Epidemiology 2 dec 2010

Epidemiology 2 Digby Tantam Clinical Professor of Psychotherapy, ScHARR Honorary Consultant Psychotherapist and Psychiatrist Honorary Senior Visiting Research Fellow, University of Cambridge

What is the prevalence of depression in the general elderly population? 0-6% 12-16% Can’t remember the rest. What is the prevalence of learning disability in the general population? 1% 2% 4% 5% 10%

Define population Define important factor in community For instance, what value do hallucinations have as a marker of psychosis in community? Or learning ‘disability’? What is a ‘disability’?

What is the prevalence of lifetime suicidal ideation in teenage children? 1% 3% 5% 10% 15%?

What is the prevalence of lifetime suicidal ideation in teenage children? 1% 3% 5% 10% 15% 37.9% Dervic et al, Vienna 20.5% Rodriguez-Figueroa& Linnett (E. Puerto Rico, 18% self harm) 17.6% Wang et al, Guanzhou

Which one of the following countries has an equal sex ratio in suicide? a) Russian Federation b) Japan c) China d) Mexico e) United Kingdom

A disease that is very rare and tends to lead to patients dying quickly would have: A high prevalence and a high incidence A low prevalence and high incidence A high prevalence and a low incidence A low prevalence and a low incidence ? A woman who has adopted a child from an alcoholic parent wants to know what the risk is of him also becoming an alcoholic.

Which of the following is the biggest risk factor for adult antisocial behaviour? Maternal depression Paternal depression Large family ADHD Autism

What is the best way to display the age and mortality of a population: Bar chart Line graph Population pyramid Stem and leaf chart Can’t remember

Stem and leaf plots are a way of presenting numerical data. The data found on the left hand side of the solid line is considered the stem, and the numbers on the right are the leaves. This particular stem and leaf plot depicts infant mortality rates in Western Africa. The rates range from 51 to 151. This stem and leaf plot was taken from http://mainland.cctt.org/mathsummer/JosephBond/StemAndPlots/images/table2.gif

Perinatal mortality Infant mortality Primary prevention Secondary prevention Tertiary prevention A patient with a manifest disease to prevent becoming a disability The ratio of infants who died under the age of 1 year to those who survived

Perinatal mortality (within 7/7 of birth) Infant mortality (within 12/12 of birth) Primary prevention Secondary prevention Tertiary prevention A patient with a manifest disease to prevent becoming a disability The ratio of infants who died under the age of 1 year to those who survived

Childhood sexual abuse Visual impairment Reduced hearing Death of a mother before the age of 14 Alcohol Which of the above is a risk factor? An elderly woman who lives alone develops auditory hallucinations A woman develops low mood, poor sleep

The prevalence of attempted suicide in Goth community is a. 53% b. 47% c. 25% d. 15% e. 10%

