Using computable evidence for priorization of interventions ��GIN 2025� �Ilkka Kunnamo 19.9.2025 ��Duodecim Publishing Company, Helsinki, Finland�Primary Care Centre of Central Finland, Karstula, Finland�Board member, Scientific Knowledge Accelerator Foundation�

1

24.9.2025

No other conflicts of interest related to this presentation

Computable effect estimates

• The summary of findings (SOF) table contains the effect estimates for all outcomes and their confidence intervals, for bot benefits and harms
• The absolute effect depends on the baseline risk of the event of interest (the higher the baseline risk the larger the effect)
• The balance of benefits and harms depends on the relative importance of each outcome
• Net effect = sum of importance-adjusted effects for all outomes.

​

Death equivalent �–a comparable outcome for all patient groups?

Pays respect to the severity of the disease*

(and, hence, helps to consider the importance of possible gains of a recommendation)

​

​

Severity = impact on quality of life (EQ-5D utility)

�Examples of utilities**�Perfect quality of life: Utility 1 �Death: Utility 0�Stroke: Utility 0.587��* Djulbegovic et al. J. of Eval in Clin practice 2025

** Saarni et al Qual Life Res 2006

Calculating the Death Equivalent (DE)

�Stroke as Death Equivalents�= perfect quality of life (1) – stroke utility (0.587) �= 0.413 DE

(= disutility caused by stroke in death equivalents) ��Practical translation

1 DE = 2.4 strokes = 3,2 AMIs = ...

https://fevir.net/resources/Composition/399948�

https://fevir.net/resources/Composition/399948�

0.35 x 2.3

https://fevir.net/resources/Composition/399948�

The corresponding net effect estimate for patients with vascular disease is 3.39 (NNT 25)

From net effect (in death equivalents) to number needed to treat (NNT)

• The meaning of the magnitude of net effect (NE) can be understood as follows:
• If the anchor outcome is death, and the NE is 1.38, this means that 1.38 out of 100 patients will avoid a composite value-adjusted outcome that is as important as death (= a death equivalent): The absolute risk reduction (ARR) for that composite outcome is 0.0138.
• The number needed to treat (NNT) to avoid that outcome is 1/ARR = 1/0.0138 = 72.
• Determining the NNT is important when the resource requirements of providing an intervention are considered.
• If the intervention (e.g drug) is cheap, the most important resource requirement is the clinicians’ time

The use of net effect estimate (1)

• A numeric value for the balance of benefits and harms -> a determinant of the strength of recommendation
• For an individual patient, interventions for various health conditions could be ordered based on their net impact on health. This is particularly useful as part of clinical decision support

The use of net effect estimate (2)

• In population health, patients can be stratified by their risks for (e.g. cardiovascular) events in order to provide interventions for those who would benefit most.

Blood pressure distribution in a population of diabetic patients

In which order should we contact patients for prescribing SGLT-2 for patients with chronic kidney disease (CKD) or heart failure (HF) or lipid-lowering drugs for patients with cardiovascular disease and LDL >= 1.8 in a population of 27000 people (preliminary calculation – will be checked and updated)

4.

Events to be prevented (death equivalents)

CKD, very high risk group NNT-DE 8

CVD, high risk group LDL >= 1.8 NNT-DE 25

HF and diabetes NNT-DE 17

Patients to be contacted

Patients per one physician/nurse team (population 1800) to be contacted: 153

37.5 1349

21102

2110

​

2299

​

77

437

189

199

Total 81

NNT to prevent one death equivalent

Moderate risk�(MVE risk, 5 years: <5%)�(Finrisk 2-9.9%)

2989 patients, �with LDL >2.6

(average 3.3)

High risk�(MVE risk, 5 years: 5-9.99%)�(Finrisk 10-14.99%)

44451 patients,�with LDL>2.6

(average 3.3)o

Very high risk

(MVE risk 10-19,99%, Finrisk > 15%)

