Practical Biostatistics

GOOD ABSTRACT: Calcium Intake and Cardiovascular Disease Risk

noreply@blogger.com (Tom Heston MD) — Fri, 28 Oct 2016 22:35:00 +0000

Calcium intake of 2000 to 2500 mg/d, regardless of source (diet vs supplements) is not associated with CVD risk in generally healthy adults. Ann Intern Med. 2016 Oct 25;

Comment: This is an example of a fair to good abstract. The notable feature of this abstract is that it includes a section on Primary Funding Source, so readers can help determine possible funding bias. Better form would have been for the authors to capitalize each section heading.

This systemic review and meta-analysis included the following sections:

Background
Purpose
Data Sources
Study Selection
Data Extraction
Data Synthesis
Limitations
Conclusion
Primary Funding Source

GOOD ABSTRACT: Vitamin D is inversely related to risk of pre-eclampsia and preterm birth

noreply@blogger.com (Tom Heston MD) — Fri, 14 Oct 2016 19:49:00 +0000

Vitamin D, pre-eclampsia, and preterm birth among pregnancies at high risk for pre-eclampsia: an analysis of data from a low-dose aspirin trial.

As maternal vitamin D status in the second trimester increased, the risk of early-onset pre-eclampsia and preterm birth decreased in women at high risk for pre-eclampsia. BJOG. 2016 Oct 5;:

Comment: This is an example of a GOOD ABSTRACT. The sections are clearly demarcated and the abstract is easily readable. This is the first research article containing a section called Tweetable Abstract. Nice!

A Tutorial on Hunting Statistical Significance by Chasing N

noreply@blogger.com (Tom Heston MD) — Mon, 10 Oct 2016 16:14:00 +0000

False positive findings (Type I errors) are greatly increased by repeated statistical testing and by manipulation of study groups post hoc. Front. Psychol., 22 September 2016

Citation bias favoring positive clinical trials

noreply@blogger.com (Tom Heston MD) — Fri, 30 Sep 2016 16:36:00 +0000

Positive trials of tPA for ischemic stroke are cited more often than neutral trials or negative trials. Trials. 2016;17(1):473

Comment: This study found that there is a significant amount of CITATION BIAS in the use of tPA in ischemic stroke. Perhaps this is because we always want to do something, like give a fancy drug or call a Code Stroke, rather than just stand by and give aspirin or clopidogrel. This citation bias may have disasterous results, so must be strongly guarded against.

MORE BIAS: Covariation bias in women with a negative body evaluation

noreply@blogger.com (Tom Heston MD) — Wed, 28 Sep 2016 18:00:00 +0000

Covariation bias was temporarily diminished by use of a computer simulation program which disconnects the two variables. J Behav Ther Exp Psychiatry. 2016 Mar;50:33-9

Comment: Covariation bias is the over-estimation of a relationship between two variables. In this study, the researchers looked at women who over-estimated the relationship between their own body and negative social feedback. The bias was decreased when computer program contingencies were manipulated so that her own body was rarely followed by negative social feedback. This abstract brings up two important concepts:

COVARIATION BIAS. Be aware that we may over-estimate relationships sometimes.
BAD ABSTRACT. This abstract is okay, but in the results section it gives a narrative analysis rather than a statistical analysis. In the results section, readers need to see the scientific, statistical analysis. We are not reading a work of literature, a poem, or a novel. This is about science, and one of the fundamental foundations of science is the statistical analysis of research data. This should be clearly presented in the results section of a scientific abstract. On the plus side, they did include a limitations section. This is optional, but a helpful addition to the standard abstract.
The RESULTS section of an abstract is where the statistical analysis of the data is presented. This section is not for a narrative summary or opinion.

Bad Abstract: Adherence to recommended exercise guidelines in patients with heart failure.

noreply@blogger.com (Tom Heston MD) — Wed, 28 Sep 2016 16:46:00 +0000

The abstract for this review states that in their manuscript, they make recommendations for the direction of future research. Heart Fail Rev. 2016 Sep 26;

Comment: Should review articles also follow structured abstract guidelines? A strong argument can be made that yes, they should. This abstract of a review article tells us almost nothing. They give some background, then vaguely state what they present in their manuscript. Thus, the abstract in effect tells the reader to just read the manuscript. What good is that? Are abstracts simply commercials, urging readers to read the full manuscript? No. Abstracts should provide a summary of the contents of the manuscript. To make it easy for readers to get the gist of the manuscript, the full article, the abstract should ideally contain these sections: a) background, b) methods, c) results, d) conclusion, and optionally e) implications. These sections should be clearly identifiable. Typically, each section should start on a new line, with the section title in all caps or bold type.