What is the prevalence of lifetime suicidal ideation in teenage children? 1% 3% 5% 10% 15%? Which one of the following countries has an equal sex ratio in suicide? a) Russian Federation b) Japan c) China d) Mexico e) United Kingdom Which of the following infections would not cause learning disability through trans-placental transmission? a) Cytomeglovirus b) Toxoplasmosis c) Treponema pallidum d) Herpes simplex e) Rubella Vascular dementia accounts for what proportion of all dementia? 0-10% 10-20% 20-30% 30-40% 40-50% The following is not a risk factor for Puerperal psychosis? A family history of Bipolar Affective Disorder Personal history of schizophrenia Personal history of Bipolar Affective Disorder Poor social support A disease that is very rare and tends to lead to patients dying quickly would have: A high prevalence and a high incidence A low prevalence and high incidence A high prevalence and a low incidence A low prevalence and a low incidence ? A woman who has adopted a child from an alcoholic parent wants to know what the risk is of him also becoming an alcoholic. 2 3 4 6 9 Which of the following is the biggest risk factor for adult antisocial behaviour? Maternal depression Paternal depression Large family ADHD Autism What is the best way to display the age and mortality of a population: Bar chart Line graph Population pyramid Stem and leaf chart Can’t remember Stubs The following is correct regarding SAD and Melatonin: No consistent evidence involving SAD Melatonin is secreted more throughout the day There is a delayed phase reaction Regarding bipolar, the following is true? Episodes get longer the older a patient gets If a patient develops BAD aged 20 it is suggestive of a strong family history The following can be considered a confounding factor: Women are more likely to suffer from depression Men have a younger age of onset for schizophrenia than women EMIs Perinatal mortality Infant mortality Primary prevention Secondary prevention Tertiary prevention A patient with a manifest disease to prevent becoming a disability The ratio of infants who died under the age of 1 year to those who survived Childhood sexual abuse Visual impairment Reduced hearing Death of a mother before the age of 14 Alcohol Which of the above is a risk factor? An elderly woman who lives alone develops auditory hallucinations A woman develops low mood, poor sleep May and November 2008 The rate of self harm in teenagers who identify with Goth culture is around a. 50% b. 10% c. 15% d. 25% e. 75% The onset of puberty is during the a. the preadolescent phase b. early adolescence c. mid adolescence d. late adolescence e. Adulthood The suicide rate in the UK in 2007 was a. 1 in 10000 b. 1 in 10 c. 1in 100 d. 1in 1000 e. 1 in 100000 The most common type of personality disorder prevalent in the UK a. Anxious-avoidant b. Borderline c. Schizoid d. Paranoid e. Dissocial The prevalence rate of deliberate self harm in adolescents is a. 6.90% b. 12.60% c. 11.20% d. 3.20% e. 25% What is the chance of children developing schizophrenia when their mother suffers from this illness? a. 5-10% b. 10-15% c. 15-20% d. 20-25% e. 40-50% The following is true about the association between schizophrenia and cannabis a. Cannabis leads to a two fold increase in the risk of schizophrenia b. Cannabis leads to a three fold increase in the risk of schizophrenia c. Cannabis leads to 5 fold increase in schizophrenia d. There is no association between cannabis and schizophrenia e. Cannabis reduces the risk of developing schizophrenia The Suicide rate in patients with Epilepsy as compared to general population is a. five folds more b. same c. five fold less d. 25 times more e. difficult to compare The prevalence of reading difficulties in Japanese children is a. 1.00% b. 3.00% c. 5.00% d. 9% e. 19% The prevalence of schizophrenia in patients over 65 years is a. 0.50% b. 1% c. 2% d. 5% e. 10% The prevalence of Alzheimer's disease in people aged more than 65 years is a. 10.00% b. 15.00% c. 20.00% d. 25.00% e. 30% The prevalence of specific reading retardation in UK is a. 4% b. 14% c. 20% d. 0.40% e. 24% The prevalence of ADHD in the UK is a. 1% b. 3% c. 5% d. 7% e. 9% The prevalence of attempted suicide in Goth community is a. 53% b. 47% c. 25% d. 15% e. 10%

Inception rate Incidence rate Prevalence Odds

Types of data e.g. categorical, ordinal, continuous. Concepts of scale of measurement, sampling methods. Problems of measurement in psychiatry, latent traits (constructs) and observed indications (symptoms). Ideas of reliability and validity. Sensitivity, specificity and predictive values of research measures. 9.15-10.45 1 hr 30 11.15-12.45 1hr 30

Scale Level Scale of Measurement Scale Qualities Example(s) 4 Ratio (scale for SPSS) Magnitude Equal Intervals Absolute Zero Age, Height, Weight, Percentage Geometric mean, log transform 3 Interval (scale for SPSS) Magnitude Equal Intervals Temperature 2 Ordinal Magnitude Likert Scale, Anything rank ordered 1 Nominal (categorical) None Names, Lists of words 0 Dichotomous: both nominal, and pseudo-interval None, but can be combined Yes/ No

Visual analogue scale Is this a Likert scale?

