Multiple imputation by predictive mean matching in cluster-randomized trials

Health And Performance Promotion in Youth (HAPPY) hybrid effectiveness-implementation cluster randomised trial: comparison of two strategies to implement an injury prevention exercise programme in Danish youth handball

OBJECTIVE

To investigate if a combination of an online and onsite implementation strategy was superior to an online-only strategy in enhancing the use of an injury prevention exercise programme (IPEP) and in reducing the risk of shoulder, knee and ankle injuries in youth community handball players (age 11-17) over a handball season.

METHODS

In this 30-week hybrid effectiveness-implementation cluster randomised type 3 study, 20 youth handball clubs were randomly assigned 1:1 to either a combined online and onsite implementation strategy (coach workshop using the health action process approach behaviour change model and health service provider (HSP) support) or an online-only strategy (control group). The primary implementation outcome was coach-reported adherence, measured as the average IPEP exercise usage by the team over 30 weeks. The primary effectiveness outcome was player-reported handball playing time to any new handball-related shoulder, knee and ankle injuries, reported weekly using the Oslo Sports Trauma Research Centre Questionnaire on Health Problems.

RESULTS

We enrolled 63 coaches (27% women) and 945 players (mean age 14.5 years, 55% girls). Intention-to-treat analyses showed no statistically significant difference between implementation strategies in adherence (between-group difference 1.4, 95% CI -0.5 to 3.4) or in cumulative injury risk (between-group difference 5.5% points, 95% CI -2.2 to 13.1).

CONCLUSION

Our findings demonstrate that in youth community handball, a combined online and onsite implementation strategy, including a coach workshop and HSP support, was not superior to an online-only strategy regarding adherence to an IPEP or in reducing shoulder, knee and ankle injury risk.

TRIAL REGISTRATION NUMBER

NCT05294237.

The Relationship Between Three Sources of Social Support and Physical Activity Level in Middle-Aged and Older Adults

This study examined how different sources of social support from family members (excluding partners), friends, and partners were associated with moderate and vigorous leisure-time physical activity (LTPA) among middle-aged and older adults. This study included married participants aged 45 or older (N = 2,155) from the Midlife in the United States secondary data set. Hierarchical linear regression was performed to investigate the relationship between the three sources of social support and moderate LTPA, and separately, with vigorous LTPA. Partner support (b = 0.19, p < .01), family support (b = -0.19, p < .01), and friend support (b = 0.26, p < .001) were all significantly associated with moderate LTPA. Only social support from friends was associated with vigorous LTPA (b = 0.24, p < .001). Our study emphasizes the significance of social support in influencing LTPA behaviors among middle-aged and older adults. Future programs promoting physical activity should incorporate social support from friends to have the greatest impact.

Clinical and neurophysiological effects of bilateral repetitive transcranial magnetic stimulation and EEG-guided neurofeedback in Parkinson's disease: a randomized, four-arm controlled trial

BACKGROUND

Repetitive Transcranial Magnetic Stimulation (rTMS) and EEG-guided neurofeedback techniques can reduce motor symptoms in Parkinson's disease (PD). However, the effects of their combination are unknown. Our objective was to determine the immediate and short-term effects on motor and non-motor symptoms, and neurophysiological measures, of rTMS and EEG-guided neurofeedback, alone or combined, compared to no intervention, in people with PD.

METHODS

A randomized, single-blinded controlled trial with 4 arms was conducted. Group A received eight bilateral, high-frequency (10 Hz) rTMS sessions over the Primary Motor Cortices; Group B received eight 30-minute EEG-guided neurofeedback sessions focused on reducing average bilateral alpha and beta bands; Group C received a combination of A and B; Group D did not receive any therapy. The primary outcome measure was the UPDRS-III at post-intervention and two weeks later. Secondary outcomes were functional mobility, limits of stability, depression, health-related quality-of-life and cortical silent periods. Treatment effects were obtained by longitudinal analysis of covariance mixed-effects models.

