Obstetric and medical factors rather than psychosocial characteristics explain why eligible women do not complete the enhanced recovery after elective caesarean (EREC) pathway: A prospective cohort study.

BACKGROUND: An Australian health-service implemented an 'enhanced recovery after elective caesarean' pathway with next-day discharge. PROBLEM: Previous anecdotal reports indicated that a large percentage of eligible women were not discharged the next day and therefore were not regarded as having completed the pathway. Psychosocial factors were expected to be the leading reason for prolonged hospitalisation. AIM: The study objectives were to: enumerate the percentage of women assessed as eligible for EREC who subsequently did not complete the pathway and the reasons; and to describe women's antenatal satisfaction with preparation, preferences, and perceived support. Women who completed the pathway versus those who did not were compared on antenatal biopsychosocial characteristics. METHODS: This exploratory prospective cohort study enrolled consenting eligible women from antenatal clinics and used patient records and questionnaire data. Comparative statistical techniques were used. FINDINGS: 62 % of women did not complete the pathway, with medical and obstetric factors being the most common reasons (80 %). There was statistically significant evidence of lower antenatal stress levels for those who completed EREC (median=5) relative to those who did not (median=8; P = 0.035); although these findings may not be of clinical importance. Antenatally, 51 % of women felt prepared for early discharge, 36 % needed more information, 19 % disliked hospital, 93 % agreed that family togetherness after birth was important. Most agreed that staff (76 %) and family (67 %) supported the pathway. CONCLUSION: This study indicated that a large percentage of women assessed as eligible did not complete EREC and that obstetric and medical factors, rather than psychosocial characteristics, largely explained this. This provides reassurance to clinicians and women that discharge home is working as intended and is useful for planning similar models of care. Higher stress levels in the antenatal period were demonstrated for women who did not complete EREC suggesting the need for further research into how to support these women.

Psychosocial outcomes from one cohort participating in the STan Australian Randomised controlled Trial (START).

BACKGROUND: In an Australian randomized controlled trial (RCT), two techniques for intrapartum fetal surveillance were compared: ST analysis (STan) as an adjunct to cardiotocography (CTG), compared with CTG alone. The aim was to determine whether CTG + STan could reduce emergency cesarean birth rates while maintaining or improving neonatal outcomes. Secondary aims were to compare clinical, economic, and psychosocial outcomes. The purpose of this paper was to present psychosocial outcomes from one cohort enrolled in the trial. METHODS: The study was conducted at one tertiary referral hospital. Participants who had taken part in the trial from the outset were invited to complete a questionnaire between March 2018 and January 2020, approximately 8 weeks after giving birth. Outcomes included depression, psychological distress, health-related quality of life, and infant feeding practices. Analysis was by intention to treat. RESULTS: N = 207/527 participants completed the questionnaire (n = 113, STan; n = 94, CTG alone). Overall, no statistically significant or clinically meaningful differences were found in the two groups for symptoms of depression, psychological distress, quality of life, or infant feeding. A statistically significant difference was observed for the subscale of pain-discomfort, where scores were higher on average in the CTG alone arm relative to that in the CTG + STan arm. CONCLUSIONS: Although STan as an adjunct to CTG constitutes a different clinical technology from CTG alone, both monitoring types appeared to produce similar results in terms of postnatal psychosocial outcomes for women. Findings from this study provide service users and staff with a comprehensive assessment of STan that can be used to make evidence-informed decisions about monitoring options should STan become more widely available.

The Adelaide Facial Bone Rule: A simple prediction model and clinical guideline for the presence of facial fractures using CT brain scans in victims of minor trauma.

