enzyme inhibitors

me again!

a brief description of how to memorise that pesky list of enyme inhibitors

enzyme inhibitors can lead to toxicity by stopping the liver from breaking down drugs. When these drugs levels rise all sorts of mayhem can ensue – and that aint no fun at all, so knowing this list will help you save lives (you clever little sausages)

The list was published on my main site but I think you’ll find this more fun

Our memory image starts with a cartoon Dwight D. Eisenhower (5 star general of the US army, 34th US president and one time supreme comander of NATO), see him strutting round his house looking rather pleased with himself admiring the shiny five stars on his jacket. We can imagine peering in through the window at him to see him waving back and mouthing a big friendly hello, grinning from ear to ear he then puts himself in one of his famous trances and meditates ……ommm ommmm (how rude)

If only we could have seen in properly to notice that he wasn’t wearing anything below the waist and how he was at the same time abusing lots of silly little cartoon ants, oh how those ants take the abuse and oh my how really hairy he is, thick luxurient hair – just imagine

We see how he stands and turns to find his valet saluting and shouting “sir general sir Eisenhower sir” and handing him a large container brimming with water and hundreds of phone sim cards bobbing about in it. The general sips and sips from that huge container with the sims spilling all around him. Before we know it he is rolling around on the floor giggling away, completely intoxicated and dunk. That funny valet – he’s played another prank on the general and given him solvent. Imagine that – see how they laugh

AODEVICES – hello (allopurinol), omm (omeprazole), ant abuse (antabuse is of course the tradename for disulfiram), hairy (hairythromycin – i mean erythromycin), valet (valproate), Eisenhower (isoniazid), sim (cimetidine), sip (ciprofloxacin), intoxicated (ethanol intoxication), solvent (sulphonamides)

Go on try it with your eyes closed? – how many did you get

Try again – get them all this time?

I hope this is useful

Dr B

(PS more pics coming when I get to update this from a faster computer)

enzyme inducing drugs

Greetings earthlings

I cordially invite you to join me on a memory adventure

Today we are going to learn how to recall (liver) enzyme inducing drugs (you know, the ones that make your p450 cytochrome oxidase pathway work faster and thus decrease the effectiveness of the oral contracetpive pills and other stuff)

They are (phenytoin, carbamazepine, barbiturates, rifampicin, alcohol and the sulphonylureas) easier to remember than you may think

We will use a few memory devices here

I originally posted on this in list form but you’ll perhaps find this easier

Imagine a PC (that’s an english police constable for those of you not in the UK) in her bra and suitably interesting lingerie. See her leaving the house (hers, yours, your choice) with pennies overflowing from her heaving bosom and out of that bra. She takes one of the shiny pennies out and pops it into the coin slot of a comedy coin operated toy car which she rides bouncing along to next door whic h turns out to be a barbershop (complete with pole) – you see her go into the shop with its wooden floor and tinkle of the bell as she enters.

She then spots that cheeky chappy the barber who is standing on the counter triumphantly urinating red urine into one cocktail glass after another. These red overflowing splashing glasses are drunk wearily by a small quiet customer who is bothered by the alcohol putting him of his crossword solving, what a bother to him (in his small spectacles and geeky outfit) he just isn’t getting the word urea despite also the urine around him

PCBRAS are the initial letters. Hence a PC in her bra. Penny (phenytoin), car (carbamazepine), barber (barbiturates), red urine (rifampicin-  turns ALL your secretions red… even those ones!), alcohol (….wait for it…… alcohol) and solver (sulphonylureas)

Close your eyes and try to recount these – you’ll be surprised at how easily it sticks and how recalling the list is more fun than learning medicine really ought to be

Lots of love

Dr B

(PS – I will add more pics etc to this post when I’m at a faster terminal!)