RESULTS

Forty people with PD participated (27 males, age = 63 ± 8.26 years, baseline UPDRS-III = 15.63 ± 6.99 points, H&Y = 1-3). Group C showed the largest effect on motor symptoms, health-related quality-of-life and cortical silent periods, followed by Group A and Group B. Negligible differences between Groups A-C and Group D for functional mobility or limits of stability were found.

CONCLUSIONS

The combination of rTMS and EEG-guided neurofeedback diminished overall motor symptoms and increased quality-of-life, but this was not reflected by changes in functional mobility, postural stability or depression levels.

TRIAL REGISTRATION

NCT04017481.

Advanced CKD detection through optimized metaheuristic modeling in healthcare informatics

Data categorization is a top concern in medical data to predict and detect illnesses; thus, it is applied in modern healthcare informatics. In modern informatics, machine learning and deep learning models have enjoyed great attention for categorizing medical data and improving illness detection. However, the existing techniques, such as features with high dimensionality, computational complexity, and long-term execution duration, raise fundamental problems. This study presents a novel classification model employing metaheuristic methods to maximize efficient positives on Chronic Kidney Disease diagnosis. The medical data is initially massively pre-processed, where the data is purified with various mechanisms, including missing values resolution, data transformation, and the employment of normalization procedures. The focus of such processes is to leverage the handling of the missing values and prepare the data for deep analysis. We adopt the Binary Grey Wolf Optimization method, a reliable subset selection feature using metaheuristics. This operation is aimed at improving illness prediction accuracy. In the classification step, the model adopts the Extreme Learning Machine with hidden nodes through data optimization to predict the presence of CKD. The complete classifier evaluation employs established measures, including recall, specificity, kappa, F-score, and accuracy, in addition to the feature selection. Data related to the study show that the proposed approach records high levels of accuracy, which is better than the existing models.

Early inflammatory markers as prognostic indicators following allogeneic stem cell transplantation

Allogeneic stem cell transplantation is used widely in the treatment of hematopoietic malignancy although graft versus host disease and relapse remain major complications. We measured the serum protein expression of 92 inflammation-related markers from 49 patients at Day 0 (D0) and 154 patients at Day 14 (D14) following transplantation and related values to subsequent clinical outcomes. Low levels of 7 proteins at D0 were linked to GvHD whilst high levels of 7 proteins were associated with relapse. The concentration of 38 proteins increased over 14 days and higher inflammatory response at D14 was strongly correlated with patient age. A marked increment in protein concentration during this period associated with GvHD but reduced risk of disease relapse, indicating a link with alloreactive immunity. In contrast, patients who demonstrated low dynamic elevation of inflammatory markers during the first 14 days were at increased risk of subsequent disease relapse. Multivariate time-to-event analysis revealed that high CCL23 at D14 was associative of AGvHD, CXCL10 with reduced rate of relapse, and high PD-L1 with reduced overall survival. This work identifies a dynamic pattern of inflammatory biomarkers in the very early post-transplantation period and reveals early protein markers that may help to guide patient management.

The structured ambulatory post-stroke care program for outpatient aftercare in patients with ischaemic stroke in Germany (SANO): an open-label, cluster-randomised controlled trial

BACKGROUND

Patients with ischaemic stroke are at risk of recurrent stroke. In this study, we aimed to compare the effect of a structured ambulatory post-stroke care programme versus usual care on recurrent vascular events and death and control of cardiovascular risk factors.

METHODS

We did a prospective, open-label, cluster-randomised controlled trial (SANO) at stroke centres in regions of Germany. A cluster was defined as a region in which acute stroke care is provided by a participating stroke centre. Patients were eligible for participation if they were aged 18 years or older, had no severe disabilities before the index stroke (modified Rankin scale 0-1), had at least one modifiable cardiovascular risk factor, and presented within 14 days of symptom onset of their first ischaemic stroke. The participating regions were randomly assigned (1:1) to the intervention and control group (usual care) by the statistician using block randomisation (block sizes of six), stratified by rural and urban regions. In intervention regions, a cross-sectoral multidisciplinary network was established to provide a 1-year organisational and patient-centred intervention. Due to the type of intervention, masking of participants and study physicians was not possible. Endpoint adjudication was performed by an independent endpoint adjudication committee who were masked to cluster allocation. The primary endpoint was a composite of recurrent stroke, myocardial infarction, and all-cause death within 12 months after baseline assessment, assessed in the modified intention-to-treat (mITT) population, which included all patients who did not withdraw consent and completed the primary endpoint assessment at 12 months. This study was registered with the German Clinical Trials Register, DRKS00015322.

