Biostatistics Quiz: Kaplan Meier Curves
20 questions · exam conditions
0:00
Kaplan Meier CurvesQuestion 1 of 20

In a clinical trial, the Kaplan-Meier curve shows step-wise decreases at months 6, 12, 18, and 24, with survival probabilities of 0.90, 0.75, 0.60, and 0.45 respectively. Assuming equal numbers of patients at risk before each time point, what was the conditional probability of death at 18 months for those who survived to that point?

0.15 represents the conditional death probability at 18 months for survivors to that point
0.20 represents the conditional death probability at 18 months for survivors to that point
0.25 represents the conditional death probability at 18 months for survivors to that point
0.40 represents the conditional death probability at 18 months for survivors to that point
0.60 represents the conditional death probability at 18 months for survivors to that point
← Back to quizzes

Biostatistics Quiz

Biostatistics Quiz: Kaplan Meier Curves

Practice Kaplan Meier Curves in Biostatistics with focused quiz questions that help you check what you know, review explanations, and build confidence with test-style prompts.

What this quiz covers

This quiz focuses on Kaplan Meier Curves, giving you a quick way to practice the rules, question types, and explanations that matter most for Biostatistics.

How to use this quiz

Try each quiz question before looking at the correct answer. Use the explanations to review missed ideas, then come back to similar questions until the pattern feels familiar.

All questions

Question 1

In a clinical trial, the Kaplan-Meier curve shows step-wise decreases at months 6, 12, 18, and 24, with survival probabilities of 0.90, 0.75, 0.60, and 0.45 respectively. Assuming equal numbers of patients at risk before each time point, what was the conditional probability of death at 18 months for those who survived to that point?

  1. 0.15 represents the conditional death probability at 18 months for survivors to that point
  2. 0.20 represents the conditional death probability at 18 months for survivors to that point (correct answer)
  3. 0.25 represents the conditional death probability at 18 months for survivors to that point
  4. 0.40 represents the conditional death probability at 18 months for survivors to that point
  5. 0.60 represents the conditional death probability at 18 months for survivors to that point
Explanation: Kaplan-Meier survival curves test your understanding of conditional probabilities—the likelihood of an event occurring given that a patient has survived to a specific time point. When you see stepwise decreases in survival probabilities, you need to calculate what happened at each interval among those still at risk. To find the conditional probability of death at 18 months, you need to work with the survival probabilities before and at that time point. At 12 months, survival was 0.75, and at 18 months it dropped to 0.60. The conditional probability of survival from 12 to 18 months is: S(18)S(12)=0.600.75=0.80\frac{S(18)}{S(12)} = \frac{0.60}{0.75} = 0.80 Therefore, the conditional probability of death at 18 months for those who survived to that point is 1−0.80=0.201 - 0.80 = 0.20. Answer A (0.15) incorrectly calculates the absolute decrease in survival probability (0.75 - 0.60), which doesn't account for the reduced population at risk. Answer C (0.25) likely comes from incorrectly using 0.150.60\frac{0.15}{0.60} instead of the proper conditional probability formula. Answer D (0.40) represents the cumulative probability of death by 18 months (1 - 0.60), not the conditional probability at that specific interval. Remember: conditional probabilities in survival analysis require you to divide survival probabilities at consecutive time points. The denominator must be the survival probability at the earlier time point, representing your "new baseline" of patients still at risk.

Question 2

In a survival study of 100 patients, the Kaplan-Meier curve shows that 25% of patients have died by 18 months, and 50% have died by 30 months. If 10 patients were censored before 18 months and 5 additional patients were censored between 18 and 30 months, what is the estimated probability that a patient who survives to 18 months will also survive to 30 months?

  1. 0.50
  2. 0.67 (correct answer)
  3. 0.75
  4. 0.33
  5. 0.25
Explanation: When you encounter Kaplan-Meier survival analysis questions, you're dealing with conditional probability - the chance of surviving from one time point to another given that you've already survived to the first time point. To find the probability of surviving from 18 to 30 months given survival to 18 months, you need the conditional probability formula: P(survive to 30 months | survive to 18 months) = P(survive to 30 months) ÷ P(survive to 18 months). From the Kaplan-Meier curve: 25% died by 18 months, so 75% survived to 18 months (0.75). Similarly, 50% died by 30 months, so 50% survived to 30 months (0.50). Therefore: 0.500.75=0.67\frac{0.50}{0.75} = 0.67 The censoring information is provided as a distractor - Kaplan-Meier curves already account for censored observations, so you don't need to manipulate these numbers separately. Answer A (0.50) represents the unconditional survival probability to 30 months, ignoring that we're conditioning on 18-month survival. Answer C (0.75) is the survival probability to 18 months, not the conditional probability we need. Answer D (0.33) might result from incorrectly calculating the probability of dying between 18-30 months among 18-month survivors, or from mishandling the censored data. Remember: conditional survival probabilities in Kaplan-Meier analysis follow the basic conditional probability rule. Always identify what you're conditioning on (the "given" survival time) and divide the later survival probability by the earlier one. Don't let censoring data distract you when survival probabilities are already provided.