Haemorrhoids – what a pain in the arse

Right everyone, sit up straight and pay attention. This is an area that is poorly understood. I will now try to shed some light where the sun don’t shine

Your bum is designed to pass poo out. It does this best if it is in well formed turds. Liquid poo comes down the digestive tract (colon) round the S-bend (sigmoid colon) and into the rectum (the straight bit near your bumhole). Here the liquid poo has some of the water reabsorbed so that it becomes a nice putty consistency and can be poo’ed out (defaecation)

Haemorrhoids

The lining of the rectum is well supplied with blood vessels (so that water can be absorbed) and these are formed into small grape like clusters near the lower end. This is called the haemorrhoidal plexus (which is where the name haemorrhoids comes from). The haemorrhoidal plexus of veins can become distended and enlarged (a bit like varicose veins in the legs) and occasionally dangle out of ones tail end. If this occurs, they are usually easily pushed back in with a finger (they can have elastic bands placed around them to make them shrivel up (done in the surgical clinic up a device called a proctoscope). Some surgeon inject them with something to make them shrink (a sclerosant such as phenol)

external haemorrhoid shoud be correctly labelled perianal haematoma

If ones haemorrhoids are enlarged (something that can happen in pregnancy) then you are prone to having a hard bit of poo (like concrete) scrape them – this tends to cause painless bright red rectal bleeding and classically comes out of the bottom just after the poo, and can even leave bright red streaks of blood on the poo.

This is what one looks like up a proctoscope. For grading (I-IV by Banov) check the wikipedia link

this is painless

Perianal haematoma

I will here mention another thing near the bottom called a peri-anal haematoma. This is important mainly because this is what most non-medical people (and a few medical ones) incorrectly call ‘piles’ or ‘haemorrhoids’. Ones anus (the crinkled skin which stretches to allow poo out and then shuts to keep ones pants clean) is a very well designed bit of kit. It needs a good blood supply and indeed gets one. The circular muscle just inside which has the same function is also well supplied with blood. This means that there are plenty of blood vessels around. The downside of this is that on ’straining at stool’ (trying to push a hard poo out) you can burst one of these blood vessels. This results in a blister under the skin right next to your anus. You can see these – they are bluish and around 1cm big. They are very tender to touch when they first form

this is painful

Haemorrhoids are painless (because they come from inside you where there aren’t stretch fibres in the skin covering it) and perianal haematomas are painful (cos they arise from the outside skin – which really doesn’t like being stretched). Another word for haemorrhoids is ‘piles’ (or Chalfont-St-Giles in English cockney rhyming slang)

Piles are painless

When a perianal haematoma shrivels up (a few weeks later – though you can expect the pain to be gone after a few days) it is because the body has reabsorbed the blood. Sometimes the skin has become over-stretched and leave a small skin tag (these may be seen with the aid of a mirror and are of no consequence

Sometimes you get dark blood (or even black like coffee-grounds) out of your bumhole (this is potentially more serious and you should go and see a doctor – it can sometimes be a sign of cancer. Particularly if you have any unexplained weight loss or abnormal tiredness)

Cream for the itching?

Before I leave perianal haematomas alone I need to dispel a few more myths. Creams that you can buy from the chemist won’t help. They won’t go away any quicker (nor will proper haemorrhoids dangling out). These creams claim to stop itch.. well the itch is usually caused by poo. That’s right. Your ring piece has got beautiful crinkled skin and it is very easy to get a few specks of poo left after you wipe. This is much more likely to happen if you have a painful blue grape there as well (perianal haematoma). Poo is painless inside (again, the skin inside is specially designed for this) but very irritant if it is contact with the delicate outside skin

Quite a successful way of cleaning if one is in pain is to use wet-wipes (moist toilet tissues) and to gently wipe till you don’t get any marmite (brown slears). This is cheaper that those creams – and more likely to work now you have a good grasp of the underlying anatomy and medical condition which you may have

No more marmite

Anal fissures (not all they’re cracked up to be)

A quick word on fissures and then we’re done. A fissure is a crack or split in the skin of your anus (bumhole). These can be exquisitely tender and typically cause pain on pooing. Also some bright red blood may be seen on the toilet tissue. Most of these get better with good cleaning (see above) as this removes the irritant material with can perpetuate the fissure. Occasionally you may need to consult a doctor (who can prescribe medication – GTN or glyceryl trinitrate ointment, which can be very effective)

The last word is just to mention some detail about the above conditions for those who like to put things up in the other direction (fingers, willies, vibrators and so forth). Haemorrhoids won’t be worsened (though if bleeding occurs you may wish to be a trifle more gentle), perianal haematomas – i recommend waiting until they aren’t painful, these also won’t be worsened. Anal fissures may be worsened – so I would suggest waiting until the fissure is completely healed. Anal bruising is painful and looks like a generalised red/blue discolouration which tends to be painful – in which case I recommend lots more lubrication, more gentleness or avoid the bum