48 937 patients with

LDL > 1,8 (av. 2,5)

LDL 1,0

LDL 1,4

LDL 1,8

LDL 2,6

3,3

2,3

1,5

2,5

1,5

1,1

NNT (DE): 108

Cost € / DE: 5731

3,3

• 6 CV deaths
• 12 strokes
• 41 CHD
• 24 DE

NNT (DE): 212

Cost € / DE: 10 145

Prevented events �per 10 years

NNT and cost

(€ per prevented DE)

NNT (DE): 88

Cost € / add. DE: 49 749

(prop. of PCSK-9: 20%)

NNT (DE): 155

Cost € / DE: 62 367

(prop. of PCSK-9: 20%)

NNT (DE): 122

Cost € / add. DE: 226 978

(prop. of PCSK-9: 30%)

0,7

ADDITIONAL

BENEFIT 2,3 -> 1,5

ADDITIONAL BENEFIT 1,5 -> 1,1

ADDITIONAL BENEFIT 1,1 -> 0,7

NNT (DE): 121

Cost € / add. DE: 583 448

(prop. of PCSK-9: 70% )

​

Finnish DM2 patients (<75-yrs) �Diabetes Registry 8/2024

HIGH RISK

3,3 -> 2,3

MODERATE

RISK 3,3 -> 2,3

ABBREVIATIONS

DE = Death equivalent

CV death = Stroke = aivohalvaus

CHD = hronic Heart Disease (coronary revascularisationi MVE = Major Vascular Event

PAD = Periferic Artery Disease

Finrisk = Finnish national cardiovascular calculator

• 169 CV deaths
• 386 strokes
• 934 CHD
• 624 DE

• 135 CV deaths
• 312 strokes
• 761 CHD
• 504 DE

• 159 CV deaths
• 167 strokes
• 540 CHD
• 12 PAD
• 402 DE

• 159 CV deaths
• 167 strokes
• 545 CHD
• 12 PAD
• 404 DE

• 396 CV deaths
• 415 strokes
• 1322 CHD
• 29 PAD
• 992 DE

VERY HIGH RISK 2,5 -> 1,5

COMPARE TO OTHER GUIDELINE RECOMMENDATIONS

This presentation and additional information https://bit.ly/4gA9Btk

Thank you!

ilkka.kunnamo@duodecim.fi

The next slides are from another presentation describing how the calculations in slide 12 were made

A tool for more ethical, effective and feasible guidelines? �A case study with LDL-targets for Finnish type 2 diabetes population��GIN 2025�Tahkola Aapo, MD Kunnamo Ilkka, MD (3,4) ��The Wellbeing Services County of Central Finland, Duodecim Publishing Company, Helsinki, Finland�Jyväskylä, Finland Primary Care Centre of Central Finland, Karstula, Finland�Finnish Insitute for Health and Welfare, Helsinki, Finland���

16

24.9.2025

No other conflicts of interest related to this presentation

Treat more and more, lower and lower risks

LDL target for (very) high risk patients

• 1988 (ATP): < 3,4 mmol/L (<130 mg/dL)
• 1993 (ATP): <2,6 mmol/L (<100 mg/dL)
• 2004 (ATP): <1,8 mmol/L (<70 mg/dL)
• 2019 -2025 (ESC/EAS): <1,4 mmol (<55 mg/dL)

(Optional <1,0 mmol/L (<40 mg/dL)

​

Jones, J et al. Journal of Clinical Medicine 2023

2025 Focused Update of the 2019 ESC/EAS Guidelines

More health? Yes

​

But are we emphasizing enough…

• Scarce resources?
• Feasibility of guidelines?
• Overall effectiveness of guidelines?
• Equity and ethical viewpoint?

​

We can’t continue like this

What to do?

1. Develop a comparable outcome for all guidelines
2. Use it to create a comparable estimate of the gains and costs of new recommendations
3. Use the estimates in guideline development

Death equivalent �–a comparable outcome for all patient groups?

