Three simple rules to ensure reasonably credible subgroup analyses

Burke, James F.; Sussman, Jeremy B.; Kent, David M.; Hayward, Rodney A.

doi:10.1136/bmj.h5651

Cited by 184 publications

(157 citation statements)

References 21 publications

Supporting

Mentioning

150

Contrasting

Order By: Relevance

“…identical pathogens) might help overcome some of the challenges faced in sepsis research (Cohen et al, 2015). Here we applied the new Sepsis-3 definition to both the discovery GWAS and the sequencing study and we wanted to avoid multiplicity issues that arise if subgroups are defined post-hoc (Sun et al, 2014, Burke et al, 2015). However, the validation GWA studies addressed a slightly different phenotype spectrum (e.g.…”

Section: Discussionmentioning

confidence: 99%

Genetic Factors of the Disease Course after Sepsis: A Genome-Wide Study for 28 Day Mortality

et al. 2016

View full text Add to dashboard Cite

Sepsis is the dysregulated host response to an infection which leads to life-threatening organ dysfunction that varies by host genomic factors. We conducted a genome-wide association study (GWAS) in 740 adult septic patients and focused on 28 day mortality as outcome. Variants with suggestive evidence for an association (p ≤ 10− 5) were validated in two additional GWA studies (n = 3470) and gene coding regions related to the variants were assessed in an independent exome sequencing study (n = 74).In the discovery GWAS, we identified 243 autosomal variants which clustered in 14 loci (p ≤ 10− 5). The best association signal (rs117983287; p = 8.16 × 10− 8) was observed for a missense variant located at chromosome 9q21.2 in the VPS13A gene. VPS13A was further supported by additional GWAS (p = 0.03) and sequencing data (p = 0.04). Furthermore, CRISPLD2 (p = 5.99 × 10− 6) and a region on chromosome 13q21.33 (p = 3.34 × 10− 7) were supported by both our data and external biological evidence.We found 14 loci with suggestive evidence for an association with 28 day mortality and found supportive, converging evidence for three of them in independent data sets. Elucidating the underlying biological mechanisms of VPS13A, CRISPLD2, and the chromosome 13 locus should be a focus of future research activities.

show abstract

Section: Discussionmentioning

confidence: 99%

Genetic Factors of the Disease Course after Sepsis: A Genome-Wide Study for 28 Day Mortality

et al. 2016

View full text Add to dashboard Cite

show abstract

“…30 Methodological recommendations for study design, analysis and interpretation of such subgroup analyses are available including the need for clear terminology. 7, 41, 59, 65, 69 In this study we attempt to distinguish between variables that demonstrated treatment effect modification for stratified care outcomes from those that were predictive of patient outcomes regardless of treatment. We use treatment effect modifiers are used for variables measured at baseline that demonstrated an interaction with the stratified care treatment outcomes (ie.…”

Section: Introductionmentioning

confidence: 99%

Identifying Treatment Effect Modifiers in the STarT Back Trial: A Secondary Analysis

Beneciuk

Hill

Campbell

et al. 2017

The Journal of Pain

View full text Add to dashboard Cite

Identification of patient characteristics influencing treatment outcomes is a top low back pain (LBP) research priority. Results from the STarT Back Trial support the effectiveness of prognostic stratified care for LBP compared to current best care, however patient characteristics associated with treatment response have not yet been explored. The purpose of this secondary analysis was to identify treatment-effect modifiers within the STarT Back Trial at 4 months follow-up (n=688). Treatment response was dichotomized using back-specific physical disability measured by the Roland-Morris Disability Questionnaire (≥7). Candidate modifiers were identified using previous literature and evaluated using logistic regression with statistical interaction terms to provide preliminary evidence of treatment-effect modification. Socioeconomic status (SES) was identified as an effect modifier for disability outcomes (OR = 1.71, P=.028). High SES patients receiving prognostic stratified care were 2.5 times less likely to have a poor outcome compared to low SES patients receiving best current care (OR = 0.40, P=.006). Education level (OR = 1.33, P=.109) and number of pain medications (OR = 0.64, P=.140) met our criteria for effect modification with weaker evidence (0.20>P≥0.05). These findings provide preliminary evidence for SES, education, and number of pain medications as treatment-effect modifiers of prognostic stratified care delivered in the STarT Back Trial. Perspective This analysis provides preliminary exploratory findings about the characteristics of patients who might least likely benefit from targeted treatment using prognostic stratified care for low back pain.