FINDINGS

Between Jan 1, 2019 and Dec 22, 2020, 36 clusters were assessed for eligibility, of which 30 were randomly assigned to the intervention group (n=15 clusters) or control group (n=15 clusters). No clusters dropped out of the study. 1203 (86%) of 1396 enrolled patients in the intervention group and 1283 (92%) of 1395 enrolled patients in the control group were included in the mITT population. The primary endpoint was confirmed in 64 (5·3%) of 1203 patients in the intervention group and 80 (6·2%) of 1283 patients in the control group (unadjusted odds ratio [OR] 0·80 [95% CI 0·49-1·30]; adjusted OR [aOR] 0·95 [95% CI 0·54-1·67]). All-cause deaths occurred in 31 (2·4%) of 1203 patients in the intervention group and 12 (1·0%) of 1283 patients in the control group. The incidence of serious adverse events was higher in the intervention group (266 [23·1%] of 1151) than the control group (106 [9·2%] of 1152). Falls (134 [11·4%] of 1203 patients in the intervention group; 39 [3·3%] of 1152 patients in the control group), hypertensive crisis (55 [4·7%]; 34 [2·8%]), and diagnosis of depression (51 [4·3%]; 13 [1·1%]) were the most frequent adverse events in both groups. No differences were identified in the rate of readmission to hospital between groups.

INTERPRETATION

No differences were identified between patients with ischaemic stroke in the intervention group and control group with regard to the incidence of vascular events 1 year after baseline assessment, despite positive effects with regard to the control of some cardiovascular risk factors. Longer-term effects and other potentially favourable effects on stroke-related sequelae and quality of life require further evaluation.

FUNDING

Innovation Fund of the Federal Joint Committee.

Accounting for expected attrition in the planning of cluster randomized trials for assessing treatment effect heterogeneity

BACKGROUND

Detecting treatment effect heterogeneity is an important objective in cluster randomized trials and implementation research. While sample size procedures for testing the average treatment effect accounting for participant attrition assuming missing completely at random or missing at random have been previously developed, the impact of attrition on the power for detecting heterogeneous treatment effects in cluster randomized trials remains unknown.

METHODS

We provide a sample size formula for testing for a heterogeneous treatment effect assuming the outcome is missing completely at random. We also propose an efficient Monte Carlo sample size procedure for assessing heterogeneous treatment effect assuming covariate-dependent outcome missingness (missing at random). We compare our sample size methods with the direct inflation method that divides the estimated sample size by the mean follow-up rate. We also evaluate our methods through simulation studies and illustrate them with a real-world example.

RESULTS

Simulation results show that our proposed sample size methods under both missing completely at random and missing at random provide sufficient power for assessing heterogeneous treatment effect. The proposed sample size methods lead to more accurate sample size estimates than the direct inflation method when the missingness rate is high (e.g., ≥ 30%). Moreover, sample size estimation under both missing completely at random and missing at random is sensitive to the missingness rate, but not sensitive to the intracluster correlation coefficient among the missingness indicators.

CONCLUSION

Our new sample size methods can assist in planning cluster randomized trials that plan to assess a heterogeneous treatment effect and participant attrition is expected to occur.

