This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
FACT was a randomized, phase 3, quadruple-masked, parallel-group prevention trial evaluating whether folic acid 4 mg throughout pregnancy reduced the occurrence of preeclampsia compared with placebo.
| Feature | FACT |
|---|---|
| Brief title | High Dose Folic Acid Supplementation Throughout Pregnancy for Preeclampsia Prevention |
| Phase | Phase 3 |
| Status | Completed |
| Therapeutic area | Women's Health & Pregnancy |
| Conditions | Pregnancy Complications; Preeclampsia |
| Design | Randomized, parallel-group, quadruple-masked |
| Primary purpose | Prevention |
| Enrollment | 2464 |
| Interventions | Folic Acid 4 mg; Placebo |
| Lead sponsor | Ottawa Hospital Research Institute |
| Start | 2011-04 |
| Primary completion | 2016-07 |
| ClinicalTrials.gov | NCT01355159 |
2. Clinical Question
The central question was whether supplementation with folic acid 4 mg throughout pregnancy changed the risk of preeclampsia compared with placebo.
Population
Participants enrolled in a phase 3 prevention trial concerning pregnancy complications and preeclampsia.
Intervention
Folic Acid 4 mg.
Comparator
Placebo.
Primary question
Does folic acid 4 mg reduce the occurrence of preeclampsia relative to placebo?
3. Trial Design
Folic Acid 4 mg
- Folic Acid 4 mg
- Pregnancy-focused prevention strategy
- Primary comparison against placebo
Placebo
- Placebo
- Parallel randomized comparison
- Reference group for risk-ratio and mean-difference analyses
The combination of randomization and quadruple masking is statistically important because it reduces opportunities for systematic differences in treatment assignment and knowledge of assignment to influence outcome assessment or participant and study-team behavior.
4. Primary Endpoint
| Endpoint | Registry definition | Time frame | Analysis |
|---|---|---|---|
| Preeclampsia | PE is defined as diastolic blood pressure ≥90 mmHg on two occasions ≥4 hours apart and proteinuria developed in women greater than 20+0 weeks of gestation. Proteinuria is defined as urinary protein ≥300mg in 24 hour urine collection OR, in the absence of 24 hour collection, ≥2+ dipstick proteinuria, OR random protein-creatinine ratio ≥30mg protein/mmol. OR HELLP syndrome defined as haemolysis, Serum LDH ≥ 600U/L, Serum AST ≥ 70U/L, and platelet count <100 x109/L OR superimposed pre-eclampsia, defined as history of pre-existing hypertension with new proteinuria. | Participants will be followed from 20+0 weeks of gestational age until 42 days postpartum (after delivery) | Chi-squared test; risk ratio |
The endpoint is binary: each participant is classified according to whether preeclampsia occurred during the specified observation period. This makes a comparison of event risk between randomized groups a natural primary analysis.
5. Primary Result: Preeclampsia
Risk ratio for preeclampsia
95% CI: 0.90–1.34 · P = 0.37
Analysis: two-sided chi-squared test; superiority hypothesis
The registry reports the primary comparison as a risk ratio of 1.10, comparing the folic acid group with placebo. The corresponding 95% confidence interval is 0.90–1.34, and the reported two-sided P-value is 0.37.
A risk ratio of 1.10 means that the estimated risk of preeclampsia in the folic acid group was 1.10 times the corresponding risk in the placebo group. Expressed as a relative measure, that estimate is consistent with a 10% higher estimated risk in the folic acid group, although the estimate itself is not a statement that every participant experienced a 10% change in risk.
The 95% CI of 0.90–1.34 is important because it describes the statistical uncertainty around the estimated risk ratio. It includes 1.00, the value corresponding to equal risks between groups. Thus the interval encompasses both a possible relative reduction and a possible relative increase in risk.
The P-value of 0.37 is a measure of compatibility between the observed data and the null hypothesis under the specified statistical test. It is not a measure of the size, clinical importance, or probability of the treatment effect. A P-value should therefore be read together with the risk ratio and confidence interval rather than used as a substitute for them.
