Category Archives: Research

Discrepâncias entre resultados pré-especificados e reportados

15 Fri Feb 2019

Posted by in Class Content, Escrita científica, Must read, Research

≈ Leave a comment

Tags

,

A mais recente publicação de Ben Goldacre é um estudo de coorte prospectivo que dá conta da real dimensão do problema das discrepâncias entre os outcomes pré-especifícados (antes do estudo) e os outcomes reportados (após o estudo) em 67 ensaios clínicos publicados num de cinco principais jornais científicos em saúde – New England Journal of Medicine, The Lancet, Journal of the American Medical Association, British Medical Journal, and Annals of Internal Medicine.[1] O estudo considera ainda as tentativas de corrigir essas discrepâncias.[2] Todos estes jornais estão publicamente comprometidos com o CONSORT (Consolidated Standards of Reporting Trials). Para ler, com preocupação, em duas partes:

1. COMPare: a prospective cohort study correcting and monitoring 58 misreported trials in real time

2. COMPare: Qualitative analysis of researchers’ responses to critical correspondence on a cohort of 58 misreported trials

 

The Global Gender Gap Report

19 Wed Dec 2018

Posted by in Must read, Research

≈ Leave a comment

Tags

,

The World Economic Forum just issued The Global Gender Gap Report and the best-ranked countries are:

GenderGap_2018_2018_top10_global-1.png

The best-ranked countries in western Europe are:

GenderGap_2018_2018_globaltop10_westerneurope-1.png

Progress has been done and there is an improvement to close the gap. However, at this rate, it will take 108 years for the world to close the Gap. In western Europe, it may be about 61 years.

2018_Regions_to_Close_Gap.png

Portugal dropped to the 37th position compared to the 33rd position in 2006. The economic and educational indicators also dropped from the 33rd to the 44th and from the 57th to the 82nd positions respectively. Only the health and survival indicators have improved from the 71st to the 54th position. Regarding gender inequality, we still have a long road ahead of us. You can further explore the data here.

gender gap portugal.jpg

 

They finally did it | Parachute use to prevent death and major trauma when jumping from aircraft

17 Mon Dec 2018

Posted by in Class Content, Must read, Research

≈ 1 Comment

Tags

,

A randomized controlled trial about parachute use to prevent death and major trauma when jumping from aircraft was finally published by Robert W Yeh in BMJ.

Since 2003 Smiths’ Systematic Review that was unable to identify any randomised controlled trials of parachute intervention and thus concluded that the effectiveness of parachutes had not been subjected to rigorous evaluation by using randomised controlled trials, the world was expecting such a trial.

aviao.jpgThe conclusions: ‘Parachute use did not reduce death or major traumatic injury when jumping from aircraft in the first randomized evaluation of this intervention. However, the trial was only able to enroll participants on small stationary aircraft on the ground, suggesting cautious extrapolation to high altitude jumps. When beliefs regarding the effectiveness of an intervention exist in the community, randomized trials might selectively enroll individuals with a lower perceived likelihood of benefit, thus diminishing the applicability of the results to clinical practice’.

Scientific Publications | Christmas Pearls

13 Thu Dec 2018

Posted by in Class Content, Escrita científica, Must read, Research

≈ 2 Comments

Tags

,

Several journals have developed a tradition of publishing slightly different papers in their December issues. These papers are intended to either make a provocative statement, making us reflect upon unusual topics, or provide unusual examples of scientific topics. I like to call them Christmas Pearls. This year my favorite Christmas Pearls:

“Death is certain, the time is not”: mortality and survival in Game of Thrones

Kaplan-Meier survival analysis with Cox proportional hazard regression modeling was used to quantify survival times and probabilities to identify independent predictors of mortality among ‘important characters’ (n = 330) appearing in Seasons 1 to 7 of Game of Thrones.

Although the statistics are quite sound and the data extensive some remarks are mandatory.

First, resuscitation bias was not considered (e.g. John snow). Neither were white walkers. Not clear what justifies the exclusion of these non-less important characters. From a mixed-methods perspective, white walkers could have participated in a focus group discussion of the results, for example.