BACKGROUND: Facial fractures bleed, resulting in high-density fluid in the sinuses (haemosinus) on computed tomography (CT) scans. A CT brain scan includes most maxillary sinuses in the scan field, which should allow detection of haemosinus as an indirect indicator of a facial fracture without the need for an additional CT facial bone scan, yet no robust evidence for this exists in the literature. The aim of this study was to determine whether the presence of haemosinus on a CT brain scan, alone or in combination with other clinical information, can predict the presence of facial fractures. METHODS: 1231 adult patients, who had both brain and facial CT scans performed on the same day, were selected from a seven year period. Patients were eligible if scans were requested for trauma. Brain and facial scans were reviewed separately for the presence of facial fractures, haemosinus, emphysema and intra-cranial haemorrhage. Prediction modelling was used to assess whether findings from brain scans could be used to identify patients requiring further CT scanning. FINDINGS: The full prediction model included four predictors and showed excellent discrimination (AUROC 0.982; 95 % CI 0.971 - 0.993). A simplified model, more suitable for clinical implementation, used only facial fractures and haemosinus as predictors. This model showed only marginally poorer discrimination (AUROC 0.964; 95 % CI 0.945 - 0.983) and excellent performance on other measures. CONCLUSION: Based on the excellent performance of the simplified prediction model, we present the Adelaide Facial Bone Rule: The absence of blood in the sinuses or facial fractures on a CT brain scan means a CT facial bone scan does not need to be routinely performed in the setting of clinically-determined minor trauma.

The weighted cumulative exposure method and its application to pharmacoepidemiology: A narrative review.

PURPOSE: The weighted cumulative exposure (WCE) method has been used in a number of fields including pharmacoepidemiology where it can account for intensity, duration and timing of exposures on the risk of an outcome. The method uses a data driven approach with flexible cubic B-splines to assign weights to past doses and select an aetiologically appropriate time window. Predictions of risk are possible for common exposure patterns encountered in real-world studies. The purpose of this study was to describe applications of the WCE method to pharmacoepidemiology and assess the strengths and limitations of the method. METHOD: A literature search was undertaken to find studies applying the WCE method to the study of medicines. Articles published in PubMed using the search term 'weighted cumulative exposure' and articles citing Sylvestre et al. (2009) in Google Scholar or Scopus up to March 2023 were subsequently reviewed. Articles were selected based on title and review of abstracts. RESULTS: Seventeen clinical applications using the data-driven WCE method with flexible cubic splines were identified in the review. These included 3 case-control studies and 14 cohort studies, of which 12 were analysed with Cox proportional hazards models and 2 with logistic regression. Thirteen studies used time windows of 1 year or longer. Of 11 studies which compared conventional models with the WCE method, 10 (91%) studies found a better fit with WCE models while one had an equivalent fit. The freely available 'WCE' software package has facilitated the applications of the WCE method with flexible cubic splines. CONCLUSIONS: The WCE method allows additional insights into the effect of cumulative exposure on outcomes, including the timing and intensity (dose) of the exposure on the risk. The flexibility of the method is particularly well suited to studies with long-term exposures that vary over time or where the current risk of an event is affected by how far the exposure is in the past, which is difficult to model with conventional definitions of exposure. Interpretation of the results can be more complex than for conventional models and would be facilitated by a standardised reporting framework.

Cardiovascular disease, associated risk factors, and risk of dementia: An umbrella review of meta-analyses.

Introduction: Cardiovascular diseases (CVDs) have been associated with an increased risk of dementia; yet the evidence is mixed. This review critically appraises and synthesises current evidence exploring associations between dementia risk and CVD and their risk factors, including coronary heart disease, heart failure, atrial fibrillation, hypertension, hyperlipidaemia, and arterial stiffness. Methods: MEDLINE, Embase, PsycINFO, and the Cochrane Database of Systematic Reviews were searched to identify systematic reviews with meta-analyses investigating the association between at least one of the CVDs of interest and dementia risk. The Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Systematic Reviews was used to assess methodological quality. Results: Twenty-five meta-analyses published between 2007 and 2021 were included. Studies largely consisted of cohorts from North America and Europe. Findings were variable, with coronary heart disease, heart failure, and atrial fibrillation consistently associated with increased risk for all-cause dementia, but results were inconsistent for Alzheimer's disease. Hypertension was more frequently associated with dementia during mid-life compared to late life. Findings concerning cholesterol were complex, and while results were inconsistent for low-density lipoprotein cholesterol and total cholesterol, there appeared to be no associations between triglycerides and high-density lipoprotein cholesterol. All meta-analyses investigating hypercholesterolaemia showed significant increases in dementia risk. There was a paucity of research on the association between arterial stiffness and dementia risk. Conclusion: Targeted CVD dementia prevention strategies could reduce dementia prevalence. Future research should determine the underpinning mechanisms linking heart and brain health to determine the most effective strategies for dementia risk reduction in CVD populations.