This is a binary endpoint analysed with a chi-squared test rather than a time-to-event endpoint, so proportional-hazards assumptions and Cox-model interpretation are not relevant to this primary analysis. The randomized comparison is also distinct from an observational comparison: randomization provides the principal basis for interpreting the between-group difference causally.
6. Statistical Methodology
Chi-squared testing for binary outcomes
The registry reports chi-squared testing for the binary maternal and infant outcomes. The chi-squared framework evaluates whether the observed distribution of participants across outcome categories differs between the randomized groups.
For a binary endpoint such as preeclampsia, the analysis compares the observed event pattern across the folic acid and placebo groups. The reported effect measure then expresses the magnitude of the between-group difference as a risk ratio.
Risk ratio
For binary outcomes, the registry reports risk ratios. Conceptually, the risk ratio compares the probability of an event in the intervention group with the probability of the same event in the comparator group.
An RR of 1 indicates equal observed risk. Values below 1 indicate lower estimated risk in the intervention group, while values above 1 indicate higher estimated risk in the intervention group.
Two-sided confidence intervals
The posted analyses report 95% two-sided confidence intervals. For a risk ratio, the value 1 is the reference point. Whether the interval includes 1 therefore has a direct relationship to the conventional null comparison for a two-sided test.
t-tests for continuous outcomes
Continuous outcomes were analysed using two-sided t-tests. The registry reports mean differences as the corresponding effect measure for outcomes such as antenatal inpatient length of stay and length of stay in a high-level neonatal care unit.
The sign of the mean difference depends on the direction of the subtraction. A negative value indicates a lower estimated mean in the folic acid group under this convention; a positive value indicates a higher estimated mean.
Missing outcome data
Several secondary infant and maternal outcomes explicitly state that not all participants had outcome data available. Some infant outcomes also state that participants with missing data were excluded from the corresponding analysis. This means the effective analysis population for those endpoints can differ from the overall randomized enrollment.
Multiple imputation / missing data
Multiple imputation / missing data is identified in the analysis text for several infant outcomes. The presence of missing data is particularly important when interpreting estimates for relatively uncommon outcomes: excluding participants with missing observations can reduce the information available for analysis and can affect precision and, depending on the missingness mechanism, potentially affect the estimate.
7. Secondary Endpoint Results
The registry reports statistical analyses for a broad set of secondary maternal and infant outcomes. The binary outcomes use risk ratios with two-sided 95% confidence intervals and chi-squared testing. The continuous outcomes use mean differences with two-sided 95% confidence intervals and t-tests.
| Secondary endpoint | Time frame | Effect | 95% CI | P-value |
|---|---|---|---|---|
| Spontaneous Abortion | Randomization until 20+0 weeks of gestation | RR 1.29 | 0.74–2.28 | 0.37 |
| Placenta Abruption | 20+0 weeks of gestation until delivery | RR 0.64 | 0.31–1.31 | 0.21 |
| Premature Rupture of Membranes | Randomization at 8-16 weeks' completed gestation until onset of labor | RR 0.97 | 0.82–1.15 | 0.71 |
| Preterm Birth | 20+0 weeks to 36+6 weeks of gestation | RR 0.99 | 0.86–1.13 | 0.87 |
| HELLP (Hemolysis, Elevated Liver Enzyme Levels & Low Platelet Count) | 20+0 weeks of gestation until delivery | RR 1.21 | 0.37–3.96 | 0.75 |
| Severe Preeclampsia | 20+0 weeks of gestation until delivery | RR 1.52 | 0.81–2.84 | 0.19 |
| Antenatal Inpatient Length of Stay | Randomization at 8-16 weeks' completed gestation until admission for delivery | Mean difference 0.34 days | -0.96 to 1.63 | 0.61 |
| Stillbirth | 20+0 weeks of gestation up to delivery | RR 0.60 | 0.30–1.19 | 0.14 |
| Intrauterine Growth Restriction (<3rd Percentile) | 20+0 weeks of gestation until delivery | RR 0.76 | 0.41–1.39 | 0.37 |