Likert scale Most commonly used scale in survey research Composed of Likert items, which are bipolar: Strongly disagree Disagree Neither agree nor disagree Agree Strongly agree Is Likert ordinal or interval? Central tendency for response Bias to positive: so have equal no. of positive and negative items

What kinds of variables are these? The place in a horse race (1 st . 2 nd . 3 rd ) The winner’s time in a horse race The handicap in horse race The means of differences Whether or not a particular horse was entered

Continous and discontinuous variables Integers are discontinuous Numbers are continuous Age in years is discontinuous GHQ score is discontinuous The average age of a population is continuous The average GHQ score of a consecutive series of OP attenders is… Depends on the N

Parametric and non-parametric Parametric statistics assume that data are samples of a normal distribution. Only continuous data can be normally distributed (or scale data to use SPSS terminology) i.e. ratio or interval data If in doubt, plot Can transform continuous data Commonest reason for transformation is ceiling effect Statistics for non-parametric data are less robust and less likely to find a statistically significant effect

Parametric and nonparametric tests Parametric Student’s t test Tests based on z, the normal variate, or (x- μ )/ σ Even if variables are not parametric their means may be (central limit theorem) Non-parametric Tests of ordinal data, or rank methods: Mann-Whitney, Spearman’s, latent trait analysis Dichotomous tests: χ 2

Sampling The whole point of sampling is to find out about the population without asking everyone in it. So generalizability is the issue This issue is fudged by qualitative research although it can rationally be replaced by representativeness and the concept of ‘saturation’ Creswell (2002) has recommended that 3-5 participants be used for case study research. Also, with respect to phenomenological studies, sample size recommendations range from 6 (Morse, 1994) to 10 (Creswell, 1998). For grounded theory research, sample size guidelines have ranged from 15-20 participants (Creswell, 2002) to 20-30 participants (Creswell, 1998). With regard to ethnographic research, Morse (1994) has recommended that 30-50 interviews be conducted. Finally, with regard to the use of focus groups, the following recommendations have been made: 6-9 participants (Krueger, 2000); 6-10 participants (Langford, Schoenfeld, & Izzo, 2002; Morgan, 1997); 6-12 participants (Johnson & Christensen, 2004); 6-12 participants (Bernard, 1995); 8-12 participants (Baumgartner, Strong, &..