Secondly, the effectiveness of wall policies for security purposes is clearly understated.

Finally, the authors conclude:

“There is great potential for preventing violent deaths in the world of Game of Thrones. Stable democratic governments, resilient institutions that deliver public goods, and implementation of evidence-based violence prevention policies can decrease the risk of violent deaths considerably“.

Given that winter is coming this is clearly an unrealistic and utopic worldview of Game of Thrones.

 

Is it time to start using the emoji in biomedical literature?

 

The lack of standardisation in emoji artwork that may cause ambiguity in interpretation is quite disturbing. Also, the emoji based alternatives to the denotation of statistical significance can’t really apprehend the need to consider the enunciation of the null hypothesis for adequate interpretation of P-values. The smiling-face-with-sunglasses_1f60e.png in a Kolmogorov-Smirnov to check distribution assumptions for parametric testing may be actually hiding

crying_emoji_small_classic_round_sticker-r4c998b077b254fe69908fa4ca57ce521_v9waf_8byvr_540.jpg.

table 01

F8.large.jpg

Research on self-treatment of “writer’s block” – a timeline critical appraisal

05 Fri Oct 2018

Posted by in Escrita científica, Must read, Opinião, Research

≈ 3 Comments

Tags

,

Research on the self-treatment of ‘Writer’s Block’ is vast and diversified. You might enjoy reading these articles, they are a fun and fast read. Please, do pay attention to the notes on them. My critical appraisal on the research about ‘Writter’s Block’ may be found at the end of this post.

1974 – the classical CASE REPORT by Dennis Upper:

  • Upper, Dennis (Fall 1974), “The unsuccessful self-treatment of a case of “writer’s block””, Journal of Applied Behavior Analysis, 7 (3): 497, doi:10.1901/jaba.1974.7-497a PDF Here

1983 – a REPLICATION study is performed by Molloy:

  • Molloy, G. N. (1983). The unsuccessful self-treatment of a case of “writer’s block”: A replication. Perceptual and Motor Skills, 57(2), 566-566. PDF Here

1984 – a new REPLICATION, however it partially failed, by Hermann:

  • Hermann, B. P. (1984). Unsuccessful self-treatment of a case of “writer’s block”: a partial failure to replicate. Perceptual and Motor Skills, 58(2), 350-350. PDF Here

1984 – a first REVIEW OF THE LITERATURE on the topic. Still, it was a narrative literature review, we have to wait until 2014 for the first systematic review and meta-analysis

  • Olson, K. R. (1984). Unsuccessful Self-Treatment of “Writer’s Block”: A Review of the Literature. Perceptual and Motor Skills, 59(1), 158–158. doi:10.2466/pms.1984.59.1.158 PDF Here

1985 – a CLINICAL TRIAL by Skinner:

  • Skinner, N. F., Perlini, A. H., Fric, L., Werstine, E. P., & Calla, J. (1985). The Unsuccessful Group-Treatment of “Writer’s Block”. Perceptual and Motor Skills, 61(1), 298-298. PDF Here

1996 – a COHORT study by Skinner:

  • Skinner, N. F., & Perlini, A. H. (1996). The unsuccessful group treatment of “writer’s block”: a ten-year follow-up. Perceptual and motor skills, 82(1), 138-138. PDF Here

2007 – a MULTICENTER REPLICATION study, an attempt to address statistical power issues by Didden:

  • Didden, R., Sigafoos, J., O’Reilly, M. F., Lancioni, G. E., & Sturmey, P. (2007). A multisite cross‐cultural replication of upper’s (1974) unsuccessful self‐treatment of writer’s block. Journal of applied behavior analysis, 40(4), 773-773. PDF Here

2014 – an evidence synthesis in a META-ANALYSIS by McLean:

  • McLean, D. C., & Thomas, B. R. (2014). Unsuccessful treatments of “Writer’s Block”: a meta-analysis. Psychological reports, 115(1), 276-278. PDF Here
Captura de ecrã 2018-10-05, às 16.11.03

