attendance at the emergency department were identified. In 227 radiographs, Bohler's and Gissane's angles were measured on two separate occasions and by two separate authors to allow calculation of inter- and intra-observer variation. Intraclass correlation coefficients were used to assess the reliability of the measurements. For Bohler's angle the overall inter-observer reliability, the intraclass correlation coefficient was 0.90 and the intra-observer reliability 0.95, giving excellent agreement. This reliability was maintained across the age groups. For Gissane's angle, inter- and intra-observer reliability was only fair or poor across most age groups. Further analysis of the Bohler's angle showed a significant variation in the mean angle with age. Contrary to published opinion, the angle is not uniformly lower than that of adults but varies with age, peaking towards the end of the first decade before attaining adult values. The age-related radiologic changes presented here may help in the interpretation of injuries to the hindfoot in children.

Function after spinal treatment, exercise and rehabilitation (FASTER): improving the functional outcome of spinal surgery.

BACKGROUND: The life-time incidence of low back pain is high and diagnoses of spinal stenosis and disc prolapse are increasing. Consequently, there is a steady rise in surgical interventions for these conditions. Current evidence suggests that while the success of surgery is incomplete, it is superior to conservative interventions. A recent survey indicates that there are large differences in the type and intensity of rehabilitation, if any, provided after spinal surgery as well as in the restrictions and advice given to patients in the post-operative period. This trial will test the hypothesis that functional outcome following two common spinal operations can be improved by a programme of post-operative rehabilitation that combines professional support and advice with graded active exercise and/or an educational booklet based on evidence-based messages and advice. METHODS/DESIGN: The study design is a multi-centre, factorial, randomised controlled trial with patients stratified by surgeon and operative procedure. The trial will compare the effectiveness and cost-effectiveness of a rehabilitation programme and an education booklet for the postoperative management of patients undergoing discectomy or lateral nerve root decompression, each compared with "usual care"using a 2 x 2 factorial design. The trial will create 4 sub-groups; rehabilitation-only, booklet-only, rehabilitation-plus-booklet, and usual care only. The trial aims to recruit 344 patients, which equates to 86 patients in each of the four sub-groups. All patients will be assessed for functional ability (through the Oswestry Disability Index - a disease specific functional questionnaire), pain (using visual analogue scales), and satisfaction pre-operatively and then at 6 weeks, 3, 6 and 9 months and 1 year post-operatively. This will be complemented by a formal analysis of cost-effectiveness. DISCUSSION: This trial will determine whether the outcome of spinal surgery can be enhanced by either a post-operative rehabilitation programme or an evidence-based advice booklet or a combination of the two and as such will contribute to our knowledge on how to manage spinal surgery patients in the post-operative period.

Early N-terminal pro-brain natriuretic peptide measurements predict clinically significant ductus arteriosus in preterm infants.

UNLABELLED: We report a blinded, prospective study of the diagnostic utility of N-terminal pro-brain natriuretic peptide (NTproBNP) measurements for predicting clinically significant patent ductus arteriosus (PDA) and assessing closure. METHODS: Plasma NTproBNP was measured during the first week in 100 preterm babies (mean gestation 28.8 +/- 2.9 weeks; mean birth weight 1224 +/- 512 g). Echocardiography was performed between days 5 and 7 by operators, blinded to NTproBNP concentration. RESULTS: NTproBNP peaked on days 2 and 3, declined by day 7. Twenty babies, later treated for PDA, had significantly higher NTproBNP levels throughout. Areas under receiver operating characteristic (ROC) curves were 0.896, 0.897 and 0.931 on days 2, 3 and 7, respectively (p < 0.0001). A concentration > 2850 pmol/L had diagnostic sensitivity of 90% and specificity of 89% (95% CI: 68, 99; likelihood ratio 8.10). Ductal closure was associated with a fall in mean NTproBNP from 3003 to 839 pmol/L (p < 0.001). CONCLUSION: N-terminal pro B-type brain natriuretic peptide (NTproBNP) concentrations peaked and then declined in the first week but remained higher in preterm babies whose PDA required treatment. NTproBNP on day 3 predicted whether a neonatal physician blinded to results would treat a PDA. Fall in plasma NTproBNP indicated closure.