Probability sampling Random sampling from whole population Cluster sampling: random sampling within naturally defined clusters within populations e.g. people in Heeley, people in Hunters Bar Systematic e.g. consecutive series of referrals, every 10 th person Stratified sampling enables over-sampling Can calculate a sampling error using a power calculation, in the same way that one can calculate in advance the necessary sample size to find a statistically significant result Statistical power depends on: the statistical significance criterion used in the test the size of the difference or the strength of the similarity (that is, the effect size ) in the population the sensitivity of the data. One of the most important determinants of sensitivity is sample size . The advantage of probability sampling is that sampling error can be calculated. Sampling error is the degree to which a sample might differ from the population. When inferring to the population, results are reported plus or minus the sampling error. In nonprobability sampling, the degree to which the sample differs from the population remains unknown. Random sampling is the purest form of probability sampling. Each member of the population has an equal and known chance of being selected. When there are very large populations, it is often difficult or impossible to identify every member of the population, so the pool of available subjects becomes biased. Systematic sampling is often used instead of random sampling. It is also called an Nth name selection technique. After the required sample size has been calculated, every Nth record is selected from a list of population members. As long as the list does not contain any hidden order, this sampling method is as good as the random sampling method. Its only advantage over the random sampling technique is simplicity. Systematic sampling is frequently used to select a specified number of records from a computer file. is commonly used probability method that is superior to random sampling because it reduces sampling error. A stratum is a subset of the population that share at least one common characteristic. Examples of stratums might be males and females, or managers and non-managers. The researcher first identifies the relevant stratums and their actual representation in the population. Random sampling is then used to select a sufficient number of subjects from each stratum. " Sufficient " refers to a sample size large enough for us to be reasonably confident that the stratum represents the population. Stratified sampling is often used when one or more of the stratums in the population have a low incidence relative to the other stratums. Convenience sampling is used in exploratory research where the researcher is interested in getting an inexpensive approximation of the truth. As the name implies, the sample is selected because they are convenient. This nonprobability method is often used during preliminary research efforts to get a gross estimate of the results, without incurring the cost or time required to select a random sample. Judgment sampling is a common nonprobability method. The researcher selects the sample based on judgment. This is usually and extension of convenience sampling. For example, a researcher may decide to draw the entire sample from one "representative" city, even though the population includes all cities. When using this method, the researcher must be confident that the chosen sample is truly representative of the entire population. Quota sampling is the nonprobability equivalent of stratified sampling. Like stratified sampling, the researcher first identifies the stratums and their proportions as they are represented in the population. Then convenience or judgment sampling is used to select the required number of subjects from each stratum. This differs from stratified sampling, where the stratums are filled by random sampling. Snowball sampling is a special nonprobability method used when the desired sample characteristic is rare. It may be extremely difficult or cost prohibitive to locate respondents in these situations. Snowball sampling relies on referrals from initial subjects to generate additional subjects. While this technique can dramatically lower search costs, it comes at the expense of introducing bias because the technique itself reduces the likelihood that the sample will represent a good cross section from the population.

Nonrandom sampling Matched samples in case-control study Convenience sampling e.g. all the people on a ward. Judgment sampling : researcher chooses sample and then has to justify that they are representative and therefore that results are generalizable Quota sampling: convenience or judgement samples taken from clusters Snowball sampling e.g. my own PhD research

Questionnaire design Think about motivating people to complete it: I never do Psychometric properties: Response bias Items only measure one thing Are unambiguous Are really necessary (ask yourself, how am I going to use this information) Do not exceed respondents language or visual abilities Are easy for you to score (beware hidden lists of dichotomous variables) Number of variables should be 1/6 the number of participants Items are not redundant Can be analysed as they stand Are internally consistent, reliable and valid

Internal consistency vs. redundancy Measure the same thing: not needed (GHQ-12 seems to do the same job as the original GHQ-40) Too different: increases variance Cronbach’s alpha is measure of individual variance contributed by each item in relation to the overall co-variance or common variance to which all the items contribute How happy are you, are you happy right now, does life feel pleasing to you, all have a common variance due to an underlying factor Do you enjoy life, do you look forward to your day, are you happy each might have their own variance although there is an underlying construct or factor of ‘happiness’

Validity Face or content: using expert panels, supervisor and supervisee, one’s own judgement, users’ groups Test: versus a criterion (often clinical judgement in psychiatry) Construct Convergent: correlations with other measures (concurrent is a special case of convergent), to be contrasted with predictive) Discriminant: no correlation with measures of independent constructs Criterion Predictor of future or of some independent corollary

Reliability: can be measured using correlations Inter-rater Test-retest Split-half

Understanding statistical terms: 4- Diagnostic tests. Drug and Therapeutics Bulletin 47 (6) June 2009, 71-72

Sensitivity is? 30/30=100% Specificity is? 70/70=100% Sample of 100 people (shown as cricles) Filled circles, people with disease Blank circles spots, people without Dotted line surrounds people who tested positive on a screening test Sensitivity of test Those who test positive and have the disease/ those who have the disease Specificity Those who test negative and do not have the disease/ those who do not have the disease

False positive is? 11 False negative is? 21 Sample of 100 people (shown as cricles) Filled circles, people with disease Blank circles spots, people without Dotted line surrounds people who tested positive on a screening test False positive Those who test positive but do not have the disease False negative Those who test negative and do have the disease

