Prognostication Tools
Evidence-based tools to estimate prognosis and guide goals-of-care discussions. PPI, PPS, and surprise question for cancer. ePrognosis for geriatrics. Prognostication is imperfect -communicate uncertainty honestly.
🔍 Overview
Key Prognostic Tools
| Tool | Population | What It Predicts |
|---|---|---|
| Surprise Question | Any serious illness | "Would I be surprised if this patient died in the next 12 months?" If no → initiate palliative discussion. |
| PPS (Palliative Performance Scale) | Cancer and non-cancer | Functional status 0–100%. PPS ≤ 50% → median survival ~6 months. PPS ≤ 30% → days to weeks. |
| PPI (Palliative Prognostic Index) | Cancer | Predicts survival < 3 weeks vs > 6 weeks based on PPS, oral intake, edema, dyspnea, delirium. |
| APACHE II/IV | ICU patients | ICU mortality prediction. Higher score = higher mortality. |
| MELD score | Liver disease | 3-month mortality in cirrhosis. Transplant prioritization. |
| Seattle Heart Failure Model | Heart failure | 1–5 year survival in chronic HF based on multiple variables. |
Physicians systematically overestimate survival -by a factor of 3–5× on average. Use validated tools to supplement clinical judgment. Communicate in ranges, not precise dates.
Key Evidence: Physicians overestimate survival by 3-5x on average Christakis, 2000. The surprise question has ~70% sensitivity/specificity for 12-month mortality Downar, 2017. PPI accurately predicts survival < 3 weeks Morita, 1999. Feeding tubes do not improve survival in advanced dementia Teno, 2012. Functional trajectories differ by disease type and shape prognostic communication Lunney, 2003.
🚨 Management
Communicating Prognosis
- Ask before telling: "How much information would you like to know about what to expect?"
- Use ranges: "We're talking about days to weeks" or "weeks to months" rather than specific dates
- Frame honestly: "I hope for the best, but I'm worried we may be looking at weeks rather than months"
- Acknowledge uncertainty: "No one can predict exactly -these are estimates based on what we know"
- Functional decline trajectory: Describe what to expect (increasing sleep, decreasing intake, less interaction)
Common Prognostic Time Frames
| Time Frame | Clinical Indicators |
|---|---|
| Months | Declining functional status, weight loss, increasing symptoms, frequent hospitalizations |
| Weeks | Bed-bound most of day, minimal oral intake, drowsy, dependent for all ADLs |
| Days | Bed-bound, minimal consciousness, mottling, Cheyne-Stokes breathing, no oral intake |
| Hours | Unresponsive, irregular breathing, cool/mottled extremities, mandibular breathing |
Functional Decline Trajectories
| Trajectory | Pattern | Typical Diseases | Prognostic Challenge |
|---|---|---|---|
| Cancer | Preserved function → rapid decline in final weeks | Solid tumors, hematologic malignancies | Relatively predictable; decline signals end-stage |
| Organ Failure | Gradual decline + acute exacerbations + partial recovery | CHF, COPD, ESRD, cirrhosis | Unpredictable; each exacerbation may be fatal |
| Dementia/Frailty | Slow progressive decline over years | Alzheimer's, vascular dementia, frailty syndrome | Very prolonged; hard to identify terminal phase |
| Sudden Death | Full function → death | MI, PE, stroke, trauma, arrhythmia | No prognostic window; emphasizes advance planning for all |
🧪 Workup
Assessment Components
- Functional status -PPS, Karnofsky, ECOG performance status
- Nutritional status -weight loss trajectory, BMI, intake history (albumin reflects inflammation, NOT nutrition per ASPEN 2021)
- Disease-specific markers -tumor markers, MELD, NYHA class, eGFR trajectory
- Hospitalization frequency -readmissions suggest declining trajectory
- Symptom burden -increasing symptom load suggests progression
- Patient/family understanding and wishes
💊 Medications
Prognostication itself does not involve medications. When prognosis guides transition to comfort care, see the Code Status & Advance Directives topic for symptom management medications (morphine for pain/dyspnea, lorazepam for anxiety, glycopyrrolate for secretions, haloperidol for agitation).