show abstract

“…This study adhered to criteria for a best practice analytical approach to subgroup analyses67 with the exception that secondary outcomes (ie, resilience protective factor outcomes) were examined in subgroups, and the proposed subgroup analyses were not published a priori. In terms of further limitations, the number of comparisons undertaken within this study may have increased the risk of type 1 error and led to the sole significant result; however, the use of a formal statistical interaction test, rather than examination of intervention effect within individual subgroups, reduces this risk 76. In the overall study, the proportion of enrolled students completing both the baseline and follow-up surveys was below 50%, and while typical for school-based research,77 it may limit the generalisability of the study results.…”

Section: Discussionmentioning

confidence: 98%

Differential intervention effectiveness of a universal school-based resilience intervention in reducing adolescent substance use within student subgroups: exploratory assessment within a cluster-randomised controlled trial

et al. 2018

View full text Add to dashboard Cite

ObjectivesInterventions addressing the individual and environmental protective factors of adolescents are suggested to have potential for reducing adolescent substance use. While universally delivered school-based substance use prevention interventions are common, previous studies have suggested variable effectiveness by subgroups of students. An exploratory study was undertaken to examine the differential effectiveness of a universal school-based resilience intervention on adolescent substance use and protective factors according to their sociodemographic and previous substance use.DesignSecondary analysis of data from a cluster-randomised controlled trial.Setting32 Australian secondary schools.ParticipantsCohort of grade 7 students (n=3155) followed up in grade 10 (aged 15–16 years; 2014; n=2105).InterventionThree-year universal school-based intervention implemented by school staff that targeted a range of student resilience protective factors (2012–2014).MeasurementsPrimary outcomes included: tobacco (recent, number of cigarettes) and alcohol (recent, ‘risk’ and number of drinks) use, and secondary outcomes included: marijuana (recent) and other illicit substance (recent) use, and aggregate individual and environmental protective factor scores. Generalised and linear mixed models examined interactions between treatment and student subgroups (gender; socioeconomic disadvantage (low/high); geographic location (major city/inner regional/outer regional-remote); and previous substance use (non-user/user)) at follow-up (36 models).ResultsAnalysis of student follow-up data showed no differential intervention effect for any substance use or protective factor outcome for any subgroup, with the exception of one differential effect found by socioeconomic status for the outcome of mean number of cigarettes smoked by recent smokers (p=0.003). There was no evidence of an intervention effect within the low (mean difference (MD) −12.89, 95% CI −26.00 to 0.23) or high (MD 16.36, 95% CI −1.03 to 33.76) socioeconomic subgroups.ConclusionsNo evidence of an intervention effect on substance use and protective factors was found according to student subgroups defined by sociodemographic characteristics or previous substance use.Trial registration numberACTRN12611000606987.

show abstract

Three simple rules to ensure reasonably credible subgroup analyses

Cited by 184 publications

References 21 publications

Genetic Factors of the Disease Course after Sepsis: A Genome-Wide Study for 28 Day Mortality

Genetic Factors of the Disease Course after Sepsis: A Genome-Wide Study for 28 Day Mortality

Identifying Treatment Effect Modifiers in the STarT Back Trial: A Secondary Analysis

Differential intervention effectiveness of a universal school-based resilience intervention in reducing adolescent substance use within student subgroups: exploratory assessment within a cluster-randomised controlled trial

Contact Info

Product

Resources

About