Keep well

dr B

Acne rosacea

Alright y’all

 

Rashes on the face come in several patterns. These can at first appear confusing. With a little understanding we will get you diagnosing people on the subway, weddings and other gregarious occasions

acne rosacea zones

Today we will look at acne rosacea. This diagnosis is one to get right because with the right treatment much improvement may be had – though with the wrong one the patient can get worse, and primum non nocere (as we docs like to say)

Famous faces

Sir Alex Ferguson

Famous folks who have this skin condition include Sir Alex ‘the hairdryer’ Ferguson (coach of Manchester Utd football club and often red faced in team talks and when discussing Chelsea). Also President Bill Clinton (I’m sure we can picture him maybe a little red-faced?)

President Bill Clinton

Sebaceous glands and blood vessels

It is not clear whether the pathology is entirely due to inappropriate response of underlying vessels in the skin. Acne rosacea widely varies in its severity and there are a few different sub-types recognised. Facial flushing is particularly associated with one type and rhinophyma is associated with another. Erythema (redness) and telangiectasia (small dilated vessels) may be found

rhinophyma - bullbous nose

Acne rosacea tends to occur later on in life. More common in ladies but more severe in men. It can be associated with excess alcohol consumption though this is not causal. The flushing associated with acne rosacea can be precipitated by sunlight and the rash may get worse with minor irritants. Those irritants can include steroid creams, benzoyl peroxide or isotretinoin (unfortunately these are often prescribed for acne vulgaris and seborrheic dermatitis for which this can be mistaken). It should go without saying that chemical peels don’t help, but aggravate acne rosacea

Stress and changes in temperature are often triggers for the flushing as are some foods (eg. spicy) and caffeine

Interestingly there is some research which shows that patients with acne rosacea are more likely to have small intestinal bacterial overgrowth and that eradication of this by a long course of antibiotics is likely to cause a resolution of the symptoms

 

Mistaken identity

Acne vulgaris tends to be more pustular and involve the neck, chest and back whereas seborrheic dermatitis tends to involve the scalp, hairline and ears more. A good rule of thumb is to think of acne rosacea as solely a facial condition

 

 

seborrheic dermatitis - involving the hairline

Note that lupus pernio (sarcoidosis of the nose) may be mistaken for rhinophyma

Treatment

The tetracyclines by mouth (eg. oxytetracycline, doxycycline) may be effective as may metronidazole (an imidazole) cream topically. If the eyelids are involved (and they are in about half of cases), then lid hygiene (regular cleaning with a baby shampoo and moisturising) is useful

For resistant cases, dermatologist specialist management is needed (indeed about 1% of all a dermatologists caseload are those with acne rosacea)

 

Finally acne rosacea is a chronic condition and most cases need some form of treatment (even if only trigger avoidance) for a lot of their lives

 

Keep well my lovelies

 

Dr B

Dermatomyositis – funny rash on the face

Hi gang

 

Dermatomyositis is a condition which is not very common – though with a bit of thinking you may be able to make a diagnosis when others around you are stumped. If you got this rash on your face and spotted it in the mirror – you may throw your hands up in horror, only to be dumbfounded by the appearance of your purple knuckles! (only of course you and I know that the girdle limb weakness meant that one couldn’t throw one’s hands up in horror – so one would probably simply sink to one’s knees weakly)

It is an inflammation of the muscles with a characteristic skin appearance. The muscle inflammation is in the big muscles (thighs and around the shoulders) – this causes weakness and raised CK (which leaks out of the damaged muscle cells)

The main symptom is a progressive symmetrical proximal muscle weakness. Weakness can be caused muscle atrophy from for example dennervation, but in dermatomyositisthe muscle bulk seems to be suprisingly preserved

Gottron's papules

 

Autoimmune associations

Dermatomyositis seems to be an autoimmune connective tissue disease. This means has associations with other autoimmune conditions. It has in biopsies changes which seem to be caused by immunoglobulins (a B cell infiltrate – humoral mediated immunity). This differentiates it from polymyositis (all the same features, but no skin involvement) as this has a T-cell infiltration indicating a cell mediated immune response seen and is therefore different

There is an association with cancer in male patients over sixty and anti-Jo1 antibody is associated too. HLA-B8 & HLA-DR3 are linked.