| Intrauterine Growth Restriction (<10th Percentile) | 20+0 weeks of gestation until delivery | RR 1.03 | 0.81–1.30 | 0.82 |
| Neonatal Death | Birth until 28 days of life | RR 0.87 | 0.31–2.44 | 0.79 |
| Perinatal Mortality | 20+0 weeks of gestation until 28 days of life | RR 0.63 | 0.37–1.05 | 0.07 |
| Retinopathy of Prematurity | Duration of hospital stay, or up to 6 weeks | RR 1.20 | 0.54–2.66 | 0.65 |
| Early Onset Sepsis | First 48 hours of life | RR 0.34 | 0.09–1.23 | 0.10 |
| Necrotising Enterocolitis | Duration of hospital stay, or up to 6 weeks | RR 2.04 | 0.49–8.57 | 0.33 |
| Intraventricular Hemorrhage (IVH) | Duration of hospital stay, or up to 6 weeks | RR 0.97 | 0.47–2.00 | 0.94 |
| Ventilation | Duration of hospital stay, or up to 6 weeks | RR 1.61 | 0.97–2.66 | 0.06 |
| Need for Oxygen at 28 Days | 28 days after birth | RR 2.37 | 0.61–9.14 | 0.21 |
| Composite Severe Adverse Fetal/Neonatal Outcome | Component-specific time frames, up to 6 weeks after birth | RR 1.20 | 0.80–1.80 | 0.38 |
| Length of Stay in 'High Level' Neonatal Care Unit | Duration of hospital stay, or up to 6 weeks | Mean difference -1.60 days | -5.84 to 2.64 | 046 |
| Neonatal Death | 28 days after birth | RR 0.87 | 0.31–2.44 | 0.79 |
| Periventricular Leukomalacia | 28 days after birth | RR 2.00 | 0.37–10.92 | 0.42 |
The registry reports 23 statistical analyses overall and identifies chi-squared tests and t-tests as the principal the statistical methods. The repeated neonatal death analysis appears twice in the posted statistical analyses with the same estimate, confidence interval, and P-value but different wording of the time-frame field.
8. How to Interpret the Secondary Risk Ratios
The secondary estimates span both sides of the null value of 1.00. For example, the reported risk ratio for spontaneous abortion is 1.29, whereas the reported risk ratio for placenta abruption is 0.64. These point estimates describe the direction and magnitude of the observed relative comparison, but the corresponding confidence intervals determine how precisely those estimates are known.
RR below 1
An RR below 1 indicates a lower estimated event risk in the folic acid group than in the placebo group for that endpoint. For example, stillbirth has an RR of 0.60.
RR above 1
An RR above 1 indicates a higher estimated event risk in the folic acid group than in the placebo group for that endpoint. For example, necrotising enterocolitis has an RR of 2.04.
Wide confidence intervals
Rare outcomes can produce imprecise estimates. Necrotising enterocolitis has a 95% CI of 0.49–8.57, illustrating substantial uncertainty around the point estimate.
Confidence interval crossing 1
When a two-sided 95% CI for a risk ratio includes 1.00, the interval includes equal risk as one statistically compatible value under the stated framework.
9. Continuous Secondary Outcomes
Antenatal Inpatient Length of Stay
Mean difference
95% CI: -0.96 to 1.63 · P = 0.61
Two-sided t-test
The positive mean difference indicates a higher estimated mean value in the folic acid group when the comparison is expressed as intervention minus placebo. The 95% confidence interval extends from a negative value to a positive value, so the statistical uncertainty includes both directions of difference.
Length of Stay in 'High Level' Neonatal Care Unit
Mean difference
95% CI: -5.84 to 2.64 · P = 046
Two-sided t-test
The negative point estimate indicates a lower estimated mean length of stay in the folic acid group under the reported comparison convention. The confidence interval is wide relative to the point estimate and crosses zero, indicating substantial uncertainty about the direction and magnitude of the mean difference.
10. Infant Outcomes and the Analysis Population
Several infant outcomes use a common analysis-population description: a total of 2738 infants were born to women recruited into the trial, with 1364 in the folic acid group and 1374 in the placebo group. The registry identifies these as infant outcomes rather than participant-level maternal outcomes.
This distinction affects how the statistical results should be read. A risk ratio for an infant outcome is not automatically interchangeable with a risk ratio for a maternal outcome, even when both use the same chi-squared framework.