📋 On Rounds
Pimp Questions
What is the "surprise question" and how accurate is it?
"Would I be surprised if this patient died in the next 12 months?" A negative answer (you would NOT be surprised) identifies patients who would benefit from palliative care discussions. Sensitivity ~70%, specificity ~70%. It performs best as a screening tool -not for precise prognostication, but for triggering earlier goals-of-care conversations. Can be applied to any illness.
Why do physicians consistently overestimate survival?
Multiple cognitive biases: (1) Optimism bias -physicians want patients to do well, (2) Therapeutic optimism -belief that treatment is helping more than it is, (3) Anchoring -to the patient's best functional state rather than current trajectory, (4) Patient expectations -pressure from patients/families who want hope. Studies show physicians overestimate by 3–5× on average. Using validated prognostic tools helps counteract these biases.
What is PPS and how do you score it?
Palliative Performance Scale: 0-100% in 10% increments. Based on 5 domains: ambulation, activity/evidence of disease, self-care, intake, and level of consciousness. PPS 50% = mainly sit/lie, unable to work, considerable assistance needed, reduced intake. PPS ≤ 30% correlates with days to weeks survival.
What are the components of the Palliative Prognostic Index (PPI)?
PPS score (≤ 20 = 4 pts, 30-50 = 2.5 pts) + oral intake (mouthfuls = 2.5, none = 1) + edema (1 pt) + dyspnea at rest (3.5 pts) + delirium (4 pts). PPI > 6 predicts survival < 3 weeks (sensitivity 83%, specificity 85%). Morita, 1999.
How do you use the "Karnofsky" vs "ECOG" performance status?
Karnofsky: 0-100 scale (100 = normal, 0 = dead). ECOG/Zubrod: 0-5 scale (0 = fully active, 5 = dead). ECOG is simpler and most used in clinical trials. KPS < 40 or ECOG ≥ 3 suggests limited treatment tolerance and poor short-term prognosis.
What is the trajectory of functional decline in cancer vs organ failure vs dementia?
Cancer: relatively preserved function, then rapid decline in last weeks/months. Organ failure (CHF, COPD): gradual decline with acute exacerbations, each potentially fatal, partial recovery between. Dementia/frailty: slow progressive decline over years. Lunney, 2003. Trajectory shapes communication approach.
What is the "PaP score" and when is it used?
Palliative Prognostic Score: uses dyspnea, anorexia, KPS, clinical prediction of survival, WBC count, and lymphocyte percentage. Divides into 3 risk groups: A (>70% chance of 30-day survival), B (30-70%), C (<30%). Pirovano, 1999. Useful for cancer patients when objective prognostication is needed.
What lab values suggest poor prognosis in advanced illness?
Albumin < 2.5, elevated LDH, leukocytosis, lymphopenia, elevated CRP/inflammatory markers, rising creatinine. Low albumin is the strongest single lab predictor across disease types. Note: albumin reflects inflammation (negative acute phase reactant), not nutritional status per ASPEN 2021.
How do you communicate prognosis when the family asks "How long does he have?"
Use ranges not specific dates: "days to weeks," "weeks to months." Frame with hope and honesty: "I hope for the best, but I'm worried we may be looking at weeks." Offer to explain what to expect functionally. Ask what they need to accomplish in that time. Never say "there's nothing more we can do" - always offer what you CAN do for comfort.
Clinical Examples
📋 Case 1 - Prognostication in Advanced Cancer
Scenario: 78M with metastatic colon cancer, no further chemo options. PPS 40% (mainly in bed, extensive disease, considerable assistance, normal-reduced intake, drowsy). Family asks: "How much time does he have?"
