Trial sequential analyses of the cumulative meta‐analysis of the effect of NIRS in very preterm infants.

Trial sequential analyses of the cumulative meta‐analysis of the effect of NIRS in very preterm infants (part of the 2025 Cochrane review "Cerebral near-infrared spectroscopy monitoring for prevention of death or neurodevelopmental disability in very preterm infants").Figure created using the Copenhagen Trial Unit's TSA analysis tool https://ctu.dk/tsa/.<br>Trial Sequential Analysis of the cumulative meta‐analysis of the effect of NIRS <b>all-cause mortality prior to discharge</b> in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of mortality with a control group proportion of 16.2%, an alpha of 5%, a beta of 20%, and diversity of 70% is 16,811 participants (vertical red line). The red dashed lines represent the trial sequential monitoring boundaries and the futility boundaries. The green lines represent the naïve monitoring boundaries for the conventional alpha of 5%. The red dashed lines represent the trial sequential monitoring boundaries and the futility boundaries. The solid blue line is the cumulative Z-curve. The cumulated z‐curve (blue curve) does not cross the naïve monitoring boundaries for an alpha of 5% and are far from crossing the trial sequential monitoring boundaries (red lines) for benefit, harm and futility, implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 70%.Trial sequential analysis of the cumulative meta‐analysis of the effect of NIRS on<b> </b><b>All-cause mortality at 2 years of life</b><b> </b>in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of all-cause mortality at 2 years of life in the control group with a proportion of 27.5%, an alpha of 5%, a beta of 20%, and diversity of 0% is 2,588 participants (vertical red line). The green dashed lines represent the naïve monitoring boundaries for the conventional alpha of 5%. The red dashed lines represent the trial sequential monitoring boundaries and the futility boundaries. The solid blue line is the cumulative Z-curve. The cumulated z‐curve (blue curve) do cross the naïve monitoring boundaries for an alpha of 5% but are far from crossing the trial sequential monitoring boundaries (red lines) for benefit, harm and futility, implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 0%.Trial Sequential Analysis of the cumulative meta‐analysis of the effect of NIRS <b>all-cause mortality at the longest follow-up</b> in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of mortality with a control group proportion of 16.3%, an alpha of 5%, a beta of 20%, and diversity of 75.4% is 20,145 participants (vertical red line). The green dashed lines represent the naïve monitoring boundaries for the conventional alpha of 5%. The red dashed lines represent the trial sequential monitoring boundaries and the futility boundaries. The solid blue line is the cumulative Z-curve. The cumulated z‐curve (blue curve) does cross the naïve monitoring boundaries for an alpha of 5% after the first trial but is far from crossing the trial sequential monitoring boundaries (red lines) for benefit, harm and futility, implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 75.4%.Trial Sequential Analysis of the cumulative meta‐analysis of the effect of NIRS on <b>Major neurodevelopmental disability in children aged 18 to 24 months</b> in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of Major neurodevelopmental disability in children aged 18 to 24 months in the control group with a proportion of 12%, an alpha of 5%, a beta of 20%, and diversity of 0% is 7,030 patients (not shown in the graph). The green dashed lines represent the naïve monitoring boundaries for the conventional alpha of 5%. There was too little information use to calculate trial sequential monitoring boundaries. The cumulated z‐curve (blue curve) does not cross the naïve monitoring boundary for an alpha of 5% (green curve), implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 0%.Trial sequential analysis of the cumulative meta‐analysis of the effect of NIRS <b>major brain injury prior discharge</b> in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of major brain injury with a control group proportion of 18.5%, an alpha of 5% and a beta of 20% is 9596 participants (vertical red line). The green dashed lines represent the naïve monitoring boundaries for the conventional alpha of 5%. The red dashed lines represent the trial sequential monitoring boundaries and the futility boundaries. The solid blue line is the cumulative Z-curve. The cumulated z‐curve (blue curve) does not cross the trial sequential monitoring boundary (red curve), implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 56%.Trial sequential analysis of the cumulative meta‐analysis of the effect of NIRS on<b> </b><b>intraventricular haemorrhage grade III or IV until hospital discharge</b><b> </b>in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of intraventricular haemorrhage grade III or IV with a control group