McLean, D. C., & Thomas, B. R. (2014). Unsuccessful treatments of “Writer’s Block”: a meta-analysis. Psychological reports, 115(1), 276-278

 

After systematic and iterative readings on the ‘Writer’s Block’ scientific literature my critical appraisal is

 

 

If you have updates on this topic, please let me know

Acknowlegements: to Professor Pedro Lopes dos Santos and Professor Isabel Menezes for sharing with me several ideas on “Writer’s Block” and useful literature about it. I would like to say to them tha

 

 

 

p-Hacking – A call for ethics

04 Thu Oct 2018

Posted by in Research, Statistics

≈ Leave a comment

Tags

p-hacking consists of the exhaustive exploitation of data through the use of different analytical models and/or the manipulation of the application criteria of these models until statistically significant results are obtained.

p-hacking can be described in a simpler manner has the art of torturing data until they confess something.

While publication bias removes from the scientific literature true or false negatives the p-hacking brings to the scientific literature true or false positives.

Conditioned scientific literature (i.e. the absence of false negatives and the presence of false positives) will bias the results of secondary studies aiming to synthesise scientific evidence, such as meta-analyses, that inform clinical guidelines and evidence-based decision making.

Here, I present a call for health ethics committees to assess the manifestation of researchers’ analytical intent in research protocols (i.e. pre-specified or exploratory) as a way to help prevent and further study the p-hacking bias.

p-Hacking – A call for ethics | PDF. Available from: https://www.researchgate.net/publication/324231282_p-Hacking_-_A_call_for_ethics

Sobre o Significado da Significância Estatística

04 Mon Jun 2018

Posted by in Class Content, Research, Statistics

≈ Leave a comment

Tags

A obtenção de resultados estatisticamente significativos (i.e. valor p < 0,05) é muitas vezes motivo de apreço e de celebração entre investigadores, mas o seu significado nem sempre é compreendido e utilizado de forma adequada. O uso do valor p como principal, senão único, elemento de suporte estatístico às conclusões em investigação científica tem sido bastante criticado e merece atenção.

A afirmação de que algo é significativo pode ser muito subjectiva. Contudo, em estatística o termo ‘resultado estatisticamente significativo’ traduz um consenso objectivo cujo significado é amplamente aceite. Porém, o que significa que o resultado é estatisticamente significativo? Considera-se um resultado significativo caso o valor p obtido seja inferior a 0,05. E o que significa que o valor p seja inferior a 0,05?

O uso de valor p ganhou notoriedade através do trabalho de Ronald Fisher que o definiu em 1925 como: ‘the probability of the observed result, plus more extreme results, if the null hypothesis were true’. Ou seja, o valor p é a probabilidade de o resultado observado, ou um resultado ainda mais extremo, ocorrer se a hipótese nula fosse verdadeira. Convencionou-se, posteriormente, que um valor p < 0,05 traduziria um resultado estatisticamente significativo (i.e. critério alpha).

Assim, um resultado observado numa dada estatística (e.g. um valor do coeficiente de correlação) com menos de 5% de probabilidade de ocorrer, por mero acaso, caso a hipótese nula se verificasse na população (i.e. H0 : r = 0 na população) sugere que a hipótese nula poderá ser rejeitada e que se poderá encarar a hipótese alternativa (i.e. H1 : r ≠ 0 na população). Assim, por exemplo, a hipótese de que na população (de onde veio a amostra) r será diferente de zero tem alguma sustentação e, de acordo com esta abordagem estatística, a melhor estimativa que temos para esse r é o valor estimado na amostra. A incerteza associada a essa estimativa pode ainda ser calculada (e.g. intervalo de confiança a 95%). A conclusão do estudo estatístico suporta assim a hipótese alternativa e aponta uma estimativa ou conjunto de estimativas (i.e. intervalo de confiança) para o valor da população em estudo (i.e. parâmetro).

Probabilidade e hipótese são termos chave do uso apropriado do valor p e da noção de significância estatística. continuar a ler…