Pays respect to the severity of the disease*

(and, hence, helps to consider the importance of possible gains of a recommendation)

​

​

Severity = impact on quality of life (EQ-5D utility)

�Examples of utilities**�Perfect quality of life: Utility 1 �Death: Utility 0�Stroke: Utility 0.587��* Djulbegovic et al. J. of Eval in Clin practice 2025

** Saarni et al Qual Life Res 2006

Calculating the Death Equivalent (DE)

�Stroke as Death Equivalents�= perfect quality of life (1) – stroke utility (0.587) �= 0.413 DE

(= disutility caused by stroke in death equivalents) ��Practical translation

1 DE = 2.4 strokes = 3,2 AMIs = ...

Estimating the gains and costs of a recommendations using Death Equivalents�Case study: LDL target for Finnish DM2 patients, real life data

Etunimi Sukunimi

20

24.9.2025

The Finnish Diabetes Registry

• All 400 000 + Finnish DM2 patients
• Stratified in two risk groups (moderate-high and very high risk)
• Last LDL measurement (coverage 76-86%)

If we treated all patients that are not-in-target

to the guideline informed targets…

Moderate risk�(MVE risk, 5 years: <5%)�(Finrisk 2-9.9%)

2989 patients, �with LDL >2.6

(average 3.3)

High risk�(MVE risk, 5 years: 5-9.99%)�(Finrisk 10-14.99%)

44451 patients,�with LDL>2.6

(average 3.3)o

Very high risk

(MVE risk 10-19,99%, Finrisk > 15%)

48 937 patients with

LDL > 1,8 (av. 2,5)

LDL 1,0

LDL 1,4

LDL 1,8

LDL 2,6

3,3

2,3

1,5

2,5

1,5

1,1

NNT (DE): 108

Cost € / DE: 5731

3,3

• 6 CV deaths
• 12 strokes
• 41 CHD
• 24 DE

NNT (DE): 212

Cost € / DE: 10 145

Prevented events �per 10 years

NNT and cost

(€ per prevented DE)

NNT (DE): 88

Cost € / add. DE: 49 749

(prop. of PCSK-9: 20%)

NNT (DE): 155

Cost € / DE: 62 367

(prop. of PCSK-9: 20%)

NNT (DE): 122

Cost € / add. DE: 226 978

(prop. of PCSK-9: 30%)

0,7

ADDITIONAL

BENEFIT 2,3 -> 1,5

ADDITIONAL BENEFIT 1,5 -> 1,1

ADDITIONAL BENEFIT 1,1 -> 0,7

NNT (DE): 121

Cost € / add. DE: 583 448

(prop. of PCSK-9: 70% )

​

Finnish DM2 patients (<75-yrs) �Diabetes Registry 8/2024

HIGH RISK

3,3 -> 2,3

MODERATE

RISK 3,3 -> 2,3

ABBREVIATIONS

DE = Death equivalent

CV death = Stroke = aivohalvaus

CHD = hronic Heart Disease (coronary revascularisationi MVE = Major Vascular Event

PAD = Periferic Artery Disease

Finrisk = Finnish national cardiovascular calculator

• 169 CV deaths
• 386 strokes
• 934 CHD
• 624 DE

• 135 CV deaths
• 312 strokes
• 761 CHD
• 504 DE

• 159 CV deaths
• 167 strokes
• 540 CHD
• 12 PAD
• 402 DE

• 159 CV deaths
• 167 strokes
• 545 CHD
• 12 PAD
• 404 DE

• 396 CV deaths
• 415 strokes
• 1322 CHD
• 29 PAD
• 992 DE

VERY HIGH RISK 2,5 -> 1,5

COMPARE TO OTHER GUIDELINE RECOMMENDATIONS

Conclusions

• With the use of common comparable outcome (death equivalent), we can help to creat more ethical (equal), effective and feasible guidelines
• These methods can be used across different health conditions and interventions to inform guideline development
• To include harms, we need to apply net health benefit –principle Alper, Oetken, Kunnamo et al. BMJ Open 2019

​

Thank you!