Because it is a generalised reaction – expect symmetry of signs

 

Skin

The skin things are a heliotrope facial rash (purply indurated facial rash is a characteristic distribution) and Gottron’s papules – purple lesions on the knuckles – these are not found in polymyositis (Heinrich Adolf Gottron was a German dermatologist living in the earlier part of the 20th centruy)

heliotrope rash

proximal muscles affected in dermatomyositis

 

Diagnosis

Proximal muscles weakness, elevated muscles enzymes, muscle biopsy and a characteristic pattern on EMG (electromyography) should just about do it. Add in the skin changes and you’ve got a winner

Mortality

The mortality of dermatomyositis used to be around 25%. Though with prednisolone and other modern immune modulators this is down to about 5%

 

I hope that was a little illuminating

 

All the best

 

 

Dr B

Making sense of medical education

Wotcha

Today I will speak a little on medical education from the point of view of teaching

a changing world in medical education

Students sometimes tell me how out of touch the medical school is and the learning situations they participate in. I think it would be worth a few lines to try and explain this

When I trained, the world was a different place (we didn’t even have computers) – the pace of change was phoenomenal, so by the time I graduated half the students were word processing their essays. The internet was coming into life and changed from being a tool where one could look up paper abstracts to one where some people knew how do download pictures (which appeared one line at a time)

Anyway – I’m waffling. I’m going to say that medical degrees evolve too. Though not quite at such a blistering pace. I heard it said that

‘changing a medical school curriculum is like trying to relocate a cemetary’

There were bits about our course which were great and others that made it seem like people were pulling off you toenails (slowly). Some lecturers I still recall many of their sentences and slides (yes many still used chalk boards too) – others I recall simply drawing their caricatures as their funny shaped ears proved the most lively part of their lecture

the purpose of a lecture was to transfer what was on the notes of the lecturer to the notes of the students without passing through the minds of either

good teaching

If people teach well, their subject lifts and inspires their students, who become self motivated informed learners (apparently the best sort and most likely to do well according to modern educational theories) – spurred on by this is I sought to try and learn how to teach too and be one of the good ones

I had no idea how tricky it is to have to meet the learning needs of all those in front of you. Each are at a different place on their own learning journey and have done differing amounts of preparation. Each learns material in a different way (visual, auditory and kinaesthetic embraces one viewpoint of learning styles)

Your students are diverse individuals (some with hangovers, some fasting, some about to menstruate) from vibrant and diverse backgrounds (socially, politically and culturally)

Each deserve the best you can produce, all the time

drawing up a curriculum

When a curriculum if formed it is a monstrosity. It is a leviathan. A huge melange of spreadsheets and bits of paper, committees and meetings. The essence is to take what you did for the last few years and make it better

You do this by teacher training (many of whom think they do quite a good job already because they are senior clinicians and that is the way they’ve always done it) – some are right, they are brilliant, some aren’t

It is difficult for a medical educationalist (often a doctor but usually in their thirties) to try and encourage a 50 year old senior head of department to change their teaching style to embrace change and flexibility. This is the task faced with improving the standards of teaching. Big ego’s (though no fault of their own – that was always how it was done, but the pace of change is too rapid now)

Changing the curriculum is tricky. Last few years was better than before and the few years before that. Most of the obsolete stuff has now gone (evidence based medicine has been practised for so long that it has just filtered through into the medical degree)

what do you include?

The challenge is that what was taught is still important. The basic sciences are evolving so their is more to learn. But the old stuff is still relevant too. Anatomy is still there, physiology still needs to be in there. All the old specialties are practised. New ones have arrived (interventional radiology and gerontology being too and I do hope you’ve got hold of palliative care). Genetics and molecular medicine are becoming increasingly important. Communication skills are now taught, telephone consultation, breaking bad news and how to write an illegible prescription with a suitably impressive flourish of a signature. Medico-legal aspects and complementary therapies are vying for attention along with multi-disciplinary team case based approaches

All of these are trying for an extra hour in the teaching timetable and clamouring for attention at committee meetings where the sandal wearing primary care physicians try (in a caring understanding way) to point out that 90 to 95% of all consultations happens in the community and they need two more weeks to spend with the students.