11. Missing Data and Multiple Imputation
Missing-data considerations appear explicitly in the analysis descriptions for several infant outcomes, including intrauterine growth restriction, neonatal death, perinatal mortality, retinopathy of prematurity, early onset sepsis, necrotising enterocolitis, intraventricular hemorrhage, ventilation, need for oxygen at 28 days, the composite severe adverse fetal/neonatal outcome, length of stay in the high-level neonatal care unit, and periventricular leukomalacia.
For several of these endpoints, the registry states that some participants had missing data and therefore were excluded from the analysis. The analysis text also identifies multiple imputation / missing data as an analysis concept.
If outcome availability differs between randomized groups or is related to outcome risk, simply analysing observed cases can produce an estimate that differs from the estimate that would have been obtained with complete outcome information. The appropriate interpretation therefore depends on the missing-data mechanism and the method used to address it.
The registry descriptions distinguish between analyses based on available outcome information and analyses in which multiple imputation / missing data is identified as an analysis concept. Those distinctions should be preserved rather than treating all secondary endpoints as if they had identical data completeness.
12. Statistical Methods Explained
Why was a chi-squared test used for preeclampsia?
Preeclampsia is a binary outcome: a participant either meets the registry definition during the specified follow-up period or does not. A chi-squared test is therefore a natural method for comparing the distribution of the binary outcome between two randomized groups. The effect size is separately expressed as a risk ratio.
What does a risk ratio of 1.10 mean?
An RR of 1.10 means that the estimated risk in the folic acid group is 1.10 times the estimated risk in the placebo group. It is a relative measure, not an absolute percentage-point difference. To understand absolute impact, the underlying event risks would also be needed.
Why does the confidence interval matter more than the point estimate alone?
The point estimate is only one estimate of the treatment effect. The 95% confidence interval describes the uncertainty around that estimate under the statistical framework. For the primary endpoint, the interval of 0.90–1.34 is considerably more informative than the value 1.10 alone because it shows that both lower and higher relative risks remain within the interval.
Why does a P-value not measure effect size?
A P-value summarizes the evidence against a specified null hypothesis under a statistical test. It does not tell you how large or clinically important the effect is. A risk ratio, mean difference, and confidence interval are required to describe magnitude and precision.
Why use a t-test for length of stay?
Length of stay is a continuous outcome, unlike a binary endpoint such as preeclampsia or neonatal death. The registry reports two-sided t-tests and mean differences for the continuous length-of-stay outcomes, providing an estimate of the difference in average values between the randomized groups.
Why do missing infant outcomes require additional caution?
Some infant outcomes do not have complete information for every participant or infant. When observations are missing, the effective sample contributing to the analysis can differ from the randomized enrollment. Missingness can therefore affect both precision and the validity of the estimated treatment comparison, depending on why observations are missing and how they are handled.
Why should multiple secondary P-values be interpreted separately from the primary analysis?
The trial has one registered primary endpoint and numerous secondary outcomes. Each additional hypothesis test provides another opportunity for a nominally small P-value to occur by chance when many comparisons are examined. Without an explicit multiplicity strategy, secondary P-values should be distinguished from the primary confirmatory comparison rather than interpreted as a collection of independent definitive findings.
13. Multiplicity and the Secondary Endpoint Set
The registry reports one primary endpoint and a large number of secondary outcomes. The statistical analyses include 23 posted analyses, covering binary and continuous endpoints. This creates an important distinction between the primary hypothesis and the many secondary comparisons.
| Analysis layer | Role | Statistical interpretation |
|---|---|---|
| Preeclampsia | Primary endpoint | Primary superiority comparison using a chi-squared test and risk ratio. |
| Maternal secondary outcomes | Secondary | Supportive comparisons using chi-squared tests or t-tests depending on endpoint type. |
| Infant outcomes | Secondary | Binary and continuous analyses with explicit attention to infant-level populations and missing data. |
| Safety outcomes | Safety | Serious adverse events summarized by randomized arm. |
The presence of a P-value below a conventional threshold in an individual secondary analysis does not, by itself, establish that the endpoint was a prespecified confirmatory success criterion. The statistical role of each endpoint matters.