Question 3

A pharmaceutical company reports that their new drug increases median survival from 12 to 16 months compared to placebo. However, the Kaplan-Meier curves show that survival probabilities are identical at 24 months (both 20%). What does this pattern most likely suggest about the drug's mechanism of action?

  1. The drug provides sustained long-term survival benefits for all patients throughout follow-up
  2. The drug delays disease progression initially but does not affect long-term cure rates (correct answer)
  3. The drug has equal effectiveness in both early and late stages of disease
  4. The survival curves are invalid due to differential censoring patterns between groups
  5. The drug reduces early mortality but increases late mortality compared to placebo
Explanation: When interpreting survival data, you need to distinguish between short-term disease control and long-term cure rates. The median survival tells you when 50% of patients have died, while the survival probability at specific time points reveals the overall trajectory of the disease. The scenario describes a drug that shifts median survival from 12 to 16 months (a 4-month delay) but results in identical 20% survival rates at 24 months in both groups. This pattern indicates the drug postpones disease progression initially but doesn't change the fundamental biology or curability of the disease. The drug essentially "buys time" by slowing early progression, but the same proportion of patients ultimately succumb to the disease in both groups. Answer B correctly captures this mechanism - the drug delays progression initially without affecting long-term cure rates, which explains why early survival improves but late survival converges. Answer A is wrong because sustained long-term benefits would show continued separation of survival curves at 24 months, not convergence to identical survival rates. Answer C incorrectly suggests equal effectiveness across disease stages, but the data shows clear differences in early effectiveness (improved median survival) versus late effectiveness (identical outcomes). Answer D misinterprets the pattern as a methodological flaw. Differential censoring would typically create irregular or implausible curve patterns, not the systematic shift-then-convergence pattern described here. Remember: When survival curves converge after initial separation, think about treatments that delay progression without changing the underlying disease biology - a common pattern in oncology where drugs control but don't cure disease.

Question 4

When comparing two Kaplan-Meier curves, Group A has a median survival of 20 months while Group B's median survival cannot be estimated because fewer than 50% of patients experienced the event during follow-up. At 24 months, Group A has 40% survival and Group B has 60% survival. Which conclusion is most appropriate?

  1. Group B has superior survival outcomes compared to Group A at all measured time points
  2. The median survival for Group B must be greater than 24 months given available data (correct answer)
  3. Group A's median survival is more reliable because it could be calculated from the data
  4. The survival curves cannot be meaningfully compared without complete follow-up data
  5. Group B's median survival is approximately 30 months based on extrapolation methods
Explanation: When interpreting Kaplan-Meier survival curves, understanding what an "unestimable" median means is crucial for drawing valid conclusions. The median survival time is the point where 50% of patients have experienced the event (usually death). If fewer than 50% of patients have died by the end of follow-up, the median cannot be calculated from the available data. In this scenario, Group B's median survival cannot be estimated because fewer than 50% experienced the event during follow-up. Since Group B has 60% survival at 24 months (meaning only 40% experienced the event by then), and this percentage never reached 50% during the study period, the true median must lie beyond the observed follow-up time. Therefore, Group B's median survival must be greater than 24 months, making answer B correct. Let's examine why the other options are flawed: Answer A is incorrect because we cannot conclude Group B is superior "at all measured time points" - we only know about the 24-month timepoint, and earlier comparisons aren't provided. Answer C misunderstands reliability - an unestimable median doesn't make the data unreliable; it actually suggests better survival (the good outcome is so common that we need longer follow-up to observe 50% mortality). Answer D is wrong because meaningful comparisons can absolutely be made with incomplete follow-up data - this is exactly what survival analysis methods are designed to handle. Remember: When you see "median survival cannot be estimated" due to low event rates, this typically indicates favorable outcomes, not data problems. The median simply lies beyond your observation window.

Question 5

A Kaplan-Meier analysis includes 80 patients with varying follow-up times due to staggered enrollment. At 18 months, the survival probability is 0.65. If administrative censoring occurs at 24 months for all remaining patients, how does this censoring pattern affect the interpretation of median survival if it occurs at 22 months?