​

aapo.j.tahkola@hyvaks.fi

ilkka.kunnamo@duodecim.fi

​

This presentation with additional info and references: http://bit.ly/46jznNG

​

​

References for the estimate presented in slide 6

Outcomes and effectiveness

• Cholesterol Treatment Trialists' (CTT) Collaborators. Lancet. 2012. DOI: 10.1016/S0140-6736(12)60367-5
• Heart Protection Study Collaborative Group. Lancet 2011. DOI: 10.1016/S0140-6736(11)61125-2
• Burger et al. Atherosclerosis. 2024. DOI: 10.1016/j.atherosclerosis.2024.118540

Health related quality of life

• Saarni et al.. Qual Life Res. 2006. doi: 10.1007/s11136-006-0020-1. Epub 2006 Sep 8. PMID: 16960751.

Target reaching

• Marret E et al. Vasc Health Risk Manag 2014. doi: 10.2147/VHRM.S54886
• C Blaum et al. European Heart Journal 2020 https://doi.org/10.1093/ehjci/ehaa946.3006
• Karalis, D et al. Cholesterol 2012 DOI: 10.1155/2012/861924
• Marret E et al. Vasc Health Risk Manag 2014. doi: 10.2147/VHRM.S54886

Cost of work

• Terveydenhuollon yksikkökustannukset 2017 (Unit costs of health care in Finland 2017)

Respecting the seriousness of illness and net health benefit

• Alper BS, Oettgen P, Kunnamo I, et al. BMJ Open 2019. DOI: 10.1136/bmjopen-2018-027445
• Djulbegovic B, Hozo I, Kunnamo I, Guyatt G.. J Eval Clin Pract. 2025. doi: 10.1111/jep.70051.

​

​

​

​

​

​

​

​

​

Costs

• Main driver: Medications
• In addition, we included the use of time of professionals
• 10 min/year of doctor’s and nurse’s time for all drug treatments�Quite conservative estimation

Cost of work�Doctor's work: 166 eur/hour�Nurse's work: 70 eur/hour�Source:Terveydenhuollon yksikkökustannukset 2017

The proportion of patients needing PCSK-9 to reach the target

25

24.9.2025

Methods for the estimation (slide 6)

• The calculation is based on a meta-analysis of the benefits of statins (1)
• Follow-up ca. 5 years, including low-risk patients and diabetics�Benefits estimated over 10 years, assuming the annual risk remains the same (2). This probably slightly underestimates the benefits of the treatment. �Baseline risk: Finrisk (Finnish national cardiovascular risk calculator) 10 years or corresponding risk category in CTT (5 years), see below
• The risk reduction per LDL unit is essentially the same regardless of the treatment (statin, ezetimibe, PCSK-9) (3)

1) Cholesterol Treatment Trialists' (CTT) Collaborators. Lancet. 2012

2) Effects on 11-year mortality and morbidity of lowering LDL cholesterol with simvastatin for about 5 years in 20,536 high-risk individuals: a randomised controlled trial. Lancet 2011

3) Burger et al. Course of the effects of LDL-cholesterol reduction on cardiovascular risk over time: A meta-analysis of 60 randomized controlled trials. Atherosclerosis. 2024

Finrisk 2-9,9%

Finrisk 10-14.99%

Finrisk 15% tai yli

Picture: CTT 2012 (1)

Comparison of the result with an 11-year follow-up study*

• In the simvastatin group, the risk for MVE was 4.2% per year for the first five years (Figure 2)
• Cumulative risk over 10 years (Figure 3) approx. 36% vs. our estimation approx. 35%

*Effects on 11-year mortality and morbidity of lowering LDL cholesterol with simvastatin for about 5 years in 20,536 high-risk individuals: a randomised controlled trial. Lancet 2011