Assessment:
- PPS 40% - median survival ~4-6 weeks
- PPI: PPS 30-50 (2.5) + reduced intake (1) + no edema (0) + no dyspnea at rest (0) + no delirium (0) = 3.5 - predicts survival > 6 weeks
- Clinical trajectory: declining over past 2 weeks, now bed-bound most of day
- Lab markers: albumin 2.1, rising LDH - poor prognostic signs
Communication: "Based on what we're seeing, I'm worried we may be looking at weeks to a few months. I hope I'm wrong, but I want to make sure you have time to do the things that matter most to your family."
Teaching point: Use multiple tools (PPS + PPI + clinical trajectory + labs) rather than relying on any single predictor. Convergence of multiple indicators increases confidence.
📋 Case 2 - Prognostication in End-Stage Heart Failure
Scenario: 82F with NYHA Class IV HF, EF 15%, on home inotropes. Third hospitalization in 2 months. Not a transplant or LVAD candidate. Surprise question: "No, I would not be surprised if she died in the next 12 months."
Assessment:
- Seattle Heart Failure Model: estimated 1-year survival < 25%
- Recurrent hospitalizations despite optimized GDMT - marker of end-stage trajectory
- NYHA IV + inotrope dependence - median survival 6-12 months without advanced therapies
- Functional trajectory: organ failure pattern with stepwise decline and incomplete recovery between exacerbations
Communication: "Her heart is getting weaker despite our best treatments. Each hospitalization is harder to recover from. I think we're looking at months rather than years. I want to make sure we're focusing on what matters most to her."
Teaching point: Heart failure prognostication is harder than cancer because of the unpredictable exacerbation pattern. Patients may die suddenly from arrhythmia even when "stable." Use the surprise question to trigger earlier GOC discussions.
📋 Case 3 - Prognostication in Advanced Dementia
Scenario: 89F with advanced Alzheimer's, FAST stage 7C (nonambulatory, minimal verbal, requires total care). Recurrent aspiration pneumonia. Nursing home asks about feeding tube placement.
Assessment:
- FAST 7C - median survival ~12 months (range 6-24 months)
- Recurrent aspiration pneumonia in advanced dementia - 6-month mortality ~50%
- CASCADE study: feeding tubes do NOT prevent aspiration, do NOT improve survival, and increase agitation in advanced dementia
- Functional trajectory: slow progressive decline typical of dementia/frailty trajectory
Communication: "Your mother's dementia has reached an advanced stage where her body is losing the ability to swallow safely. A feeding tube would not prevent aspiration and studies show it doesn't help people with advanced dementia live longer. I'd recommend we focus on careful hand feeding for comfort and pleasure, and discuss what your mother would have wanted."
Teaching point: The Choosing Wisely campaign and multiple studies recommend against feeding tubes in advanced dementia. This is one of the clearest evidence-based recommendations in palliative care. Frame as what you CAN do (hand feeding, comfort), not what you won't do.
⚡ Summary
Why Tools Beat Intuition
Physicians systematically overestimate survival, on average by a factor of 3 to 5. The error grows with a longer doctor-patient relationship, and it delays hospice referral, which is how patients spend their last months in hospital rather than at home. Use a validated tool to supplement judgment, not to replace it.
Start With the Surprise Question
'Would I be surprised if this patient died in the next 12 months?' A 'no' should trigger goals-of-care planning. Sensitivity and specificity are each around 70%, which is imperfect but far better than unaided estimation, and it takes seconds. It works across cancer and non-cancer illness.
Match the Tool to the Population
PPS scores function from 0 to 100% in both cancer and non-cancer disease; PPS 50% or below suggests a median survival of about 6 months. PPI is cancer-specific and separates a survival under 3 weeks from over 6 weeks using performance status, oral intake, edema, dyspnea at rest and delirium. ePrognosis for community-dwelling older adults. APACHE for ICU mortality. Using a cancer tool on a heart failure patient gives a number with no validity.