Effect of a multi-faceted rapid response system re-design on repeat calling of the rapid response team.

BACKGROUND: Repeat Rapid Response Team (RRT) calls are associated with increased in-hospital mortality risk and pose an organisation-level resource burden. Use of Non-Technical Skills (NTS) at calls has the potential to reduce potentially preventable repeat calling. NTS are usually improved through training, although this consumes time and financial resources. Re-designing the Rapid Response System (RRS) to promote use of NTS may provide a feasible alternative. METHODS: A pre-post observational study was undertaken to assess the effect of an RRS re-design that aimed to promote use of NTS during RRT calls. The primary outcome was the proportion of admissions each month subject to repeat RRT calling, and the average number of repeat calls per admission each month was the secondary outcome of interest. Univariate and multivariable interrupted time series analyses compared outcomes between the two study phases. RESULTS: The proportion of admissions with repeat calls each month increased across both phases of the study period, but the increase was lower in the post re-design phase (change in regression slope -0.12 (standard error 0.07) post versus pre re-design). The multivariable model predicted a 6% reduction (95% confidence interval -15.1-3.1; P = 0.19) in the proportion of admissions having repeat calls at the end of the post redesign phase study compared to the predicted proportion in the absence of the re-design. The average number of calls per admission was also predicted to decrease in the post re-design phase, with an estimated difference of -0.07 calls per admission (equivalent to one fewer repeat call per 14 patients who had RRT calls) at the end of the post re-design phase (95% confidence interval -0.23-0.08, P = 0.35). CONCLUSION: This study of an RRS re-design showed modest, but not statistically significant, reductions in the proportion of admissions with repeat calls and the mean number of repeat calls per admission. Given the economic and workforce capacity issues that all health care systems now face, even small improvements in the RRS may have lasting impact across the organisation. For the potential interest of RRS managers, this paper presents a pragmatic, low-cost initiative intended to enhance communication and cooperation at RRT calls.

Extrapolation of Survival Curves Using Standard Parametric Models and Flexible Parametric Spline Models: Comparisons in Large Registry Cohorts with Advanced Cancer.

BACKGROUND: It is often important to extrapolate survival estimates beyond the limited follow-up times of clinical trials. Extrapolated survival estimates can be highly sensitive to model choice; thus, appropriate model selection is crucial. Flexible parametric spline models have been suggested as an alternative to standard parametric models; however, their ability to extrapolate is not well understood. AIM: To determine how well standard parametric and flexible parametric spline models predict survival when fitted to registry cohorts with artificially right-censored follow-up times. METHODS: Adults with advanced breast, colorectal, small cell lung, non-small cell lung, or pancreatic cancer with a potential follow-up time of 10 y were selected from the SEER 1973-2015 registry data set. Patients were classified into 15 cohorts by cancer and age group at diagnosis (18-59, 60-69, 70+ y). Follow-up times for each cohort were right censored at 20%, 35%, and 50% survival. Standard parametric models (exponential, Weibull, Gompertz, log-logistic, log-normal, generalized gamma) and spline models (proportional hazards, proportional odds, normal/probit) were fitted to the 10-y data set and the 3 right-censored data sets. Predicted 10-y restricted mean survival time and percentage surviving at 10 y were compared with the observed values. RESULTS: Across all data sets, the spline odds and spline normal models most frequently gave accurate predictions of 10-y survival outcomes. Visually, spline models tended to demonstrate better fit to the observed hazard functions than standard parametric models, both in the censored and 10-y data. CONCLUSIONS: In these cohorts, where there was little uncertainty in the observed data, the spline models performed well when extrapolating beyond the observed data. Spline models should be routinely included in the set of models that are fitted when extrapolating cancer survival data.

Reduced length of hospital stay after caesarean section: A systematic review examining women's experiences and psychosocial outcomes.

