Talk:Assessment of the health impacts of H1N1 vaccination
Evaluation of the H1N1 assessment
The evaluation was originally made for the purpose of demonstrating the application of the properties of good assessment framework in an article considering evaluation of model and assessment effectiveness, but in the end was omitted from the manuscript. Here the assessment of the health impacts of H1N1 vaccination is evaluated according to the application of the Properties of Good Assessment framework, originally published by Tuomisto and Pohjola (2007). For the purpose of this evaluation, the framework has been slightly updated as is described in Properties of good assessment.
On the course the assessment primarily served the purposes of illustrating essential aspects of decision analysis and risk management, as well as an explicating the swine flu (AH1N1 influenza) pandemic and related vaccination campaign in Finland 2009 - 2010 as examples of practical contexts for decision analysis and risk management. Here we focus on its secondary purpose: to evaluate the decision to launch a nationwide vaccination campaign to alleviate the pandemic in Finland. The discourse erupted during autumn 2010 when suspicions regarding a relationship between the AH1N1 vaccine and the sudden increase in prevalence of narcolepsy in Finland were publicized in the media. Soon after that the National Institute for Health and Welfare (THL) in Finland set up a task force to determine whether such a causal relationship exists (THL, 2011a,b).
The setting for evaluating the assessment:
- Time of assessment: spring 2011.
- Assessors: organizers of the DARM course, participants of the course.
- Intended primary user: Ministry of Social Affairs and Health in Finland.
- Intended use: basis for communication about public concerns regarding swine flu vaccines and narcolepsy.
- Evaluation method: qualitative expressions on a 5-point scale from very low to very high.
- Evaluation focus: whole assessment (main message, supported by all other information).
- Basis for evaluation: information provided on the assessment page in Opasnet, complemented with additional information obtained from assessment participants where necessary.
|Quality of content||Informativeness||Medium||The conclusion that vaccinating whole population was a better alternative than no vaccination is well supported by the model results. The conclusion that vaccinating whole population was a better alternative than vaccinating all except 5-19 old is more vaguely supported. The uncertainty of some variables in the model is high.|
|Calibration||Medium||Model results and assessment conclusions are in line with the analyses by the European Medical Agency (EMA, 2011) and the National narcolepsy task force in Finland (THL, 2011b). The calibration of some variables in the model, e.g. DALY weight for narcolepsy, may well be questioned as they are based on assumptions rather than data.|
|Coherence||High||The question is well addressed and the answer is reasoned with a model that takes account of the most important factors known to have effect on the outcome. Limitations in the comprehensiveness of the model and its parts exist, e.g. in terms of assumptions, but many of them identified and explicated. Value of information analysis indicates high coherence within the assessment.|
|Applicability||Relevance||Very low (potentially high)||The assessment addresses an issue that can be considered as underlying much of the discussion regarding the swine flu pandemic and the vaccination campaign. The assessment could thus be claimed to serve a real, existing need and the potential relevance could be considered high. However, in reality there was practically no interaction between the assessors and intended users and there was no demand from the users for the assessors to address the question. Consequently, the actual relevance is very low.|
|Availability||Low (potentially very high)||The model was developed and presented on a freely accessible assessment page in Opasnet at a time when many of the issues related to the case were still unresolved and under active public discussion. Practically no technical limits to availability exist. However, awareness about the assessment among intended users remained low despite (or due to only) the minor efforts of informing them by e-mail. Consequently, the relevance is actually low, although potentially very high.|
|Usability||Medium||The main message and its basis is presented in a structured way and is relatively easily perceivable even for non-experts. Assumptions and limitations are described and access to the data and calculations is provided and easy. However, obtaining a deeper and detailed understanding of the model requires specific knowledge, detailed scrutiny, and possibly also assistance from the developers. Explication of the intended use and guidance of use are omitted from the assessment (cf. use plans in Vermaas and Houkes, 2006). Due to lack of interaction with users the actual usability in intended use is unknown.|
|Acceptability||Medium||The open approach can be considered to have increased acceptability in a situation where authorities were accused for non-warranted withholding of important information. On the other hand the model was developed by non-experts regarding infectious diseases and vaccines. Also the credibility of the organization, THL, that the main developers represented was strongly questioned in public at the time of developing and delivering the model. Due to lack of interaction with users the actual acceptability in intended use is unknown.|
|Efficiency||Intra-assessment efficiency||High||The assessment was developed as a side product of the DARM course. The development of the model consumed about 2 person months work, consisting mostly of the efforts of the course assistant, a high school graduate with good mathematical and computer skills, but no prior specific expertise on vaccines or infectious diseases.|
|Inter-assessment efficiency||High||The assessment is mostly structured as independent variables that are applicable in other assessments. The assessment also applies some variables that were developed in previous assessments. However, the calculation in the model for the most part was not coded as independently applicable modules.|
|Overall evaluation||Effectiveness||Low||Despite mostly relatively good scores with regard to many properties, the overall effectiveness remains low because the intended use did not take place in reality. The potential for outcome effectiveness can be seen, but the failure of the delivery, i.e. lack of interaction between the assessment and its use, prevents it from becoming realized. The realized impacts are mostly process effects, increasing the knowledge among the participants of the assessment. As many of the participants work in roles that are relevant to the interests of the intended user, the Ministry of Social Affairs and Health, it can be assumed that some of that knowledge will eventually trickle to its intended use, but indirectly and with delay.|
The example assessment can be considered as somewhat typical in the sense that it fails to convey its as such good results into practice. Although the evaluation example above can be considered somewhat superficial, and is made only qualitatively, it highlights some important aspects of assessment and model performance:
- In terms of outcome effectiveness, there is a major difference between the potential of an assessment or model to deliver its intended outcomes and the actual delivery thereof.