Bad Abstract: Comparison of risk and protective factors associated with smartphone addiction and Internet addiction.

noreply@blogger.com (Tom Heston MD) — Wed, 28 Sep 2016 16:36:00 +0000

The authors of this study conclude: "Our findings will aid clinicians in distinguishing between predictive factors for smartphone and Internet addiction and can consequently be utilized in the prevention and treatment of smartphone addiction." J Behav Addict. 2015 Dec;4(4):308-14

Comment: This is an example of a poor abstract. The conclusion to an abstract should state what the scientific study found, not opinion regarding implications of the research. In this case, the researchers presented a conclusion that has nothing to do with the hard scientific data that they found and analyzed. Rather, their conclusion was a mish-mash of vague wishy washy ideas about what they imagine regarding the effects of their research. I challenge anyone to tell me what their research study actually found, from a scientific standpoint, based upon their conclusion. Deeper analysis of the abstract reveals that their results section was also flawed; what was presented here should have been in the conclusion. The authors did not present a statistical analysis of their data in the abstract (this should be presented in the results section).

How social media affects the health of children and young people.

noreply@blogger.com (Tom Heston MD) — Wed, 28 Sep 2016 16:20:00 +0000

Cyber bullying and Facebook Depression are associated with mental health problems in children and young people, but no cause and effect relationship has been proven. The authors recommend more research into this area. J Paediatr Child Health. 2015 Dec;51(12):1152-7

Comment: The authors raise an important statistical fact: association does not equal causation. There is conflicting evidence regarding the benefits versus cons of social media. Benefits include greater social interactions along with less loneliness for the elderly and isolated. Cons include negative self-esteem and mood disorders, primarily in younger users. While these observations are interesting, the authors rightly point out that no cause and effect relationships have been scientifically proven.

Mindfulness Practice to Reduce Bias

noreply@blogger.com (Tom Heston MD) — Mon, 26 Sep 2016 14:34:00 +0000

Practicing mindfulness meditation may be a way to reduce harmful unconscious biases, such as racial or ethnic biases, in health care providers. Patient Educ Couns. 2016 Sep 15;

Comment: We all have several potentially harmful unconscious cognitive processes. The authors of this article argue that mindfulness practice promotes a nonjudgmental approach. Note, however, that this is a theoretical article that does not scientifically prove in any way that mindfulness practice is effective at reducing bias. We cannot exclude the possibility that the opposite, undesired effect may occur. Scientific research has a way of often disproving the best sounding theories.

Attentional Bias for Emotional Stimuli in Borderline Personality Disorder: A Meta-Analysis.

noreply@blogger.com (Tom Heston MD) — Sun, 25 Sep 2016 15:11:00 +0000

Patients with borderline personality disorder tend to have an ATTENTION BIAS towards emotionally negative words. Psychopathology. 2016 Sep 20

Comment: Attention bias is another way to say excessive focus upon, or to have blinders. Not only patients can suffer from attention bias, scientific researchers also can suffer from this bias. When analyzing research papers, it is important to see the big picture as well as the fine details. You may discover that researchers are so close to their area of expertise that they excessively focus on some detail, resulting in attention bias.

Belief-confirming reasoning bias in social anxiety disorder.

noreply@blogger.com (Tom Heston MD) — Sun, 25 Sep 2016 14:40:00 +0000

BELIEF BIAS is when in spite of the facts, people will tend to consider believable conclusions as true and unbelievable conclusions as not true. J Behav Ther Exp Psychiatry. 2016 Dec;53:9-16

Comment: This study found a belief bias among those with social anxiety disorder. The beliefs tended to reinforce false ideas with the result of affirming their social anxieties.

Why Most Published Research Findings Are False

noreply@blogger.com (Tom Heston MD) — Sat, 24 Sep 2016 21:14:00 +0000

This is an interesting article on bias, from a professor of medicine at Stanford(John P. A. Ioannidis)

http://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.0020124

Many scientific “truths” are, in fact, false

noreply@blogger.com (Tom Heston MD) — Sat, 24 Sep 2016 21:13:00 +0000

Mythology is alive and well in the field of science.

http://qz.com/638059/many-scientific-truths-are-in-fact-false/

Voices that cannot be heard: Can shyness explain how we communicate on Facebook versus face-to-face?

noreply@blogger.com (Unknown) — Tue, 23 Aug 2016 14:29:00 +0000

This study found that shyness was negatively correlated with self-disclosure. Computers in Human Behavior 29(4):1402–1407 · July 2013

Comment: This study found NEGATIVE correlations, meaning that as one variable goes up, the other goes down. This is the meaning of negative in terms of statistical correlation; it does not imply good or bad. Thus, we conclude that since shyness and self-disclosure on Facebook are negatively correlated, shy people do less self-disclosure on Facebook.