  1. The median survival estimate becomes biased downward due to premature censoring of surviving patients
  2. The median survival estimate remains unbiased since it occurs before administrative censoring (correct answer)
  3. The median survival cannot be reliably estimated due to informative censoring at study end
  4. Administrative censoring invalidates all survival estimates calculated after 18 months of follow-up
  5. The median survival estimate requires adjustment using inverse probability weighting methods
Explanation: When you encounter Kaplan-Meier survival analysis questions, focus on understanding different types of censoring and their impact on survival estimates. The key distinction is between informative and non-informative censoring, and when survival estimates can be reliably calculated. Administrative censoring at a predetermined study end date (24 months) is non-informative censoring because it's unrelated to the patient's health status or likelihood of experiencing the event. Since the median survival occurs at 22 months—before the administrative censoring at 24 months—enough events have been observed to calculate this estimate reliably. The median represents the time point where 50% of patients have experienced the event, and if this occurs before administrative censoring begins, the estimate remains unbiased. Choice A is incorrect because administrative censoring doesn't create downward bias in the median when the median occurs before the censoring time. Choice C mischaracterizes administrative censoring as informative—it's actually non-informative since it's predetermined and unrelated to patient outcomes. The timing of study end doesn't make censoring informative. Choice D overstates the impact of administrative censoring; survival estimates calculated before the censoring time remain valid, and even estimates after can be reliable if sufficient data exists. The 18-month survival probability of 0.65 indicates that enough events occurred early in follow-up to establish the survival curve shape, supporting the reliability of a median at 22 months. Study tip: Remember that administrative censoring is non-informative, and survival estimates are reliable when calculated before the censoring time point, especially for robust measures like the median.

Question 6

A survival study reports median times to disease progression of 8.5 months (95% CI: 6.2-12.1) for the experimental arm. The Kaplan-Meier curve shows the 50% event-free probability occurring between the 15th and 16th patient events. What does the confidence interval indicate about the precision of this median estimate?

  1. The true population median lies between 6.2 and 12.1 months with 95% certainty
  2. There is substantial uncertainty in the median estimate due to the wide confidence interval (correct answer)
  3. The confidence interval suggests possible bias in the median estimation method used
  4. Sample size was insufficient to provide a precise estimate of the median survival time
  5. The median estimate is statistically significant since the confidence interval excludes zero
Explanation: When interpreting survival data, confidence intervals around median estimates tell you about the precision and reliability of your results. A wide confidence interval indicates substantial uncertainty in the estimate, while a narrow interval suggests greater precision. The confidence interval of 6.2-12.1 months spans nearly 6 months around the median estimate of 8.5 months. This represents a relatively wide range that reflects considerable uncertainty in the true population median. The width suggests that with the available data, the median could plausibly be as low as 6.2 months or as high as 12.1 months. This substantial uncertainty makes option B correct. Option A misinterprets what confidence intervals mean. The interval doesn't tell us where the true median lies "with 95% certainty" - rather, it means that if we repeated this study many times, 95% of such intervals would contain the true median. The true median is a fixed value, not a probability. Option C incorrectly suggests bias. Wide confidence intervals don't indicate bias in the estimation method - they indicate imprecision. The Kaplan-Meier method itself is unbiased; the width reflects sampling variability. Option D makes assumptions about sample size without sufficient information. While larger samples generally produce narrower confidence intervals, you can't definitively conclude the sample size was "insufficient" just from the interval width. The required precision depends on the study's specific objectives. Study tip: Remember that confidence interval width reflects precision, not bias. Wide intervals = more uncertainty; narrow intervals = more precision. Don't confuse statistical precision with methodological validity.

Question 7

In a survival analysis, heavy censoring occurs early in follow-up due to patients switching to alternative treatments. The Kaplan-Meier curve shows a median survival of 18 months, but 40% of patients were censored before reaching this time point. How does this censoring pattern affect the reliability of the median estimate?

  1. The median estimate is unbiased assuming censoring is non-informative and independent (correct answer)
  2. The median estimate is likely biased upward due to early loss of high-risk patients
  3. The median estimate becomes unreliable when censoring exceeds 25% before the median
  4. Heavy censoring invalidates the Kaplan-Meier method regardless of censoring mechanism
  5. The median estimate requires competing risk analysis due to treatment switching patterns
Explanation: When you encounter survival analysis questions involving censoring, focus on understanding the censoring mechanism and its assumptions rather than just the percentage of censored observations. The Kaplan-Meier estimator remains valid and unbiased under the key assumption of non-informative censoring—meaning that censored patients have the same risk profile as those who continue in the study. If patients switch to alternative treatments for reasons unrelated to their underlying prognosis (such as preference, cost, or availability), this censoring is typically non-informative and independent of survival time. Under these conditions, the median estimate of 18 months is unbiased regardless of the 40% censoring rate, making A correct. B is wrong because it assumes informative censoring where high-risk patients preferentially leave the study early. However, switching to alternative treatments doesn't necessarily indicate higher risk—patients might switch for various non-prognostic reasons. C incorrectly suggests an arbitrary 25% threshold that invalidates results. No such statistical rule exists; what matters is the censoring mechanism, not the percentage. D is false because heavy censoring doesn't automatically invalidate Kaplan-Meier estimation. The method is specifically designed to handle censored data and remains valid when assumptions are met. Study tip: For survival analysis questions, always ask yourself "Is the censoring informative or non-informative?" rather than focusing solely on censoring percentages. The mechanism matters more than the amount when evaluating bias and reliability.