BACKGROUND: Globally, reducing hospital stays after caesarean section is becoming more prevalent. Whilst this reduction in length of stay after caesarean section has not been found to be associated with adverse maternal health outcomes, the psychosocial impact and women's experiences have not been systematically reviewed. This review aims to evaluate the literature on women's experiences and psychosocial outcomes (including infant feeding) associated with a reduced hospital stay after caesarean section. METHODS: A mixed methods systematic review examining records between 1980 and 2019 was undertaken. The review included research which defines a reduced length of stay in comparison with standard care or a comparator with a longer discharge time. It considered data related to the antenatal period, time of discharge and postnatal period. The following databases were searched: PsycINFO, CINAHL, PubMed, Embase and ProQuest Dissertations and Theses. 13,760 records were identified, after duplicates were removed, 10,902 articles were reviewed for suitability by title and abstract. 78 full text articles were assessed, and the final review included 8 articles. RESULTS: A total of 8 articles were included, and four areas were examined: satisfaction with care, mental wellbeing, infant feeding and pain. Articles were of mixed quality when assessed using the Mixed Methods Appraisal Tool. CONCLUSIONS: This review indicated no evidence of a systematic negative impact on women's psychosocial outcomes and experiences. The review also identifies a number of characteristics of care associated with more positive experiences and psychosocial outcomes. These include the provision of support systems, access to pain management before and after discharge and continued care with home midwifery. The limited number of studies point to the need for more research, and especially those using qualitative methods.

Re-designing a rapid response system: effect on staff experiences and perceptions of rapid response team calls.

BACKGROUND: Rapid Response Team (RRT) calls are clinical crises. Clinical and time pressures can hinder effective liaison between staff who call the RRT ('users') and those responding as part of the RRT ('members'). Non-technical skills (NTS) training has been shown to improve communication and cooperation but requires time and financial resources that may not be available in acute care hospitals. Rapid Response System (RRS) re-design, aiming to promote use of NTS, may provide an alternative approach to improving interactions within RRTs and between members and users. METHODS: Re-design of an existing mature RRS was undertaken in a tertiary, metropolitan hospital incorporating the addition of: 1) regular RRT meetings 2) RRT role badges and 3) a structured member-to-user patient care responsibility "hand-off" process. To compare experiences and perceptions of calls, users and members were surveyed pre and post re-design. RESULTS: Post re-design there were improvements in members' understanding of RRT roles (P = 0.03) and responsibilities (P < 0.01), and recollection of introducing themselves to users (P = 0.02). For users, after the re-design, there were improvements in identification of the RRT leader (P < 0.01), and in the development of clinical plans for patients remaining on the ward at the end of an RRT call (P < 0.01). However, post-re-design, fewer users agreed that the structured hand-off was useful or that they should be involved in the process. Both members and users reported fewer experiences of conflict at RRT calls post-re-design (both P < 0.01). CONCLUSION: The RRS re-design yielded improvements in interactions between members in RRTs and between RRT members and users. However, some unintended consequences arose, particularly around user satisfaction with the structured hand-off. These findings suggest that refinement and improvement of the RRS is possible, but should be an ongoing iterative effort, ideally supported by staff training. TRIAL REGISTRATION: NCT01551160. Registered: 12th March 2012.

Adjusting for Treatment Switching in Oncology Trials: A Systematic Review and Recommendations for Reporting.

OBJECTIVES: To systematically review the quality of reporting on the application of switching adjustment approaches in published oncology trials and industry submissions to the National Institute for Health and Care Excellence Although methods such as the rank preserving structural failure time model (RPSFTM) and inverse probability of censoring weights (IPCW) have been developed to address treatment switching, the approaches are not widely accepted within health technology assessment. This limited acceptance may partly be a consequence of poor reporting on their application. METHODS: Published trials and industry submissions were obtained from searches of PubMed and nice.org.uk, respectively. The quality of reporting in these studies was judged against a checklist of reporting recommendations, which was developed by the authors based on detailed considerations of the methods. RESULTS: Thirteen published trials and 8 submissions to nice.org.uk satisfied inclusion criteria. The quality of reporting around the implementation of the RPSFTM and IPCW methods was generally poor. Few studies stated whether the adjustment approach was prespecified, more than a third failed to provide any justification for the chosen method, and nearly half neglected to perform sensitivity analyses. Further, it was often unclear how the RPSFTM and IPCW methods were implemented. CONCLUSIONS: Inadequate reporting on the application of switching adjustment methods increases uncertainty around results, which may contribute to the limited acceptance of these methods by decision makers. The proposed reporting recommendations aim to support the improved interpretation of analyses undertaken to adjust for treatment switching.