Biostatistics Faculty and NIH Awards at U.S. Medical Schools.

noreply@blogger.com (Tom Heston MD) — Sat, 18 Apr 2015 03:40:00 +0000

The number of biostatistics faculty is positively associated with the amount of NIH research awards in medical schools in the U.S. (Am Stat. 2015 Feb;69(1):34-40.)

Plastic Surgery Residents' Understanding and Attitudes Toward Biostatistics: A National Survey.

noreply@blogger.com (Tom Heston MD) — Mon, 16 Feb 2015 15:40:00 +0000

This study found that plastic surgery residents felt that knowledge of biostatistics was important, however upon objective testing they only had a fair understanding of statistical principles. The residents had difficulty with study study design, ANOVA, regression, and identification of a statistically significant result. Confidence was not a good predictor of objective performance. (Ann Plast Surg. 2015 Jan 30.)

Bonferroni multiple comparison test

noreply@blogger.com (Tom Heston MD) — Tue, 10 Feb 2015 22:49:00 +0000

When performing multiple tests of significance upon a single dataset, there is an ever increasing chance that at least one of these tests will be statistically significant. For example, a p-value of 0.05 means there is a 1-in-20 likelihood of the statistical result occurring by chance alone. If we perform 20 such tests, then there is a 1-in-1 likelihood of at least one result having a p-value of 0.05 or less.

The Bonferroni correction is simple. When multiple tests of significance are performed, then the cutoff p-value for statistical significance is set to equal 0.05 divided by the number of tests performed.

For example, if 10 tests of significance are performed, the cutoff p-value for statistical significance would be 0.05 / 10 = 0.005 when using the Bonferroni correction. If 5 tests are performed, then the cutoff p-value would be 0.05 / 5 = 0.01.

Recommended Books to Help You Prepare for the USMLE Biostatistics Section

noreply@blogger.com (Tom Heston MD) — Mon, 26 Jan 2015 22:01:00 +0000

The USMLE, as well as board certification examinations, almost all contain questions on basic statistics. A good review of biostatistics thus is helpful for both physicians in training and practicing physicians. As a teacher of bioastatistics for medical students, I've had the opportunity to review a large number of biostatistic review guides, and as a researcher frequently use several references which are helpful for the physician interested in more in-depth learning.

First Aid for the USMLE Step 1. This review has a nice, short section on biostatistics that covers most of the basics. You don't need to get the latest version of this book, obviously, because the basic concepts of biostatistics doesn't change.
High-Yield Biostatistics. This is a very well written book, which addresses biostatistics, epidemiology, and public health with the USMLE Step 1 exam in mind. This contains more in-depth information on biostatistics than most people will require or want, but it is a good reference and highly recommended.
PDQ Statistics. This small book covers biostatistics with the researcher in mind. It has good chapter summaries, and especially good pitfalls to watch out for when performing statistics or reading the medical literature.
How to Lie with Statistics. Required reading for all physicians.
Basic and Clinical Biostatistics. This is a fantastic reference book. It covers the topic too extensively to be used as a quick review guide, however, it is an excellent reference that physicians engaged in research should have on their bookshelf.

The quantum revolution

noreply@blogger.com (Unknown) — Tue, 02 Dec 2014 15:50:00 +0000

Scientists have developed a better way to run a quantum algorithm. This new method uses much simpler methods than previously algorithms, and could possibly be a major advance in the development of a quantum computer. Read More

How to Pass Your Standardized Examination in Biostatistics: Fear and Reward

noreply@blogger.com (test) — Mon, 10 Nov 2014 22:32:00 +0000

Pleasure and fear are primary human motivators. You will need to learn how to use these properly in order to pass your statistics examination.

First, lets take a close look at your fears.

Fear isn't a weakness, but rather when used properly it is a gift. Fear can paralyze you, or it can motivate you to take action. Usually, paralysis comes from when we do not face our fears head-on, and let them simmer in the recesses of our thoughts. So take a good inventory of your fear. In the case of a standardized examination, the fear is obvious: it is the fear of failure.

The fear of failure is very real when it comes to standardized examinations. In order to pass classes, graduate, and become licensed, students need to pass and do well on standardized examinations covering epidemiology and biostatistics. It is an essential part of medical training because the basic concepts do not change over time, and nearly all general standardized examinations contain at least a few questions on epidemiology and biostatistics.

As a practicing physician in the United States, I was required to pass multiple standardized examinations in order to pass my medical school classes, graduate from medical school, get a medical license, and become Board Certified. Now, with the new Maintenance of Certification requirements, I am required to continue to take and pass standardized examinations. Doing well means I can continue to work. Not passing an examination means in many cases a loss of Board Certification and subsequent loss of hospital privileges. And what do all of these examinations have in common? First, they are all structured in a similar manner, and secondly, they all contain questions on epidemiology and biostatistics.