- The properties that have been least addressed within the common contemporary approaches to performance, namely relevance and availability, are critical for transforming the potential of an assessment or model to effectiveness.
- The delivery of the assessment or model outputs to their intended use is necessary to take account of in considering assessment and model performance.
- Improving effectiveness of assessments and models is not an issue to be addressed within the communities of assessment and modelling, but requires simultaneous development of the use processes and the capacity of policy making to make use of what assessments and models can deliver.
The major limitations of the assessment indicated by the evaluation according to the properties of good assessment may seem apparent, but they would probably not show up in evaluations applying more conventional approaches. Altogether, the example shows that, despite still lacking explicit methods for its application, the Properties of Good Assessment framework can already be a useful and powerful means for evaluating and managing assessment and model performance.
EMA, 2011. Press release 27 July 2011: European Medicines Agency recommends restricting use of Pandemrix. European Medicines Agency. Available: http://www.ema.europa.eu/docs/en_GB/document_library/Press_release/2011/07/WC500109182.pdf
THL, 2011a. National narcolepsy task force interim report 31 January 2011. National Institute for Health and Welfare (THL), Helsinki. Available: http://www.thl.fi/thl-client/pdfs/dce182fb-651e-48a1-b018-3f774d6d1875
THL, 2011b. National narcolepsy task force final report 31 August 2011 (in Finnish). National Institute for Health and Welfare (THL), Helsinki. Available: http://www.thl.fi/thl-client/pdfs/c02a3788-a691-47a4-bca8-5161b6cff077
Vermaas, P.E., Houkes, W., 2006. Technical functions: a drawbridge between the intentional and structural natures of technical artefacts. Studies in History and Philosophy of Science 37, 5-18.
Pandemrix should not be used because of narcolepsy risk
|Fact discussion: . (Pandemrix)|
|Statements: Pandemrix should not be used any more anywhere because its narcolepsy risk is too high.
Resolution: Not accepted. Pandemrix is still an effective and safe vaccine. However, due to precautionary reasons, other alternatives should be used when available, because the occurrence of narcolepsy is not understood.
(Resolved, i.e., the resolution has been updated to the main page.)
Discussion groups (Decision analysis and risk management 2011):
- Talk:Assessment of the health impacts of H1N1 vaccination/Group A
- Talk:Assessment of the health impacts of H1N1 vaccination/Group B
- Talk:Assessment of the health impacts of H1N1 vaccination/Group C
- Talk:Assessment of the health impacts of H1N1 vaccination/Group D (Finnish material)
- Talk:Assessment of the health impacts of H1N1 vaccination/Group E (Finnish material)
- Rokotusinfo: Swine flu
- YLE: EU agency does not find link between Pandemrix and narcolepsy
- European Centre for Disease Prevention and Control (ECDC): Questions and answers
- THL press release 9 Dec 2010
- THL press release 25 Aug 2010
- Helsingin Sanomat: Arkkiatri moittii sikainfluenssarokotteen vastustajia (in Finnish)
- WHO Europe: Pandemrix® vaccine and increased risk of narcolepsy
- WHO Global Advisory Committee on Vaccine Safety: Statement on narcolepsy and vaccination
- European Medicines Agency (EMA): Information page on Swine flu
- PreventDisease.com: Total of 2300 Reports of Adverse Reactions From Pandemrix Vaccine in Sweden
- THL recommends to stop the use of Pandemrix