A qualitative study of a sample of women participating in an Australian randomised controlled trial of intrapartum fetal surveillance.

BACKGROUND: The STan Australian Randomised controlled Trial (START), the first of its kind in Australia, compares two techniques of intrapartum fetal surveillance (cardiotocographic electronic fetal monitoring (CTG) plus analysis of the ST segment of the fetal electrocardiogram (STan+CTG) with CTG alone) with the aim of reducing unnecessary obstetric intervention. It is also the first comprehensive intrapartum fetal surveillance (IFS) trial worldwide, including qualitative examination of psychosocial outcomes and cost-effectiveness. In evaluating and implementing healthcare interventions, the perspectives and experiences of individuals directly receiving them is an integral part of a comprehensive assessment. Furthermore, the added value of using qualitative research alongside randomised controlled trials (RCTs) is becoming widely acknowledged. OBJECTIVE: This study aimed to examine women's experiences with the type of IFS they received in the START trial. METHODS: Using a qualitative research design, a sample of thirty-two women were interviewed about their experiences with the fetal monitoring they received. Data were analysed using thematic analysis. FINDINGS: Six themes emerged from analysis: reassurance, mobility, discomfort, perception of the fetal Scalp Electrode (FSE), and overall positive experience. CONCLUSION: Interestingly, it was found that women who had an FSE in the CTG alone arm of the trial reported very similar experiences to women in the STan+CTG arm of the trial. Despite STan and CTG differing clinically, from women's perspectives, the primary difference between the two techniques was the utilisation (or not) of the FSE. Women were very accepting of STan+CTG as it was perceived and experienced as a more accurate form of monitoring than CTG alone. Findings from this study have significant implications for health professionals including midwives and obstetricians and implications for standard practice and care. The study has demonstrated the importance and significance of incorporating qualitative enquiry within RCTs.

Reasons for Repeat Rapid Response Team Calls, and Associations with In-Hospital Mortality.

BACKGROUND: Previous publications noted increased mortality risk in patients subject to repeat rapid response team (RRT) calls. These patients were examined as a homogenous group, but there may be many reasons for repeat calls. Those potentially preventable by the rapid response system have not been investigated. METHODS: In a retrospective cohort study, patients with potentially preventable repeat calls were classified into two categories: type 1 (patients who had a repeat call following an initial call that ended despite the patient still triggering RRT calling criteria [T1-PRC]) and type 2 (patients with a repeat call within 24 hours of an initial call and for the same reason [T2-PRC]). In-hospital mortality for these patients and for those with repeat calls for all other reasons (ORC) were compared to patients with only a single call during their admission (SC). RESULTS: Mortality occurred in 31 (43.7%) T1-PRC, 13 (15.1%) T2-PRC, 56 (28.9%) ORC, and 289 (13.9%) SC patients. Univariate odds ratios (ORs), in comparison to SC patients, were 4.81 (95% confidence interval [CI]: 2.96-7.81; p < 0.001), 1.10 (95% CI: 0.60-2.02; p = 0.75), and 2.52 (95% CI: 1.80-3.52; p < 0.001), respectively. Mortality effects persisted for the T1-PRC and ORC groups after adjustment for patient, admission, and initial call characteristics with ORs of 4.07 (95% CI: 2.36-7.01; p < 0.001) and 2.29 (95% CI: 1.57-3.34; p < 0.001), respectively. CONCLUSION: This study found that repeat calls following an initial call that ended with ongoing breach of predefined calling criteria were strongly associated with increased mortality. This highlights the risk to patients when the RRT leaves reversible clinical deterioration unresolved at the end of a call.

Women's psychosocial outcomes following an emergency caesarean section: A systematic literature review.