Sensitivity is? 9/30= 30% Specificity is? 59/70=84% Sample of 100 people (shown as cricles) Filled circles, people with disease Blank circles spots, people without Dotted line surrounds people who tested positive on a screening test Sensitivity of test Those who test positive and have the disease/ those who have the disease Specificity Those who test negative and do not have the disease/ those who do not have the disease

Positive likelihood value: How much likely than not , a person who tested positive has the disease (true positive/ false positive or sensitivity/ 1-specificity) Negative likelihood value How much less likely than not a person who tested negative has the disease (false negative/ true negative or 1-sensitivity/specificity) True positive= 9 False positive=11 Positive likelihood value is? 0.82 False negative= 21 True negative = 70 Negative likelihood value is? 0.30

Positive predictive value (PPV): Probability of having disease if positive test Negative predictive value (NPV) Probability of not having disease if test negative Diagnostic test accuracy Number with disease who test positive+ number without disease who test negative/ Total sample N True positive= 9 False positive=11 Positive predictive value is? 9/(9+11)= 0.45 True negative= 48 All negative =80 Negative likelihood value is? 0.60 Diagnostic accuracy is? 9+59/100=68%

Copyright ©1994 BMJ Publishing Group Ltd. Altman, D G et al. BMJ 1994;309:188 FIG 1 (left) – Receiver operating characteristic curve Distribution of values of an index of mixed epidermal cell lymphocyte reactions in patients who did or did not develop grafts versus host disease3 Area under the curve is the ‘diagnostic accuracy’

Which has the greater diagnostic accuracy?

Familial aggregation studies: Is there a genetic component to the disease, and what are the relative contributions of genes and environment? Segregation studies: What is the pattern of inheritance of the disease (e.g. dominant or recessive)? Linkage studies: On which part of which chromosome is the disease gene located? Association studies: Which allele of which gene is associated with the disease?

How might these studies be carried out epidemiologically? Familial aggregation Segregation (e.g. dominant or recessive)? Linkage studies: On which part of which chromosome is the disease gene located? Association studies: Which allele of which gene is associated with the disease?

Classification of the trinucleotide repeat, and resulting disease status, depends on the number of CAG repeats [13] Repeat count Classification Disease status <28 Normal Unaffected 28–35 Intermediate Unaffected 36–40 Reduced Penetrance +/- Affected >40 Full Penetrance Affected

Genomic epidemiology What polymorphisms in human genome can be linked to disease? Counter-intuitive findings: CNVs and other polymorphisms common

Segregation The first law of Mendel Each gamete of a parent has one ‘hereditary unit’ or allele. Alleles may be dominant (expressed even if heterozygous) or repressed (expressed only if homozygous) Huntington’s chorea, gene, codes for mHuntingtin that interferes with normal action of huntingtin on ?DNA transcription. The presence of a small amount of mHuntingtin blocks Huntingtin and so Huntington’s chorea is ??

The authors performed a meta-analysis of studies examining the association between polymorphisms in the 5,10-methylenetetrahydrofolate reductase (MTHFR) gene, including MTHFR C677T and A1298C, and common psychiatric disorders, including unipolar depression, anxiety disorders, bipolar disorder, and schizophrenia. The primary comparison was between homozygote variants and the wild type for MTHFR C677T and A1298C. For unipolar depression and the MTHFR C677T polymorphism, the fixed-effects odds ratio for homozygote variants (TT) versus the wild type (CC) was 1.36 (95% confidence interval (CI): 1.11, 1.67), with no residual between-study heterogeneity (I2 = 0%)--based on 1,280 cases and 10,429 controls. For schizophrenia and MTHFR C677T, the fixed-effects odds ratio for TT versus CC was 1.44 (95% CI: 1.21, 1.70), with low heterogeneity (I2 = 42%)--based on 2,762 cases and 3,363 controls. For bipolar disorder and MTHFR C677T, the fixed-effects odds ratio for TT versus CC was 1.82 (95% CI: 1.22, 2.70), with low heterogeneity (I2 = 42%)--based on 550 cases and 1,098 controls. These results were robust to various sensitively analyses. This meta-analysis demonstrates an association between the MTHFR C677T variant and depression, schizophrenia, and bipolar disorder, raising the possibility of the use of folate in treatment and prevention. Gilbody, S., S. Lewis, et al. (2007). "Methylenetetrahydrofolate Reductase (MTHFR) Genetic Polymorphisms and Psychiatric Disorders: A HuGE Review." Am. J. Epidemiol. 165 (1): 1-13.