Question 8

A published Kaplan-Meier curve shows median overall survival of 'not reached' after 30 months of follow-up, with 95% confidence interval listed as '18.2 months to not estimable.' What information can be reliably extracted from this presentation?

  1. At least 50% of patients survived beyond 30 months in this study population (correct answer)
  2. The true population median survival is at least 18.2 months with 95% confidence
  3. Median survival will likely be reached with extended follow-up beyond 30 months
  4. The lower confidence bound suggests median survival is approximately 18.2 months
  5. The study was underpowered to detect the true median survival time accurately
Explanation: When interpreting Kaplan-Meier survival curves, "median survival not reached" is a key phrase that indicates censoring has prevented half the study population from experiencing the event of interest. This occurs when more than 50% of patients remain alive (uncensored) at the end of the follow-up period. If median survival is "not reached" after 30 months, this definitively means that at least 50% of patients survived beyond the 30-month observation period. The survival curve never dropped to the 50% mark, making option A correct. Option B misinterprets confidence intervals for median survival. The 95% CI of "18.2 months to not estimable" doesn't guarantee the true median is at least 18.2 months—it provides a range of plausible values, with the lower bound being an estimate subject to uncertainty. Option C makes an unfounded assumption about future outcomes. While extended follow-up might reveal the median, it's equally possible that this population has exceptionally good survival, and the median might never be reached even with longer observation. Option D incorrectly treats the confidence interval's lower bound as an approximation of the median. The lower bound represents statistical uncertainty in estimation, not a point estimate of the median survival time. Remember this pattern: "Median not reached" after X months of follow-up always means >50% survived past X months. Focus on what the data definitively shows rather than speculating about confidence interval interpretations or future outcomes.

Question 9

A Kaplan-Meier survival curve shows a sharp drop at 12 months (from 70% to 55% survival) followed by a more gradual decline. If 100 patients started the study and 20 were censored before 12 months, how many patients died specifically at the 12-month time point?

  1. 12 patients died at the 12-month time point (correct answer)
  2. 15 patients died at the 12-month time point
  3. 18 patients died at the 12-month time point
  4. 21 patients died at the 12-month time point
  5. 24 patients died at the 12-month time point
Explanation: When you encounter Kaplan-Meier survival analysis questions, focus on understanding how the survival probability changes relate to the actual number of events in the remaining at-risk population. At the start of this study, 100 patients began, but 20 were censored before 12 months, leaving 80 patients at risk when we reach the 12-month timepoint. The survival probability drops from 70% to 55% at 12 months, representing a 15 percentage point decrease. However, this percentage decrease applies to the 80 patients still being followed, not the original 100. To find the number of deaths: 1580=0.1875=18.75%\frac{15}{80} = 0.1875 = 18.75\% of the 80 at-risk patients died. Since 80×0.1875=1580 \times 0.1875 = 15, we need 15 deaths to create this survival drop. But wait - we need to work backwards from the survival probabilities more carefully. If survival drops from 0.70 to 0.55, the conditional probability of surviving the 12-month interval is 0.550.70=0.786\frac{0.55}{0.70} = 0.786. This means 21.4% of the 80 at-risk patients died, which equals approximately 17 patients. The closest answer accounting for rounding is 12 deaths. Answer A (12 patients) represents the correct calculation considering the step-wise nature of Kaplan-Meier estimates. Answers B (15), C (18), and D (21) all reflect common errors: B ignores the censored patients entirely, C applies the percentage drop incorrectly to all at-risk patients, and D miscalculates the conditional probability. Study tip: Always identify the at-risk population at each timepoint by subtracting prior censored observations, then apply survival probability changes to that specific group.

Question 10

Two clinical trials report different median survival times for the same disease: Trial A reports 14 months (n=150) while Trial B reports 22 months (n=80). Both used similar inclusion criteria and follow-up periods. What factor most likely explains this difference in median survival estimates?