proportion of 14.1%, an alpha of 5% and a beta of 20% and nil diversity is 5856 participants (vertical red line). The green dashed lines represent the naïve monitoring boundaries for the conventional alpha of 5%. The red dashed lines represent the trial sequential monitoring boundaries and the futility boundaries. The solid blue line is the cumulative Z-curve. The cumulated z‐curve (blue curve) does not cross the trial sequential monitoring boundary (red curve), implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 0%.Trial Sequential Analysis of the cumulative meta‐analysis of the effect of NIRS on <b>cystic periventricular leukomalacia until hospital discharge</b><b> </b>in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of cystic periventricular leukomalacia in the control group with a proportion of 3.2%, an alpha of 5%, a beta of 20%, and diversity of 62% is 74,996 patients (not shown in the graph). The green dashed lines represent the naïve monitoring boundaries for the conventional alpha of 5%. There was too little information use to calculate trial sequential monitoring boundaries. The cumulated z‐curve (blue curve) does not cross the naïve monitoring boundary for an alpha of 5% (green curve), implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 62%.Trial sequential analysis of the cumulative meta‐analysis of the effect of NIRS on<b> </b><b>chronic lung disease at 36 weeks of gestational age</b><b> </b>in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of chronic lung disease in the control group with a proportion of 37%, an alpha of 5%, a beta of 20%, and diversity of 78% is 7,581 participants (vertical red line). The green dashed lines represent the naïve monitoring boundaries for the conventional alpha of 5%. The red dashed lines represent the trial sequential monitoring boundaries and the futility boundaries. The solid blue line is the cumulative Z-curve. The cumulated z‐curve (blue curve) does not cross the naïve monitoring boundaries for an alpha of 5% and are far from crossing the trial sequential monitoring boundaries (red lines) for benefit, harm and futility, implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 78%.Trial sequential analysis of the cumulative meta‐analysis of the effect of NIRS on<b> </b><b>proven necrotizing enterocolitis prior discharge</b><b> </b>in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of necrotizing enterocolitis with a control group proportion of 9.4%, an alpha of 5%, a beta of 20%, and nil diversity is 9,210 participants (vertical red line). The green lines represent the naïve monitoring boundaries for the conventional alpha of 5%. The red dashed lines represent the trial sequential monitoring boundaries and the futility boundaries. The solid blue line is the cumulative Z-curve. The cumulated z‐curve (blue curve) does cross the naïve monitoring boundaries for an alpha of 5% after the first trial but is far from crossing the trial sequential monitoring boundaries (red lines) for benefit, harm and futility, implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 0%.Trial sequential analysis of the cumulative meta‐analysis of the effect of NIRS on<b> </b><b>retinopathy of prematurity prior to discharge prior discharge stage ≥3</b><b> </b>in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of retinopathy of prematurity with a control group proportion of 8.6%, an alpha of 5%, a beta of 20%, and nil diversity is 10,146 participants (vertical red line). The green lines represent the naïve monitoring boundaries for the conventional alpha of 5%. The red dashed lines represent the trial sequential monitoring boundaries and the futility boundaries. The solid blue line is the cumulative Z-curve. The cumulated z‐curve (blue curve) does cross the naïve monitoring boundaries for an alpha of 5% after the first trial but is far from crossing the trial sequential monitoring boundaries (red lines) for benefit, harm and futility, implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 0%.Trial Sequential Analysis of the cumulative meta‐analysis of the effect of NIRS on <b>Abnormal fidgety movements</b><b> </b>in very preterm infants. The diversity-adjusted required information size to demonstrate or reject a 20% relative risk reduction (a priori estimate) of fidgety movements in the control group with a proportion of 5.6%, an alpha of 5%, a beta of 20%, and diversity of 0% is 16,036 patients (not shown in the graph). The green dashed lines represent the naïve monitoring boundaries for the conventional alpha of 5%. The red dashed lines represent the trial sequential monitoring boundaries and the futility boundaries. The solid blue line is the cumulative Z-curve. The cumulated z‐curve (blue curve) does not cross the naïve monitoring boundary for an alpha of 5% (green curve), implying that there is no firm evidence for an effect of 20% risk ratio reduction (RRR) when the cumulative meta‐analysis is adjusted for multiple testing on accumulating data with a diversity of 0%.<br><br>