Polymorphism Homozygous Gene Meta-analysis Fixed effects model: all studies come from a common population Between studies heterogeneity: different methods, different populations in studies included in meta-analysis Sensitivity analysis: looking at how assumptions in model might affect results. A simple method is to reanalyse excluding outliers

Suppose we assume that 5% of the people are drug-users. A test is 95% accurate, which we'll say means that if a person is a user, the result is positive 95% of the time; and if she or he isn't, it's negative 95% of the time. A randomly chosen person tests positive. Is the individual highly likely to be a drug-user? Given your conditions, once the person has tested positive, you may as well flip a coin to determine whether she or he is a drug-user. The chances are only 50-50.

Odds ratios Odds: The number of times an event occurs/ the number of times it does not If I throw a dice 6 times and get a ‘5’ on 2 occasions, the odds of getting a five vs not getting a five are? 2/4 or a half: but what’s wrong with that answer?

Association between hay fever and eczema in 11 year old children 1 From BMJ 2000;320:1468 ( 27 May ) Eczema Hay fever Total Yes No Yes 141 420 561 No 928 13 525 14 453 Total 1069 13 945 15 522

If a child has eczema (Estimated) probability that they will have hay fever is 141/561 (25.1%) Odds that they have hay fever is 141/420 If a child does not have eczema, Probability of having hay fever is 928/14 453 (6.4%) Odds of the child having hay fever is 928/13 525 Odds that a child with eczema will also have eczema is the ratio of the two odds is (141*13 525)/ (928*420) i.e. 4.89 Odds cannot be negative so it’s distribution is skewed, but ln(odds) is normally distributed, and so a standard error and confidence interval can be calculated

The Medicines and Healthcare Products Regulatory Agency announced the withdrawal in 2005 of co-proxamol over a 2 year period Keith Hawton said that co-proxamol was responsible for a fifth of all drug-related suicides before 2005, but his study showed that by 2007 prescribing of the drug had fallen by 59% Over the two-year period, deaths from co-proxamol fell by 62%. Specifically there were 295 fewer suicides and 349 fewer deaths from the drug including accidental overdoses. BBCwebsite 19 June 09

Cause cannot be inferred from associations/ correlation does not imply causation The more firemen fighting a fire, the bigger the fire is going to be. Therefore firemen cause fire. Sleeping with one's shoes on is strongly correlated with waking up with a headache . Therefore sleeping in one’s shoes causes headache Young children who sleep with the light on are much more likely to develop myopia in later life. As ice cream sales increase, the rate of drowning deaths increases sharply. Therefore, ice cream causes drowning. Since the 1950s, both the atmospheric CO 2 level and crime levels have increased sharply. Hence, atmospheric CO 2 causes crime.

The third factor problem Violence and unemployment occur together. If one does not cause another, then there must be a third factor that causes both The third factor is (pick whichever is ideologically desirable) Discrimination Social disadvantage Parenting Personality disorder

Symptoms vs. ‘constructs’ Factor analysis Latent traits analysis Categorical classifications using weighting combinations Neural net methods Are there underlying constructs?

Epidemiology 2 dec 2010

More Related Content

What's hot

Viewers also liked

Similar to Epidemiology 2 dec 2010

Recently uploaded

Epidemiology 2 dec 2010