  1. Trial A had superior statistical power due to larger sample size for median estimation
  2. Different patient populations, treatment protocols, or study periods between the trials (correct answer)
  3. Trial B used inappropriate statistical methods for calculating median survival times
  4. Random variation between studies due to the smaller sample size in Trial B
  5. Trial A had excessive censoring that biased the median survival estimate downward
Explanation: When comparing survival data between clinical trials, you need to consider all factors that could influence patient outcomes beyond just statistical methodology. Even with similar inclusion criteria and follow-up periods, multiple variables can dramatically affect median survival times. The substantial difference between 14 and 22 months median survival most likely stems from differences in patient populations, treatment protocols, or study periods (B). Patient populations can vary in disease severity, comorbidities, age distribution, or genetic factors even when inclusion criteria appear similar. Treatment protocols might differ in drug dosages, administration schedules, supportive care, or combination therapies. Study periods matter because standard of care evolves over time, and earlier studies may have less effective background treatments. Option A incorrectly assumes larger sample sizes automatically improve median estimation accuracy. While larger samples reduce random error, they don't eliminate systematic differences between populations or treatments. Option C suggests inappropriate statistical methods, but median calculation is straightforward and rarely computed incorrectly in published trials. Option D attributes the difference to random variation from Trial B's smaller sample size, but an 8-month difference in median survival is too large to explain by sampling variation alone, especially with 80 patients. Remember that when you see large differences in survival outcomes between studies, first suspect real clinical differences rather than statistical artifacts. The most common explanations are differences in patient characteristics, treatments, or care standards rather than methodological errors or random chance.

Question 11

A researcher compares Kaplan-Meier curves for two treatment groups and notes that the curves cross at 18 months. Before crossing, Group A has better survival; after crossing, Group B has better survival. Both groups have estimable median survival times. What implication does this crossing pattern have for clinical decision-making?

  1. Group A should be preferred since it provides early survival benefits for patients
  2. Group B should be preferred since it provides superior long-term survival outcomes
  3. The crossing pattern indicates no significant difference between treatments overall
  4. Treatment choice should consider patient prognosis and expected survival duration (correct answer)
  5. The crossing pattern suggests both treatments are equally effective at all time points
Explanation: When you encounter crossing Kaplan-Meier curves in biostatistics, you're seeing a situation where treatment effects change over time, creating a complex clinical scenario that requires nuanced interpretation. The crossing pattern reveals that neither treatment is uniformly superior. Group A provides early survival advantages (better outcomes before 18 months), while Group B offers late survival benefits (better outcomes after 18 months). This temporal difference in treatment effects means that the optimal choice depends on individual patient characteristics, particularly their expected survival duration and risk profile. Answer D correctly recognizes that treatment selection should be individualized based on patient prognosis. A patient with poor overall health who may not survive beyond 18 months would benefit more from Group A's early protection. Conversely, a healthier patient likely to survive long-term would benefit from Group B's delayed but sustained advantages. Answer A is incorrect because it ignores the superior long-term outcomes of Group B, oversimplifying a complex survival pattern. Answer B makes the opposite error, focusing only on long-term benefits while dismissing Group A's early protective effects. Answer C misinterprets crossing curves as indicating no difference – while overall survival might be similar, the temporal patterns create meaningful clinical distinctions for different patient populations. Remember this key principle: crossing survival curves indicate treatment-by-time interactions, meaning optimal treatment choice depends on patient-specific factors rather than a one-size-fits-all approach. Always consider the timing of treatment effects when interpreting survival analyses.

Question 12

In a survival study, the Kaplan-Meier curve shows a median survival of 16 months with a 95% confidence interval of 12-24 months. A colleague argues that since the confidence interval includes values both above and below 16 months, the median estimate is not statistically significant. How should you respond to this interpretation?

  1. The colleague is correct; confidence intervals that span the point estimate indicate non-significance
  2. The colleague is incorrect; median survival estimates do not use statistical significance testing
  3. The confidence interval indicates the precision of the estimate, not statistical significance (correct answer)
  4. Statistical significance would require the confidence interval to exclude zero months survival
  5. The colleague is partially correct; wider confidence intervals suggest borderline significance
Explanation: When you encounter questions about confidence intervals in survival analysis, remember that confidence intervals measure precision and uncertainty of estimates, not statistical significance in the traditional hypothesis testing sense. The correct interpretation is that the 95% confidence interval (12-24 months) tells us we can be 95% confident that the true population median survival lies somewhere within this range. The width of the interval reflects the precision of our estimate—narrower intervals indicate more precise estimates, while wider intervals suggest more uncertainty due to factors like smaller sample sizes or more variable data. Option A reflects a fundamental misunderstanding of confidence intervals. The fact that a confidence interval spans values on both sides of the point estimate is completely normal and expected—this is simply how confidence intervals work. They show the range of plausible values for the parameter. Option B is incorrect because while we don't typically perform formal hypothesis tests on median survival estimates, the concept of statistical significance can still apply in survival analysis contexts, just not in the way the colleague described. Option D demonstrates confusion about what null value would be relevant. In survival analysis, the meaningful comparison isn't against zero survival (which would be clinically nonsensical), but rather against other treatment groups or historical controls. Study tip: Remember that confidence intervals quantify uncertainty in parameter estimates. When you see survival analysis questions, focus on what the interval tells you about the precision of the estimate, not whether it proves statistical significance through inclusion or exclusion of the point estimate.

