A short presentation I gave to Clincial Officers working in a Tanzanian Hospital about the importance of reading and thinking yourself about new medical ideas and insights.
Book Paid Powai Call Girls Mumbai 𖠋 9930245274 𖠋Low Budget Full Independent H...
Valuing research
1. Valuing Research by Clinicians Dr Christian van Rij @ 18th February 2010 Continuing Education for Clinical Officers at Wasso Hospital
2. Research Biochemical Inventions Treatment Insights Renewed ideas for the Best Current Treatment Plans Highest Patient Benefit Lowest Side Effects Cost Efficient
3. Guided by Guidelines Agreed summery of Best Treatment Plans Low value if based on opinions/ideas High value if based on Research Evidence Should be up to date: old ideas are not modern!
4. Guided by Seminars To remind or update your knowledge Should be applicable Should be reflecting the newest Evidence Based Medicine (and not merely opinions)
5. Being Told vs Reading yourself http://www.flickr.com/photos/90306674@N00/3196584510
6. Being Told vs Reading yourself Not always somebody present to guide you in the issue Road signs may be outdated Overcome by getting updated maps & reading them yourself! http://www.flickr.com/photos/waltzer/2428713783
8. Use proper statistical methods A man with red colors is 129 times more likely to be gored by a bull! But what is the Base Rate? www.flickr.com/photos/jagf/4249757103/in/pool-toros
9. Type of Research Study Observational Experience (Retrospective) Cohort Study Randomized Clinical Trial Preferably Prospective and Double Blind A Meta-analysis Study can be applied Likeliness of describing a true causative relationship
10. How does a RCT work? Patient group Intervention group Control group PICO Outcome Outcome Statistical Difference?
11. Critical Appraisal Satisfied with the patient group? Confounding factors which were not randomized? Sufficient external validation of the group? Proper hypothesis for a causative relationship? True correlation?
12. Stork Population and number of newborns; January 2004; Paediatric and Perinatal Epidemiology; volume 18:1:P88-09 http://nl.wikipedia.org/wiki/Bestand:Stork_with_new-born_child.png Stork Population & Baby Birth Rates
14. Positive Correlation Increasing the intervention increases the outcome Example: adrenaline as intervention and heart rate as outcome Relative Risk > 1 Effect
15. Negative Correlation Increasing the intervention decreases the outcome Example: beta-blocker as intervention and heart rate as outcome Relative Risk 0 < 1 Effect
17. Correlation maybe by Chance only? P-Value Describes the Probability that the correlation was caused by chance only It is agreed that a P-value below 0.05 is small enough to assume that the correlation did not happen by chance
18. Key Points Don’t go for personal experience as evidence: this is the level of traditional healers Instead of reading only road signs (which might not always be present or up to date), learn to read maps yourself! If you want to be a sharp doctor: make it a habit to sharpen yourself regular! Not updating yourself will make you outdated!
New Biochemical DevelopmentsACE-inhibitorsAntionesin-Receptor AntagonistsTreatment DevelopmentsLower Uterine Segment incisions for CSLaparoscopic procedures instead of LaparotomyRenewed Ideas about best current treatment plans
Although cultural values can be reflected in guidelines: A drug for infertility might be acceptable in a Maasai community but not in European communityA treatment in a NICU that causes blindness but prevents mental retardation: what is experienced worse can differ per cultureBut cultural values should not be simply assumed: they should studied, understood and quantified by Social Sciences.
Obervational Experience:It is my experience that … we treated 8 patients with …. and sawRetrospective Study:We identified a group of 1000 people who ate a lot of fruit, andWe identified a group of 1000 people who were not eating fruitWe saw that in the group of fruit eaters there was a lot of malaria, So we conclude that eating fruit is a risk factor for developing malariaWhen there are several randomized double blind clinical trial, a meta-analysis study can be applied to achieve the highest certainty of truth.
Splitting into the two groups should be truly randomized; control for confounders
Example of the questionable causative relationship:Prevalence of storks and baby birthsIncrease of antenna’s or cellphones and the increase of cancers
Hypothesis can not give an understandable causative relationship with the outcome.In order to prove a paradigm, you should actually also come up with a hypothesis with which you can deny the paradigm._______________________________Data from Berlin (Germany) show a significant correlation between the increase in the stork population around the city and the increase in deliveries outside city hospitals (out-of-hospital deliveries). However, there is no correlation between deliveries in hospital buildings (clinical deliveries) and the stork population. The decline in the number of pairs of storks in the German state of Lower Saxony between 1970 and 1985 correlated with the decrease of deliveries in that area. The nearly constant number of deliveries from 1985 to 1995 was associated with an unchanged stork population (no statistical significance). However, the relevance of the stork for the birth rate in that part of Germany remains unclear, because the number of out-of-hospital deliveries in this area is not well documented. A lack of statistical information on out-of-hospital deliveries in general is a severe handicap for further proof for the Theory of the Stork.Paediatric and Perinatal EpidemiologyVolume 18 Issue 1, Pages 88 - 92Published Online: 22 Jan 2004http://www3.interscience.wiley.com/journal/118774766/abstract?CRETRY=1&SRETRY=0
The intervention increases the risk for the measured outcome.For example adrenaline will increase heart rate.
The intervention reduces the risk for the measured outcome.For example betablocker andheartrate
Correlation can be caused by chance, but if there is good randomization, a higher number of tests will make the chance that the correlation seen is simply caused by chance.Also point out that no correlation would mean a Relative Risk of 1: which means that there is no difference to have the risk or not in respect to the measured outcome.