Classification of breast cancer recurrence based on imputed data: a simulation study

Several studies have been conducted to classify various real life events but few are in medical fields; particularly about breast recurrence under statistical techniques. To our knowledge, there is no reported comparison of statistical classification accuracy and classifiers' discriminative ability on breast cancer recurrence in presence of imputed missing data. Therefore, this article aims to fill this analysis gap by comparing the performance of binary classifiers (logistic regression, linear and quadratic discriminant analysis) using several datasets resulted from imputation process using various simulation conditions. Our study aids the knowledge about how classifiers' accuracy and discriminative ability in classifying a binary outcome variable are affected by the presence of imputed numerical missing data. We simulated incomplete datasets with 15, 30, 45 and 60% of missingness under Missing At Random (MAR) and Missing Completely At Random (MCAR) mechanisms. Mean imputation, hot deck, k-nearest neighbour, multiple imputations via chained equation, expected-maximisation, and predictive mean matching were used to impute incomplete datasets. For each classifier, correct classification accuracy and area under the Receiver Operating Characteristic (ROC) curves under MAR and MCAR mechanisms were compared. The linear discriminant classifier attained the highest classification accuracy (73.9%) based on mean-imputed data at 45% of missing data under MCAR mechanism. As a classifier, the logistic regression based on predictive mean matching imputed-data yields the greatest areas under ROC curves (0.6418) at 30% missingness while k-nearest neighbour tops the value (0.6428) at 60% of missing data under MCAR mechanism.

Multiple Imputation to Salvage Partial Respondents: Analysis of the Forgotten Joint Score-12 After Total Hip Arthroplasty

BACKGROUND

Missing responses are common when Asian patients complete the Forgotten Joint Score-12 (FJS-12), which is widely used to evaluate total hip arthroplasty (THA). We aimed to provide orthopaedic researchers with a solution for handling missing values in such patient-reported outcome measures (PROMs).

METHODS

Patients who had undergone primary THA between 1998 and 2016 (n = 1,021) were investigated in 2020. The FJS-12 and 9 other PROMs, including questions related to Asian lifestyle activities, were administered. Risk factors for missing FJS-12 items were investigated. Partial respondents were matched with complete respondents; then, in each pair, the items not completed by the partial respondent were deleted from the responses of the complete respondent. Predictive mean matching (PMM) was performed in an attempt to recover the deleted items, using 65 sets of imputation models. After the missing values had been imputed, we explored patient characteristics that affected the FJS-12, using data from all complete and partial respondents.

RESULTS

A total of 652 patients responded to the survey (393 complete and 193 partial respondents). Partial respondents were older, more often female, and less active. Older respondents were more likely to skip items involving the bed, while those who reported a better ability to sit in the seiza style (traditional Japanese floor sitting) were more likely to skip items about chair sitting. The imputed FJS-12 value exhibited excellent reliability (intraclass correlation coefficient for agreement with the true scores, 0.985). FJS-12 values of complete respondents were significantly higher than those of respondents with 4 to 11 missing items (51.6 versus 32.8, p < 0.001). Older age was associated with higher FJS-12 values, which was revealed only via analysis of the multiply imputed data sets (p < 0.001).

CONCLUSIONS

Analysis of only complete FJS-12 responses after THA resulted in a nonresponse bias, preferentially excluding older, female, and less active individuals and those with a traditional floor living style. Multiple imputation could provide a solution to scoring and analyzing PROMs with missing responses by permitting the inclusion of partial respondents.

LEVEL OF EVIDENCE

Therapeutic Level III. See Instructions for Authors for a complete description of levels of evidence.

Machine learning of COVID-19 clinical data identifies population structures with therapeutic potential

Clinical outcomes for patients with COVID-19 are heterogeneous and there is interest in defining subgroups for prognostic modeling and development of treatment algorithms. We obtained 28 demographic and laboratory variables in patients admitted to hospital with COVID-19. These comprised a training cohort (n = 6099) and two validation cohorts during the first and second waves of the pandemic (n = 996; n = 1011). Uniform manifold approximation and projection (UMAP) dimension reduction and Gaussian mixture model (GMM) analysis was used to define patient clusters. 29 clusters were defined in the training cohort and associated with markedly different mortality rates, which were predictive within confirmation datasets. Deconvolution of clinical features within clusters identified unexpected relationships between variables. Integration of large datasets using UMAP-assisted clustering can therefore identify patient subgroups with prognostic information and uncovers unexpected interactions between clinical variables. This application of machine learning represents a powerful approach for delineating disease pathogenesis and potential therapeutic interventions.