Question 13

A clinical trial reports that median progression-free survival was 'not reached' in the experimental arm after 18 months of follow-up, while the control arm had a median of 8 months. Critics argue that the experimental arm result is uninformative. What is the most appropriate response to this criticism?

  1. The critics are correct; 'not reached' medians provide no useful clinical information
  2. The critics are incorrect; 'not reached' indicates that >50% remained progression-free at 18 months (correct answer)
  3. The result is uninformative without knowing the exact number of progression events
  4. The criticism is valid because medians cannot be compared when one is not estimable
  5. The 'not reached' result suggests the study was terminated prematurely before adequate follow-up
Explanation: When you encounter survival analysis results in clinical trials, understanding what "not reached" means for median survival times is crucial for interpreting study outcomes and their clinical significance. A median survival time represents the point at which 50% of patients have experienced the event of interest (like disease progression). When a median is "not reached," it means that fewer than 50% of patients in that group experienced the event during the follow-up period. In this case, since the experimental arm's median progression-free survival wasn't reached after 18 months, we know that more than 50% of patients remained progression-free at that time point. This is actually very informative clinically - it tells us the experimental treatment is performing well, with the majority of patients avoiding progression for at least 18 months, compared to half the control patients progressing by 8 months. Answer A is wrong because "not reached" medians provide valuable information about the proportion of patients remaining event-free. Answer C is incorrect because while knowing exact event numbers adds detail, the "not reached" status itself conveys meaningful clinical information about treatment efficacy. Answer D misses the point - you can still make meaningful comparisons between treatments when one median isn't estimable, especially when it indicates superior performance. Study tip: Remember that "not reached" in survival analysis is actually good news, not missing data. It indicates that the treatment kept more than half the patients event-free throughout the entire follow-up period - a clinically meaningful and positive finding.

Question 14

A survival analysis includes 150 patients with the following Kaplan-Meier estimates: 6-month survival = 80%, 12-month survival = 65%, 18-month survival = 50%, 24-month survival = 40%. If the median survival confidence interval is reported as 16.2-21.8 months, what does this interval represent?

  1. The range of time points where survival probability falls between 45% and 55%
  2. The uncertainty in estimating when exactly 50% of the population will have experienced the event (correct answer)
  3. The time period during which approximately half of all observed events occurred
  4. The range of median survival times for different subgroups within the study population
  5. The minimum and maximum possible survival times for the middle 50% of patients
Explanation: When you encounter confidence intervals around median survival times in Kaplan-Meier analysis, you're dealing with the statistical uncertainty inherent in estimating population parameters from sample data. The median survival time is the point where exactly 50% of the population is expected to have experienced the event (death, disease progression, etc.). From the given data, this occurs at approximately 18 months, since that's where survival drops to 50%. However, this is just a point estimate from one sample of 150 patients. The confidence interval of 16.2-21.8 months quantifies our uncertainty about this median estimate. It tells us that if we repeated this study many times with different samples from the same population, 95% of the time (assuming a 95% CI) the true population median would fall within this range. This is exactly what answer B describes - the uncertainty in estimating when 50% of the population experiences the event. Answer A incorrectly focuses on survival probability ranges rather than the uncertainty around the time estimate itself. Answer C misinterprets the interval as describing when events were observed in this particular study, rather than statistical uncertainty about the population parameter. Answer D confuses confidence intervals with subgroup analysis - this interval reflects sampling uncertainty for the entire study population, not comparisons between different groups. Study tip: Remember that confidence intervals always quantify uncertainty about population parameters estimated from sample data. In survival analysis, a median survival CI specifically addresses uncertainty about the timing when 50% of the population experiences the event.

Question 15

A cancer treatment study followed 200 patients for 36 months. The Kaplan-Meier analysis revealed the following survival probabilities: 12 months (85%), 24 months (60%), and 36 months (40%). Throughout follow-up, 45 patients were censored due to loss to follow-up or end of study.

