Research method
Kaplan-Meier Survival Analysis
Kaplan–Meier analysis estimates the probability of remaining event-free over time while keeping censored people in the risk set until they leave. The plot is a step-down survival curve; a log-rank test or a Cox model often supplies the contrast between groups. Five-year overall survival of about 92% in both trial arms, or a 10-year breast-cancer-specific HR of 1.70 in a biomarker subgroup, are KM-family results. An eight-week PHQ-9 app trial is not, even if a lexicon tagged it here.
Oncology papers reach for KM when the outcome is death or cancer-specific death over years, not a yes/no at week 9. It answers 'what fraction has survived to this time, allowing for incomplete follow-up?' Its main limitation is causal: overlapping KM curves after randomised anaesthetic assignment are a treatment contrast, whereas KM curves by ANXA1 expression or by screen-detected versus symptomatic cancer remain observational and open to lead time, length bias and confounding.
Evidence
What the evidence shows
Drawn from 4 studies in this library. Each finding starts with a plain-language takeaway, then the denser detail. Supports means evidence for a finding; Challenges means evidence against a stated position; Qualifies marks scope with a short note on each study’s contribution. Challenged positions are labeled — they are not findings.
Patients having curative primary breast-cancer surgery were randomised 1:1 to propofol or sevoflurane. Among 1,670 analysed by intention to treat (841 propofol, 829 sevoflurane), five-year overall survival was about 92% in both arms, with a hazard ratio near 1 and no statistically significant difference. Per-protocol analysis and adjustment for uneven triple-negative receptor status agreed. Rare mortality still makes equivalence hard to prove; opioid co-analgesia differed by arm.
A pooled BCAC tumour series (5,752 patients plus familial BRCA cases) related ANXA1 expression to subtype and survival. High-grade tumours had higher ANXA1 odds (adjusted OR 1.59); ANXA1 tracked basal-like/triple-negative features; among HER2-positive cases, ANXA1 overexpression had worse 10-year breast-cancer-specific survival (adjusted HR 1.70). That is an observational biomarker association, not a ready clinical assay or a chemotherapy-predictive trial.
Among 1,846 New Zealand women with first primary breast cancer, screen-detection was 62.7% in NZ European versus 49.2% in Māori cancers. Māori women had worse crude five- and ten-year survival overall, but among screen-detected cancers survival was similar or numerically better for Māori. Observational screening–survival associations cannot fully separate selection, lead time and length bias from screening benefit.
Not every tagged paper plotted deaths. An IntelliCare 2×2 factorial RCT followed PHQ-9 and GAD-7 over about 8 weeks in adults above symptom thresholds; coaching favoured GAD-7, recommendations favoured PHQ-9, and engagement was high (median 216 sessions; median last use day 56). Those are symptom and usage trajectories, not Kaplan–Meier overall survival. Lexicon over-recruitment is why the paper sits in this method list.
Open questions
Tensions and limits
Some items are genuine disagreements on the same question. Others mark different assays, populations, or outcomes — limits on how far one study travels — not a forced fight between papers.
KM in a randomised anaesthetic trial, KM in a biomarker series, KM in a screening cohort, and an 8-week app RCT are not one method applied four times. Propofol versus sevoflurane can say assigned anaesthetic did not shift 5-year OS (~92%, HR near 1). ANXA1's HR 1.70 in HER2-positive overexpression is an observational 10-year BCSS association. NZ ethnic survival gaps change when the sample is restricted to screen-detected cancers. IntelliCare did not analyse cancer death at all.
- Propofol vs sevoflurane and breast cancer survival
- ANXA1 in breast cancer prognosis
- Screening and breast cancer equity in NZ
- Coaching vs recommendations in IntelliCare
Study Role Design N Population Outcome Propofol vs sevoflurane and breast cancer survival Supports RCT1:1 propofol vs sevoflurane for anaesthesia maintenance in curative primary breast cancer surgery N=1670 · ITT: 841 propofol, 829 sevoflurane Patients undergoing curative primary breast cancer surgery 5-year overall survival ANXA1 in breast cancer prognosis Supports CohortPooled BCAC tumour series plus familial BRCA cases; ANXA1 IHC associations N=5752 · BCAC patients in primary analyses; plus familial BRCA cases in secondary analyses Women with breast cancer in pooled BCAC cohorts (mostly non-BRCA carriers) ANXA1 associations with tumour subtype markers and breast-cancer-specific survival Screening and breast cancer equity in NZ Supports CohortNew Zealand screening-age women with first primary breast cancer; screen-detected vs symptomatic N=1846 · 1,548 invasive and 298 in-situ; 1,064 (57.6%) screen-detected Screening-age women in New Zealand with newly diagnosed first primary breast cancer Screen detection rates and cancer-specific survival by ethnicity Coaching vs recommendations in IntelliCare Supports RCT2×2 factorial: coaching vs self-guided × weekly app recommendations vs none N=301 · Adults with elevated PHQ-9 or GAD-7 using IntelliCare suite Adults with elevated depression or anxiety symptoms using IntelliCare Android apps PHQ-9/GAD-7 change and app engagement over ~8 weeks Crude and stratum-specific survival can tell opposite equity stories. Māori women had worse crude five- and ten-year survival overall, yet among screen-detected cancers survival was similar or numerically better for Māori, while screen-detection itself was 49.2% versus 62.7%. A KM curve that ignores detection route will not match a curve that conditions on it, and neither fully isolates screening's causal effect.