Know the Illness Trajectory You Are In
Cancer: high function preserved until a fairly sharp terminal decline over weeks to months, which is what makes prognosis most predictable here. Organ failure (heart failure, COPD): a saw-tooth of exacerbations with incomplete recovery, and death often occurs during an exacerbation that looked survivable. Dementia and frailty: a slow decline over years, where the prognostic markers are weight loss, recurrent infection and loss of ambulation rather than any single test.
Communicate in Ranges
'Hours to days', 'days to weeks', 'weeks to months', 'months to years'. Never give a precise date: it will be wrong, and families anchor on it and lose trust when it passes. Ranges convey the same planning information while being honest about the uncertainty.
Ask Permission and Check Understanding
'Would it be helpful to talk about what to expect with time?' Some patients want detail, some want the broad picture, and some want the family told instead. Ask before you tell, then check what they took away. Name the uncertainty explicitly rather than hiding it: saying you cannot be certain builds credibility rather than undermining it.
Hope and Prognosis Are Not in Conflict
Use 'hope for, prepare for' framing: hope the treatment works while preparing for the possibility it does not. Do not offer false reassurance and do not strip hope away; both damage the relationship at the moment it matters most. Respond to emotion before adding information, since nothing said after a shock is retained.
Let the Estimate Change Something
A prognosis is only useful if it drives a decision: hospice referral, advance directive completion, deprescribing preventive medications, stopping burdensome monitoring, or bringing a family conversation forward. Revisit it as the clinical picture changes, because a prognosis given at diagnosis is stale a month later, and document it so the next team is not starting from scratch.
📄 One Pager
Text version
PALLIATIVE CARE · One Pager
Prognostication Tools
Each tool is validated for one population and answers one question. Use tools to correct clinical judgment, which systematically runs optimistic, then communicate in ranges, never in dates.
Which Tool, Which Population
- Surprise question (any serious illness): "Would I be surprised if this patient died in the next 12 months?" If no, start the palliative discussion. Sensitivity and specificity ~70%, so it is a screening trigger, not a prognosis.
- PPS, Palliative Performance Scale (cancer and non-cancer): functional status 0–100% in 10% increments across ambulation, activity and evidence of disease, self-care, intake, and consciousness. PPS ≤ 50% predicts median survival ~6 months; PPS ≤ 30% means days to weeks.
- PPI, Palliative Prognostic Index (cancer): combines PPS, oral intake, edema, dyspnea at rest, and delirium. PPI > 6 predicts survival < 3 weeks (sensitivity 83%, specificity 85%).
- PaP, Palliative Prognostic Score (cancer): dyspnea, anorexia, KPS, clinical prediction of survival, WBC, lymphocyte %. Sorts into group A (> 70% chance of 30-day survival), B (30–70%), C (< 30%).
- APACHE II/IV (ICU): ICU mortality prediction, higher score means higher mortality.
- MELD (liver disease): 3-month mortality in cirrhosis and transplant prioritization.
- Seattle Heart Failure Model (heart failure): 1–5 year survival in chronic HF.
- KPS vs ECOG (performance status): KPS 0–100 (100 normal), ECOG 0–5 (0 fully active, 5 dead). KPS < 40 or ECOG ≥ 3 signals limited treatment tolerance and poor short-term prognosis.
Bedside Time Frames
- Months: declining functional status, weight loss, increasing symptoms, frequent hospitalizations.
- Weeks: bed-bound most of the day, minimal oral intake, drowsy, dependent for all ADLs.
- Days: bed-bound, minimal consciousness, mottling, Cheyne-Stokes breathing, no oral intake.
- Hours: unresponsive, irregular breathing, cool and mottled extremities, mandibular breathing.
- These physical signs are what let you narrow the range you give the family, which is why the exam at the door matters more than any score in the last days.
Communicating Prognosis, Step by Step
1
Ask before telling: "How much information would you like to know about what to expect?" Not everyone wants numbers, and unwanted numbers close the conversation down.
2
Use ranges: "days to weeks," "weeks to months," never a specific date, because a date will be heard as a promise and remembered as one.