The students point out that the course is too long, costs them too much (for they are now empowered consumers) and could they please have a half day on Wednesdays for sport and the option to study at home because of childcare commitments). The education is moving increasingly online and some are even delivering lecture material over the internet – this is not evidence based, but seems quite gadgety and may save all concerned time and money – but at what cost?

on trying to be a good teacher

My expounding is over (polydirectional and pointless though it may have been) – changing a curriculum is not easy. All those who are involved think they know best. The time is limited and there are pulls from all sides including the end-users. Patient groups too I’m sure will be chipping in before long

As a medical teacher it is important you know your material well enough teach without notes and really communicate with who you have in front of you. Keep up to date with educational theory and embrace change. Be aware of their needs (upcoming social events in the students calendar and individuals on personal crisis and accept that some of them can’t even remember what on earth succinyl CoA is for). Make it fun and they will learn better. You don’t have to teach them everything, but get them leaving wanting more so that they will want to take their own journeys

on being a good learner

As a student, know the difficulties the medical school had coordinating everyone to the right place at the right time. Know the challenges of an evolving curriculum. Become involved in the future and try to be the best for your patients and future generations of doctors. Read up on your subject before you attend any teaching. Keep your notes brief – long stuff is when you are in the library. Focus on the learning and review afterwards, review it all at the end of the week. Preview the next sections. Teach it to others in small groups – only by knowing it well enough to teach it well do you fully assimilate your learning

Above all, keep yourself well. Eat well, exercise, play, socialise, get enough rest. Your body is the most important tool you will ever own, treat it well. You cannot learn well if you aren’t looking after yourself – and that is not fair on the patients you will come to treat, they deserve your best. You must look after yourself then too – same rules. You are more important than them – because you cannot give to people from an empty cup. Full your cup full to brimming and then and then you will overflow and give in abundance

Keep well

Dr B

Vital statistics (medical student stats – part 5)

Hello hello hello

More on stats. I hope this is of some use

The story so far:

Part 1 covered sensitivity and specificity and touched on predictive values

Part 2 looked at p values and null hypothesis (with type I and II errors)

Part 3 looked at some stats tests

Part 4 looked at types and distribution of data

This one will look at skewing, correlation coefficients and touch on meta-analyses

skewed data

Some data isn’t distributed in the pretty ‘normal’ manner it is skewed. This means that there is a disproportionate amount of samples at one end or the other. The skew is described with respect to where all this extra stuff is. So a ‘negative’ skew has a big left tail and and a ‘positive’ skew has a big right tail

skewed data

correlation coefficients

When one can plot data on a graph as a scatter plot. One can either look at the dots and try and draw a line (looking for a relationship between the dots) – or one can use a statistical test. The test to use is Pearson’s coefficient of linear correlation. This gives an ‘r’ value. The closer the dots lie to a line the closer the value of ‘r’ to 1.

r=0 & r=1

In the above graphic the top plot has seemingly random dots (r=o) and the lower one seems to have the dots in a line (r=1). If the line sloped down rather than up, then r= -1.

linear regression

This is a posh term and sounds really scary. It ain’t. If you look at the bottom of the two scatter plots above, you can see that a line becomes apparent. This is from the data we’ve collected and then analysed to produce this picture. Linear regression is the use of a formula to predict other values in the population from what we already have. So if we have a line – we can extend the line in either direction to predict what we might find. Simple. By the way – the regression equation would look like (Y = a + bX)

meta-analysis

Meta-analyses are where clever doctors and statisticians combine several related studies to draw conclusions that are more powerful than the sum of their parts. They make use of clever statistical sums such as meta-regression and other fancy stuff. The Cochrane database library is perhaps the best known medical compiler of this type of information

That’s it – I reckon that covers most of the important medical stats. Here’s a picture of a cute kitten to help soothe your well exercised mind. Now go and make a cup of tea and have a little lie down

cute kitten

All the best

Dr B

Vital statistics (medical student stats – part 4)

Hi once more

I true blockbuster stylee it turns out that trilogy turns out not to mean three parts

I reckon I could make some passing comments about a couple more things…

The story so far:

Part 1 covered sensitivity and specificity and touched on predictive values

Part 2 looked at p values and null hypothesis (with type I and II errors)

Part 3 looked at some stats tests

types of data

Data can be put into boxes or on a line. Qualitative or quantitative. Qualitative data might be types of heartsink patients (entitled demander, dependant clinger, manipulative help-rejector etc) and quantitative data might be the resting heart rates of doctors when greeting these daily challenges.

mean, median and mode

These are three ways of describing an ‘average’ – they may all be the same number, but may yield very different results (cf politician and the honourable way they handle statistics to be helpful and open!)