14. Safety Results
The registry reports serious adverse events by randomized arm. The counts are presented as affected participants over participants at risk.
| Arm | Serious adverse events | Affected / at risk |
|---|---|---|
| Folic Acid 4 mg | Serious adverse events | 227 / 1227 |
| Placebo | Serious adverse events | 195 / 1236 |
These figures describe the reported serious-adverse-event burden within each arm. They should not be converted into an overall safety conclusion using only the raw counts because the relevant denominator, event definition, timing, and distribution of individual adverse events all matter.
The registry provides serious-adverse-event counts by arm but does not provide a formal effect estimate for this safety summary in the statistical-analysis record presented here. The appropriate descriptive comparison is therefore the reported number affected relative to the reported number at risk in each arm.
15. Endpoint Timing and Estimand Structure
One of the most important statistical features of FACT is that the outcomes are measured over different biological and clinical periods. The primary preeclampsia endpoint extends from 20+0 weeks of gestational age through 42 days postpartum, while several maternal outcomes end at delivery and infant outcomes extend into the neonatal period.
Preeclampsia
Participants are followed from 20+0 weeks of gestational age until 42 days postpartum.
Pregnancy through delivery
Outcomes such as placenta abruption, HELLP, and severe preeclampsia use follow-up through delivery.
Delivery and early neonatal period
Stillbirth and growth-related outcomes are assessed through delivery, while neonatal outcomes extend after birth.
Up to 28 days or 6 weeks
Several infant outcomes are followed for 28 days, while others use hospital stay or up to 6 weeks.
These different time frames mean that the secondary endpoints should not be interpreted as if they represented a single common follow-up period. The endpoint definition determines the observation window and therefore the population of events that can contribute to that particular analysis.
16. Primary vs Secondary Analysis: A Statistical Reading
Primary endpoint
Preeclampsia was analysed with a chi-squared test and reported as RR 1.10 with a 95% CI of 0.90–1.34 and P = 0.37.
Secondary endpoints
Multiple maternal and infant outcomes were analysed using the same broad categorical or continuous-data frameworks, with risk ratios or mean differences.
Rare outcomes
Several neonatal outcomes have wide confidence intervals, illustrating limited precision for relatively uncommon events.
Missing data
Several infant analyses explicitly identify missing outcome information and exclusions, making endpoint-specific analysis populations important.
17. Limitations
- Complex diagnosis: the diagnosis of preeclampsia is complex because of its heterogeneous aetiology.
- Diagnostic criteria: the registry states that the criteria for preeclampsia remained consistent with NICE guidelines, while revisions occurred in other settings. It notes that additional women in the study might therefore have had preeclampsia.
- Missing secondary outcomes: not all participants had secondary outcome data for several analyses, and some infant analyses excluded participants with missing data.
- Rare outcomes: several neonatal outcomes have wide confidence intervals, limiting precision around their risk-ratio estimates.
- Multiple comparisons: the large secondary endpoint set creates a distinction between the primary hypothesis and individual secondary P-values.
- Different follow-up periods: maternal and infant endpoints use different observation windows, so estimates should be interpreted according to their endpoint-specific time frames.
- Infant-level outcomes: several analyses concern infants born to women recruited into the trial, so the analysis unit and outcome level should be kept distinct from maternal endpoints.
- Safety denominator: serious adverse-event counts are reported as affected participants over participants at risk; the registry summary does not provide a corresponding formal effect estimate for this safety comparison.
18. What the Primary Risk Ratio Does — and Does Not — Mean
The primary risk ratio of 1.10 represents the estimated relative risk of preeclampsia in the folic acid group compared with placebo. Because the estimate is above 1, its point estimate corresponds to a higher estimated risk in the intervention group.
The 95% CI of 0.90–1.34 indicates that the point estimate is not precise enough to exclude relative risks below 1 or above 1 under the stated confidence-interval framework. It is therefore inappropriate to interpret 1.10 as a fixed or exact treatment effect.
The RR of 1.10 does not mean that every participant experienced a 10% increase in risk. It is a group-level relative measure. It also does not provide the absolute risk difference between treatment groups.