3
Frame honestly: "I hope for the best, but I'm worried we may be looking at weeks rather than months." Hope and honesty in the same sentence.
4
Acknowledge uncertainty: "No one can predict exactly, these are estimates based on what we know."
5
Describe the trajectory: increasing sleep, decreasing intake, less interaction. Telling families what dying looks like ahead of time prevents each change being read as a new crisis.
6
Never say "there's nothing more we can do": ask what they need to accomplish in the time available and offer what you can still do for comfort.
Decline Trajectories
- Cancer: preserved function, then rapid decline in the final weeks. Relatively predictable, so the onset of decline itself signals end-stage.
- Organ failure (CHF, COPD, ESRD, cirrhosis): gradual decline with acute exacerbations and partial recovery. Unpredictable, because any exacerbation may be the fatal one, which is why the surprise question earns its place here.
- Dementia and frailty: slow progressive decline over years. Very prolonged, so the terminal phase is hard to identify at all.
- Sudden death (MI, PE, stroke, trauma, arrhythmia): full function to death, no prognostic window, which is the argument for advance care planning in everyone.
What To Assess
- Functional status: PPS, Karnofsky, or ECOG, the strongest single bedside predictor across diseases.
- Nutrition: weight loss trajectory, BMI, intake history. Albumin reflects inflammation, not nutrition, per ASPEN 2021.
- Disease-specific markers: tumor markers, MELD, NYHA class, eGFR trajectory.
- Hospitalization frequency: readmissions mark a declining trajectory even when each individual admission looks survivable.
- Symptom burden: a rising symptom load suggests progression.
- Labs: albumin < 2.5, elevated LDH, leukocytosis, lymphopenia, elevated CRP, rising creatinine. Low albumin is the strongest single lab predictor across disease types.
- Patient and family understanding and what they want to know.
Pitfalls
- Trusting your own estimate: physicians overestimate survival by 3–5× on average, driven by optimism bias, therapeutic optimism, anchoring to the patient's best functional state instead of the current trajectory, and pressure from families who want hope.
- Applying a tool outside its population: PPI and PaP are validated in cancer, MELD in liver disease, Seattle in heart failure, APACHE in the ICU. A score's threshold does not transfer to another score's decision.
- Relying on one predictor: converge PPS plus PPI plus clinical trajectory plus labs; agreement between them is what raises confidence.
- Treating the surprise question as a prognosis: it screens for who needs the conversation, it does not estimate survival.
- Reading albumin as a nutrition marker and then treating the number instead of the disease.
- Feeding tubes in advanced dementia: they do not prevent aspiration, do not improve survival, and increase agitation. Offer careful hand feeding for comfort and pleasure instead.
Trigger The Goals Conversation When
- You would not be surprised if the patient died within 12 months.
- PPS ≤ 50% (median ~6 months) and certainly at PPS ≤ 30% (days to weeks), because at that point the window for planning is closing.
- PPI > 6 in a cancer patient, predicting survival < 3 weeks.
- Recurrent hospitalizations despite optimized therapy in organ failure, since each exacerbation may be fatal and a stable-looking patient can die suddenly from arrhythmia.
- Advanced dementia with recurrent aspiration, when the question of a feeding tube is being raised.
- Family asks "how long?" Answer with a range, not a refusal.
Key Evidence
- Christakis, 2000 physicians overestimate survival by 3–5× on average.
- Downar, 2017 surprise question performance for 12-month mortality, good enough to screen, not to prognosticate.
- Morita, 1999 PPI > 6 predicts survival < 3 weeks in cancer.
- Pirovano, 1999 PaP score validation, three 30-day survival groups.
- Lunney, 2003 four end-of-life trajectories, which is why prognostic communication differs by disease.
- Teno, 2012 feeding tubes do not improve survival in advanced dementia.
Related Topics
Code Status & Advance DirectivesFamily Meeting FrameworkGoals of Care & Symptom ManagementHospice Eligibility CriteriaNon-Opioid Symptom ManagementOpioid Rotation Conversion