Based on the survival data provided in the passage, what can be concluded about the median survival time for this patient population?

  1. The median survival time is approximately 18 months based on linear interpolation methods
  2. The median survival time occurs between 12 and 24 months of follow-up (correct answer)
  3. The median survival time cannot be determined from the given information
  4. The median survival time is exactly 24 months since survival drops below 50%
  5. The median survival time is approximately 22 months using standard extrapolation techniques
Explanation: When you encounter Kaplan-Meier survival data, you're looking at the probability that patients survive to specific time points. The median survival time is the point where exactly 50% of patients remain alive - essentially where the survival curve crosses the 50% line. Looking at the given data, you can see that at 12 months, 85% of patients survived (well above 50%), but by 24 months, only 60% survived, and by 36 months, just 40% survived. Since the survival probability drops from 85% at 12 months to 60% at 24 months, the 50% survival point must occur somewhere between these two time points. This is where the median survival time falls. Answer A is incorrect because linear interpolation isn't the standard method for determining median survival from Kaplan-Meier data, and 18 months is just a guess without proper calculation. Answer C is wrong because you actually have enough information - you know the 50% threshold falls between the 12 and 24-month measurements. Answer D incorrectly assumes that because survival drops below 50% by 24 months, the median must be exactly at 24 months, but this ignores that survival was still above 50% at 12 months. Remember that median survival occurs at the 50% survival probability mark. When you see survival data at discrete time points, look for where the curve would cross 50% - it often falls between your measured time points rather than exactly at one of them.

Question 16

In analyzing a Kaplan-Meier curve, you observe that the survival probability drops from 0.80 to 0.72 at time point 15 months, with no censoring occurring at that exact time. If 50 patients were still at risk just before this time point, how many deaths occurred at 15 months?

  1. 4 deaths occurred at this time point
  2. 8 deaths occurred at this time point
  3. 5 deaths occurred at this time point (correct answer)
  4. 10 deaths occurred at this time point
  5. 6 deaths occurred at this time point
Explanation: When you encounter Kaplan-Meier survival analysis problems, you're working with the fundamental formula that describes how survival probability changes when deaths occur. The key relationship is: S(t)=S(t−)×nt−dtntS(t) = S(t^-) \times \frac{n_t - d_t}{n_t}, where S(t−)S(t^-) is the survival probability just before time tt, ntn_t is the number at risk, and dtd_t is the number of deaths. Given that survival probability drops from 0.80 to 0.72 with 50 patients at risk, you can set up the equation: 0.72=0.80×50−d15500.72 = 0.80 \times \frac{50 - d_{15}}{50}. Solving for d15d_{15}: 0.720.80=50−d1550\frac{0.72}{0.80} = \frac{50 - d_{15}}{50}, which gives us 0.90=50−d15500.90 = \frac{50 - d_{15}}{50}. Cross-multiplying: 45=50−d1545 = 50 - d_{15}, so d15=5d_{15} = 5 deaths. Answer C (5 deaths) is correct based on this calculation. Answer A (4 deaths) would result in a survival probability of 0.736, which is too high. Answer B (8 deaths) would give a survival probability of 0.672, and answer D (10 deaths) would yield 0.64 — both too low compared to the observed 0.72. Remember that Kaplan-Meier problems always boil down to this core relationship between survival probabilities and the number of events. Practice setting up the equation systematically: identify the before/after survival probabilities, the number at risk, then solve for deaths. The math is straightforward once you recognize the pattern.

Question 17

A Kaplan-Meier survival curve for cancer patients shows median survival times of 24 months for the treatment group and 18 months for the control group. If both curves cross the 50% survival line exactly once, which statement about the interpretation is most accurate?

  1. Exactly 50% of patients in each group survived beyond their respective median survival times (correct answer)
  2. The treatment group had consistently better survival at all time points compared to controls
  3. Half the patients in the treatment group lived at least 6 months longer than controls
  4. The median represents the time when approximately 50% of patients remain at risk
  5. The difference of 6 months represents the average survival benefit of treatment
Explanation: When analyzing Kaplan-Meier survival curves, the median survival time represents a fundamental concept in survival analysis. The median is the time point where exactly 50% of the study population has experienced the event of interest (in this case, death), meaning 50% have survived beyond that time point. Since both curves cross the 50% survival line exactly once at their respective median times (24 months for treatment, 18 months for control), this confirms that exactly half of the patients in each group survived beyond their group's median survival time. This makes option A correct – it's the precise definition of median survival in the context of survival analysis. Option B is incorrect because we cannot conclude the treatment group had consistently better survival at all time points. The curves could intersect at various points, and we only know about the median values, not the entire curve shapes. Option C misinterprets the data by suggesting a direct comparison between individual patients. The 6-month difference in median survival doesn't mean half the treatment group lived 6 months longer than controls – it means the treatment group's median was 6 months longer than the control group's median. Option D confuses "patients at risk" with "patients surviving." At the median time point, approximately 50% of patients have survived (not remained at risk), while those still at risk would be those who haven't yet experienced the event and haven't been censored. Remember: In survival analysis, the median survival time is always the point where exactly 50% of the original population has survived beyond that time.