The mode is the number in a data which occurs most often. eg pulse rate of 67 (which occurred an amazing 5 times in our fictitious sample of stressed primary care physicians)

The median is the number in a data set which occurs in the middle if you wrote out all the numbers out in order (not really a very useful thing)

The mean is what people normally talk about as being the ‘average’. It is the arithmetic average (done by sums!), add up all the numbers and then divide that by the number in the sample and voila – the mean

normal distribution

This is a beautiful mathematical thing. It says that is you have a ‘normal’ set of data it will be distributed a little like an upturned bell. This makes a lot of snese sense. Take IQ (intelligence quotiant). This says that if you have a sample big enough (not six people, but six million for example) and tested all of them, then their values would show ‘normal’ distribution. So most people would be near the middle with a few clever clogs at the top end and a few thickies down the bottom end. It translates therefore that the number in the middle is the average (if you paying attention earlier you may notice that for normally distributed data the mean, median and the mode are all the same. In IQ’s this number is given the score 100. I like claiming when out in public in a surprised voice that I read in the paper only just today that there is research which proves that half the population is below average intelligence (you’ll smile much to yourselves when you find people are amazed to hear that and sometimes try to claim that it isn’t so!)

Normal distribution of IQ's

standard deviations

The standard deviation gives you a measure of the spread of the values

This is how much the numbers in the population you are looking at deviate from the middle numbers. A small standard deviation would be if nearly all the numbers are in the middle – so for the IQ example, say 98% are between 99 and 101 with 1% lying outside this range, then the above graph would look really pointy in the middle like a huge thin needle. If there was a large standard deviation then what we are saying, is that the numbers are spread about over the whole range and the above graph would look wide and flat (if this was the case then the data would no longer be normally distributed

standard deviation in normally distributed data

Normally distributed data is really pretty – this means that we can know stuff (and calculate useful stuff like confidence intervals). Sigma σ is often used to denote one standard deviation (sd). We see from the above graph that two sd’s covers 68.2% of the data (2 x 34.1%). Two sd’s covers 97.8%.

standard error

The standard error is a measure of how accurately the mean of the sample approximates the population mean. So if you sample all of the population is will be pretty accurate – and if you sample 4 people it won’t! There is a formula to calculate this

s = standard deviation & n = sample size

confidence intervals

Confidence intervals are a way of stating the reliability of an estimate. They are calculated from standard errors of the mean

A 95% confidence interval (often used) is the mean ± 1.96 standard errors of the mean (SEM). We often approximate it though and say 2 SEM.

In the above graphic for standard deviation, the 98% confidence interval is the numbers on the bottom line which are at -2σ and at 2σ. So you are saying that you are 98% confident that the correct number will fall into this range

More on skewed data in the next part…. (c’mon, I know you’re excited really!)

Ciao

Dr B

Vital statistics (medical student stats – part 3)

Greeting once more dear chums

This is the 3rd of a three part trilogy (Part1 here, Part2 here)

Part 1 covered sensitivity and specificity and touched on predictive values

Part 2 looked at p values and null hypothesis (with type I and II errors)

Part 3 will look at the various stats test which you might be reasonably expected to know a little about

the point of statistical tests

These have a purpose. They are to analyse data sets. If you use the wrong test then the conclusions you draw from your analysis are not likely to be correct. You may end up recommending that all women use the new ‘wonderpill’ which makes them all better looking and maintains a youthful appearance without side effects (it turns out you used the wrong stats test and actually there are side effects – the pill turns them all into frogs in a bizarre quirk of fate). Now that will have consequences for you professional registration if anyone reads your paper and works out you’ve been a numpty. So lets look at the various tests we might employ

'wonderpill'

whoopsiedaisy

parametric and non-parametric

To select which stats test to use for your data you have first to decide is your data set is parametric…. or if it not (parametric means having parameters and often makes a few assumptions)

You need to know what sort of data you have to analyse and whether the data is paired

parametric tests

Parametric tests usually assume normal distribution

A paired (1-sample) t-test is used if you have paired members of the groups. If you matched people in the two sets – so that each person with the disease is age and sex matched with someone without the disease then you have a paired data set. In a crossover trial the person is looked at in each group before they cross over into the other arm – they are thus paired with themselves