BACKGROUND: Given the sudden and unexpected nature of an emergency caesarean section (EmCS) coupled with an increased risk of psychological distress, it is particularly important to understand the psychosocial outcomes for women. The aim of this systematic literature review was to identify, collate and examine the evidence surrounding women's psychosocial outcomes of EmCS worldwide. METHODS: The electronic databases of EMBASE, PubMed, Scopus, and PsycINFO were searched between November 2017 and March 2018. To ensure articles were reflective of original and recently published research, the search criteria included peer-reviewed research articles published within the last 20 years (1998 to 2018). All study designs were included if they incorporated an examination of women's psychosocial outcomes after EmCS. Due to inherent heterogeneity of study data, extraction and synthesis of both qualitative and quantitative data pertaining to key psychosocial outcomes were organised into coherent themes and analysis was attempted. RESULTS: In total 17,189 articles were identified. Of these, 208 full text articles were assessed for eligibility. One hundred forty-nine articles were further excluded, resulting in the inclusion of 66 articles in the current systematic literature review. While meta-analyses were not possible due to the nature of the heterogeneity, key psychosocial outcomes identified that were negatively impacted by EmCS included post-traumatic stress, health-related quality of life, experiences, infant-feeding, satisfaction, and self-esteem. Post-traumatic stress was one of the most commonly examined psychosocial outcomes, with a strong consensus that EmCS contributes to both symptoms and diagnosis. CONCLUSIONS: EmCS was found to negatively impact several psychosocial outcomes for women in particular post-traumatic stress. While investment in technologies and clinical practice to minimise the number of EmCSs is crucial, further investigations are needed to develop effective strategies to prepare and support women who experience this type of birth.

Assessing Therapeutic Alliance in the Context of mHealth Interventions for Mental Health Problems: Development of the Mobile Agnew Relationship Measure (mARM) Questionnaire.

BACKGROUND: Digital health interventions in the form of smartphone apps aim to improve mental health and enable people access to support as and when needed without having to face the stigma they may experience in accessing services. If we are to evaluate mobile health (mHealth) apps and advance scientific understanding, we also need tools to help us understand in what ways mHealth interventions are effective or not. The concept of therapeutic alliance, a measure of the quality of the relationship between a health care provider and a service user, is a key factor in explaining the effects of mental health interventions. The Agnew Relationship Measure (ARM) is a well-validated measure of therapeutic alliance in face-to-face therapy. OBJECTIVE: This study presented the first attempt to (1) explore service users' views of the concept of relationship within mHealth mental health interventions and (2) adapt a well-validated face-to-face measure of therapeutic alliance, the Agnew Relationship Measure (ARM), for use with mHealth interventions. METHODS: In stage 1, we interviewed 9 mental health service users about the concept of therapeutic alliance in the context of a digital health intervention and derived key themes from interview transcripts using thematic analysis. In stage 2, we used rating scales and open-ended questions to elicit views from 14 service users and 10 mental health staff about the content and face validity of the scale, which replaced the word "therapist" with the word "app." In stage 3, we used the findings from stages 1 and 2 to adapt the measure with the support of a decision-making algorithm about which items to drop, retain, or adapt. RESULTS: Findings suggested that service users do identify relationship concepts when thinking about mHealth interventions, including forming a bond with an app and the ability to be open with an app. However, there were key differences between relationships with health professionals and relationships with apps. For example, apps were not as tailored and responsive to each person's unique needs. Furthermore, apps were not capable of portraying uniquely human-like qualities such as friendliness, collaboration, and agreement. We made a number of changes to the ARM that included revising 16 items; removing 4 items due to lack of suitable alternatives; and adding 1 item to capture a key theme derived from stage 1 of the study ("The app is like having a member of my care team in my pocket"). CONCLUSIONS: This study introduces the mHealth version of the ARM, the mARM, that has good face and content validity. We encourage researchers to include this easy-to-use tool in digital health intervention studies to gather further data about its psychometric properties and advance our understanding of how therapeutic alliance influences the efficacy of mHealth interventions. TRIAL REGISTRATION: International Standard Randomized Controlled Trial Number (ISRCTN) 34966555; http://www.isrctn.com/ISRCTN34966555 (Archived by WebCite at http://www.webcitation.org/6ymBVwKif).

Should multiple imputation be the method of choice for handling missing data in randomized trials?