Common misconceptions
If KM curves overlap at 92% five-year survival in 1,670 patients, the two anaesthetics have been proven equivalent.
The propofol–sevoflurane ITT and per-protocol results were null with HR near 1, but rare deaths make equivalence hard to prove, opioid co-analgesia differed by arm, and the finding is for primary breast-cancer surgery, not all cancers.
Better KM survival among screen-detected cancers proves screening caused the gain, especially if Māori look similar in that stratum.
Among 1,846 women, Māori cancers were less often screen-detected (49.2% vs 62.7%). Survival that conditions on screen-detection can reflect lead time, length bias and who attends screening, not only treatment of the same tumours found earlier. Crude overall survival was still worse for Māori.
ANXA1's adjusted HR 1.70 means the stain is ready for routine prognosis, and any paper in this method used KM on deaths.
The HR is 10-year BCSS among HER2-positive ANXA1-overexpressing cases in a pooled observational series (high-grade OR 1.59). It does not establish a clinical assay or chemotherapy prediction. IntelliCare's 8-week PHQ-9/GAD-7 factorial trial is not a survival analysis.
Exam-style questions
Short-answer questions that ask you to explain or compare, not recall.
What does Kaplan–Meier add to a simple 5-year dead/alive table in the propofol–sevoflurane trial, and what does ~92% OS in both arms still not prove?
KM uses time to death and censoring so people with shorter follow-up still contribute until they leave, rather than collapsing everyone to a yes/no at a common calendar date. Among 1,670 ITT patients the curves and HR near 1 did not differ, but rare mortality, differing opioids, and restriction to primary breast-cancer surgery block a claim of proven equivalence for all cancer anaesthesia.
Contrast ANXA1's 10-year BCSS HR of 1.70 with the anaesthetic trial's HR near 1. Which is a randomised treatment contrast?
Only anaesthetic assignment was randomised (841 vs 829). ANXA1 expression was observed in a pooled series of 5,752 patients plus familial BRCA cases; the 1.70 is an adjusted association in a HER2-positive overexpression subgroup, and the high-grade OR of 1.59 is likewise not a treatment effect.
NZ European screen-detection was 62.7% versus 49.2% in Māori. How can Māori have worse crude 5- and 10-year survival overall and similar or better survival among screen-detected cancers?
Overall KM mixes detection routes. If Māori cancers are more often symptomatic, crude survival looks worse. Restricting to screen-detected tumours compares a selected subset in which survival was similar or numerically better for Māori. That stratum does not cancel the detection gap or prove screening caused ethnic survival equality.
A methods index lists IntelliCare under Kaplan–Meier. What outcomes did that trial actually analyse, and why does the mis-tag matter?
PHQ-9 and GAD-7 over about 8 weeks, plus engagement (median 216 sessions; last use day 56), in a factorial coaching×recommendations RCT. Those are not time-to-death curves. Mixing them with 5- and 10-year breast-cancer survival teaches the wrong estimand.
The studies
4 studies in this library bear on Kaplan-Meier Survival Analysis, ordered by citations.
- Coaching vs recommendations in IntelliCare
In a factorial RCT, coaching lowered anxiety more than self-guided use, while weekly app recommendations boosted depression improvement and app sessions.
- ANXA1 in breast cancer prognosis
ANXA1-positive breast tumors clustered with poor-prognosis features (high grade, basal/TN, BRCA1/2) and worse outcomes in some high-risk subgroups such as HER2+.
- Propofol vs sevoflurane and breast cancer survival
In the CAN randomised trial, five-year overall survival after primary breast cancer surgery was essentially the same with propofol or sevoflurane maintenance anaesthesia.
- Screening and breast cancer equity in NZ
Screen-detected cancers were less common in Māori women, and crude survival was worse, but survival gaps narrowed among screen-detected cases.
Learn alongside