An unpaired (2-sample) t-test is a good one to use for unpaired normally distributed data – comparing the average values of two independent groups (eg the height of medical students against the height of law students)(eg shoe size of patients with the disease and shoe size of the control group whom don’t have the disease)

non-parametric tests

Non-parametric tests tend to be based on ranks. This is better for non-normally distributed data. I think I’d better tackle the concept of normal distribution for those of you who’re staring blankly at the screen – check out my next post

Non- normally distributed data can be ranked and compared using test such as

• Wilcoxon
• Chi-squared (χ2)
• Mann-Whitney U

If you see a question which has some information in a 2×2 box – then you should probably use a chi-squared test. If its in a box you only have 2 values in each direction – so it can’t be normally distributed can it. Chi-squared is good at comparing percentages between groups too

Wilcoxon could be used for repeated measurements on a single sample

I’ll stop here – but think I’ll continue the trilogy as I’ve thought of some more stuff to tell you about

Much love

Dr B

Vital statistics (medical student stats – part 2)

Hi there

In Part 1, we talked about sensitivity and specificity. We also looked at what predictive values were. Today we are going to cover statistical tests, how to choose the right one and p values. We also will look at the type I error stuff which crops up occasionally in exam questions for finals, USMLE, MRCP(UK) and MRCGP

p values

A p value is the chance that the result you’ve got could have arisen by chance. That means that the result you have could actually not reflect the real world and you’d be a numpty to believe it. If you flip a coin and I think that it is ‘heads’ as a result of having run this particular analysis on it (nothing but guesswork). The chances of my guesswork getting it right are 0.5 – so the ‘p’ value of this analysis is 0.5 – rubbish.

So rubbish in fact that I’m not going to bet my car or the life of my patient on it. If I had a test which could give me the answer with 90% accuracy, 0.9. The chance of that being wrong are 1 in 10 – a p value of 0.1

This is a little better, but I’m still not convinced. In medicine we like to know our data is significant when reading journals (ok, if we’re being honest – skim reading the abstracts). To be significant (likely to represent sensible information which is accurate in the real world) a p value of <0.01 is what we’re after. That means that there is a less than a 1 in 100 chance that this result could have arisen by chance ( = 99% chance of being accurate)

pisum sativum (Pea)

I think we’ll cover statistical test selection in the next posting – to prevent hypnotic induction and give you at least the interactive input of clicking your mouse to try and keep you awake (if my carefully thought out picture selection isn’t quite doing it for you)

Type I and type II statistical errors

Type I and type II statistical errors (Type 1 and type 2) are a discussion about the null hypothesis. Get your head around this as it makes sense – it will make it easier that memorising and make you less prone to errors regurgitating in the exam!

null hypotheses

The null hypothesis states that there is not a difference between the two sets of statistical data

This means they are the same. Null – nothing. Hypothesis – we’ve having a guess to see if this theory fits. More info on null hypothesis here.

type I errors

If you reject the null hypothesis and it was right – this is a type I error. So you reject that null (null says: no difference, you say ‘Oh yes there is..’), turns out that you’re wrong (oops). That means that there is not a difference between the data and you said there is. So you have just made a type I error. This crops up in medicine because it is seen in ‘false positive’ tests – see below. eg the test saying the patient has HIV, whereas they they don’t really

type II errors

If you accept the null hypothesis and it was wrong – this is a type II error. So you accept the null (null says: no difference between the groups and you say ‘I agree – Oh no there isn’t..’), turns out that you’re wrong (oops again). That means that there is a difference between the two sets of data and you’ve said they are the same. You have just made a type II error. In medicine we find this as ‘false negative’ tests – see below. eg the test says that the patient doesn’t have HIV when actually they do

false positives and false negatives

		Condition (eg. HIV)
		Present	Absent
Test result	Positive	Condition Present + Positive result = True Positive	Condition absent + Positive result = False Positive Type I error
	Negative	Condition present + Negative result = False (invalid) Negative Type II error	Condition absent + Negative result = True (accurate) Negative

This table shows a way of looking at tests (eg. HIV test). The false positive is a type I error and the false negative is a type I error. A type I error is where the null hypothesis is falsely rejected. So a positive test  but no HIV (you say there ‘is a difference’ or here the test says there ‘is a difference’, but it is wrong)

That’ll do for now

Chin up

Dr B