The first step in doing well on these standardized examinations is to really understand at a deep and personal level just how important it is to do well. Yes, it is enjoyable to learn and master epidemiology and statistics, but you must do more. You must be able to apply your learning, and pass your standardized examination.

For medical students, achieving a high score on the USMLE Step 1 examination is critical in the determination of where students will end up doing their residency program. Those who score well will be much more likely to get into their number one residency choice. Those who score poorly and those who fail face a much greater challenge in getting into the specialty of choice, their residency of choice, and even getting into a residency program at all. Understanding at a fundamental, personal level the great importance of this examination helps motivate you to study and prepare properly. Would you like to take Friday night off to go dancing? Maybe that's okay, but maybe you need to study instead. First, think about how important the USMLE examination is, then decide if it is in your best interests to go out or if you should stay in and prepare for the USMLE.

So, we know one thing quite clearly. Failing your standardized examination in statistics will be painful.

The second great motivator is pleasure. In the case of numbers and statistics, it may be difficult to understand at first, but upon really understanding the concepts a great pleasure results. It's the pleasure of learning, the pleasure of knowing, the pleasure of realizing that your hard work ultimately will result in you helping people better and more effectively.

Understanding basic concepts in epidemiology and biostatistics means you won't be fooled by sales reps that want you to use their product. You will know better, and see past their sales pitch. Your mastery of statistics will help you view one of the greatest things in this universe --- the truth.

A great pleasure of learning statistics in a manner that will help you pass your standardized examination comes from passing your test. Passing your test with flying colors means more recognition, more options, and greater control over your future. A high score means more job opportunities, more residency opportunities, more options.

Be smart. Understand the basic motivations of fear and pleasure. Use the fear of failure to motivate you to study more, and study better. Use the pleasure of learning to motivate you to study more, and study better. Most of us respond better to either fear or pleasure. Use both if you can, but at least be sure to tap into your primary motivation source. Use this to light that fire in your belly in order to learn statistics, pass your test, and help your patients.

Proper preparation for the USMLE exam requires that you get started on day #1 of medical school.

Primer in statistics part 2

noreply@blogger.com (Unknown) — Wed, 22 Oct 2014 13:51:00 +0000

This is the second part of a two part series.

First, a guide in selecting the proper
statistical tests based on the research question will be laid out in
text and with a table so that researchers choose the univariable
statistical test by answering five simple questions.

Second, the
importance of utilizing repeated measures analysis will be illustrated.
This is a key component of data analysis as in many dental studies,
observations are considered repeated in a single patient (several teeth
are measured in a single patient).

Third, concepts of confounding and
the use of regression analysis are explained by going over a famous
observational cohort study.

Lastly, the use of proper agreement analysis
vs. correlation for study of agreement will be discussed to avoid a
common pitfall in dental research.

Primer of statistics in dental research: P... [J Prosthodont Res. 2014] - PubMed - NCBI

Primer in statistics part 1

noreply@blogger.com (Unknown) — Wed, 22 Oct 2014 13:47:00 +0000

This series of 2 articles aim to introduce dental researchers to 9 essential topics in statistics to conduct EBD with intuitive examples. The part I of the series includes the first 5 topics

(1) statistical graph

(2) how to deal with outliers

(3) p-value and confidence interval

(4) testing equivalence

(5) multiplicity adjustment.

Part II will follow to cover the remaining topics including

(6) selecting the proper statistical tests

(7) repeated measures analysis

(8) epidemiological consideration for causal association

(9) analysis of agreement.

J Prosthodont Res. 2014 Jan;58(1):11-6

Vikings and Superheroes: building a statistical network

noreply@blogger.com (Unknown) — Fri, 17 Oct 2014 16:18:00 +0000

The Icelandic sagas of the Norse people are thousand-year-old chronicles of brave deeds and timeless romances, but how true to Viking life were they? Researchers used a statistical network of associations between characters to find out. Read More

What drives activity on Pinterest?

noreply@blogger.com (Unknown) — Sun, 30 Mar 2014 04:48:00 +0000

Researchers have released a new study that uses statistical data to help understand the motivations behind Pinterest activity, the roles gender plays among users and the factors that distinguish Pinterest from other popular social networking sites. Read More

Sporting events: Clear your memory to pick a winner

noreply@blogger.com (Unknown) — Sun, 30 Mar 2014 04:48:00 +0000

Predicting the winner of a sporting event with accuracy close to that of a statistical computer program could be possible with proper training, according to researchers. Read More