The reported P-value of 0.37 does not measure the size of the treatment effect and should not be interpreted as the probability that the null hypothesis is true. It summarizes the evidence against the specified null comparison under the two-sided chi-squared test.
19. Why This Trial Matters Statistically
FACT is a useful teaching case because it illustrates how a randomized prevention trial can combine a binary primary endpoint with a large set of maternal and infant secondary outcomes, while requiring careful attention to endpoint timing, analysis populations, missing data, and multiplicity.
| Concept | How it appears in FACT |
|---|---|
| Randomization | Randomized parallel-group allocation of 2464 participants. |
| Blinding | Quadruple-masked trial design. |
| Binary endpoint analysis | Preeclampsia and numerous secondary outcomes analysed with chi-squared tests. |
| Risk ratio | Primary and secondary binary effects reported as risk ratios. |
| Confidence interval | Two-sided 95% intervals quantify uncertainty around risk ratios and mean differences. |
| Continuous outcomes | Length-of-stay endpoints analysed using two-sided t-tests and mean differences. |
| Missing data | Several secondary analyses explicitly identify missing outcome information and exclusions. |
| Multiple imputation | Identified as an analysis concept for several infant outcomes. |
| Multiplicity | One primary endpoint is accompanied by numerous secondary comparisons. |
| Endpoint timing | Maternal and infant outcomes use distinct follow-up windows. |
| Safety analysis | Serious adverse events reported by randomized arm. |
20. Planned Analysis vs Reported Results
For the primary endpoint, the registry provides both the prespecified endpoint definition and a posted formal statistical analysis. That analysis uses a chi-squared test and reports a risk ratio with a two-sided 95% confidence interval and P-value.
The secondary analyses similarly provide formal posted comparisons for binary and continuous outcomes. Their interpretation requires additional attention to endpoint-specific follow-up periods, available outcome data, and the distinction between maternal and infant outcomes.
21. Overall Statistical Synthesis
The statistical structure of FACT is straightforward at the primary-endpoint level: randomization creates two comparison groups, preeclampsia is a binary endpoint, and the registry reports a chi-squared test with a risk ratio as the effect measure. The primary estimate is RR 1.10, with a 95% CI of 0.90–1.34 and P = 0.37.
The secondary analysis set is substantially broader. Most secondary endpoints are binary and use the same chi-squared/risk-ratio framework, while length-of-stay outcomes are continuous and use t-tests with mean differences. Several infant endpoints have incomplete outcome information, and the registry explicitly identifies missing-data considerations and multiple imputation for a number of those analyses.
Statistically, the most important distinction is between the primary comparison and the large collection of secondary endpoints. The primary result provides the principal formal test of the registered question. Secondary estimates provide additional information about maternal, fetal, and neonatal outcomes, but their interpretation should account for multiplicity, endpoint-specific data completeness, and the precision of each estimate.
22. Related Tutorials
Learn more about the methods used in this trial:
23. Related Calculators
24. Sources
- ClinicalTrials.gov: NCT01355159.
- PubMed: PMID 37073421.
- PubMed: PMID 33851221.
- PubMed: PMID 32067533.
- PubMed: PMID 30209050.
- PubMed: PMID 24349782.
Continue through the Clinical Biostats clinical-trial resources
Explore statistical tutorials, calculators, and additional clinical trial analyses covering study design, endpoints, effect measures, and statistical methodology.
25. Record Summary
FACT provides a clear example of statistical analysis for a randomized prevention trial in pregnancy. The primary endpoint is binary and is evaluated with a chi-squared test and risk ratio, while continuous secondary outcomes use t-tests and mean differences. The primary preeclampsia analysis reports RR 1.10 with a 95% CI of 0.90–1.34 and P = 0.37. The secondary results extend the analysis across maternal, fetal, and neonatal outcomes, with several endpoints requiring particular attention to missing data and the size of the available analysis population.
The most useful statistical reading therefore combines the effect estimate, confidence interval, and P-value with the underlying endpoint definition, observation period, analysis population, and role of the endpoint within the overall trial. For infant outcomes, the distinction between maternal randomization and infant-level measurement is also essential. This framework allows the numerical results to be interpreted without treating every secondary comparison as equivalent to the primary hypothesis.