The use of multiple imputation has increased markedly in recent years, and journal reviewers may expect to see multiple imputation used to handle missing data. However in randomized trials, where treatment group is always observed and independent of baseline covariates, other approaches may be preferable. Using data simulation we evaluated multiple imputation, performed both overall and separately by randomized group, across a range of commonly encountered scenarios. We considered both missing outcome and missing baseline data, with missing outcome data induced under missing at random mechanisms. Provided the analysis model was correctly specified, multiple imputation produced unbiased treatment effect estimates, but alternative unbiased approaches were often more efficient. When the analysis model overlooked an interaction effect involving randomized group, multiple imputation produced biased estimates of the average treatment effect when applied to missing outcome data, unless imputation was performed separately by randomized group. Based on these results, we conclude that multiple imputation should not be seen as the only acceptable way to handle missing data in randomized trials. In settings where multiple imputation is adopted, we recommend that imputation is carried out separately by randomized group.

Multiple imputation for handling missing outcome data when estimating the relative risk.

BACKGROUND: Multiple imputation is a popular approach to handling missing data in medical research, yet little is known about its applicability for estimating the relative risk. Standard methods for imputing incomplete binary outcomes involve logistic regression or an assumption of multivariate normality, whereas relative risks are typically estimated using log binomial models. It is unclear whether misspecification of the imputation model in this setting could lead to biased parameter estimates. METHODS: Using simulated data, we evaluated the performance of multiple imputation for handling missing data prior to estimating adjusted relative risks from a correctly specified multivariable log binomial model. We considered an arbitrary pattern of missing data in both outcome and exposure variables, with missing data induced under missing at random mechanisms. Focusing on standard model-based methods of multiple imputation, missing data were imputed using multivariate normal imputation or fully conditional specification with a logistic imputation model for the outcome. RESULTS: Multivariate normal imputation performed poorly in the simulation study, consistently producing estimates of the relative risk that were biased towards the null. Despite outperforming multivariate normal imputation, fully conditional specification also produced somewhat biased estimates, with greater bias observed for higher outcome prevalences and larger relative risks. Deleting imputed outcomes from analysis datasets did not improve the performance of fully conditional specification. CONCLUSIONS: Both multivariate normal imputation and fully conditional specification produced biased estimates of the relative risk, presumably since both use a misspecified imputation model. Based on simulation results, we recommend researchers use fully conditional specification rather than multivariate normal imputation and retain imputed outcomes in the analysis when estimating relative risks. However fully conditional specification is not without its shortcomings, and so further research is needed to identify optimal approaches for relative risk estimation within the multiple imputation framework.

Treatment of missing data in follow-up studies of randomised controlled trials: A systematic review of the literature.

BACKGROUND/AIMS: After completion of a randomised controlled trial, an extended follow-up period may be initiated to learn about longer term impacts of the intervention. Since extended follow-up studies often involve additional eligibility restrictions and consent processes for participation, and a longer duration of follow-up entails a greater risk of participant attrition, missing data can be a considerable threat in this setting. As a potential source of bias, it is critical that missing data are appropriately handled in the statistical analysis, yet little is known about the treatment of missing data in extended follow-up studies. The aims of this review were to summarise the extent of missing data in extended follow-up studies and the use of statistical approaches to address this potentially serious problem. METHODS: We performed a systematic literature search in PubMed to identify extended follow-up studies published from January to June 2015. Studies were eligible for inclusion if the original randomised controlled trial results were also published and if the main objective of extended follow-up was to compare the original randomised groups. We recorded information on the extent of missing data and the approach used to treat missing data in the statistical analysis of the primary outcome of the extended follow-up study. RESULTS: Of the 81 studies included in the review, 36 (44%) reported additional eligibility restrictions and 24 (30%) consent processes for entry into extended follow-up. Data were collected at a median of 7 years after randomisation. Excluding 28 studies with a time to event primary outcome, 51/53 studies (96%) reported missing data on the primary outcome. The median percentage of randomised participants with complete data on the primary outcome was just 66% in these studies. The most common statistical approach to address missing data was complete case analysis (51% of studies), while likelihood-based analyses were also well represented (25%). Sensitivity analyses around the missing data mechanism were rarely performed (25% of studies), and when they were, they often involved unrealistic assumptions about the mechanism. CONCLUSION: Despite missing data being a serious problem in extended follow-up studies, statistical approaches to addressing missing data were often inadequate. We recommend researchers clearly specify all sources of missing data in follow-up studies and use statistical methods that are valid under a plausible assumption about the missing data mechanism. Sensitivity analyses should also be undertaken to assess the robustness of findings to assumptions about the missing data mechanism.