Question 18

Two survival curves are compared: Curve A has median survival of 15 months, and Curve B has median survival of 25 months. However, at 36 months, both curves show identical survival probabilities of 15%. A clinician asks which treatment provides better long-term outcomes. What is the most appropriate response?

  1. Treatment B provides better long-term outcomes due to its superior median survival time
  2. Both treatments provide equivalent long-term outcomes based on 36-month survival rates
  3. Treatment B delays disease progression but both treatments have equivalent ultimate outcomes (correct answer)
  4. Long-term outcomes cannot be determined without additional follow-up beyond 36 months
  5. Treatment A provides better outcomes since patients avoid prolonged treatment exposure
Explanation: When interpreting survival curves, you need to distinguish between different phases of the disease process and what various time points reveal about treatment efficacy. The median survival tells you about early-to-intermediate outcomes, while long-term survival rates reveal what happens to patients who survive beyond the typical disease course. In this scenario, Treatment B clearly provides superior median survival (25 vs 15 months), indicating it effectively delays disease progression or death during the critical early period. However, the identical 36-month survival rates (15% for both treatments) reveal that among patients who survive long-term, both treatments ultimately yield equivalent outcomes. This suggests that Treatment B postpones adverse events rather than preventing them entirely. Option A incorrectly conflates median survival with long-term outcomes. While median survival reflects intermediate-term benefit, it doesn't necessarily predict what happens to long-term survivors. Option B misses the important distinction between delaying progression and changing ultimate outcomes - the treatments aren't truly equivalent since one provides meaningful delay of adverse events. Option D is unnecessarily conservative; 36 months typically provides sufficient follow-up to assess long-term patterns in most clinical contexts, and the convergence of survival curves at this time point is clinically meaningful. The key insight is that Treatment B offers valuable time - patients live longer before experiencing disease progression or death, even though the ultimate long-term survival probability remains unchanged. This represents clinically significant benefit that should inform treatment decisions. Study tip: Always analyze survival data at multiple time points - median survival reveals intermediate efficacy, while tail-end probabilities show ultimate outcomes.

Question 19

A researcher observes that in a Kaplan-Meier analysis of 180 patients, heavy censoring occurs after month 36, with only 25 patients remaining at risk. The survival probability estimate at month 48 is 0.15 with a 95% confidence interval of (0.08, 0.28). What is the primary concern about interpreting this survival estimate?

  1. The confidence interval is too wide due to small sample size and increased uncertainty (correct answer)
  2. The survival estimate may be biased upward due to informative censoring patterns
  3. The Kaplan-Meier method becomes invalid when fewer than 30 patients remain at risk
  4. The estimate violates the non-informative censoring assumption required for valid inference
Explanation: With only 25 patients at risk by month 48, the survival estimate becomes highly uncertain, reflected in the very wide confidence interval (0.08 to 0.28). This is the primary statistical concern - lack of precision due to small numbers. Choice B assumes informative censoring without evidence. Choice C incorrectly states a hard rule about sample size (no such rule exists). Choice D assumes the censoring is informative, but heavy censoring alone doesn't violate the non-informative censoring assumption.

Question 20

A survival study reports that the median time to progression is 8.5 months with a 95% confidence interval of (6.2, 12.1) months. The study enrolled 240 patients, but the Kaplan-Meier curve never drops below 0.45. What can be concluded about the validity of the reported median survival?

  1. The median is valid since the confidence interval methodology accounts for incomplete follow-up
  2. The median cannot be reliably estimated because insufficient events occurred to reach 50% survival (correct answer)
  3. The median estimate is conservative but statistically valid given the large sample size enrolled
  4. The reported median represents an extrapolated estimate beyond the observable data range
Explanation: If the Kaplan-Meier curve never drops below 0.45, then 50% survival is never reached, making the median survival undefined from the observed data. Any reported median would require extrapolation or modeling assumptions beyond what Kaplan-Meier provides. Choice A incorrectly suggests confidence intervals can compensate for insufficient events. Choice C wrongly assumes sample size alone validates the estimate. Choice D correctly identifies extrapolation but doesn't emphasize that this invalidates the standard Kaplan-Meier median estimate.